Author`s personal copy - Laboratoire d`étude des Transferts en

Transcription

Université Joseph Fourier de Grenoble
Spécialité : Sciences de la Planète
Synthèse des travaux de recherche
Présentée en vue de l’obtention de
L’Habilitation à Diriger des Recherches
Aquifères, recharges
et transferts d’eau en zone non-saturée:
Caractérisation par spatialisation
et suivi temporel géophysique
Marc Descloitres
Ingénieur de Recherche à l’IRD
LTHE (UMR IRD/UJF/CNRS/G-INP)
Equipe Hydrogéophysique
CERMO, 460 Rue de la piscine
BP 53, 38041 Grenoble Cedex 9
Jury :
MM
Michel CHOUTEAU, Ecole Polytechnique de Montréal,
Rapporteur
Michel DIAMENT, IPG Paris,
Rapporteur
Jean Pierre GOURC, LTHE Grenoble,
Examinateur
Anatoly LEGCHENKO, LTHE Grenoble,
Examinateur
Olivier RIBOLZI, LMTG Toulouse,
Examinateur
Pascal SAILHAC, EOST Strasbourg,
Rapporteur
Date de soutenance : 31 mai 2010
Dossier HDR – M. Descloitres, LTHE, 2010
1
Résumé
Ce document présente la synthèse de mes travaux de recherche, qui portent sur l’étude
des aquifères et des processus de transferts d’eau dans le sous-sol par les outils géophysiques.
Ces recherches ont été menées sur les chantiers de l’IRD dans les pays du Sud, souvent
confrontés à la rareté des eaux de surface, et qui se tournent de plus en plus vers l’exploitation
des ressources en eau souterraine. Les études antérieures permettent de décrire les principaux
modèles conceptuels des aquifères et des processus hydrologiques associés (recharge,
infiltration dans la proche surface). Néanmoins, selon le contexte géologique et climatique,
des questions restent en suspens, comme le rôle que pourraient jouer les ravines de versant
dans la recharge, ou comme la description fine du cycle de l’eau à l’échelle des tous premiers
décimètres des sols sableux, support de la végétation. Pour mieux gérer la ressource, il est
nécessaire d’avoir une meilleure connaissance i) des processus de recharge à l’échelle
régionale, et ii) de la structure du sous-sol, notamment des formations argileuses, moins
perméables. En zone de socle, les ressources en eau sont modestes et leur prospection est
difficile sans l’aide des outils de la géophysique.
Mes activités de recherches sont déclinées selon trois axes thématiques: i) la
spatialisation des aquifères, ii) l’étude de leur recharge dans différents contextes géologiques
et climatiques, et iii) la spatialisation des processus élémentaires de transfert de l’eau en zone
non saturée. Un quatrième axe, méthodologique, permet l’adaptation de certains outils
géophysiques aux spécificités des aquifères et des processus hydrologiques en jeu. La
démarche scientifique utilisée comprend des allers-retours entre modélisations numériques et
études de terrain menées sur des bassins versants expérimentaux permettant de disposer de
données extérieures, et de croiser les interprétations avec d’autres disciplines. Les méthodes
géophysiques utilisées sont principalement les méthodes d’imagerie de résistivité électrique
(méthodes à courant continu ou électromagnétiques), et la Résonance Magnétique des
Protons, méthode récente de l’hydrogéophysique. Un aspect méthodologique important est
l’utilisation des suivis temporels pour traquer les lieux de recharge et les transferts d’eau dans
les sols, ceux-ci créant des contrastes géophysiques suffisants pour les suivre spatialement et
temporellement. Les résultats peuvent être synthétisés ainsi :
i) La spatialisation géophysique des aquifères de socle à l’échelle du bassin versant permet
de mieux comprendre l’organisation des régoliths (réservoir) et de leurs propriétés
hydriques. A l’échelle régionale, ce sont les systèmes aquifères sédimentaires qui se
prêtent le mieux à une spatialisation, en particulier s’ils comprennent des formations
argileuses. Dans tous les cas, l’utilisation de la RMP permet de quantifier la ressource
en présence.
ii) Les lieux de recharge dans les versants dépendent du contexte géologique et climatique.
Au sahel nous avons mis en évidence des processus d’infiltrations sous les versants
sableux en zone sédimentaire, alors que dans les zones de socle, les versants ne
participent pas à la recharge. En zone plus humide, les suivis temporels géophysiques
permettent de connaître la forme des infiltrations temporaires sous les ravines.
iii) Les processus hydriques dans les sols sableux peuvent être appréhendés par des suivis
temporels de résistivité. Ceux-ci permettent de connaître les zones préférentielles
d’infiltration et de dessiccation lors de cycles courts. Dans les sols argileux, les relations
entre résistivité et variables hydrologiques sont plus difficiles à établir et cela limite les
possibilités d’emploi de la résistivité.
2
iv) D’un point de vue méthodologique, ces études mettent en évidence les sérieuses
difficultés rencontrées lors de certains suivis temporels de résistivité. Nos études
récentes permettent de proposer une fiabilisation de ces suivis. L’apport de la RMP a été
déterminant, par sa capacité à quantifier des paramètres hydriques des terrains étudiés.
Dans le cadre des prolongements du programme AMMA en Afrique de l’Ouest, les
perspectives de mon travail auront principalement pour objectif la spatialisation des
ressources en eau souterraine des zones de socle, l’identification des zones de recharge à plus
grande échelle, en utilisant un couplage accru entre la RMP et les méthodes de résistivité,
ainsi que les potentialités offertes par les suivis temporels électromagnétiques.
3
Liste des abréviations utilisées
•
AMMA : Analyse Multidisciplinaire de la Mousson Africaine
•
AMMA-Catch : Observatoire ORE « Couplage de l’ATmosphére et du Cycle Hydrologique »
•
ANR Ghyraf: Agence Nationale de la Recherche, projet « Gravimétrie et Hydrologie en
Afrique »
•
BRGM : Bureau de Recherche Géologique et Minière
•
BVET : ORE « Bassin Versants Expérimentaux Tropicaux »
•
EAGE : European Association of Geoscientists and Engineers
•
EC2CO « Ondine » : projet Ecosphère Continentale et Côtière » Impact des changements
d’usages des terres sur la genèse des crues »
•
ERT : Electrical Resistivity Tomography
•
GEOFCAN : Réseau de Recherche sur la Géophysique des Couvertures Anthropisées et
Naturelles
•
HSM : Laboratoire HydroSciences Montpellier
•
IRD : Institut de recherche pour le Développement
•
LMTG : Laboratoire de Mécanique des Transferts en Géologie
•
LTHE : Laboratoire d’Etude des Transferts en Hydrologie et Environnement
•
ORE : Observatoire Régional en Environnement (maintenant SO : Service d’Observation)
•
OSUG : Observatoire des Sciences de l’Univers de Grenoble
•
RMP : Résonance Magnétique des Protons
•
TDEM : Time Domain Electromagnetism
•
ZNS : zone non-saturée
4
Table des matières
Pages
1ère partie : CURRICULUM VITAE
1 Diplômes
7
2. Etapes principales de ma carrière
7
7
8
9
9
10
11
2.1
2.2
2.3.
2.4.
2.5.
2.6.
Résumé
1992-1996: Affectation au centre IRD de Dakar, Sénégal
1996-1999: Préparation d’une thèse à Paris
2000-2003: Affectation au Burkina Faso, UR Geovast
2003-2006: Visiting scientist à l’Indian Institute of Science, Bangalore, Inde
2006-2010: Affectation au LTHE à Grenoble
3. Synthèse des publications
12
4. Encadrements et enseignements
13
13
13
16
4.1 Direction de thèses
4.2 Tableau des encadrements d’étudiants de 3ème cycle, 1992-2009
4.3 Enseignements
5. Participation à des projets scientifiques
16
2ème partie : TRAVAUX DE RECHERCHE
Introduction
18
1. Questions scientifiques abordées
1.1. Les aquifères de socle
a) Distribution verticale des propriétés hydrauliques
b) Aquifères de socle sous climat sahélien
c) Aquifères de socle sous climat soudanien à guinéen
1.2 Les aquifères sédimentaires en zone semi-aride
1.3 Les processus élémentaires à l’échelle de la parcelle ou de la séquence de sols
2. Méthodologie
2.1. Approche générale
2.2. Principales méthodes géophysiques employées
a) Méthodes de résistivité
b) La Résonance Magnétique des Protons (RMP)
c) Autres méthodes utilisées
2.3. Exemples de mise en œuvre des méthodes géophysiques
a) Site de Katchari, Burkina Faso
b) Site de Moole Hole, Inde du Sud
c) Site d’Ara, Nord Bénin
d) Autres sites étudiés
3. Synthèse des résultats
3.1. Spatialisation des aquifères
a) Spatialisation des altérites de socle et de leurs paramètres hydriques
b) Spatialisation régionale des aquifères sédimentaires
c) En résumé
3.2 Recharge des aquifères
a) Les ravines de versant sont-elles des lieux de recharge ?
b) Comment s’effectue la recharge des aquifères par les ravines de bas fond ?
19
19
20
21
22
23
25
27
27
29
29
31
32
32
33
34
34
35
36
36
36
40
43
44
44
48
5
c) En résumé
3.3. Etude des transferts d’eau dans les premiers décimètres du sol
a) Les transferts d’eau dans une micro-dune lors de cycles de pluie
b) Les transferts d’eau dans un système de sols
c) En résumé
3.4. Apports méthodologiques
a) Conception d’une sonde de diagraphie de résistivité en zone non-saturée
b) Vers une fiabilisation des imageries de suivi temporel de résistivité
4. Perspectives
Introduction
4.1 Projet de recherche au Bénin
a) Spatialiser les ressources en eau en zone de socle à l’échelle régionale
b) Spatialiser la recharge à l’échelle du bassin versant
c) Mieux quantifier le bilan de l’eau souterraine à l’échelle du site
4.2 Autres projets
a) Spatialiser la dynamique des infiltrations par suivi temporel de résistivité
b) La résistivité est-elle un marqueur de la dégradation des déchets
50
51
51
55
57
58
58
58
62
62
63
63
65
66
68
68
69
Conclusion
71
Références bibliographiques
72
Annexe 1. Liste des publications
77
1.
2.
3.
4.
5.
6.
7.
8.
Publications dans des journaux à comité de lecture
Articles soumis
Brevet
Conférences internationales avec actes
Conférences nationales, colloques et réunions scientifiques
Rapports de contrats, de mission
Contributions diverses
Relectures d’articles pour revues à comité de lecture
Annexe 2. Tirés à part des principaux articles
78
79
80
80
82
84
85
85
86
6
CURRICULUM VITAE
Marc DESCLOITRES
> Géophysicien, Ingénieur de Recherche 1ère classe à l’IRD
> Depuis 2006 : membre de l’équipe Hydrogéophysique (HGP) au Laboratoire d’étude des
Transferts en Hydrologie et Environnement (LTHE), Université de Grenoble.
Bat. CERMO, bureau 333
460 Rue de la Piscine
BP 53, 38041 GRENOBLE
Tel 04.76.63.56.59
marc.descloitres@ird.fr
1 Diplômes
¾
1998 : Thèse de Doctorat, Université de Paris 6, Laboratoire de Géophysique
Appliquée. Sujet : « Les sondages électromagnétiques en domaine temporel (TDEM) :
Application à la prospection d’aquifères sur les volcans de Fogo (Cap Vert) et du Piton
de la Fournaise (la Réunion) ». Thèse de Doctorat de l’Université de Paris 6, direction A.
Tabbagh.
¾
1986 : DEA « Mécanique des Milieux Géophysiques et Environnement », IRIGM,
Grenoble. Sujet : « Modélisation analytique des contraintes dans une formation calcaire
soumise à des poussées horizontales».
¾
1985 : Maîtrise de Géologie Expérimentale, Université Grenoble 1.
2. Etapes principales de ma carrière
2.1 Résumé
Depuis 1986, j’ai exercé les fonctions d’ingénieur en géophysique dans diverses
structures (CEA, bureau d’étude) et depuis 1991 à l’ORSTOM (IRD). Après ma thèse de
doctorat en 1998, mon activité s’est orientée vers la conception, la réalisation et la
valorisation de recherches en géophysique appliquée aux eaux souterraines, ce qu’on appelle
maintenant l’hydrogéophysique. La plupart de mes activités se sont déroulées dans les pays
du Sud où l’IRD développe des partenariats, déploie des dispositifs d’observation de
l’environnement, et nous incite à former nos partenaires par la recherche. J’ai ainsi alterné des
périodes d’expatriation avec des périodes dans des laboratoires en France (tableau 1).
7
Poste
Organisme
Structure
Equipe
Projet
« Houillères de
Provence »
Ingénieur
CEA
VSN
CEA
Ingénieur
SIMECSOL
Ingénieur
d’étude
2ème classe
(IE2)
ORSTOM
(IRD)
Laboratoire de
Détection
Géophysique
Observatorio
San Calixto
Bureau d’étude
(120
ingénieurs)
Centre IRD
d’île de France
ORSTOM
(IRD)
ORSTOM
(IRD)
Centre IRD de
Dakar
Université de
Paris 6
Géophysique
IRD
Centre IRD de
Ouagadougou
IRD
Indian Institute
of Science
IRD
LTHE
IE2
IE2
Ingénieur de
Recherche
2ème classe
(IR2)
Ingénieur de
Recherche
1ère classe
(IR1)
IR1
Lieux
Dates
Paris
1987
La Paz,
Bolivie
Paris
1988
1990
Bondy
1991
Dakar,
Sénégal
Paris
19921996
19961999
UR Geovast
Ouagadougou,
Burkina Faso
20002003
Cellule Franco
Indienne de
Recherche sur
l’Eau
Hydrogéophysique (HGP)
Bangalore,
Inde
20032006
Grenoble
20062010
« Mesures
dynamiques »
Géophysique
Laboratoire de
géophysique
appliquée
Tableau 1. Postes occupés durant ma carrière.
2.2. 1992-1996 : Affectation au centre IRD de Dakar, Sénégal
Ingénieur géophysicien au centre IRD de Dakar, je développe, sous la responsabilité
de Michel Ritz, une activité basée sur l’application de méthodes géophysiques aux problèmes
d’eau souterraine. Les projets sur les aquifères du Sénégal en zone côtière, puis sur l’aquifère
profond du Maastrichien étudié par l’équipe de Michel Chouteau de l’école Polytechnique de
Montréal (Giroud et al., 1997) nous confortent sur la nécessité de renforcer à l’IRD une
thématique géophysique dédiée aux eaux souterraines. Yves Albouy, géophysicien au centre
IRD de Bondy, structure ces projets avec le programme « Géaquif » (Géophysique appliquée
aux aquifères), et deux projets majeurs vont émerger en dehors du Sénégal:
¾ Le projet « Hydrofournaise » à l’île de La Réunion permet d’appliquer des sondages
électromagnétiques de résistivité aux aquifères: Audio-Magnétotéllurique (AMT) et
Very Low Frequency (VLF) tout d’abord, puis utilisation du Time Domain ElectroMagnetism (TDEM) sous l’impulsion de Pierre Andrieux (Paris 6) et d’Yves Albouy
(IRD Bondy). A la suite de ces études, l’application de ces méthodes pour la
prospection d’aquifères volcaniques est validée (Courteaud et al., 1996 ; Robineau et
al., 1997 ; Ritz et al., 1997 ; Descloitres et al., 1997).
¾ Le projet « Aquifères du volcan Fogo » (archipel des îles du Cap Vert) étudie les
aquifères de la caldeira et des flancs du volcan avec la méthode TDEM. Nos mesures
TDEM montrent dans la caldeira des problèmes majeurs. Je décide de monter un
8
projet de recherche avec la mission de coopération Française au Cap-Vert, et les
données collectées serviront pour la réalisation de ma thèse.
Parallèlement à la réalisation de ces projets, des collègues travaillant au Cameroun (H.
Robain et J. J. Braun) nous sollicitent pour prospecter les sols de bassins versants en forêt
équatoriale (Robain et al., 1996). Grâce à l’émergence de nouvelles méthodes de mesures
(tomographie de résistivité électrique, radar géologique), je contribue à la constitution d’un
pôle d’étude des sols tropicaux par méthodes géophysiques. Ces études se poursuivront plus
tard avec la création de l’équipe «Geovast», faisant suite au programme «Géaquif».
2.3 . 1996-1999: Préparation d’une thèse à Paris
La mise en évidence de courbes de sondage TDEM anormales dans la caldeira du
volcan Fogo me conduit à réaliser une thèse au Laboratoire de Géophysique Appliquée de
l’Université de Paris 6 sous la direction d’Alain Tabbagh. Cette étude ne sera pas relatée dans
ce document. Pour résumer, elle m’amène à proposer une méthodologie adaptée à la
reconnaissance et au traitement des anomalies de résistivité complexe en TDEM (Descloitres
et al, 2000).
En parallèle, une nouvelle idée naît dans l’équipe Géaquif : celle d’introduire une
paramétrisation à base de données géophysiques dans les modèles hydrogéologiques. Je
monte, avec Roger Guérin de Paris 6, une action qui vient renforcer le programme PNRH de
notre collègue Anne Coudrain (Paris 6) portant sur la dynamique d’un aquifère salé
d’extension régionale en Bolivie (altiplano). Nos collègues hydrogéologues comptent
contraindre la modélisation des flux souterrains par les limites géométriques données par la
géophysique. Grâce aux données acquises en TDEM, j’établis une relation régionale
permettant de délimiter des zones argileuses peu conductrices hydrauliquement (Guérin et al.,
2001). Enfin, divers projets d’applications du TDEM et du radar géologique voient le jour, à
mon initiative:
• Premières prospections radar appliquées à la reconnaissance des épaisseurs des
glaciers en France et en Bolivie au sein de l’équipe « Great Ice » de l’IRD
(Descloitres et al., 1999). Les résultats sont largement utilisés dans la thèse d’Edson
Ramirez sur le glacier de Chacaltaya en Bolivie (Ramirez et al., 2000).
• Utilisation de la méthode TDEM pour l’étude des glissements de terrain argileux,
dans le cadre de la thèse de Myriam Schmutz (Université de Strasbourg, Schmutz et
al., 1999 et 2000).
2.4 . 2000-2003 : Affectation au Burkina Faso, UR Geovast
Fin 99, je contribue à définir les objectifs d’une Unité de Recherche IRD montée par
Henri Robain, l’UR « Geovast », qui réunit pédologues et géophysiciens. Cette équipe compte
étudier la spatialisation des sols tropicaux et leur dynamique hydrique par le suivi temporel de
paramètres géophysiques. Lors de mon affectation au Burkina Faso, je travaille
principalement sur le terrain de l’équipe « Erosion et Changements d’Usage des terres »
(ECU) de l’IRD où j’organise les opérations géophysiques. Elles sont couplées avec les
études hydrologiques menées par Olivier Ribolzi et Jean Pierre Delhoume. Nous étudions
avec nos méthodes respectives les mêmes processus hydriques de ces milieux semi-arides à
9
différentes échelles spatiales (de la parcelle au bassin versant), et temporelles (de l’évènement
pluvieux à la saison hydrologique). Les résultats obtenus montrent les avantages, mais surtout
les limites de certains suivis temporels de résistivité (Descloitres et al., 2003). La
problématique de recharge des aquifères est aussi abordée grâce aux suivis temporels de
résistivité en coupe 2D. Nous étudions aussi certains processus d’infiltration à fine échelle en
conditions semi contrôlées (Descloitres et al., 2008). Ces expériences seront décrites dans la
deuxième partie de ce mémoire car elles ont fondé une partie importante de mon travail de
recherche des années suivantes. A ce moment (2003), la question méthodologique que je me
pose est la suivante : « la méthode de suivi temporel de résistivité est-elle suffisamment fiable
pour en tirer des renseignements utiles à notre compréhension des cycles souterrains de
l’eau? ».
Parallèlement à ces études sur le processus, je poursuis aussi mes recherches sur la
caractérisation géophysique des aquifères de socle. J’encadre localement la thèse de Ghislain
Toé (Paris 6) dirigée par Pierre Andrieux et Yves Albouy, portant sur une thématique
d’imagerie géophysique des aquifères de socle (Toé et al., conf. EAGE, Paris, 2004). Je
conçois une étude sur l’imagerie en suivi temporel de résistivité lors de l’hydrofracturation
des forages de socle et encadre à cette occasion deux diplômants de Lausanne, Mathieu Beck
et Denis Girardet, avec Dominique Chapellier (Beck et al., conférence Geofcan, 2001). Enfin,
l’étude des aquifères de socle est abordée avec la Résonance Magnétique des Protons,
méthode émergente et sujet de la thèse de mon collègue Jean Michel Vouillamoz : nous
réalisons la première expérimentation de la méthode en zone de socle en Afrique (Vouillamoz
et al., 2005). D’autres thématiques voient le jour au Niger et du Bénin :
•
•
Au Niger, à l’occasion de la thèse de Sylvain Massuel (direction Guillaume Favreau)
nous essayons de mettre en évidence les lieux d’infiltration profonde le bassin versant
expérimental de Wankama (Massuel et al., 2006). A cette occasion, et en
collaboration avec mon collègue Yann Le Troquer, je conçois et fabrique un outil de
diagraphie, la sonde gonflable qui sera breveté par l’IRD. Il sera largement utilisé par
la suite, par exemple en Inde (Braun et al., 2009).
Au Bénin je réalise les premières mesures géophysiques sur le bassin versant du
Service d’Observation (SO « AMMA-Catch ») avec l’aide de Maxime Wubda que
j’encadre pour son stage de DESS. Les résultats, permettent d’enrichir la première
compréhension des processus hydrologiques au Bénin (Kamagaté et al., 2007). Aussi,
ce travail fournit des hypothèses sur le rôle des nappes perchées pour la modélisation
hydrologique faite par Mathieu Le Lay, en thèse au LTHE (Lelay et al., 2008).
2.5 . 2003-2006 : Visiting scientist à l’Indian Institute of Science, Bangalore,
Inde.
L’expertise acquise au Burkina Faso m’amène à participer à un projet initié par Jean
Jacques Braun et Henri Robain en collaboration avec nos collègues indiens de l’Indian
Institute of Science (IISc). Différentes problématiques hydrologiques attendent la contribution
des outils géophysiques sur les bassins versants expérimentaux de la Cellule Franco-Indienne
de Recherche en Science de l’Eau (CEFIRSE), à Bangalore. Je propose de réaliser une
spatialisation géophysique selon 3 échelles :
10
a) La parcelle de sols (collaborations L. Barbiéro, L. Ruiz et M. S. Mohan Kumar), qui
nous permet de réaliser des expériences de suivi temporel de résistivité sur tout le cycle de
mousson. Nous nous heurtons de nouveau à des problèmes d’artefacts d’imagerie de
résistivité. La cartographie géophysique des sols à l’échelle du bassin permet de compléter la
compréhension du fonctionnement des sols (Barbiéro et al., 2007).
b) Le versant (collaborations L. Ruiz et M. Sekhar). A cette échelle, il faut comprendre
la recharge des aquifères. L’utilisation de la méthode RMP par A. Legchenko montre l’apport
de la modélisation 2D (Legchenko et al., 2006). Les résultats sont comparés avec succès avec
les images de résistivité 2D, et je montre la faisabilité du suivi temporel en RMP pour l’étude
de la recharge localisée (Descloitres et al., 2008). Ces études concrétisent nos espoirs de voir
cette méthode applicable à des géométries d’aquifère complexes Pour le suivi temporel de
résistivité, j’ai retrouvé à cette échelle les mêmes problèmes que ceux rencontrés au Burkina
Faso. D’autres études montrent des problèmes similaires (Kemna et al., 2004). Je décide d’en
faire un thème de recherche méthodologique et la thèse de Rémi Clément que je co-encadre
au LTHE portera partiellement sur cette thématique.
c) Le bassin versant (collaborations J. J Braun, J. Riotte et nos collègues de l’IISc).
L’objectif final est de réaliser le bilan géochimique d’altération de ces roches cristallines sous
climat tropical. Nous avons réalisé une étude couplée géochimie/géophysique, et Simon
Fleury, stagiaire de l’EOST de Strasbourg, a réalisé sous ma direction une étude sur les
incertitudes des paramètres d’inversion en imagerie de résistivité (Braun et al., 2009).
Enfin, depuis l’Inde, je participe au montage de la nouvelle équipe Hydro-GéoPhysique (HGP) au LTHE, que je rejoins en 2006, pour approfondir ma formation
d’utilisateur de la méthode RMP.
2.6 . 2006-2010 : Affectation au LTHE à Grenoble
A mon arrivée au LTHE mes objectifs sont de fiabiliser le suivi temporel de résistivité,
et contribuer ainsi à l’étude du bilan de l’eau. Outre la valorisation de nos actions « Inde », je
participe à plusieurs projets :
a) La spatialisation des transferts d’eau dans le sous-sol par suivi temporel de
résistivité nécessite une amélioration des procédures d’imagerie existantes. C’est
un des volets de la thèse de Rémi Clément. Récemment, des avancées très
significatives ont été obtenues (Clément et al., 2009), grâce à l’outil d’inversion
développé par le Dr Thomas Gunther du Leibniz Institute of Applied Geophysics
de Hanovre avec qui nous avons monté une collaboration. D’autre part, nous
profitons du projet ANR « Bioréacteur » (coordination J. P Gourc, LTHE) pour
appliquer cette méthodologie à une thématique environnementale majeure, celle de
l’optimisation de la biodégradation des déchets ménagers (Clément et al., 2010).
J’y greffe aussi un suivi temporel de résistivité par sondage électromagnétique
TDEM.
11
b) Une approche couplée entre géophysique et modélisation est de nouveau tentée au
Niger avec mes collègues IRD d’HydroSciences Montpellier (HSM). Cette
approche est initiée au Niger par J. M. Vouillamoz et G. Favreau et j’apporte ma
contribution à ce projet en réalisant des sondages électromagnétiques TDEM
(Boucher et al., 2009). Un autre projet, près du Lac Tchad (coord. P Genthon)
m’offre la possibilité de coupler le TDEM à la RMP. Avec Kostas Chalikakis
(post-doc LTHE), nous réalisons la première prospection des aquifères de la vallée
de la rivière Komadougou. Pierre Genthon (HSM) me propose de co-encadrer la
thèse de Abdou Moumouni, hydrogéologue nigérien, commencée en 2009.
c) Le programme Ghyraf (Gravimétrie et Hydrologie en Afrique de l’Ouest,
coordination J. Hinderer, EOST Strasbourg) m’offre la possibilité de coupler les
méthodes RMP et électriques sur des sites expérimentaux de l’ORE Amma-Catch,
notamment au Bénin (Descloitres et al., 2010, soumis).
d) La capacité de la RMP à détecter de l’eau liquide en fait une méthode idéale pour
le projet d’étude de l’eau liquide dans les glaciers, monté en collaboration avec
Christian Vincent du LGGE (Descloitres et al., 2010, rapport de mission LTHE).
3. Synthèse des publications
La liste complète de mes publications est retranscrite en annexe 1. J’en donne la
synthèse sous forme d’histogramme sur le tableau 2 ci-dessous.
Tableau 2. Synthèse des publications depuis 1992 (mise à jour : 21/02/2010)
12
4. Encadrements et enseignements
4. 1 Direction de thèses
Clément, R. Caractérisation de l’infiltration par imagerie et suivi temporel géophysique.
Méthodes électriques, électromagnétiques et RMP. Thèse sur bourse Ministère de la
Recherche, Directeur de thèse : J. P. Laurent ; LTHE, co-directeur de thèse : M. Descloitres,
LTHE. 2007-2010.
Moumouni, A. Hydrogéophysique des invasions salées de l’aquifère de Bosso, Est Niger.
Caractérisation, modélisation et possibilités d’exploitation. Directeur de thèse : Pierre
Genthon, DR IRD, HSM (partie hydrogéologie) ; co-directeur de thèse : M. Descloitres,
LTHE (partie géophysique). Thèse préparée à l’Université de Niamey, en cours 2009-2012.
Toé, G. 2004. Apport de nouvelles méthodes géophysiques à la connaissance des aquifères de
socle. Tomographie électrique, électromagnétisme fréquentiel, sondages par résonance
magnétique des protons. Applications au Burkina Faso. Thèse de doctorat de l’Université
de Paris 6, soutenue le 11 juin 2004. Directeurs de thèse : Andrieux, P., Albouy, Y.,
Legchenko, A. Co-encadrant au Burkina Faso pendant 3x6 mois, membre du jury.
4.2 Tableau des encadrements d’étudiants de 3ème cycle, 1992-2009
Depuis 1992, j’ai encadré 24 étudiants de 3ème cycle, surtout de 5ème année universitaire,
mais aussi lors de thèses, pour lesquelles j’ai apporté mon expertise sur certains outils
géophysiques, l’étudiant mettant ensuite en œuvre lui-même la méthode sur le terrain. Je
recense dans le tableau 3 ci-dessous le détail de ces encadrements, en précisant les sujets traités
ainsi que les co-publications issues de leurs travaux (y compris pour les 3 étudiants en thèse
cités plus haut).
13
THESES
PARTENAIRES
Post
Docs
THESES FRANCE
Nom de
l’étudiant
Année/
durée
Responsabilité
Titre du mémoire
Co-publications
CLEMENT
Rémi
2007-2010
LTHE
Co-directeur
Caractérisation de l’infiltration par imagerie et suivi
temporel géophysique. Méthodes électriques,
électromagnétiques et RMP
* 2 articles: C. R. Geosciences (1), Waste M.(1),
* 2 articles soumis (Near Surface Geophysics, Waste
management), 4 conférences : EAGE (1), EGU (1), Geofcan
(1), MRS2009 (1)
TOE
Ghislain
2002-2004
Paris 6
Co-encadrant
Apport de nouvelles méthodes géophysiques à la
connaissance des aquifères de socle au Burkina Faso
* 1 conférence: EAGE (1)
MASSUEL
Sylvain
2002 HSM
Participation
* 1 article : Catena (1),
* 3 conférences : MRS 2009 (2), Geofcan (1)
SCHMUTZ
Myriam
1998-1999
Univ.
Strasbourg
Participation
Évolution récente de la ressource en eau consécutive aux
changements climatiques et environnementaux du sudouest du Niger
Apport des méthodes géophysiques à la connaisance des
glissements-coulées.
RAMIREZ
Edson
1998 Paris 6
Participation
* 3 articles : Journal of Glaciology (1), Pangea (1), Houille
Blanche (1)
COURTEAUD
Michel
Participation
BOUCHER
Marie
1994-95
Univ.
La
Réunion
2008-2009
HSM
Influence de la variabilité climatique sur un glacier de la
Cordillère Royale de Bolivie : le Glacier de Chacaltaya
(16°S).
Etude des structures géologiques et hydrogéologiques du
massif de la Fournaise par la méthode AMT
Participation
Hydrogéophysique
CHALIKAKIS
Konstantinos
MOUMOUNI
Abdou Moussa
2008-2009
LTHE
2009-2012
U. Niamey
Participation
Hydrogéophysique
* 1 article en cours de rédaction
Co-directeur
Les nappes salées de l’aquifère de la Komadougou, Est
Niger
* 1 article en cours de rédaction
PARATE
Harshad
2004-2005
IISC Inde
Co-encadrant
Modélisation des flux d’eau en zone non saturée
* 1 article en review à Current Science (Inde)
CHAUDURY
Abhijit
2004-2005
IISC Inde
Participation
Modélisation stochastique d’un aquifère de socle
* 1 article en review à Mathematical Geology
* 2 articles : Revue Géotechnique (1), Surveys in Geophysics
(1), * 1 conférence : EGS (1)
* 4 articles : Geophysics (1), Groundwater (1), Comptes
rendus Geosciences (1), Water Resources Research (1)
*4 conférences : IAH (2), EEG (1), EEGS (1)
* 1 article: Comptes Rendus Geoscience (1),
* 2 conférences : MRS2009 (1), AMMA (1)
Tableau 3. 1ère partie Etudiants encadrés en thèse
14
5ème année DEA / DESS/ M2P / Ingénieurs
5ème année
universitaire :
DEA Ingénieurs,
Partenaires de
l’IRD
Volont.
Internat
SIMONOVICI
Patrick Denis
2009, 5 mois
INPG
Responsable
Application de la tomographie électrique au suivi
infiltrométrique
.
Géophysique appliquée au suivi des déchets. Chatuzange.
PARRA
Johan
2007 6 mois
M2P
Coresponsable
FLEURY
Simon
2005, 6 mois
EOST
Responsable
Determination of the weathered thickness at Moole Hole
and Maddur watersheds using 2D electrical imaging
SIMONATO
Nelly
2003 2 mois
DESS
Participation
Prospection géophysique du bassin versant de Moole
Hole, Inde
WUBDA
Maxime
2003 DESS
Responsable
Prospections géophysiques sur le bassin versant d’ARA,
Nord Bénin
* 1 article soumis (Near Surface Geophysics)
* 2 conférences : EAGE (1), AMMA (1)
BECK
Matthieu
2001-2002
10
mois
Lausanne
2001-2002
10
mois
Lausanne
1996 DESS
Responsable
Diagraphies électriques pour l’optimisation de
l’hydrofracturation au Burkina Faso
* 1 conférence : Geofcan (1)
Responsable
Diagraphies électriques pour l’optimisation de
l’hydrofracturation au Burkina Faso
Coresponsable
LAMY
Violaine
1995 DESS
4 mois
CoResponsable
Etude hydrogéologique et géophysique des aquifères de
la zone du Ngalenka, périmètre irrigué de la vallée du
fleuve Sénégal
Application du radar géologique à l’étude des formations
superficielles en régions sahéliennes et méditerranéennes
FORGET
Francis
TCHANI
Joseph
1994 DESS
4 mois
1995-96,
Univ. Dakar
CoResponsable
Responsable
GOMIS
Raymond
1995-96
Univ. Dakar
Responsable
DIOUF
Same
1995-96
Univ. Dakar
CoResponsable
KOUSSOUBE
Youssouf
1992, 2 mois
Dakar
Coresponsable
BOST
Adelphe
2006-2007
Responsable
* 1 article : Geochimica et Cosmochimica Acta (1)
* 1 conférence : Goldschmitd Conférence (1)
GIRARDET
Denis
ZANOLIN
Anne
Utilisation du VLF pour la reconnaissance de la
structure du volcan Fogo (Cap-Vert)
L'aquifère des sables quaternaires au Nord de la
presqu'île du Cap-Vert (Sénégal): Analyse d'un cas
d'invasion saline.
L’aquifère des sables quaternaires de la presqu’île du
Cap-Vert (Sénégal) : Morphologie déduite des données
hydrogéologiques et géophysiques
Application de la géophysique (électrique et sismique)
pour l’étude du réservoir de l'aquifère du littoral Nord
Sénégalais
Etude géophysique du Nord Sénégal
Geophysical, geological and cartographic survey at the
Moole Hole and Maddur watersheds, South India
* 3 conférences : EAGE (1), Goldschmidt (1), MRS2007 (1)
Tableau 3. 2ème partie : Etudiants encadrés en 5ème année universitaire.
15
4.3. Enseignements
Impliqué depuis 1998 dans quelques formations professionnelles, principalement à
destination de nos partenaires africains ou indiens, je participe depuis mon arrivée au LTHE à
des sessions de cours sur l’hydrogéophysique, avec mes collègues S. Garambois et A.
Legchenko. Le but est de donner une formation pratique adaptée à des ingénieurs, leur
permettant de comprendre les techniques récentes de l’hydrogéophysique et d’en mesurer les
avantages et les limites, dans le cadre de projets environnementaux intégrant plusieurs
disciplines.
•
•
•
•
Co-responsable du module « Hydrogéophysique» du Master2P « Eaux Souterraines »
de l’UFR « OSUG » à l’Université J. Fourrier de Grenoble. Cours
« Hydrogéophysique, méthodes électriques », 25 heures de cours –TD- TP terrain).
Cours de géophysique de 3ème année du cycle Universitaire « Polytech » 2008 et 2009
à l’Université J. Fourrier de Grenoble (15 heures de cours –TD). « Géophysique
appliquée aux problèmes environnementaux »
Formation à l’imagerie par tomographie de résistivité, 2007, Institut Technologique
du Karnataka, Suratkhal, Inde, 15 étudiants niveau 3ème cycle, 2 jours de cours et TD.
Séminaire de cours « Géophysique Appliquée en Prospection Minière », 1999, IRD,
Guinée Conakry, 5 jours.
5. Participation à des projets scientifiques
Depuis 1992, j’ai participé à différents programmes d’expertise et de recherche. Depuis
1998, je suis responsable de volets de recherche au sein de programmes principalement
nationaux. Ces programmes sont décrits dans le tableau 4 ci-dessous.
Nom du programme
et thématique
Année(s)
Origine des
financements
Montant
géré
Volets de recherche en tant que responsable,
publications et conférences issues de ces actions
(euros)
Interreg
Risques glaciaires
20102011
Projet Européen
15000
De l’eau liquide dans le glacier
de Tête Rousse ?
2009
« TUNES » UJF
France
12000
20082010
EC2CO
France
5000
20092010
ADEME
France
13000
« Bioréacteur »
Optimisation de la gestion des
déchets
20072009
ANR
France
10 000
Ghyraf : Gravimétrie et
Hydrologie en Afrique.
20082010
ANR
Niger/Bénin
5 000
Ondine : Impact des
changements d’usages des
terres sur la
genèse des crues
Paraphyme : gestion des
déchets anciens
Resp : Christian Vincent, LGGE,
Mesure de la quantité d’eau liquide dans un glacier
par RMP.
Resp : Marc Descloitres,
Mesure de la quantité d’eau liquide dans un glacier
par RMP.
Coord. Olivier Ribolzi, LMTG Toulouse
Suivi temporel d’une infiltration provoquée en
tomographie électrique
Publication(s) : 2 ; conférence(s) : 2
Coord : Jean Pierre Gourc,
Mesures TDEM et RMP sur les déchets anciens
Coord. :Jean Pierre GOURC, LTHE, Grenoble
Suivi temporel de la résistivité des déchets par
sondage TDEM
Publication(s): 2; conférence(s) : 3
Coord : Jacques Hinderer, EOPGS, Strasbourg
Caractérisation géophysique des sites des
gravimètres : Lac Tchad, Bénin
Publication(s) : en cours ; conférence(s) : 1
16
IRD Programme
Structurant
Pilote (PSP)
Niger
Lac Tchad : Ressources en eau
et impacts environnementaux
20082010
Fonctionnement hydrologique
des bassins expérimentaux du
Bénin
2003
puis
20062008
AMMA
Bénin
15 000
L’aquifère du continental
terminal au Niger : limites
géométriques
2006
AMMA
Niger
10 000
Fonctionnement
biogéochimique des bassins de
la rivière Kabini, Inde du Sud.
20032006
ECCO et
EC2CO
Inde
15 000
Les aquifères granitiques de la
région d’Hyderabad, Inde
20032005
CEFIPRA
Inde
-
Relation entre sols et
hydrologie en Afrique subsaharienne
20002003
PNSE
Burkina Faso
30 000
Neige et Glaciers tropicaux
1998
Programme
NGT IRD
Bolivie
Fonctionnement hydrologique
de l’aquifère de l’altiplano
Bolivien
1998
PNRH
Bolivie
15 000
Etude du glissement coulée de
Super Sauze
19981999
PNRN
France
3000
Forage d’eau en milieu
volcanique : apports des
diagraphies géophysiques
1999
Conseil Général
de l’île de La
Réunion
-
19941997
Conseil Général
de l’île de la
Réunion
-
14 000
5000
« HydroFournaise »
Les aquifères du Piton de la
Fournaise (La Réunion)
Les aquifères du volcan Fogo
(Archipel du Capt-Vert)
Indurations des mines de
phosphate de Taïba
1995
1992
Mission de
coopération
Française
Archipel duCapVert
Mines de Taïba
Sénégal
7000
5000
Coord. Pierre Genthon, HSM, Montpellier
Volet hydrogéophysique : caractérisation des
aquifères de la Komadougou
Coord. Marc Descloitres
Apport de l’hydrogéophysique à l’hydrologie des
bassins versants expérimentaux du programme
AMMA
Coord. Marc Descloitres. Guillaume Favreau
Caractérisation électromagnétique TDEM des sites
de mesures RMP
Coord. Jean Jacques Braun, LMTG, Toulouse
Apport de l’hydrogéophysique à l’hydrologie des
bassins versants expérimentaux du programme ORE
« Bassins versant expérimentaux tropicaux »
Coord. Jean Michel Baltassat , BRGM
Apport du TDEM à la reconnaissance des altérites
de granite
Publication(s) : - ; conférence(s) : 1
Coord. Henri Robain et Olivier Ribolzi
Responsable des suivis temporels géophysique au
Burkina Faso, gestion crédits de l’Unité IRD
« Geovast » pendant 3ans,.
Coord. Pierre Ribstein et Bernard Francou
Apport du radar géologique à la détermination de
l’épaisseur des glaciers.
Coord. Anne Coudrain
Paramétrisation des terrains aquifères de
l’Altiplano Bolivien par sondage TDEM
Coord. Olivier Maquaire, Strasbourg
Sondages TDEM sur les glissements, thèse de M.
Schmutz
Coord. Pierre Andrieux et Marc Descloitres
(Paris 6)
Apport des diagraphies électriques, montage du
programme.
Publication(s) : - ; conférence(s) : Coord. Bernard Robineau et Jean Coudray (Univ
La Réunion)
Apport des sondages TDEM et des cartographies
VLF à la détection de l’eau douce en milieu
volcanique insulaire.
Publication(s) 4 ; conférence(s) : 3
Apport des sondages TDEM et influence des
conductivités complexes (thèse)
3 méthodes de détection des indurations latéritiques
Publication(s) : - ; conférence(s) : -
Tableau 4. Synthèse de ma participation à des programmes de recherche.
17
TRAVAUX DE RECHERCHE
Introduction
Certains pays du Sud sont confrontés à la rareté de leurs ressources en eaux de surface,
particulièrement en zone semi-aride. D’autres, favorisés par une pluviométrie plus grande,
voient pourtant le débit de leurs cours d’eau diminuer fortement depuis les dernières
décennies (Descroix, 2009). Alors, pour l’alimentation en eau des populations rurales ou
urbaines, ou pour des besoins d’irrigation, ils se tournent vers l’exploitation des eaux
souterraines par puits ou par forages. La gestion durable de ces aquifères nécessite la
quantification des ressources, la compréhension de leur renouvellement et de leur
vulnérabilité et ce, dans un contexte d’urbanisation rapide et de risque de pollution agricole ou
industrielle. Certains de ces pays du Sud sont aussi confrontés à des problèmes de dégradation
ou d’appauvrissement des sols par l’érosion, la déforestation, le changement d’usage des
terres. Ces phénomènes peuvent être amplifiés du fait des changements climatiques, et des
pressions anthropiques ou agro-pastorales accrues.
Dans ce contexte, mes activités de recherches ont pour objectif d’apporter des
connaissances sur les aquifères et sur les processus de transferts d’eau dans le sous-sol au
moyen des outils géophysiques. Ces activités se situent dans différents contextes géologiques
et climatiques, et sont déclinées suivant trois axes thématiques :
i)
La spatialisation des aquifères. En effet, il faut localiser les aquifères à partir de la
surface et estimer du mieux possible la quantité et la qualité de l’eau en présence,
avant de réaliser des forages.
ii)
L’étude de la recharge des aquifères. Pour modéliser les impacts que pourraient
avoir dans le futur les changements climatiques et anthropiques sur la ressource en
eau souterraine, il faut construire des modèles conceptuels les plus précis possible.
Pour cela, la question de la localisation des recharges doit être abordée. Mes
recherches concernent surtout l’échelle locale (l’hectare, le petit bassin versant).
iii)
La spatialisation de processus élémentaires de transfert de l’eau se produisant dans
les premiers décimètres des sols, en zone non saturée. Ces processus sont
généralement observés par des dispositifs localisés et la question de leur
représentativité spatiale se pose.
Un quatrième axe, méthodologique, permet l’adaptation de certains outils géophysiques
aux spécificités des aquifères et des processus hydrologiques étudiés.
Ces recherches ont été menées sur les chantiers de l’IRD dans les pays du sud, en
collaboration avec des hydrologues, hydrogéologues, hydrogéochimistes, pédologues ou
modélisateurs des zones vadoses et saturées. Ces différentes disciplines ont besoin de la
géophysique de sub-surface car elles ont une difficulté en commun : passer d’une observation
localisée (l’échantillon, la parcelle, le forage) à la compréhension du fonctionnement d’une
unité plus grande (aquifère, versant, bassin versant, couverture de sols, région), jugée comme
pertinente pour les processus étudiés et les modèles qui en découlent. En utilisant la
géophysique, on bénéficie de ses facultés à spatialiser et quantifier d’une manière nondestructive certaines propriétés physiques du sous-sol.
18
Ces études abordent nécessairement différentes échelles d'espace et de temps. Echelles
spatiales, tout d’abord, puisque l’extension latérale des aquifères et les processus de recharge
peuvent être très locaux (nappes perchées temporaires de quelques hectares, recharge
indirecte localisée dans les axes drainants), ou régionaux (aquifère des grands bassins
sédimentaires de plusieurs centaines de kilomètres carrés, recharges directes sur l’ensemble
du paysage), et concerne des profondeurs allant de la surface à plusieurs dizaines de mètres de
profondeur. Echelles temporelles ensuite, car pour comprendre les processus de sub-surface
en zone non-saturée ou les flux dans les aquifères il faut s’intéresser à des échelles allant de
l’évènement pluvieux de quelques heures à l’année hydrologique, voire interannuelles.
Mon mémoire s’articulera autour de 4 chapitres principaux:
•
•
•
•
Le premier décrira les questions scientifiques que j’ai abordées, en s’appuyant sur la
description des principaux modèles conceptuels relatifs aux aquifères et aux processus.
Le second présentera la démarche que j’ai suivie, décrira succinctement les principales
méthodes géophysiques employées, et montrera leur mise en œuvre sur trois chantiers
majeurs.
Le troisième présentera les principaux résultats scientifiques obtenus pour les 3 axes
thématiques (la spatialisation des aquifères, leur recharge, les processus élémentaires)
et pour l’axe méthodologique. Pour cela je m’appuierai sur les publications, jointes en
annexe 2.
Le quatrième chapitre présentera les perspectives de recherche que je compte
développer dans les prochaines années.
1. Questions scientifiques abordées
Ce paragraphe décrit les principales questions scientifiques concernant la spatialisation
des aquifères que j’ai étudiés, leurs modèles de recharge, et certains processus élémentaires en
zone non-saturée. Pour les illustrer, je m’appuie sur des modèles conceptuels classiques de
l’Afrique de l’Ouest. Ces questions restent similaires pour les autres chantiers que j’ai étudiés,
notamment en Inde (aquifères de socle de la région de Mysore en climat soudanien) et en
Bolivie (aquifère sédimentaire de l’Altiplano Bolivien en climat semi-aride). Ils suivent en
général les mêmes schémas conceptuels. Leurs particularités sont décrites dans les
publications.
1.1. Les aquifères de socle
Les aquifères de socle recouvrent une surface non négligeable en Afrique de l’Ouest
(Figure 1). Les roches qui les contiennent regroupent l’ensemble des grandes familles
géologiques : granites, gneiss, roches vertes, roches métamorphiques, d’âge précambrien en
général. Sur cette figure, j’ai placé les sites que j’ai étudiés.
19
Figure 1. Répartition des zones aquifères de socle (couleur brune) et des bassins sédimentaires en Afrique de
l’Ouest. Les cercles indiquent les sites étudiés. Pour les aquifères de socle, il s’agit des sites de Katchari au
Burkina Faso (1), des environs de Ouagadougou, Burkina Faso (2), des environs de Djougou, Bénin (3). Pour
les aquifères sédimentaires, il s’agit des sites des environs de Niamey, Niger (4), de la vallée de la rivière
Komadougou Yobé, Lac Tchad, Niger (5), des zones côtières près de Dakar, Sénégal (6). La ligne pointillée
sépare approximativement les zones climatiques semi-aride et sahéliennes (pluies < 750 mm/an) au nord, des
zones soudanienne et guinéenne, au sud (pluies > 750 mm/an). Fond de carte d’après MacDonald and Davies
(2000).
a) Distribution verticale des propriétés hydrauliques
Les ressources en eau en zone de socle sont modestes, en particulier en contexte aride
ou semi-aride (Lachassagne et Wyns, doc. BRGM, 2005). Les schémas de distribution
verticale des propriétés hydrauliques les plus simples représentent les aquifères de socle selon
des colonnes de sol comme celles de la figure 2. L’altération chimique des roches crée un
profil d’altération (le regolith) surmontant le bedrock (ou protolith). Selon le type de roche, sa
texture (grosseur des grains,), sa structure (foliation par exemple) ou sa fracturation initiale,
divers matériaux d’altération se mettent en place, allant d’une arène (sable grossier) à des
argiles de néoformation. Les auteurs s’accordent à dire que la conductivité hydraulique de
l’aquifère augmente en général avec la profondeur. Généralement plus il y aura d’argile, plus
la conductivité hydraulique diminuera.
Figure 2. Schémas conceptuels simplifiés d’un aquifère de socle. A gauche, variations de la conductivité
hydraulique et de la porosité cinématique avec la profondeur, selon Chilton et Foster (1995). Il ne s’agit pas
d’une distribution immuable, et d’autres scénarios de distribution peuvent se rencontrer : Jones propose en effet
un accroissement de la porosité de drainage (specific yield) avec la profondeur, à droite. D’après Jones (1985).
20
Ces auteurs soulignent aussi le rôle important que jouent les fractures situées au sein
du protolith, qui conduisent le flux plus facilement. La conductivité hydraulique et l’épaisseur
vont fixer le débit potentiel pour l’exploitation de l’aquifère. La porosité cinématique
(effective porosity), qui aurait tendance à diminuer avec la profondeur selon certains auteurs,
va constituer la partie mobilisable de l’eau par pompage. Le stock d’eau disponible est
quantifié par la porosité de drainage (specific yield), généralement inférieure à la porosité
cinématique. Selon d’autres auteurs, cette dernière pourrait au contraire augmenter avec la
profondeur. C’est le réservoir de l’aquifère.
On voit que la distribution verticale des propriétés hydriques des regoliths n’est pas
forcément généralisable et qu’il convient de les préciser si l’on veut en apprendre plus sur le
fonctionnement de l’aquifère sous le site étudié. La connaissance du profil d’altération, dont
l’épaisseur peut atteindre plusieurs dizaines de mètres en climat tropical, est fondamentale
pour l’exploitation de l’aquifère : si le regolith est argileux, la porosité de drainage sera faible,
sa conductivité hydraulique aussi, et les réserves d’eau mobilisable par pompage par
conséquent limitées. Si la partie inférieure du profil est peu épaisse, la transmissivité de
l’aquifère (produit de la conductivité hydraulique par l’épaisseur) sera faible et le forage peu
productif.
En géophysique, la caractérisation du profil d’altération par des mesures de surface,
avant le forage, constitue un véritable défi. En général, les méthodes géophysiques voient leur
résolution diminuer avec la profondeur. Les argiles, électriquement conductrices, jouent le
rôle d’un écran, car elles confinent les circulations des courants électriques créées par nos
appareils. Enfin, les fractures du protolith sont profondes, et constituent des objets très
difficilement détectables avec précision à partir de la surface. Seules les zones de fractures
suffisamment larges pour avoir favorisé l’approfondissement de l’altération argileuse (zone de
faille ou de cisaillement par exemple) peuvent être détectées depuis la surface.
b) Aquifères de socle sous climat sahélien
Modèle conceptuel
La situation décrite sur la figure 3 correspond aux zones sahéliennes (400 à 700 mm
de pluie annuelle). Dans ces régions, l’évaporation potentielle est très importante (supérieure
à 2000 mm/an). L’absence quasi-totale de végétation, et d’arbres en particulier, rend le terme
de transpiration très faible. En zone de socle, le proche sous-sol est généralement plus
argileux qu’en zone sédimentaire. Cette situation défavorise l’infiltration. De plus, la
formation de croûtes de dessiccation en surface favorise les ruissellements lors de la saison
des pluies. L’eau qui arrive à s’infiltrer avant peut être reprise par évaporation directe.
Descroix et al (2009) montrent que ce sont les zones de socle et les ruissellements accrus
depuis les dernières décennies qui sont responsables de la modification du régime du fleuve
Niger, rendant le pic de crue à Niamey plus court et plus intense. Les zones de socle de la
zone sahélienne forment ainsi une vaste région d’exoréisme, à opposer aux zones
sédimentaires, endoréiques.
21
Figure 3. Schéma conceptuel des systèmes aquifères de socle en zone semi aride. Les rectangles délimités en
pointillé correspondent aux zones clefs que j’ai étudiées au Burkina Faso.
Questions scientifiques abordées
En zone de socle, on considère qu’il n’y a pas de recharge directe des aquifères par
infiltration généralisée sur l’ensemble du paysage. On admet généralement que les aquifères
se rechargent de façon indirecte par les axes drainants et les mares de bas fond, mais ce
schéma est cependant mal connu. Avec mes collègues hydrologues, nous avons abordé deux
questions du cycle de l’eau :
¾ Les ravines de versant participent-elles à la recharge des aquifères? En effet, si ce rôle
a pu être mis en évidence en contexte sédimentaire à la même latitude au Niger
(Peugeot et al., 1997) en raison de l’épaisseur importante des sols sableux, aucune
étude n’a été menée sur les versants en zone de socle.
¾ Comment se rechargent les aquifères dans les axes drainants (bas fond) ? Quelle est la
géométrie de la recharge et sa dynamique temporelle?
A une échelle plus superficielle, on admet que l’eau qui ne ruisselle pas se stocke
préférentiellement dans des couvertures peu épaisses de sable éolien. Nous nous sommes
intéressés à évaluer les capacités de stockage hydrique de ces sols sableux, question que je
détaillerai plus loin.
c) Aquifères de socle sous climat soudanien à guinéen
Modèle conceptuel
La figure 4 décrit les processus dans une situation climatique plus humide (climat
soudanien à guinéen, c'est-à-dire avec des pluviométries de 750 mm/an à plus de 1200
mm/an). Deux différences fondamentales avec les zones plus arides apparaissent : a) les
nappes peuvent, suivant la saison, alimenter les cours d’eau, et b) la végétation joue un rôle
majeur dans le cycle de l’eau. La transpiration est augmentée, les ruissellements diminués, et
la recharge peut être directe (elle se produit sur l’ensemble du paysage). La recharge indirecte
par les cours d’eau de bas fond peut se rencontrer, particulièrement dans les zones à saisons
contrastées lorsque les drains coulent de façon intermittente (situation rencontrée en Inde et au
Bénin sur les bassins versants des Observatoires Régionaux en Environnement (ORE)
22
« BVET » et « AMMA-Catch ». Le battement de la nappe peut être important (plusieurs
mètres) selon la saison en régime de mousson, comme c’est le cas au Nord Bénin. En zone
forestière, les variations interannuelles du niveau des nappes peuvent être déphasées par
rapport à l’année en cours, par la transpiration de la végétation (Ruiz et al, 2009).
Figure 4. Schéma conceptuel des systèmes aquifères de socle en zone soudano-guinéenne. Les rectangles
délimités en pointillé correspondent aux zones clefs que j’ai étudiées en Inde et au Bénin.
Le cycle de l’eau semble plus complexe dans ce contexte climatique : en effet, la
nappe participe aux écoulements des rivières de façon sporadique, du moins sur les sites
étudiés. Les déconvolutions des hydrogrammes de crues sont difficiles à réaliser, car plusieurs
compartiments apportent leur contribution de façon déphasée. La reprise par la transpiration
peut être considérable et affecter des épaisseurs importantes de zone non saturée. Mes
collègues du LMTG se posent aussi la question du taux d’altération de ces roches sous climat
tropical. Grâce à un couplage entre nos outils, nous avons abordé les questions suivantes :
¾
¾
¾
¾
¾
¾
Quelle est la géométrie de l’infiltration sous les axes drainants ?
Quelle est l’épaisseur du régolith et le modelé du toit du bedrock ?
Peut-on mieux quantifier les taux d’altération tropicale des roches cristallines ?
Peut-on mettre en évidence la recharge directe à l’échelle d’un versant ?
Quelle est la contribution de la nappe à l’évapotranspiration ?
Quelle est la porosité de drainage du regolith ? dépend-t-elle du type de roche mère ?
1.2. Les aquifères sédimentaires en zone semi-aride
Les aquifères sédimentaires de l’Afrique de l’Ouest sont situés à plus haute latitude
que les aquifères de socle (figure 1) au sein de formations sableuses, gréseuses, ou alluviales,
d’âges tertiaire et quaternaire. Ces formations couvrent une surface importante des territoires
du Sénégal, du Mali, de la Mauritanie et du Niger.
23
Modèle conceptuel
Les aquifères sédimentaires apparaissent plus simples, car leur géométrie peut être
considérée généralement comme tabulaire. La nappe est dite « libre », et la géométrie de son
toit est plus régulière. Sur la figure 5, je prends l’exemple de l’aquifère du continental
terminal de la région de Niamey au Niger. Mais d’autres aquifères de ce type ont été étudiés
(aquifère de l’altiplano bolivien, aquifère de la rivière Komadougou près du Lac Tchad au
Niger). A l’échelle régionale, les études de Leduc et al (2001) font apparaître des remontées
des niveaux des nappes suite à la période de sécheresse au Sahel. Ce paradoxe est expliqué
par les changements intervenus en surface : augmentation du ruissellement, concentration des
eaux dans les mares des points bas, dont le nombre augmente, et favorise ainsi un
accroissement de la recharge indirecte. L’évaporation directe des sols est aussi favorisée par
la disparition de la végétation, avec une diminution de la transpiration.
Figure 5. Illustration de la géométrie et du fonctionnement des systèmes aquifères sédimentaires en zone semiaride : exemple de l’aquifère du continental terminal à l’est de Niamey, Niger. Les rectangles délimités en
pointillé correspondent aux problématiques que j’ai abordées. Figure modifiée d‘après Massuel et al,
conférence Geofcan (2003).
Les variations latérales éventuelles à l’échelle locale et régionale peuvent être étudiées
avec un pas d’échantillonnage plus lâche. Pour le géophysicien, cette situation est favorable,
car certaines méthodes se prêtent mieux à une approximation 1D.
Les études entreprises en géophysique à l’échelle régionale concernent des questions assez
classiques, mais qui peuvent être revisitées grâce au développement des outils géophysiques
dans la dernière décennie :
24
¾ Quelle est la profondeur de la nappe ?
¾ Le mur de l’aquifère (généralement argileux), présente-t-il un modelé régulier ?
¾ Au Niger, pour mieux comprendre et quantifier la hausse des nappes, quelles sont les
variations régionales de la porosité de drainage de l’aquifère principal ?
¾ En Bolivie et au Lac Tchad, quelle est la répartition des formations sédimentaires
argileuses pouvant faire obstacle aux flux souterrains ?
A l’échelle locale du bassin versant, les questions abordées concernent l’identification
des processus du cycle de l’eau. S’il est communément admis que les mares temporaires
constituent la zone de recharge principale, certains auteurs (Peugeot, 1997) suggèrent aussi le
rôle possible des versants qui, à la faveur de zones d’épandage sableux, pourraient favoriser
une recharge indirecte. Il est donc important de savoir si ces versants contribuent, ou non, à la
recharge. Avec mes collègues hydrogéologues d’Hydrosciences Montpellier, nous nous
sommes intéressés à cette question sur le bassin versant expérimental de Wankama au Niger
(ORE « AMMA-Catch »).
1.3 . Les processus élémentaires à l’échelle de la parcelle ou de la séquence
de sols
Modèle conceptuel
La connaissance du cycle de l’eau commence souvent par documenter les processus
localement. Ces processus élémentaires concernent la zone non-saturée (ZNS) à des petites
échelles (parcelles, séquence de sols). Je prends ici l’exemple des micro-dunes sableuses du
nord du Burkina Faso qui sont le support de la végétation herbacée naturelle (pâturage), ou
des cultures ensemencées par l’homme (Ribolzi et al. 2006). Comprendre le cycle de l’eau au
sein de ces objets fragiles, éléments incontournables de l’hydrologie sahélienne, a constitué
un de nos objectifs au Burkina Faso. Sur la figure 6, nous voyons que ces micro-dunes sont
stratifiées par l’effet des dépôts éoliens successifs. Elles présentent un bord abrupt, érodé par
le vent, et un coté abrité où peut croître l’herbe en saison des pluies.
Figure 6. Les micro-dunes en zone
semi-aride. A : paysage en saison
sèche au nord du Burkina Faso : les
micro-dunes sont entourées d’un
pointillé blanc. B : le foisonnement
de l’herbe sur les micro-dunes en
saison des pluies. C : coupe d’une
micro-dune
montrant
sa
stratification, l’action érosive du
vent, et les premières herbes,
poussant sur le coté sous le vent. D :
Schéma conceptuel supposé des
transferts d’eau au sein des microdunes sous une pluie : le coté au
vent produirait des écoulements
d’eau nouvelle, alors que le coté
sous le vent produirait des eaux
mélangées. d‘après Descloitres et
al, conférence EAGE, (2006).
25
Pour l’hydrologue, les questions qui se posent concernent un micro cycle de l’eau, car
il s’agit par exemple de déterminer les fractions d’eau libre et d’eau liée dans ce milieu
poreux. S’il s’agit de déconvoluer un hydrogramme de crue, il faut faire la partition, dans le
ruissellement, entre l’eau nouvelle de la pluie et l’eau plus ancienne ayant acquis des
signatures géochimiques différentes. Sur la figure 6D, le schéma conceptuel de
fonctionnement lors d’une pluie implique un rôle important de la micro-stratification. Ce
schéma est cependant hypothétique et nous avons essayé de le préciser. Ces processus
peuvent être appréhendés par l’implantation, dans le proche sous-sol, de capteurs intrusifs.
Cette intrusion n’est pas forcément sans conséquence sur les processus étudiés et, comme
l’étude des aquifères, les mesures ponctuelles nécessitent souvent une spatialisation. Nous
avons tenté d’apporter des réponses par la géophysique aux questions suivantes :
¾ Lors d’une pluie, quelle est l’influence des versants? l’eau pénètre-t-elle
préférentiellement sur le coté herbeux?
¾ La micro-stratification canalise-t-elle l’écoulement ?
¾ Est-il possible de visualiser au sein de la dune les eaux nouvelles des eaux anciennes,
piégées dans la dune lors des pluies précédentes ?
¾ Comment se répartit l’évaporation au sein de la dune et à quelle vitesse ?
D’autres expériences à cette échelle ont été menées, en Inde notamment, sur des
séquences de sols. Elles concernent aussi la spatialisation des transferts d’eau en ZNS, et le
fonctionnement des sols argileux lors des périodes de mousson.
26
2. Méthodologie
2.1. Approche générale
Que ce soit en contexte sahélien ou soudanien, contribuer à répondre, avec la
géophysique, aux questions précédentes nécessite des méthodes qui soient capables :
a) d’estimer les volumes d’eau stockés en géométrie complexe (2D, voire 3D),
b) de spatialiser les propriétés hydriques du regolith (souvent liées à la teneur en argile),
c) de localiser les lieux de recharge, en prospectant de vastes zones pour identifier des
recharges très localisées au sahel, aider à l’implantation des forages de reconnaissance,
d) d’appréhender différentes échelles, de la microdune de quelques mètres carrés au
bassin versant de quelques kilomètres carrés,
e) d’estimer la dynamique des processus autrement qu’en quelques points de mesure, à
différentes échelles temporelles,
f) de fournir des paramètres clefs (géométriques ou propriétés hydriques notamment)
pour contribuer aux modélisations hydrologiques.
Au cours de mes recherches, j’ai mis en œuvre une méthodologie s’appuyant sur :
•
•
•
•
Une description des sites d’étude en réalisant des spatialisations des paramètres
géophysiques, de manière à faciliter l’implantation de méthodes ou de capteurs donnant
des résultats quantitatifs sur les processus étudiés.
Une étude des variations temporelles de résistivité électrique et d’autres paramètres
géophysiques pertinents, à des échelles variables, de la parcelle au versant et de
l’évènement pluvieux à la saison hydrologique complète.
Une utilisation couplée de méthodes géophysiques, avec une préférence marquée pour les
méthodes de résistivité: tomographie de résistivité électrique (ERT), cartographie ou
sondages électromagnétiques et plus récemment, Résonance Magnétique des Protons
(RMP). D’autres paramètres géophysiques ont aussi été considérés plus ponctuellement :
permittivité diélectrique (méthode radar), potentiels électriques (méthode de polarisation
spontanée) notamment.
Une recherche des relations entre paramètres géophysiques et variables d’intérêt pour
l’hydrologue, généralement au cours d’expérience de terrain semi-contrôlées d’échelle
réduite.
L’ensemble de ma démarche est schématisé sur la figure 7.
27
Figure 7. Schéma de la méthodologie suivie au cours de mes recherches. Depuis la conception de l’étude (en
gris), la modélisation numérique (en jaune), les mesures de terrain (en vert), jusqu’à l’interprétation (en bleu).
En général, je m’intéresse dans un premier temps à des mesures obtenues à partir de la
surface. C’est l’approche classique du prospecteur géophysicien, qui relève de considérations
pratiques : ces méthodes sont moins coûteuses que la réalisation de forages, même si les
méthodes de surface subissent une perte de résolution avec la profondeur.
Mais, afin de contrôler les résultats des méthodes de surface, j’ai utilisé dans un deuxième
temps, quand c’était possible, des données obtenues en profondeur, mises en place
spécifiquement pour les besoins des expérimentations (diagraphies en forage par exemple).
Ces vérifications par des données extérieures indépendantes ont des conséquences majeures
dans mes travaux car elles ont générées des travaux méthodologiques sur les inversions
géophysiques, notamment en suivi temporel de résistivité par imagerie en 2 ou 3D. Lorsque
des reconnaissances en forage profond n’étaient pas possibles, j’ai parfois entrepris la
réalisation d’expériences de taille réduite, toujours sur le terrain, afin de faciliter les contrôles
au sein du milieu.
Un autre aspect méthodologique implique une approche classique d’aller-retour entre les
résultats de terrain et modélisations numériques, ces dernières étant utilisées pour a) tester
l’adéquation de la méthode géophysique avec la spatialisation des modèles conceptuels de
fonctionnement des processus hydriques, b) évaluer les équivalences possibles des modèles
géophysiques issus des inversions.
28
La plupart de mes travaux ont été entrepris à l’échelle du terrain, de quelques mètres à
plusieurs kilomètres en surface, et jusqu’à environ 100 mètres de profondeur. En effet, établir
des fonctions de transfert entre paramètres géophysiques et variable hydrologiques qui soient
représentatives de l’échelle des phénomènes (ou au contraire montrer qu’elles sont trop
complexes pour être utiles) réclame des expériences qui ne sont pas forcément reproductibles
en laboratoire. Ma démarche comprend aussi l’acquisition d’informations clefs pour
construire des modèles numériques géophysiques correctement renseignés, comme par
exemple l’évaluation i) de la gamme de variation des paramètres, ii) des anisotropies
éventuelles, iii) des effets de changements d’échelles, et enfin iv) des conditions de bruit
électromagnétique relatives au site de mesure.
2.2. Principales méthodes géophysiques employées
Les principes physiques des méthodes employées, les différentes configurations de
mesure employées sur le terrain, et les procédures d’inversion sont décrits en détail dans les
publications jointes en annexe 2 de ce document. Cependant, je reprends ici quelques
éléments sur les méthodes de résistivité et de Résonance Magnétique des Protons (RMP),
méthodes incontournables de l’hydrogéophysique.
a) Méthodes de résistivité
La résistivité d’un milieu quantifie sa capacité à s’opposer au passage d’un courant
électrique. La résistivité d’un sol ou des formations géologiques dépend de facteurs très
intéressants pour l’hydrogéophysique: porosité, teneur en eau, conductivité électrique de
l’eau, tortuosité des pores, température, présence ou non de minéraux argileux. S’il n’y a pas
d’argile, c’est la loi empirique d’Archie (Archie, 1942) qui décrit le mieux l’influence de ces
différents facteurs. Les gammes de variations de la résistivité sont très étendues. En présence
d’argile, la loi d’Archie n’est généralement pas applicable. En dehors de son intérêt évident
dû à sa large gamme de variation selon la nature et l’état hydrique du sous-sol, l’emploi de la
résistivité en hydrogéophysique est incontournable en raison de quatre avantages décisifs:
• Toutes les échelles d’espace, de 10 cm à plurikilométriques, peuvent être abordées, à la
fois sous forme cartographique ou par la réalisation des tomographies du sous sol, sur une
profondeur qui peut atteindre plusieurs dizaines de mètres,
• On peut prendre en compte, aux échelles citées précédemment, des géométries complexes
1, 2 et 3D,
• On peut adapter le rythme d’acquisition à des échelles de temps variant de la minute à
l’année hydrologique,
• On peut mesurer la résistivité par induction électromagnétique, ce qui est d’un avantage
considérable sur des sols secs, ce qui est souvent le cas en zone semi-aride.
Mais l’emploi de la résistivité présente un inconvénient majeur dû à sa dépendance
aux multiples facteurs cités plus haut. La déconvolution du signal de résistivité est une tâche
complexe étant donné le nombre important de paramètres entrant en jeu (sans même compter
29
l’hétérogénéité naturelle du sous-sol, les anisotropies éventuelles, ou les effets de changement
d’échelle). Cet inconvénient est contourné, du moins partiellement, par deux approches
utilisées dans ma démarche: les suivis temporels de résistivité, et le couplage avec la
Résonance Magnétique de Protons.
En ce qui concerne les suivis temporels de résistivité en zone non saturée, on bénéficie
d’un avantage : entre deux instants, les variations de la résistivité seront au premier degré
uniquement dues à la variation de teneur en eau (si la conductivité électrique de l’eau
d’imbibition ne change pas). De même, en zone saturée, les variations de résistivité dépendent
au premier degré essentiellement des changements de la conductivité de l’eau. En suivi
temporel de résistivité, on s’affranchit des variations spatiales et temporelles de facteurs
restant invariants dans le sol : porosité, tortuosité, teneur en argile par exemple. Les variations
de température restent à prendre en compte dans certains cas. Un suivi temporel de résistivité
nécessite cependant une bonne résolution de ce paramètre : La figure 8 présente les résultats
obtenus par Knight (1991). Dans la gamme de saturation 50 à 100%, les variations de
résistivité restent faibles, particulièrement en phase de dessiccation (variations assez réduites
de l’ordre de 10 à 25%). Le challenge pour le géophysicien est donc de pouvoir mesurer avec
fiabilité de faibles variations de résistivité sur le terrain, de manière à espérer pouvoir les
traduire en variations de saturation correspondantes.
Figure 8. Variations de la résistivité
en fonction de la saturation pour des
échantillons de grès. On note le
comportement hystérétique de la
résistivité
selon
la
phase
d’humectation
(carrés)
et
de
dessiccation (cercles), d’après les
travaux de Knight (1991)
Hubbard et Rubin (2006) précisent que les imageries géophysiques en suivi temporel
diminuent la dépendance des mesures géophysiques aux variations géologiques statiques du
milieu, ainsi qu’aux procédures d’inversion et artefacts associés. Si je suis d’accord avec le
fait que le suivi temporel permet de s’affranchir par exemple des variations de porosité (pour
la résistivité), je suis plus prudent en revanche sur les améliorations que le suivi temporel
apporterait lors des inversions. On verra lors de la synthèse des résultats que l’étude de la
variation de résistivité dans le temps présente des difficultés majeures (Descloitres et al.,
2004, Descloitres et al., 2008). Celles-ci sont essentiellement dues à la non-unicité des
modèles de résistivité et aux processus d’inversion qui sont la plupart du temps conduits avec
des facteurs d’amortissement importants, comme le remarquent aussi Kemna et al. (2004).
30
b) La Résonance Magnétique des Protons (RMP)
La méthode RMP est une méthode géophysique récente et non destructive de détection
des aquifères. Elle se démarque des autres méthodes géophysiques qui analysent des
anomalies de structure ou de paramètres physiques qui ne sont qu’indirectement liés à la
présence d'eau. Le phénomène de résonance magnétique concerne les protons +H de la
molécule d’eau H2O. Lorsqu’on détecte un signal de résonance, il y a de l’eau liquide dans le
sous-sol. Son principe physique est présenté dans de nombreuses publications, Legchenko et
al. (2002a et 2002b) ou Lubczynski et Roy (2003 et 2004). La RMP apporte des
renseignements importants pour les aquifères complexes de socle (Vouillamoz et al, 2004,
Wyns et al., 2004) ou pour les aquifères sédimentaires (Vouillamoz et al, 2008, Chalikakis et
al, 2009). On réalise un sondage RMP en auscultant le terrain selon des profondeurs
croissantes en augmentant l’intensité du courant. Avec l’appareillage existant actuellement,
seules les formations saturées sont étudiées. Deux paramètres sont déduits des acquisitions:
•
L’amplitude initiale du signal, qui est proportionnelle au volume d’eau présent dans le
sous-sol. Un signal d’amplitude importante témoignera d’un volume d’eau liquide important.
Dans un milieu poreux saturé on parle de « teneur en eau RMP ». Seul le signal de l’eau dite
« libre » contenue dans le milieu est mesuré. L’eau « liée » aux parois des grains de la matrice
(liaison par les forces capillaires - si le milieu est désaturé- et eau adsorbée sur les parois des
grains), produisent aussi un signal de résonance, mais il se manifeste avec des temps de
relaxation trop courts pour être mesurés avec l’appareillage RMP de terrain.
•
La forme de la décroissance de l’enveloppe du signal, caractérisée par la constante de
temps T2*, peut être affectée par les propriétés magnétiques des grains. Ces propriétés
magnétiques n’étant pas connues facilement, on préfère analyser la constante de temps de
relaxation longitudinale T1, calculée grâce à l’acquisition de 2 signaux de relaxation appelés
FID 1 et FID2, résultants de l’injection de 2 pulses de courant successifs. T1 est
caractéristique de la taille des pores : plus la valeur de T1est grande, plus les pores sont gros.
Un aquifère constitué de sable très grossier ou de graviers sera ainsi caractérisé par des temps
de décroissance T1 supérieurs à 400 millisecondes. Les formations argileuses ne produisent
pas de signal détectable par l’équipement actuel.
L’utilisation de la RMP dans mes travaux date de 2003, lorsque j’ai participé aux
mesures RMP sur les aquifères de socle au Burkina Faso dans le cadre de la thèse de mon
collègue J.M. Vouillamoz. En 2007, A. Legchenko et moi avons employé cette méthode sur
des aquifères complexes en Inde sur le site de Moole Hole, où la RMP apporte une image
convaincante des propriétés hydriques des aquifères de gneiss et amphibolite altérés. Le
couplage avec les méthodes d’imagerie électrique est particulièrement fertile, comme on le
verra dans la synthèse des résultats (Legchenko et al., 2007). Je réalise aussi sur ce site les
premières expérimentations de suivi temporel RMP (Descloitres et al., 2008). Si je ne
développe pas moi-même la méthode, je m’attache depuis 2006 à l’employer dans des
contextes novateurs qui seront décrits dans la synthèse de mes travaux. Mes travaux
méthodologiques se situent à l’aval des développements de la méthode RMP, dans l’intercomparaison du signal RMP avec les autres méthodes sur des sites choisis.
31
c) Autres méthodes utilisées
D’autres outils utiles aux études hydrogéophysiques ont été employés ponctuellement:
• Le radar géologique, sensible au premier degré à la permittivité diélectrique des milieux
que traverse l’onde radar (et donc à la teneur en eau) a été employé sur les couvertures
d’altération en 1995 (Descloitres et al., 1997, Geofcan), puis pour sonder les glaciers
(Descloitres et al, 1999). Cette méthode s’est révélée également utile pour caractériser le
toit de la nappe phréatique au sein de grandes dunes au Burkina Faso (Savadogo et al,
conférence Geofcan, 2001), pour caractériser des stratifications au sein des micro dunes
(Rejiba et al., conférence Geofcan, 2001), et enfin sur des filons de quartz aurifères au
Sénégal (Lamy, 1995). J’ai finalement délaissé le radar au profit d’autres méthodes car la
profondeur d’investigation était trop souvent limitée à quelques mètres sous la surface en
raison du caractère très argileux des altérites tropicales, dont la faible résistivité (quelques
ohm.m) atténue trop fortement les ondes radar.
• La polarisation spontanée a été testée afin de repérer les lieux de recharge sur les bassins
versants au Burkina Faso et en Inde, avant et après des pluies intenses. Ces tests n’ont pas
produit des résultats interprétables.
• Les méthodes de diagraphies nucléaires (neutron-neutron, gamma densimétrie) me
permettent une comparaison des résultats, lorsqu’il est possible de forer des tubes d’accès
dans le sol.
2.3.
Exemples de mise en œuvre des méthodes géophysiques
La majorité des études géophysiques que j’ai entreprises ont été réalisées sur trois
bassins versants expérimentaux de l’IRD. Tous se situent en zone de socle : le bassin versant
de Katchari, en zone sahélienne du Burkina Faso, celui de Moole Hole au Sud de l’Inde, et
celui d’Ara au Nord du Bénin.
D’une manière générale, j’ai privilégié les sites sur lesquels des validations des
résultats géophysiques pouvaient être faites de façon indépendante avec d’autres méthodes.
Sur ces bassins versants expérimentaux, la démarche d’implantation des méthodes
géophysiques suit en général la logique suivante:
a) Réalisation d’une cartographie de résistivité à maille serrée, de quelques coupes de
résistivité, d’une reconnaissance géologique détaillée, conduisant à l’identification des
zones clefs où se posent les questions hydrologiques. Ces prospections sont réalisées
généralement avant la saison des pluies et constituent un « état zéro » géophysique.
b) Sur les zones clefs, mise en place des dispositifs de suivi temporel (en général, des
tomographies de résistivité).
c) Suivi temporel de résistivité à pas de temps approprié sur les zones clefs.
d) Réalisation d’expériences semi–contrôlées in-situ, dédiées à l’établissement des
fonctions de transferts entre géophysique et hydrologie, ou au contrôle des
interprétations géophysiques en profondeur (forages et diagraphies)
e) Mise en œuvre d’autres méthodes géophysiques complémentaires, RMP notamment.
32
a) Site de Katchari, Burkina Faso
Le site de Katchari au nord du Burkina Faso (figure 9), est constitué de bassins
versants emboîtés, de quelques hectares à plusieurs kilomètres carrés, permettant d’étudier les
changements d’échelle des processus. Les états de surface sont représentatifs de toute la zone
nord-sahélienne.
Figure 9 .Bassin versant de Katchari au Burkina Faso. Les fonds de carte correspondent aux valeurs des
résistivités des sols et des altérites de socle granitique à différentes échelles de profondeur. Quatre opérations
majeures sont représentées : 1 et 2 : La recherche des infiltrations sous une ravine de versant caractéristique,
avec une approche cartographique en suivi temporel de la résistivité apparente (Descloitres et al, 2003), et en
coupe au travers de la ravine (Clément et al, 2009). 3 : La dynamique des infiltrations dans une micro-dune sous
pluie simulée par tomographie de résistivité électrique (Descloitres et al, 2008). 4 : La spatialisation de la
recharge sous une mare de bas-fond par suivi temporel par tomographie de résistivité au travers de la mare.
Ce site présente un avantage important pour la géophysique: les sols, très secs après
quasiment six mois de saison sèche, se trouvent brusquement inondés par les ruissellements
consécutifs aux premières pluies de la mousson africaine. Les variations de résistivité sont
donc très importantes (plusieurs décades) et cela procure un avantage méthodologique : ce
fort contraste génère une réponse géophysique très prononcée, propice à la spatialisation des
processus d’infiltration et de recharge. Le corollaire à cette situation est de subir de fortes
distorsions géométriques des lignes de courant dans le sous-sol pouvant générer des effets
d’équivalence indésirables lors des inversions des sondages géophysiques.
33
b) Site de Moole Hole, Inde du Sud
Le site expérimental instrumenté de Moole Hole en Inde (ORE « BVET ») est situé en
zone tropicale humide, à régime de mousson (figure 10). Vierge de toute intervention
humaine, il permet l’étude des processus hydrologiques naturels. La zone non saturée (ZNS)
présente une épaisseur très variable latéralement, situation difficile à étudier pour la
géophysique. Les versants sont équipés d’un réseau de forages, ce qui permet la comparaison
avec les images de résistivité. Les variations de résistivité attendues en ZNS lors de la saison
des pluies favorisent l’étude de la recharge directe saisonnière sur les versants, qui s’effectue
au travers des systèmes de sols rouges. Les recharges indirectes sont étudiées sous la ravine
principale, intermittente.
Figure 10 .Bassin versant de Moole Hole en Inde du Sud. Le fond de carte correspond aux valeurs des
résistivités des sols et des altérites de socle gneissique. Cinq opérations majeures sont représentées : 1 et 2 :
L’étude des systèmes de sols et de l’infiltration durant la mousson (Barbiéro et al. 2007). 3 : L’étude du taux
d’altération chimique des gneiss par la réalisation de coupes de tomographies de résistivité sur tout le bassin
versant (Braun et al, 2009) 4 : La spatialisation de la recharge sous la ravine durant la mousson (Descloitres et
al, 2008) 5. La caractérisation des propriétés hydriques des altérites de socle par RMP (Legchenko et al, 2006).
c) Site d’Ara, Nord Bénin
Le site expérimental d’Ara (ORE « AMMA-Catch ») au nord Bénin (figure 11)
présente une situation climatique et géologique similaire à celle de Moole Hole (régime de
mousson et socle constitué de gneiss et d’amphibolites). Les variations du niveau des nappes
de plusieurs mètres entre la saison sèche et la fin de la mousson permettent d’étudier les
34
couvertures d’altération lorsqu’elles sont saturées, puis deviennent non saturées, situations
idéales pour appréhender leurs propriétés hydriques. Ce site est instrumenté intensivement
dans le cadre du programme AMMA (et suivi dans le cadre de l’ORE Amma-Catch), ce qui
permet en particulier des comparaisons avec les mesures d’évapotranspiration. Ce chantier est
encore en cours d’étude et constituera un site important de mes futures activités.
Figure 11 .Bassin versant d’Ara, Nord Bénin. Le fond de carte correspond aux valeurs des résistivités des sols et
des altérites de socle gneissique. 3 opérations majeures sont représentées : 1: La spatialisation des unités
géologiques par cartographie de résistivité, couplée à des prospections géologiques, 2: La mise en évidence des
altérites argileuses des versant par coupe de résistivité 2D (Kamagaté et al, 2007) 3 : La spatialisation des
propriétés des altérites de socle par sondage de Résonance Magnétique des Protons autour du site de
monitoring du gravimètre du programme ANR Ghyraf (Descloitres et al, soumis).
d) Autres sites étudiés
Sur les aquifères sédimentaires, en Bolivie et au Niger, mon intervention s’est faite
sous forme de missions ponctuelles. Le site de l’altiplano Bolivien se prête bien à la
35
déconvolution du signal de résistivité, grâce à la présence de plusieurs dizaines de forages où
la conductivité de l’eau de la nappe est connue. Cette situation se retrouve aussi pour
l’aquifère du continental terminal de la région de Niamey, où la profondeur du substratum
géologique repéré par géophysique peut être comparé à des données de plus de 30 forages.
Dans la même région, le site de Wankama au Niger (ORE « Amma-Catch ») présente des
zones de versant sableuses idéales pour vérifier les hypothèses d’infiltration profonde dans les
épandages sableux. Nous y avons réalisé des forages de contrôle permettant un couplage de la
géophysique avec les analyses géochimiques.
3. Synthèse des résultats
Ce chapitre présente la synthèse des principaux résultats, en tentant, lorsque cela est
possible, des parallèles entre les sites étudiés. 4 thèmes sont traités :
¾
¾
¾
¾
La spatialisation des aquifères,
L’étude des recharges dans différents contextes,
L’étude des transferts d’eau dans les premiers décimètres du sol,
Les principaux apports méthodologiques.
3.1. Spatialisation des aquifères
Six chantiers illustrent ce thème, trois en contexte de socle (sud de l’Inde, nord du
Burkina Faso et du Bénin), trois en contexte sédimentaire (altiplano bolivien, Est de Niamey
et rivière Komadougou près du Lac Tchad, tous deux au Niger).
a) Spatialisation des altérites de socle et de leurs paramètres hydriques
Initiée dans les années 1995 au Cameroun (Robain et al, 1996), l’étude des
couvertures d’altérations de socle et des aquifères qu’elles renferment a été approfondie lors
de la thèse de G. Toé que j’ai encadré au Burkina Faso et qui portait en particulier sur les
géométries d’électrodes adaptées à la reconnaissance des failles de socle (Toé, 2004). Ces
études se renforcent aussi grâce à la mise en œuvre de la résonance magnétique des protons au
Burkina Faso par J. M. Vouillamoz en 2003, avec mon appui. Mes travaux dans ce domaine
sont illustrés ici par trois études récentes.
La première avait pour objectif de contraindre le bilan géochimique d’altération du
bassin versant de Moole Hole en Inde par la connaissance de l’épaisseur du regolith. Cette
question des géochimistes m’a conduit à proposer des méthodes de cartographie et de
tomographie de résistivité (illustrées sur la figure 10 présentée page 34), couplées à des
mesures de résistivité en forage. Pour cette étude, nous avons établi une relation entre
résistivité et indices d’altération chimique « WIP » et « CIA » pour ces roches
métamorphiques, présentée sur la figure 12. Nous avons constaté que la valeur de résistivité
36
de 400 ohm.m permettait, sur ces roches, de séparer le domaine du protolith du domaine
d’altération, le regolith (Braun et al, 2009).
Figure 12. Relations entre
résistivité
mesurée
par
diagraphie en forage et les
indices
d’altération
chimiques des échantillons.
CIA= Chemical Index of
Alteration WIP = Weathering
Index of Parker D’après
Braun et al.( 2009).
L’intérêt de coupler le paramètre de résistivité avec le degré d’altération permet de
mieux comprendre comment séparer le réservoir de l’aquifère de sa partie fissurée / fracturée.
A cette fin, nous avons réalisé une étude numérique de sensibilité des paramètres d’inversion
de tomographie électrique pour déterminer l’incertitude associée à l’imagerie de la profondeur
du regolith avec une limite de 400 ohm.m (figure 13).
Figure 13 .Analyse de
l’incertitude associée à
l’inversion
des
tomographies électriques
sur
différentes
géométries de regolith
(travail de DEA de S.
Fleury, d’après Braun et
al, 2009).
Au total, l’incertitude due à l’inversion nous conduit à ré-évaluer l’épaisseur
d’altération de 17%. Grâce à la réalisation de douze transects géophysiques au travers du
bassin versant (montrés sur la figure 10, page 34) nous déduisons une épaisseur moyenne
d’altération de 17 mètres, ce qui pourrait correspondre à plus d’un million d’années
d’altération chimique sous climat tropical.
Même si la connaissance de l’épaisseur du regolith est d’un grand intérêt pour la
compréhension de la dynamique d’altération des roches, elle ne suffit pas à estimer la
ressource en eau stockée, ni les paramètres de flux, comme la conductivité hydraulique. Pour
37
cela, nous avons proposé d’utiliser la RMP. En raison de la complexité de l’aquifère, A.
Legchenko a développé une modélisation 2D (Legchenko et al, 2006). Comparée à l’imagerie
électrique (figure 14), la RMP apporte une information indispensable à la compréhension des
aquifères (Descloitres et al, 2008).
Figure 14 .Comparaison entre l’interprétation de la conductivité hydraulique déduite des mesures RMP (sans
calibration par forage) et les résultats de la tomographie électrique réalisée à l’exutoire du bassin versant de
Moole Hole (figure 10), d’après Descloitres et al. (2008). L’iso-contour 400 ohm.m marquant la limite
régolith/protolith est tracé en rouge pointillé (Braun et al., 2009). La teneur en eau RMP suit presque les mêmes
iso-contours, et montre des valeurs de l’ordre de 2.5% du volume total au maximum
La RMP, même avec une faible résolution spatiale, identifie la partie aquifère sous
l’isocontour 400 ohm.m. Les zones argileuses, détectées par la tomographie de résistivité,
occupent la majeure partie du regolith, au dessus. Par cette comparaison, nous apprenons que
le regolith ne forme pas ici un stock d’eau souterraine, et que la faible réserve d’eau (quelques
pourcents du volume total) se situe au sein du protolith, probablement dans la partie fissurée
mais non encore altérée de la roche. Ces résultats obtenus en Inde corroborent partiellement
ceux obtenus sur des formations granitiques altérées (Vouillamoz et al, 2005). Sur le site du
forage « KB 203 » au Burkina Faso par exemple, la résistivité électrique permet de distinguer,
grâce à des diagraphies, les roches fissurées/fracturées des altérites (figure 15).
0
clay
20
weathered
granite
30
fissured
granite
casing
10
00
10
10
10
0
00
0
-4
-5
Ro (ohm.m)
00
1x
10
4x
10
7x
10
1x
10
-5
-5
T (m/²s)
12
0
0
80
6
0
40
1.
0
Tpumping test = 3.10-4 m²/s
T1 decay constant
(ms)
8
water content
(%)
0.
Kombissiri KB203
Figure
15.
Résultats
géophysiques RMP et de
diagraphie de résistivité
obtenus sur le site test de
Kombissiri au Burkina Faso.
D’après Vouillamoz et al.
(2005)
40
granite
pegmatite
50
60
Q+
fresh
granite
fractured
zones ?
screen
70
Transmissivity
1
6 . 0 -6
0x
10 6
0x
0x
2.
Hydraulic conductivity
10
-6
80
4.
Depth (m)
Q+
K (m/s)
38
A la différence des altérites des roches gneissiques de l’Inde, les altérites du granite
apparaissent un peu plus poreuses (teneur en eau RMP 1.5%). On voit ici se dégager une
discrimination possible des types d’altérites en fonction de la roche mère à l’aide de leur
signature RMP.
Cette faculté discriminatoire de la RMP m’a conduit à entreprendre des
recherches au Bénin, avec pour objectif d’étudier les propriétés hydriques des altérites en
contexte de roches métamorphiques pour contribuer à la compréhension du bilan
hydrologique. En particulier, les parts respectives de la recharge et de l’évapotranspiration sur
le bilan total sont mal connues. En connaissant les variations des niveaux des nappes, et à
condition de connaître la porosité de drainage du regolith et ses variations spatiales, il est
possible de mieux contraindre la recharge (Kamagate et al, 2007, Guyot et al., 2009). J’ai
choisi pour ce site une approche impliquant une reconnaissance géologique, la cartographie
électromagnétique du regolith (représentée figure 11, page 35) et la caractérisation par des
sondages de résonance magnétique des protons des formations géologiques identifiées. La
figure 16 montre le résultat de la spatialisation des teneurs en eau RMP sur le site grâce à
l’identification géologique et la cartographie électromagnétique et ce, à l’échelle des mesures
de scintillométrie micro-onde mises en œuvre dans notre équipe par J. M. Cohard et A. Guyot
(Guyot et al, 2009).
Figure 16. Spatialisation des teneurs en
eau RMP à l’échelle des mesures
scintillométriques sur le bassin versant
d’Ara au Bénin (ORE Amma Catch). Les
chiffres en caractères gras indiquent la
teneur en eau RMP des altérites des
formations
métamorphiques.
Les
contours pointillés indiquent l’empreinte
de sensibilité du scintillomètre en
fonction du secteur du vent dominant au
long du trajet optique de l’instrument, en
rouge (voir Guyot et al, 2009).
A : Micaschists
B : Amphibolites
C : Quartzites
D : Gneiss migmatitiques
d’après Descloitres et al. (2010),
soumis.
On constate que les teneurs en eau RMP sont assez variables d’une formation
géologique altérée à l’autre : de moins de 1.5% pour les amphibolites (zone « A ») à plus de
10 % du volume total pour les quartzites (« C »), en passant par les gneiss migmatitiques
(« D »). Pour l’instant (2010), les teneurs en eau RMP ne sont pas calibrées par essai de
pompage, ce qui rend ce résultat relatif. En faisant l’hypothèse que les teneurs en eau RMP
maximisent les valeurs de porosité de drainage, l’analyse des signaux du scintillomètre peut
déjà être mieux comprise grâce à cette spatialisation (J. M. Cohard, étude en cours). D’un
point de vue appliqué, cette étude confirme aussi l’intérêt de choisir des formations de
quartzite fracturée pour l’implantation des forages ou des puits villageois, ouvrages qui
39
draineront des quantités d’eau 2 à 4 fois supérieures à celles des formations gneissiques ou
schisteuses.
b) Spatialisation régionale des aquifères sédimentaires
Ces études entrent dans la continuité de ma thèse sur la méthode électromagnétique en
domaine temporel (TDEM) appliquée aux aquifères volcaniques, montrant les capacités de
cette méthode à reconnaître les substratums argileux électriquement conducteurs ou les
biseaux salés (Ritz et al, 1997, Descloitres et al, 1997).
La première étude décrite ici avait pour but de contribuer à l’élaboration d’un modèle
conceptuel de fonctionnement d’un grand aquifère de l’altiplano bolivien (bassin endoréique
en climat semi-aride, situé entre le lac Titicaca et les zones des lacs salés d’Uyuni). La salinité
de l’eau souterraine varie notablement dans cette zone de 3000 km². Les eaux salées sont
repoussées lentement vers le sud par les recharges du fleuve Desaguadero. Si la géochimie
permet de quantifier la vitesse d’écoulement, très lente (1m/an), du flux souterrain et les
évaporations de l’aquifère (Coudrain et al., 2001), les chemins souterrains de l’eau restaient
inconnus, car tributaires de la répartition spatiale et en profondeur des formations argileuses,
entremêlées avec des formations sableuses plus perméables. Nous avons réalisé une centaine
de sondages électromagnétiques TDEM sur l’ensemble de la zone. La cartographie des
géométries des formations argileuses et du substratum conducteur (probablement formé par
des aquifères très salés) a été réalisée grâce à la relation que j’ai établie entre les valeurs de
résistivité et la salinité de la nappe, connue par une trentaine de forages. La figure 17 permet
de discriminer 3 grands types de terrain en présence : A) les formations sableuses, B) les
formations sablo argileuses, C) les formations d’argile à eau douce. Un quatrième domaine
(D) présente à la fois des valeurs de salinité élevées (> 3mS/cm) et des résistivités très basses
(< 10 ohm.m). Cette situation conduit à une indétermination sur le type de formation en
présence.
Figure 17. Relation entre la résistivité
interprétée des sondages TDEM et la
conductivité de l’eau souterraine sur
l’Altiplano Bolivien (Guérin et al.,
2004). Les porosités des sables déduites
de la loi d’Archie (figure 8) sont
importantes, de l’ordre de 30%, ce qui
représente un stock d’eau souterraine
important à l’échelle régionale.
40
Une telle relation, spécifique au site, permet de tracer les chemins probables de l’eau
souterraine, empruntant préférentiellement les chenaux sableux dans le sous-sol : la figure 18
présente les cartes de répartition des formations du sous-sol, à 10 et 50 m de profondeur. Les
formations sableuses favorisent un écoulement peu profond (10m) dans des chenaux au sudouest. Les formations argileuses font obstacle à la circulation de l’eau à l’est et au sud-est.
Figure 18. Cartographie des
formations
sableuses
et
argileuses à 10 (a) et 50 mètres
(b) de profondeur déduite de la
figure 17.
L’aquifère
sableux
est
nettement développé à 50
mètres de profondeur sur la
moitié de la surface étudiée,
confirmant les potentialités
d’exploitation de cet aquifère
(Guérin et al., 2004). Les
flèches bleues indiquent les flux
souterrains possibles.
Cette étude démontre les possibilités de spatialisation offertes par les sondages
TDEM, et propose les contraintes géométriques de modèles hydrogéologiques d’écoulement
(testés dans la thèse de A. Talbi, 2001, mais non publiés). Une conclusion importante ressort
de cette étude : il existe une zone d’indétermination pour les formations électriquement
conductrices : on ne discrimine pas les argiles des sables à eau salée, limitation très classique
des méthodes de résistivité.
Pour s’affranchir de cette indétermination, nos études se sont ultérieurement portées
sur l’utilisation de la RMP, capable de discriminer les deux cas. Cette complémentarité a
d’abord été montrée par nos études au Cambodge à la même époque (Vouillamoz et al, 2002)
pour une zone aquifère où la présence de lentilles argileuses disposées très aléatoirement
compromettait l’efficacité des programmes d’alimentation en eau potable des populations.
Plus récemment, deux études du même type confirment l’intérêt du couplage des
deux méthodes. La première concerne l’étude de l’aquifère du continental terminal de la
région de Niamey (figure 5, page 24) initiée par mes collègues J.M. Vouillamoz et G.
Favreau. Cette étude montre que le TDEM apporte une information clef sur l’épaisseur de
l’aquifère. En effet, cette méthode permet la détermination très fiable de la profondeur du mur
argileux, comme en témoigne la figure 19, où l’on voit que la corrélation entre le toit des
argiles et les profondeurs repérées par forages est excellente (Boucher et al., 2009).
41
Figure 19. relation entre la profondeur du toit des
argiles situées sous l’aquifère de Niamey calculées
par TDEM et profondeur repérées par forage
(compilation de G. Favreau, dans Boucher et al,
2009)
Ce résultat, attendu, permet d’entrevoir une spatialisation régionale de l’épaisseur de
l’aquifère, qui pourrait être faite sans difficultés par des prospections TDEM aéroportées.
Néanmoins, les études de sensibilité du modèle hydrogéologique de recharge menées par M.
Boucher montrent que la connaissance très précise de la profondeur du mur de l’aquifère n’est
pas cruciale pour le calage du modèle, alors que les résultats RMP sont, eux, très précieux.
Cette constatation m’a amené à proposer récemment une seconde étude dans le cadre du
programme IRD « Lac Tchad », coordonné par P. Genthon. L’objectif de cette étude est
d’améliorer le modèle conceptuel hydrogéologique de l’aquifère de la vallée de la
Komadougou au Niger, proche du Lac Tchad. Autour de la ville de Diffa, seule la rivière
participe de façon intermittente à la recharge. Les modifications de l’usage des terres (cultures
intensives de poivrons, développements projetés de l’agriculture irriguée) nécessitent
d’évaluer la ressource en eau souterraine disponible, ce qui constitue un second objectif, plus
appliqué. Une coupe géoélectrique TDEM faite au travers de la vallée, accompagnée de
sondages RMP, permet d’établir les bases du modèle conceptuel de l’aquifère. Sur la figure
20, on peut voir que la coupe géoélectrique délimite de grandes unités, surmontant un
substratum argileux. Ces classes de résistivité indiquent des variations entre un pôle argileux
(vert et bleu clair) et un pôle sableux (couleur rouge foncée). L’agencement des formations
étant très complexe dans le détail (feuillets de lits argileux au sein d’une matrice sableuse), la
méthode TDEM n’a pas la résolution suffisante pour les discriminer finement. Si la résolution
de la RMP est aussi insuffisante pour cela, la teneur totale en argile de la colonne aquifère
peut être calculée à l’aide des données RMP.
42
Figure 20. Coupe géoélectrique TDEM de l’aquifère de la vallée de la Komadougou, Niger, en haut. Les
pourcentages d’argile sur la colonne de terrain déduite des modélisations et des mesures RMP sont superposés
à la coupe TDEM (Chalikakis et al., en préparation). En bas, modèle conceptuel proposé pour les futures
modélisations hydrogéologiques (thèse de A. Moumouni Moussa, direction Genthon/Descloitres)
En réalisant une modélisation numérique RMP, on peut calculer une équivalence en
terme de pourcentage total d’argile de la colonne de terrain explorée sous le sondage. Cette
modélisation montre latéralement des variations importantes d’argile (de 8 à 45 %). Cette
information est cruciale si on veut appliquer un modèle d’écoulement dans ce système, car la
transmissivité de l’aquifère pourrait varier significativement latéralement. En terme de
ressource en eau pour l’irrigation, le couplage TDEM/RMP permet l’identification des zones
favorables aux forages, clairement situées dans des formations de sables grossiers situées
entre la surface et 40 mètres de profondeur. Notre étude montre aussi que l’absence de terrain
argileux en surface rend cet aquifère très vulnérable à la pollution.
c) En résumé
Pour les systèmes aquifères de socle, les études présentées ici démontrent la richesse
d’une approche multi méthodes. Les possibilités d’imagerie offertes par les tomographies de
résistivité permettent d’appréhender des géométries complexes du sous sol. Nous avons
constaté que les regoliths présentent, sur les sites étudiés, des résistivités inférieures à 400
ohm.m. Mais ces études souffrent de la difficulté à établir des fonctions de transfert entre la
résistivité et les variables hydrologiques. Pour pallier cette situation, l’utilisation de la RMP
apporte un éclairage essentiel sur les propriétés hydriques du regolith, principal compartiment
de stockage des aquifères. Lorsque les conditions expérimentales sont excellentes, la RMP
permet de reconnaître les zones fissurées. Néanmoins, la RMP ne permet pas encore d’obtenir
rapidement une résolution suffisante dans les aquifères complexes (du moins sans réaliser une
prospection RMP à maille très serrée), et la comparaison avec les méthodes d’imagerie
électrique est alors idéale. Nous avons aussi constaté que la capacité de stockage de ces
regoliths (spatialisée grâce à la teneur en eau RMP) semble bien dépendre du type de roche
mère. Ce point mérite d’être approfondi et discuté à l’avenir, car cela permettrait à terme une
cartographie des capacités de stockage des altérites de socle, et donc l’évaluation des
ressources en eau à partir d’approches couplées géologie/géophysique.
Les études en contexte sédimentaire montrent l’importance de l’identification des
zones argileuses (y compris les substratums) pour la compréhension des systèmes aquifères.
43
Si l’altiplano bolivien montre l’avantage du TDEM pour une spatialisation régionale et
l’identification des eaux salées, le cas de Diffa montre lui la capacité de la RMP à discriminer
les formations argileuses des formations plus sableuses, dans les limites de résolution de la
méthode. Ainsi, en combinant la rapidité des sondages TDEM de résistivité à la possibilité de
quantification des paramètres hydriques de la RMP, la construction des modèles
hydrogéologiques régionaux est facilitée.
3.2. Recharge des aquifères
Ce thème concerne les sites en contexte de socle du Burkina Faso et d’Inde, et en
contexte sédimentaire à Wankama, au Niger. C’est principalement le suivi temporel de
résistivité qui est utilisé, mais la RMP est mise aussi à contribution sur le site de Moole Hole
(Inde). Ce thème illustre en particulier les difficultés des suivis temporels de résistivité, ainsi
que les apports d’une approche multidisciplinaire.
a) Les ravines de versant sont-elles des lieux de recharge ?
S’il est généralement établi que les mares et les axes drainants de bas-fond sont les
lieux de la recharge indirecte des aquifères au Sahel, les ravines de versant pourraient jouer
aussi un rôle comme le suggèrent Peugeot et al. (1997) au Niger sahélien. Ils observent que
l’infiltration s’accroît lorsque les écoulements sont localisés dans des ravines temporaires, et
particulièrement lorsqu’ils traversent des sols sableux et caillouteux. Esteves et Lapetite
(2003) concluent aussi que des infiltrations profondes peuvent exister lorsque les ravines de
versant traversent des sols sableux épais (supérieurs à 10 m).
La première étude est menée dans la zone sahélienne du nord du Burkina Faso, qui se
trouve exactement dans la même situation climatique qu’à Wankama à l’est de Niamey au
Niger. Mais au Burkina Faso, le contexte géologique est différent : la couverture de sols est
très argileuse en raison d’une altération des roches cristallines comportant des minéraux plus
sensibles à l’altération chimique. De plus nous avons observé, partout dans le paysage, que les
ruissellements concentrés détruisaient les croûtes indurées et imperméables des horizons de
surface pour former des ravines d’érosion. L’objectif de cette première étude était donc de
vérifier si les ravines de versant pouvaient être infiltrantes, ou non, et si des infiltrations
profondes pouvaient être influencées par certains états de surface. La ravine de l’exutoire du
bassin versant sur le site de Katchari (figure 21) a été choisie parce qu’elle présente tous les
états de surface de la zone, et traverse un filon de quartz fracturé qui pouvait favoriser
l’infiltration.
44
Figure 21. Ravine de
l’exutoire de Katchari.
Carte des états de
surface
et
des
dispositifs
géophysiques. La ligne
rouge représente le
tracé
des
lignes
d’électrodes pour la
coupe géoélectrique.
Rectangles
rouges :
fosses pédologiques de
contrôle.
D’après
Descloitres
et
al.
(2003)
J’ai mis en œuvre une approche de suivi des résistivités apparentes sur toute la saison
des pluies. En parallèle, un suivi temporel a été réalisé en tomographie de résistivité au travers
de la ravine. L’ensemble a été contrôlé par des mesures neutroniques implantées dans 6 tubes
d’accès à 6 mètres de profondeur, et par l’excavation de fosses pédologiques. Pour le suivi
cartographique, les résultats obtenus sont présentés sur la figure 22.
Figure 22. Variations des
résistivité
apparentes
depuis le début de la saison
de mousson (juin) jusqu’ au
milieu de la saison sèche
(mars). Dispositif Wenner,
écartement d’électrodes :
5m. Les diminutions de
résistivité apparaissent en
bleu, les augmentations en
orange et rouge. Le tracé
de la ravine est souligné
par des lignes noires
continues.
D’après
Descloitres et al. (2003)
Ces résultats montrent que, juste après les premières pluies, la résistivité apparente
mesurée décroît dans certaines zones à l’ouest, y compris en dehors du tracé de la ravine, et
augmente plutôt à l’est. Le tracé de la ravine n’est marqué par aucune anomalie de variation
de résistivité apparente. La tentation est forte de traduire les décroissances mesurées
directement en zones d’infiltration. Mais comment expliquer que la résistivité apparente
augmente dans d’autres zones ? S’agit-il d’une dessiccation plus profonde ? Pourquoi à cette
époque des premières pluies ? Illogisme, confirmé par les suivis neutroniques : aucune
dessiccation n’a eu lieu en profondeur. Dès lors, tout notre travail a été d’expliquer ce résultat.
Pour cela, j’ai construit un modèle synthétique des infiltrations possibles. Olivier Ribolzi a
proposé de réaliser des expérimentations contrôlées dans des fosses de contrôle, implantées
grâce aux cartes géophysiques. Nous avons constaté : a) dans la zone de diminution de
résistivité apparente la présence de carbonates se diluant rapidement lors des premières pluies,
45
faisant considérablement chuter la résistivité des sols, b) des infiltrations très superficielles
dans les tubes neutroniques. Les modélisations synthétiques représentant la réponse d’un
sondage électrique à des scénarios d’infiltration, présentées sur la figure 23, montrent que, en
cas de chute de la résistivité vraie dans les tous premiers décimètres du sous sol, la résistivité
apparente peut augmenter pour les mesures à longueur de ligne intermédiaires. Or, c’est
justement ces longueurs que nous avons utilisées pour la prospection de résistivité apparente !
Ce phénomène, dû à ce que les géophysiciens de terrain appellent le « retard à la remontée des
courbes de sondage électrique », avait déjà été constaté par Louis Cagniard en 1959 pour des
mesures sur glacier, où la neige recouvrant le glacier jouait le rôle de couche conductrice
superficielle au dessus d’un milieu plus résistant, la glace! Dans notre cas, nous nous sommes
fait « piégés » par ce phénomène, intrinsèque à la méthode.
Figure 23. Courbes de
sondage
électrique
synthétique construites à
partir
d’un
scénario
d’infiltration superficielle
dans
des
sols
non
carbonatés. On montre la
chute
de
résistivité
apparente pour les courtes
longueurs de ligne. La zone
rouge
souligne
l’augmentation pour des
longueurs
de
ligne
intermédiaires Descloitres
et al. (2003)
Finalement, cette étude montre que les ravines de versant ne sont pas le siège
d’infiltrations profondes dans ce contexte sahélien, avec des sols issus de l’altération de roche
de socle. Notre cartographie de suivi temporel permet néanmoins d’identifier les zones de sols
carbonatés. De plus petites longueurs de lignes d’électrodes auraient permis une cartographie
plus fine de ces sols, grâce à leur facilité à faire chuter drastiquement la résistivité par la
dissolution rapide de carbonates. Nous pensons que c’est bien la nature très argileuse des sols,
associée aux temps de transit trop rapides des écoulements sur les surfaces potentiellement
infiltrantes, qui sont responsables de l’absence d’infiltration dans les versants sahéliens de
socle.
Les résultats des suivis temporels en coupe de résistivité au travers de la ravine,
interprétés à la même époque (2003) sont aussi problématiques. Même s’ils utilisent des
écartements d’électrodes réduits pour tenir compte des infiltrations superficielles, ils mettent
en évidence un problème sérieux pour le suivi temporel en tomographie : les augmentations
de résistivité apparente dues à des infiltrations très superficielles semblent se répercuter dans
l’inversion : La figure 24 montre en effet des zones d’augmentation de résistivités calculées
apparaissant en profondeur dans les coupes, zones d’augmentation contredites par les mesures
de sonde à neutron.
46
Figure 24. Coupe de
résistivité en suivi temporel
au travers de la ravine de
Katchari)
D’après
Descloitres
et
al.
(conférence Geofcan, 2001)
Les différentes solutions imaginées avec les outils d’inversion disponibles à l’époque,
comme contraindre les inversions par les données de diagraphies, n’améliorent pas cette
situation (Descloitres et al., conférence Geofcan, 2001). Je montrerai plus loin comment ce
problème a été résolu depuis, dans le cadre de la thèse de Rémi Clément.
Pour la deuxième étude, nous changeons de contexte géologique pour nous retrouver
au Niger en zone sédimentaire. Cette étude avait pour objectif d’étudier les recharges des
aquifères sur un autre élément morphologique du paysage sahélien de versant, les épandages
sableux. Ces épandages étaient identifiés par mes collègues G. Favreau et S. Massuel comme
potentiellement infiltrants, en plus des ravines localisées déjà connues (figure 25). Cette étude
a été réalisée sur le bassin versant expérimental de Wankama au Niger (ORE « AMMACatch ») qui présente un épandage sableux caractéristique. Afin de faciliter l’implantation de
forages destinés à prélever des échantillons pour l’analyse géochimique, j’ai proposé une
approche par cartographie de résistivité électromagnétique à différentes profondeurs
d’investigation, et une imagerie par coupe de résistivité positionnée grâce à ces cartes. Nous
comptions ainsi reconnaître la zone non saturée sableuse très épaisse jusqu’à la nappe située à
plus de 25 mètres de profondeur. Pour cette étude, j’ai imaginé, puis conçu et breveté avec
mon collègue Y. Le Troquer, un outil de diagraphie en forage adapté à la zone non saturée.
47
Figure 25. Bassin versant de Wankama, Niger. A et B : localisation des forages, cartographies de résistivité. C :
coupe de résistivité Les diagraphies géophysiques réalisées dans les forages 1 et 2, placés grâce à la
géophysique, sont comparées aux analyses géochimiques (D). D’après Massuel et al. (2006)
Les résultats montrent que le positionnement très précis des forages par la
géophysique a permis aux hydro-géochimistes d’identifier clairement des zones où le
lessivage des éléments chimiques vers la nappe était prédominant. De plus, les résultats
géophysiques montrent que les chemins de percolation à la nappe sont discontinus, de
largeurs variables : à l’image des phénomènes d’infiltration préférentielle en forme de doigt
(« fingering ») observés dans la proche surface de certains sols lors d’une infiltration, nous
avons ici un phénomène analogue, mais à l’échelle d’un versant. Cette interprétation, qui
enrichit le modèle conceptuel de la recharge dans ces zones, a été rendue possible grâce au
couplage des données de résistivité en forage et de géochimie (Massuel et al., 2006).
b) Comment s’effectue la recharge des aquifères par les ravines de bas fond ?
Après les versants, l’objectif était de tenter de localiser et de quantifier les recharges
indirectes au niveau des axes drainants de bas-fond. Au Sahel, ces axes concentrent l’essentiel
des ruissellements. En milieu soudanien, il faut savoir faire la part entre une recharge directe
de l’aquifère sur les versants, et des recharges indirectes dans les zones où les ravines peuvent
être intermittentes, du moins sur les sites étudiés.
Au Burkina Faso, au travers de la ravine de bas fond du site de Katchari (figure 9,
page 33), un suivi temporel en coupe de résistivité a été tenté. Je ne le présenterai pas ici, car
ce suivi, entaché probablement d’artefact d’imagerie, doit être réinterprété. Pour illustrer mes
travaux, je m’appuie sur le cas de l’Inde, à Moole Hole, où les suivis de résistivité ont pu être
comparés à des mesures en forages et enrichis par une tentative réussie de suivi temporel
RMP.
A Moole Hole, c’est aussi la zone de l’exutoire (figure 10, page 34) qui a été choisie
en raison de la piste le longeant, seul endroit accessible aux camions pour les forages de
contrôle dans cette forêt vierge. Sur la figure 26, je présente les résultats de la coupe de
48
résistivité électrique faite au travers de la ravine de l’exutoire. Pour obtenir une image
géophysique fiable, il a d’abord fallu ajuster les paramètres d’inversion en comparant les
modèles géophysiques avec les données de résistivité mesurées en forage. L’image obtenue
est complexe, et montre que le socle rocheux pointe sous la surface en de nombreux endroits,
pointements entrecoupés par des incisions profondes de matériaux plus conducteurs. La
longueur d’onde des variations latérales de résistivité est très courte : par exemple, l’image
électrique montre le socle à 5 mètres de profondeur au niveau du piezomètre 7, alors qu’il est
situé à plus de 25 mètres sous le piézomètre 8, situé seulement 35 mètres à coté !
Figure 26. En haut ; coupe
de tomographie électrique
au travers de la ravine de
Moole Hole, localisation
des forages de contrôle,
avec l’exemple de la
diagraphie de résistivité
dans le piézomètre n°7
comparée aux inversions.
En bas, suivi temporel de
résistivité dans la zone
centrale.
L’image
représente les valeurs du
rapport des résistivités
calculées (état avant les
pluies : état après les
pluies).D’après Descloitres
et al. (2008)
Les paramètres d’inversion géophysique, ajustés par les données de forage, sont
utilisés pour analyser les différences de résistivité obtenues après les pluies (en bas de la
figure 26). Des zones de diminutions de résistivité sont identifiées en zone non saturée, sous
la ravine, indiquant une recharge localisée, indirecte. Dans les versants, nous remarquons
aussi des diminutions de résistivité dans les premiers mètres sous la surface, indiquant cette
fois ci une possibilité de recharge directe, généralisée sur l’ensemble du bassin versant. Cette
infiltration profonde est confirmée par des mesures neutroniques dans des trous de tarière. En
zone saturée, dans la nappe, l’image géophysique montre que la résistivité diminue, ce qui
pourrait correspondre à une concentration en solutés qui s’accroît. Ce résultat n’est pas
corroboré par la mesure de la résistivité des eaux, qui augmente au contraire : en effet, la
concentration en solutés diminue, suite à la dilution des eaux anciennes par l’arrivée de l’eau
de pluie nouvelle dans l’aquifère. Un autre artéfact de l’imagerie géophysique est repéré :
dans les versants en zone non saturée, le calcul nous montre une augmentation de la résistivité
en dessous de l’infiltration de surface. Cela n’est pas très logique et ne pourrait être expliqué
que par un prélèvement profond par les racines… Il existe une autre façon d’expliquer cette
image : ce résultat peut être critiqué de la même façon que sous la ravine de Katchari (figure
49
24, page 47): un effet d’infiltration superficielle pourrait être mal pris en compte par
l’inversion, en raison de modèles équivalents possibles. Je montre dans cette étude, par une
modélisation numérique, que c’est bien le cas : on peut reproduire l’image de variation de
résistivité uniquement par l’introduction d’une infiltration de surface, sans qu’il y ait de
dessiccation en profondeur ! Cette conclusion renforce mon avis que la procédure
d’interprétation de suivi temporel doit être l’objet d’une étude méthodologique spécifique, qui
doit en particulier tenir compte des variations de surface, même quand il y a des fronts
d’infiltration plus profonds. L’étude de l’infiltration a été aussi réalisée par deux sondages
RMP centrés sur la ravine à deux époques différentes : l’objectif était de quantifier les
volumes d’eau que le regolith peut contenir lors de la recharge. La figure 27 montre le résultat
de l’interprétation des sondages RMP.
Figure 27. Suivi temporel RMP, depuis la fin de la saison des pluies (novembre, aquifère à son plus haut niveau)
et le milieu de la saison sèche (fin janvier, niveau minimum). A gauche, variation de la teneur en eau selon la
profondeur. La partie poreuse de l’aquifère (2.5%) se retrouve non saturée, et fait apparaître la partie fissurée,
qui contient nettement moins d’eau (0.5%) et présente une conductivité hydraulique plus faible (à droite).
D’après Descloitres et al. (2008).
Les conditions de mesure exceptionnelles de ce site (bruit de fond électromagnétique
extrêmement faible) permettent de connaître la variation de stock d’eau possible dans
l’aquifère, environ 2.5 % du volume total. Une fois le niveau de la nappe redescendu, seule la
partie fissurée de la roche contient encore de l’eau, environ 0.5% du volume total. La RMP
apparaît ainsi être un outil de suivi temporel intéressant pour quantifier les variations de stock
d’eau lors de la recharge à l’échelle des ravines.
c) En résumé
Les suivis temporels géophysiques ont été utilisés pour localiser et spatialiser, sous
forme de carte ou de coupes, les recharges indirectes des aquifères par les ravines. Pour
préciser les modèles conceptuels en zone sahélienne, nous nous sommes intéressés aux zones
de versants, mal connues. Deux situations se rencontrent selon la géologie: en zone de socle,
les versants ne jouent probablement pas un rôle infiltrant malgré les multitudes de ravines
50
d’érosion existantes. En zone sédimentaire, en particulier lorsque les terrains sous jacents sont
à dominante sableuse, les épandages sableux peuvent jouer un rôle dans la recharge. Une fois
cartographiés, ces épandages pourraient être inclus dans les sources de recharge des modèles
hydrologiques régionaux.
Les suivis temporels géophysiques en zone plus humide ont été utilisés pour
caractériser les recharges indirectes par les ravines de bas fond, mais aussi pour mettre en
évidence des recharges directes dans les versants. Nous avons appris que les axes drainants
intermittents sont bien le siège de recharges indirectes localisées. Dans les versants, nos
études montrent des infiltrations généralisées. Si les études géophysiques menées ne
quantifient pas les taux de recharge, elles permettent la spatialisation des phénomènes, et
facilitent l’implantation de reconnaissances plus ponctuelles par forage à des fins d’analyses
géochimiques ou autres. Les études ultérieures de Maréchal et al. (2009) et de Ruiz et al.
(2009) s’appuieront en partie sur les résultats géophysiques pour étudier les termes de
recharge directe et indirecte sur ce site, et dégager le rôle clef que joue le regolith, qui stocke
des réserves d’eau utilisées par la forêt lors des années de sécheresse.
D’un point de vue méthodologique, nous avons montré que les suivis temporels de
résistivité peuvent présenter des artéfacts notables. Nous avons identifié des causes possibles
à ces effets indésirables, notamment la variation de résistivité dans les premiers décimètres
des sols, la non prise en compte des modèles équivalents et/ou les effets de régularisation dans
les inversions.
3.3. Etude des transferts d’eau dans les premiers décimètres du sol
Ce thème concerne l’étude des processus élémentaires du cycle de l’eau. Il s'agit de
recherches menées à de petites échelles sur le terrain (parcelles, séquence de sols) avec pour
objectif de contribuer à établir des modèles de description et de quantification de ces
processus. Etant donné qu’on s’intéresse à des échelles d’espace et de temps réduites (de
quelques décimètres à quelques mètres, de quelques minutes à quelques jours), ces études
utilisent principalement la méthode d’imagerie de résistivité, adaptable à ces échelles.
J’illustre ces études par l’expérience la plus complète, celle réalisée sur une micro-dune sous
pluie simulée au Burkina Faso. Une autre expérience a été menée en Inde montrant les
limitations du suivi temporel de résistivité dans des sols plus argileux
a) Les transferts d’eau dans une micro-dune lors de cycles de pluie
L’expérience de la microdune avait pour objectif de comprendre les processus
d’infiltration et d’évaporation (ou de drainage) dans les sols sableux hétérogènes. Comme je
l’ai expliqué en présentant la figure 6 (page 25), les microdunes présentent une structure
interne complexe héritée de l'interaction entre les phénomènes hydriques et éoliens, structure
qui pourrait également jouer un rôle clef dans l'infiltration car il existe des couches
superposées de perméabilité différente. La végétation sur le coté sous le vent pourrait
favoriser l’infiltration. Lors d’une pluie, on ne connaît pas la redistribution de l’eau en son
sein, et il faut donc connaître la part d’eau ancienne et d’eau nouvelle participant aux
51
ruissellements démarrant de la micro-dune. Cela est nécessaire pour mieux comprendre
comment déconvoluer les hydrogrammes de crues mesurés en aval du bassin. De même, pour
comprendre la dynamique des ruissellements, on aimerait savoir si une microdune peut
générer des écoulements rapides (démarrage de ruissellement de surface par engorgement
rapide des premiers centimètres du sol) ou jouer le rôle de « tampon » en absorbant l’eau de
pluie plus profondément. Nous avons décidé de coupler une approche hydrochimique et
d’imagerie de résistivité, en contrôlant non seulement l’intensité des pluies, mais aussi leur
chimie. Le dispositif expérimental est décrit sur la figure 28.
Figure 28. Dispositif expérimental
l’expérience « micro-dune ».
A : simulateur de pluie sur le site
Katchari (Burkina Faso).
B : implantation des dispositifs
surface et aperçu des rigoles
récupération des écoulements, coté
vent et sous le vent.
C. Plan de position des capteurs
surface. D’après Descloitres et
(2008)
de
de
de
de
au
en
al.
Pour le géophysicien, cette expérience présente l’avantage de pouvoir accéder au cœur
du terrain juste après la dernière pluie, en coupant la microdune en deux. L’image de la figure
29 ci-dessous a donc été prise après l’expérience. Nous avons implanté deux micro-forages
pour mesurer la résistivité au sein du sol. De manière à prévenir des infiltrations
préférentielles au long de ces tubes, ceux-ci ont été placés à la perceuse 8 mois avant
l’expérience, permettant ainsi au terrain de se réaménager de façon naturelle autour des
sondes géophysiques. Nous avons aussi mesuré les variations de température dans les 30
premiers centimètres. Cela permet de corriger les valeurs de résistivité dues aux amplitudes de
variation thermique très importantes au sahel dans les premiers décimètres (plus de 15°C, soit
plus de 30% de variation de la valeur de résistivité).
52
Figure 29. Dispositif expérimental
géophysique, vu en coupe.
a) principaux interfaces pédologiques
repérés après l’expérience.
b) implantation des électrodes, des
tubes de diagraphie électrique, de
mesure de température, et tracé des
cellules de prélèvement (conception O.
Ribolzi et J.P. Thiébaux). D’après
Plusieurs pluies simulées ont été réalisées, la dernière utilisant un traceur salin pour
amplifier le contraste de résistivité (et hydrochimique) et faciliter ainsi le traçage de l’eau
nouvelle par rapport à l’eau ancienne. Dans un premier temps, les inversions géophysiques
réalisées en routine ne permettaient pas d’être validées par la comparaison avec les valeurs
des diagraphies de résistivité dans les micro-forages. J’ai conduit alors une étude sur
l’optimisation des paramètres d’inversion. Il en ressort les conclusions suivantes :
a) la connaissance de la zone invariante de résistivité en profondeur est essentielle pour
limiter le domaine de calcul des résistivités à des endroits réalistes,
b) il faut limiter l’inversion à 2 itérations pour ne pas amplifier les ajustements excessifs
des valeurs de résistivité, pilotés uniquement par le critère de convergence.
Malgré cette optimisation de l’inversion, les résultats des images de résistivité
présentent, en profondeur notamment, des écarts de plus de 20% avec la réalité. C’est donc la
limite de résolution obtenue. La figure 30 présente le résultat du suivi temporel sous les pluies
successives, corrigé des variations de température. Après la pluie n°2, réalisée après 2 jours
d’assèchement, on constate que l’infiltration se produit préférentiellement sous le coté au vent
(à droite de la microdune sur ces images). L’évaporation se produit de façon plus homogène.
Après la pluie n°3, salée, on constate que l’eau s’infiltre plutôt de façon homogène,
remplissant les zones préalablement évaporées partiellement. De même, l’évaporation se
produit ensuite de façon homogène. Les différences de dynamique de « versant » sont-elles
dues à la fréquence des pluies ? L’hydrogéochimie nous apprend aussi que l’eau présente dans
la microdune avant la pluie salée avait une concentration en soluté supérieure à la
53
concentration de l’eau de pluie salée…Dès lors, il devenait difficile de tracer les eaux
anciennes des eaux nouvelles, malheureusement.
Figure 30. Résultats du suivi temporel de résistivité au sein de la microdune, pour deux pluies successives, la
première avec une eau déminéralisée (« rain 2 ») et la seconde après la pluie salée (« rain 3 ») En haut, les
images de la résistivité calculée après optimisation de l’inversion. En bas, les rapports de résistivité permettant
d’appréhender les zones d’humectation (diminution de la résistivité, couleur bleue à violette) et les zones de
dessiccation (évaporation, en couleur orange à noire).
Dans un deuxième temps, notre objectif était de tenter une spatialisation de la teneur
en eau grâce à la spatialisation de la tension, elle-même déduite de la résistivité… Mais une
relation directe entre résistivité et tension ne peut pas être établie en raison des variations
importantes de porosité dans ces sables, influant significativement sur la valeur de résistivité.
J’ai alors comparé les valeurs de variation de la résistivité avec les différences de tension
mesurées par des micro-tensiomètres au sein de la dune. Cette comparaison donne les
résultats montrés sur la figure 31.
1.2
Rain 1 : evaporation
afteronly
rain
6
Rain 2 : rain
after rain
evaporation
1
0.8
2
fit: Equation
Y = -1.61 * ln (X) + 0.04
R-squared = 0.797
7
3
5
INFILTRATION
DOMAIN
0.4
5
8
4
9
4
0
-4
T1
2
4
T6
T2
-8
T7
T3
T8
1
-12
T9
-16
-20
T10
3
7
6
T5
1.1
0.6
8
EVAPORATION
(or drainage)
DOMAIN
T4
-0.8
0.5
2 1
6
7
tensiometers
0
8
5
9
-0.4
9
Figure 31. Relation expérimentale entre les
rapports de résistivité (après / avant la pluie) et les
différences de tension mesurée au sein de la
microdune.
Le domaine d’infiltration (en fond bleu) est défini
par des différences de tension positives.
La localisation des tensiomètres utilisés pour
établir cette relation est montrée en encart en bas à
gauche.
Les données utilisées sont celles après la pluie n°1,
avant et après la pluie n°2, toutes deux non salées.
D’après Descloitres et al. (2008)
0.7
DECREASE
0.8
0.9
1.2
1.3 1.4 1.5 1.6 1.7 1.8 1.9
INCREASE
1
resistivity ratio (final / initial)
2
54
On établit ainsi expérimentalement une relation linéaire entre les rapports des
résistivités et les différences de tension. Cette relation n’est pas vérifiée lors de la pluie salée,
car les variations de conductivité de l’eau dues au traceur brouillent complètement le signal de
variation de résistivité, en introduisant un deuxième paramètre dans la loi d’Archie, en plus du
facteur saturation. Nous n’avons pas tenté de dériver cette relation vers la teneur en eau, en
l’absence de relation expérimentale de tension/saturation bien établie, comme le montre la
figure 32.
Figure 32. Relation expérimentale entre la tension et
la saturation sur les sables de la microdune. D’après
b) Les transferts d’eau dans un système de sols argileux : les limites de l’emploi de
la résistivité
L’objectif de l’étude entreprise en Inde avec L. Barbiéro et L. Ruiz (cellule IRD de
Bangalore) était d’établir le fonctionnement hydrique d’un système de sols complexe, où des
sols argileux gonflants (les « sols noirs ») se développent au sein de sols rouges. Ces sols
noirs alternent des phases de rétractation ou de gonflement selon les périodes de la saison
hydrologique. Ce fonctionnement est important à analyser car il génère des phénomènes
d’érosion en forme de « cuillère » (Barbiéro et al, 2007). Sur le site de Moole Hole, nous
avons choisi la séquence de sols « T3 » (repérée figure 10, page 34), pour mettre en œuvre les
suivis temporels géophysiques. Cette séquence présente une coupe naturelle permettant la
description des horizons pédologiques. J’ai implanté des tubes d’accès neutroniques en arrière
plan de cette séquence, comme le montre la figure 33. Les relations comptage
neutronique/teneur en eau ont été obtenues par calibration à différents états hydriques. Des
mesures du potentiel hydrique et de prélèvement des eaux interstitielles ont été faites à
différentes profondeurs.
55
Figure 33. Description pédologique de
la séquence de sols T3 en Inde. Les
tubes neutroniques, ainsi que les
variations de teneur en eau entre mars
et juin 2004 y sont représentés, et
montrent les infiltrations préférentielles
dans le tube A4 traversant les sols
noirs.
Nous avons constaté que les sols noirs (à l’aval, sous les sols rouges) ont une activité
hydrique importante en début de saison des pluies, car les fentes de retrait y favorisent
l’infiltration profonde. Ces fentes se referment ensuite, et les sols noirs jouent alors le rôle
d’écran à l’infiltration. Avant d’utiliser les suivis temporels de résistivité, connaissant
l’influence de la présence d’argiles sur la valeur de la résistivité, j’ai cherché à établir des
relations entre résistivité et teneur en eau, en utilisant l’outil de diagraphie de résistivité à
membrane gonflable créé au Burkina Faso. Un exemple en est donné sur la figure 34.
Figure 34. Relation expérimentale entre
la résistivité et la saturation sur les sols
rouges et noirs de la séquence de sols
T3. La photo montre le site en saison
sèche, avec au premier plan la sonde à
neutron. Les mesures de teneur en eau
proviennent du tube neutronique A4
traversant les sols noirs (voir figure 33).
Les données de résistivité ont été
acquises par diagraphie dans un autre
forage situé à 25 cm du tube A4.
D’après ces résultats, on peut voir que ces relations ne sont pas simples. Tout d’abord,
dans les horizons de sols rouges proches de la surface, il semble exister une décroissance de la
résistivité en fonction de l’augmentation de la teneur en eau massique. La forme exacte de la
relation est de plus très incertaine. Ensuite, au sein des sols noirs argileux à 200 et 250 cm de
profondeur, la relation est très dispersée, et présente une tendance grossièrement inverse…
Nous pensons que cette tendance pourrait être due à la diminution de la concentration des
eaux interstitielles, qu’il nous a été techniquement impossible de mesurer malgré nos essais de
56
prélèvement à ces profondeurs. Ces relations mal établies et trop dispersées nous ont fait
écarter la possibilité de spatialiser les variations de teneur en eau par la résistivité.
c) En résumé
D’un point de vue méthodologique, l’expérience menée sur les sols sableux montre
que l’obtention d’une image géophysique des variations de résistivité mieux contrainte passe
par l’optimisation des paramètres d’inversion d’une part, et par la connaissance de zones
invariantes dans le milieu d’autre part, ce qui permet de limiter le calcul aux endroits-clefs.
Cela a aussi pour corollaire d’avoir à disposer de données acquises au sein même du milieu,
ce qui est contraignant. L’analyse de la résistivité doit s’affranchir des variations spatiales de
porosité, et il faut donc traiter les données sous forme de rapport de résistivité.
Les fonctions de transfert entre résistivité et variables hydrologiques d’intérêt sont
difficiles à obtenir in situ. Une piste est esquissée pour les sols sableux en proposant une
relation linéaire entre les différences de tension hydrique et les rapports de résistivité, mais
elle doit être explorée à l’avenir pour la confirmer. Cette piste est intéressante car elle
permettrait de spatialiser qualitativement par la géophysique les zones d’humectation ou de
dessiccation dans un sol sableux. Cela donne l’espoir de pouvoir tracer les mouvements de
l’eau par les suivis temporels de résistivité sans avoir à mettre en place de nombreux
tensiomètres. Une spatialisation quantitative serait par contre indirecte puisque les tensions de
l’eau doivent elles-mêmes être retranscrites en teneur en eau.
D’un point de vue apport de connaissance sur les processus hydriques dans les micro
dunes, nous mettons en évidence, grâce à l’imagerie géophysique, un rôle important du
versant au vent, semblant favoriser les infiltrations, surtout après une période d’assèchement
assez longue. Nous constatons aussi que la micro stratification ne semble pas jouer un rôle de
guide des écoulements de sub-surface. Ces observations déduites de la géophysique ont
conduit mes collègues à généraliser cette observation sur l’ensemble du paysage (Ribolzi et
al, 2006)
Pour les sols plus argileux, l’étude menée en Inde illustre les difficultés à vouloir
suivre la dynamique des processus élémentaires dans les sols argileux à l’aide des variations
de résistivité électrique, puisque la relation entre résistivité et teneur en eau est complexe. De
plus établir ces relations in situ, à différentes profondeurs, est très difficile techniquement.
Seules les données de sonde à neutron (et la cartographie des résistivités des sols à l’échelle
du bassin, montrée sur la figure 10, page 34) seront utilisées par Barbiéro et al. (2007) pour
compléter l’étude dynamique de ces systèmes de sol. Les études futures devront se tourner
vers l’emploi de méthodes complémentaires, comme la résistivité complexe (IP fréquentielle
par exemple) pour avancer dans ce domaine. Finalement, seule l’expérience menée sur les
sols sableux permet de dégager l’utilité et les promesses du suivi temporel de résistivité pour
l’étude non destructive des processus élémentaires.
57
3.4. Apports méthodologiques
Ce thème est illustré par a) la conception faible coût d’un outil de diagraphie de
résistivité en zone non saturée et b) nos récents travaux sur l’amélioration des suivis
temporels de résistivité.
a) Conception d’une sonde de diagraphie de résistivité en zone non-saturée.
S’il est facile de mesurer la résistivité en forage lorsque celui-ci est rempli d’eau,
grâce au contact électrique ainsi réalisé entre l’électrode et le terrain, c’est en revanche
impossible en zone non saturée, à moins de plaquer les électrodes sur la paroi nue du forage.
Les outils existants sont issus du domaine pétrolier et utilisent des mécanismes de placage des
électrodes (sorte de patins métalliques) actionnés mécaniquement ou électriquement. Depuis
les années 50, d’autres concepteurs ont imaginé divers systèmes (par exemple des sortes de
baleines de parapluie actionnées mécaniquement). Difficilement miniaturisables à l’échelle du
trou de tarière (diamètre de 5 à 7 cm environ), ils restaient compliqués et donc peu attractifs.
D’où l’idée d’utiliser des membranes gonflables et des électrodes constituées de colliers
métalliques extensibles destinés à être plaqués contre la paroi. Avec mon collègue Y. Le
Troquer, nous avons conçu l’outil de diagraphie montré sur la figure 35.
Figure 35. A gauche, sonde 2 électrodes dégonflée, au milieu, une fois gonflée. A droite, la sonde est reliée au
résistivimètre, et à la valve de gonflage pour réaliser les diagraphies dans les forages de Wankama (résultats
des diagraphies, figure 25)
Cet outil, adaptable à de multiples diamètres (depuis le trou de tarière de 4 cm
jusqu’au forage d’eau de plus de 25 cm de diamètre) peut être construit à très faible coût. La
plupart des diagraphies de résistivité réalisées pour nos études en zone non saturée ont utilisé
cette sonde gonflable, qui a été brevetée par l’IRD. Les tentatives d’intéresser des
constructeurs n’ont cependant pas abouties en raison du très faible marché potentiel.
b) Vers une fiabilisation des imageries de suivi temporel de résistivité
L’idée de ce travail a germé au cours des difficultés successives rencontrées lors des
suivis temporels de résistivité, et des tentatives plus ou moins réussies pour les améliorer. Le
diagnostique fait à l’issu de ces premières études suggère que les artéfacts obtenus sont dus i)
58
à la non-unicité des modèles de résistivité (difficulté intrinsèque à la méthode), ii) à des jeux
de données de terrain incomplets qui renseignent mal les inversions, et iii) aux processus
d’inversion qui sont la plupart du temps conduits avec des facteurs d’amortissement trop
importants (mais qui sont nécessaires pour faire converger les calculs). D’autres auteurs
soulignent aussi ces difficultés. Kemna et al (2004), ou Singha et Gorelick (2006), font état de
leurs difficultés à traduire leurs images de résistivité en terme de concentration en solutés, à
cause i) de la décroissance de sensibilité plus on écarte les électrodes et ii) des effets
d’amortissement consécutifs au facteur de régularisation de l’inversion.
Le travail méthodologique réalisé depuis fin 2007, concrétisé par Rémi Clément (en
thèse au LTHE sous mon co-encadrement), comprend deux phases principales. La première
consiste à traiter le problème de l’effet des infiltrations de surface (géométrie 1D ou
approchante). La seconde s’attaque au problème des infiltrations pour des géométries 2 ou
3D. En effet, des études récentes réalisées lors d’injection de fluides conducteurs dans les
déchets montrent des images géophysiques difficilement interprétables. Pensant qu’il
s’agissait d’artefacts, nous avons cherché a) à les reproduire en modèle numérique, et b) à les
éliminer. De plus, les infiltrations sous les ravines présentent aussi des géométries 2D, voire
3D, qu’il s’agit de fiabiliser : en effet, nous l’avons vu en Inde, ces images peuvent être
distordues (approfondissement démesuré des diminutions de résistivité en zone saturée sous la
ravine par exemple). Notre démarche générale utilise à la fois des modélisations numériques
et des vérifications par des applications de terrain, à l’échelle d’un site industriel, mais aussi à
échelle plus réduite, en conditions semi contrôlées dans une fosse remplie de sable.
Effet des infiltrations superficielles sur le suivi temporel de résistivité
Nous avons repris le problème posé par les suivis temporels en coupe 2D au travers de
la ravine de Katchari au Burkina Faso. En profitant des possibilités récentes offertes par les
logiciels de modélisation, en particulier celles de découpler le calcul d’inversion selon une
ligne continue entre deux groupes de cellules, ligne dont la géométrie peut être déterminée par
d’autres méthodes (sismique ou radar par exemple), nous avons cherché à prendre en compte
la géométrie du front d’infiltration dans l’inversion. Ce front délimite en effet la frontière
entre une zone de résistivité invariante dessous, et une zone d’humectation superficielle au
dessus, où la résistivité chute après les premières pluies. Si l’on prend soin de bien
caractériser la profondeur du front d’infiltration, en particulier en utilisant des dispositifs avec
des électrodes très rapprochées, on arrive à forcer l’inversion vers une solution équivalente
qui élimine presque totalement les augmentations fictives de résistivité en profondeur.
L’exemple de la ravine de Katchari est présenté figure 36.
59
Figure 36. En haut, interprétation
standard du suivi temporel de Katchari.
L’ellipse
rouge
souligne
les
augmentations
considérables
de
résistivité calculées par le modèle (+ de
250%), irréalistes et non repérées par le
suivi neutronique (au centre)
En bas, ré-interprétation en utilisant le
front d’infiltration comme ligne de
découplage dans l’inversion. Les
augmentations se cantonnent dans la
limite de + 20%, considérée comme une
limite de résolution pour ce jeu de
données. D’après Clément et al. (2009)
Le résultat montre que la prise en compte du front d’infiltration comme ligne de
découplage permet d’éliminer presque totalement les artefacts d’augmentation de résistivité
dans la zone superficielle. De plus, les valeurs de résistivité calculées au sein de la frange
d’infiltration sont en meilleur accord avec les valeurs mesurées. Il subsiste tout de même des
zones d’artefact en profondeur. Ce travail est donc perfectible, mais il confirme l’importance
de l’introduction d’informations extérieures dans les modélisations, de manière à restreindre
le domaine des solutions équivalentes. Il ouvre ainsi la voie à des fiabilisations des suivis
temporels en conditions naturelles, en particulier sur les sites sahéliens ou soudaniens, où les
fronts d’infiltration sont marqués par une chute de résistivité très forte lors des premières
pluies de mousson. Il montre surtout que la très proche surface joue un rôle considérable dans
l’inversion, et par voie de conséquence, oblige le géophysicien à acquérir des données avec
des électrodes très rapprochées (quelque décimètres), ce qui est contraignant sur le terrain.
Cas des infiltrations 2 ou 3D
Dans le but d’optimiser la dégradation des déchets ménagers, la visualisation du
contenu en eau et le contrôle de sa quantité dans les massifs de stockage sont des problèmes
pratiques cruciaux pour les exploitants des sites, car la dégradation des déchets est contrôlée
par la teneur en eau. On cherche donc à créer les conditions optimales de teneur en eau au sein
des alvéoles de stockage en ré-injectant des lixiviats par des dispositifs d’injection qu’il faut
optimiser. Pour ce faire, de nombreuses études ont proposé l’utilisation des suivis temporels
de résistivité (voir par exemple Guérin et al, 2004). Une étude récente (Marcoux, thèse, 2008)
a été conduite au LTHE sur l’imagerie géophysique des panaches de lixiviat issus d’injections
provoquées au sein des déchets ménagers. Cette étude fait apparaître des images de résistivité
suspectes, des « halos » d’augmentation de résistivité entourant les panaches d’injection. Un
certain nombre de publications tentent d’expliquer ces images par le fait que le panache de
lixiviat repousse le gaz autour de lui (par exemple Guérin et al., 2004). Cette migration de gaz
60
pourrait en effet désaturer partiellement le milieu, le rendant électriquement plus résistant.
Cependant L. Oxarango, modélisateur des transferts hydriques, émet des doutes sur
l’existence de migrations massives de gaz. Cela nous conduit, Rémi Clément et moi-même, à
évaluer la production possible d’artefact d’imagerie électrique autour de ces panaches
d’injection de lixiviat, qui sont de géométrie 2 ou 3D. Nous employons la démarche
méthodologique suivante : 1) s’appuyer en premier lieu sur des modélisations numériques et
2) conduire une vérification in-situ à l’échelle du massif de déchets (sur le site de Chatuzange,
Véolia Environnement). Grâce aux développements des outils numériques d’inversion des
tomographies de résistivité électrique proposés par T. Günther (LIAG Hanovre) que j’ai
sollicité, nous proposons une boite à outils spéciale « gommage des artefacts ». Quatre
solutions techniques majeures, testées numériquement, et validées sur le terrain pour certaines
d’entre elles, sont proposées:
a) Si possible, on doit renseigner l’inversion par la prise en compte des zones invariantes.
Ce résultat, esquissé déjà sous la microdune par la prise en compte d’un substratum
invariant, se confirme en 2D.
b) Si le site s’y prête, on préférera une acquisition comportant des injections de courant
asymétriques et inversées (de type pole-dipôle direct et inverse par exemple), qui
nécessitent la mise en place d’électrodes à l’infini. Cela permet de créer des jeux de
données comportant des augmentations contradictoires de résistivité apparente dans
l’espace de calcul, et guider ainsi l’inversion vers des solutions équivalentes.
c) Lorsque les deux premières solutions ne peuvent être appliquées, on peut utiliser une
procédure d’inversion qui limite les variations de résistivité entre deux itérations,
technique proposée par Loke (2000) et implantée dans les logiciels de T. Gunther
(2004). Selon ce dernier, cette procédure utilise une contrainte « minimum length »
qui minimise entre deux itérations la différence du vecteur modèle et ce, selon la
norme L2. Cette norme considère le minimum de la somme des carrés des différences
entre les résistivités apparentes mesurées et calculées, utilisée pour analyser la
convergence du calcul d’inversion.
d) Si on s’intéresse à des panaches d’infiltration 3D, on utilisera une géométrie
d’électrodes en étoile au lieu de lignes parallèles classiques, ce qui a pour effet i) de
renseigner d’autres secteurs du terrain de façon plus homogène et ii) de croiser les
lignes de courant d’une ligne à l’autre si nécessaire.
e) En cas d’infiltrations superficielles enfin, la boite à outils comporte le découplage par
la connaissance du front d’infiltration, déjà décrit plus haut, si des pluies ont eu lieu
par exemple entre deux acquisitions.
Pour illustrer l’efficacité d’un de ces outils, je reprends sur la figure 37 un exemple tiré
de la récente publication de Clément et al. (2010) qui montre l’amélioration des images
géophysiques obtenues pour un scénario d’injection simulé numériquement, lorsqu’on utilise
un dispositif géophysique asymétrique.
61
Figure 37. Effet de la symétrie des
dispositifs d’électrodes sur l’imagerie de
suivi temporel de résistivité. L’objet
simulé ici est une infiltration de lixiviat
2D réalisée dans un massif de déchets à
partir d’une tranchée placée en surface.
Le panache, de taille réduite, est souligné
par un rectangle pointillé.
Seul le dispositif asymétrique permet de
reconstruire l’image d’infiltration, dans
les limites de résolution de la méthode,
qui a toujours tendance à approfondir les
panaches sous la limite réelle de
pénétration. D’après Clément et al.,
(2010, in press)
On voit sur cette image qu’en utilisant des dispositifs d’injection symétriques comme
le Wenner ou le dipôle-dipôle, les zones d’augmentation de résistivité peuvent apparaître
autour du bulbe d’injection (en rouge et noir sur l’image). Ces zones, interprétées par les
exploitants des sites de stockage comme des migrations de gaz repoussés vers l’extérieur lors
de l’injection, disparaissent complètement avec l’utilisation d’un dispositif asymétrique, le
pôle-dipôle. Notre étude montre aussi que si des migrations de gaz augmentant la résistivité se
produisent réellement, elles sont correctement vues par nos méthodes d’inversion.
Les améliorations obtenues ces deux dernières années, si elles restent pour l’instant
appliquées à des cas précis, sont extrêmement prometteuses pour relancer l’utilisation des
techniques d’imagerie par suivi temporel de résistivité. Je compte à l’avenir reprendre certains
jeux de données grâce de ces améliorations.
4. Perspectives
Introduction
Au cours des dernières années, je me suis progressivement spécialisé dans la recherche
sur les aquifères en zone de socle et sur les processus hydrologiques associés, dans des
régions à climat semi-aride ou soudanien. Je me suis aussi intéressé à des problématiques
méthodologiques, en particulier sur l’emploi des suivis temporels de résistivité et leur
amélioration lors des interprétations des tomographies de résistivité.
Je compte, dans les années à venir, poursuivre mes études des aquifères de socle et des
processus de recharge, et m’impliquer dans l’amélioration des outils géophysiques. Ces
recherches auront pour finalité non seulement de mieux comprendre ces systèmes d’aquifères
complexes mais aussi d’en améliorer les modèles conceptuels, par le croisement des outils de
la géophysique avec ceux des autres disciplines. Cette approche multidisciplinaire devrait
permettre à terme une meilleure paramétrisation des modèles numériques hydro(géo)logiques
et améliorer leur fiabilité prédictive.
Pour mener à bien ces recherches, je vais m’appuyer sur l’environnement scientifique
procuré par l’ORE « Amma Catch » autour des bassins versants expérimentaux de cet
62
observatoire. Je compte renforcer l’implication de l’hydrogéophysique dans le futur projet
régional faisant suite au projet AMMA (Analyse Multidisciplinaire de la Mousson Africaine).
Au sein de la future équipe HyBiS (Hydrogéophysique et Bilans Spatialisés) du LTHE, je
devrais être affecté fin 2010 au Bénin avec mon collègue J.M Vouillamoz pour travailler au
sein des services de la Direction Générale de l’Eau (DG-Eau) de Cotonou, et avec le
Laboratoire d’Hydrologie Appliquée (LHA) de l’Université d’Abomey-Calavi.
4.1 Projet de recherche au Bénin
Dans le nord du Bénin, les questions hydrologiques restant en suspens ont été bien
identifiées par nos collègues hydrologues du LTHE, de HSM et de la DG-Eau. La
géophysique pourrait aider à répondre aux questions suivantes, posées à l’échelle du bassin
versant:
¾
¾
Quel est le cycle de l’eau dans les bas fonds herbeux, éléments clefs du cycle
hydrologique ? Quels sont les échanges hydriques entre ces bas-fonds et les versants ?
Quels sont les lieux de recharge directe, et dépendent-ils de la végétation ?
Comment faire la partition du bilan de l’eau souterraine entre transpiration et recharge ?
En effet, même si les dernières études de Guyot (thèse LTHE, 2010) concluent à une
part importante du bilan due à la transpiration, le cycle d’évapotranspiration reste mal
compris à certaines périodes de l’année.
En plus de ces études sur les processus hydrologiques, notre équipe compte s’intéresser
aussi à la spatialisation locale et régionale des ressources en eau souterraine. Pour des pays
comme le Bénin où tombent pourtant des quantités de pluie importantes, le recours à l’eau
souterraine est de plus en plus fréquent, pour l’alimentation en eau potable des centres urbains
en particulier. Cela nécessite de planifier l’utilisation de la ressource, et donc de connaître les
volumes d’eau qui pourraient être utilisés. J.M. Vouillamoz et moi proposons de réaliser cette
étude au Bénin, et éventuellement sur d’autres zones de socle représentatives de la région.
Pour répondre à ces questions, nous comptons mener des actions de recherche selon 3 axes:
a) La spatialisation des ressources en eau à l’échelle régionale par une approche couplée
géologie et géophysique au sol, voire en utilisant aussi des données existantes de
géophysique aéroportée.
b) L’étude des processus de recharge à l’échelle du bassin versant, en évaluant les
potentialités des méthodes électromagnétiques.
c) La quantification du bilan de l’eau souterraine à l’échelle locale par un couplage accru
entre méthodes, en profitant notamment des développements de la méthode RMP en
zone non-saturée, et de la présence au Bénin du gravimètre du projet «Ghyraf».
a) Spatialiser les ressources en eau en zone de socle à l’échelle régionale
Le constat que je dresse à l’issu de mes recherches est qu’il est possible de spatialiser,
de la parcelle au versant, certains processus d’infiltration et de recharge des aquifères. De
63
plus, il est possible d’approcher, grâce notamment à la RMP, la caractérisation des propriétés
hydrodynamiques des régoliths. Pour l’instant, la spatialisation des ressources en eau
souterraine à l’échelle du bassin versant, et plus encore, à l’échelle régionale, n’est pas
réalisée. L’objectif étant de caractériser les capacités de stockage, on s’intéresse donc
essentiellement au réservoir de l’aquifère, le regolith. Notre approche impliquera l’utilisation
de la RMP, dont on a reconnu les avantages en terme de quantification, et qui sera implantée
grâce à une approche couplée résistivité/géologie. Le Bénin se prête relativement bien à ce
projet, car il présente des catégories de roches de socle très variées (figure 38). Le nord du
Bénin procure aussi un autre avantage méthodologique, car cette région est située en zone
climatique soudano-guinéenne, par conséquent les regoliths sont saturés une partie de l’année
seulement, et désaturés ensuite. C’est idéal pour étudier la réponse de résistivité et de RMP
alternativement en situation saturée puis non saturée, ce qui augmente les contrastes
géophysiques.
Figure 38. Carte géologique synthétique du Bénin.
En rouge, j’ai indiqué les principales unités de
roches de socle présentes sur le bassin de la
Donga, étudié dans le cadre de l’ORE AMMACatch. Les quartzites et les micaschistes ont déjà
été partiellement caractérisés à petite échelle sur
le bassin versant d’Ara (voir figure 16).
Carte reproduite d’après un document du
gouvernement du Bénin (« Orientations et plan
d’actions stratégiques de développement du
secteur minier en République de Bénin, Cotonou,
2007)
Pour commencer, un certain nombre de sites de socle seront choisis dans les
principales unités géologiques sur le bassin versant de la Donga (ORE AMMA-Catch). Dans
un premier temps, une caractérisation hydrogéophysique sera faite à l’échelle du site de
forage. Il faudra guider l’implantation de la RMP avec des méthodes de résistivité, car nous
avons vu que la résistivité permet une spatialisation rapide et offre des possibilités
intéressantes de changement d’échelle. Les sites choisis comprendront aussi des forages
existants, dotés de mesures d’essais de pompages, de manière à croiser les interprétations des
paramètres hydriques issues des méthodes géophysiques et hydrogéologiques.
64
Dans un deuxième temps, je compte étudier la possibilité de spatialiser régionalement
les ressources en eaux souterraines grâce aux données électromagnétiques aéroportées, les
seules capables de couvrir l’échelle de la région. De nombreux pays africains se sont dotés de
cartes électromagnétiques aéroportées en zone de socle (Burkina Faso, Niger, Ghana par
exemple). Réalisées à des fins d’exploration minière, l’idée de les réutiliser à des fins
hydrogéologiques n’est pas nouvelle : Paterson et Bosschart proposent, en 1987 déjà,
d’utiliser ces cartes pour définir des sites de prospection au sol. Plus récemment, Bromlet et al
(1994) utilisent cette approche au Botswana. Pour spatialiser les ressources en eau
souterraine, Paterson et Bosschart conseillent de procéder à la vérification, sur le terrain, des
« cibles » hydrogéologiques identifiées à partir d’une première phase de couplage entre carte
géophysique aéroportée et autres « couches » d’informations (géologie, végétation, ou
topographie par exemple). Ils préconisent d’utiliser des forages et des essais de pompage pour
tenter ensuite une spatialisation régionale des ressources. Je pense que cette phase de contrôle
sur le terrain pourrait avantageusement utiliser la RMP plutôt que des forages, et à des coûts
nettement moindres. Mais avant de se lancer dans un projet de cartographie qui utilise les
cartes issues des prospections géophysiques aéroportées, il conviendra de s’assurer, au sol, de
la faisabilité de cette approche. Pour cela, il faudra :
i) S’assurer que les cartes établies à partir de profil de résistivité mesuré en altitude
représentent convenablement les distributions de résistivité au sol. Pour cela, des
modélisations numériques construites avec notre connaissance du terrain devraient
suffire.
ii) Evaluer si les variations de résistivité pourraient traduire des propriétés
hydrodynamiques différentes. Ce n’est pas forcement le cas, nous le voyons déjà sur le
bassin d’Ara au Bénin : des formations de même résistivité peuvent avoir des
signatures RMP très différentes (entre quartzites et micaschistes par exemple). La
caractérisation par RMP au sol sera donc impérative.
iii) Mieux connaître la réponse RMP de ces formations altérées, et réaliser un travail de
calibration de la méthode sur des sites choisis.
b) Spatialiser la recharge à l’échelle du bassin versant
Si les recharges localisées peuvent être spatialisées localement à l’aide des mesures
géophysiques au travers d’une ravine par exemple, la question se pose de la représentativité
de ces mesures à l’échelle du bassin versant. Ces changements d’échelle impliquent
l’utilisation des méthodes électromagnétiques, dont les potentialités en mode de suivi
temporel restent encore largement inexploitées. J’ai déjà testé en 2003 des suivis temporels
par méthodes électromagnétiques fréquentielles, sur le bassin versant d’Ara au Bénin. Les
résultats, très préliminaires, sont montrés sur la figure 39 sur une petite zone test de 1km².
65
Figure 39. Suivi temporel électromagnétique EM34 réalisé sur 1km² au centre du bassin versant d’Ara au Bénin
entre mai et août 2003 (données DESS de M. Wubda, 2003). Les données intègrent une tranche de profondeur
jusqu’à 20 mètres.
Sur ces cartes, on peut voir que seule la zone au sud de la prospection montre après
deux mois de saison des pluies une extension d’une anomalie conductrice, ce qui pourrait
indiquer une recharge dans cette zone. L’anomalie conductrice à l’ouest (en bleu sombre sur
la figure) est située dans un bas fond herbeux. Cette anomalie ne change pas significativement
de forme, indiquant que les sols sont toujours saturés et que les eaux de pluie nouvelles ne se
marquent pas par des différences notables de résistivité électrique. Nous montrons dans cette
étude qu’il serait possible de spatialiser des infiltrations et des recharges si des contrastes de
résistivité sont présents. Poursuivre ces recherches nécessitera un travail méthodologique sur
la calibration des mesures, la réalisation de mesures de vérification in-situ dans les sols, et la
prise en compte de la réponse des méthodes EM à des structures 3D. Pour ce dernier point, je
compte engager des collaborations avec les hydrogéophysiciens de l’UMR Sisyphe, qui
possèdent des codes de calcul appropriés.
c) Mieux quantifier le bilan de l’eau souterraine à l’échelle du site
Sur le bassin versant d’Ara, la poursuite des observations hydrologiques dans les
prochaines années procure l’avantage de pouvoir engager des actions géophysiques
complémentaires, dont les résultats pourront être croisés avec d’autres méthodes. En
particulier, des questions méthodologiques sur la RMP restent ouvertes. Je propose de mettre
en œuvre 2 actions de recherche complémentaires destinées à mieux comprendre le signal de
RMP produit par ces regoliths, avec pour objectif de connaître leur porosité de drainage et
contribuer à mieux comprendre le bilan de l’eau. Deux échelles vont être étudiées :
Echelle de l’échantillon
Si la teneur en eau mesurée par la RMP peut, en première analyse, être considérée
comme maximisant la porosité cinématique d’un aquifère (Vouillamoz et al, 2005), il reste à
vérifier cette assertion dans le contexte des altérites de roches métamorphiques du site d’Ara.
Nos prospections RMP montrent que ces altérites de micaschistes ont une teneur en eau RMP
d’environ 3%. Or, l’analyse par porosimétrie mercure de cinq échantillons d’altérite prélevés
66
en puits sous un sondage RMP (figure 40) montre que la porosité utile pour restituer de l’eau
à la végétation par transpiration est bien plus élevée, de 20 à plus de 40% du volume total.
Ainsi, l’eau « vue » par la RMP n’est qu’une fraction de l’eau potentiellement utilisable par la
végétation dans ces sols. Finalement, quelle fraction de la porosité mesure la RMP, avec
l’appareillage dont nous disposons actuellement?
Figure 40. Courbes de succion déduites
d’une analyse de porosimétrie mercure
pour 5 échantillons non remaniés
d’altérite de micaschistes prélevés à cinq
profondeurs dans un puits sur un versant
au Bénin (analyses de J.F Daian et H.
Denis, LTHE, non publiées). Le sondage
de résonance magnétique effectué autour
de ce puits montre une porosité RMP de
l’ordre de 3% (volumes hachurés sur la
figure). On note que la majorité du
volume d’eau (plus de 30%) n’est pas
« vue » par la RMP. Cette eau, non
disponible pour les pompages, le reste
cependant pour la végétation.
Pour répondre à cette question, je compte analyser en laboratoire la réponse RMP
d’échantillons d’altérites qui seront prélevés dans des fosses pédologiques réalisées durant la
saison sèche au Bénin. Ces mesures peuvent désormais être faites grâce aux très récents
développements instrumentaux au LTHE (A. Legchenko et H. Guyard). Ainsi, il devrait être
possible de comparer les réponses de l’instrument RMP de terrain aux signaux obtenus sur
échantillons, ceux-ci pouvant ensuite être analysés par d’autres méthodes classiques, telle que
la porosimétrie.
Echelle du site
La future implantation d’un gravimètre supra conducteur sur le site d’Ara au Bénin
(programme ANR « Ghyraf », Hinderer et al., 2009) sera une occasion de coupler les
méthodes RMP et électriques avec la méthode gravimétrique, avec l’objectif de caractériser la
variation du stock d’eau durant la mousson. Pour évaluer la faisabilité de ce couplage de
méthodes, j’ai réalisé une première modélisation numérique. Celle-ci vise à reproduire les
variations de signal gravimétrique que l’on obtiendrait par vidange de la nappe uniquement.
Pour cela, il faut connaître la porosité de drainage de l’altérite dans la zone de battement de la
nappe. En faisant l’hypothèse, encore critiquable, que la teneur en eau RMP maximise la
valeur de porosité de drainage, le modèle numérique gravimétrique peut être reconstruit. J’ai
rajouté une information sur le pendage faible des formations géologiques, qui reste encore à
vérifier. Le résultat, très préliminaire, est présenté sur la figure 41.
67
Figure 41. Modélisation 3D RMP en
haut, et traduction de la géométrie RMP
en terme d’anomalie gravimétrique en
bas. On prend en compte la variation du
niveau des nappes entre deux périodes
contrastée.
Le signal synthétique
calculé par le modèle gravi est de
l’ordre de 10 µgal ; alors que sur la
même durée, le gravimètre mesurait
13µgal.
La modélisation gravimétrique synthétique nous apprend que le signal théorique est de
l’ordre de 10 µgal. Le gravimètre mesure 13 µgal sur le terrain pour la même période. La
question est alors posée : la RMP ne minimise-t-elle pas la porosité de drainage,
contrairement à notre hypothèse de départ? Quelle est la part du stock en zone non saturée qui
pourrait contribuer au signal gravimétrique ? Cette modélisation nous apprend aussi
l’influence du pendage : en déplaçant virtuellement les formations de 25 mètres sous le
gravimètre, on peut changer le signal gravimétrique de 2 µgal. Ces études devront être
affinées par une modélisation gravimétrique plus poussée ainsi que par des mesures
complémentaires de terrain. Nous pourrons aussi tenter de caractériser les variations
d’humidité dans les sols non saturés en réalisant des mesures RMP à petite échelle dans les
sols, de manière à les comparer aux mesures d’humidité faites avec d’autres méthodes.
4.2 Autres projets
Dans le cadre de projets existants, d’autres actions de recherche sont actuellement
entreprises. Elles portent i) sur la spatialisation des infiltrations lors d’une expérience
d’infiltrométrie (projet « EC2CO Ondine »), ii) sur l’étude de la résistivité comme marqueur
éventuel de la dégradation des déchets anciens (projet « Paraphyme »). Les premiers résultats
réclament d’approfondir un certain nombre de questions méthodologiques liées aux outils
géophysiques.
a) Spatialiser la dynamique des infiltrations par suivi temporel de résistivité
Rémi Clément et moi avons commencé une étude de tomographie de résistivité 3D
sous un infiltromètre dans le cadre du programme EC2CO « Ondine » (coord. O. Ribolzi).
L’objectif est d’obtenir non destructivement des paramètres dynamiques utiles à
l’interprétation de l’infiltration, comme la forme et la vitesse de propagation du front
68
d’infiltration. Cette expérience est un demi succès : le protocole de mesure géophysique
permet effectivement de réaliser un film de la propagation de l’infiltration grâce à l’obtention
d’environ 70 images successives du bulbe sous l’infiltromètre (thèse de R. Clément, à venir).
En revanche, la comparaison avec les variations de tension a échoué à cause de problèmes
instrumentaux. Nous comptons reprendre cette expérience inachevée, mais prometteuse. La
possibilité d’obtenir des données de résistivité à haute cadence, permet de dégager de
nombreuses perspectives pour l’analyse des courbes de variations temporelles de résistivité :
la décroissance dans le temps présente t-elle la même forme pour un sol sableux que pour des
sols sablo-argileux ? Telle pourrait être une des questions abordées dans le futur. Pour cela, je
propose d’approfondir les relations entre résistivité et tension. Mais, pour avancer dans ce
domaine, il faudra imaginer des systèmes couplant, aux mêmes endroits exactement, des
mesures de tension, de résistivité et de température. Cette piste de recherche nécessitera donc
des adaptations de capteurs existants.
b) La résistivité est-elle un marqueur de la dégradation des déchets?
En 2009, à l’initiative de J. P. Gourc de l’équipe Transpore du LTHE, nous avons
obtenus des financements pour un projet d’étude sur la dégradation des déchets : La question
cruciale pour un exploitant d’une décharge est de connaître jusqu’à quand les déchets sont
totalement dégradés. Pensant que la résistivité pouvait être un marqueur de choix pour évaluer
cette dégradation, nous avons couplé des sondages TDEM à de l’imagerie électrique sur un site
présentant des âges de déchets différents. Les premiers résultats obtenus dégagent une question
méthodologique importante pour la géophysique : la comparaison des mesures TDEM et
électriques montrent en effet que les valeurs de résistivité obtenues en courant continu sont 5 à
10 fois supérieures à celle obtenues en TDEM, comme l’illustre la figure 42.
Figure 42. Coupe de résistivité au travers d‘un site de stockage de déchets composé de 5 cellules juxtaposées.
Comparaison entre la coupe de résistivité obtenue par TDEM (en haut) et celle obtenue par mesures à courant
continu classique (en bas), sur laquelle est reportée les limites observées par TDEM.
69
Cette discordance, déjà remarquée par notre équipe en 2008 (Descloitres et al.,
Conférence EAGE, 2008), n’a jamais été remarquée dans la littérature, et doit être expliquée.
Une publication de Zhdanov et Pavlov (2001) montrent que des effets magnétiques pourraient
conduire à distordre des courbes de sondages TDEM. Or, les déchets étudiés présentent des en
surface des anomalies de champ total significatives (plus de 1000 nT). Autre possibilité, des
anisotropies dans la distribution de résistivité des déchets pourraient aussi être à l’origine de
cette discordance. Je compte réaliser des modélisations de la réponse TDEM en présence de
terrains magnétiques et évaluer les effets 3D grâce aux nouveaux outils de modélisation qui se
développent actuellement.
70
Conclusion
Les études menées permettent des avancées dans la connaissance des aquifères de
socle et sédimentaires, situés dans des zones climatiques semi-arides ou plus humides, ainsi
que de certains processus hydrologiques associés. Les méthodes géophysiques, capables de
spatialiser, quantifier et suivre dans le temps les variations des paramètres géophysiques, à
différentes échelles, apportent une contribution aux questions posées i) sur le rôle des ravines
de versant dans la recharge des aquifères, qui se distingue selon le contexte géologique,
particulièrement au Sahel, ii) sur l’infiltration dans les sols sableux, et plus largement iii) sur
les propriétés des altérites, réservoirs des aquifères de socle. L’établissement des modèles
conceptuels de fonctionnement est facilité par la spatialisation des formations argileuses,
notamment pour les systèmes aquifères sédimentaires.
D’un point de vue méthodologique, j’ai montré que l’étude des phénomènes de
recharge des aquifères et d’infiltration dans les sols sableux peut être entreprise par le suivi
temporel de résistivité, s’il existe des contrastes suffisants. Nos études récentes permettent de
fiabiliser ces suivis temporels, longtemps handicapés par la production d’artéfacts
indésirables, qui compromettaient une interprétation hydrologique fiable. Nous avons appris
que le succès d’un suivi temporel de résistivité dépend beaucoup du soin apporté i) à
l’acquisition de données produites par des dispositifs géométriques adaptés, ii) aux choix des
paramètres d’inversion et iii) à l’incorporation d’informations extérieures dans la
modélisation. L’utilisation de la modélisation numérique reste incontournable pour préparer
l’application sur le terrain et pour conforter les interprétations. La difficulté à obtenir des
relations expérimentales liant la résistivité aux variables hydrologiques d’intérêt a été
soulignée, notamment pour les sols et les formations à dominante argileuse. L’utilisation de la
RMP apporte la possibilité de discriminer la nature des formations en présence à l’échelle du
sondage en quantifiant leurs paramètres hydrologiques, et offre des potentialités intéressantes
en suivi temporel pour quantifier les volumes impliqués dans les recharges localisées sous les
axes drainants.
Les perspectives dégagées par mes travaux concernent principalement l’évaluation des
ressources en eau souterraine des zones de socle, plus difficiles à spatialiser que celles des
aquifères sédimentaires, et l’étude de leur recharge. Elles s’appuieront sur des suivis
temporels de résistivité, avec notamment l’exploration des potentialités des méthodes
électromagnétiques dans ce domaine, et sur l’utilisation de la RMP. Je compte mener ces
recherches au sein de la nouvelle équipe HyBis du LTHE, dans le cadre d’un projet au Bénin,
pays qui présente une grande variété de roches de socle. Les applications de ces recherches
concerneront, outre des apports méthodologiques pour la prospection géophysique des
aquifères et pour l’implantation des forages, une contribution aux modélisations
hydro(géo)logiques. D’autres projets, plus méthodologiques, seront poursuivis sur des
thématiques variées, allant de l’étude d’infiltration en situation semi contrôlée à l’étude de la
dégradation des déchets ménagers, par suivi temporel de résistivité.
71
Références bibliographiques
Albouy, Y., Andrieux, P., Rakotondrasoa, G., Ritz, M., Descloitres, M., Join, J. L., Rasolomanana, E.,
2001. Mapping Coastal Aquifers by Joint Inversion of DC and TEM Soundings. Three Case
Histories. Groundwater, Vol. 39, n°1,pp 87-97.
Archie E. 1942. The Electrical Resistivity Log as an Aid in Determining Some Reservoir
Characteristics. Technical Publication 1422, Petroleum Technology, American Institute of Mining
and Metallurgical Engineer, New York, USA; 8.
Baltassat J.M., Krishnamurthy, N.S., Girard, J.F., Dutta S., Dewandel, B., Chandra S., Descloitres, M.,
Legchenko, A., Robain, H. 2006. Geophysical characterisation of weathered granite aquifer in the
Hyderabad region, Andra Pradesh, India Cefipra project final report, BRGM report RP-54538, 117
p., 57 fig., 11 tabl., 17 appendices
Barbiéro L., Parate, H. R., Descloitres, M., Bost A., Furian S., Mohan Kumar M.S, Kumar C., Braun J.
J., 2007. Using a structural approach to identify relationships between soil and erosion in a semihumid forested area, South India. CATENA, vol. 70 , pp 313–329
Beck, M., Girardet, D., Chapellier, D., Descloitres, M., 2001. Diagraphies électriques pour
l’optimisation de l’hydrofracturation en zone de socle. Premiers résultats au Burkina Faso. Actes
du 3ème Colloque GEOFCAN (Géophysique des sols et des Formations Superficielles), 25-26
septembre, Orléans, France, pp 55-59.
Boucher M., Favreau, G. Descloitres, M., Vouillamoz, J. M., Massuel, S., Nazoumou, Y., Legchenko,
A., 2009. Contribution of geophysical surveys to groundwater modelling of a porous aquifer in
semiarid Niger: An overview. Comptes Rendus Geoscience, 341, 800-809.
Braun, J-J., Descloitres, M., Riotte, J., Fleury, S., Barbiero, L., Boeglin, J-L., Violette, A., Lacarce, E.,
Ruiz, L., Sekhar, M., Mohan Kumar, M.S., Subramanian, S., Dupre, B., 2009. Regolith mass
balance inferred from combined mineralogical, geochemical and geophysical studies: Mule Hole
gneissic watershed, South India, Geochimica et Cosmochimica Acta, doi: 10.1016
/j.gca.2008.11.013.
Bromley, J., Mannstrom, B., Nisca, D., Jamtlid, A., 1994. Airborne geophysics: Application to a
grunwater Study in Botswana. Groundwater, Vol 32, N°1, jan-feb 1994.
Cagniard, L., 1959. Abaque pour sondages électriques sur glace Annales de géophysique vol 15., N° 4,
pp 561-563
Chalikakis, K., Ryom Nielsen M., Legchenko, A., Feldberg Hagensen, T. 2009. Investigation of
sedimentary aquifers in Denmark using the magnetic resonance sounding method (MRS) Comptes
Rendus Géoscience, Vol. 341, N°10-11, pp 918-927, Doi : 10.1016/j.crte.2009.07.007
Chilton P J and Foster S S D 1995. Hydrogeological characterisation and water-supply potential of
basement aquifers in tropical Africa. Hydrogeology Journal 3 (1), 36-49.
Clément, R., Descloitres, M., Günther, T., Morra, C., Oxarango, L. Artefact removal in time-lapse
ERT interpretation. Application to leachate injection experiment in landfills., 2010. Waste
management, in press.
Clément, R., Descloitres, M., Günther, T., Ribolzi, Legchenko, A. (2009), Influence of shallow
infiltration on time-lapse ERT. Experience of advanced interpretation. Comptes Rendus
Geosciences, 341, pp 886-898.
Coudrain, A., Loubet, M., Guérin, R., Descloitres, M., Talbi, A., Quintanilla, J, Ledoux, E. 2000.
Reconstitution des écoulements souterrains de l’Altiplano bolivien pendant l’Holocène par
modélisation hydrogéochimique. Colloque PNRH 2000, Toulouse, 16-17 mai, pp 217-223.
Coudrain, A. Amal Talbi, Emmanuel Ledoux, Loubet M, Vacher J, Ramirez E. 2001. Subsurface
transfer to chloride after a lake retreat in the Central Andes. Ground Water, Vol. 39, No5, p. 751759.
Courteaud, M., Ritz, M., Descloitres, M., Robineau, B., Coudray, J., 1996. Cartographie
audiomagnétotellurique du biseau salé dans la zone côtière du Piton de la Fournaise (Ile de la
Réunion). C.R. Acad. Sci. Paris, t.322, série II a, pp 93 -100.
Descloitres, M., Ritz, M., Courteaud, M., Robineau, B., 1997. Electrical structure beneath the
collapsed eastern flank of Fournaise volcano, Réunion Island: implication to the quest for
groundwater. Water Resources Research, 33(1), pp 13-19.
Descloitres, M., Robain, H., Dabas, M., Camerlynck, C., Albouy, Y., 1997. Apport des imageries
électriques et radar à la reconnaissance des couvertures d’altération. Bassin versant de Nsimi,
72
Cameroun. Colloque GEOFCAN “ Géophysique des sols et des formations superficielles ”,
Bondy, 11-12 septembre.
Descloitres, M., Guérin, R., Ramirez, E., Gallaire, R., Ribstein, P., Valla, F., 1999. Détermination de
l’épaisseur des glaciers de Sarenne (Alpes) et de Chacaltaya (Bolivie) par prospection radar à 50
MHz. La Houille Blanche, Société Hydrotechnique de France, n°5, pp 29-33.
Descloitres, M., Guérin, R., Albouy, A., Tabbagh, A., Ritz, M., 2000. Improvement in TDEM
sounding interpretation in presence of induced polarization. A case study in resistive rocks of
Fogo volcano, Cape Verde Islands. Journal of Applied Geophysics, 45, pp 1-18.
Descloitres, M., Ribolzi, O., Le Troquer, Y., 2001. Variations saisonnières de la résistivité des sols
d’une ravine sur un versant sahélien. II. Interprétation des panneaux électriques 2D. Actes du 3ème
Colloque GEOFCAN (Géophysique des sols et des Formations Superficielles), 25-26 septembre,
Orléans, France, pp 31-34.
Descloitres, M., Ribolzi, O., Le Troquer, Y, 2003. Study of infiltration in a gully erosion sahelian area
using time-lapse electrical resistivity mapping. CATENA 53, pp 229-253.
Descloitres, M., Ribolzi, O., Le Troquer, Y., Thiébaux, J. P., 2006. Spatializing Water Tension in
Heterogeneous Sandy Soils with Surface ERT During Rain-Evaporation Cycles. 12th European
Meeting of EAGE “Near Surface 2006”, Helsinki, Finland 4 - 6 September 2006. Extended
abstract, Paper B039.
Descloitres, M., Séguis, L., Wubda, M., Legchenko, A., 2007. Discrimination of rocks with different
hydrodynamic properties using MRS, EM and resistivity methods. EAGE International
Conference “Near Surface 2007”, Istambul, 3-5 sept. 2007.
Descloitres, M., Laurent, J. P., Morra, C., Clément, R., Oxarango, L., Gourc, J. P., 2008. Monitoring
resistivity in non-hazardous waste landfill using Time Domain Electromagnetism (Drôme,
France). International Conference EAGE “Near Surface 2008”, Karkow.
Descloitres, M., Ruiz, L., Sekhar, M., Legchenko, A., Braun, J. J., Mohan Kumar, M.S., Subramanian,
S., 2008. Characterization of seasonal local recharge using Electrical Resistivity Tomography and
Magnetic Resonance Sounding. Hydrological Processes, Vol. 22, pp 384 – 394.
Descloitres, M., Ribolzi, O., Le Troquer, Y., Thiebaux, J.P. 2008, Study of water tension differences
in heterogeneous sandy soils using surface ERT.. Journal of Applied Geophysics, Vol 64/3-4, pp
83-98 DOI information: 10.1016/j.jappgeo.2008.12.007.
Descloitres, M., Séguis, L Legchenko, A., Wubda, M. Hydrogeophysical identification of
representative geological units as the first step of hydrodynamic modelling at a watershed scale in
metamorphic context (North Bénin), Near Surface Geophysics, submitted december 2009
Descloitres, M., Legchenko, A., Vincent, C., Guyard, H., Chalikakis, K., 2010. Recherche d’eau
liquide dans le glacier de Tête Rousse par sondage de Résonance Magnétique des Protons.
Rapport de mission, Projet Pôle « TUNES », LTHE-LGGE, Université Joseph Fourier, Grenoble
I, 38 pages, 9 figures, 2 tableaux, 4 annexes.
Descroix L, Mahe G, Lebel T, Favreau G, Galle S, Gautier E, Olivry J.C., Albergel J, Amogu O,
Cappelaere B, Dessouassi R, Diedhiou A, Le Breton E, Mamadou I, Sighomnou D., 2009. Spatiotemporal variability of hydrological regimes around the boundaries between Sahelian and
Sudanian areas of West Africa: A synthesis Journal of Hydrology, Vol. 375, Issue: 1-2, pp 90-102
Esteves, M., Lapetite, J. M., 2003. A multi-scale approach of runoff generation in a Sahelian gully
catchment: a case study in Niger. Catena, Vol. 50 (2003), pp. 255– 271
Giroux, B., Chouteau, M., Descloitres, M., Ritz, M., 1997. Use of the magnetotelluric method in the
study of the deep Maestrichtian aquifer in Senegal. Journal of Applied Geophysics, 38, pp 77-96.
Gouvernement du Bénin, 2007. Document « Orientations et plan d’actions stratégiques de
développement du secteur minier en République de Bénin », Cotonou.
Guérin R., Descloitres, M., Coudrain-Ribstein A., Talbi A., Gallaire R., 2001. Geophysical surveys for
identifying saline groundwater in the semi-arid region of the central Altiplano, Bolivia.
Hydrological Processes, 15, 17, pp 3287-3301.
Guérin, R., Munoz, M.L., Aran, C., Laperrelle, C., Hidra, M., Drouart, E., Grellier, S., 2004. Leachate
recirculation: moisture content assessment by means of a geophysical technique. Waste
Management 24 (8), 785–794.
Guyot, A., Cohard, J. M., Anquetin, S., Galle, S., Lloyd, C., 2009. Combined analysis of energy and
water balances to estimate latent heat flux of a sudanian small catchment. Journal of Hydrology,
Vol. 375, Issue: 1-2, Special Issue, pp. 227-240.
73
Hinderer J., de Linage C., Boy J.P., Gegout P., Masson F., Rogister Y., Amalvict M., Pfeffer J., Littel
F., Luck B., Bayer R., Champollion C., Collard P., Le Moigne N., Diament M., Deroussi S., de
Viron O., Biancale R., Lemoine J.M., Bonvalot S., Gabalda G., Bock O., Genthon P., Boucher M.,
Favreau G., Séguis L., Descloitres M., Galle S., 2009. The GHYRAF (Gravity and Hydrology in
Africa) experiment: description and first results. Journal of Geodynamics 48, Issues 3-5, pp: 172181.
Hubbard, S.S., et Rubin, Y., 2006. Introduction to Hydrogeophysics. In Hydrogeophysics, Rubin, Y.,
and Hubbard, S.S., Ed., Springer, pp. 129-156.
Jones, M.J. 1985. The weathered zone aquifers of the Basement Complex areas of Africa, Q. J. ENG.
Geol., London, V. 18, pp 35-46.
Kamagate, B., Favreau, G., Séguis, L., Seidel, J. L., Descloitres, M., Affaton, P, 2007. Hydrological
processes and water balance of a tropical crystalline bedrock catchment in Benin (Donga, upper
Ouémé River). Comptes Rendus Geosciences, Volume 339, Issue 6, pp. 418-429
Keller, G.V., 1988. Rock and mineral properties. In Electromagnetic methods in Applied Geophysics,
vol. 1, Nabighian, M., Editor. Society of Exploration Geophysicists.
Kemna, A., Vanderborght, J., Hardelauf, H., Veerecken, H., 2004. Quantitative imaging of 3-D solute
transport using 2-D time-lapse ERT: A synthetic feasability study. Proc. Symp. Applications of
Geophysics to Engineering and Environmental Problems, Environ. Eng. Geophys. Soc., 342-353.
Knight, R., 1991. Hysteresis in the electrical resistivity of partially saturated sandstones Geophysics,
Vol. 56, N°, 12, pp 2139-2147.
Lachassagne, P., et Wyns, R., L’eau des socles, revue « Géosciences », doc. BRGM numéro 2,
septembre 2005.
Lamy, V., 1995. Application du radar géologique à l’étude des formations superficielles en régions
sahéliennes et méditerranéennes. Rapport de stage de DESS, ORSTOM Dakar, Université de Paris
6, Laboratoire de géophysique appliquée.
Leduc, C., Favreau, G., Shroeter, P., 2001. Long term rise in a Sahelian water-table: the continental
terminal in South-West Niger. Journal of Hydrology 243, 43–54.
Legchenko A, Valla P. 2002a. A review of the basic principles for proton magnetic resonance
sounding measurements. Journal of Applied Geophysics 50: 3–19.
Legchenko A, Baltassat J-M, Beauce A, Bernard J. 2002b. Nuclear magnetic resonance as a
geophysical tool for hydrogeologists. Journal of Applied Geophysics 50: 21–46.
Legchenko, A., Descloitres, M., Bost, A., Ruiz L., Reddy, M., Girard, J-F., Sekhar, M., Mohan
Kumar, M.S., Braun, J. J., 2006 Resolution of MR Soundings applied to the characterization of
hard rock aquifers. Groundwater 44(4), pp 547-554.
Le Lay, M.; Saulnier, G. M.; Galle, S.; Seguis, L.; Metadier, M.; Peugeot, C., 2008. Model
representation of the Sudanian hydrological processes: Application on the Donga catchment
(Benin) Journal of Hydrology , 363 (1-4 ): 32-41.
Lubczynski, M.W., Roy, J., 2003. Hydrogeological interpretation and potential of the new magnetic
resonance sounding (MRS) method. Journal of Hydrology 283/1–4, 19–40.
Lubczynski, M.W., Roy, J., 2004. Magnetic resonance sounding: new method for groundwater
assessment. Ground Water 42 (2), 291–303.
MacDonald and Davies, 2000. A brief review of groundwater for rural water supply in sub-Saharan
Africa, BGS Technical Report WC/00/33.
Maréchal, J.C., Varma, M.R.R., Riotte, J., Vouillamoz, J.M., Mohan Kumar, M.S., Ruiz, L.,Sekhar,
M., Braun, J.J., 2009 Indirect and direct recharges in a tropical forested watershed: Mule Hole,
India. Journal of Hydrology, 364 (3-4), 272-284.
Marcoux, M. A., 2008. Optimisation des performances hydro-bio-mécaniques d’une ISDND exploitée
en mode bioréacteur :étude à l’échelle du site. Thèse Doctorat UJF, Grenoble.
Massuel, S., Favreau, G., Descloitres, M., Le Troquer, Y., Albouy, Y., Cappelaere, B. Infiltration
profonde à travers une zone d’épandage sableuse de versant au Niger semi-aride: évidence par
modélisation hydrologique et reconnaissance géophysique. Colloque GEOFCAN, Paris, septembre
2003.
Massuel, S., Favreau, G., Descloitres, M., Le Troquer, Y., Albouy, Y., Cappelaere, B. Deep
infiltration through a sandy alluvial fan in semiarid Niger inferred from electrical conductivity
survey, vadose zone chemistry and hydrological modelling. 2006. CATENA, 67 (2), pp 105-118.
74
Paterson, N., R., Bosschart, R., A., 1987. Airborne geophysical Exploration for Groundwater.
Groundwater, Vol. 25, N°1, jan-feb 1987.
Peugeot, C., Esteves, M., Galle, S., Rajot, J.L., Vandervaere, J.P., 1997. Runoff generation processes:
results and analysis of field data collected at the East Central Supersite of the HAPEX-Sahel
experiment. J. Hydrol. 188–189, 179–202.
Ramirez, E., Francou, B., Ribstein, P., Descloitres, M., Guérin, R., Mendoza, J., Gallaire, R., Joradan,
E., 2001. Small-sized glaciers disappearing in the Tropical Andes: A case study in Bolivia, The
Chacaltaya Glacier, 16°S. Journal of Glaciology, vol 47, Issue 157, p 187-194.
Rejiba, F., Descloitres, M., Ribolzi, O., Camerlynck, C., 2001. Apport du radar haute résolution pour
la reconnaissance des placages sableux au Sahel. Actes du 3ème Colloque GEOFCAN
(Géophysique des sols et des Formations Superficielles), 25-26 septembre, Orléans, France, pp 6771.
Ribolzi, O., Hermida, M. Karambiri, H., Delhoume, J. P., Thiombiano, L., 2006. Effects of aeolian
processes on water infiltration in sandy Sahelian rangeland in Burkina Faso. Catena 67 (2006) 145
– 154.
Ritz, M., Descloitres, M., Courteaud, M., Robineau, B., 1997. Audiomagnetotelluric prospecting for
groundwater in the Baril coastal area. Piton de la Fournaise. Réunion Island. Geophysics, Vol
62(3), pp 758-762.
Robain, H., Descloitres, M., Ritz M., Yene Atangana, Q., 1996. A multiscale electrical survey of a
lateritical soil system in African rain forest. Journal of Applied Geophysics 34, pp 237-253.
Robineau, B., Ritz, M., Courteaud, M., Descloitres, M., 1997. Electromagnetic Investigations of
Aquifers in the Grand Brulé Coastal Area of Piton de la Fournaise Volcano, Reunion Island.
Groundwater, 35(4), pp 585-592.
Ruiz, L., Murari Varma, R. R., Mohan Kumar, M.S., Sekhar, M., Maréchal, J. C., Descloitres, M.,
Riotte, J., Kumar, S., Kumar, C., Braun J.J., 2009. Water balance modelling in a tropical
watershed under deciduous forest (Mule Hole, India): regolith matric storage buffers the
groundwater recharge process. Journal of Hydrology, Vol. 380, Issues 3-4, pp. 460-472.
Savadogo, A. N., Descloitres, M., Nakolendousse, S., Camerlynck, C., Bazie, P., Le Troquer, Y.,
Koussoube, Y., 2001. Etude géophysique du tracé de la digue du futur barrage de Yakouta au
Burkina Faso. Complémentarité des méthodes électriques et radar en milieu dunaire. Actes du 3ème
Colloque GEOFCAN (Géophysique des sols et des Formations Superficielles), 25-26 septembre,
Orléans, France, pp 131-134.
Schmutz, M., Guérin, R., Maquaire, O., Descloitres, M., Schott, J.J., Albouy, Y., 1999. Apport de
l’association des méthodes TDEM et électrique pour la connaissance de la structure du glissementcoulée de Super Sauze (Bassin de Barcelonnette, Alpes de Haute-Provence, France). C.R. Acad.
Sci. Paris, IIa, 328, pp 797-800.
Schmutz, M., Albouy, Y., Guérin, R., Maquaire, O., Vassal, J., Schott, J. J., Descloitres, M., 2000.
Joint electrical and Time Domain Electromagnetism (TDEM) data inversion applied to the Super
Sauze earthflow (France). Surveys in Geophysics, 21, pp 371-390.
Singha, K., Gorelick, S. M. 2006. Effects of spatially variable resolution on field-scale estimates of
tracer concentration from electrical inversions using Archie’s law. Geophysics, Vol. 71, N°. 3, pp
G83–G91
Talbi , A. , 2001. Etude du cycle de sel dans un bassin endoréique. Modélisation hydrogéochimique
des écoulements souterrains dans les Andes centrales depuis le retrait du Lac tauca (11 000 ans) .
thèse de l’Université de Paris 06, Paris, 182 p.
Toé, G., Vouillamoz, J.M., Descloitres, M., Robain H., Andrieux, P. 2004 New Geophysical Tools to
Study Hard Rock Aquifers Case Studies from Burkina Faso, W. Africa. International Conference
EAGE Paris, 7-10 june 2004.
Toé, G., 2004. Apport de nouvelles techniques géophysiques à la connaissance des aquifères de socle
tomographie électrique : électromagnétisme fréquentiel : résonance magnétique protonique:
applications au Burkina Faso. Thèse de l’Université Pierre et Marie Curie, 1 vol., 271 p.
Vouillamoz, J. M., Descloitres, M., Bernard, J., Fourcassier, P., Romagny, L., 2002. Application of
integrated magnetic resonance sounding and resistivity methods for borehole implementation. A
case study in Cambodia. Journal of Applied Geophysics, 50, pp 67-81.
75
Vouillamoz, J. M., Descloitres, M., Toe, G., Legchenko, A., 2005. Characterization of crystalline
basement aquifers with MRS: comparison with boreholes and pumping tests data in Burkina Faso.
Near Surface Geophysics, 3: 107-111.
Vouillamoz, J. M., G. Favreau, S. Massuel, M. Boucher, Y. Nazoumou, A. Legchenko, 2008.
Contribution of magnetic resonance sounding to aquifer characterization and recharge estimate in
semiarid Niger, J. Appl. Geophys. 64, pp 99–108.
Wubda, M., 2003. Prospections géophysiques sur le bassin versant d’Ara, Nord Bénin. Rapport de
stage de DESS, IRD Ouagadougou, Université de Paris 6.
Wyns, R., Baltassat, J., M., Lachassagne, P., Legchenko, A., Vairon, J., Mathieu, F., 2004.
Application of proton magnetic resonance soundings to groundwater reserve mapping in
weathered basement rocks (Brittany, France) Bull. Soc. géol. Fr., 2004, t. 175, n1, pp. 21-34.
Zhdanov, M., S., Pavlov, D., A., 2001. Analysis and interpretation of anomalous conductivity and
magnetic permeability effects in time domain electromagnetic data. Journal of Applied
Geophysics, vol. 46, no4, pp. 235-248.
76
ANNEXE 1
LISTE DES PUBLICATIONS
77
1. Publications dans des journaux à comité de lecture
2010
Clément, R., Descloitres, M., Günther, T., Morra, C., Oxarango, L. Artefact removal in time-lapse ERT
interpretation. Application to leachate injection experiment in landfills., 2010. Waste management, in press.
2009
Boucher M., Favreau, G. Descloitres, M., Vouillamoz, J. M., Massuel, S., Nazoumou, Y., Legchenko, A., 2009.
Contribution of geophysical surveys to groundwater modelling of a porous aquifer in semiarid Niger: An
overview. Comptes Rendus Geoscience, 341, 800-809.
Braun, J-J., Descloitres, M., Riotte, J., Fleury, S., Barbiero, L., Boeglin, J-L., Violette, A., Lacarce, E., Ruiz, L.,
Sekhar, M., Mohan Kumar, M.S., Subramanian, S., Dupre, B., 2009. Regolith mass balance inferred from
combined mineralogical, geochemical and geophysical studies: Mule Hole gneissic watershed, South India,
Geochimica et Cosmochimica Acta, doi: 10.1016 /j.gca.2008.11.013.
Clément, R., Descloitres, M., Günther, T., Ribolzi, Legchenko, A. (2009), Influence of shallow infiltration on timelapse ERT. Experience of advanced interpretation. Comptes Rendus Geosciences, 341, pp 886-898.
Hinderer J., de Linage C., Boy J.P., Gegout P., Masson F., Rogister Y., Amalvict M., Pfeffer J., Littel F., Luck B.,
Bayer R., Champollion C., Collard P., Le Moigne N., Diament M., Deroussi S., de Viron O., Biancale R.,
Lemoine J.M., Bonvalot S., Gabalda G., Bock O., Genthon P., Boucher M., Favreau G., Séguis L., Descloitres
M., Galle S., 2009. The GHYRAF (Gravity and Hydrology in Africa) experiment: description and first results.
Journal of Geodynamics 48, Issues 3-5, pp: 172-181.
Ruiz, L., Murari Varma, R. R., Mohan Kumar, M.S., Sekhar, M., Maréchal, J. C., Descloitres, M., Riotte, J.,
Kumar, S., Kumar, C., Braun J.J., 2009. Water balance modelling in a tropical watershed under deciduous forest
(Mule Hole, India): regolith matric storage buffers the groundwater recharge process. Journal of Hydrology, Vol.
380, Issues 3-4, pp. 460-472
2008
Descloitres, M., Ribolzi, O., Le Troquer, Y., Thiebaux, J.P. Study of water tension differences in heterogeneous
sandy soils using surface ERT. Journal of Applied Geophysics, Vol 64/3-4, pp 83-98 DOI information:
10.1016/j.jappgeo.2007.12.007.
Descloitres, M., Ruiz, L., Sekhar, M., Legchenko, A., Braun, J. J., Mohan Kumar, M.S., Subramanian, S., 2008.
Characterization of seasonal local recharge using Electrical Resistivity Tomography and Magnetic Resonance
Sounding. Hydrological Processes, Vol 22, pp 384 – 394.
2007
Barbiéro L., Parate, H. R., Descloitres, M., Bost A., Furian S., Mohan Kumar M.S, Kumar C., Braun J. J., 2007.
Using a structural approach to identify relationships between soil and erosion in a semi-humid forested area,
South India. CATENA, vol. 70 , pp 313–329
Kamagate, B., Favreau, G., Séguis, L., Seidel, J. L., Descloitres, M., Affaton, P, 2007. Hydrological processes and
water balance of a tropical crystalline bedrock catchment in Benin (Donga, upper Ouémé River). Comptes Rendus
Geosciences, Volume 339, Issue 6, pp. 418-429.
Ribolzi, O., Karambiri, H., Bariac, T., Benedetti, M., Caquineaux, S., Descloitres, M., Aventurier, A. 2007. Soil
surface characteristics and suspended load affect storm solutes behaviour in a semi-arid catchment. Journal of
Hydrology, Volume 337, Issues 1-2, 15 April 2007, Pages 104-116.
2006
Legchenko, A., Descloitres, M., Bost, A., Ruiz L., Reddy, M., Girard, J-F., Sekhar, M., Mohan Kumar, M.S.,
Braun, J. J. Resolution of MR Soundings applied to the characterization of hard rock aquifers. Groundwater
44(4), pp 547-554.
Massuel, S., Favreau, G., Descloitres, M., Le Troquer, Y., Albouy, Y., Cappelaere, B. Deep infiltration through a
sandy alluvial fan in semiarid Niger inferred from electrical conductivity survey, vadose zone chemistry and
hydrological modelling. 2006. CATENA, 67 (2), pp 105-118.
2005
Vouillamoz, J. M., Descloitres, M., Toe, G., Legchenko, A., 2005. Characterization of crystalline basement aquifers
with MRS: comparison with boreholes and pumping tests data in Burkina Faso. Near Surface Geophysics, 3: 107111.
2003
Descloitres, M., Ribolzi, O., Le Troquer, Y, 2003. Study of infiltration in a gully erosion sahelian area using timelapse electrical resistivity mapping. CATENA 53, pp 229-253.
2002
Vouillamoz, J. M., Descloitres, M., Bernard, J., Fourcassier, P., Romagny, L., 2002. Application of integrated
magnetic resonance sounding and resistivity methods for borehole implementation. A case study in Cambodia.
Journal of Applied Geophysics, 50, pp 67-81.
78
2001
Albouy, Y., Andrieux, P., Rakotondrasoa, G., Ritz, M., Descloitres, M., Join, J. L., Rasolomanana, E., 2001.
Mapping Coastal Aquifers by Joint Inversion of DC and TEM Soundings. Three Case Histories. Groundwater,
Vol. 39, n°1,pp 87-97.
Guérin R., Descloitres, M., Coudrain-Ribstein A., Talbi A., Gallaire R., 2001. Geophysical surveys for identifying
saline groundwater in the semi-arid region of the central Altiplano, Bolivia. Hydrological Processes, 15, 17, pp
3287-3301.
Maquaire, O., Flageollet, J. C., Malet, J. P., Schmutz, M., Webder, D., Klotz, S., Albouy, Y., Descloitres, M.,
Dietrich, M., Guérin, R., Schott, J. J., 2001. Une approche multidisciplinaire pour la connaissance d’un glissement
coulée dans les marnes noires du Callovo-Oxfordien (Super Sauze, Alpes de Haute-Provences, France). Revue
Française de Géotechnique, 95/96, 15-32.
2000
Descloitres, M., Guérin, R., Albouy, A., Tabbagh, A., Ritz, M., 2000. Improvement in TDEM sounding
interpretation in presence of induced polarization. A case study in resistive rocks of Fogo volcano, Cape Verde
Islands. Journal of Applied Geophysics, 45, pp 1-18.
Ramirez, E., Francou, B., Ribstein, P., Descloitres, M., Guérin, R., Mendoza, J., Gallaire, R., Joradan, E., 2001.
Small-sized glaciers disappearing in the Tropical Andes: A case study in Bolivia, The Chacaltaya Glacier, 16°S.
Journal of Glaciology, vol 47, Issue 157, p 187-194.
Schmutz, M., Albouy, Y., Guérin, R., Maquaire, O., Vassal, J., Schott, J. J., Descloitres, M., 2000. Joint electrical
and Time Domain Electromagnetism (TDEM) data inversion applied to the Super Sauze earthflow (France).
Surveys in Geophysics, 21, pp 371-390.
1999
Descloitres, M., Guérin, R., Ramirez, E., Gallaire, R., Ribstein, P., Valla, F., 1999. Détermination de l’épaisseur des
glaciers de Sarenne (Alpes) et de Chacaltaya (Bolivie) par prospection radar à 50 MHz. La Houille Blanche,
Société Hydrotechnique de France, n°5, pp 29-33
Guérin, R., Descloitres, M., Coudrain-Ribstein, A., Talbi, A., Ramirez, E., Gallaire, R., 1999. Etude d’un aquifère
salé de l’Altiplano Bolivien par prospection TDEM. Pangea, 31/32, pp 26-29.
Ritz, M., Robain, H., Pervago, E., Albouy, A., Camerlynck, C., Descloitres, M., Mariko, A., 1999. Improvement to
resistivity pseudosection modelling by removal of near surface inhomogeneity effects. Application to a soil
system in south Cameroon. Geophysical Prospecting, vol. 47, pp 85-101.
Robain, H., Albouy, Y., Camerlynck, C., Dabas, M., Descloitres, M., Mechler, P., Tabbagh, A., 1999. The location
of infinite electrodes in pole-pole electrical surveys : consequences for 2D imaging. Journal of Applied
Geophysics, 41, pp 313-333.
Schmutz, M., Guérin, R., Maquaire, O., Descloitres, M., Schott, J.J., Albouy, Y., 1999. Apport de l’association des
méthodes TDEM et électrique pour la connaissance de la structure du glissement-coulée de Super Sauze (Bassin
de Barcelonnette, Alpes de Haute-Provence, France). C.R. Acad. Sci. Paris, IIa, 328, pp 797-800.
1998
Descloitres, M., 1998. Les sondages électromagnétiques en domaine temporel (TDEM) : Application à la
prospection d’aquifères sur les volcans de Fogo (Cap Vert) et du Piton de la Fournaise (la Réunion). Thèse de
Doctorat de l’Université de Paris 6, 238 p.
Courteaud, M., Robineau, B., Ritz, M., Descloitres, M., 1998. Electromagnetic Mapping of Subsurface Formations
in the Lower Northeast Rift Zone of Piton de la Fournaise Volcano : Geological and Hydrogeological
Implications. J. Engineering and Environmental Geophysics, 2(3), pp 181-187.
1997
Descloitres, M., Ritz, M., Courteaud, M., Robineau, B., 1997. Electrical structure beneath the collapsed eastern
flank of Fournaise volcano, Réunion Island: implication to the quest for groundwater. Water Resources Research,
33(1), pp 13-19.
Giroux, B., Chouteau, M., Descloitres, M., Ritz, M., 1997. Use of the magnetotelluric method in the study of the
deep Maestrichtian aquifer in Senegal. Journal of Applied Geophysics, 38, pp 77-96.
Ritz, M., Descloitres, M., Courteaud, M., Robineau, B., 1997. Audiomagnetotelluric prospecting for groundwater in
the Baril coastal area. Piton de la Fournaise. Réunion Island. Geophysics, Vol 62(3), pp 758-762.
Robineau, B., Ritz, M., Courteaud, M., Descloitres, M., 1997. Electromagnetic Investigations of Aquifers in the
Grand Brulé Coastal Area of Piton de la Fournaise Volcano, Reunion Island. Groundwater, 35(4), pp 585-592.
1996
Courteaud, M., Ritz, M., Descloitres, M., Robineau, B., Coudray, J., 1996. Cartographie audiomagnétotellurique du
biseau salé dans la zone côtière du Piton de la Fournaise (Ile de la Réunion). C.R. Acad. Sci. Paris, t.322, série II
a, pp 93 -100.
Robain, H., Descloitres, M., Ritz M., Yene Atangana, Q., 1996. A multiscale electrical survey of a lateritical soil
system in African rain forest. Journal of Applied Geophysics 34, pp 237-253.
2
Articles soumis
Chaudury, A., Sekhar, M., Descloitres, M., Godderis, Y., Braun, J. J. Stochastic modelling of steady groundwater
flow in complewx aquifer with geoemetry cosntrain derived from 2-D ERT geophysics., submitted to Advances in
water sciences, February 2009.
79
Clément, R., Oxarango, L., Descloitres, M., 2010. Contribution of 3-D Time Lapse ERT to Study Plume Migration
in Landfill. Waste management, submitted March 2010.
Clément, R., Legchenko,A., Quetu,M., Descloitres, M., Oxarango, L., Guyard., H. Laboratory and in-situ study of
landfilled domestic waste using magnetic resonance measurements, submitted to Near Surface Geophysics,
janvier 2010.
Descloitres, M., Séguis, L Legchenko, A., Wubda, M. Hydrogeophysical identification of representative geological
units as the first step of hydrodynamic modelling at a watershed scale in metamorphic context (North Bénin),
Near Surface Geophysics, submitted december 2009
Parate, R. P. , Mohan Kumar, M,.S., Descloitres, M., Barbiero, L., Ruiz, L., Braun, J. J., Sekhar, M., Kumar, C.,
Comparison of electrical resistivity by geophysical method and neutron probe logging for soil moisture
monitoring in forested watershed. Submitted to Current Sciences, janvier 2009.
Séguis, M., Kamagaté, B., Favreau, G., Descloitres, M., Seidel, J.L., Galle, S., Peugeot, C., Gosset, M., Le Barbé,
L., Malinur, F., Van Exter, S., Arjounin, M., Wubda, M. Origins of streamflow in a crystalline basement
catchment in the sub-humid Soudanian one: the Donga basin (Benin, West Africa). Inter annual variability of
water budget. Journal of Hydrology, submitted January 2009.
Séguis, L., Boulain, N., Cappelaere, B., Cohard, J.M., Descroix, L., Descloitres, M., Favreau, G., Galle, Guyot, A.,
S., Hiernaux, P., Kamagaté, B., Lebel, T., Le Lay, M., Mougins, É., Peugeot, C., Ramier, D., Seghieri, J., Timouk,
F. Contrasted land surface processes along a West African meridian rainfall gradient. Submitted to Atmospheric
Science Letters, January 2010.
3
Brevet
Descloitres, M., Le Troquer, Y., 2004. Sonde de diagraphie géophysique pour la mesure de la résistivité sur la paroi
d’un forage. Brevet d’invention, Institut National de la Propriété Industrielle, N° publication 2 845 416, N°
d’enregistrement national: 02 12191, date de mise à disposition: 24/12/2004.
4. Conférences internationales avec actes
2010
Ruiz, L., Varma., M. R. R., Mohan Kumar, M. S., Sekhar, M., Molenat, J., Marechal, J. C., Descloitres, M., Riotte,
J., Kumar, S., Braun, J.J., 2010. Transpiration by tree roots in the deep unsaturated regolith buffers the recharge
process in a tropical watershed under deciduous forest (Mule Hole, India) Geophysical Research Abstracts Vol.
12, EGU2010-2719, 2010, EGU General Assembly 2010
2009
Boucher M., Favreau G., Massuel S., Vouillamoz J.M., Descloitres, M., Cappelaere B., Nazoumou Y., Legchenko
A. – Subsurface geophysics for constraining surface water - groundwater modeling in SW Niger. 3rd
International AMMA Conference, July 20-24, 2009, Ouagadougou, Burkina Faso, Po.2B.14, p. 66.
Boucher M., Favreau G., Massuel S., Vouillamoz J.M., Descloitres M., Cappelaere B., Nazoumou Y., Legchenko
A. – Contribution of MRS to groundwater modelling of an unconfined aquifer in SW Niger. 4th International
Workshop on the Magnetic Resonance Sounding, October 20-22, 2009, Grenoble, France, pp. 5-10.
Braun, J. J., Descloitres, M., Riotte, J., Deschamps, P., Violette, A., Marechal, J.C., Sekhar, M., Mohan Kumar,
M.S., Subramanian, S., 2009. Contemporary versus long-term weathering rates in Tropics:
Mule Hole , South India. Goldschmidt Conference at Davos, Switzerland, june 2009.
Clement, R., Descloitres, M., Günther, T., Oxarango, L., 2009. Comparison of three arrays in time-lapse ERT:
Simulation of a leachate injection experiment., 7th Colloque “GEOFCAN” and 8th International Conference on
Archaeological Prospection, 9-12 september 2009, Conservatoire National des Arts et Métiers, Paris, France.
Clement, R., Oxarango, L., Descloitres, M., 2009. Hydrodynamic of Leachate Plume in Bioreactor Landfill,
contribution of 3D Time-lapse ERT. 15th European Meeting of Environmental and Engineering Geophysics,
“Near Surface 2009”, 7 - 9 September, Dublin, Ireland.
Clément, R., Legchenko, A., Quetu, M., Descloitres, M., Oxarango, L., Guyard H., 2009. Laboratory and in-situ
study of landfill domestic waste using magnetic resonance measurements. 4th International Workshop on the
Magnetic Resonance Sounding, October 20-22, 2009, Grenoble, France, pp. 5-10
Descloitres M., Legchenko A., Séguis, L., Wubda, M., 2009. Contribution of MRS and resistivity surveys to local
scale watershed hydrology. A case study in metamorphic context, North Bénin., Extended Abstract, 4th
International Workshop on the Magnetic Resonance Sounding, October 20-22, 2009, Grenoble, France.
Genthon, P., Sylvestre, F., Favreau, G., and the LAKE CHAD PROJECT Team. Water Resources in the Lake Chad
Basin, Assessment, Uses and Social Organizations. Geophysical Research Abstracts, Vol. 11, EGU2009-0, 2009
EGU General Assembly 2009.
Hinderer, J. and the GHYRAF team. The GHYRAF (Gravity and HYdrology in Africa) experiment: first results
from GPS, GRACE and surface gravity observations in relation with water storage changes. Geophysical
Research Abstracts, Vol. 11, EGU2009-0, 2009 EGU General Assembly 2009
Séguis, L., Galle, S., Descloitres, M., Laurent, J.-P., Grippa, M., Pfeffer, J., Luck, B., Genthon, P., Hinderer J.
Monitoring water stock variations by gravimetry in Benin. Geophysical Research Abstracts, Vol. 11, EGU2009-0,
2009 EGU General Assembly 2009.
80
Wubda, M., Descloitres, M., Séguis, L., Legchenko, A., 2009. Hydrogeophysical surveys for local scale hydrology
in north Benin. Poster session, 3rd International AMMA Conference, July 20-24, 2009, Ouagadougou, Burkina
Faso.
2008
Clément, R., Descloitres, M., Günther, T., 2008. Influence of shallow infiltration on time-lapse ERT. Geophysical
Research Abstracts, Vol. 10, EGU2008-A-09082, 2008 EGU General Assembly 2008, Vienna, april 2008.
Descloitres, M., Laurent, J. P., Morra, C., Clément, R., Oxarango, L., Gourc, J. P. Monitoring resistivity in nonhazardous waste landfill using Time Domain Electromagnetism (Drôme, France). International Conference
EAGE “Near Surface 2008”, Karkow, Poland, September 2008.
Hoareau, J., Vouillamoz, J. M., Grammare, M., Descloitres, M., Kumar, C., Nandagiri, L., 2008. Fresh water
mapping and quantification in coastal aquifers using Magnetic Resonance Soundings and Time Domain
Electromagnetism. EGU abstract, Vienna, April 2008.
2007
Descloitres, M., Séguis, L., Wubda, M., Legchenko, A., 2007. Discrimination of rocks with different hydrodynamic
properties using MRS, EM and resistivity methods. International Conference EAGE “Near Surface 2007”,
Istambul, sept. 2007.
2006
Baltassat, J.M., Krishnamurthy, N.S., Girard, J. F., Dutta, S., Dewandel, B., Chandra S., Descloitres, M.,
Legchenko, A., Robain, H., Ahmed, S., 2006. Geophysical Characterization of weathered granite aquifers in the
Hyderabad region, Andra Pradesh, India. Extended abstract of the 3rd Magnetic Resonance Sounding
International workshop, MRS2006, a reality in applied geophysics, Madrid, 25-27 october, 2006.
Braun, J.-J., Descloitres, M., Riotte, J., Barbiéro L., Fleury S., Boeglin, J.L., Ruiz, L., Muddu, S., Mohan Kumar,
M.S., Kumar, C. Regolith thickness inferred from geophysical and geochemical studies in a tropical watershed
developed on gneissic basement: Moole Hole, Western Ghâts (South India). Geochimica et Cosmochimica Acta,
Volume 70, Issue 18, Supplement 1, August-September 2006, page A65.
Descloitres, M., Ruiz, L., Sekhar, M., Legchenko, A., Bost, A., Mohan Reddy, M., Parate, H., 2006. MRS and ERT
for localizing temporary recharge in heterogeneous aquifer. Extended abstract of the 3rd Magnetic Resonance
Sounding International workshop, MRS2006, a reality in applied geophysics, Madrid, 25-27 october, 2006.
Descloitres, M., Ribolzi, O., Le Troquer, Y., Thiébaux, J. P., 2006. Spatializing Water Tension in Heterogeneous
Sandy Soils with Surface ERT During Rain-Evaporation Cycles. 12th European Meeting of EAGE “Near Surface
2006”, Helsinki, Finland 4 - 6 September 2006. Extended abstract, Paper B039.
Legchenko, A., Baltassat, J.M., Boucher, M., Descloitres, M., Girard, .J.F., Ezerski, M., Vouillamoz, J.M., 2006.
Magnetic resonance sounding as a tool for imaging of hard Rock and karst aquifers., in Hydrogeophysical
Workshop - Vancouver - British Columbia.
2005
Braun, J.J., Ruiz, L., Riotte, J., Mohan Kumar, M.S., MurariI, V., Sekhar, M., Barbiéro, L., Descloitres, M., Bost,
A., Dupré, B., Lagane, C., 2005. Chemical and physical weathering in the Kabini River Basin, South India.
Goldschmidt Conference Abstracts, Paper n° A691, Moscow, USA.
Chaudurry, A., Sekhar, M., Descloitres, M., Legchenko, A, 2005. Stochastic modelling combined with geophysical
investigations for groundwater fluxes at watershed scale in the weathered gneissic formations of South India. Prepublished Proceedings, ModelCARE 2005, 5th International Conference on Calibration and Reliability in
Groundwater Modelling. The Hague, The Netherlands, 6-9 June 2005, pp 35-41.
Favreau, G., Guéro, A., Massuel, S., Nazoumou, Y., Descloitres, M., Leblanc, M., Cappelaere, B., Descroix, L.,
2005. La nappe phréatique en hausse du SO Niger, un paradoxe sahélien ? Nouveau bilan et perspectives.
Conférence AMMA, Dakar, décembre 2005
Legchenko, A., Descloitres, M., Bost, A., Ruiz, L., Reddy, M., Girard, J.F., Sekhar, M., Mohan Kumar, M.S.,
Braun, J.J., 2005. Characterization of fractured rock aquifers by surface geophysical methods. Abstracts of the
EAGE European conférence « Near Surface 2005 », Palermo, Italy, 4-7 september.
2004
Barbiéro, L., Bost, A., Camerlynck, C., Descloitres, M., Kumar, C., Mohan Kumar, M.S., Sekhar, M., 2004.
Understanding natural soil erosion vulnerability with dense electromagnetic mapping Example of the Moole Hole
forested watershed, South India. Proceedings of GEORISK-2004, International Workshop on Risk Assessment in
Site Characterization and Geotechnical Design, Bangalore, India. November 26-27th, p. 265-272.
Robain, H., Camerlynck, C., Baltassat, J. M., Descloitres, M., Legchenko, A., Dewandel, B., Krishnamurthy, N. S.,
RAO, P., 2004. Groundwater depletion of a heavily irrigated watershed in Southern India: Detailed assessment
using MRS and ERT. Eos Trans. AGU, 85(17), Jt. Assem. Suppl., Abstract NS41A-03.
Toé, G., Vouillamoz, J.M., Descloitres, M., Robain H., Andrieux, P. 2004 New Geophysical Tools to Study Hard
Rock Aquifers Case Studies from Burkina Faso, W. Africa. International Conference EAGE Paris, 7-10 june
2004.
2003
Krishnamurthy, N. S., Baltassat, J.M., Robain, H., Legchenko, A., Descloitres, M., Lachassagne, P., Kumar, D.,
Ahmed, S. 2004. MRS and electrical imagery for characterizing weathered and fractured hard rock aquifer in
theMaheswaram watershed, Hyderabad, India. 2nd International Workshop on the Magnetic Resonance Sounding
method applied to non-invasive groundwater investigations, November 19-21, 2003,Orléans, France, pp 53-56.
2000
81
Ramirez, E., Francou, B., Ribstein, P., Descloitres, M., Guérin, R., Pouyaud, B., Jordan, E, 2000. La recesion del
glaciar de Chacaltaya (Bolivia, 16°S) desde la pequeña edad de hielo, y su aceleracion actual. Xème Congreso
Peruano de Geologia. Simposium de Glaciologia, Lima, Peru, 22/06/2000.
1999
Guérin, R., Descloitres, M., Coudrain-Ribstein, A., Ramirez, E., 1999. Sondeos electromagneticos en el dominio
del tiempo (TDEM) en Bolivia (Altiplano y Salar de Uyuni). Jornadas sobre Actualidad de las Técnicas Aplicadas
en Hidrogeología. Grenade. 10-12 Junio.
Schmutz, M., Guérin, R., Maquaire, O., Descloitres, M., Schott, J. J., Albouy, Y., 1999. Contribution of the TDEM
and electrical combination to a flowslide study. European Geophysical Society (EGS), La Haye, 19-23/04/1999.
1998
Robain, H., Albouy, Y., Camerlynck, C., Dabas, M., Descloitres, M., Tabbagh, A., 1998. Geophysical surveys
contribution to structural and behavioural knowledge of tropical soils. Application to mapping purpose. World
congress of soil science. AISS. Montpellier 20-26/08/1998.
1997
Boutard, G., Camerlynck, C., Dabas, M., Descloitres, M., Robain, H., 1997. Aerial features removing from Ground
Penetrating Radar profiles. EEGS 3rd Conference, Aarhus, September 5-7th .
Robain, H., Albouy, Y., Dabas, M., Descloitres, M., Camerlynck, C., Mechler, P., Tabbagh, A.,1997. The location
of infinite electrodes in pole-pole electrical surveys and the resulting error for 2D electrical imaging. A practical
point of view. EEGS 3rd Conference, Aarhus, September 5-7th.
1996
Courteaud, M., Robineau, B., Coudray, J., Ritz, M., Descloitres, M., Albouy, Y., 1996. Audiomagnetotelluric
evaluation of saline water intrusion : Ste-Rose coastal area, piton de la fournaise, Reunion Island. 2nd Meeting on
Environmental and Engineering Geophysics, Nantes, September 2-5th.
Descloitres, M., Albouy, Y., Bouvier, A., Andrieux, P., Rakotondrasoa, G., Join, J. L., Coudray, J., 1996. DC and
transient soundings to map coastal aquifers. The International Congress on Environment and Climate IGEC-96.
Rome 04-08/03/96.
1995
Courteaud, M., Descloitres, M., Join, J. L., Albouy, Y., Coudray, J., 1995. TDEM survey in heterogeneous volcanic
aquifers: correlation between basic one dimensional models and hydrogeological data at 15 borehole test sites of
Reunion Island. Congress of the International Association of Hydrogeology, Edmonton, June 4-10th.
Descloitres, M., Ritz, M., Mourgues, P. 1995. TDEM soundings for locating aquifers inside the caldeira of Fogo
active volcano. Cape Verde Islands. First Meeting on Environmental and Engineering Geophysics, Torino,
September 25-27th.
Ritz, M., Robineau, B., Descloitres, M., Courteaud, M., Coudray, J., 1995. AMT and TDEM
groundwaterprospecting on the eastern collapsed flank of Piton de la Fournaise volcano. Reunion Island.
Congress of the International Association of Hydrogeology, Edmonton, June 4-10th.
5 Conférences nationales, colloques et réunions scientifiques
2009
Descloitres, M., Chalikakis, K., Legchenko, A., Moumouni, A., Favreau, G., Genthon, P., Le Coz, M., Oï, M., 2009
Ressources en eau souterraine de la vallée de la Komadougou (Diffa, Est Niger) Contribution géophysique.,
Workshop de restitution du projet PSP « Lac Tchad », Colloque de restitution , Niamey, 30 novembre- 1er
décembre 2009
2007
Descloitres, M., 2007. Spatialisation et suivi temporel de la tension dans les sols sableux hétérogènes par méthode
électrique DC : succès, difficultés et perspectives. Conférence lors du Colloque de restitution du projet
WATERSCAN, Coord P. Sailhac – A. Legchenko, Autrans, 10-12 octobre 2007
Descloitres, M., 2007. Suivi des processus hydrologiques par méthodes électriques (DC) et Résonance Magnétique
des Protons : succès, difficultés et perspectives. Conférence lors du Colloque de restitution du projet
WATERSCAN, Coord P. Sailhac – A. Legchenko, Autrans, 10-12 octobre 2007.
Descloitres, M., Legchenko, A., 2007. Caractérisation géophysique des aquifères par tomographie de résistivité
électrique (ERT) et sondage de résonance magnétique (MRS). Journées scientifiques « Imageries » du
groupement inter-laboratoires « GEMME », Grenoble, 8 février 2007.
Descloitres, M., Séguis, L., Wubda., M., 2007. Caractérisation des Aquifères sur les sites du Bénin. Apport de la
résonance magnétique des Protons. Premiers Résultats. Journées SO « AMMA CATCH », Mont Saint Odile, mars
2007.
Descloitres, M., Boucher, M., Favreau, G., Vouillamoz, J. M., 2007. Apport des sondages Electromagnétiques
temporels (TDEM) à la reconnaissance du substratum de l’aquifère du CT3 au Niger. Premiers résultats.
Journées SO « AMMA CATCH », Mont Saint Odile, mars 2007.
Favreau, G., Boucher, M., Vouillamoz, J.M., Descloitres, M., Massuel, S., Nazoumou, Y., Legchenko, A., (2007). Apport des sondages TDEM et RMP à une meilleure estimation des paramètres de la modélisation d’un aquifère
libre en milieu semi-aride (Niger). GEOFCAN, AGAP, qualité, Géophysique des Sols et des Formations
Superficielles 6 Colloque, Bondy, France,25-26-septembre-2007.
82
Favreau, G., Boucher, M., Descloitres, M., Vouillamoz, J.M., Massuel, S., Nazoumou, Y., Legchenko, A., 2007.
Apport des sondages TDEM et RMP à une meilleure estimation des paramètres de la modélisation d’un aquifère
libre en milieu semi-aride (Niger). Colloque GEOFCAN, septembre 2007.
Hoareau, J., Vouillamoz, J.M., Beck, M., Reddy, M., Descloitres, M., Legchenko, A., Sekhar M., Mohan Kumar,
M.S., and Braun, J.J., 2007 - Application des mesures de résistivité électrique et de résonance magnétique
protonique pour décrire les structures et les écoulements souterrains d’un aquifère complexe. GEOFCAN, AGAP,
qualité, Géophysique des Sols et des Formations Superficielles 6 Colloque, Bondy, France 25 et 26 Septembre
2007.
2006
Favreau, G., Vouillamoz, J. M., Boucher, M., Descloitres, M., 2006. Premiers résultats des prospections
hydrogéophysiques au Niger, perspective au Bénin, Journées ORE AMMA CATCH, Toulouse, octobre 2006.
Sekhar, M., Chauduri, A., Fleury, S., Descloitres, M., 2006. Stochastic modeling of groundwater flow in the
saprolite of a tropical gneissic watershed, Colloque IAHR-GW Toulouse 2006.
2005
Braun, J.J., Riotte, J., Ruiz, L., Barbiéro, L., Descloitres, M., Mohan Kumar, M.S., Sekhar, M., Bost, A., Dupré, B.,
Feydier, R., Lacaux Galy, C., Godderis, Y., Labat, D., Lagabe, C., Fritsch, E., Camerlynck, C., Murari, V., Parate,
H., Chaudurry, A., Veena, S., 2005. Etude intégrée du bassin de la rivière Kabini (Inde du Sud) Influence des
facteurs environnementaux sur les processus de fractionnement et les transferts hydro-biogéochimiques. Colloque
ECCO-PNRH, Toulouse, décembre 2005.
Favreau, G., Guéro, A., Massuel, S., Nazoumou, Y., Descloitres, M., Leblanc, M., Cappelaere, B., Descroix, L.,
2005. La nappe phréatique en hausse du SO Niger, un paradoxe sahélien ? Nouveau bilan et perspectives.
Conférence AMMA, Dakar, décembre 2005
Galle, S., Seguis, L., Arjounin, M., Bariac, T., Bouchez, J.M., Braud, I., Cohard, J.M., Descloitres, M., Favreau, G.,
Kamagate, B., Laurent, J.P., Le Lay, M., Malinur, F., Peugeot, C., Robain, H., Seghieri, J., Seidel, J.L., Varado,
N., Zin, I., Zribi, M., 2005. Evaluation des termes du bilan hydrologique sur le bassin versant de la Donga par
mesure et modélisation. Colloque ECCO PNRH, Toulouse, décembre 2005.
Legchenko, A., Descloitres, M., Bost, A., Ruiz, L., Reddy, M., Girard, J.F., Sekhar, M., 2005. Etude sur la capacité
des sondages RMP à localiser les aquifères de socle. Colloque Geofcan, Orléans, septembre 2005.
2004
Sekhar, M., Rasmi, S.N., Ruiz, L., Descloitres, M., 2004. Regional groundwater flow modeling in Kabini river
basin: Issues and challenges. Seminar on Assessment and Management of Water Resources (AMWR-2004),
Bangalore, India.
2003
Massuel, S., Favreau, G., Descloitres, M., Le Troquer, Y., Albouy, Y., Cappelaere, B. Infiltration profonde à travers
une zone d’épandage sableuse de versant au Niger semi-aride: évidence par modélisation hydrologique et
reconnaissance géophysique. Colloque GEOFCAN, Paris, septembre 2003.
Vouillamoz, J. M., Descloitres, M., Toé, G., 2003. La caractérisation des aquifères de socle du Burkina Faso par
sondages RMP. Colloque GEOFCAN, Paris, septembre 2003.
2001
Beck, M., Girardet, D., Chapellier, D., Descloitres, M., 2001. Diagraphies électriques pour l’optimisation de
l’hydrofracturation en zone de socle. Premiers résultats au Burkina Faso. Actes du 3ème Colloque GEOFCAN
(Géophysique des sols et des Formations Superficielles), 25-26 septembre, Orléans, France, pp 55-59.
Descloitres, M., Ribolzi, O., Le Troquer, Y., 2001. Variations saisonnières de la résistivité des sols d’une ravine sur
un versant sahélien. I. Etude cartographique par traîné Wenner. Actes du 3ème Colloque GEOFCAN (Géophysique
des sols et des Formations Superficielles), 25-26 septembre, Orléans, France, pp 25-29.
Descloitres, M., Ribolzi, O., Le Troquer, Y., 2001. Variations saisonnières de la résistivité des sols d’une ravine sur
un versant sahélien. II. Interprétation des panneaux électriques 2D. Actes du 3ème Colloque GEOFCAN
(Géophysique des sols et des Formations Superficielles), 25-26 septembre, Orléans, France, pp 31-34.
Rejiba, F., Descloitres, M., Ribolzi, O., Camerlynck, C., 2001. Apport du radar haute résolution pour la
reconnaissance des placages sableux au Sahel. Actes du 3ème Colloque GEOFCAN (Géophysique des sols et des
Formations Superficielles), 25-26 septembre, Orléans, France, pp 67-71.
Savadogo, A. N., Descloitres, M., Nakolendousse, S., Camerlynck, C., Bazie, P., Le Troquer, Y., Koussoube, Y.,
2001. Etude géophysique du tracé de la digue du futur barrage de Yakouta au Burkina Faso. Complémentarité des
méthodes électriques et radar en milieu dunaire. Actes du 3ème Colloque GEOFCAN (Géophysique des sols et des
Formations Superficielles), 25-26 septembre, Orléans, France, pp 131-134.
2000
Coudrain, A., Loubet, M., Guérin, R., Descloitres, M., Talbi, A., Quintanilla, J, Ledoux, E. 2000. Reconstitution
des écoulements souterrains de l’Altiplano bolivien pendant l’Holocène par modélisation hydrogéochimique.
Colloque PNRH 2000, Toulouse, 16-17 mai, pp 217-223.
1999
Guérin, R., Descloitres, M., Coudrain-Ribstein, A., Talbi, A., Ramirez, E., Gallaire, R., 1999. Etude d'un aquifère
salé de l'Altiplano bolivien par prospection TDEM. Colloque GEOFCAN, Géophysique des sols et des formations
superficielles. 21-22/09/1999. Orléans (France).
83
Vouillamoz, J. M., Bernard, J., Descloitres, M., Fourcassier, P., Romagny, L. 1999. Implantation de forages d’eau
au Cambodge. Utilisation conjointe des méthodes électriques, TDEM et RMP. Colloque GEOFCAN, Orléans, 2122 septembre.
1998
Schmutz, M., Guérin, R., Maquaire, O., Descloitres, M., Schott, J. J., Albouy, Y., 1998. Apport du TDEM à la
connaissance du glissement-coulée de Super Sauze (Alpes, France). Séminaire "MAST", Ecole de Physique des
Houches, 5-6/10/1998, pp 38-40.
1997
Descloitres, M., Robain, H., Dabas, M., Camerlynck, C., Albouy, Y., 1997. Apport des imageries électriques et
radar à la reconnaissance des couvertures d’altération. Bassin versant de Nsimi, Cameroun. Colloque GEOFCAN
“ Géophysique des sols et des formations superficielles ”, Bondy, 11-12 septembre.
Zanolin, A., Tchani, J., Barbiéro, L., Boivin, P., Descloitres, M., 1997. Apport de la méthode électrique pour la
reconnaissance hydrogéologique et l’étude des variabilités superficielles en zone sédimentaire subsaharienne.
Colloque GEOFCAN “ Géophysique des sols et des formations superficielles ”, Bondy, 11-12 septembre.
1996
Join, J. L., Descloitres, M., Ritz, M., 1996. Reconnaissance volcano-structurale de la phase ante caldeira du volcan
Fogo, Iles du Cap-Vert . 16ème Réunion des Sciences de la Terre, Orléans, 10-12 avril.
1995
Courteaud, M., Descloitres, M., Join, J. L., Albouy, Y., Coudray, J., 1995. TDEM survey in heterogeneous volcanic
aquifers: correlation between basic one dimensional models and hydrogeological data at 15 borehole test sites of
Reunion Island. Congress of the International Association of Hydrogeology, Edmonton, June 4-10th.
Descloitres, M., Ritz, M., Mourgues, P. 1995. TDEM soundings for locating aquifers inside the caldeira of Fogo
active volcano. Cape Verde Islands. First Meeting on Environmental and Engineering Geophysics, Torino,
September 25-27th.
Ritz, M., Robineau, B., Descloitres, M., Courteaud, M., Coudray, J., 1995. AMT and TDEM groundwater
prospecting on the eastern collapsed flank of Piton de la Fournaise volcano. Reunion Island. Congress of the
International Association of Hydrogeology, Edmonton, June 4-10th.
6 Rapports de contrats, de mission
Descloitres, M., Legchenko, A., Vincent, C., Guyard, H., Chalikakis, K., 2010. Recherche d’eau liquide dans le
glacier de Tête Rousse par sondage de Résonance Magnétique des Protons. Rapport de mission, Projet Pôle
« TUNES », LTHE-LGGE, Université Joseph Fourier, Grenoble I, 38 pages, 9 figures, 2 tableaux, 4 annexes.
Descloitres, M., Legchenko, A., Clément, R., Quetu, M., Oxarango, O., 2009. Prospections géophysiques sur le site
de Villiers sur Tholon, Rapport de mission du projet ADEM « Paraphyme ». LTHE, 40 pages.
Descloitres, M., Chalikakis, K., 2008. Caractérisation géophysique de l’aquifère sur le site de Bagara (Diffa, Est
Niger). Rapport de mission, Programme ANR Ghyraf, IRD-LTHE-HSM, 67 pages, 17 figures, 5 annexes.
Descloitres, M., Wubda, M., Séguis, L., 2008. Caractérisation géophysique de l’aquifère sur le site de Nalohou
(Djougou, Nord Bénin). Rapport de mission, Programme ANR Ghyraf, IRD-LTHE-HSM, 58 pages, 20 figures, 5
tableaux, 3 annexes.
Descloitres, M., 2008. Projet ANR « Bioréacteur», site de stockage de Chatuzange. Prospections géophysiques
(méthodes de résistivité). Rapport intermédiaire, 2008. Rapport interne LTHE, 26 pages, 18 figures, 1 tableau.
Descloitres, M., Favreau G., Boucher, M., Vouillamoz, J.M, 2007. Rapport de mission de prospection TDEM au
Niger. IRD-LTHE, BRGM, IRIS Instruments, Grenoble, février 2007.
Descloitres, M., Séguis, L., Wubda, M., 2007. Caractérisation des aquifères sur les sites Amma-Catch au Bénin.
Apport de la Résonance Magnétique des protons. Rapport de mission IRD-LTHE, Grenoble, mai 2007.
Baltassat J.M., Krishnamurthy, N.S., Girard, J.F., Dutta S., Dewandel, B., Chandra S., Descloitres, M., Legchenko,
A., Robain, H. 2006. Geophysical characterisation of weathered granite aquifer in the Hyderabad region, Andra
Pradesh, India Cefipra project final report, BRGM report RP-54538, 117 p., 57 fig., 11 tabl., 17 appendices.
Descloitres, M., Bost, A., Legchenko, A., Ruiz, L., Sekhar, M., 2005. Characterization of anistropic crystalline
basement aquifers using Magnetic Resonance Soundings (Southern India). Complementary survey. Report of the
Indo-French Cell for Water Science, IRD/IISc, Indian Institute of Science, Bangalore, february 2005.
Baltassat, J.M., Robain, H., Descloitres, M., 2004. Geophysical investigation on the Maheswaram watershed,
Hyderabad, India. Field report., BRGM/IRD/NGRI Cefipra project, Hyderabad.
Legchenko, A., Descloitres, M., Bost, A., Ruiz, L., Sekhar, M., 2004. Characterization of anistropic crystalline
basement aquifers using Magnetic Resonance Soundings (Southern India). Report of the Indo-French Cell for
Water Science, IRD/IISc, Indian Institute of Science, Bangalore, IRD-IISc report, 106 p., 51 fig., 9 tabl., 2 ann.
Descloitres, M., Wubda, M., Le Troquer, Y., 2003. Prospections géophysiques sur le bassin versant d’Ara, Nord
Bénin. Electrique 2D et électromagnétisme EM34. Compte rendu de mission 5 –14 mai 2003. Projet AMMA, UR
027 Geovast, IRD Ouagadougou, mai 2003.
Vouillamoz, J.M., Descloitres, M., 2003. La caractérisation des aquifères de socle par les sondages de Résonance
Magnétique des Protons (RMP). Première mise en œuvre au Burkina Faso. Compte rendu de mission, décembre
2002 et janvier 2003. Projet PNRH N°01/22 Document IRD / AcF, UR 027 Geovast, IRD Ouagadougou, juillet
2003.
84
Albouy, Y., Andrieux, P., Descloitres, M., Robain, H., Sorensen, K, Vassal, J, 2001. Time Domain
Electromagnetism. Comparison between 3 commercial systems. Centre de Recherche Géophysique de Garchy,
IRD Bondy, France.
Descloitres, M., Le Troquer, Y., 2000. Prospection géophysique sur le site du futur barrage de Yakouta, Burkina
Faso. Rapport interne IRD Ouagadougou.
Descloitres, M., Robain, H., 1999. Multi-electrode electrical and time domain electromagnetism survey at
Maheswaram catchement (India). Field report,. CEFIPRA Project 2013-1.
Descloitres, M., Join, J. L., Ritz, M., Dukhan, M., 1996. Prospection géophysique par la méthode TDEM du volcan
Fogo, archipel des îles du Cap-Vert. Rapport final. Multigraphié ORSTOM Dakar, 30 p., 17 figures, 10 p. en
annexe.
Courteaud, M., Descloitres, M., Ritz, M., Robineau, B., 1995. Secteur pilote de Ste Rose. Etude géophysique par les
méthodes TDEM et AMT. Rapport récapitulatif sur les mesures et données de terrain. Programme
hydrogéologique du Massif de la Fournaise (Ile de la Réunion), Conseil Général de l'île de la Réunion.
Multigraphié Université de St Denis de la Réunion. 55 p.
Courteaud , M., Descloitres, M., Ritz, M. et Robineau, B. 1994. Secteur pilote du Grand Brûlé : étude géophysique
par les méthodes TDEM et AMT. Implications géologiques et hydrogéologiques. Conseil Général de l'île de la
Réunion. Multigraphié Université de St Denis de la Réunion. 43 p., 25 fig., 22 p. en annexe.
Ritz M., Descloitres, M., Courteaud, M., Robineau, B., 1993. Etude géophysique VLF et AMT du secteur pilote du
Baril. Conseil Général de l'île de la Réunion. Multigraphié Université de St Denis de la Réunion. 49 p., 35 fig., 4
planches, 7 p. en annexe.
Descloitres, M., Ritz, M., Sylla, M., Samb, M., 1992. Etude des niveaux d'indurations du recouvrement du gisement
de phosphate de Tobène-Taïba. Méthodes géologiques et géophysiques. ORSTOM Dakar, IST Dakar.
7. Contributions diverses
¾
Participation à l’élaboration de site Internet, audiovisuel
• Descloitres, M., Dukhan, M., 1995. Quelques Méthodes de Géophysique Appliquée. Cellule audiovisuelle Dakar,
VHS, 20 mn. Public : étudiants 2ème cycle.
• Conception et réalisation du site web de l’équipe HGP au LTHE
¾
•
•
•
•
•
•
Organisation de formations / ateliers/ séminaires
4th International Workshop on Magnetic Resonance Sounding « MRS 2009 » à Grenoble, 80 participants:
membre du comité d’organisation de la conférence, logistique locale.
Co-organisateur du colloque de restitution du programme INSU « Waterscan », Autrans, LTHE, 10-12
octobre 2007, programme scientifique et logistique locale.
Organisateur des journées de formation des partenaires Burkinabé à la Résonance Magnétique des Protons,
Ouagadougou, décembre 2002.
Organisateur de la formation collective aux techniques de forage, 1 semaine, IRD Ouagadougou, 2001.
Formateur dans l’atelier « prospection électrique 2D », centre de formation continue CEFOC,
Ouagadougou, 2000.
Intervenant dans l’atelier Formation continue IRD sur le technique TDEM, Garchy, 1996, 1 jour.
8. Relectures d’articles pour revues à comité de lecture
Années
2000
2001
2002
2004
2003
2006
2007
2008
2009
2010
2009
2010
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Journaux
Revue des Sciences de l’Eau
Water International
Revue des Sciences de l’Eau
Near Surface Geophysics
Journal of Applied Geophysics
Sud-Sciences et Technologie
Hydrogeology Journal
Comptes Rendus Geosciences
Comptes Rendus Geosciences
Total :
Nombre
1
1
2
1
1
2
2
1
2
1
1
1
1
1
1
13 rang A, 6 rang B
85
ANNEXE 2
Tirés à part des principaux articles
1.
Spatialisation des aquifères
Braun, J-J., Descloitres, M., Riotte, J., Fleury, S., Barbiero, L., Boeglin, J-L., Violette, A., Lacarce, E., Ruiz, L.,
Sekhar, M., Mohan Kumar, M.S., Subramanian, S., Dupre, B., 2009. Regolith mass balance inferred from
combined mineralogical, geochemical and geophysical studies: Mule Hole gneissic watershed, South India,
Geochimica et Cosmochimica Acta, doi: 10.1016 /j.gca.2008.11.013.
Guérin R., Descloitres, M., Coudrain-Ribstein A., Talbi A., Gallaire R., 2001. Geophysical surveys for identifying
saline groundwater in the semi-arid region of the central Altiplano, Bolivia. Hydrological Processes, 15, 17, pp
3287-3301.
Legchenko, A., Descloitres, M., Bost, A., Ruiz L., Reddy, M., Girard, J-F., Sekhar, M., Mohan Kumar, M.S., Braun,
J. J. Resolution of MR Soundings applied to the characterization of hard rock aquifers. Groundwater 44(4), pp
547-554.
2.
Etude de la recharge des aquifères
Descloitres, M., Ribolzi, O., Le Troquer, Y, 2003. Study of infiltration in a gully erosion sahelian area using timelapse electrical resistivity mapping. CATENA 53, pp 229-253.
Descloitres, M., Ruiz, L., Sekhar, M., Legchenko, A., Braun, J. J., Mohan Kumar, M.S., Subramanian, S., 2008.
Characterization of seasonal local recharge using Electrical Resistivity Tomography and Magnetic Resonance
Sounding. Hydrological Processes, Vol 22, pp 384-394
Massuel, S., Favreau, G., Descloitres, M., Le Troquer, Y., Albouy, Y., Cappelaere, B. Deep infiltration through a
sandy alluvial fan in semiarid Niger inferred from electrical conductivity survey, vadose zone chemistry and
hydrological modelling. 2006. CATENA, 67 (2), pp 105-118.
3.
Etude des processus de transferts d’eau en zone non-saturée
Barbiéro L., Parate, H. R., Descloitres, M., Bost A., Furian S., Mohan Kumar M.S, Kumar C., Braun J. J., 2007.
Using a structural approach to identify relationships between soil and erosion in a semi-humid forested area,
South India. CATENA, vol. 70 , pp 313–329
Descloitres, M., Ribolzi, O., Le Troquer, Y., Thiebaux, J.P. Study of water tension differences in heterogeneous
sandy soils using surface ERT. Journal of Applied Geophysics, Vol 64/3-4, pp 83-98
4.
Apports méthodologiques
Clément, R., Descloitres, M., Günther, T., Ribolzi, Legchenko, A., 2009. Influence of shallow infiltration on timelapse ERT. Experience of advanced interpretation. Comptes Rendus Geosciences, 341, pp 886-898.
Clément, R., Descloitres, M., Günther, T., Morra, C., Oxarango, L., 2010. Artefact removal in time-lapse ERT
interpretation. Application to leachate injection experiment in landfills. Waste management, in press.
86
Available online at www.sciencedirect.com
Geochimica et Cosmochimica Acta 73 (2009) 935–961
www.elsevier.com/locate/gca
Regolith mass balance inferred from combined
mineralogical, geochemical and geophysical studies: Mule
Hole gneissic watershed, South India
Jean-Jacques Braun a,b,*, Marc Descloitres a,c, Jean Riotte a,b, Simon Fleury a,
Laurent Barbiéro a,b, Jean-Loup Boeglin b, Aurélie Violette b, Eva Lacarce d,
Laurent Ruiz e, M. Sekhar a,f, M.S. Mohan Kumar a,f,
S. Subramanian a,g, Bernard Dupré b
a
Indo-French Cell for Water Sciences (IRD/IISc Joint Laboratory), Indian Institute of Science, 560012 Bangalore, India
b
LMTG, Université de Toulouse, CNRS, IRD, OMP, 14, Avenue E. Belin, F-31400 Toulouse, France
c
IRD, Laboratoire d’Etude des Transferts en Hydrologie et Environnement (LTHE), UMR/CNRS-IRD-INPG-UJF, BP53,
F-38041 Grenoble, Cedex 09, France
d
INRA, US1106, Unité INFOSOL, 2163 Av. Pomme de Pin, F45075 Orleans Cedex 2, France
e
Sol-Agronomie-Spatialisation (SAS), UMR INRA, 65, rue de Saint-Brieuc CS 84215, F-35042 Rennes Cedex, France
f
Indian Institute of Science, Department of Civil Engineering, 560012 Bangalore, India
g
Indian Institute of Science, Department of Materials Engineering, 560012 Bangalore, India
Received 4 February 2008; accepted in revised form 4 November 2008; available online 20 November 2008
Abstract
The aim of this study is to propose a method to assess the long-term chemical weathering mass balance for a regolith developed on a heterogeneous silicate substratum at the small experimental watershed scale by adopting a combined approach of
geophysics, geochemistry and mineralogy. We initiated in 2003 a study of the steep climatic gradient and associated geomorphologic features of the edge of the rifted continental passive margin of the Karnataka Plateau, Peninsular India. In the transition sub-humid zone of this climatic gradient we have studied the pristine forested small watershed of Mule Hole (4.3 km2)
mainly developed on gneissic substratum. Mineralogical, geochemical and geophysical investigations were carried out (i) in
characteristic red soil profiles and (ii) in boreholes up to 60 m deep in order to take into account the effect of the weathering
mantle roots. In addition, 12 Electrical Resistivity Tomography profiles (ERT), with an investigation depth of 30 m, were generated at the watershed scale to spatially characterize the information gathered in boreholes and soil profiles. The location of
the ERT profiles is based on a previous electromagnetic survey, with an investigation depth of about 6 m. The soil cover thickness was inferred from the electromagnetic survey combined with a geological/pedological survey.
Taking into account the parent rock heterogeneity, the degree of weathering of each of the regolith samples has been
defined using both the mineralogical composition and the geochemical indices (Loss on Ignition, Weathering Index of Parker,
Chemical Index of Alteration). Comparing these indices with electrical resistivity logs, it has been found that a value of
400 Ohm m delineates clearly the parent rocks and the weathered materials. Then the 12 inverted ERT profiles were constrained with this value after verifying the uncertainty due to the inversion procedure. Synthetic models based on the field
data were used for this purpose. The estimated average regolith thickness at the watershed scale is 17.2 m, including
15.2 m of saprolite and 2 m of soil cover.
*
Corresponding author. Address: Indo-French Cell for Water Sciences (IRD/IISc Joint Laboratory), Indian Institute of Science, 560012
Bangalore, India.
E-mail address: braun@civil.iisc.ernet.in (J.-J. Braun).
0016-7037/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.gca.2008.11.013
936
J.-J. Braun et al. / Geochimica et Cosmochimica Acta 73 (2009) 935–961
Finally, using these estimations of the thicknesses, the long-term mass balance is calculated for the average gneiss-derived
saprolite and red soil. In the saprolite, the open-system mass-transport function s indicates that all the major elements except
Ca are depleted. The chlorite and biotite crystals, the chief sources for Mg (95%), Fe (84%), Mn (86%) and K (57%, biotite
only), are the first to undergo weathering and the oligoclase crystals are relatively intact within the saprolite with a loss of only
18%. The Ca accumulation can be attributed to the precipitation of CaCO3 from the percolating solution due to the current
and/or the paleoclimatic conditions. Overall, the most important losses occur for Si, Mg and Na with 286 106 mol/ha
(62% of the total mass loss), 67 106 mol/ha (15% of the total mass loss) and 39 106 mol/ha (9% of the total mass loss),
respectively. Al, Fe and K account for 7%, 4% and 3% of the total mass loss, respectively. In the red soil profiles, the opensystem mass-transport functions point out that all major elements except Mn are depleted. Most of the oligoclase crystals
have broken down with a loss of 90%. The most important losses occur for Si, Na and Mg with 55 106 mol/ha (47%
of the total mass loss), 22 106 mol/ha (19% of the total mass loss) and 16 106 mol/ha (14% of the total mass loss),
respectively. Ca, Al, K and Fe account for 8%, 6%, 4% and 2% of the total mass loss, respectively.
Overall these findings confirm the immaturity of the saprolite at the watershed scale. The soil profiles are more evolved
than saprolite but still contain primary minerals that can further undergo weathering and hence consume atmospheric CO2.
Ó 2008 Elsevier Ltd. All rights reserved.
1. INTRODUCTION
Understanding the relative controls of forcing factors on
the long-term silicate chemical weathering rates and the
associated atmospheric CO2 consumption remains a major
challenge (White and Brantley, 1995; Kump et al., 2000).
Several publications based on small to medium granitogneissic watershed studies (1–100 km2) examined the
relationships between temperature and runoff for different
climatic and tectonic settings (Bluth and Kump, 1994;
White and Blum, 1995; White et al., 1999). The authors
stressed that the silicate weathering rates were not governed
by any single parameter. In addition to climate, the importance of the thickness and nature of the blanket of loose
and transportable weathered material, namely regolith,
which overlies the intact bedrocks, was also recently invoked, especially in the tropical environment (Millot
et al., 2002; Oliva et al., 2003; Braun et al., 2005; West
et al., 2005).
At the watershed scale the regolith cover is produced
either by in situ weathering or by deposition (downslope
colluviums and valley-floor alluviums) (Taylor and Eggleton, 2001). However, as the ubiquitous terms saprolite
and soil used to describe the regolith compartments from
bottom to top have often various meanings in the literature
because of their trans-disciplinary usage (Ehlen, 2005;
Dethier and Lazarus, 2006; Dewandel et al., 2006) it is germane to give here straightaway consensual definitions. The
saprolite corresponds to the lower part of in situ regolith
covers. It develops downward (weathering front) at the expense of the underlying fractured parent rock from which it
does retain the structure and the fabric, i.e. isovolumetric
weathering. The soils develop at the expense of either saprolite or colluviums/alluviums at the uppermost part of
the regolith where the perturbation brought by both physical and biological processes lead to (i) the differentiation
into horizons and (ii) the loss of the existing isovolumetric
weathering features.
The regolith thickness depends on the balance between
deepening at the weathering front by chemical weathering
and skimming off by mechanical erosion at the topsoil
(Riebe et al., 2003; Anderson et al., 2007; Burke et al.,
2007, and references therein). Chemical weathering rates
are highly sensitive to the availability of fresh mineral surfaces, which would tend to be enhanced by increased physical erosion. A thin, immature regolith still containing a
large amount of primary minerals able to weather, would
increase the chemical weathering flux while a thick, mature
regolith poor in weatherable primary minerals (e.g. strongly
depleted lateritic cover) would slow it down (Oliva et al.,
2003). Regolith characterization is therefore of fundamental interest to improve the model of the groundwater flow
paths and to assess the long-term geochemical mass balance. Regardless, the assessment of the three-dimensional
structure of regolith is still challenging (Thomas, 1994;
Taylor and Eggleton, 2001; Anderson et al., 2004, 2007).
Understanding where chemical weathering takes place
within a landscape remains a critical missing piece in the
complicated puzzle of this fundamental Earth surface
process.
Until now, only a few integrated watershed approaches
have been carried out in tropical regions. The two most
studied sites in terms of contemporary and long-term chemical silicate weathering have focused on humid tropics (i) in
the Luquillo mountain tropical forest, Puerto Rico (Rio
Icacos site, Water, Energy and Biogeochemical Budget;
http://pr.water.usgs.gov/public/webb/) and (ii) in the South
Cameroon plateau Nsimi site, developed as part of the project ‘Observatoire de Recherche en Environnement – Bassin
Versant Expérimentaux Tropicaux, http://bvet.ore.fr/.
Both sites are characterized by rather thick regolith, i.e.
larger than 5 m, still rich in weatherable minerals in the
first case, while strongly depleted in the second case
(laterites).
A study of the steep climatic gradient and associated
geomorphologic features of the Western Ghâts rain shadow
located on the edge of the rifted continental passive margin
of the Karnataka Plateau, Peninsular India was initiated in
2003 (Gunnell and Bourgeon, 1997; Gunnell, 1998a,b,
2000; Gunnell et al., 2003, 2007). This combined gradient
from humid to semi-arid provides unique conditions to
study the shift from deep mature to shallow immature regolith covers as well as the influence of their thickness and
nature on the silicate chemical weathering.
Regolith mass balance in a gneissic watershed, South India
937
We started our investigations on the pristine forested
Mule Hole Small Experimental Watershed, SEW
(4.3 km2). The first results were published on the soil distribution and erosion processes (Barbiéro et al., 2007), on the
performance of Magnetic Resonance Sounding method applied to the hard-rock aquifer (Legchenko et al., 2006) and
on the seasonal local recharge processes at the stream outlet
(Descloitres et al., 2007).
The present paper focuses on the nature and the degree
of weathering and the thickness of the regolith developed
on the heterogeneous silicate substratum of the Mule Hole
SEW. Our key issues are (i) to geochemically distinguish between fresh rocks and weathered materials since the initial
parent rock composition is found to be variable, (ii) to spatially characterize the information at the watershed scale
and (iii) to evaluate the long-term weathering mass balance.
We have adopted a combined approach using geophysics,
geochemistry and mineralogy.
First, the gneissic protolith is differentiated from its
weathering products by comparing geochemical and mineralogical compositions with electrical resistivity at the sample scale. Second, a detailed non-destructive Electrical
Resistivity Tomography (ERT) survey is performed on
the watershed to produce representative ERT profiles that
give resistivity distribution from the sub-surface down to
30 m. ERT uncertainty is then analyzed using a synthetic
modeling approach that allows us to spatially characterize
the distribution of soil and saprolite at the watershed scale.
Finally, the volume of the weathered material is assessed
and the main saprolitization and soil chemical processes
are deciphered by a mass balance approach based on the
ERT mapping. The impact on long-term chemical weathering rates at the watershed scale are then discussed. Forthcoming companion papers will address the contemporary
chemical and erosion fluxes based on climatic, hydrological,
hydrogeological and geochemical time series and long-term
denudation rates determined with cosmogenic nuclides.
Dharwar craton (Naqvi and Rogers, 1987), is dominated
by complexly folded, heterogeneous Precambrian peninsular gneiss intermingled with mafic and ultramafic rocks of
the volcano-sedimentary Sargur serie (Shadakshara Swamy
et al., 1995). The Peninsular gneiss represents at least 85%
of the watershed basement. The gneiss foliation is mainly
oriented at N75° in the Northern part and at N100°–
N120° in the Southern part. The dip angle of the gneissic
units ranges from 75° to the vertical. Usually mafic and
ultramafic rocks (hornblendite, amphibolite and serpentinite) come into sight as metric enclaves or seams intermingle with the gneiss layers. However an amphibolite body
occurs in the southeastern part of the watershed and represents roughly 7% of the whole watershed area. At outcrop
level, the basement rocks appear more or less fissured.
The soil cover of the watershed has been mapped by
Barbiéro et al. (2007) based on the FAO terminology
(IUSS-Working-Group-WRB, 2006). The gneissic saprolite, cohesive to loose sandy, crops out both in the streambed and at the mid-slope in approximately 22% of the
watershed area. Shallow red soils (Ferralsols and Chromic
Luvisols) from 1 to 2 m in depth cover 66% of the whole
watershed area (hillslopes). A stone line, composed of
quartz pebbles and ferruginous nodules, often occurs at
the boundary between the topsoil and saprolite. The total
area covered by black soil is 12%. The lower part of the
slope and the flat valley bottoms are covered by, on average, 2 m of black soils (Vertisols and Vertic intergrades).
They are developed on both the gneiss and the mafic rocks.
The other occurrence of the black soil is lithodependant
with development of deeper soils (2.5 m) on gneissic zones
rich in amphibolite layers located in the depressions on
the crest line.
2. FIELD SETTINGS
Previous soil and geological maps at the Mule Hole watershed scale are based on the observations of the parent
rock and the saprolite outcrops and the electromagnetic
surveys using a GeonicsÒ EM31 instrument coupled with
both structural approach on a selected soil catena and an
auger survey (Barbiéro et al., 2007). The EM31 equipment
measures the electrical conductivity in milliSiemens per meter (mS/m) with a penetration depth typically ranging from
4 to 6 m (McNeill, 1980). Conductivities below 2.5 mS/m
correspond to fresh rock occurrence between surface and
2 m depth. Values between 2.5 and 10 mS/m are characteristic of red soils, and above 10 mS/m of black soils or
weathered amphibolite (Fig. 2).
Red soil samples developed on gneiss were collected in
both sites S1 and S2 (Fig. 2). S1 is located downslope close
to the streambed and including the T1 soil catena described
in Barbiéro et al. (2007) while S2 is located upslope on the
North ridge crest. Two soil profiles, namely S1-P and S2-P
were sampled down to the top of the saprolite (19 samples).
The profiles S1-P and S2-P are 3.2 and 2.4 m thick, respectively. Fourteen soil samples (S1-T1) were also collected at
different depths in the T1 soil catena (Barbiéro et al., 2007).
The Mule Hole SEW (11°430 N–76°260 E) lies in the
sub-humid zone of the climatic gradient of the Kabini river
basin (Fig. 1). The morphology of the watershed is highly
incised by the temporary stream network. The edge slope
is relatively low with small depressions. The slope convexity
is high upslope and concave by the streambeds. The streambeds are steep-sided up to 2 m down compared to valley
floor.
The average annual rainfall at the Mule Hole SEW is
1090 ± 230 mm/yr with a dry season lasting an average of
5.5 months. The average annual air temperature is
21.8 °C. The watershed is covered by dry deciduous forest
with different facies linked to the soil distribution (Barbiéro
et al., 2007). Currently, the Mule Hole SEW is dedicated to
wildlife and biodiversity preservation (Bandipur National
Park).
The Mule Hole protolith presents high concentration of
lithological, structural and compositional heterogeneities,
which favor the water circulation and therefore the weathering processes. The lithology, representative of the West
3. MATERIALS AND METHODOLOGY
3.1. Previous studies and sampling
938
Fig. 1. Location of the Kabini river basin and the Mule Hole experimental watershed. The shaded area represents the boundaries of the subhumid zone with the 900 and 1500 mm/yr isohyets.
The fresh gneiss samples were collected in the Mule Hole
streambed.
The deep regolith was studied through a network of thirteen boreholes (BH1–13) distributed on the watershed
edges along the main roads (Fig. 2). Composite samples
(i.e. cuttings) of saprolite and of protolith were collected
for every 2 m along the depth of the eight boreholes
(BH1–2–3–4–5–6–12–13). All boreholes were drilled in the
gneissic basement but BH6 is mainly in the amphibolite
body. Borehole depths range between 20 and 60 m.
3.2. Protolith/regolith geochemistry and mineralogy
The mineralogy of 157 powdered composite samples of
boreholes BH1–2–3–4–5–6–12–13 was determined by
X-ray diffraction (XRD) at LMTG (Laboratoire des
Mécanismes de Transfert en Géologie, Toulouse). Thin sections were also prepared from outcrop samples of gneiss
and BH6 amphibolite. They were observed with optical
microscope and SEM coupled with backscattered electrons
and EDX. Major and accessory minerals were analyzed
939
Fig. 2. Results of the electromagnetic survey (EM) and location of the 12 ERT profiles. The colored background is the soil electrical
conductivity measured with electromagnetic devices (EM31) along N–S oriented profiles implemented every 100 m on the watershed (Barbiéro
et al., 2007). ERT profiles are implemented to sample the main pedological units deduced from electrical conductivity distribution and
pedological survey. Shaded areas indicate the zones of occurrence of seams and enclaves of amphibolite mingled with gneiss. (For
interpretation of the references to color in this figure legend, the reader is referred to the web version of this paper.)
with a SX100 Cameca microprobe at the LMTG. Bulk densities (q) were determined by the paraffin method with a
SartoriusÒ density kit (10 replicates).
Bulk chemical analyses were carried out on BH1, BH5,
BH6 and BH12 (109 samples) at the SARM (Centre de
Recherche Pétrographique et Géochimique-CNRS,
Vanduvre-lès-Nancy). After LiBO2 fusion and HNO3 dissolution, Si, Al, Fe, Mn, Mg, Ca, Na, K, Ti, P were analyzed by ICP-AES and Zr, Th and Nb by ICP-MS. The
detection limits (in wt%) are 0.8 for SiO2, 0.3 for Al2O3,
0.1 for Fe2O3, 0.03 for MnO, 0.4 for MgO, 0.5 for CaO,
0.08 for Na2O, 0.05 for K2O, 0.09 for TiO2 and 0.2 for
P2O5 and (in ppm) 0.5 for Zr, Nb and Th. The detection
limit of the Loss on Ignition (LoI), the measure of volatile
H2O, CO2, F, Cl and S, is found to be 0.02 wt%.
3.3. Geophysical investigations
The bedrock and regolith cover was studied at the watershed scale through Direct Current electrical methods
using a SYSCAL R2 resistivitimeter from IRIS Instruments
(Descloitres et al., 2007). Our study benefitted from previous attempts to assess the geometry of regolith in the Tropics using these geophysical methods (Robain et al., 1996;
Beauvais et al., 1999, 2004, 2007). Electrical resistivity of
the regolith varies with porosity (bulk density), amount of
clay minerals, temperature and both the water content
and the salinity (Telford et al., 1990). These parameters
make the electrical resistivity convenient for characterizing
the regolith since it presents a lower density than the protolith, as well as significant clay occurrence and water content. Nevertheless, the separation limit between regolith
and protolith cannot be based on uncalibrated resistivity
measurements alone as the resistivity of the regolith can
be site-specific. Three complementary electrical methods
were carried out on the watershed:
(i) Resistivity measurements on typical protolith outcrops and soils. For this, a Wenner array with an
electrode spacing of 0.20 m was used. The resistivity
calculated using such a small array is considered as
the true resistivity of the medium.
(ii) Resistivity logging in boreholes BH5, 6, 12 and 13
with a pole–pole (also called log ‘‘normal”) array
with 0.30 m spacing between electrodes and measurement for each 0.25–0.5 m. The logging was carried
out just after drilling, before casing when possible,
with an inflatable probe in the vadose zone (Descloitres and Le Troquer, 2004) and with steel electrodes
below the water table. The resistivity calculated using
such a small electrode spacing is considered as the
true resistivity of ground around the probe.
(iii) 2D Electrical Resistivity Tomography (ERT) survey
(Loke, 2000; Seaton and Burbey, 2002) to investigate
the first 30 m of the sub-surface. ERT was carried out
with two geometric arrays. The first one is the Wenner array, more sensitive to the vertical variations of
the electrical resistivity (Loke, 2000). The second
array is the dipole–dipole, more sensitive to the lateral variations of the electrical resistivity. The latter
is particularly suitable in fractured hard rock studies
(Seaton and Burbey, 2002) because of the 2D distribution of resistivity in such a medium. Twelve ERT
profiles totaling 7600 m were setup according to the
940
topographical, EM31 and geological surveys (Fig. 2).
ERT profiles 1, 2, 3, 4, 7, 8, 12 and 9 are located
above gneissic basement, whereas profiles 5, 10, 11
and 6 are mostly above the mingled amphibolitegneiss basement. Profiles 1, 2, 3 and 4 focused on
the outlet area and crossed two patches of black soils.
For comparison with true resistivity logging in boreholes, ERT profiles 1 and 2 are situated near BH1
and BH12, ERT profile 12 near BH5 and ERT profile
11 near BH6.
4. RESULTS
4.1. Boreholes and soil profiles
Table 1 displays the bulk analyses of major and selected
trace inert elements (Zr, Th, Nb), borehole electrical resistivity and mineralogy of major minerals for the composite
samples of boreholes BH6, BH1, BH5 and BH12. Table 2
shows the same information for the soil samples of the S1
and S2 sites. The compositions of the main gneiss minerals,
in wt%, are reported in Table 3.
4.1.1. Fresh gneiss and weathering products
Combined XRD patterns, SEM–EDX observations and
microprobe analyses indicate that the major minerals of
gneiss are quartz, oligoclase (An14), sericite, biotite and
chlorite. The accessory minerals are apatite, epidote, allanite, titanite, magnetite, ilmenite, pyrite and zircon.
Fig. 3a shows the fine layering of the gneiss with both leucosome and melanosome. Sericitization of oligoclase crystals and chloritization of biotite crystals are frequent
(Fig. 3b–d). Titanite and apatite crystals are closely associated with the presence of biotite (Fig. 3e). The zircon
crystals observed in thin sections are particularly tiny
(10–15 lm in length) (Fig. 3f). The gneiss is veined with
epidote-rich quartz seams of hydrothermal origin.
The bulk chemical compositions of the fresh to weathered gneiss reflect both the primary and the secondary mineralogical variability. The parent gneiss composition varies
from felsic (oligoclase/quartz-rich) to mafic (biotite/chlorite-rich) end-members. Gneiss dominates in boreholes 1,
2, 3, 4, 5, 12 and 13 and is present at a depth of 58 m in
BH6. Significant clay mineral contents (kaolinite, smectite)
occur up to a depth of 4, 20, 5, 15, 22, 14, 15 m in BH1, 2, 3,
4, 5, 12 and 13, respectively.
The range of the chemical compositions for fresh rocks
and saprolite in the borehole samples was calculated without taking into account the near surface samples (less than
4 m in depth) because of the blend of soil layers with saprolitic materials. SiO2 ranges between 50 and 76 wt%, Al2O3
between 9 and 16 wt%, Fe2O3 between 1.3 and 16.0 wt%,
MnO between the detection limit and 0.2 wt%, MgO between 0.3 and 8.5 wt%, CaO between 0.2 and 9.7 wt%,
Na2O between 0.6 and 6.6 wt%, K2O between 0.2 and
9.7 wt%, TiO2 between 0.2 and 2.2 wt%, P2O5 between
the detection limit and 0.5 wt% and LoI between 1.0 and
6.8 wt%. Negative correlations, with coefficient r2 P 0.6
(n = 63) exist (i) between SiO2 and Fe2O3, MnO, MgO
and CaO and (ii) between Fe2O3 and Na2O. Positive correlations exist (i) between Al2O3 and Na2O, (ii) between
Fe2O3 and MnO and MgO (iii) between MnO and CaO
and (iv) between TiO2 and P2O5. The borehole electrical
resistivity varies from 50 to 4500 Ohm m.
In the soil samples derived from the gneiss (S1-T1, S1P and S2-P), SiO2 ranges between 61.9 and 78.8 wt%,
Al2O3 between 8.5 and 16.7 wt%, Fe2O3 between 1.6 and
6.9 wt%, MnO between the detection limit and 0.13 wt%,
MgO between 0.3 and 1.5 wt%, CaO between 0.5 and
2.1 wt%, Na2O between 0.4 and 4.5 wt%, K2O between
0.6 and 3.2 wt%, TiO2 between 0.1 and 0.7 wt%, P2O5 between the detection limit and 0.1 wt% and LoI between 2.3
and 11.6 wt%. Negative correlations, with coefficient
r2 P 0.6 (n = 32) exist (i) between SiO2 and Al2O3,
Fe2O3 and LoI and (ii) between Na2O and TiO2 and
LoI. Positive correlations exist between Fe2O3 and MnO,
TiO2 and LoI. The electrical resistivity varies from 10 to
100 Ohm m.
4.1.2. Fresh and weathered amphibolite (BH6)
The major minerals of the mafic body are labradorite,
Mg-hornblende, tremolite and chlorite. Fig. 4a and b portray a fractured seam of amphibolite and the corresponding
minor and accessory mineral assemblage as Mg-rich calcite,
serpentine and iron oxides containing Cr and Ti. The occurrence of quartz and epidote in the BH6 samples detected by
XRD patterns suggests that the amphibolite is also veined
with hydrothermal seams. SiO2 ranges between 46.3 and
51.1 wt%, Al2O3 between 10.9 and 16.0 wt%, Fe2O3 between 10.4 and 20.4 wt%, MnO between 0.2 and 0.3 wt%,
MgO between 3.2 and 9.6 wt%, CaO between 2.0 and
10.7 wt%, Na2O between 0.8 and 3.6 wt%, K2O between
0.1 and 0.6 wt%, TiO2 between 0.7 and 1.7 wt%, P2O5 between 0.1 and 0.3 wt% and LoI between 1.5 and
14.1 wt%. Two positive correlations, with coefficient
r2 P +0.6 (n = 35), exist between Fe2O3 and TiO2 and between TiO2 and P2O5. The borehole electrical resistivity
varies from 10 to 10,000 Ohm m. In the borehole samples,
large amounts of clay minerals occur between 10 and
12 m in BH6. However, observation carried out in pit
shows that the first 3 m are composed of rock with a millimetric to centimetric fissured network filled with loose
clayey materials that do not appear in the composite
XRD analysis. There is no soil horizon topping this
saprolite.
4.2. ERT profiles
Twelve ERT profiles were analyzed and a routine inversion method was applied to the apparent resistivity field
data. The distribution of the calculated resistivity is displayed along profiles between surface and 30 m depth
(Fig. 5). Calculated resistivity ranges from 10 Ohm m near
the surface to more than 5000 Ohm m downward with a
strong lateral variability, showing high amplitude corrugations. Such complex geometry prevents any simple estimate
of regolith thickness based on the resistivity alone: the resistivity limit between weathered and fresh rocks will be determined in the following section.
Table 1
Bulk chemical analyses for major and selected trace inert elements (Zr, Th, Nb), electrical resistivity and mineralogy of major minerals based on XRD patterns for the composite samples of
boreholes BH6, BH1, BH5 and BH12 and the soil samples of the S1 and S2 sites. Chemical Index of Alteration (CIA) and Weathering Index of Parker (WIP) are also mentioned. CIA is defined
with molecular proportion of major element oxides by CIA = 100[Al2O3/(Al2O3 + CaO* + Na2O + K2O)] with CaO* = CaO 10/3P2O5; CaO is restricted to that derived from silicate minerals.
WIP is calculated with the atomic proportion of Na, Mg, K and Ca divided by weighting factors corresponding to the bond strengths of the elements with oxygen: WIP = 100[(Na/0.35) + (Mg/
0.90) + (K/0.25) + (Ca/0.70)]. The groundwater table level and the conductivity are indicated in each borehole. Key for XRD analysis: empty cell: absence, x: presence (<5%), xx: abundant and
xxx: very abundant.
Depth, m
SiO2, %
Al2O3, %
Fe2O3, %
MnO, %
MgO, %
CaO, %
Na2O, %
K2O, %
TiO2, %
P2O5, %
LoI, %
Total, %
Zr, ppm
Th, ppm
Nb, ppm
BH6
BH6
BH6
BH6
BH6
BH6
BH6
BH6
BH6
BH6
BH6
BH6
BH6
BH6
BH6
BH6
BH6
BH6
BH6
BH6
BH6
BH6
BH6
BH6
2.0
4.0
6.0
8.0
10.0
12.0
14.0
16.0
18.0
20.0
22.0
24.0
26.0
28.0
30.0
32.0
34.0
36.0
38.0
40.0
42.0
44.0
46.0
48.0
48.78
51.15
49.37
46.99
49.65
50.74
51.09
50.44
48.97
50.72
49.21
50.84
49.99
50.43
50.68
47.06
47.27
49.53
49.78
50.31
49.71
50.62
50.51
50.86
10.87
11.02
13.53
14.15
12.89
12.78
13.23
13.65
13.50
13.28
13.92
13.41
13.22
13.53
13.31
11.01
13.27
12.93
12.16
13.30
13.50
15.99
13.74
14.65
15.67
14.84
15.81
15.36
14.79
14.95
15.21
16.42
15.52
16.81
16.69
15.36
16.32
15.33
14.63
20.38
17.08
17.02
13.88
11.64
12.71
10.43
12.91
11.77
0.23
0.23
0.23
0.20
0.28
0.19
0.22
0.27
0.23
0.27
0.28
0.24
0.19
0.18
0.20
0.24
0.21
0.20
0.19
0.19
0.17
0.18
0.17
0.16
7.11
7.33
5.20
6.08
3.21
3.63
5.12
5.02
4.92
4.16
4.47
4.17
6.30
5.50
5.60
6.61
7.01
5.29
9.60
7.05
7.05
3.95
7.60
4.87
10.10
8.67
8.20
8.52
2.04
3.68
6.83
8.56
10.41
6.51
10.54
10.60
5.74
7.70
9.11
6.76
8.14
7.13
7.90
10.44
9.92
8.81
10.17
7.17
0.82
2.02
2.26
2.73
1.28
1.65
1.97
2.08
2.09
2.27
2.45
2.33
2.71
3.33
2.95
1.32
3.20
3.60
2.18
2.63
2.73
3.52
2.42
2.35
0.58
0.23
0.17
0.24
0.10
0.11
0.14
0.26
0.14
0.28
0.22
0.15
0.38
0.39
0.27
0.65
0.32
0.45
0.32
0.20
0.31
0.62
0.19
0.10
0.96
1.12
1.30
1.03
1.21
1.22
1.16
1.33
1.22
1.38
1.35
1.27
1.30
1.30
1.26
1.62
1.55
1.67
0.80
0.76
0.74
1.00
0.81
0.89
0.07
0.09
0.13
0.10
0.10
0.15
0.13
0.13
0.12
0.13
0.12
0.12
0.13
0.12
0.13
0.26
0.14
0.18
0.08
0.06
0.09
0.10
0.08
0.10
5.12
3.03
2.74
4.19
14.11
10.76
4.70
1.85
2.53
3.74
1.46
1.73
4.50
2.32
2.85
4.50
2.49
2.69
3.11
2.77
3.66
5.49
1.93
6.58
100.30
99.72
98.94
99.57
99.67
99.85
99.81
100.01
99.65
99.55
100.71
100.21
100.77
100.11
100.99
100.40
100.68
100.69
100.00
99.34
100.58
100.73
100.54
99.50
56
58
86
56
82
85
79
81
87
86
84
74
76
82
72
84
85
96
36
47
41
60
53
68
0.4
0.3
0.8
0.3
0.5
0.5
0.6
0.6
0.5
0.5
0.4
0.5
0.3
0.6
0.4
0.4
0.4
0.5
0.2
0.2
0.2
0.3
0.2
0.3
1.7
1.7
3.1
2.2
2.4
2.3
2.3
2.7
2.7
2.7
2.5
2.3
2.4
2.4
2.4
3.7
3.6
3.8
1.6
1.6
1.6
2.7
2.2
2.4
BH6
BH6
BH6
BH6
BH6
50.0
52.0
54.0
56.0
58.0
46.35
49.52
48.67
57.78
67.22
14.07
14.00
14.13
16.69
15.93
11.41
11.40
13.23
7.01
3.57
0.16
0.19
0.20
0.14
0.06
5.42
5.94
5.91
2.40
1.37
9.13
10.70
10.53
7.79
4.37
2.32
2.25
2.46
4.21
5.18
0.11
0.14
0.24
0.54
0.88
0.89
0.83
1.10
0.83
0.41
0.08
0.08
0.12
0.09
0.11
8.95
4.08
4.04
1.95
1.14
98.90
99.11
100.63
99.42
100.24
59
58
80
56
86
0.3
0.3
0.4
1.2
2.2
2.4
2.1
3.3
2.7
1.6
Borehole
Depth, m
WIP
CIA
ER, Ohm m
BH6
BH6
BH6
BH6
BH6
BH6
2.0
4.0
6.0
8.0
10.0
12.0
58
63
57
65
26
35
35
37
42
41
69
58
10
24
45
11
8
19
Quartz
Oligoclase
XXX
XXX
XX
XX
XX
X
XXX
XXX
X
X
XX
XX
XX
XX
Anorthite
Biotite
Sericite
Epidote
Mg-hornblende
X
X
XXX
XXX
XXX
XXX
X
X
X
Tremolite
XXX
X
XX
Chlorite
Calcite
X
X
X
(continued on next page)
941
Clays 2:1–1:1
Borehole
942
Table 1 (continued)
Borehole Depth, m WIP CIA ER, Ohm m Clays 2:1–1:1 Quartz Oligoclase Anorthite Biotite Sericite Epidote Mg-hornblende Tremolite Chlorite Calcite
14.0
16.0
18.0
20.0
22.0
24.0
26.0
28.0
30.0
32.0
34.0
36.0
38.0
40.0
42.0
44.0
46.0
48.0
50
57
60
51
63
61
60
68
68
52
72
69
69
72
72
71
71
54
46
42
38
46
38
37
47
41
38
43
40
41
40
36
37
42
38
47
162
1755
206
7547
8532
227
1416
3907
na
na
na
6329
6071
6461
3829
2654
3047
1450
XXX
XXX
XXX
XXX
XX
XXX
XXX
XXX
XXX
XXX
XX
XXX
XX
XX
XXX
XXX
XX
XXX
BH6
BH6
BH6
BH6
BH6
50.0
52.0
54.0
56.0
58.0
60
65
67
70
70
41
38
38
44
48
5996
10,222
15,264
17,038
na
XXX
XXX
XXX
XXX
XXX
Borehole
Depth, m
SiO2, %
Al2O3, %
BH5
BH5
BH5
BH5
BH5
BH5
BH5
BH5
BH5
BH5
BH5
BH5
BH5
BH5
BH5
BH5
BH5
BH5
2.0
4.0
6.0
8.0
10.0
12.0
14.0
16.0
18.0
20.0
22.0
24.0
26.0
28.0
30.0
32.0
34.0
36.0
74.28
70.29
72.42
74.81
67.86
73.63
64.89
71.64
68.80
68.99
63.86
63.56
62.07
71.68
74.00
70.93
71.17
68.83
14.00
16.34
13.57
14.57
15.03
13.42
13.48
13.50
12.53
15.14
11.43
15.10
14.16
13.53
13.93
13.96
13.99
13.38
Fe2O3, %
4.05
4.31
4.93
1.91
4.70
3.91
8.47
4.23
5.99
3.32
8.13
7.16
7.49
3.69
2.64
4.05
3.47
4.57
X
XXX
XXX
XX
X
XX
XXX
XX
X
XX
XX
XX
XX
XX
X
XXX
XX
X
X
XX
X
X
XX
XX
X
X
X
X
XXX
X
XX
MnO, %
MgO, %
0.60
0.50
1.17
0.68
1.84
0.96
2.33
1.53
2.20
1.65
6.57
3.16
3.79
1.60
1.03
1.64
1.44
1.63
XXX
XXX
XX
XX
XX
XX
X
X
X
X
X
X
X
X
X
X
X
X
X
X
XXX
XXX
X
XXX
XXX
XX
X
XXX
XXX
XX
X
XX
X
X
XXX
X
XXX
XXX
XXX
X
X
XX
XXX
XX
XX
XXX
XXX
XXX
XX
XXX
XXX
XX
XXX
XXX
XXX
XXX
X
XX
<L.D.
<L.D.
<L.D.
<L.D.
0.04
<L.D.
0.04
<L.D.
0.05
0.03
0.07
0.07
0.07
0.04
<L.D.
0.03
0.03
0.06
XXX
X
XX
X
X
XX
XXX
XXX
XXX
XXX
X
Groundwater table
XX
X
600–700 lS cm1
X
CaO, %
Na2O, %
K2O, %
TiO2, %
P2O5, %
LoI, %
Total, %
Zr, ppm
Th, ppm
Nb, ppm
0.11
0.16
0.15
0.33
0.18
0.66
0.37
0.79
0.65
0.57
1.46
2.31
1.71
0.82
1.51
1.52
2.86
1.79
1.10
2.59
4.15
3.62
3.31
2.64
3.19
4.30
5.95
2.13
4.84
3.72
4.49
4.88
4.27
4.59
5.20
2.11
1.95
2.21
1.55
1.63
1.74
1.59
2.11
1.14
1.08
1.40
1.55
2.54
1.88
1.92
2.35
2.28
0.95
0.34
0.35
0.46
0.18
0.50
0.37
0.80
0.46
0.61
0.31
0.36
0.59
0.84
0.50
0.20
0.47
0.38
0.40
0.04
0.04
0.04
<DL
0.04
0.05
0.06
0.05
0.13
0.12
0.07
0.15
0.14
0.18
0.04
0.15
0.09
0.09
3.57
5.88
3.43
2.86
4.72
3.11
5.90
3.40
3.78
3.04
5.50
3.37
3.42
1.68
1.36
1.62
1.64
2.45
100.77
100.88
100.98
100.85
100.31
100.66
100.86
100.48
100.32
100.29
100.08
101.00
100.55
100.97
100.82
100.99
100.59
100.42
728
610
476
184
334
661
469
521
377
213
85
137
257
257
139
236
243
162
10.6
15.7
10.3
14.9
8.0
9.1
5.8
10.1
5.3
8.2
1.9
1.7
6.1
9.7
9.7
11.7
7.9
2.3
22.3
10.8
12.7
4.3
11.4
15.3
19.2
13.7
14.0
6.3
10.1
13.7
17.1
8.1
7.6
14.3
6.4
8.5
BH6
BH6
BH6
BH6
BH6
BH6
BH6
BH6
BH6
BH6
BH6
BH6
BH6
BH6
BH6
BH6
BH6
BH6
BH5
BH5
BH5
BH5
BH5
BH5
BH5
38.0
40.0
42.0
44.0
46.0
48.0
50.0
72.93
64.99
70.88
57.79
74.43
73.91
76.00
14.14
14.36
13.39
11.01
14.44
13.85
12.90
2.86
5.94
4.26
11.33
1.77
1.86
1.70
0.04
0.05
0.03
0.10
0.04
<L.D.
<L.D.
1.29
3.11
1.84
8.53
0.78
0.80
0.61
0.71
1.20
0.85
1.46
0.48
0.96
0.88
5.31
4.85
5.32
1.77
5.57
5.36
4.71
1.82
2.53
1.33
2.51
1.58
1.38
1.46
0.29
0.59
0.46
0.62
0.18
0.20
0.22
0.08
0.14
0.08
0.11
0.05
0.08
0.06
1.36
2.22
1.86
4.29
1.11
1.04
1.08
100.83
99.96
100.30
99.52
100.41
99.43
99.63
121
175
268
140
137
380
405
14.6
6.5
9.7
17.7
7.2
9.8
32.0
9.2
11.9
10.4
15.0
3.9
5.3
5.1
BH5
BH5
BH5
52.0
54.0
56.0
72.71
68.90
72.77
14.06
10.85
13.89
2.58
6.89
2.36
<L.D.
0.07
<L.D.
0.92
5.95
1.00
0.93
0.73
0.72
4.93
2.31
5.39
1.78
2.03
1.62
0.30
0.29
0.19
0.06
0.06
0.03
1.41
2.48
1.04
99.68
100.54
99.00
258
310
256
9.8
16.2
14.8
11.3
9.3
3.7
2.0
4.0
6.0
8.0
10.0
12.0
14.0
16.0
18.0
20.0
22.0
24.0
26.0
28.0
30.0
32.0
34.0
36.0
38.0
40.0
42.0
44.0
46.0
48.0
50.0
36
28
46
54
53
48
46
52
57
70
51
70
72
66
66
67
69
67
69
77
67
65
68
66
60
73
80
66
62
65
64
65
63
57
56
66
56
53
53
55
54
53
48
55
53
54
58
55
54
54
142
106
66
56
108
75
93
81
92
80
127
184
395
490
na
na
na
na
na
579
1125
1565
1759
1585
3247
BH5
BH5
BH5
52.0
54.0
56.0
65
57
68
55
60
54
2102
2959
3450
X
X
X
X
X
X
X
X
X
X
X
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
X
X
XX
XXX
XX
XX
XX
XX
XXX
XXX
XX
XXX
XXX
XXX
XXX
XX
XXX
XXX
XXX
XXX
XXX
X
XXX
XXX
XX
XXX
XXX
XXX
XXX
XX
XXX
X
X
X
X
X
X
XX
X
X
X
X
XX
XX
X
X
X
X
X
X
X
X
X
X
X
X
XX
X
X
X
X
XX
X
X
X
X
XX
XX
X
X
X
XX
X
X
XX
X
XXX
XX
X
X
X
X
XX
X
X
X
XX
X
X
X
BH5
BH5
BH5
BH5
BH5
BH5
BH5
BH5
BH5
BH5
BH5
BH5
BH5
BH5
BH5
BH5
BH5
BH5
BH5
BH5
BH5
BH5
BH5
BH5
BH5
X
XX
X
X
XX
Groundwater table
600–700 lS cm1
XX
X
X
Depth, m
SiO2, %
Al2O3, %
Fe2O3, %
MnO, %
MgO, %
CaO, %
Na2O, %
K2O, %
TiO2, %
P2O5, %
LoI, %
Total, %
Zr, ppm
2.0
4.0
71.32
69.22
10.47
15.92
4.75
1.97
0.13
<L.D.
1.04
0.64
1.29
1.56
1.37
6.32
0.64
1.31
0.49
0.24
0.04
0.09
7.47
1.64
99.01
98.90
254
114
Th, ppm
Nb, ppm
10.3
5.5
5.1
3.5
943
Borehole
BH1
BH1
944
Table 1 (continued)
Depth, m
SiO2, %
Al2O3, %
Fe2O3, %
MnO, %
MgO, %
CaO, %
Na2O, %
K2O, %
TiO2, %
P2O5, %
LoI, %
Total, %
Zr, ppm
Th, ppm
Nb, ppm
BH1
BH1
BH1
BH1
BH1
BH1
BH1
BH1
BH1
BH1
BH1
BH1
BH1
BH1
BH1
BH1
BH1
6.0
8.0
10.0
12.0
14.0
16.0
18.0
20.0
22.0
24.0
26.0
28.0
30.0
32.0
34.0
36.0
38.0
70.36
69.29
70.45
70.85
67.45
72.35
71.88
68.41
50.19
68.08
62.91
63.86
71.69
53.02
53.11
53.14
57.21
16.29
16.16
16.47
15.81
15.52
14.61
14.72
13.27
14.13
13.57
11.82
14.14
14.24
9.74
10.57
10.52
9.03
1.74
2.25
1.59
2.25
2.06
1.58
1.30
3.57
8.31
3.90
6.40
3.98
2.89
11.71
15.83
16.04
10.98
<L.D.
<L.D.
<L.D.
<L.D.
<L.D.
<L.D.
<L.D.
0.04
0.10
0.04
0.08
0.07
0.03
0.19
0.20
0.20
0.22
0.47
0.86
0.36
0.34
0.91
0.60
0.46
2.36
4.99
2.61
6.41
1.90
1.57
4.96
5.64
5.73
2.61
0.91
0.87
0.68
0.47
1.86
0.88
1.74
2.20
7.27
1.98
2.72
3.98
1.61
8.88
5.56
5.41
9.68
6.54
6.28
6.63
5.96
5.62
5.50
5.62
4.50
3.37
4.56
2.70
3.80
5.77
0.75
0.64
0.64
1.91
1.50
1.71
1.66
1.93
2.09
1.78
1.60
2.05
3.57
1.98
1.37
2.74
1.44
2.18
3.30
3.29
1.19
0.18
0.22
0.17
0.23
0.26
0.18
0.17
0.32
2.20
0.38
0.29
0.52
0.29
0.84
0.78
0.77
0.43
0.07
0.07
0.07
0.09
0.12
0.06
0.05
0.09
0.46
0.08
0.06
0.15
0.08
0.15
0.16
0.16
0.12
1.34
1.56
1.14
1.21
2.92
1.48
1.75
2.21
4.09
2.03
4.29
4.01
1.30
6.80
4.16
3.89
5.30
99.39
99.27
99.21
99.14
98.80
99.02
99.27
99.02
98.68
99.21
99.05
99.14
100.91
99.21
99.95
99.78
98.68
107
103
99
121
119
103
111
105
213
195
94
114
127
89
102
107
96
10.9
4.5
2.7
6.7
4.5
7.7
8.5
5.1
2.9
5.0
4.8
3.5
7.0
2.1
1.8
2.2
4.3
3.0
3.2
1.8
3.1
3.5
4.0
4.5
4.6
19.5
5.8
4.5
5.5
5.7
5.9
4.6
4.6
4.9
BH1
BH1
2.0
4.0
24
75
67
52
na
na
BH1
BH1
BH1
BH1
BH1
BH1
BH1
BH1
BH1
BH1
BH1
BH1
BH1
BH1
BH1
BH1
BH1
6.0
8.0
10.0
12.0
14.0
16.0
18.0
20.0
22.0
24.0
26.0
28.0
30.0
32.0
34.0
36.0
38.0
77
77
78
73
76
69
71
71
92
71
61
73
74
61
63
63
59
54
54
54
56
51
54
51
50
40
51
52
47
51
33
42
43
29
na
na
na
na
na
na
na
na
na
na
na
na
na
na
na
na
na
Borehole
BH12
BH12
Depth, m
2.5
3.0
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XX
XXX
XXX
XXX
XX
XXX
XX
XXX
XXX
X
X
XX
XX
XX
X
X
X
X
X
X
X
XX
XXX
XX
X
X
XX
XXX
XXX
XX
XX
XX
XX
XX
XX
X
XX
X
X
X
X
X
XX
X
Groundwater table
200–300 lS cm1
X
X
X
XXX
X
X
XX
XX
X
X
X
X
XX
SiO2, %
Al2O3, %
Fe2O3, %
MnO, %
MgO, %
CaO, %
Na2O, %
K2O, %
TiO2, %
P2O5, %
LoI, %
Total, %
Zr, ppm
69.64
68.98
12.60
16.12
6.76
3.52
0.22
0.04
0.47
0.67
0.81
1.48
0.96
3.62
0.67
1.06
0.46
0.36
0.07
0.04
7.73
4.90
100.38
100.79
269
156
Th, ppm
6.5
5.3
Nb, ppm
5.6
4.0
Borehole
4.0
6.0
7.0
72.02
64.51
67.24
14.64
14.85
15.54
2.61
5.59
4.12
0.02
0.06
0.04
1.11
2.42
1.62
1.84
3.24
3.65
4.59
4.54
4.97
1.00
0.39
0.45
0.29
0.49
0.46
0.03
0.06
0.09
2.47
4.22
2.75
100.64
100.37
100.92
139
104
128
3.9
10.7
2.0
2.9
3.7
3.9
BH12
BH12
BH12
BH12
BH12
BH12
BH12
BH12
BH12
BH12
BH12
BH12
BH12
BH12
BH12
BH12
BH12
BH12
BH12
BH12
BH12
BH12
BH12
BH12
BH12
8.5
9.0
10.0
10.5
11.0
13.5
14.0
15.0
17.5
18.5
20.5
22.0
23.5
25.0
26.5
28.0
29.5
31.0
32.5
34.0
35.5
37.0
37.5
40.0
41.0
71.43
66.82
71.04
70.23
70.37
67.93
66.91
71.61
70.81
63.24
62.13
69.97
69.50
70.14
66.52
67.39
70.28
68.63
57.50
64.72
64.59
63.74
67.55
69.72
67.49
14.20
13.28
14.52
14.39
12.78
10.20
10.29
12.87
15.34
12.46
12.24
14.47
13.44
13.14
14.00
13.75
14.52
13.88
13.92
13.00
14.21
13.99
14.17
13.25
14.30
2.76
5.38
3.05
3.22
5.13
8.40
8.07
3.81
2.50
6.57
7.81
3.23
3.83
3.79
5.33
5.02
3.28
4.38
9.59
6.94
5.05
6.43
4.29
4.24
4.56
0.02
0.05
0.03
0.03
0.04
0.05
0.05
0.03
0.02
0.09
0.08
0.03
0.04
0.04
0.05
0.05
0.03
0.04
0.11
0.08
0.06
0.07
0.05
0.04
0.05
1.33
2.65
1.39
1.91
2.03
4.35
5.09
1.14
1.31
3.82
6.77
2.93
3.23
2.73
4.11
3.34
1.89
2.95
4.34
3.38
3.07
3.30
3.05
2.59
2.91
2.23
2.41
2.42
2.01
2.55
1.40
1.26
2.95
1.67
5.41
2.18
0.69
1.10
1.70
1.55
2.28
1.87
2.04
6.43
4.12
3.12
3.93
2.17
1.80
2.39
4.88
4.04
4.86
5.06
3.34
1.54
2.03
4.33
5.78
3.43
2.73
6.16
5.02
4.47
4.16
4.06
5.26
4.88
2.80
3.74
5.30
4.20
4.57
3.94
4.61
0.89
0.42
0.98
0.69
1.62
2.39
1.93
0.96
1.15
1.13
1.99
0.49
0.83
0.92
1.69
1.63
1.29
1.06
1.73
1.18
1.38
1.47
1.39
2.13
1.56
0.31
0.42
0.34
0.26
0.48
0.70
0.46
0.45
0.26
0.60
0.60
0.25
0.39
0.36
0.45
0.39
0.31
0.37
0.85
0.61
0.46
0.58
0.41
0.43
0.46
0.08
0.07
0.09
0.06
0.11
0.08
0.09
0.15
0.08
0.12
0.11
0.06
0.11
0.11
0.12
0.11
0.08
0.10
0.13
0.11
0.13
0.12
0.12
0.13
0.12
2.17
5.29
2.13
2.54
2.32
3.95
4.41
1.45
1.36
3.12
2.82
1.57
1.71
1.59
1.94
1.68
1.47
1.70
2.31
1.98
2.26
2.16
2.18
1.73
1.95
100.30
100.82
100.86
100.39
100.75
100.97
100.60
99.73
100.26
99.98
99.45
99.86
99.18
98.97
99.91
99.70
100.29
100.02
99.71
99.85
99.63
99.98
99.95
100.00
100.39
159
125
221
121
243
500
598
334
175
100
87
61
118
116
107
83
76
136
123
153
178
173
288
244
205
4.5
4.7
6.7
4.0
5.2
11.2
6.0
9.4
8.5
4.1
2.5
1.9
4.5
7.1
6.1
4.7
3.4
7.5
3.8
6.2
11.0
10.6
25.4
14.0
14.7
3.5
4.5
3.6
2.6
7.3
15.6
15.6
6.3
3.8
7.3
8.7
3.2
5.1
3.7
5.7
5.9
4.1
5.0
5.3
4.9
5.5
5.3
5.5
7.1
6.2
BH12
BH12
BH12
BH12
BH12
2.5
3.0
4.0
6.0
7.0
18
48
58
60
63
78
62
59
55
55
60
60
60
90
150
X
X
X
X
X
XXX
XXX
XXX
XXX
XXX
X
XX
XX
XXX
XXX
X
X
X
X
X
X
X
X
X
BH12
BH12
8.5
9.0
62
54
57
56
220
300
X
X
XXX
XXX
XX
XXX
X
X
X
X
X
X
BH12
BH12
BH12
BH12
BH12
BH12
BH12
10.0
10.5
11.0
13.5
14.0
15.0
17.5
63
63
56
50
52
58
71
56
58
51
48
50
52
60
400
na
na
155
167
165
370
X
X
X
X
X
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XX
XXX
XX
X
X
XX
XXX
X
X
X
X
XX
X
X
X
X
X
X
X
X
XX
X
X
X
X
X
X
X
X
X
BH12
BH12
BH12
Groundwater table
200–300 lS cm1
XX
X
XX
945
946
XX
XXX
XX
XX
XX
XX
XX
XX
XX
XX
XX
XX
XX
XX
X
XX
5.1. Determination of fresh and weathered materials
X
X
X
X
X
X
X
X
X
X
X
X
X
X
XX
X
X
XX
X
X
XX
Mg-hornblende
Biotite
Oligoclase
XX
XX
XXX
XXX
XX
XX
XX
XXX
XX
XX
XX
XXX
XXX
XX
XX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
43
51
66
61
57
58
54
58
57
42
47
53
49
56
54
55
ER, Ohm m
CIA
WIP
18.5
20.5
22.0
23.5
25.0
26.5
28.0
29.5
31.0
32.5
34.0
35.5
37.0
37.5
40.0
41.0
BH12
BH12
BH12
BH12
BH12
BH12
BH12
BH12
BH12
BH12
BH12
BH12
BH12
BH12
BH12
BH12
65
66
71
65
61
68
66
69
67
68
64
77
70
68
66
69
Depth, m
Borehole
Table 1 (continued)
1221
1703
2425
3481
3128
3445
3952
3823
4538
3003
2445
1303
2221
2475
1621
na
Clays 2:1–1:1
Quartz
Anorthite
X
XX
X
X
X
X
X
X
X
XX
X
XX
XX
X
X
X
Sericite
Epidote
X
Tremolite
Chlorite
Calcite
5. DISCUSSION
Since the nature of the parent lithology is highly variable, we first distinguished the fresh parent material and
saprolite samples from the boreholes using both LoI and
[Fe2O3 + MgO] contents (Fig. 6). The LoI of fresh parent
rocks depends on the relative abundance of the primary hydrated minerals, given the low carbon content of the silicate
bedrock. When the rocks weather, the LoI accordingly increases with the formation of clays and clay minerals. Since
iron and magnesium are present in all primary hydrated
minerals of the watershed bedrocks, the comparison between [Fe2O3 + MgO] and LoI defines theoretical domains
for fresh rocks, based on the XRD mineralogy for the borehole samples. LoI, used as a weathering index, will then be
compared to the Chemical Index of Alteration, CIA (Nesbitt and Young, 1982) and the Weathering Index of Parker,
WIP (Parker, 1970), more classically used. Both indices reflect the mobility of base cations. CIA considers aluminum
as a conservative element and reflects the extent of plagioclase weathering, i.e. leaching of K, Na and Ca, and transformation into clay minerals such as kaolinite. As
weathering progresses, CIA increases from about 50 for
fresh rocks to 100 for optimum weathering. WIP differs
from CIA in that it relies on all major mobile alkali and
alkaline earth (K, Na, Ca, Mg) and can therefore be applied
to both acid and basic igneous rocks. WIP may be upper
than 100 for fresh rocks and tends towards 0 for weathered
materials. However, its application to highly weathered
material such as laterite is not recommended (Price and Velbel, 2003). Finally, the comparison will be carried out with
the electrical resistivity measured in boreholes, rock outcrops and soil catena.
5.1.1. Determination of fresh gneiss, gneiss-derived saprolite
and red soil
The determination of fresh gneiss samples is made with
the a priori condition that the samples should be located, in
the bivariate plot [Fe2O3 + MgO] versus LoI, in the domain
defined by three components: sericite-rich, biotite-rich and
chlorite-rich. The sericite-rich component represents the
leucocratic pool while both biotite-rich and chlorite-rich
components represent the melanocratic pool. In the leucocratic pool, in which the lowest [Fe2O3 + MgO] content is
2 wt%, the LoI content is prominently influenced by sericite
(LoI = 4.0 wt%; [Fe2O3 + MgO] = 7.2 wt%). For a
[Fe2O3 + MgO] content of 2 wt%, the maximum amount
of sericite is 27 wt%, so the corresponding maximum LoI
is 1 wt%. Towards the melanocratic pool, in which
[Fe2O3 + MgO] reaches 22 wt%, the LoI content is mainly
influenced by the abundances and the relative proportions
of chlorite and biotite. Knowing that LoI of chlorite is
11 wt% and LoI of biotite is 4 wt%, the maximum LoI values for the fresh melanocratic component range from
2.6 wt% for a biotite-rich sample to 5.9 wt% for a chlorite-rich sample. The samples above the sericite-chlorite
mixing line obviously correspond to weathered samples,
i.e. gneiss-derived saprolite and red soil. The samples within
Table 2
Bulk chemical analyses for major and selected inert trace elements (Zr, Th, Nb) and electrical resistivity for the soil samples of the S1 and S2 sites. CIA and WIP are indicated.
Soils
SiO2, % Al2O3, % Fe2O3, % MnO, % MgO, % CaO, % Na2O, % K2O, % TiO2, % P2O5, % LoI, % Total, % Zr, ppm Th, ppm Nb, ppm WIP CIA ER, Ohm m
downslope T1 catena
40–60
64.52
40–80
63.12
60–80
62.18
80–100
62.34
80–120
63.06
80–120
63.56
100–120
63.72
100–120
63.53
120–140
61.89
120–140
62.53
140–160
64.81
160–180
64.42
180–200
65.66
200–220
73.19
S1-P – profile sampled close
S1-P
0–15
78.79
S1-P
25–35
77.10
S1-P
45–55
71.01
S1-P
65–75
67.09
S1-P
95–105
70.07
S1-P
120–130
72.67
S1-P
145–155
73.00
S1-P
170–180
71.22
S1-P
205–215
67.28
S1-P
225–235
67.36
S1-P
230–240
71.09
S2-P – upslope profile
S2-P
0–5
71.31
S2-P
8–18
73.29
S2-P
25–35
76.60
S2-P
90–100
64.88
S2-P
190–200
68.85
S2-P
240–250
76.18
S2-P
285–295
76.96
S2-P
310–320
73.41
(Barbiéro
15.51
15.59
16.57
16.46
15.78
15.50
15.78
16.28
16.67
16.50
15.17
14.78
15.22
14.57
et al., 2007)
6.18
0.10
6.31
0.12
6.65
0.10
6.94
0.08
6.46
0.13
6.49
0.10
6.58
0.10
6.54
0.09
6.92
0.08
6.38
0.06
6.11
0.11
5.82
0.09
6.06
0.11
1.81
<L.D.
0.52
0.51
0.75
0.63
0.58
0.51
0.56
0.58
0.65
0.75
0.50
0.97
0.53
0.70
0.62
0.65
0.68
0.59
0.68
0.70
0.74
0.63
0.56
0.70
0.68
0.83
0.70
0.77
0.45
0.45
0.58
0.50
0.63
0.41
0.46
0.45
0.50
0.60
0.46
0.82
0.57
4.16
0.90
0.87
0.97
0.95
0.94
0.92
0.95
0.94
0.97
0.94
0.88
0.92
0.84
1.35
0.61
0.62
0.67
0.67
0.60
0.63
0.65
0.67
0.67
0.66
0.60
0.64
0.62
0.15
0.05
0.05
0.04
0.04
0.05
0.05
0.05
0.04
0.04
0.04
0.04
0.04
0.04
<DL
11.13
11.38
11.56
11.28
10.52
10.61
10.89
11.31
11.30
11.16
9.69
10.62
9.23
2.76
100.59
99.67
100.74
100.48
99.43
99.48
100.47
101.07
100.23
100.31
99.05
99.93
99.57
99.45
283
229
223
232
274
264
236
273
222
265
271
224
295
88
15.1
12.9
8.6
8.4
8.9
10.0
8.5
9.0
9.2
9.5
9.3
9.5
10.5
11.2
3.5
8.9
8.5
8.9
8.4
8.6
8.6
9.1
9.5
9.7
9.7
9.7
9.5
9.3
15
14
17
16
17
15
16
15
16
17
15
20
15
54
85
85
84
85
83
85
84
85
86
84
84
80
83
60
60
60
60
60
60
30
30
30
30
30
30
30
150
150
to the downslope T1 catena
8.50
3.03
0.04
11.26
3.81
0.04
13.31
4.62
0.07
15.21
5.15
0.03
14.57
3.39
0.03
12.69
4.15
0.03
12.30
3.87
0.03
13.06
3.76
0.04
14.81
4.55
0.05
16.69
3.31
0.06
16.51
1.61
0.02
0.61
0.65
0.66
0.66
0.75
0.95
1.34
1.20
1.53
1.13
0.51
0.74
0.68
0.56
0.49
0.55
0.76
1.08
1.32
1.50
1.76
2.08
0.69
0.62
0.48
0.42
0.51
0.90
1.26
1.63
2.02
3.79
4.55
0.79
0.82
0.81
0.79
0.81
0.72
0.81
0.79
0.76
0.58
1.05
0.36
0.42
0.45
0.50
0.51
0.40
0.36
0.37
0.40
0.29
0.21
0.05
0.04
0.05
0.04
0.02
0.03
0.02
0.02
0.02
0.02
0.02
6.76
4.35
8.10
9.01
8.35
6.72
6.15
6.16
6.72
4.67
2.27
100.36
99.80
100.12
99.40
99.55
100.02
100.23
99.56
99.64
99.66
99.92
260
306
230
221
285
234
195
236
144
109
124
na
na
na
na
na
na
na
na
na
na
na
na
na
na
na
na
na
na
na
na
na
na
16
16
14
14
15
19
25
28
33
47
58
73
79
84
87
85
78
72
69
68
62
57
60
60
60
60
30
30
30
30
150
150
150
0.60
0.64
0.53
1.44
1.17
0.79
0.74
0.35
1.11
0.86
0.69
1.05
1.10
0.81
0.79
0.63
1.15
1.22
1.22
1.11
1.75
1.72
1.73
3.96
0.81
0.77
0.75
0.90
0.85
0.89
0.90
3.16
0.55
0.58
0.51
0.65
0.53
0.35
0.37
0.14
0.08 10.65
0.05 7.56
0.04 5.69
0.02 9.59
0.02 7.28
0.02 4.92
0.02 4.82
0.02 2.44
99.37
99.48
100.27
99.42
99.94
100.09
100.33
100.09
292
304
313
193
170
177
216
79
na
na
na
na
na
na
na
na
na
na
na
na
na
na
na
na
22
22
21
24
29
28
28
66
66
71
71
75
70
66
67
56
60
60
60
60
150
150
150
150
8.99
10.30
9.77
13.96
13.19
10.22
10.74
14.15
4.03
4.13
4.41
5.77
5.14
4.16
3.23
1.83
0.09
0.07
0.06
0.06
0.06
0.04
0.04
0.01
S1-T –
S1-T1
S1-T1
S1-T1
S1-T1
S1-T1
S1-T1
S1-T1
S1-T1
S1-T1
S1-T1
S1-T1
S1-T1
S1-T1
S1-T1
Depth, m
947
99
49
98
71
74
98
97.65
99.70
98.43
11.97
<DL
12.64
<DL
0.04
<DL
17.18
29.86
2.37
29.80
52.46
41.77
1.50
<DL
0.17
<DL
<DL
5.85
11.04
<DL
<DL
0.14
<DL
0.10
25.67
<DL
35.37
0.07
12.85
0.03
<DL
4.44
<DL
0.03
<DL
0.04
9
30
90
77
98.93
99.33
2.07
2.03
0.06
0.21
2.23
10.27
Mafic to ultramafic rocks
Tremolite
n = 18
Mgn = 10
hornblende
Chlorite
n=5
Labradorite
n = 18
Serpentine
n = 47
54.93
46.28
0.22
<DL
2.13
4.00
3.87
10.20
0.19
0.21
19.93
12.10
12.78
11.65
0.25
1.28
0.09
0.88
0
100
105
96
43
57
57
99.94
97.57
99.78
96.08
100.09
101.69
#
<DL
4.00
4.42
11.69
1.89
1.18
#
0.09
9.26
10.19
0.02
<DL
<DL
Gneiss
Quartz
Oligoclase
Biotite
Sericite
Chlorite
Epidote
Titanite
#
n = 30
n = 15
n=7
n=6
n = 16
n=2
100.00
66.61
36.72
49.25
27.25
38.45
31.53
#
20.75
15.22
29.18
20.87
23.19
2.23
#
<DL
<DL
<DL
<DL
<DL
<DL
#
<DL
<DL
<DL
<DL
13.25
2.03
#
<DL
20.71
3.30
20.08
<DL
<DL
#
<DL
0.25
<DL
0.23
0.17
0.17
#
<DL
9.53
2.12
18.67
<DL
1.02
#
1.63
0.12
<DL
0.07
23.02
26.47
#
10.75
<DL
0.63
<DL
<DL
<DL
#
<DL
1.75
<DL
0.05
<DL
37.05
WIP
Total,
wt%
LoI,
wt%
TiO2,
wt%
K2O,
wt%
Na2O,
wt%
CaO,
wt%
MgO,
wt%
MnO,
wt%
FeO,
wt%
Fe2O3,
wt%
Cr2O3,
wt%
Al2O3,
wt%
SiO2,
wt%
Table 3
Chemical composition, CIA and WIP of the main minerals from gneiss and mafic to ultramafic rocks. <DL: below detection limit; #: not determined. Quartz is assumed 100% SiO2.
0
50
60
73
100
34
34
CIA
948
the domain containing chlorite are fresh while those containing only biotite and above the sericite–biotite mixing
line can be considered as weathered. Their selection is based
on the XRD patterns. Both groups of fresh gneiss and saprolite samples spread on a wide range and show positive
correlations (Fig. 6). Most of fresh gneiss and saprolite
samples are however in the range of 2–10 wt% [Fe2O3 +
MgO]. The average for fresh gneiss and saprolite therefore
represents the parent rock at the watershed scale.
The estimation of the modal abundances for the average
gneiss takes into account the mineral occurrence on the
XRD patterns. Assuming that apatite controls 100% of
P2O5, its modal abundance is first determined; then the proportion of Ca linked to apatite is deducted from the bulk
analysis. To estimate the modal abundances with the apatite-corrected bulk analysis, we apply a linear inverse method using least squares criterion (Tarantola and Valette,
1982). The solution and error is given by equations 47
and 48 in Tarantola and Valette (1982):
1
^x ¼ AT C 1
AT C 1
ð1Þ
y0 y0 A
y0 y0 y 0
1
C^x^x ¼ AT C 1
ð2Þ
y0 y0 A
where y0 is the chemical composition vector of the rock, A
is the matrix of the main mineral compositions and ^x the a
posteriori solution (modal abundance vector), C 1
y 0 y 0 is the
inverse of the covariance matrix and C^x^x is the a posteriori
error covariance of the solution. The residuals are calculated by y 0 ^y , where ^y ¼ A ^x. The minerals selected in
the matrix A are quartz, oligoclase, biotite, sericite, chlorite,
epidote, titanite for the average gneiss. Both bulk compositions (±r), calculated modal abundances and their errors,
estimated bulk compositions (^y ) and associated residuals,
and the estimated contributions from each mineral to the
whole rock are summed up in Table 4 for the average
gneiss.
It appears that CIA and WIP are not able to separate
the saprolite samples from the fresh gneiss samples, while
the red soil samples are distinguished (Fig. 6). The comparison of all three weathering indices to the electrical resistivity shows however a threshold between the fresh gneiss
samples and the weathering materials at 400 Ohm m
(Fig. 7). Compared to the measurements of electrical resistivity on fresh gneiss outcrops, which are in the range 1000–
2000 Ohm m, this threshold seems to be relatively low. An
explanation could be the integration of the fissured
unweathered bedrock layers common in hard-rock aquifers
(Dewandel et al., 2006). More precise measurements, i.e.
drill core sampling instead of cuttings, should be done to
characterize with accuracy the boundary between fresh,
unfractured rock, fissured rock and saprolite. The
400 Ohm m threshold will therefore be used in the modeling
of the ERT profiles in the next section.
5.1.2. Determination of the fresh amphibolite and
amphibolite-derived saprolite
In the fresh amphibolite, in which [Fe2O3 + MgO] ranges
between 17 and 28 wt%, the LoI results from the mixing between chlorite, serpentine and Mg-hornblende. We may
949
Fig. 3. Petrographical features of the gneiss. (a) Handpicked sample of gneiss showing melanocratic and leucocratic parts at the decimetric
and centrimetric scale, (b) SEM-BE microphotograph of gneiss section at lower magnification including (c) and (d), (c) detail showing a
chlorite crystal, (d) detail showing sericite sticks within an oligoclase crystal, (e) biotite crystal with exsolution of titanite crystals, epidote
crystals are also present, (f) apatite and zircon crystals. Note the small size of the latter (10 lm).
consider that the chlorite/serpentine line delineates fresh
and weathered samples as both minerals have similar LoI
(–12 wt%) while the Mg-hornblende line (LoI – 2 wt%)
delineates the lower boundary (Fig. 6). However some fresh
samples can contain a large amount of carbonates in which
the LoI goes up.
Six points related to the fresh samples are clearly observed in the weathered domain. These points correspond
to shallow samples up to 10 m in depth. Observations carried out in a 3 m deep pit dug close to BH6 shows clearly
fresh highly fissured rock in which the joints between the
angular boulders are filled with clayey materials (similar
to the outcrop shown in Fig. 4). We suppose that this conductive material however does not show a difference in
terms of chemical signature with the unweathered parent
bedrock and is responsible for the low resistivity. The resistivity of the fresh amphibolite ranges between 10,000 and
1000 Ohm m but a clear resistivity limit between weathered
and fresh amphibolite cannot be extracted from this data
set and cannot be taken into account in the further mass
950
Fig. 4. Petrographical features of the BH6 amphibolite. (a) Outcrop of fissured dyke of amphibolite near BH6, (b) detailed SEMbackscattered electron (BE) section of the BH6 amphibolite showing the presence of crystals of Mg-hornblende, chlorite, serpentine, Fe–Cr
oxide and Mg-rich calcite.
Fig. 5. Result of ERT survey. Calculated resistivity resulting from ERT inversion of field data is presented versus depth for the 12 ERT crosssections. Bore hole locations (BH1, 5, 6 and 12) are noted.
a
951
Loss on Ignition (wt%)
15
10
Fresh gneiss
Red soil
Saprolite
Fresh amphibolite
Weathered amphibolite
chlorite
serpentine
1
chlorite
2
5
2
1
3
3
biotite
sericite
LoI = 1.7(Fe2O3+MgO) - 1.3
r 2 = 0.78
LoI = 0.3(Fe2O3+MgO) + 1.5
r 2 = 0.62
LoI = 0.1(Fe2O3+MgO) + 0.8
r 2 = 0.87
Mg-hornblende
0
0
10
20
30
Fe2O3+MgO (wt%)
b
15
c
10
5
0
0
20
40
60
80
100
Chemical Index of Alteration (CIA)
0
20
40
60
80
100
Weathering Index of Parker (WIP)
Fig. 6. Property–property diagrams for Loss on Ignition versus Fe2O3 + MgO, WIP and CIA.
balance calculation. Besides fresh gneiss and amphibolite
have a similar resistivity range. Consequently it is not feasible to differentiate and to quantify the volumes of both
lithologies with the resistivity alone at the watershed scale.
5.2. Assessment of regolith thickness with ERT
The electrical resistivity measured in four boreholes located on ERT profile 2 (BH13, 7, 8 and 9) was earlier compared to the calculated resistivity of this ERT profile
(Descloitres et al., 2007). This comparison shows the good
agreement between both measured and calculated resistivities for a restricted data set. But, since the threshold of
400 Ohm m delineates fresh and weathered rocks, the uncertainty linked to the calculation of this resistivity value in the
ERT profile has to be assessed. For this purpose, we carried
out a modeling, based on typical geometries and resistivity
ranges encountered in the watershed. Four geometries have
been tested (Fig. 8): (i) one step in the regolith, (ii) three
steps in the regolith, (iii) two thin resistive dykes and (iv)
two deep conductive dykes. To fix the model resistivity values we chose the weathered materials with resistivities just
below 400 Ohm m, and above 400 Ohm m for the fresh
rock. Four materials and corresponding resistivities were
defined: (i) thin topsoil of 100 Ohm m, (ii) clayey-sandy
materials of 60 Ohm m, (iii) sandy-clayey materials of
350 Ohm m and (iv) fresh bedrock of 5000 Ohm m. The first
step of the modeling is to generate a synthetic apparent resistivity data set, similar to field data with RES2DMOD software (Loke, 2000). Then this synthetic data set is inverted
with RES2DINV software (Loke, 2000). We calculated simplified ERT profiles with the threshold of 400 Ohm m
(Fig. 8). Inversions roughly reproduce the geometry of initial models but they are more reliable to reproduce intrusions of protolith in regolith than to detect intrusions of
regolith in protolith. This may be due to the loss of accuracy
of ERT with depth. For each model the uncertainty of ERT
is noted as a deviation of regolith thickness from the model
952
Density,
g/cm3
SiO2,
wt%
Al2O3,
wt%
Fe2O3,
wt%
MnO,
wt%
MgO,
wt%
CaO,
wt%
Na2O,
wt%
K2O,
wt%
TiO2,
wt%
P2O5,
wt%
LoI,
wt%
Total,
wt%
Zr,
ppm
Th,
ppm
Nb,
ppm
WIP
CIA
ER,
Ohm m
Average
2.74
68.21
13.68
4.78
0.05
2.60
2.02
4.49
1.67
0.40
0.10
1.93
99.92
172
8.8
6.2
68
53
400–
5000
±r
Estimate
Residuals
0.05
5.27
68.21
0.00
1.35
13.94
0.25
3.32
5.28
0.50
0.05
0.05
0.01
1.86
2.57
0.04
1.50
1.63
0.39
1.40
3.98
0.51
0.56
1.59
0.08
0.17
0.40
0.00
0.03
0.79
1.65
0.27
83
6.3
2.9
5
5
Mode
(%)
±r
Gneissic protolith
average gneiss
n = 29
Mineral contribution to the whole rock composition
SiO2
Al2O3
Fe2O3
MnO
MgO
CaO
Na2O
K2O
TiO2
P2O5
LoI
Major minerals
Quartz
Oligoclase
Biotite
Chlorite
Sericite
32 ± 7
38 ± 11
10 ± 12
8±6
6 ± 11
47
37
5
3
5
0
57
11
13
14
0
0
46
38
5
0
0
48
38
0
0
0
36
59
5
0
35
1
0
0
0
99
0
0
1
0
2
57
0
41
0
0
43
1
0
0
0
0
0
0
0
0
23
56
16
Accessories
Titanite
Apatite
Epidote
0.60 ± 0.73
0.23 ± 0.08
3.74 ± 6.03
0
0
2
0
0
6
0
0
10
2
0
12
0
0
0
9
7
48
0
0
0
0
0
0
56
0
0
0
100
0
0
0
4
Table 4
Average composition, modal abundance and contribution to the whole rock for each mineral of the parent rock.
Electrical Resistivity (Ohm.m)
a
953
100000
Fresh gneiss
Red soil
10000
Saprolite
Fresh amphibolite
Weathered amphibolite
1000
400 Ohm.m
100
soil
10
1
0
5
10
15
b
100000
10000
1000
400 Ohm.m
100
soil
10
1
20
30
40
50
60
70
80
90
Chemical Index of Alteration (CIA)
c
100000
10000
1000
400 Ohm.m
100
soil
10
1
10
20
30
40
50
60
70
80
Weathering Index of Parker (WIP)
Fig. 7. Property–property diagrams for electrical resistivity versus Loss on Ignition, WIP and CIA.
value. In the three-step model, ERT inversion underestimates the regolith thickness by 9.7%. The highest deviation
is noted for the dyke model, 21%. Based on these models, the
average underestimation of the regolith thickness by the
ERT inversion procedure would be 15.8%. If this value is taken into account for the calculation of the regolith thickness
954
ERT forward modelling
and inversion
average
regolith
thickness
for models
calculated
models
synthetic
models
Distance (m)
21.70 m
depth (m)
100
-10
-10
a
-30
100
200
depth (m)
depth (m)
0
-10
-10
depth (m)
150
150
-20
150
-20
100
200
-10
d
150
-20
model resistivity (Ohm.m)
regolith
fresh rock
materials
350
9.70%
12.40 m
- 5.55 m
15.80%
16.75 m
- 7.35 m
21.00%
200
h
-30
60
- 3.40 m
g
100
0
100
12.55 m
200
-30
-10
-30
150
-20
0
-20
14.70%
f
-30
200
c
-30
- 5.15 m
200
0
b
-30
16.55 m
e
-30
0
-20
ratio to
total model
thickness
200
-10
100
24.10 m
150
150
-20
0
-10
100
17.95 m
100
0
-20
difference
with model
Distance (m)
200
0
100
15.95 m
150
average
regolith
thickness
for ERT
5000
total model
thickness : 35 m
calculated resistivity
with ERT (Ohm.m)
regolith
fresh rock
< 400
> 400
Fig. 8. ERT modeling: synthetic models are (a) one step in the regolith, (b) three steps in the regolith, (c) two thin resistive dykes, (d) two deep
conductive dykes. The models are computed with ERT forward modeling procedure. The resulting apparent resistivity cross-sections (not
shown) are inverted with the same ERT procedure as field data to produce calculated resistivity profile. ERT final results (e), (f), (g) and (h)
are presented using the resistivity threshold of 400 Ohm m (deduced from geochemical and mineralogical analysis) that separates regolith
domain, in blue, from fresh rock, in red. The differences between the model regolith thickness and calculated ERT regolith thickness and their
respective ratio related to the total model thickness (35 m) are indicated on the right for each model. (For interpretation of the references to
color in this figure legend, the reader is referred to the web version of this paper.)
V w qw C j;w V p qp C j;p
¼
þ mj;flux
100
100
in the 12 profiles (Fig. 9), then the regolith thickness would
range from 13.5 m at the outlet of the watershed (profile 1)
to 23.7 m at the top of the watershed (profile 9). The distribution of the regolith thickness along the 12 ERT sections is
not related to the altitude (Fig. 10). The variation in thickness in each profile might be related to the heterogeneity
in the structure and the fractures of the gneissic substratum
and its availability to weather. On average, the regolith
thickness in the Mule Hole watershed is 17.2 m, which corresponds to a volume of 74 106 m3 of weathered materials.
As the average thickness of the soil cover estimated from
both EM31 investigations and pedological survey is about
2 m, the saprolite thickness can be deduced to be 15.2 m.
where the subscripts p and w refer to the parent and weathered materials, respectively. V is volume in cm3, q is bulk
density in g/cm3 and Cj is chemical concentration of any
element j in weight percent (wt%). The mj,flux represents
the mass of an element j moving into or out of the system.
The mj,flux is positive if the element j is accumulating in the
system and negative if j is leaching from the system.
The volumetric strain (e) or volume change is calculated
from the density ratios q and conservative element concentrations Ci in the regolith by
5.3. Mass balance calculation
ei;w ¼
The mass balance equation set is based on the principle
of mass conservation (Brimhall et al., 1991; Oh and Richter, 2005). For a chemical element j
Positive values of ei,w indicate expansion, negative ones
indicate collapse and values around zero, isovolumic
weathering.
qp C i;p
1
qw C i;w
ð3Þ
ð4Þ
955
Fig. 9. Interpretation of the 12 ERT profiles and corresponding average thickness of the regolith.
0
830
840
850
860
870
880
890
900
910
UPSLOPE
5
10
15
average thickness
20
25
30
35
40
DOWNSLOPE
average altitude
Regolith thickness inferred from ERT
(meters)
Altitude (in meters, above sea level)
820
Fig. 10. Relationship between regolith thickness and altitude along the 12 ERT profiles. The average regolith thickness (17.2 m) and altitude
(860 m) lines are also marked.
The addition or subtraction of a chemical element j,
either by solute migration or mechanical translocation, is
quantified by the open-system mass fraction transport function (sj,w)
!
qw C j;w
ðei;w þ 1Þ 1
ð5Þ
sj;w ¼
qp C j;p
Because the calculation of sj,w takes into account both
residual enrichment and deformation, a positive value for
sj,w reflects a true mass gain in element j of the weathered
rock compared to the parent rock and a negative value indicates a mass loss. If sj,w = 0, the element is immobile during
weathering with respect to the volume of regolith consid-
ered. Moreover, quantification of the overall mass transfers
during both saprolitization and soil processes can be approached by the estimation of the chemical component
transfer. The total mass of any mobile element j (DMj)
transferred through the weathering system with thickness
z (cm), expressed in mol/ha, is given by
Z z
DM j ðmol=haÞ ¼ 106 qp C j;p sj;w dz
ð6Þ
0
Mass balance requires precise verifications regarding the
determination of the parent material composition prior to
the chemical weathering onset and the choice of an inert
element, which should be very insoluble and resistant to
weathering. For the mass balance calculation, we propose
956
to consider the average thickness of saprolite and soil determined above and the average gneiss composition as parent
material.
5.3.1. Selection of the inert element
Even if it would be preferable to assess mass balance in
weathering profiles based on the usual inert trace elements
as Zr, Th and Nb (Braun et al., 1993; White and Brantley,
1995; Kurtz et al., 2000), the heterogeneous distribution of
these elements in the Mule Hole parental gneiss prevents
them from being used as references (see Table 1). In the
gneiss, TiO2 is chiefly controlled by titanite and biotite (Table 4). Both minerals are among the first to breakdown in
the incipient weathering stage and this leads to the in-situ
precipitation of insoluble Ti-oxides. TiO2 may constitute
the most suitable reference for the mass balance calculation
even if we cannot dismiss a slight mobility in the weathering
profile as shown by Tripathi and Rajamani (2007) in similar
weathering profiles from the Mysore Plateau and by Cornu
et al. (1999) and Taboada et al. (2006). Chemical weathering fluxes would be therefore slightly underestimated (Table
5).
5.3.2. Strain and elemental gain or loss in the average gneissderived saprolite
The mass balance calculation for the saprolite first requires estimating its average bulk density, which obviously
spatially varies according to the degree of weathering of the
gneissic domains. For instance, the bulk density is
1.9 ± 0.1 g/cm3 for the saprolite samples derived from the
gneiss at the bottom of the T1 soil catena. One way to estimate the average saprolite bulk density is to assume isovolumetric weathering. If so, eTiO2 ;w equals to 0 and then
qsaprolite ¼ qp ðC TiO2 ;parent =C TiO2 ;saprolite Þ. The calculated average bulk density of the saprolite is 2.4 g/cm3.
Within the Mule Hole saprolite the open-system masstransport functions indicate that all major elements except
Ca are depleted with the following sequence: Mg
(s = 0.42) > K (0.26) > Mn (0.22) > Fe (0.20) > Na
(0.18) > P (0.17) > Si (0.13) > Al (0.11) (Fig. 11).
Similar calculations carried out on the Rio Icacos quartz
diorite provided different sequences with P > Ca = Na >
Fe(II) > K > Mn > Si = Mg > Fe (total) > Al for a spheroid corestone/rindlet system (Buss et al., 2008) and Ca =
Na > Mg > Si > K > Al > Fe (total) for the underlying saprolite, respectively (White et al., 1998). Buss et al. (2008)
concluded that both sequences indicate (i) the rapid dissolution of plagioclase and apatite and slower weathering of
Fe–Mg-silicates in the incipient weathering rindlets and
(ii) the further weathering of biotite in the saprolite with
loss of Mg. They argue that biotite oxidation is the most
likely fracture inducing reaction in the rindlets allowing
the solutions to dissolve the other mineral phases. At Mule
Hole the s sequence primarily supports that chlorite and
biotite, the chief sources for Mg (95%), Fe (84%), Mn
(86%) and K (57%, biotite only) are the first to weather during saprolitization. The biotite loss may be estimated with
s(K) if we consider (i) the stability of sericite and (ii) the total K leaching. It means that, at least, 49% of the biotite
crystals are weathered in the average saprolite and trans-
formed into smectite and kaolinite/smectite interstratified
(Bourgeon and Larqué, 1992). The second information
borne in the s sequence is that the oligoclase crystals are
quite preserved into the saprolite. Oligoclase is the chief
source for Na (99%) and Al (57%) and the second source
for Si (38%) after quartz (48%) and for Ca (35%) after epidote (48%). As sericite, quartz is stable in the saprolite.
Therefore, the chief sources of Al, Na and Si during weathering are the breakdown of oligoclase. The loss of oligoclase in the saprolite can be assessed if we assume that
Na is congruently leached from the regolith; if so, s(Na)
corresponds to the amount of oligoclase loss in the saprolite, i.e. 18%. The s(P) is also moderate, meaning that apatite, as the only P-bearing mineral, is partly conserved in the
saprolite. A significant leaching of Fe and Mn also occurs
in the saprolite.
Ca is slightly accumulated in the saprolite. In the average gneiss, the chief Ca sources are oligoclase (35%), epidote (48%), titanite (9%) and apatite (7%). All these
phases are weathered to some degree and should lead to
the leaching of Ca. A differential weathering pathway of
the primary Ca-bearing minerals cannot explain the Ca
accumulation. Another explanation would be the precipitation of CaCO3 from the percolating solution due to current
and/or paleoclimatic conditions; carbonate nodules formed
within the saprolite are common in the watershed.
Overall when integrated over the average saprolite depth
of 15.2 m, the losses by total mass occur for Si, Mg and Na
with 286 106 mol/ha (62% of the total mass loss),
67 106 mol/ha (15% of the total mass loss) and
39 106 mol/ha (9% of the total mass loss), respectively.
Al, Fe and K account for 7%, 4% and 3% of the total mass
loss, respectively. P and Mn account for only 0.04% and
0.10%, respectively.
5.3.3. Strain and elemental gain or loss in the average gneissderived red soil
The calculated strain in the average red soil indicates a
collapse of 38% of the volume due to bio-pedoturbation
processes. The open-system mass-transport functions point
out that all major elements except Mn are depleted within
the red soil profiles: Na = Mg (s = 0.76) > P
(0.70) > Ca (0.65) > K (0.55) > Si = Fe (0.19) > Al
(0.17) (Fig. 11). The s sequence indicates that Na-plagioclase weathering is enhanced compared to saprolite; from
s(Na), at least 80% of oligoclase crystals have broken down.
The low s(K) emphasized that biotite is completely transformed as shown by its absence on XRD soil patterns.
The remaining K may be attributed to the persistence of
sericite crystals. Fe is slightly leached from the soil and
Mn is accumulated. Both elements are precipitated as oxides and oxyhydroxides during soil formation. Their mobility is linked to climate changes (Tripathi and Rajamani,
2007). When integrated over the average red soil depth of
2 m, the most important losses occur for Si, Na and Mg
with 55 106 mol/ha (47% of the total mass loss),
22 106 mol/ha (19% of the total mass loss) and
16 106 mol/ha (14% of the total mass loss), respectively.
Ca, Al and K account for 7.9%, 5.8% and 3.8% of the total
mass loss, respectively. Fe and P account for only 1.9% and
30–100
Loss () or gain (+)
% Of the sum (loss only)
Loss () or gain (+)
% Of the sum (loss only)
Total
P
Ti
K
Na
Ca
Mg
Mn
Fe
Al
Si
sj,red soil
Av. thickness = 15.2 m
Av. red soil
Av. thickness = 2 m
sj,saprolite
±r
n = 25 Average
±r
mol/ha(106)
%
mol/ha(106)
%
1.60
0.10
286
32
18
0.48 67
7
39
16
NR
62.2
7.0
3.9
0.10
14.7
8.6
3.4
55
7
2
0.04 16
9
22
4
NR
47.1
5.8
1.9
14.0
7.90 19.3
3.8
0.40 0.09
0.49 0.04
0.16 0.01
0.00 0.17
0.00 0.70
1.30
0.73
1.34
0.93
1.24
0.42
0.18 0.26
0.76 0.55
1.19 2.60
0.77 0.88
0.30 0.39
0.42 0.06
0.76 0.65
6.17
1.69
2.68 0.06
68.65 14.04
4.78 0.07
5.15
2.45
1.57 0.03
0.13 0.11 0.20 0.22
0.19 0.17 0.19 0.15
0.56
1.42
1.40
4.23
1.50
2.46
1.86
1.74
0.05
0.04
3.32
4.35
1.35
13.96
5.27
67.99
0.05
2.40
±r
n = 18 Average
Av. saprolite
0.21 459
0.05 100
0.12 116
0.10 100
11
25
14
3.2 4.8
10.0 8.7
1.9 1.6
1.39
155
7.95 99.92 223
2.93
65
9
76
10
100–
400
5
62
0.17
0.45
0.03
0.09
0.79
83
3.27 100.01 217
6.3 2.9
6.6 6.8
5
54
400–
5000
53
68
8.8 6.2
1.93 99.92 172
0.10
0.40
1.67
4.49
2.02
2.60
0.05
4.78
13.68
2.74
n = 29 Average
Av. gneiss
68.21
Total, Zr, Th, Nb, WIP CIA ER,
wt% ppm ppm ppm
Ohm m
TiO2, P2O5, LoI,
wt% wt% wt%
0.1%, respectively. Overall the soil profiles are more evolved
than saprolite but still contain primary minerals able to
weather. If the mass balance is computed within the soil
zone only, 80% of the losses of Si, Al, Fe would be neglected. It thus becomes crucial to assess the weathering
across the full depth of the regolith profile.
Chemical weathering rates for landscapes are difficult to
quantify because the timescales over which weathering occurs are often unknown. For an eroding landscape where
the weathering system is adjusted to hydrobioclimatic conditions it is reasonable to assume that the rate of conversion
of rock into saprolite equals the average long-term physical
erosion rate, i.e. that the system has reached a steady state.
This assumption supposes that the mass of weathered material in storage on the landscape is approximately constant
through time (Green et al., 2006). Based on an approach
combining landform, vegetation, water balance index, clay
mineral and soil studies, the steady state assumption was
argued for the landscapes of the rain shadow of the Western
Ghâts in spite of inevitable fluctuations of erosion rates
around median statistical values (Gunnell and Bourgeon,
1997; Gunnell, 2000; Gunnell et al., 2007). Subsequently
the erosion rates of the gneissic substratum of the Karnataka Plateau were assessed based on cosmogenic 10Be measurements and steady state assumption (Gunnell et al.,
2007). The average erosion rate is 13.6 ± 2.9 mm/kyr
(Table 2 from Gunnell et al. (2007)) and consequently the
average long-term chemical weathering, i.e. deepening of
the weathering front, is of the same order. That supposes
an average time span of 1.1 Ma to form 15 m of saprolite
at the watershed scale.
Density, SiO2,
wt%
g/cm3
Al2O3, Fe2O3, MnO, MgO, CaO, Na2O, K2O,
wt%
wt%
wt% wt%
wt% wt%
wt%
957
5.4. Long-term chemical weathering rate and minimum age of
the saprolite
Gneissic regolith Av.
thickness = 17.2 m
Table 5
Average parent rock, saprolite and soil compositions used in the mass balance calculations. Open-system mass-transport function s and estimated elemental mass flux in mol/ha over the mean
sampling depth during saprolite and soil weathering. The percentage of the sum of loss is also indicated. NR: not relevant.
5.5. Consequence of chemical weathering on the alkalinity
production potential on the Karnataka Plateau
Even if the Mule Hole watershed is representative of
only a very narrow bioclimatic transition zone wedged between the comparatively far more extensive humid and
semi-arid zones of the rain shadow gradient (Fig. 1) it is
worth discussing the potential of alkalinity production of
weathering covers according to the regional climatic variability associated with alternating periods of depletion
and intensification of the monsoon. Because of low reserves
in unweathered base cation-rich primary minerals, we can
argue that, whatever the intensity of the monsoon, the deeply depleted lateritic cover of the West end of the gradient
will have a limited potential for producing alkalinity. However, in the event of increased mean rainfall over the region,
one would assume that both the transition zone and the
very extensive semi-arid zone containing a significant stock
of unweathered primary minerals would significantly contribute to produce alkalinity and therefore to consume
atmospheric CO2. It could be added that the vegetation,
at least in the transition zone, would also probably change
to evergreen forest instead of moist deciduous, with ecological parameters such as increased biomass and carbon
958
reconstructed
weathering
profile 1
ρbulk
1.5
0
average soil
2.5
3
2
Depth (meters)
average saprolite
2
4
6
8
10
12
14
average parent gneiss
gain
loss
-1
-0.8
-0.6
-0.4
16
-0.2
0
0.2 -1
gain
loss
-0.8
-0.6
-0.4
-0.2
0
0.2 -1
gain
loss
-0.8
-0.6
-0.4
-0.2
0
0.2
0
Depth (meters)
2
4
τ
Si
τ
Mg
τ
Na
τ
τ
τ
Mn
τ
τ
6
Fe
8
Al
10
12
14
16
0
2
4
6
8
τ
K
10
12
14
16
0
2
4
6
8
10
Ca
P
12
14
16
Fig. 11. Estimated s for major elements referenced to Ti in the reconstructed weathering profile developed on gneiss of the Mule Hole
watershed. Bulk density profile is also indicated.
storage potential. Quantifying and modeling the contemporary chemical weathering fluxes along this ecocline based on
hydrological and geochemical time series will be the scope
of future papers.
6. CONCLUSION
By combining investigations of geophysics, mineralogy
and geochemistry on the SEW of Mule Hole the following
conclusions can be arrived at:
A relationship is found between weathering indices (LoI,
WIP, CIA) and electrical resistivity in gneissic weathering
profiles, which helps to constraint the ERT profile modeling and to define the most likely limit between protolith
and regolith at the watershed scale.
ERT is a suitable method to assess protolith/regolith
geometry even in heterogeneous terrains. The average
regolith thickness calculated from the 12 ERT profiles
is 17.2 m. This result was obtained after correcting routine ERT with an estimate of ERT uncertainty using a
synthetic modeling approach. It showed that routine
ERT inversion underestimates regolith thickness by
15%. This underestimation is however related to the typical resistivity arrangement encountered in the Mule Hole
watershed. For other watersheds, the synthetic modeling
approach could lead to a different result.
Saprolitization processes at Mule Hole are limited and
lead to an immature material with low porosity and moderate base cation losses. In the incipient stages, biotite
and chlorite are broken down leading to the transfer of
Mg, Fe and K. Quartz and sericite are stable. Oligoclase
moderately weathers as indicated by the Na and Si transfer functions. Nonetheless due to its abundance, Na, Al
and Si are the elements that are the most significantly leached away.
Soil processes lead to more mature material with a 90%
loss of Na-plagioclase and 100% loss of biotite.
The immature soil and saprolite of the sub-humid zone of
the Kabini climatic gradient associated with geomorphologic features have a great potential to produce alkalinity
by chemical weathering. Depending on the runoff and
therefore climate variability with a more humid gradient
(i.e. intensification of the monsoon), the production of
alkalinity would increase and consequently increase the
atmospheric CO2 consumption.
ACKNOWLEDGMENTS
The Kabini river basin is part of the ORE-BVET project
(Observatoire de Recherche en Environnement – Bassin Versant
Expérimentaux Tropicaux, www.orebvet.fr). Apart from the specific support from the French Institute of Research for Development (IRD), the Embassy of France in India and the Indian
Institute of Science, our project benefited from funding from
IRD and INSU/CNRS (Institut National des Sciences de l’Univers/Centre National de la Recherche Scientifique) through the
French programmes ECCO-PNRH (Ecosphère Continentale: Processus et Modélisation – Programme National Recherche Hydrologique), EC2CO (Ecosphère Continentale et Côtière) and ACIEau. It is also funded by the Indo-French programme IFCPAR
(Indo-French Center for the Promotion of Advanced Research
W-3000). The multidisciplinary research carried on the Mule Hole
watershed began in 2002 under the aegis of the IFCWS (IndoFrench Cell for Water Sciences), joint laboratory IISc/IRD. We
959
thank the Karnataka Forest Department and the staff of the Bandipur National Park for all the facilities and support they provided.
P. de Parseval (SEM, microprobe), M. Thibaut (XRD), R. Wyns,
A. Bost and C. Kumar are thanked for their technical assistance.
A special thank to P. Mazzega for his help for the mineralogical inverse problem solutions.
We wish to express our sincere gratitude to Y. Gunnell and J.A.
West for providing thorough and valuable reviews.
M. Novak is thanked for his editorial handling of this paper.
REFERENCES
Anderson S. P., Blum J., Brantley S. L., Chadwick O. A., Chorover
J., Derry L. A., Drever J. I., Hering J. G., Kirchner J. W.,
Kump L. R., Richter D. and White A. F. (2004) Proposed
initiative would study Earth’s weathering engine. EOS (abs)
Trans. Am. Geophys. Union 85, 265–272.
Anderson S. P., von Blanckenburg F. and White A. F. (2007)
Physical and chemical controls on the critical zone. Elements 3,
315–319.
Barbiéro L., Parate H. R., Descloitres M., Bost A., Furian S.,
Mohan Kumar M. S., Kumar C. and Braun J.-J. (2007) Using a
structural approach to identify relationships between soil and
erosion in a non-anthropic forested area, South India. Catena
70, 313–329.
Beauvais A., Parisot J.-C. and Savin C. (2007) Ultramafic rock
weathering and slope erosion processes in a South West pacific
tropical environment. Geomorphology 83, 1–13.
Beauvais A., Ritz M., Parisot J.-C., Bantsimba C. and Dukhan M.
(2004) Combined ERT and GPR methods for investigating
two-stepped lateritic weathering system. Geoderma 119, 121–
132.
Beauvais A., Ritz M., Parisot J.-C., Dukhan M. and Bantsimba C.
(1999) Analysis of poorly stratified lateritic terrains overlying a
granitic bedrock in West Africa, using 2-D electrical resistivity
tomography. Earth Planet. Sci. Lett. 173, 413–424.
Bluth G. J. S. and Kump L. R. (1994) Lithologic and climatologic
controls of river chemistry. Geochim. Cosmochim. Acta 58,
2341–2359.
Bourgeon G. and Larqué P. (1992) Trait originaux de l’arénisation
des gneiss en ambiance carbonatée-confinée au sud-ouest du
plateau de Mysore (Inde). C. R. Acad. Sci. Paris 314, 299–304.
Braun J.-J., Ndam Ngoupayou J. R., Viers J., Dupre B., Bedimo
Bedimo J. P., Boeglin J. L., Robain H., Nyeck B., Freydier R.,
Sigha Nkamdjou L., Rouiller J. and Muller J.-P. (2005) Present
weathering rates in a humid tropical watershed: Nsimi site
(South Cameroon). Geochim. Cosmochim. Acta 69, 357–387.
Braun J. J., Pagel M., Herbillon A. and Rosin C. (1993)
Mobilization and redistribution of REE and Th in a syenitic
lateritic profile: a mass balance study. Geochim. Cosmochim.
Acta 57, 4419–4434.
Brimhall G. H., Lewis C. J., Ford C., Bratt J., Taylor G. and Warin
O. (1991) Quantitative geochemical approach to pedogenesis:
importance of parent material reduction, volumetric expansion
and eolian influx in laterization. Geoderma 51, 51–91.
Burke B. C., Heimstath A. M. and White A. F. (2007) Coupling
chemical weathering with soil production across soil-mantled
landscapes. Earth Surf. Proc. Land. 32, 853–873.
Buss H. L., Sak P. B., Webb S. M. and Brantley S. L. (2008)
Weathering of the Rio Blanco quartz diorite, Luquillo Mountains, Puerto Rico: Coupling oxidation, dissolution, and
fracturing. Geochim. Cosmochim. Acta 72, 4488–4507.
Cornu S., Lucas Y., Lebon E., Ambrosi J. P., Luizao F., Rouiller
J., Bonnay M. and Neal C. (1999) Evidence of titanium
960
mobility in soil profiles, Manaus, central Amazonia. Geoderma
91, 281–295.
Descloitres M. and Le Troquer Y. (2004) Sonde de diagraphie
électrique pour la mesure de la résistivité sur la paroi d’un
forage. Bulletin Officiel de la Propriété Industrielle. Institut
National de la Propriété Industrielle (INPI), French Patent No.
2845416.
Descloitres M., Ruiz L., Sekhar M., Legchenko A., Braun J. J.,
Mohan Kumar M. S. and Subramanian S. (2007) Characterization of seasonal local recharge using electrical resistivity
tomography and magnetic resonance sounding. Hydrol. Process. 22, 384–394.
Dethier D. P. and Lazarus E. D. (2006) Geomorphic inferences
from regolith thickness, chemical denudation and CRN erosion
rates near the glacial limit, Boulder Creek catchment and
vicinity, Colorado. Geomorphology 75, 384–399.
Dewandel B., Lachassagne P., Wyns R., MarÉchal J. C. and
Krishnamurthy N. S. (2006) A generalized 3-D geological and
hydrogeological conceptual model of granite aquifers
controlled by single or multiphase weathering. J. Hydrol. 330,
260–284.
Ehlen J. (2005) Above the weathering front: contrasting
approaches to the study and classification of weathered mantle.
Geomorphology 67, 7–21.
Green E. G., Dietrich W. E. and Banfield J. F. (2006) Quantification of chemical weathering rates across an actively eroding
hillslope. Earth Planet. Sci. Lett. 242, 155–169.
Gunnell Y. (1998a) Passive margin uplifts and their influence on
climatic change and weathering patterns of tropical shield
regions. Global Planet. Change 18, 47–57.
Gunnell Y. (1998b) Present, past and potential denudation rates: is
there a link? Tentative evidence from fission-track data, river
sediment loads and terrain analysis in the South Indian shield.
Geomorphology 25, 135–153.
Gunnell Y. (2000) The characterization of steady state in Earth
surface systems: findings from the gradient modelling of an
Indian climosequence. Geomorphology 35, 11–20.
Gunnell Y. and Bourgeon G. (1997) Soils and climatic geomorphology on the Karnataka plateau, peninsular India. Catena
29, 239–262.
Gunnell Y., Braucher R., Bourles D. and Andre G. (2007)
Quantitative and qualitative insights into bedrock landform
erosion on the South Indian craton using cosmogenic
nuclides and apatite fission tracks. Geol. Soc. Am. Bull.
119, 576–585.
Gunnell Y., Gallagher K., Carter A., Widdowson M. and Hurford
A. J. (2003) Denudation history of the continental margin of
western peninsular India since the early Mesozoic – reconciling
apatite fission-track data with geomorphology. Earth Planet.
Sci. Lett. 215, 187–201.
IUSS-Working-Group-WRB (2006) World reference base for soil
resources 2006. World Soil Resources Food and Agriculture
Organization of the United Nations, Rome.
Kump L. R., Brantley S. L. and Arthur M. A. (2000) Chemical
weathering, atmospheric CO2, and climate. Annu. Rev. Earth
Planet. Sci. 28, 611–667.
Kurtz A. C., Derry L. A., Chadwick O. A. and Alfano M. J. (2000)
Refractory element mobility in volcanic soil. Geology 28(8),
683–686.
Legchenko A., Descloitres M., Bost A., Ruiz L., Reddy M., Girard
J. P., Sekhar M., Mohan Kumar M. S. and Braun J.-J. (2006)
Resolution of MRS applied to the characterization of hardrock aquifers. Ground Water 44, 547–554.
Loke M. H. (2000) Electrical imaging surveys for environmental
and engineering studies. Technical note of Geotomo Software,
www.geoelectrical.com.
McNeill D. J. (1980) Electromagnetic terrain conductivity measurements at low induction numbers. Technical Note TN-6,
GEONICS.
Millot R., Gaillardet J., Dupre B. and Allegre C. J. (2002) The
global control of silicate weathering rates and the coupling with
physical erosion: new insights from rivers of the Canadian
Shield. Earth Planet. Sci. Lett. 196, 83–98.
Naqvi S. M. and Rogers J. W. (1987) Precambrian Geology of
India. Clarendon Press, Oxford University Press, New York.
Nesbitt H. W. and Young G. M. (1982) Early Proterozoic climates
and plate motions inferred from major element chemistry of
lutites. Nature 299, 715–717.
Oh N.-H. and Richter D. D. (2005) Elemental translocation and
loss from three highly weathered soil-bedrock profiles in the
southeastern United States. Geoderma 126, 5–25.
Oliva P., Viers J. and Dupre B. (2003) Chemical weathering in
granitic environments. Chem. Geol. 202, 225–256.
Parker A. (1970) An index of weathering for silicate rocks. Geol.
Mag., 501–504.
Price J. R. and Velbel M. A. (2003) Chemical weathering indices
applied to weathering profiles developed on heterogeneous
felsic metamorphic parent rocks. Chem. Geol. 202, 397–416.
Riebe C. S., Kirchner J. W. and Finkel R. C. (2003) Long-term
rates of chemical weathering and physical erosion from
cosmogenic nuclides and geochemical mass balance. Geochim.
Cosmochim. Acta 67, 4411–4427.
Robain H., Descloitres M., Ritz M. and Atangana Q. Y. (1996) A
multiscale electrical survey of a lateritic soil system in the rain
forest of Cameroon. J. Appl. Geophys. 34, 237–253.
Seaton W. J. and Burbey T. J. (2002) Evaluation of twodimensional resistivity methods in a fractured crystalline-rock
terrane. J. Appl. Geophys. 51, 21–41.
Shadakshara Swamy N., Jayananda M. and Janardhan A. S.
(1995) Geochemistry of Gundlupet gneisses, Southern Karnataka: a 2.5 Ga old reworked sialic crust. In India as a Fragment
of East Gondwana (eds. M. Yoshida, M. Santosh and A. T.
Rao). Gondwana Research Group.
Taboada T., Cortizas A. M., Garcia C. and Garcia-Rodeja E.
(2006) Particle-size fractionation of titanium and zirconium
during weathering and pedogenesis of granitic rocks in NW
Spain. Geoderma 131, 218–236.
Tarantola A. and Valette B. (1982) Generalized nonlinear inverse
problems solved using the least squares criterion. Rev. Geophys.
Space Phys. 20, 219–232.
Taylor G. and Eggleton R. A. (2001) Regolith Geology and
Geomorphology. John Wiley & Sons, Ltd.
Telford W. M., Geldart L. P. and Sheriff R. E. (1990) Applied
Geophysics, Second ed. Cambridge University Press, UK, p.
792.
Thomas M. F. (1994) Geomorphology in the Tropics. A Study of
Weathering and Denudation in Low Latitudes. John Wiley &
Sons.
Tripathi J. K. and Rajamani V. (2007) Geochemistry and origin of
ferruginous nodules in weathered granodioritic gneisses,
Mysore Plateau, Southern India. Geochim. Cosmochim. Acta
71, 1674–1688.
West A. J., Galy A. and Bickle M. (2005) Tectonic and climatic
controls on silicate weathering. Earth Planet. Sci. Lett. 235,
211–228.
White A. F., Blum A. E., Bullen T. D., Vivit D. V., Schulz M. and
Fitzpatrick J. (1999) The effect of temperature on experimental
and natural chemical weathering rates of granitoid rocks.
Geochim. Cosmochim. Acta 63, 3277–3291.
White A. F. and Brantley S. L. (1995) Weathering rates of silicate
minerals: an overview. In Reviews in Mineralogy (eds. A. F.
White and S. L. Brantley). Mineralogical Society of America.
White A. T. and Blum A. E. (1995) Effects of climate on chemical
weathering in watersheds. Geochim. Cosmochim. Acta 59, 1729–
1747.
White A. T., Blum A. E., Schulz M. S., Vivit D. V., Stonestrom D.
A., Larsen M., Murphy S. F. and Eberl D. (1998) Chemical
weathering in a tropical watershed, Luquillo Mountains, Puerto
961
Rico: I Long-term versus short-term weathering fluxes. Geochim. Cosmochim. Acta 62, 209–226.
Associate editor: Martin Novak
HYDROLOGICAL PROCESSES
Hydrol. Process. 15, 3287– 3301 (2001)
DOI: 10.1002/hyp.284
Geophysical surveys for identifying saline groundwater in
the semi-arid region of the central Altiplano, Bolivia
Roger Guérin,1 * Marc Descloitres,2 Anne Coudrain,3 Amal Talbi3 and Robert Gallaire4
1
UMR 7619 Sisyphe, Département de Géophysique Appliquée, Université Pierre et Marie Curie (Paris 6), case 105, 4 place Jussieu, 75252
Paris Cedex 05, France
2 IRD, Institut de Recherche pour le Développement, BP 182, Ouagadougou 01, Burkina Faso
3 UMR 7619 Sisyphe, Centre National de la Recherche Scientifique (CNRS), Université Pierre et Marie Curie (Paris 6), case 123, 4 place
Jussieu, 75252 Paris Cedex 05, France
4 IRD, Institut de Recherche pour le Développement, casilla 9214, La Paz, Bolivia
Abstract:
In the central part of the Bolivian Altiplano, the shallow groundwater presents electrical conductivities ranging from
0Ð1 to 20 mS/cm. In order to study the origin of this salinity pattern, a good knowledge is required of the geometry of
the aquifer at depth. In this study, geophysics has been used to complement the sparse data available from drill holes.
One hundred time-domain electromagnetic (TDEM) soundings were carried out over an area of 1750 km2 . About
20 geological logs were available close to some of the TDEM soundings. Three intermediate results were obtained
from the combined data: (i) the relationship between the electrical conductivity of the groundwater and the formation
resistivity, (ii) geoelectrical cross-sections and (iii) geoelectrical maps at various depths. The limited data set shows
a relationship between resistivity and the nature of the rock. From the cross-sections, a conductive substratum with a
resistivity of less than 1 Ðm was identified at most of the sites at depths ranging from 50 to 350 m. This substratum
could be a clay-rich formation containing brines. Using derived relationships, maps of the nature of the formation
(sandy, intermediate and clayey sediments) were established at depths of 10 and 50 m. Discrimination between sand
and clays was impossible where groundwater conductivity is high (>3 mS/cm). In the central part of the area, where
the groundwater conductivity is low, sandy sediments are likely to be present from the surface to a depth of more
than 200 m. Clayey sediments are more likely to be present in the south-east and probably constitute a hydraulic
barrier to groundwater flow. In conclusion, the study demonstrates the efficiency of the TDEM sounding method to
map conductive zones. Copyright  2001 John Wiley & Sons, Ltd.
KEY WORDS
Bolivian Altiplano; hydrogeology; TDEM sounding; formation resistivity and water conductivity
relationship; geophysical surveys
INTRODUCTION
Transient modelling of flow and transport is an important aid to understanding trends in groundwater quality
(Loftis, 1996) related to the effects of climate change (Jones, 1999) or to increasing groundwater uptake
(e.g. Gore et al., 1998). A transient quantitative approach of this type requires a good knowledge of the
spatial distribution of the aquifer thickness and of the nature of the sediments. Where such information is not
available, geophysical surveys should be considered.
The study area is situated in the central part of the Altiplano, a plateau covering 190 000 km2 . Lake Titicaca
is located in the north of this closed basin at an altitude of 3810 m, and the salt crust of Uyuni is located to
the south at an altitude of 3650 m (Figure 1). The Andes, rising to altitudes of over 4500 m, run along its
western and eastern boundaries. The economy is divided between mining and agriculture (Montes de Oca,
* Correspondence to: Dr Roger Guérin, UMR 7619 Sisyphe, Département de Géophysique Appliquée, Université Pierre et Marie Curie
(Paris 6), case 105, 4 place Jussieu, 75252 Paris Cedex 05, France. E-mail: guerin@ccr.jussieu.fr
Copyright  2001 John Wiley & Sons, Ltd.
Received 27 July 2000
Accepted 6 February 2001
3288
67°30'
67°35'
67°40'
67°45'
N
adero (1.8
3740
8060000
0
5
W
mS/cm)
E
30
37
8055000
17°30'
10 km
3720
8050000
8045000
water conductivity (mS/cm)
m
Rio Desag
u
S
Northing UTM Zone 19K (m)
8065000
67°50'
67°55'
R. GUÉRIN ET AL.
10
3.9
17°35'
2
1
17°40'
site 1
3720
8040000
30
8035000
17°45'
4
o
37
ALTIPLANO BASIN
e
fil
pr
3710
N
8030000
W
Titicaca
lake
Willa
Khara
E
17°50'
S
Ea
rn
ste
8025000
yll
o
pr
electrical
ssoundings
ra
am
ca
Poopo
lake
ra
8015000
highlands
and outcrops
ra
ng
e
30
68° W
68°
67°
655000
650000
645000
640000
635000
630000
625000
Uyuni
salar
620000
615000
18° S
610000
37
8010000
17°55'
piezometry
(m)
18°
665000
ille
o
er
ad
Hu
TDEM
soundings
1
ar
ille
ord
study
area
e
fil
660000
rd
8020000
nC
18°
Solito
TDEM5
Co
ster
17°
u
rio esag
D
We
LA PAZ
Easting UTM Zone 19K (m)
Figure 1. Location map of the study area. The main hydrogeological data (hydraulic head and water conductivity) are taken from
Coudrain-Ribstein et al. (1995), and have been completed to the south by some data from the 1998 survey. Ellipses indicate the 16
sites where geological logs, water conductivity and TDEM sounding data were available. The data obtained at these sites are used to
construct the diagram in Figure 3
1997). The scarcity and the salinity of water resources increase from north to south and represent a serious
obstacle in the development of the region.
Mean annual rainfall in the central part of the Altiplano is about 350 mm/year. The only permanent surface
water in this region is the Rio Desaguadero, which is fed mainly by the outlet of Lake Titicaca and, in the
study area, has a mean electrical conductivity of 1Ð8 mS/cm. Since the severe drought of 1982, a Bolivian
Hydrol. Process. 15, 3287– 3301 (2001)
GEOPHYSICAL SURVEYS FOR IDENTIFYING SALINE GROUNDWATER IN BOLIVIA
3289
non-governmental organization (YUNTA) has been active in the region attempting to provide clean, fresh
water by drilling wells and installing hand pumps to replace open hand-dug wells in the main aquifer, which
lies in fluvio-lacustrine sediments. The existing 100 wells have a limited penetration (10 to 20 m in general)
and provide only limited point information where the study area extends beyond 1750 km2 . The aims of this
study were to achieve a better knowledge of the geometry of this phreatic aquifer at depth and to contribute
to the study on the origin of the salinity of the aquifer, which varies widely in space (Coudrain-Ribstein
et al., 1995). In terms of electrical conductivity, the groundwater presents a wide range of values from 0Ð1 to
20 mS/cm.
Geophysical methods have been used to identify the nature and the geometry of the aquifer at depths of at
least a few hundred metres. Electrical resistivity expressed in Ðm (or its inverse, conductivity theoretically
expressed in S/m and in practice in mS/cm, 1 Ðm corresponds to 1 S/m D 10 mS/cm), is widely considered
to be a relevant parameter for hydrogeological studies. Resistivity values are indeed particularly sensitive to
the porosity, the water content, the mineralization of the water and the nature of the rocks (McNeill, 1980).
Many site-specific geophysical studies have tried to link resistivity to the hydraulic properties of the aquifer
(Kelly, 1977; Mazác et al., 1990; Cassiani and Medina, 1997). Among geophysical resistivity methods, direct
current soundings are generally well suited for shallow groundwater investigations down to the first 100 m
below the soil surface (Ebraheem et al., 1997). In our study area, 24 electrical soundings were carried out
over a period of 15 days by YUNTA to site a number of drill holes (Ledezma et al., 1995). However, these
data showed severe limitations with regard to the investigation depth. In almost all the locations, the shallow
layers present very low values of resistivity (<10 Ðm) and they channel the electric current lines close to
the surface. This phenomenon limits the investigation depth to the first few dozen metres for the maximum
distance between the current electrodes (AB), which was limited to 1000 m. Penetration to more than 100 m
in depth requires a minimum AB separation of about 5000 m, which not only would be very time consuming
but also would require a powerful current source.
In such ground conditions, electromagnetic methods provide an alternative approach. An audiomagnetotelluric (AMT) survey had been carried out previously 50 km north of our survey area by Ritz et al. (1991)
but the depth of exploration of AMT does not allow good resolution of the shallow layers and, in addition,
provides information to a depth of several kilometers, which is beyond the scope of this study. Consequently,
we decided to use time-domain electromagnetic (TDEM) soundings as proposed by Fitterman and Stewart
(1987). This method has been used successfully in saline water environments (Goldman et al., 1996) and in
difficult areas with dry surface conditions (Robineau et al., 1997). Using TDEM, the geophysicist can explore
the ground at depths ranging between a few metres and a few hundred metres. Moreover, this method is
particularly sensitive and efficient in conductive environments.
STUDY AREA AND HYDROGEOLOGICAL BACKGROUND
The central Bolivian Altiplano is a complex Tertiary–Quaternary region of intermountain foreland basins.
The very thick (3000 m) accumulation of sediments was deposited during Late Oligocene and Miocene times
(Baby et al., 1990).
The western part of the area under study is hilly and formed by the flank of a Tertiary overthrust fault
system. All the Tertiary formations crop out through this fault, called the Chuquichambi Fault (Hérail et al.,
1997). Immediately to the east of these hills, Tertiary sandstone units form confined aquifers. The eastern half
of the province is a remarkably flat plain that consists of a series of largely lacustrine Quaternary sediments
(GEOBOL, 1996) overlying the same Tertiary system that outcrops to the west. The area under study is
bounded by the Rio Desaguadero to the north and east and by rugged hills to the west. The faulted Palaeozoic
structure outcrops to the east of the area and underlies the Tertiary formation at great depth. Several dacite
domes, some of which are mineralized, exist in the eastern part of the area (Columba and Cunningham, 1993).
Hydrol. Process. 15, 3287– 3301 (2001)
3290
R. GUÉRIN ET AL.
The results of the hydrogeological study carried out in the area, which are published elsewhere (Coudrain
et al., 2000), can be summarized as follows. The Quaternary sediments comprise coarse to fine sands, clayey
sands and clays. The groundwater level generally is between 2 and 10 m below the soil surface. From the
piezometric map (Figure 1), it can be seen that the aquifer is continuous with the Rio Desaguadero, constituting
a hydraulic limit of the aquifer. This map reveals the different types of recharge and discharge of the aquifer.
Fresh water (with electrical conductivity of less than 0Ð3 mS/cm) infiltrates from temporary runoff at the hinge
line between the western hills and the plain. The relatively saline water of the Rio Desaguadero (with a mean
value of electrical conductivity of 1Ð8 mS/cm) recharges the aquifer downstream of piezometric level 3735 m.
The two main types of discharge are the outflow towards the south through the arbitrary south-eastern boundary
corresponding to profile 1 in Figure 1, and the evaporative outflow from the aquifer (Coudrain-Ribstein et al.,
1998). Steady-state hydrogeological modelling with present-day conditions enabled each of these recharge
and discharge terms to be computed; they were found to be of the same order of magnitude, between 10 and
30 million cubic metres per year.
Soil salinity is high in the flat eastern part of the area and is related mainly to the evaporative outflow
from the aquifer. This is confirmed by LANDSAT and SPOT satellite images and soil cores (Ledezma et al.,
1995). The groundwater conductivity of the Quaternary aquifer increases regularly from west to east, ranging
respectively from 0Ð1 to 20 mS/cm (Figure 1). Extremely high values of 300 mS/cm are limited to the brines
of the dacite domes that were sampled by the mining company.
TDEM SURVEY AND DATA INTERPRETATION
Method
TDEM is a time domain controlled-source method that uses transient electromagnetic field diffusion
(Nabighian and Macnae, 1991; McNeill, 1994). A current is alternatively turned on and off in a rectangular
loop of wire laid out on the ground as a transmitter source. A static primary magnetic field, perpendicular
to the plane of the transmitter loop, is created during current-on time. At turn-off time, an electromotive
force is induced in the ground by the decaying primary field, producing eddy currents in conductive bodies.
These induced currents penetrate into the ground. They create a secondary magnetic field with an amplitude
that decreases over time. This is measured at the surface by a receiver coil or loop at several pre-set times
during the turn-off period. The decay shape reflects the resistivity depth distribution. By increasing the period
over which the decaying voltages are observed, information is obtained about deeper formations. The TDEM
method uses a variety of transmitter and receiver configurations, the most common being the central loop
configuration, which has a small receiver at the centre of the transmitter loop. The advantages of TDEM
are its good sensitivity to conductive formations, a depth of investigation that is greater than the transmitter
loop side length, good lateral and vertical resolution and, in addition, the convenience of not requiring any
galvanic contact with the ground. The main disadvantages are poor sensitivity to resistive formations (above
about 500 Ðm) and its limitation to depths of more than 15 m.
Field survey
The survey was carried out with a Protem47 system (Geonics Ltd) using a transmitter loop of 100 m ð
100 m, with a central receiver loop of 15 m ð 15 m, and an injection current of 1Ð8 A. More than
100 soundings were measured on a grid comprising several profiles orientated SW–NE (Figure 1) over a
period of 15 days. The distance between profiles was approximately 5 km. A sounding was made every 3 km
along each profile. The observation interval was sometimes varied to take advantage of the proximity of drill
holes.
Data interpretation
All the TDEM soundings were interpreted as one-dimensional layered models using Temix interpretation
software (Interpex Ltd). Two arguments allow corroboration of our assumption of layered earth. In one location,
Hydrol. Process. 15, 3287– 3301 (2001)
3291
measurements carried out in the three main directions showed that the structure is mainly one-dimensional,
i.e. the layers are horizontal. Profiles made in the central part of the basin using an electromagnetic induction
system (with an EM34 from Geonics Ltd) with an intercoil spacing of 40 m did not show any two-dimensional
structure. However, at one site in the vicinity of an outcrop of magmatic rock in the south-eastern part of
the study area, the sounding curve cannot be interpreted without considering a three-dimensional structure.
Considering that the Quaternary sediments are mostly lacustrine, the layered earth assumption is quite probably
valid. Two types of interpretation were conducted. The first involved a user-defined starting model (generally
three or four layers following the form of the sounding curves) and used a curve-matching algorithm that
yields a best-fit solution. This procedure also gives numerous equivalent solutions, as can be seen in Figure 2.
The results of these interpretations were used to draw the resistivity cross-sections. The second method of
interpretation involved a computer-generated model (using 15 layers) that defines a smooth variation of the
resistivity with depth. This procedure is based on an Occam-type inversion (Constable et al., 1987) and was
used in our study to construct resistivity maps at different depths.
Data interpretations of TDEM and electrical soundings from site 1 in the centre of the study area are
compared in Figure 2. As mentioned previously, the electrical soundings in the present environment give
TDEM1
Resistivity Thickness
(Ω.m)
(m)
3.9
2.3
20
6
326
13
0.23
103
100
0
0
100
100
Depth (m)
Apparent resistivity (Ω.m)
104
Depth (m)
(a)
200
300
200
300
10
400
400
1
0.001
0.01
0.1
1
10
100
500
0.01
Time (ms)
500
0.1
1
10
0.1
100
Resistivity (Ω.m)
1
10
100
Resistivity (Ω.m)
SE1
(b)
0
Resistivity Thickness
(Ω.m)
(m)
0.8
5
5.5
1
14
65
8
10
Depth (m)
Apparent resistivity (Ω.m)
100
10
20
30
1
1
10
AB/2 (m)
100
1000
0.1
1
10
100
1000
Resistivity (Ω.m)
Figure 2. Example of TDEM interpretation procedures and comparison with electrical sounding interpretation at site 1: (a) left, TDEM
sounding curve; centre, model with curve matching (solid line) and equivalent solutions (dashed lines); right: smoothed 15 layer interpretation;
(b) left, electrical sounding curve; right, model (solid line) and equivalent solutions (dashed lines)
Hydrol. Process. 15, 3287– 3301 (2001)
3292
R. GUÉRIN ET AL.
only well-defined information in the top 6 m or so. The two conductive shallow layers (5 Ðm, 0Ð8 m thick;
1 Ðm, 5Ð5 m thick) preclude the measurement of information below this depth. On the other hand, the
TDEM sounding yields information from deeper formations. A third layer appears at a depth of 23Ð9 m
(13 Ðm, 326 m thick) overlying a very conductive substratum (0Ð23 Ðm). The smooth model confirms the
considerable thickness of resistive layers and a very conductive substratum. The depth of the substratum is not
well-defined (ranging from about 270 to 420 m) as shown by equivalent curves and the smooth interpretation,
whereas the parameters of the upper layers are well-defined by both types of interpretation. This example
demonstrates the advantages of using the TDEM method in such conductive areas; in direct current (electrical
sounding) the shallow conductive layers mask deeper formations, whereas deep conductive layers are detected
using the TDEM method even with a limited transmitter loop size (e.g. 100 m ð 100 m).
RESULTS
Relationship between water conductivity, formation resistivity and the nature of the rocks
Within the study area are16 sites (see Figure 1) where the following information was available: (i) a good
geological description obtained previously by drilling, (ii) a subsurface groundwater conductivity value and
(iii) a TDEM sounding from nearby (and sometimes also an electrical sounding). These combined data
sets allowed calibration of the geophysical interpretation. Moreover they also were used to investigate the
relationships between groundwater conductivity w , formation resistivity f and the nature of the sediments.
For each site where the sediments were known as a result of drilling, the TDEM interpretation was
constrained to fit the interface depths. The resistivity, f , of each formation was then calculated. An example
of this procedure is illustrated in Figure 3 for the Solito drill hole and the corresponding sounding TDEM
5. From the drill hole information, a major sharp interface can be seen at a depth of 40 m between sands
and massive clays. From the TDEM interpretation, below the superficial conductive layer, the sandy layer
presents a resistivity of 22 Ðm. A more conductive layer (9 Ðm) corresponds to the underlying clays. In
this well, the water conductivity w is 2 mS/cm. Hence this site provides two points on the graph of formation
resistivity f versus water conductivity w (Figure 3).
It is interesting to note that the 21 points available identify four domains. Domain A corresponds to sandy
formations for which it is possible to calculate the porosity  using the well-known Archie’s law (Archie,
1942),
f w D am
where f , w and  are expressed in Ðm, S/m and by a fraction, respectively. The dimensionless coefficients
a and m depend on the rock type. In our case, we have taken the values a D 0Ð88 and m D 1Ð37 following the
values given by Keller (1988) for sands, sandstone and some limestone. In domain A, the calculated porosity
 varies from 29 to 36%, which is an acceptable range for such media. One point shows a value of 60% that
is too high to be realistic. It is situated close to domain B and probably corresponds to more clayey sands for
which Archie’s law is not applicable.
Domain B corresponds to intermediate formations and marks the transition between sands and clays. In
this geological context, the transition between sand and clay is not sharp as shown by some drill holes logs.
Consequently the delimitation of domain B in Figure 3 should not be considered as final, particularly because
it is only defined by four points.
Domain C corresponds to clayey formations with a resistivity ranging from about 3 to 12 Ðm.
Domain D, i.e. the intersection of domains A and C, corresponds to sandy and clayey formations with
values of groundwater conductivity greater than 3 mS/cm. It is remarkable that sandy formations saturated
with saline water (10 mS/cm) display resistivity values as low as 3 Ðm. The porosities of 40 and 44%
calculated from Archie’s law for two pairs of values are too high to be realistic. These sandy formations may
include a significant percentage of clay.
Hydrol. Process. 15, 3287– 3301 (2001)
3293
100
Type of formation
90
80
70
porosity (%),
calculated from
Archie Law
Sandy
Sand
Coarse sand
50
Intermediate
Clayey sandstone
Clayey sand
40
Clayey
Clay
60
36
60
B
22
Undetermined
35
31
20
A
30
29
Resistivity
(Ω.m)
10 20 30
pumping =>
σw = 2 mS/cm
water level
30
sand
22
10
−40 m
40
(fixed)
99
8
7
34
Depth (m)
Formation resistivity ρf(Ω.m)
30
6
44
5
C
4
massive
clays
9
D
40
3
−150 m
sounding
TDEM 5
2
Solito
drill hole
0.1
1
10
Water conductivity σw (mS/cm)
40
50
60
70
80
90
30
20
4
5
6
7
8
9
3
2
2
0.4
0.5
0.6
0.7
0.8
0.9
0.3
0.2
1
100
Figure 3. Plot of formation resistivity against water conductivity for different types of formations. Inset at the right is an example of a
constrained interpretation of the TDEM sounding number 5 with the geological log obtained at the Solito drill hole. The specific values
inferred from this example are shown in squares on the diagram. The possible range of formation resistivities calculated using the Temix
software are shown as deviation bars
This diagram is a key tool for transforming the geophysical results into hydrogeological information on
the nature of the sediments. However, the boundaries between the four domains should not be considered as
final because of the small number of data points available. Moreover, for any pair of values situated inside
the saline domain D, it is impossible to discriminate between sand and clay.
Geoelectrical cross-section
The two cross-sections presented in Figure 4 were established on the basis of the TDEM soundings
interpreted with the assumption of one-dimensional structures. These cross-sections were drawn using the
Hydrol. Process. 15, 3287– 3301 (2001)
121
1.04
16
46
241
7
28
0.3
3.5
65
0
205
1.4
9
66
90 16
17 20
0
1.04
15
2.8
1.06
12
2.8
6 km
132
61
14
48
14
255
74
1.7
7
51
189
118
0.5
2.8
15
201
3.5
27
68
67
68bis
5.8 4.2
4.7 8.2
20
33
112
42
12
47
5.2
348
8
1
20
240
3.2
0.6
4.4
10
49
69
130
33
6 km
1.5
20
70
132
40
13
1.8
13
71
9
22
5
10
Solito
drill-hole
78
47
6
(2 km to
north)
2
14
3.1
8
166
12
0.43
4.7
14
10. 3
164
22
50
0.3
4.5
72
305
2.5
0.35
6.5
0.9
0.4
5.3
73
170
50
25
51
6.5
27
4.9
74
7.6
145
14
121
11
71
4.4
75
0.52
2
6.2
52
6.2
76
20
135
4.2
76
0.2
1.8
53
124
2.5
268
60
1.57
2.9
77
2.5
18
NE
54
ρ(Ω.m)
2.5
4
9
16
60
Figure 4. Geoelectrical cross-sections derived from TDEM modelling along profiles 1 and 4. The location of the profiles is shown in Figure 1. Numbers along the top of each
profile are TDEM sounding stations; those in italics and underlined inside the graphs are depth with equivalency bars to indicate the possible depth ranges of the interpretation;
other numbers inside the graphs are resistivity values inferred from a classic layered interpretation. The vertical scale is exaggerated with respect to the horizontal scale
300
m
250
200
150
100
50
0
Profile 4
150
m
100
50
0
SW
Profile 1
3294
R. GUÉRIN ET AL.
Hydrol. Process. 15, 3287– 3301 (2001)
3295
results of interpretation with a minimum number of layers. The soil surface of the area investigated is flat.
For several thousand years, the area was covered by a lake. The highest water surface level was reached
15 000 years ago. The profiles are presented with respect to their depth below the soil surface. Altitude
was deduced using topographical maps of the area (contours every 20 m). In profile 1, the altitude of the
soil surface varies between 3715 m (TDEM 46) and 3706 m (TDEM 5). In profile 4, the first two TDEM
soundings (65 and 66) were carried out close to the hinge line with the relief. Their altitude reaches 3740
and 3730 m respectively. The cross-sections show the lateral resistivity variation with an exaggerated vertical
scale. The apparently steep slope between two geophysical units is lower than 1Ð5° in the study area, and the
juxtaposition of each one-dimensional interpretation to obtain the geoelectrical section therefore is justified.
The resistivity range is divided into six major units. In the north-eastern part of profile 4, a high resistivity unit
is plotted in black and has values greater than 60 Ðm. This could correspond to the underlying Palaeozoic
formation that crops out as foothills oriented NW–SE (shown in Figure 1). The deepest unit has resistivity
values lower than 2Ð5 Ðm. The resistivity of this conductive substratum presents values as low as 0Ð2 Ðm
in the southern part of the study area at a depth between 112 and 170 m (profile 1). In profile 4, the resistivity
and the depth of this conductive substratum show greater variations ranging from 0Ð3 to 2Ð5 Ðm and 78 to
348 m in depth. Above this substratum, the structure is more complex. The four resistivity units (between 2Ð5
and 4 Ðm, between 4 and 9 Ðm, between 9 and 16 Ðm, between 16 and 60 Ðm) do not show a simple
spatial distribution.
In profile 1, formation resistivity decreases from south-west to north-east, although remaining in a
conductive range when water conductivity increases (Figure 1). In the central part of profile 1, close to
the Solito drill hole, a small resistive unit (22 Ðm) appears at a relatively shallow depth. This unit is the
only aquifer able to contain freshwater.
In profile 4, the resistivity values of all units above the conductive substratum are relatively high. The
conductive substratum is shallowest in the central part of the section (near site 8). Resistivities are higher to
the south-west of this point than to the north-east. This could reflect an increase in water conductivity from
the foothills of Huyllamarca located in the south-west towards the north-east.
Resistivity maps at different depths
Based on the results of geostatistical analysis, we compiled resistivity maps at four depths derived from an
Occam type interpretation of each TDEM sounding (Figure 5). At a depth of 10 m, a resistive channel (with
resistivity greater than 20 Ðm) orientated NW–SE, probably corresponds to a freshwater channel. The water
supply in this channel seems to come from the Huyllamarca foothills to the west. Still, at a depth of 10 m,
the eastern part shows a conductive body in the same area as the saline groundwater plume (Figure 1). At
greater depth, the resistive channel does not reach the south-eastern boundary of the study area (see map at
50 m depth) but is closed off in the south. It almost disappears on the deeper maps (110 and 160 m). On these
deeper maps, a conductive plug is present in the southern part. This may correspond to either a formation
saturated with saline water or to a clay-rich formation. The resistivity contours become more tortuous with
depth. The deeper map (depth: 160 m) shows well-defined irregularities (boundaries between conductive and
resistive formations, dykes, etc.).
Maps of the types of formation
Lateral variation in the electrical conductivity of groundwater is generally smooth and a good estimation
of this value can be inferred from a limited number of measurements. This is the case in the present study.
Using conductivity maps of the groundwater, the value of the formation resistivity at a given point can be
related to a given type of formation by referring to the diagram in Figure 3. In this way the resistivity maps
at depths of 10 and 50 m have been converted into maps showing the type of formation, but it is important
to bear in mind the following limitations and hypotheses:
Hydrol. Process. 15, 3287– 3301 (2001)
3296
R. GUÉRIN ET AL.
660000
8060000
8050000
650000
8040000
640000
8030000
630000
8020000
620000
610000
10 m
8010000
ρ (Ω.m)
98.5
50 m
50.5
32.3
20.7
13.3
8.5
5.4
110 m
3.5
2.2
1.4
0.9
0.5
160 m
660000
8060000
650000
8050000
N
30 km
630000
8030000
E
620000
S
8020000
W
40 km
640000
8040000
610000
8010000
20 km
10 km
0 km
Figure 5. Resistivity maps at different depths (10, 50, 110 and 160 m) derived from smooth Occam TDEM interpretation
Hydrol. Process. 15, 3287– 3301 (2001)
3297
1. Most of the drill holes are less than 25–35 m deep. Consequently we consider that the water conductivity
map in Figure 1 is valid only for this depth range. However, we consider that this map provides a reliable
image of the variation in water conductivity for the entire area, even though it is constructed by interpolation
between irregularly distributed drill holes. The water conductivity map is coherent even with a poor sampling
because of the spatial continuity of this feature. This hypothesis can be challenged in the southern and
south-eastern parts of the zone where the drill holes are scarce.
2. The diagram in Figure 3 is considered sufficiently reliable for the entire area but only for subsurface
formations.
3. The information about the nature of the formation inferred from each TDEM sounding location is considered
to be representative of a wide zone around the sounding (circles of 2 to 5 km for example). This hypothesis
can be challenged if the lateral variations in the facies of the formations occur within a distance of less
than 2 km throughout.
The maps at 110 or 160 m were not drawn because the first of the two above-mentioned hypotheses are
no longer valid. The maps at 10 and 50 m presented in Figure 6 show that the type of formation (sand and
coarse sand, clayey sandstone and clayey sand, and clay) may vary spatially very rapidly.
In the map at 10 m, the global distribution follows an interlocking scheme. The sandy formations are present
in the central part of the zone and extend to the south. The intermediate formations are located mainly in the
north-western part, whereas the clayey formations surround a wide zone of undetermined formations (sandy
or clayey) located in the east. As mentioned previously, it is not possible to predict the type of formation
in this zone from the geophysical results without additional information. A more clayey deposit along the
Rio Desaguadero may roughly explain this distribution, whereas sandy formations could be related to the
proximity of the hills at the west. However, this scheme is no longer valid for deeper deposits.
In the map at 50 m, the results are related more closely to the main aquifer because the data used are only
representative of the saturated zone. The overall arrangement shows a wide sandy zone in the north, reduced
in the centre to a 10 km-wide channel bordered on both sides by clayey or intermediate formations, although
a clayey zone almost blocks its extension to the south. The eastern part is made up mainly of undetermined
sediments surrounded by three clayey zones. This arrangement suggests that this area also may be clayey.
Moreover a clayey formation was identified between 40 and 135 m at the Solito drill hole located in the
south. A thick formation of this type should extend far beyond the Solito area. However, even if the presence
of sandy sediments cannot be excluded in this undetermined zone, the surrounding clayey formations as well
as the clayey zone in the south would seriously limit groundwater flow at this depth.
DISCUSSION
Validity of the relationship between the resistivity of the formation and water conductivity
Many studies have attempted to link the resistivity obtained by geophysical measurements to hydraulic
parameters. When hydraulic conductivity data are available from pumping tests, a relationship is sometimes
found between hydraulic conductivity and the resistivity of the formation in some site-specific cases (Mazác
et al., 1990). In the area under study, the pumping tests available provided transmissivity values ranging from
5 ð 104 m2 /s close to the fault line in the Tertiary formation to 102 m2 /s in the central part of the area. The
only geological logs available in this area are provided by drill hole reports. From such limited information,
the different rock formations could be divided only into four types. Nevertheless, the results shown in Figure 3
indicate a rough arrangement of the data points that satisfy the resistivity ranges indicated in other studies (see,
for example, resistivity ranges in the table given by Reynolds, 1997). In our case, the points corresponding to
the clay domain indicate a resistivity constantly below 12 Ðm, whereas for the sand domain they generally
remain in a higher range (10–80 Ðm). Logically, intermediate formations would be located between these
two domains. Domain D (undetermined formations) corresponds either to sands saturated by water with a
Hydrol. Process. 15, 3287– 3301 (2001)
3298
R. GUÉRIN ET AL.
8065000
0
5
N
10 km
8060000
W
E
S
8055000
8050000
8045000
8040000
8035000
8030000
8025000
8020000
8015000
TDEM
soundings
10 m
8010000
a
8065000
8060000
8055000
8050000
8045000
8040000
8035000
8030000
8025000 Formation type
Sandy
8020000
Intermediate
8015000
Clayey
50 m
Undetermined
665000
660000
655000
650000
645000
640000
635000
630000
625000
620000
615000
610000
8010000
b
Figure 6. Maps of the interpreted type of formations at 10 and 50 m (a and b, respectively). These maps are inferred from the resistivity
maps for depths of 10 and 50 m (Figure 5), from the water conductivity map (Figure 1) and from the diagram in Figure 3
Hydrol. Process. 15, 3287– 3301 (2001)
3299
conductivity of more than 3 mS/cm or to clayey formations. The decrease in the resistivity value to below
a few Ðm for sandy formations containing salt or brackish waters has been noted in many studies and is
predicted by Archie’s law. Consequently we believe that, despite the small number of data points, Figure 3
can be used with confidence for preliminary identification of the formations.
Three (or four) -layer model versus a 15-layer smooth model
As stated previously, only a few drill holes show a variation in resistivity with depth. For example,
the Solito drill log (shown in Figure 3) shows an abrupt transition between sands and clays at a depth of
40 m. On the other hand, the Willa Khara drill hole (not presented in this paper) indicates a succession
of a few metres of clays and intermediate coarser sediments. The complexity of such deposits cannot be
imaged by the geophysical data, particularly at depths of 50 to 100 m. To illustrate this point, we calculated
the TDEM response to a 40-m-thick intrastratified medium made up of eight layers, each 5 m thick (clay,
5 Ðm and sand, 20 Ðm), overlain by a 50-m-thick sandy formation. The substratum is clayey. The resulting
synthetic sounding curve was interpreted as either a three-layered model or a smooth model. Neither of the
interpretations was able to image the intrastratified medium correctly. The three-layer interpretation indicates
an intermediate formation with a resistivity value (8 Ðm) closer to the clayey domain. This illustrates one
limitation of the resolution of the TDEM sounding and also indicates that an intrastratified medium would
be identified as a clayey zone rather than as a sandy one. Hence, in our case, the resistivity cross-sections
or maps can be considered only as a generalized image of resistivity variations without providing detailed
information.
Geological and hydrogeological implications of the geophysical results
The hydrogeological studies of Coudrain et al. (2000) showed that the aquifer is recharged by the Rio
Desaguadero and by infiltration of runoff from the western hills. The general direction of the groundwater
flow is from north-west to south-east. The present-day salinity may be related to three processes: (i) the
diffusion from the saline palaeolake that covered the area 13 000 years BP, (ii) the accumulation of salt in
the unsaturated zone by evaporation from the aquifer over long periods and the subsequent return of the salt
into the aquifer during short humid periods, (iii) upward leakage from deep brines. In order to quantify these
three processes, it is important to know the geometry of the aquifer and its transmissivity variations. These
two parameters are essential to determine the available quantity of water (fresh or saline) and the velocity of
the groundwater.
At this point, the hydrogeological information that may be derived from the geophysical results can be
summarized as follows:
The southern part of the zone is almost closed off at a depth of 50 m by clayey formations that limit
groundwater flow towards the south. However, close to the surface, a shallow permeable path may channel
fresh groundwater through sandy formations.
In the centre, resistivity values at 110 m indicate formations between 10 and 30 Ðm. If the water
conductivity values measured near the surface were extended to more than 100 m (this assumption is,
however, highly hypothetical), this zone would correspond to formations between intermediate and sandy
types (Figure 3). This would suggest the presence of a thick layer of sandy material in the central part of
the area.
In the eastern part, the occurrence of clayey formations surrounding some undetermined formations would
again limit groundwater flow to a few paths. If this last undetermined formation were in fact a clay-rich
formation, it would go a long way towards explaining the hydrochemical and isotopic data showing that the
groundwater is several thousand years old and that the substitution of brackish waters by fresh waters is
very slow.
The conductive substratum identified by most of the TDEM soundings at depths ranging from 50 to 350 m
is probably not closely related to the groundwater flow regime. However, its nature is questionable because
Hydrol. Process. 15, 3287– 3301 (2001)
3300
R. GUÉRIN ET AL.
the resistivity values are quite low for a sedimentary context of this type; less than 1 Ðm and sometimes
less than 0Ð3 Ðm, comparable to the resistivity value of saline water at 3%, i.e. 0Ð15 Ðm (Telford et al.,
1990; Hurwitz et al., 1999). To the north of our study area, Ritz et al. (1991) have identified very conductive
zones at depths of more than 1 km. They propose an explanation involving Tertiary sand and clay deposits
(or volcanoclastic deposits). In our case, because this substratum appeared to be shallower, we hypothesized
a clayey formation containing brines. In order to corroborate this hypothesis, we performed three TDEM
soundings 600 km to the south on the hard crust of the Uyuni dry salt lake (Uyuni salar), where a 123 m hole
had been drilled by Risacher (1992). Our results indicate a layer between 6 and 70 m thick, with a resistivity
of 0Ð3 Ðm that corresponds to an intercalation of clayey sediments and brines within crystallized salt layers.
However, in our area, the geometry of the top of the substratum is not as flat as the salt lake crust, and does
not explain how brines and clays could occur in a sedimentary context of this type. We conclude that a deep
drill hole is required in the central Altiplano to determine the actual nature of this substratum.
CONCLUSION
The efficiency of the TDEM method to map accurately conductive zones between the surface and a depth of
a few hundred metres in a short survey period has been confirmed.
For each type of formation, a relationship between the resistivity of the formations and groundwater
conductivity has been derived using a limited set of control data. Even if the data set is not complete, it
enables a rough classification of the nature of the formation into four domains (sandy, intermediate, clayey or
undetermined) following the range of the groundwater conductivity. It has been used extensively to convert
the geophysical results into hydrological information.
The geoelectrical cross-sections as well as the resistivity maps at different depths allow delineation of
the boundaries between fresh and saline waters, and between sandy and clayey formations. From north to
south, from west to east and from shallow to greater depths, the resistivity of the sedimentary formations
decreases and they are underlain by a conductive substratum (clay and/or brine). A paleochannel containing
fresh groundwater is indicated, narrowing from north to south and with depth. This may correspond to a path
along which fresh water pushes the shallow saline water towards the south-east, while a deeper conductive
plug, probably clay, presents an obstacle to a fresh water outlet towards the south.
Two maps of the type of formation at depths of 10 and 50 m were compiled using the conductivity–resistivity relationship and the geophysical interpretation. They provide a preliminary rough estimate of the
distribution of the sandy formations, mostly present in the north and the centre of the zone and limited
in the south by a clayey barrier at a depth of 50 m. Intermediate or clayey formations surround this zone to
the west and to the east, while the sediments in the eastern part remain undetermined. This point illustrates one
limitation to geophysical mapping in a zone where conductive waters (>3 mS/cm) are present. The deeper
part of the aquifer (100 to 300 m) only remains difficult to interpret from TDEM results owing to the lack of
geological information.
ACKNOWLEDGEMENTS
This study was funded by the French National Programme for Research in Hydrology (PNRH) from INSU,
involving the Institut de Recherche pour le Développement (IRD) and the Unité Mixte de Recherche (UMR
7619 Sisyphe) from Paris 6 University. The authors wish to thank J. C. Salinas, M. Guzman, A. Osco and
the IRD team in La Paz for their field contributions.
REFERENCES
Archie E. 1942. The Electrical Resistivity Log as an Aid in Determining Some Reservoir Characteristics. Technical Publication 1422,
Petroleum Technology, American Institute of Mining and Metallurgical Engineer, New York, USA; 8.
Hydrol. Process. 15, 3287– 3301 (2001)
3301
Baby P, Sempere T, Oller J, Barrios L, Hérail G, Marocco R. 1990. Un bassin en compression d’âge ologo-miocène dans le sud de l’Altiplano
bolivien. Comptes Rendus de l’ Académie des Sciences, Paris (Séries IIA) 311: 341– 347.
Cassiani G, Medina MA. 1997. Incorporating auxiliary geophysical data into ground-water flow parameter estimation. Groundwater 35:
79–91.
Columba MC, Cunningham CG. 1993. Geologic model for the mineral deposits of the La Joya district, Oruro, Bolivia. Economic Geology
88: 701–708.
Constable SC, Parker RL, Constable CG. 1987. Occam’s inversion: a practical algorithm for generating smooth models from electromagnetic
sounding data. Geophysics 52: 289–300.
Coudrain A, Talbi A, Ledoux E, Ribstein P, Marsily G de. 2000. Changements climatiques et recyclage sur plusieurs milliers d’années du
chlorure entre lacs et eau souterraine: exemple de l’Altiplano bolivien. International Association of Hydrological Sciences Bulletin 262:
409–416.
Coudrain-Ribstein A, Pratx B, Quintanilla J, Zuppi GM, Cahuaya D. 1995. Salinidad del recurso hidrico subterraneo del Altiplano central.
Bulletin de l’ Institut Français d’ Études Andines 24: 483– 493.
Coudrain-Ribstein A, Pratx B, Talbi A, Jusserand C. 1998. Is the evaporation from phreatic aquifers in arid zones independent of the soil
characteristics? Comptes Rendus de l’ Académie des Sciences, Paris (Séries IIA) 326: 159–165.
Ebraheem AA, Senosy MM, Dahab KA. 1997. Geoelectrical and hydrogeochemical studies for delineating ground-water contamination due
to salt water intrusion in the northern part of the Nile delta, Egypt. Groundwater 35: 216– 222.
Fitterman DV, Stewart MT. 1987. Transient electromagnetic sounding for groundwater. Geophysics 51: 995– 1005.
GEOBOL. 1996. Geological map (mapa de geolog ı́a). Oruro SE19. GEOBOL (Servicio Geológico de Bolivia) La Paz, Bolivia.
Goldman M, Hurwitz S, Gvirtzman H, Rabinovich B, Rotstein Y. 1996. Application of the marine time-domain electromagnetic method in
lakes: the sea of Galilee, Israel. European Journal of Environmental and Engineering Geophysics 1: 125–138.
Gore KP, Pendke MS, Gurunadha R, Gupta CP. 1998. Groundwater modelling to quantify the effect of water harvesting structures in
Wagarwadi watershed, Parbhani district, Maharashtra, India. Hydrological Processes 12: 1043– 1052.
Hérail G, Rochat P, Baby P, Aranibar O, Lavenu A, Mascle G. 1997. El Altiplano norte de Bolivia: evolución geológica terciaria. In El
Altiplano, ciencia y conciencia en los Andes, Gonzalez RC (ed.). Universidad de Chile: Santiago; 33–44.
Hurwitz S, Goldman M, Ezersky M, Gvirtzman H. 1999. Geophysical (time domain electromagnetic model) delineation of a shallow brine
beneath a freshwater lake, the sea of Galilee, Israel. Water Resources Research 35: 3631– 3638.
Keller GV. 1988. Rock and mineral properties. In Electromagnetic Methods in Applied Geophysics 1: Theory, Nabighian MN (ed.). Society
of Exploration Geophysicists: Tulsa, Oklahoma, USA; 13–51.
Kelly WE. 1977. Geoelectrical sounding for estimating hydraulic conductivity. Groundwater 15: 420–425.
Ledezma RD, Jiménez A, Moreau S, Coudrain-Ribstein A. 1995. Recursos hı́dricos para riego en la provincia Villarroel, Altiplano boliviano.
Bulletin de l’ Institut Français d’ Études Andines 24: 473– 482.
Loftis JC. 1996. Trends in groundwater quality. Hydrological Processes 10: 335– 355.
Jones JAA. 1999. Climate change and sustainable water resources: placing the threat of global warming in perspective. Hydrological Sciences
Journal 44: 541– 557.
Mazác O, Cislerova M, Kelly WE, Landa I, Venhodova D. 1990. Determination of hydraulic conductivities by surface geoelectrical methods.
In Geotechnical and Environmental Geophysics 2: Environmental and Groundwater , Ward SH (ed.). Society of Exploration Gepphysicists:
125–131.
McNeill JD. 1980. Electrical Conductivity of Soils and Rocks. Technical Note TN-5, Geonics Limited, Mississauga, Ontario, Canada; 22.
McNeill JD. 1994. Principles and Application of Time Domain Electromagnetic Techniques for Resistivity Sounding. Technical Note TN-27,
Geonics Limited, Mississauga, Ontario, Canada; 15.
Montes de Oca I. 1997. Geograf ı́a y recursos naturales de Bolivia. EDOBOL Press: La Paz, Bolivia; 614.
Nabighian MN, Macnae JC. 1991. Time domain electromagnetic prospecting methods. In Electromagnetic Methods in Applied Geophysics
2: Applications, Nabighian MN (ed.). Society of Exploration Gepphysicists: 427– 520.
Reynolds JM. 1997. An Introduction to Applied and Environmental Geophysics. Wiley: New York; 796.
Risacher F. 1992. Géochimie des lacs salés et croûtes de sel de l’Altiplano bolivien. Sciences Géologiques Bulletin (Strasbourg, France) 45:
135–214.
Ritz M, Bondoux F, Hérail G, Sempéré T. 1991. A magnetotelluric survey in the northern Bolivian Altiplano. Geophysical Research Letters
18: 475–478.
Robineau B, Ritz M, Courteaud M, Descloitres M. 1997. Electromagnetic investigations of aquifers in the Grand Brûlé coastal area of Piton
de la Fournaise volcano, Reunion island. Groundwater 35: 585– 592.
Telford WM, Geldart LP, Sheriff RE. 1990. Applied Geophysics 2nd edn. Cambridge University Press: Cambridge; 770.
Hydrol. Process. 15, 3287– 3301 (2001)
Resolution of MRS Applied to the Characterization
of Hard-Rock Aquifers
by Anatoly Legchenko1, Marc Descloitres2, Adelphe Bost2, Laurent Ruiz2, Mohan Reddy2, Jean-Franc
xois Girard3,
2
2
2
Muddu Sekhar , M.S. Mohan Kumar , and Jean-Jacques Braun
Abstract
The performance of the Magnetic Resonance Sounding (MRS) method applied to the investigation of heterogeneous hard-rock aquifers was studied. It was shown using both numerical modeling and field measurements that
MRS could be applied to the investigation of the weathered part of hard-rock aquifers when the product of the free
water content multiplied by the thickness of the aquifer is >0.2 (for example, 10-m-thick layer with a 2% water
content). Using a currently available one-dimensional MRS system, the method allows the characterization of
two-dimensional subsurface structures with acceptable accuracy when the size of the subsurface anomaly is equal
to or greater than the MRS loop. However, the fractured part of hard-rock aquifers characterized by low effective
porosity (<0.5%) cannot be resolved using currently available MRS equipment. It was found that shallow water in
the weathered part of the aquifer may screen MRS signals from deeper water-saturated layers, thus further reducing the possibility of investigating deeper fractured aquifers. A field study using the NUMISplus MRS system
developed by IRIS Instruments was carried out on an experimental watershed in southern India. A heterogeneous
unconfined aquifer in a gneissic formation was successfully localized, and MRS results were confirmed by drilling
shortly after the geophysical study. The top of the aquifer revealed by MRS was found to be in a good agreement
with observed static water level measurements in boreholes.
Introduction
Magnetic Resonance Sounding (MRS) is sensitive specifically to ground water. The method allows a noninvasive
detection of subsurface water using magnetic resonance
measurements. Thus, a direct detection of subsurface water
is the main advantage of MRS compared to other geophysical tools used for hydrogeological investigation. MRS is
a large-scale method, and the investigated volume can be
1Corresponding author: Institut de Recherche pour le Développement (IRD), LTHE, BP53, 38041, Grenoble Cedex 9, France;
(33) 4 76 82 50 63; fax (33) 4 76 82 50 14; anatoli.legtchenko@
hmg.inpg.fr
2Indo French Cell for Water Science, Department of Civil Engineering, Indian Institute of Science, Bangalore 560012, India
3Bureau de Recherches Géologiques et Minières (BRGM), 3,
Avenue Claude Guillemin, BP 6009, 45060, Orléans Cedex 2, France
Received April 2005, accepted November 2005.
Copyright ª 2006 The Author(s)
Journal compilation ª 2006 National Ground Water Association.
doi: 10.1111/j.1745-6584.2006.00198.x
approximated by a cube of 1.5 3 a where 20 a 150 m
is the side length of a square loop. The geometry and water
content of water-saturated layers can be obtained after inversion of MRS data.
However, when applied to the investigation of hardrock aquifers, MRS has some specific limitations that
should be taken into account. The usual conceptual model
of the hard-rock aquifer describes several zones that
together form the same reservoir (Lachassagne et al. 2001;
Wyns et al. 2004). The upper zone consists of weathered
and decayed rocks of clayey-sandy composition. Their hydraulic conductivity is usually low, but their water-retention
capacity can be significant, and they play the major part of
storativity in the functioning of the aquifer. The underlying
weathered-fissured zone is characterized by almost horizontal fractures that diminish in density with depth and often
vertical fractures and fissures that enhance flow relationships with the fractures in the bedrock. This zone is characterized by increasing values of hydraulic conductivity. The
deeper zone is represented by fractured bedrock. It is highly
Vol. 44, No. 4—GROUND WATER—July–August 2006 (pages 547–554)
547
permeable only locally, where affected by tectonic fracturing, and it has very limited storativity. The geometry of
these parts is guided mainly by the geological context and
the history of the weathering processes and often exhibits
two-dimensional (2D) and three-dimensional (3D) features.
In this paper, we investigate whether MRS is able to
characterize such a heterogeneous aquifer. We theoretically analyze the resolution of MRS over one-dimensional
(1D) and 2D structures using numerical modeling. An
example of a MRS study of a heterogeneous hard-rock
aquifer in southern India is presented. Geophysical results
are compared with those measured in boreholes.
MRS Method
A brief description of MRS is provided subsequently.
A more detailed presentation of the method can be found
in the literature (for example, Legchenko and Valla 2002;
Legchenko et al. 2004; Lubczynski and Roy 2004; Roy
and Lubczynski 2003; Weichman et al. 2000; Yaramanci
et al. 2002).
To an outside observer, the MRS field setup appears
very similar to that of the transient electromagnetic
method with a coincident transmitting/receiving loop
(Figure 1). It consists of a wire loop laid out on the
ground, usually in a square with the side length between
20 and 150 m. The depth of investigation is proportional
to the loop size. The loop is then energized by a pulse of
alternating current i(t) ¼ I0cos(x0t). The frequency of the
current is equal to the Larmor frequency of the protons in
the geomagnetic field. The Larmor frequency x0 ¼ 2pf0
is given by the spin Larmor resonance condition:
x 0 ¼ c p B0
ð1Þ
with B0 being the magnitude of the geomagnetic field and
cp/2p ¼ 4.257707 3 107 Hz/T, the gyromagnetic ratio for
protons. The Larmor frequency is obtained from measurements of the geomagnetic field (B0) on the surface using
a proton magnetometer. Depending on the global geographical location of the investigated area, the geomagnetic
field varies between ~20,000 and 60,000 nT, and the Larmor frequency correspondingly varies between 800 and
2800 Hz. For example, in southern India, the geomagnetic
field is around 40,500 nT and the Larmor frequency is correspondingly around 1730 Hz. The pulse causes precession
of spin magnetization of the protons in ground water
around the geomagnetic field, which creates an alternating
electromagnetic field that can be detected using the same
loop after the pulse is terminated. Oscillating with the Larmor frequency, the MRS signal has an exponential envelope
and depends on the pulse moment q ¼ I0s with I0 and s
being, respectively, the amplitude and duration of the pulse.
We assume the spin system to be linear, which is an
approximation, but it allows calculating the MRS
response using the first three harmonics generated by the
pulse. The signal induced in the receiver loop is proportional to the sum of the flux of all precessing magnetic
moments of protons. Considering Equation 1 and the
pulse harmonics, the signal induced in the receiving loop
becomes (Legchenko 2004):
548
A. Legchenko et al. GROUND WATER 44, no. 4: 547–554
Figure 1. MRS method for ground water investigation:
(a) MRS system in a field; (b) two aquifers and MRS loop;
(c) MRS results (vertical distribution of the water content).
e0 ðqÞ ¼
Z X
K 21
B1k ðrÞeju0k ðrÞ M’k ðq; rÞ wðrÞdVðrÞ ð2Þ
I0k
x0
V k¼1
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
where B1k ðrÞ ¼ ReðB1k Þ2 1 ImðB1k Þ2 is the magnetic
field transmitted by kth harmonic of the pulse and
I0k
is
the
amplitude
of
this
harmonic,
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2
2
M’k ðq; rÞ ¼ ReðM’k Þ 1 ImðM’k Þ is the perpendicular to the geomagnetic field component of the spin magnetization for protons per unit volume for each harmonic
with the corresponding frequency offset xk, K being the
number of harmonics, and r ¼ r(x, y, z) is the coordinate
vector. Considering that both the magnetic field generated
by spin magnetization and the transmitted magnetic field
B1 are complex and assuming a coincident transmitting/
receiving loop, we can express the phase of the signal
generated by volume dV as
u0k ¼ tan21 ðImðM’k Þ=ReðM’k ÞÞ
1 2 tan21 ðImðB1k Þ=ReðB1k ÞÞ ¼ uxk 1 2uqk
ð3Þ
where uxk and uqk are the phase shifts due to, respectively, the frequency offset and the electromagnetic shift
caused by the electrical conductivity of rocks. Water
distribution in the subsurface w(r) is the solution of
Equation 2.
Measurements of the magnetic resonance signal
are performed while varying the pulse moment q. The
currently available 1D inversion scheme assumes a horizontal stratification and reveals a vertical distribution of
the water content w(z), where z is the depth. An increase
in the water content in the MRS log corresponds to an
aquifer (Figure 1).
MRS provides an estimate of the total amount of
water inR the subsurface VMRS (m3). In a general case,
VMRS ¼ V wðrÞdVðrÞ. Assuming a horizontal stratification, an estimate of the volume of water per surface unit
in a layer of thickness z (a column of a height z) can
be obtained from
VMRS 2 H ¼
Z
wðzÞdz
ð4Þ
z
This volume VMRS2H (m3/m2) provides an estimate
of the amount of water in horizontally stratified earth and
corresponds to a hydrostatic water column used in hydrogeology.
Numerical Modeling
In order to investigate the capability of MRS to
characterize fractured rock aquifers, numerical modeling was carried out. Considering the usual conceptual
model of the hard-rock aquifer (Lachassagne et al.
2001), the weathered zone was assumed to exhibit 2D
features and to contain more water than the underlying
weathered-fissured and fractured zones. For modeling,
the weathered-fissured and fractured zones are united in
one fissured-fractured zone considered as a 1D structure
with low effective porosity. A geomagnetic field of
40,500 nT with an inclination of 20 (which corresponds
to southern India) and a two-turns 50-m-side square loop
(which corresponds to a standard configuration of the
NUMISplus system manufactured by IRIS Instruments)
were assumed.
reliably detected by MRS (VMRS2H ¼ 0.01 3 30 ¼ 0.3 m).
A synthetic data set was computed using Equation 2.
Inversion of the synthetic signals was carried out using
the well-known Tikhonov regularization method (Tikhonov and Arsenin 1977). It is known from the previous
study (Legchenko and Shushakov 1998) that the MRS
inverse problem is ill posed and it is characterized by a
decrease in resolution with increasing depth. When the
aquifer is deeper than approximately the loop side (in our
case, 50 m), it cannot be resolved. Taking it into account,
one can see that the second aquifer is generally well resolved when there is no water in the first aquifer (Figure 3a).
However, when the water content in the first aquifer is
increasing, it corrupts the resolution of the second aquifer
(Figures 3b through 3d). Synthetic data are well fitted by
the theoretical signals calculated using inversion results,
thus demonstrating the accuracy of inversion (Figure 4).
The misfit was calculated as average absolute (eabs) and
relative (erel) errors. If we assume J soundings ( j ¼ 1, 2,
., J) with Nj (Nj ¼ N1, N2, ., NJ) pulse moments in
each sounding and MRS signal measured (or simulated)
for each pulse moment, e(d)i, j (i ¼ 1, 2, ., Nj), then the
errors can be calculated as
eabs ¼
J
X
j¼1
1D Modeling
1D modeling enables a threshold for the MRS instrument applied to the detection of homogeneous fractured
aquifers to be established. Moreover, using a 1D synthetic
data set, one can estimate the accuracy of the inversion
algorithm, currently restricted to the 1D case. Considering the NUMISplus system, it is known that the aquifer
would be reliably detected if the signal from this aquifer
is >10 nV.
Since the amplitude of the MRS signal is proportional to the total volume of water in the subsurface VMRS
and for the horizontally stratified earth to its estimate
VMRS2H (Equation 4), this parameter (VMRS2H) will be
used for establishing a threshold of aquifer detection. In
Figure 2, the maximum amplitude of the MRS signal
against the depth to the top of the aquifer for different
values of VMRS2H is depicted for a 50- 3 50-m square
loop with two turns. The results show that MRS is able to
detect, for example, a 20-m-thick aquifer with a water
content of 1% down to 20 m and with a water content of
2% down to ~50 m. Consequently, if the fractured part of
the hard-rock aquifers is characterized by an average
effective porosity of 0.1% to 0.2%, then the total water
volume is small (VMRS2H < 0.1, even for thick structures)
and it would be barely detectable. Hence, only the weathered part of the aquifer with usually larger effective
porosity and consequently larger water volume (VMRS2H
> 0.15) could be a target for the MRS method (Figure 2).
Considering the conceptual model of the hard-rock
aquifer, let us assume an aquifer composed of two parts:
(1) a 10-m-thick aquifer at a depth from 5 to 15 m with
varying water content and (2) a 30-m-thick aquifer with a
1% water content at a depth from 15 to 45 m (Figure 3).
According to previous results, the second part could be
!
21
Nj
Nj
J X
X
jeðdÞi; j 2 eðtÞi; j j;
j¼1 i¼1
erel ¼ ðeabs =emax Þ100%
ð5Þ
where e(t)i, j are the theoretical amplitudes and emax ¼
maxi; j ðeðdÞi; j Þ is the maximum of the measured signal
(Figure 3).
For the examples presented in Figure 3 (models 1
through 4), the relative errors of inversion are 2.7%,
1.6%, 1.7%, and 2.7%, respectively. We observe that for
all the models, the accuracy of inversion is comparable,
even for model 4 (Figure 3d), which was not able to
Figure 2. Maximum amplitude of the MRS signal vs. depth
of the aquifer for different volumes of water in the subsurface (50- 3 50-m loop, two turns).
549
Figure 3. Example of inversion of synthetic signals: models and resolved water content w(z).
detect the second aquifer. This can be explained by
a larger signal generated by the first aquifer, which
screens a small signal from the second aquifer (the signal
from the second aquifer alone is given by model 1). Some
more details about the inversion of MRS signals can be
found in the literature (Legchenko and Shushakov 1998;
Legchenko 2005; Mohnke and Yaramanci 2002; Weichman
et al. 2002).
2D Modeling
At present, in the only commercially available MRS
instrument (NUMISplus), the same loop is used as the
transmitting and receiving antenna (a coincident Tx/Rx
loop configuration). A multichannel measuring system
for a separate transmitting (Tx) and receiving (Rx) loop
configuration does not exist, and 2D inversion software
is still under development (Hertrich et al. 2005). For this
reason, it is a matter of practical importance to study
errors introduced by the 1D inversion algorithm when
Figure 4. Synthetic signals and theoretical fit after inversion
of the models presented in Figure 3.
550
measuring MRS data from a 2D target using a coincident
Tx-Rx loop. For modeling, a 2D water content anomaly
that is simulated by a rectangular 10-m-thick parallelepiped with varying width and with a 5% water content set
in a matrix with a 0.5% water content was used. Its location, thickness, and width are given by Figure 5. The 2D
MRS response was computed in two steps using Equation 2. First, the signal from the matrix with a 0.5%
water content was calculated assuming the horizontal
stratification, and then it was added to the signal calculated by limiting the integration in Equation 2 by the
volume occupied by the parallelepiped with a 4.5%
water content.
For the inversion, the solution is composed of a set
of infinite horizontal layers with varying water content
(1D inversion). The goal of this exercise is to see how
large the errors caused by a target smaller than the loop
size would be. The inversion results of three individual
soundings (models M1 through M3) with the loop position shown in Figure 5 reveal that the depth and thickness
of the anomaly are reasonably well resolved, but the error
Figure 5. Models of 2D water-saturated structure: a rectangular parallelepiped with varying width (25, 40, and 55 m) at
a depth of 5 m and with a water content of 5% in a matrix
with a water content of 0.5% from 5 to 50 m.
Figure 6. Numerical modeling: (a) inversion of 2D models M1 through M3 presented in Figure 5; (b) the theoretical fit calculated
using inversion results.
in the water content depends on the size of the anomaly
and may be very large (Figure 6a). In all cases, water in
the matrix below the anomaly was not detected. Despite
large errors in the water content, the misfit between the
synthetic signals and the theoretical fits (Figure 6b) calculated using Equation 5 shows that the fitting errors are
smaller than 8%, being larger when the target is smaller
(7.9%, 6.5%, and 6.2% for models M1, M2, and M3,
respectively). Hence, MRS signals from a 2D target can
be well fitted by a 1D solution, thus introducing large errors in the results when the target is smaller than the loop.
We then moved the loop over the target with a step
of 25 m. For each sounding position, a 1D inversion was
performed and the results of all 1D inversions were interpolated in the x direction (distance). The results presented
in a cross section (Figure 7) show that the model was
resolved. The misfit calculated using Equation 5 is 3.4%.
Two important limitations are highlighted. First, the water
content was underestimated by a factor of 2. Second, one
can observe that the screening effect consists of attenuation of the signal from the deeper aquifer by that from
water in the shallow aquifer, thus underestimating the
water content in the deeper aquifer.
Figure 7. Numerical modeling: a cross section of 2D water
content distribution obtained by interpolation of the 1D
inversion of MRS stations along a profile over the target presented on the plot as a rectangle (Figure 5, model M1).
In this paper, we study the relationship between the
size of subsurface structures and errors caused by 2D effects
considering the hard-rock aquifers. More comprehensive
MRS modeling considering the resolution of structures
smaller than loop 2D and 3D structures (water-filled
karst) can be found in the literature (Boucher et al. 2005;
Girard et al. 2005; Vouillamoz et al. 2003).
Field Example
MRS was applied to the characterization of a heterogeneous hard-rock aquifer in southern India (Figure 8).
The bedrock is mainly represented by gneiss with local
amphibolite and hornblendite inclusions. At some outcrops, the rock exhibits a highly foliated structure. Observations at outcrops show that the foliation is close to
the vertical (75 dip angle). Boreholes drilled in the outlet
area of the watershed (Figures 8 and 9) allow quasicontinuous monitoring of the static water level. The depth
of the boreholes, the corresponding MRS stations, and the
depth to the water are presented in Table 1. In the 2004
rainy season, the depth to ground water varied between 5
Figure 8. Location map of the investigated area in southern
India.
551
Table 1
Moole Hole: Boreholes and MRS Stations in the Outlet Area
Static Water
Level (m)
Depth of the
Borehole (m)
P1
P2
P3
P5
P6
P7
3.86
21.66
28.63
40.21
38.54
4.43
P8
P9
P10
P11
P12
P13
Borehole
Date of Water Level
Measurement
MRS
Station
Depth to the
Top of Aquifer (m)
Date of MRS
Measurement
45
60
24
54
55
18
November 26, 2004
November 26, 2004
November 26, 2004
November 13, 2004
November 26, 2004
November 26, 2004
9.53
18
November 13, 2004
18.84
18.42
5.01
11.76
11.01
23
29
16
40
40
November 26, 2004
November 13, 2004
November 26, 2004
December 4, 2004
December 4, 2004
MRS9
MRS12
MRS15
MRS2
MRS3
MRS1
MRS9
MRS8
MRS10
MRS11
MRS11
MRS9
MRS6
MRS5
4
No MRS signal
No MRS signal
No MRS signal
No MRS signal
3
4
2
10
16
16
4
11
10
November 13, 2004
November 24, 2004
November 27, 2004
November 12, 2004
November 29, 2004
November 13, 2004
November 21, 2004
November 20, 2004
November 17, 2004
November 23, 2004
November 23, 2004
November 13, 2004
November 22, 2004
November 16, 2004
and 35 m around the watershed. Thus, the hydrogeological parameters of the weathered zone were expected to be highly variable.
All MRS measurements were carried out using
NUMISplus equipment. The data processing was performed using NUMIS standard interpretation software.
Fifteen MRS stations with a 50- 3 50-m square loop were
carried out in the area (Figure 9). For this loop, the depth
of investigation of MRS could be considered as ~60 m. In
a more heterogeneous area (known from boreholes),
a 25-m distance between the neighboring stations was
maintained. Along the profile, MRS revealed significant
variations in the aquifer characteristics. For demonstration
Figure 9. Moole Hole watershed: position map of MRS
loops and boreholes along the profile.
552
purposes, we present the 1D inversion results (Figure 10a)
and measured signals (Figure 10b) for three soundings
(MRS1, MRS13, and MRS17). The noise measurements
show the signal to noise ratio for the data. The fitting
error calculated for each sounding using Equation 5 reveals 6.2%, 9.2%, and 7% for the MRS1, MRS13, and
MRS17 stations, respectively. The repeatability of the
MRS measurements was carefully verified. It was confirmed that the internal instability of the instrument or
external electromagnetic noise could not cause these
variations. Consequently, the observed variations can
only be explained by a lateral heterogeneity of the subsurface.
The MRS results along the profile in the outlet area
are presented in Figure 11a. In each of the two plots, the
elevation is represented along the y-axis and the distance
along the profile along the x-axis. The color scale in each
plot presents the MRS water content. In the southern part
of the profile, a highly heterogeneous aquifer was detected. Its thickness varies between 4 and >40 m. This
thick structure was unsuspected before the MRS survey.
The aquifer is unconfined, and the top of this aquifer
revealed from MRS data corresponds well with the static
water level as measured in boreholes. The aquifer is not
continuous, and it was not detected by MRS in the northern part of the profile.
Considering the outlet area, a 2D block model of
the water content distribution was created (Figure 11b).
To do this, the water content cross section derived from
MRS measurements was used as a first estimate. MRS
did not reveal any water in the unsaturated zone and in
fractured rocks. However, using hydrogeological considerations, we assumed that the water content above the
water table is equal to zero and it is equal to 0.2% in
fractured rocks. Each block in the model was set considering the size, position, and water content derived from
MRS results. In each block, the water content was allowed to attain one of the discrete values 0.2%, 0.5%,
1.0%, 1.5%, or 2%. Then, in all blocks the water content
Figure 10. (a) Example of the 1D inversion of individual soundings (MRS1, MRS13, and MRS17) demonstrating variations in
the water content distribution along the profile in the Moole Hole area. (b) Measured signals for these soundings vs. the pulse
parameter and the theoretical fit calculated using inversion results.
was kept fixed and the size of each block was iteratively
adjusted aiming to minimize the misfit between the field
measurements and the theoretical signal calculated from
the block model. For every loop position, synthetic signals from each block were calculated separately using
Equation 2 and limiting the integration by the volume of
the block. The total signal measured by the loop is
a sum of these signals. The misfit was calculated using
Equation 5. For the model presented in Figure 11, the
misfit was found at 9.1%. However, as has been shown
Figure 11. MRS results in the outlet area: (a) cross section
of the water content distribution derived from inversion
of field measurements; (b) a 2D block model of the MRS
water content. The thick and thin dashed lines show the
static water level measured in boreholes and the depth of
investigation of the MRS, respectively.
in the previous sections, while the geometry and relative
variations of the water content along the profile are
rather reliable even in a 2D environment, a good fit of
data does not guarantee a correct value of the water content revealed in the investigated aquifers. The errors can
be estimated only if the exact geometry of the subsurface structures is known. The screening effect may
also corrupt the reliability of the MRS results concerning the deeper part of the subsurface located between 75
and 350 m along the profile.
A comparison between the static water level measured in boreholes and the depth to the top of the aquifer
revealed by MRS is presented in Table 1. The correspondence between these two data sets is depicted in Figure 12.
In the investigated area, the static water level measured in
two boreholes both inside one MRS loop may vary
Figure 12. Moole Hole: comparison of static water level
measurements in boreholes with depth to the top of the aquifer given by the corresponding MRS station.
553
significantly (for example, P7 and P8, Figures 9 and 12).
MRS provides results averaged over the volume affected
by the loop that often integrates different subsurface
structures. For the same reason, two MRS stations (MRS8
and MRS10) around borehole P8 reveal very different
results depending on what volume of the subsurface contributes more to the measured signal. Thus, when the
aquifer cannot be approximated by the horizontally stratified media considering the loop size, the MRS system
operating under 1D assumption can provide only qualitative results.
Conclusions
Numerical modeling reveals that MRS can reliably
detect water-saturated rocks when the water volume produces a signal larger than the threshold of the instrument.
The water volume can be estimated by a product of the
MRS water content and the thickness of the structure.
For example, a weathered rock aquifer with a water
content of 2% can be detected if its thickness is >10 m.
The fractured part of hard-rock aquifers characterized
by low effective porosity (<0.5%) yields a very small
MRS response, which is below the threshold of currently
available MRS instruments, and thus these aquifers cannot be detected. It was shown that water in a shallow
aquifer (i.e., between 0 and 15 m deep) may act as
a screen for a deeper aquifer, if this deeper aquifer
contains less water in comparison with the shallow aquifer. The screening effect may corrupt the reliability of
the MRS results concerning the deeper part of the subsurface.
A currently available measuring device and inversion
routine operating within a 1D assumption is able to give
a satisfactory image of the aquifers when the size of subsurface anomalies is equal to or greater than the MRS
loop. Otherwise, larger errors should be expected.
MRS was applied in southern India to the investigation of a heterogeneous aquifer with the bedrock represented mainly by gneiss. It was shown that MRS could
be an efficient tool for characterizing the weathered part
of this aquifer. The fissured part remains undetectable
with currently available MRS equipment. Generally,
a good correlation between MRS results and borehole
measurements was observed. The top of this unconfined
aquifer is correctly determined by MRS, except when the
water level varies significantly at a distance smaller than
the loop size.
Acknowledgments
The field work was supported by the French national
research program ECCO-PNRH, IRD, and the Watershed
Project of the Indo French Cell for Water Sciences of
IISc and IRD in Bangalore. The authors are grateful to
the Karnataka Forest Department for providing access to
the site and to C. Kumar for his assistance. We specially
thank P. Lachassagne, M. Lubczynski, A. Mazzela, and
B. Steinich for their comments, which improved the clarity and readability of the paper.
554
References
Boucher, M., K. Chalikakis, J.-M. Baltassat, A. Legchenko, and
J.-F. Girard. 2005. Localization of a karst conduit using
magnetic resonance soundings, a case study. In Extended
Abstracts of the 11th European Meeting of Environmental
and Engineering Geophysics, Palermo, Italy, 5-8 September
2005, paper A027, CD ROM edition. Co Production,
Houten, the Netherlands.
Girard, J.-F., M. Boucher, A. Legchenko, and J.-M. Baltassat.
2005. Numerical modeling of magnetic resonance signal
from water-filled cavities. In Extended Abstracts of the 11th
European Meeting of Environmental and Engineering Geophysics, Palermo, Italy, 5-8 September 2005, paper B037,
CD ROM edition. Co Production, Houten, the Netherlands.
Hertrich, M., M. Braun, and U. Yaramanci. 2005. Magnetic
resonance soundings with separated transmitter and
receiver loops. Near Surface Geophysics 2005, 3, no. 3:
141–154.
Lachassagne, P., R. Wyns, P. Bérard, T. Bruel, L. Chéry, T.
Coutand, J.F. Desprats, and P. Le Strat. 2001. Exploitation
of high-yields in hard-rock aquifers: Downscaling methodology combining GIS and multicriteria analysis to
delineate field prospecting zones. Ground Water 39, no. 4:
568–582.
Legchenko, A. 2005. Improved modeling of the magnetic resonance signal in presence of shallow aquifers. Near Surface
Geophysics 3, no. 3: 121–130.
Legchenko, A. 2004. Magnetic resonance sounding: Enhanced
modeling of a phase shift. Applied Magnetic Resonance 25,
no. 3–4: 621–636.
Legchenko, A., J.-M. Baltassat, A. Bobachev, C. Martin,
H. Robain, and J.-M. Vouillamoz. 2004. Magnetic resonance sounding applied to aquifer characterization.
Ground Water 42, no. 3: 363–373.
Legchenko, A., and P. Valla. 2002. A review of the basic principles for proton magnetic resonance sounding measurements. Journal of Applied Geophysics 50, no. 1–2: 3–19.
Legchenko, A.V., and O.A. Shushakov. 1998. Inversion of surface NMR data. Geophysics 63, no. 1: 75–84.
Lubczynski, M., and J. Roy. 2004. Magnetic resonance sounding: New method for ground water assessment. Ground
Water 42, no. 1: 291–303.
Mohnke, O., and U. Yaramanci. 2002. Smooth and block inversion of surface NMR amplitudes and decay times using
simulated annealing. Journal of Applied Geophysics 50,
no. 1–2: 163–177.
Roy, J., and M. Lubczynski. 2003. The magnetic resonance
sounding technique and its use for groundwater investigations. Hydrogeology Journal 4, no. 11: 455–465.
Tikhonov, A., and V. Arsenin. 1977. Solution of Ill-Posed Problems. John Wiley & Sons Inc. New York.
Vouillamoz, J.-M., A. Legchenko, Y. Albouy, M. Bakalowicz,
J.-M. Baltassat, and W. Al-Fares. 2003. Localization of
saturated karst aquifer with magnetic resonance sounding
and resistivity imagery. Ground Water 41, no. 5: 578–587.
Weichman, P.B., D.R. Lun, M.H. Ritzwoller, and E.M. Lavely. 2002.
Study of surface nuclear magnetic resonance inverse problems.
Journal of Applied Geophysics 50, no. 1–2: 131–147.
Weichman, P.B., E.M. Lavely, and M. Ritzwoller. 2000. Theory
of surface nuclear magnetic resonance with applications to
geophysical imaging problems. Physical Review E 62, no.
1: 1290–1312.
Wyns, R., J.-M. Baltassat, P. Lachassagne, A. Legtchenko,
and J. Vairon. 2004. Application of proton magnetic resonance soundings to groundwater reserve mapping in weathered basement rocks (Brittany, France).
Bulletin de Socie´te´ geólogique de France 175, no. 1:
21–34.
Yaramanci, U., G. Lange, and M. Hertrich. 2002. Aquifer characterization using surface NMR jointly with other geophysical techniques at the Nauen/Berlin test site. Journal of
Applied Geophysics 50, no. 1–2: 47–65.
Catena 53 (2003) 229 – 253
www.elsevier.com/locate/catena
Study of infiltration in a Sahelian gully erosion area
using time-lapse resistivity mapping
Marc Descloitres a,*, Olivier Ribolzi b, Yann Le Troquer a
a
Unité de Recherche 027 Geovast, Institut de Recherche pour le Développement (IRD), BP 182,
Ouagadougou 01, Burkina Faso
b
Unité de Recherche 049 ECU, IRD, BP 182, Ouagadougou 01, Burkina Faso
Received 28 March 2002; accepted 18 March 2003
Abstract
Observing that concentrated runoff destroys indurate and impermeable surface horizons to form
gullies on Sahelian slopes, we investigated whether these gullies are preferential places for deep
infiltration and groundwater recharge processes. The primary aim of this study is to determine if
resistivity mapping is an appropriate method to use for locating recharge zones from the surface. The
study area, in northern Burkina Faso, is a typical (1 ha) gully erosion area located at the outlet of an
82-ha catchment with solonetz soils and a crystalline basement. Taking advantage of a long dry
season followed by a short rainy season, we made use of a time-lapse approach to carry out electrical
resistivity mapping and monitor apparent resistivity variations that occurred in the soils during the
rainy season, between June and September. We made nine apparent resistivity maps in the year 2000
and two in January and March 2001. To monitor expected infiltration and percolation to depths of 5
m or more, we laid out Wenner array profiles with an inter-electrode spacing of 5 m. The time-lapse
mapping was also controlled with: (i) neutron probe measurements; (ii) resistivity measurements on
outcrops during infiltration tests; (iii) electrical resistivity logging in auger holes. Geophysical results
showed that the apparent resistivity parameter can either decrease (typical case) or increase
(unexpected case) after a rain. Neutron probe measurements indicated that infiltration varies within a
few decimeters even at the centre of the main gully. Using one dimensional (1D) modelling based on
resistivity variations monitored during infiltration tests, we concluded that apparent resistivity
variations are linked to the presence of carbonate in the soils. When soluble carbonates are present,
the resistivity of the infiltrated layer varies from 220 V m (dry state) to less than 5 V m (wet state),
bringing about a decrease in apparent resistivity value for the 5m spacing. In the absence of
carbonate, resistivity varies from 1500 to 180 V m, but produces an increase of the apparent
resistivity value for the same spacing. Consequently, we found time-lapse apparent resistivity
mapping to be an efficient way to delineate certain soil properties. It also provided additional
* Corresponding author.
E-mail addresses: Marc.Descloitres@ird.bf (M. Descloitres), Olivier.Ribolzi@ird.bf (O. Ribolzi).
0341-8162/03/$ - see front matter D 2003 Elsevier Science B.V. All rights reserved.
doi:10.1016/S0341-8162(03)00038-9
230
M. Descloitres et al. / Catena 53 (2003) 229–253
information about punctual observations. However, our results have led us to conclude that the 5-m
inter-electrode spacing is too large to monitor this type of shallow infiltration phenomenon and that
the effect of temperature on resistivity should be considered when comparing maps over the period
of a few months. Furthermore, this type of survey should be controlled using electrical loggings in
auger holes, or electrical soundings in order to get a better understanding of in-depth resistivity
variations. Finally, this survey indicated that deep infiltration processes are not occurring below the
gully situated on the slope. Further studies are required downstream to identify the location of
groundwater recharge in Sahelian crystalline contexts.
D 2003 Elsevier Science B.V. All rights reserved.
Keywords: Gully erosion; Infiltration; Resistivity mapping; Wenner array; Time-lapse measurements; Sahelian
zone; Burkina Faso
1. Introduction
Soil surface sealing is a common feature of most soils in arid and semi-arid regions
(Valentin and Bresson, 1992; Fedoroff and Coutry, 1999). It reduces infiltration rates,
triggers runoff, and hence increases soil erosion (Casenave and Valentin, 1992). In
northern Burkina Faso, overgrazing, extensive farming (Marchal, 1983) and increasing
climatic dryness (Albergel et al., 1984) aggravate erosion. The present study is a part of an
interdisciplinary research program focused on erosion processes occurring in this area.
Limiting our study to the scale of a small catchment located on a slope, we evaluated
spatial flow variations during the rainy season. It is particularly important to be able to
evaluate the impact of soil surface conditions and gullies on infiltration processes. Poesen
et al. (2003) present an overview on gully erosion processes and its impacts on environmental change. They note that if the gully channel develops into more permeable horizons,
it can increase infiltration. As described by Tooth (2000), dryland river floods are
generally subject to downstream volume decreases. This phenomenon is primarily due
to transmission losses resulting from the infiltration of floodwater into channel boundaries,
and over bank floodings. On the other hand, Poesen et al. (2003) also report that the gullies
can enhance the drainage of the hillslopes, and consequently dried out the soil profiles in
the intergully area, lowering water tables. Generally, on Sahelian slopes, concentrated
runoff destroys indurate and impermeable surface horizons to form gullies (Vuillaume,
1969; Mietton, 1988). Only a few other papers treat the subject of infiltration and
groundwater recharge in the Sahelian zone of West Africa. To balance the hydrological
budget of a small catchment, Peugeot et al. (1997) made a hypothesis based on their
observations in Niger: Infiltration is increased when flows are located at the bottom of
temporary gullies and streams, particularly when they cross gravelly and sandy surfaces.
Downstream pools also contribute to groundwater recharge. Those observations also
support the conclusions of the study made by Le Gal La Salle et al. (2001) in Niger: the
wide variation in the electrical conductivity and oxygen-18 content of the groundwater
indicate a heterogeneous recharge, occurring mainly through a drainage system of
temporary streams and pools. The role played by the gullies in the Sahelian part of Niger
for infiltration is also evidenced by Esteves and Lapetite (2003). At the scale of a
231
catchment (0.2 km2), they noted the reduction of runoff coefficients between two gauging
stations. They concluded that a significant infiltration is active trough the sandy soil cover
(up to 10 m thick) crossed by the gully.
Although the Sahelian zone of Burkina Faso has the same climatic conditions as in
Niger, the geological context differs. The soil cover is very clayey, due to the alteration of
the crystalline substratum that forms the majority of the basement of the country. In
comparison with Niger, thick permeable sandy soils are not widely encountered. Thus, the
primary objective of this study was to determine if gullies located on slopes promote
infiltration during the rainy season as evidenced in the sedimentary context of Niger. We
had to find answers to the following set of questions: (i) Are the gullies actually the main
zones of infiltration? (ii) What is the influence of surface conditions? (iii) Is infiltration
concentrated in some places?
Several authors have already demonstrated the utility of geophysics in soil studies. In a
study dealing with tropical soils, Lamotte et al. (1994) used electrical resistivity mapping
in order to delineate sandy horizons within the first 2 m associated with vegetation in the
northern Cameroon. Robain et al. (1996) have investigated an elementary catchment
located in the rain forest of southern Cameroon, using resistivity mapping and soundings
in order to map the bedrock topography and soil cover thickness. Lateritic weathering over
granite and metamorphic basement has been investigated in Senegal by Ritz et al. (1999),
using two-dimensional (2D) electrical imaging.
These previous studies advocate the use of electrical resistivity, as well as other
geophysical parameters. This parameter (or its inverse, conductivity) is highly dependant
on porosity, water content, the conductivity of the water, and the percentage of clayey
minerals, as well as other factors such as temperature and the conductivity of non-clayey
minerals (Telford et al., 1990, pp. 289 – 291). Electrical resistivity can vary on a wide scale
of values, ranging from less than 10 V.m for clayey saturated material to more than 1000
or even 10,000 V.m for dry sand. Previous studies have also concluded that an in situ
calibration is required to be able to link the resistivity to the natural and the hydric state of
sub-surface layers.
An increasing number of recent studies deal with the time variation of electrical
resistivity. For example, Daily and Ramirez (1995, 2000) and Slater et al. (2000) present
the successful monitoring of resistivity variations with cross-hole electrical imaging.
These three studies deal with in situ trichlorethylene remediation, engineered hydraulic
barrier testing and saline tracer injection, respectively. Yoon and Park (2001) have
investigated the sensitivity of leachate and clay contents of sandy soils using electrical
resistivity. They concluded that the variation of soil resistivity is highly influenced by
both variations in the water content and chemical composition of the pore fluid. The
resistivity parameter is found to be well suited for water infiltration monitoring.
Benderitter and Schott (1999) have investigated the short time variation of the resistivity
in an unsaturated soil and its relation to rainfall. They concluded that slight resistivity
variations (of a few percent) can be measured during rainfall. Asch and Morrison (1989)
have investigated the use of apparent resistivity measurements, using surface and subsurface electrodes. Their purpose was to demonstrate the interest of using sub-surface
(buried) electrodes when trying to locate a target hidden by surface inhomogeneity
(conductive overburden). They limited their work to the analysis of apparent resistivity
232
differences. In the study of Robain et al. (1998), electrical resistivity mapping was
employed with a time-lapse approach and Wenner profiling techniques with three
spacings (10, 20 and 40 m) over lateritic soil covers in a forest in Cameroon. The
apparent resistivity variations were analysed, using a statistical discriminant analysis
approach. This allowed them to delineate the main units of the soil cover. Furthermore,
the apparent resistivities varied in the same way as the mineralisation of the groundwater,
giving information about the dynamic of the soil units.
In our study, the main idea was to take advantage of a very long dry season and
short rainy season, which is characteristic of the Sahelian climate, to monitor spatial
and temporal electrical resistivity variations which could be linked with infiltration and
recharge processes. We opted for a mapping approach using apparent resistivity
because our purpose was to estimate the influence of heterogeneous surface conditions
on infiltration processes and groundwater recharge. Resistivity mapping is a wellknown technique, which can be used to map a large surface in a limited survey time
(in this case, 1 ha in a half-day survey). It is also an indirect and non-destructive
surface method, which allowed us to choose relevant investigation sites to implement
auger holes and pedological pits. Moreover, it allowed us to extend local information
laterally.
The primary aim of this paper is to present the results obtained by mapping the spatial
and temporal variations of apparent resistivity during the rainy season of 2000 in the area
of the main gully, and to explain the practical consequences deduced from this example.
2. Material and methods
2.1. Site description
The experimental site is located in northern Burkina Faso, 13 km to the west of the
town of Dori (14jN, 0jW, Fig. 1), near the village of Katchari. It is a small, degraded 82ha catchment crossed and overgrazed by livestock. The climate is Sahelian, with only one
rainy season (June to September). The average annual rainfall recorded at Dori is 512 mm,
with a maximum of 181 mm in August (Casenave, 1998). There is wide year-to-year
rainfall variability (244 mm minimum, 784 mm maximum). Mean annual potential
evapotranspiration is about 2396 mm. The site soils are fundamentally haplic solonetz
(Boulet, 1968; FAO – UNESCO, 1989) developed from granitic and amphibolitic rocks.
No water table was found at less than 40 m deep in this zone.
The (1 ha) study area is affected by a typical gully erosion (Fig. 1). It is located at the
outlet of the catchment. We selected this area because the main soil surface features of a
catchment are present and because some of them are more permeable horizons. Six surface
features were identified according to the classification of Casenave and Valentin (1992): (1)
erosion crust surfaces low permeable and bare, (2) structural surfaces also low permeable,
Fig. 1. Location of the study area and map of the soil surface features around the main gully of the studied
catchment. The locations of pits, auger holes and the data points of the Wenner profiles are indicated.
233
234
(3) drying surfaces corresponding to sandy aeolian deposits, (4) permeable sandy runoff
surfaces inside the main gullies, (5) pavement surfaces mainly composed of quartz and
ferruginous gravels, and (6) the outcrop of a highly fractured quartz vein inside the main
gully.
2.2. Direct current (DC) resistivity measurements
2.2.1. Wenner mapping
The objective was to map the area with a profiling technique, using as simple an array as
possible to avoid the use of infinite remote electrodes. Consequently, we chose the Wenner
array, a quadripole made with two current electrodes A and B and two potential electrodes
M and N, using a regular inter-electrode spacing ‘‘a’’ in the order (A, M, N, B). A 1D
Wenner electrical sounding at the bottom of the gully was made before the time-lapse
mapping. The centre of this sounding is situated at the coordinates (42.5, 32.5) in Fig. 1.
The directions of the electrical lines were chosen perpendicular and parallel to the gully.
The resulting two curves were then interpreted in terms of 1D variations using a curvematching technique. The resulting models were used as initial models to evaluate a 20%
resistivity decrease within the first 5 m, the expected phenomena when infiltration occurs.
Both sounding curves exhibited a 30– 50% decrease of apparent resistivity between dry and
wet states for inter-electrode spacings, between 1 and more than 10 m. The spacing ‘‘a’’ of 5
m was selected for the mapping, representing a compromise between a relatively shallow
depth of investigation, estimated here at 2.5 m according to Loke (2000), and a reasonable
spatial averaging to map the area (90 70 m) following a square grid of 5 5 m (266
measurements) within a half-day survey. Moreover, surface conditions had to remain as
undisturbed as possible, which excluded the intensive use of a smaller array. Furthermore, a
Wenner array with a = 1 m would have been extremely time consuming (more than 6000
measurements in a regular grid of 1 1 m) as well as quite ‘‘destructive’’ to surface
conditions. Nevertheless, even though the array (a = 5 m) was simple and fast to lay out, it
does not allow us to make a quantitative interpretation of our results, as other authors have
proposed with three dipole separations (Robain et al., 2001).
Between June 8, 2000 and March 15, 2001, 11 maps of apparent resistivity were made
in the area, using a Syscal R2 resistivity meter. The results are presented in apparent
resistivity (in V m), which is defined as the product of the resistance (R) calculated using
the ratio voltage/intensity given by the quadripole AMNB (in V) by a geometric factor, K
(in m). K = 31.4 m for a Wenner array with an electrode separation of 5 m. The apparent
resistivity is equal to the resistivity if the medium is a homogeneous half space. But it is
important to note that apparent resistivity is not equal to the resistivity of the ground at the
investigation depth, when the ground is horizontally layered (1D case) or more complicated (2D and 3D cases). To recover the resistivity value at depth, direct or inverse
modelling is necessary, but this can only be done if more than one electrode separation is
used (i.e., soundings).
2.2.2. DC measurements inside auger holes and on outcrops
A resistivity logging was done in the two auger holes (GEOP 1 and GEOP 2) at 5 m
deep located along a profile crossing the gully (Fig. 1). For those measurements, the pole –
235
pole array (two buried electrodes A and M and two remote electrodes B and N) was used,
placing the remote electrodes more than 150 m away, thus avoiding any geometric
distortion (Robain et al., 1999). A sketch of the device is presented in Fig. 2. In the holes,
the electrodes were buried every 0.5 m. These measurements allow a rapid and reliable
indication of resistivity variations at depth. The results are in apparent resistivity, with
K = 6.28 m, but, as the inter-electrode spacing remains short, the values are close to the
resistivity.
Some measurements were made after the time-lapse mapping on small outcrops near by
and in the pits using nail-electrodes with Wenner array and an inter-electrode spacing of
0.05 m as shown in Fig. 2. The purpose of these measurements was to evaluate resistivity
variations over time under induced infiltration and natural desiccation processes. They
were made inside 0.25-m-diameter PVC rings, where infiltration tests were performed
using a strip of 0.03 m thick of demineralised water, which corresponds to a mean rain.
Again in this case, the apparent resistivity obtained with such a small array can be
considered to be the actual resistivity, assuming that the medium remains homogeneous.
The variation of resistivity was monitored during the infiltration phase (20 – 30 min) and
for up to 1 week following, until the initial value of the resistivity was recovered.
2.3. Neutron probe measurements
Neutron measurements were made inside the six auger holes (TN 22– 27) with PVC
casing from the surface down to 4– 5 m deep as shown in Fig. 2. The holes are located
along a profile crossing the gully (Fig. 1). Care was taken to prevent any preferential
infiltration. The use of PVC instead of metal was necessary to prevent any distortion
during geophysical measurements due to metallic conductors near the electrodes. To verify
that PVC does not significantly alter the measurements, we did a preliminary measurement
without any tube. The measurements were made down each tube with a 20-cm interval
Fig. 2. Set-up of electrical measurements on outcrops and in auger holes. (A) Infiltration tests and small Wenner
measurements. The tests were done after the time-lapse mapping near the pits and on their steps, for example at
locations 1, 2 and 3. (B) Auger hole with buried electrodes for time-lapse pole – pole measurements. Couples of
(A, M) electrodes were connected to the resistivity-meter successively. B and N infinite electrodes were placed
150 m away (not shown). (C) Neutron probe measurements in a PVC encased auger hole.
236
Fig. 3. Massic water content from neutron measurements before the rainy season (May 2000) and granulometry
profile in the auger hole TN22.
after every rainfall to assess if any infiltration occurred. Calibrations with water content
were made, using two series of dry/humid weight measurements on soil samples correlated
with the number of counts before and after the rainy season. Fig. 3 presents a
representative example of the granulometry and neutron profiles for auger hole TN22 in
its initial state in May 2000. The typical soil profile shows a sandy to clayey horizon from
0 to 20 cm, a gravelly horizon from 20 to 60 cm (from where infiltration was expected to
be favoured), a clayey layer from 60 to 1.6 m, a sandy silt from 1.6 to 3.7 m, and a sandy
layer after 3.7 m. The water content is low, between 2.5% and 7.5%.
3. Results
3.1. Apparent resistivity versus soil surface conditions
Fig. 4 presents the results of the first map made in June 8, 2000. The contours of soil
surface features have been superimposed on the same map. The apparent resistivity ranges
from 20 to 130 V m. There is a good agreement between lower apparent resistivity zones
and the water flow paths, which are generally sandy. The gully banks, generally more
clayey surfaces, are more resistive. There are two notable exceptions to this pattern. The
237
Fig. 4. Apparent resistivity map of June 8, 2000, drawn from Wenner profiles with inter-electrode spacing of 5 m.
The data are dotted. Contours of soil surface features (Fig. 1) are superimposed using continuous black lines.
first is located at the coordinates (75,25), where the resistive zone crosses the gully. The
area has an outcrop of the gravelly horizon. The second exception is located at the
coordinates (30,20), where a conductive zone is found within a clayey bank. The analysis of
the map shows that apparent resistivity is not in a clear correlation with surface conditions.
Sandy soils would have been detected as resistive rather than conductive. One possible
explanation is that the sand layer is very shallow (10 –20 cm), concealing clayey layers
below. Clayey surface conditions would have been detected as conductive rather than
resistive. But below this horizon, a gravel horizon (20 – 60 cm thick, more resistive) has
been noted below a few decimeters, when digging neutron probe access holes, and by direct
surface observations. This layer outcrops in some places within the gully, for example at the
coordinates (65,45) near the fractured quartz vein. These observations indicate that the
array is more sensitive to deeper layers (range 0.5– 2 m). Thus, the best explanation for the
general pattern deduced from the apparent resistivity map (i.e., lower apparent resistivity
values inside the gully, but higher ones on the banks of the gully) can be given as follows:
The gully (25 – 40 cm deeper than its banks) has eroded the gravel horizon in most places
and therefore only the deeper clayey material remains below the sandy cover at the gully
bottom. In general, the presence of the more resistive gravel horizon and its thickness
control the value of the apparent resistivity, even increasing it. At this first stage of the
work, despite some exceptions, the general agreement between isocontours of apparent
resistivity and water flow path was evidenced. The apparent resistivity arrangement can be
explained, when considering the erosion processes that modify the arrangement of subsurface layers. However, we should note that the apparent resistivity using a Wenner array,
with a = 5 m and a measurement each 5 m, does not make fine distinctions among detailed
238
surface conditions. In fact, the apparent resistivity map is more representative of the first
meters of the sub-surface and cannot be considered to be a surface condition map.
3.2. Monitoring apparent resistivity variations
Eleven apparent resistivity maps were produced at about a 15-day sampling interval
during the rainy season. Each of them is quite identical to the others, if we consider the
geometry of isocontours as well as the apparent resistivity ranges. Therefore, they are not
presented in this paper. To compare these maps, we calculated the ratio between one map
over the first one. Fig. 5 presents the results obtained when monitoring the area between
the initial state (June 8), which corresponds to the drier state before the first rainfall of the
year 2000, and the middle of the dry season in 2001 (March 15). The values of the
variations have been limited to values above F 5%, to focus on significant phenomena. In
Fig. 5. Time-lapse apparent resistivity ratios (current date/initial state, June 8, 2000) for four selected dates.
Simplified contours of soil surface features are indicated using continuous black lines. The location of the pits and
auger holes are also drawn. The cumulative rain for the rainy season of 2000 is also shown with the dates of
measurements.
239
the same figure, we have drawn the curve of cumulative rain, which reached a total of 420
mm for year 2000. Five dates are selected to illustrate the focus of this work.
3.2.1. June 21 versus June 8
This map represents apparent resistivity variations 2 days after the first rain, which
occurred on June 19 with an exceptional amplitude (80 mm). The variations are noticeable
and range between
20% and + 20%. Half the surface does not show any variations
above F 5%. If apparent resistivity may be considered here as the resistivity of the ground
(this case occurs only if we consider a homogeneous half space), a decrease of the
resistivity when the rainwater penetrates the dry soil becomes a classical situation. The
majority of the decreases are located on the left side of the zone. A resistivity increase is
abnormal, if the temperature remains constant. The majority of the increases are located on
the right side. None of the decrease/increase zones fit the limit of the gully, neither do they
fit any particular surface conditions.
3.2.2. August 2 versus June 8
This map corresponds to variations measured in the middle of the rainy season (when
half the total rain had fallen). Again, decreases and increases of apparent resistivity are
seen, while more than half the surface exhibit variations of less than F 5%.
3.2.3. September 27 versus June 8
This map represents the variations obtained at the end of the rainy season. Slight
differences are noted, for example the persistence of a decrease zone on the left side of the
map. Only a few increase zones and a majority of invariant zones (ranging F 5%) are
noted.
3.2.4. March 15, 2001 versus June 8, 2000
This map is representative of the differences between two dry states. Almost the entire
surface exhibits a general apparent resistivity increase. Temperature variation aside, if we
consider also the apparent resistivity to indicate the actual resistivity of the media, this
finding could be interpreted as the result of a global desiccation of the soil.
The preliminary analysis of apparent resistivity variations obtained during the rainy season
with Wenner profiling (with a = 5 m) has shown: (i) the areas where the apparent resistivity
increases (to more than 20%); (ii) the areas where the apparent resistivity decreases (to less
than 20%); and (iii) the areas where it remains in the range of F 5% (the majority of the
surface).
3.3. Evaluation of infiltration using neutron probe measurements
Fig. 6 shows the results of infiltration monitoring during the rainy season, using the
data of the neutron probe in the three profile tubes crossing the gully. These measurements
were made after each significant rain (i.e., >15 mm). The massic water content of the soil
is plotted each 20 cm only for the first meter, because the values remain identical below.
Only the dry and the final states are presented here, as all intermediate results remain
240
Fig. 6. Soil massic water content measured using a neutron probe at auger holes 22, 23 and 26 for the initial state
(June 8) and the end of the rainy season (September 27).
between those values. For tube TN 26, situated outside the major gully flow, the massic
water content increases from 1.5% to less than 4% and above 60 cm deep. Tube TN 22
does not change significantly, remaining around 3% (the same results is obtained for tubes
TN 25 and TN 27, not shown here). In the contrary, for tube TN 23, situated inside the
flow area, we have monitored an increase of the water content from 3% to 9% within the
first 80 cm (the same result is obtained for tube TN 24, not shown here). This result clearly
indicates that apart from the first decimeters, no major infiltration process (i.e., phenomena
which lead to several meters of moistening and groundwater recharge) takes place below
the gully.
These results are in agreement with the low infiltrability values obtained by Casenave
and Valentin (1992) in other parts of the Sahel. Measurements in this gully, situated on a
slope over a crystalline basement, yielded results that contradict the observations of
241
Peugeot (1997) or Esteves and Lapetite (2003) in Niger for other gullies also situated on
slopes, but over sedimentary basements.
3.4. Pedological data
We dug two pedological pits A and B (2 m deep) in the area further downstream on one
bank of the gully, with the same surface conditions and a clayey erosion crust. They are
located in two different zones identified on the map of the apparent resistivity ratio of
August 2 to June 8. Pit A is inside a zone showing a decrease of apparent resistivity, while
pit B is inside a zone of increasing apparent resistivity. The granulometry, pH and the
electrical conductivity (EC) of the soils (measured from a massic water/soil ratio of 2.5)
are presented, versus depth in Fig. 7. From the granulometry analysis, the two profiles are
quite similar. On the other hand, the curves of the pH and EC are rather different. For pit
A, where the apparent resistivity decreases, the pH remains above 7 with an EC above 200
AS/cm (0.02 S/m). For pit B, where the resistivity increases, the pH varies between 5 and
6, the EC decreases from 100 AS/cm (0.01 S/m) at the surface down to 30 –40 AS/cm
(0.003 –0.004 S/m) at depth. These differences can be explained by the presence of
secondary carbonates inside the soils for pit A.
Fig. 7. Electrical conductivity (saturated paste), pH and granulometry measurements of the soil versus depth in the
two pedological pits, A and B.
242
These carbonates primarily take the form of micrite of calcite, and of thin white coatings
around ferruginous gravel. No carbonate minerals were found in the soil of pit B. This
lateral heterogeneity is no doubt linked to differences in bedrock composition. The granitic
rocks include some dykes or thin lodes of basic rock (mainly amphibolite) that release an
significant amount of divalent cations under alteration processes, leading to secondary
carbonate concentration. An alternate explanation could be provided by the evolution of the
soils from solonetz to solod types. This could account for the superficial acidification
observed at pit B and the low values of pH ( < 5) observed in other places outside the area.
Considering the above observations, our preliminary conclusion is that zones where
apparent resistivity increases are related to lower pH and EC zones (no carbonates); while
higher pH and EC zones (carbonates) are related to zones where the apparent resistivity
decreases. The index could be rather satisfactory if the relationship is observed elsewhere
in the area. Fig. 8 presents the values of the pH and the soil conductivity, using soil
samples from auger holes located on the profile across the gully. The general trend (low
pH, low EC – high pH, high EC) is roughly respected. However, this set of data shows a
notable dispersion. In the case of high pH samples, the dispersion of the EC from 200 to
400 AS/cm (0.02 – 0.04 S/m) could be due to a variable—a limited amount of carbonated
minerals in one sample as opposed to another. In the case of low pH samples, the
dissolution of other soluble mineral traces can modify the EC without increasing the pH.
These results have led us to the preliminary conclusion that the presence of carbonates
plays an important role in explaining the geophysical results. When it rains, the water that
infiltrates the soil quickly dissolves these carbonates. Consequently, water conductivity is
greater in zones where carbonates are present than in zones where they are absent. This
Fig. 8. Relationship between mean pH and EC calculated from five soil samples collected between 0 and 1 m
deep inside eight auger holes and pits A and B.
243
process is rapid because the calcite crystals are probably poorly crystallised and very small
(Bouzigues et al., 1997). As a result, soil resistivity may drastically decrease. To assess this
assumption, some tests on outcrops inside the pits were carried out. They are presented
below.
3.5. DC resistivity variations during simulated infiltration on outcrops
On the surface and on the steps of pits A and B, we monitored resistivity variations
during a simulated infiltration, using a small Wenner array (a = 0.05 m). As previously
mentioned, this spacing is small enough to enable us make the assumption that the
apparent resistivity is equal to the resistivity because at this scale the medium is considered
to be homogeneous. The results are presented in Fig. 9 for the shallower horizon. Other
experiments on the surface of the gravel horizon showed the same tendency. In a dry state,
Fig. 9. Comparison between variations in resistivity monitored during simulated infiltration tests on clayey
superficial horizon of pits A and B. The measurements of the resistivity were done using a small Wenner array
with an inter-electrode spacing of 5 cm.
244
before infiltration, the initial value for pit A was 220 V m. However, in pit B it was not
possible to make good contact with the soil for a measurement. One minute after the
introduction of the water into the PVC ring, both resistivities began to decrease. For pit A,
with carbonates, the resistivity decreases noticeably from 110 V m to less than 6 V m
within 30 min. The infiltration of the 0.03 m water strip was taken for 33 min. For pit B,
the same tendency is noted, but it starts at 400 V m and decreases down to 180 V m within
10 min. The infiltration was taken for 20 min. At the end of the experiment, 7 days after,
the resistivity level had reached values of 100 V m and more than 1500 V m for pits A and
B, respectively. Some oscillations of the curves are due to temperature variations because
the measurements were not taken regularly at the same time each the day.
These results confirm that in the infiltration process, resistivity is always decreasing.
These results have been confirmed by seven other experiments on various horizons, not
presented here. Therefore, the increases of the apparent resistivity deduced from the maps
are in complete contradiction with the observed resistivity decrease. The following
questions are raised regarding the geophysical data: Is the Wenner array with a = 5 m
able to accurately detect shallow infiltration? Is the increase of apparent resistivity a direct
consequence of shallow infiltration? Could the temperature have influenced the results?
4. Interpretation and discussion
4.1. Temperature influence
During a rainfall, cold rain penetrating hot soil may increase its resistivity. This point
must be considered because of the wide daily temperature variations of the Sahelian
climate. We have monitored surface and sub-surface temperatures (down to 40 cm) as well
as apparent resistivity variations for three Wenner arrays with inter-electrode spacing
a = 0.5, 1 and 5 m. This was done during 1 day in the dry season, when temperature
variations are maximum (14 jC in the night, 37 jC in the day) over some of the areas
where increase of apparent resistivity had been detected. The results (not shown here)
indicate that apparent resistivity with a = 0.5 m increases from about 4% to 5%, when the
surface temperature is at its lowest value in the night. For a = 1m, the increase remains less
than 1%. This value is lower than 1% for the spacing a = 5 m used for the time-lapse
survey. The conclusion is that even if a cold rain (25 jC) falls on a hot soil (35 jC), it is
highly improbable that this phenomenon could lead to an increase of apparent resistivity of
more than 1% for a large array.
Giving an overview of the period of few months, the map of the apparent resistivity ratio
between March 15, 2001 and June 8, 2000 (shown in Fig. 5) exhibits a general increase in
apparent resistivity. Is this result the consequence of the soil drying up, or could it be
attributed to annual temperature variations inside the soil? The monitoring of resistivity
every 50 cm down to 5 m inside the auger holes GEOP 1 and 2 in the centre of the profile is
shown in Fig. 10. At the depth of 2 m, for example, resistivity has increased between
September 27, 2000 and March 15, 2001, when the deeper section of the clayey soil has
received, with a phase shift, the cold wave issued from the coldest months (December–
January, mean air temperature 24 jC). Resistivity decreased afterwards ( 9%) in June
245
Fig. 10. Variation of resistivity with depth measured with a pole – pole array inside auger holes GEOP 1 and 2 for
3 selected dates: end of the rainy season (September 2000), middle of the dry season (March 2001), and end of the
dry season (June 2001).
2001 without any rain, when the soils received the heat wave from the hottest months
(April – May, mean air temperature 36 jC). Moreover, the neutron probe measurements
have not shown any variation between those dates. Knowing that a 2% resistivity variation
corresponds to a 1 jC variation (Keller and Frischknecht, 1966), a 9% resistivity variation
can correspond to a temperature variation of about 4 jC at 2 m deep. This is a rather high
variation at this depth, but remains plausible variation in dry clayey materials. So, it is
demonstrated that temperature variation cannot be neglected when we consider the maps
over the period of a few months. The difference between the March 15, 2001 and June 8,
2000 map readings is likely to be the result of the temperature decrease at depth.
4.2. 1D synthetic modelling of DC resistivity variations
A synthetic 1D resistivity model was built in order to calculate a series of Wenner
sounding curves for shallow infiltration phenomenon. For this model, we made the
following assumptions deduced from the neutron measurements and infiltration tests:
Infiltration remains within the initial decimeters of the soil. We chose the value of 35
cm depth (mean value deduced from neutron probe measurements);
246
In the absence of carbonates, the first layer presents a resistivity variation from 1500
V m (dry state) down to 180 V m (wet state). In the presence of carbonates, the same
layer presents a decrease in resistivity from 220 V m (dry state) down to 5 V.m (wet
state).
This allowed us to model the influence of resistivity variation within the infiltrated
layer on the apparent resistivity value for the Wenner spacing a = 5 m. Fig. 11 presents the
Fig. 11. Electrical sounding curves (Wenner) calculated from a 1D model for cases 1 (no carbonates) and 2 (with
carbonates) as described in the text. Apparent resistivity is plotted versus the inter-electrode spacing a. The
variation in apparent resistivity is indicated for the particular inter-electrode spacing of 5 m used in this study for
drawing the maps.
247
results of this modelling. The first two models are calculated for dry soil, with the
following 1D geometry:
The first layer, clayey, 55 cm thick with a resistivity of 1500 V m (case 1, no
carbonates) or 220 V.m (case 2, carbonates),
The second layer, gravel, 300 V m, is 50 cm thick,
The third layer, clayey, 80 V m, is 15 m thick,
The final layer, 3000 V m, is the altered bedrock.
The second two models, for after the rain, were calculated assuming that only the first
35 cm of the first layer is wet, with resistivity decreasing to 180 or 5 V.m (cases 1 and 2,
respectively). The other deeper horizons remained invariable.
In case 1, the curve clearly exhibits a decrease of apparent resistivity before the spacing
a = 1.5 m, an increase between 1.5 and 10 m, and finally joins the previous curve for dry
state. This proves that an increase of apparent resistivity can be produced for a spacing of 5
m by decreasing a shallow horizon’s resistivity under special arrangement (i.e., combination of thickness, resistivity and depth of infiltration).
For case 2, representative of the soil with carbonates, the curve exhibits a general
decrease of apparent resistivity.
Despite the possibility that a 1D layered ground may be questionable, this model is a
simple yet demonstrative example of the ‘‘shift to the right’’ of the electrical sounding
curves for some particular geometry. This case is commonly encountered when carrying
out Wenner soundings when overburden thickness is varying. This phenomenon is rather
troublesome because it can create an increase in the measured parameter, when in fact, the
resistivity may decrease only in the first decimeters, while the other layers remain
unchanged.
4.3. Wenner resistivity mapping
The Wenner array was chosen because pole – pole or pole – dipole arrays are less
practical to lie out, the operator having to use remote electrodes (‘‘infinite’’ electrodes)
even if they have a higher investigation depth. Similarly, the Schlumberger or dipole –
dipole arrays have been avoided, because it is sometimes more difficult to get a good
signal/noise ratio with them. But the major problem for all these arrays, including the
Wenner, remains their sensitivity to shallow resistivity variations. For all the arrays cited
above, we have calculated the resulting sounding curves for cases 1 and 2. For all of them,
the value of apparent resistivity for a medium inter-electrode spacing (a = 5 m, for
example) is influenced in the same way as with the Wenner array. It is clear that only a
small inter-electrode spacing (a < 0.5 m) is able to restore the phenomena. Using spacing
of 0.5 m or less and a measurement square grid was impracticable, because we wanted to
preserve natural surface conditions.
The Wenner array presents one small disadvantage compared to pole –pole array. It is
more sensitive to lateral resistivity variations (2D near surface effects). In Fig. 12, we
present the result of a calculation of a 2D structure, a resistive superficial patch in a layered
half space, when making a profile with three arrays (Wenner, pole – pole and dipole –
248
Fig. 12. Theoretical electrical profiles for three different arrays (Wenner, pole – pole and dipole – dipole) calculated
from a 2D model (near-surface resistive block included in a layered medium) for an inter-electrode spacing of 5 m.
dipole). The responses show some ‘‘à-coup de prise’’, well known in geophysical
prospecting (see, for example, Kunetz, 1966, pp. 43– 49). This is sometimes troublesome,
and explains why we have some ‘‘eye-like’’ anomalies in the maps. All arrays are sensitive
to shallow resistivity variations. These variations, even in the initial centimeters, have to
be estimated for a correct understanding of the results. However, if deep infiltration
processes (i.e., more than 1 –2 m) are encountered, reliable results can be obtained using
arrays with large inter-electrode spacings.
Frequency Electromagnetic (EM) surveys could be an alternative that would overcome
(i) sensitivity to first layer resistivity changes and (ii) major eye-like features. However,
these techniques are difficult to use if the ground resistivity is over 200 – 300 V m (Mc
Neill, 1980, pp. 10– 11), which is sometimes the case for the dry horizons of the area.
Some tests, which were made in the area using the well-known EM 38 equipment, have
demonstrated that instrumental dispersion could be as large as 5 –10% even if a careful
calibration is made many times a day at the same place. This instrumental dispersion is
249
attributed to electronic drift due to high temperature variations encountered during the day
and this is obviously a problem for a time-lapse survey. However, further tests have to be
made to avoid this problem. For example, the instrument could be protected against the
wide variations of outside temperature. So, we do believe that time-lapse EM mapping
could be an alternative to direct current methods when resistivity remains low and when
conditions can be more carefully controlled.
Using electrostatic equipment, we also made electrical measurements. This equipment
used 0.5 m2 metal sheet electrodes isolated from the ground by a rubber film. The
electrodes have to be dragged along the ground to get a good signal to noise ratio,
destroying the surface conditions in the process. Because of this, the electrodes were not
suitable for our case. However, we believe that the electrostatic method could be used in
other cases, using, for example, cylinder electrodes.
4.4. Choice of time reference
The results of time-lapse resistivity imaging are typically presented in reference to the
initial state. This approach has been used by us in this paper and by Barker and Moore
(1998), among others. However, we have observed that the annual temperature variation at
depth can influence mapping results. In this case, temperature monitoring is also required.
For shallow surveys, care should be taken with regard to daily temperature variations.
One way to overcome this problem, and to also get different information, is to use a
sequential representation, i.e. current date over preceding date. This is illustrated in this
paper by the first map, which shows conditions just after the first rain. This way, it is
possible to monitor the dynamic of a rainfall, as well as the return to the dry state if no rain
occurs.
A third approach has been tried by other authors (Robain et al., 1998) who are
monitoring the apparent resistivity of the soils of a catchment in the rain forest of
Cameroon. For each location, they took the median of the values obtained for all the
readings and then drew the ratio between each map over the median value. This method is
effective for smoothing out results and, importantly, it gives a lesser weight to errors.
Moreover, in most parts of Cameroon, there is no long dry season, thus the ‘‘initial’’ state
is more difficult to determine. This third way was not considered for use in our study, as
we saw that we would lose the advantage of having a long dry season that fixes the initial
state as a reference. Furthermore, we believe that in the present study, using a median
value approach would have probably masked or smoothed out the temperature effect noted
between the two dry seasons.
5. Conclusion
In the gully area, the use of time-lapse apparent resistivity mapping using a Wenner
array with an inter-electrode spacing of 5 m has shown that information about the physicochemical properties of soils can be deduced. Wenner time-lapse mapping alone can be
difficult to interpret if a soil analysis is not done after the experiment. The geophysical
anomalies can then be compared to the physico-chemical properties of the soils. However,
250
soil sampling is easier to locate, considering the geophysical results. For our study, the
results can be summarized as follows:
A decrease of apparent resistivity was detected in some zones, where carbonates may
be present and greatly lower the resistivity of the infiltrated horizon when rainfall
occurs.
An increase of apparent resistivity was also been detected, but not due to any increase
of resistivity. The 1D modelling demonstrated that this phenomenon could be the result
of a ‘‘shift to the right’’ of the sounding curve. In this case, the resistivity of the
infiltrated horizon does not decrease enough to produce a decrease of the apparent
resistivity curve, as is seen when carbonates are present, but on the contrary, increases
the apparent resistivity.
Temperature variation can influence the results. This was monitored in this study, where
the array was sensitive to annual temperature variations ( < 4 jC) occurring at a depth of
2 m between the 2 dry seasons of 2000 and 2001.
Some practical conclusions can be drawn from the results.
From a methodological point of view, it is obvious that the Wenner array (or other
arrays) with an inter-electrode spacing of 5 m is not well adapted to measure very
shallow infiltration processes, where deeper infiltration were initially expected. Such an
array can show unexpected apparent resistivity variations (i.e., increases after rain),
which complicate the qualitative interpretation. One of the ways to overcome this
problem would be to reduce inter-electrode spacing (for example 0.5 m or even less), but
this would have meant much more work, as well as a degradation of surface conditions.
In future studies, in cases where infiltration is shallow and the soil surface condition is
less fragile, smaller arrays (and a multi-spacing survey) should be employed to obtain
more reliable qualitative interpretations from time lapse-apparent resistivity mapping
surveys.
The temperature must be recorded at depth, in order to estimate its influence,
particularly when resistivity variations are expected to be within the range of F 20%. If
an outside air temperature is recorded near a site, one can estimate the temperature
variations at depth, given the thermic capacity and conductivity of the soil. As the
influence of temperature on resistivity decreases with depth, but could still exist down
to 2– 3 m in clayey dry soils, care should be taken when interpreting resistivity
variations separated by several months.
This type of time-lapse mapping survey must be controlled in a few places by electrical
logging in auger holes or electrical soundings, which give the actual or the apparent
resistivity variations versus depth, respectively. We did so in our study with 2 auger
holes, which allowed us to better understand the results. This can be a definite
advantage when interpreting the data.
Resistivity measurements during infiltration tests on outcrops allow us to know the range
of resistivity variation. This supports the quantitative modelling of the data. If
representative outcrops can be identified at the beginning of the experiment, tests must
be done before the time-lapse mapping is carried out. This can aid greatly in choosing an
251
appropriate ‘‘a’’ spacing. It was difficult in our case, as the surface conditions of the
carbonated zones were no different from those of other zones.
These practical conclusions can be applied when making 2D measurements, i.e.
electrical tomography.
Finally, we conclude that the deep infiltration (greater than 2– 3 m) expected in this gully
area did not occur during the rainy season. Only shallow infiltration was seen in the first
decimeters down to 80 cm maximum. Even though this study was limited to the main gully,
we should note that this conclusion could be extended to all other minor gullies of the
catchment, which exhibit the same surface conditions. Our results indicate that, in the
context of this study (a semi-arid zone and a clayey soil cover over a crystalline basement),
gullies situated on the slopes are not the primary location of the deep infiltration processes
that recharge aquifers. Therefore, the location of this infiltration has still to be determined.
Further studies should be carried out in the gullies situated downstream, where local ponds
are filled during the rainy season, and a shallower water table (10 – 20 m deep) is present.
Acknowledgements
This work was funded and conducted by the Unités de Recherche 027 «GEOVAST»
and 049 «ECU» of the Institut de Recherche pour le Développement (IRD), and the INSU
Programme National Sol Erosion (PNSE) project no. 99/44.
We would like to thank William E. Kelly, J. Poesen and one anonymous reviewer for
their constructive comments on the manuscript, the Institut National de l’Environnement et
de la Recherche Agricole (INERA) of Burkina Faso for providing access to the site as well
as the team of the Laboratoire de Géophysique Appliquée of Paris 6 University for their
active participation in this study. We also thank the team of the hydrological laboratory of
IRD at Ouagadougou, with a special mention of Maxime Wubda, Yves Dzouali and
Boureima Tou for their help in the field.
References
Albergel, J., Carbonel, J.P, Grouzis, M., 1984. Péjoration climatique au Burkina Faso: Incidences sur les ressources en eau et les productions végétales. Cah. ORSTOM, Ser. Hydrol. XXI (1), 3 – 19.
Asch, T.H., Morrison, H.F., 1989. Mapping and monitoring electrical resistivity with surface and sub-surface
electrode arrays. Geophysics 54, 235 – 244.
Barker, R., Moore, J., 1998. The application of time-lapse electrical tomography in groundwater studies. Lead.
Edge 17 (10), 1454 – 1458.
Benderitter, Y., Schott, J.J., 1999. Short time variation of the resistivity in an unsaturated soil: the relationship
with rainfall. Eur. J. Environ. Eng. Geophys. 4, 37 – 49.
Boulet, R., 1968. Carte Pédologique de Reconnaissance de la Rép. de Haute-Volta (Centre Nord) (1/500 000).
Bouzigues, R., Ribolzi, O., Favrot, J.C., Vallès, V., 1997. Carbonate redistribution and hydro-geochemical processes in two calcareous soils with groundwater in a Mediterranean environment. Eur. J. Soil Sci. 48, 201 – 211.
Casenave, A., 1998. Dynamique des milieux de savane sèche au Burkina Faso: dégradation des systèmes
écologiques et conséquences sur le cycle de l’eau. Rapport Préliminaire: Analyse de la Pluviométrie, Documents Internes de l’IRD. IRD, Ouagadougou, Burkina Faso. 49 pp.
252
Casenave, A., Valentin, C., 1992. A runoff capability classification system based on surface features criteria in
semi-arid areas of West Africa. J. Hydrol. 130, 231 – 249.
Daily, W., Ramirez, A.L., 1995. Electrical resistance tomography during in-situ trichloethylene remediation at the
Savannah River Site. J. Appl. Geophys. 33, 239 – 249.
Daily, W., Ramirez, A.L., 2000. Electrical imaging of engineered hydraulic barriers. Geophysics 65 (1), 83 – 94.
Esteves, M., Lapetite, J.M., 2003. A multi-scale approach of runoff generation in a Sahelian gully catchment: a
case study in Niger. Catena 50, 255 – 271.
FAO – UNESCO, 1989. Soil Map of the World at 1/5.000.000, 2nd ed. Food and Agriculture Organization,
Rome.
Fedoroff, N., Coutry, M.A., 1999. Soil and soil forming processes under increasing aridity. In: Singhvi, A.K.,
Derbyshire, E. (Eds.), Paleoenvironmental Reconstitution in Arid Lands. Oxford and IBH Publishing PVT,
New Delhi, India, pp. 78 – 108.
Keller, G.V., Frischknecht, F.C., 1966. Electrical methods in geophysical prospecting. International Series of
Monographs in Electromagnetic Waves, vol. 10. Pergamon Press, Oxford, UK.
Kunetz, G., 1966. Principles of direct current resistivity prospecting. Geopublication Associates. Geoexplor.
Monogr., Ser. 1, vol. 1. Gebrüder Borntraeger, Berlin.
Lamotte, M., Bruand, A., Dabas, M., Donfack, P., Gabalda, G., Hesse, A., Humbel, F.X., Robain, H., 1994.
Distribution d’un horizon à forte cohésion au sein d’une couverture de sol aride du Nord Cameroun. Apport
d’une prospection électrique. C. R. Acad. Sci. Paris, 318 (II), 961 – 968.
Le Gal La Salle, C., Marlin, C., Leduc, C., Taupin, J.D., Massault, M., Favreau, G., 2001. Renewal rate
estimation of groundwater based on radioactive tracers (3H, 14C) in an unconfined aquifer in a semi-arid
area, Iullemeden Basin, Niger. J. Hydrol. 254, 145 – 156.
Loke, M.H., 2000. Electrical imaging surveys for environmental and engineering studies. Technical note of
Geotomo Software. Available on line at www.geoelectrical.com.
Marchal, J.Y., 1983. Yatênga. Nord Haute Volta. La dynamique d’un espace rural soudano – sahélien. Trav. Doc.
ORSTOM 167, 872 (ORSTOM, Paris, France).
Mc Neill, D.J., 1980. Electromagnetic terrain conductivity measurement at low induction numbers. Technical
note TN 6 from Geonics Limited. 1745 Meyerside Drive, Unit 8, Mississauga, Ontario, Canada.
Mietton, M., 1988. Dynamique de l’interface lithosphere – atmosphere au Burkina Faso: l’érosion en zone de
savane. Thèse d’Etat, Université Joseph Fourrier, Grenoble. 738 pp.
Peugeot, C., Esteves, M., Galle, S., Rajot, J.L., Vandervaere, J.P., 1997. Runoff generation processes:
results and analysis of field data collected at the East Central Supersite of the HAPEX-Sahel experiment.
J. Hydrol. 188 – 189, 179 – 202.
Poesen, J., Nachtergaele, J., Verstraeten, G., Valentin, C., 2003. Gully erosion and environmental change:
importance and research needs. Catena 50, 91 – 133.
Ritz, M., Parisot, J.C., Diouf, S., Beauvais, A., Dione, F., Niang, M., 1999. Electrical imaging of lateritic
weathering mantles over granitic and metamorphic basement of eastern Senegal, West Africa. J. Appl. Geophys. 41, 335 – 344.
Robain, H., Descloitres, M., Ritz, M., Yene Atangana, Q., 1996. A multiscale electrical survey of a lateritical soil
system in African rain forest. J. Appl. Geophys. 34, 237 – 253.
Robain, H., Albouy, Y., Camerlynck, C., Dabas, M., Descloitres, M., Tabbagh, A., 1998. Geophysical surveys
contribution to structural and behavioural knowledge of tropical soils. Application to mapping purpose. World
Congress of Soil Science. AISS, Montpellier. 20-26/08/1998.
Robain, H., Albouy, Y., Camerlynck, C., Dabas, M., Descloitres, M., Mechler, P., Tabbagh, A., 1999. The
location of infinite electrodes in pole – pole electrical surveys: consequences for 2D imaging. J. Appl. Geophys. 41, 313 – 333.
Robain, H., Lajarthe, M., Florsch, N., 2001. A rapid sounding method, the three point method: a Bayesian
approach. J. Appl. Geophys. 47, 83 – 96.
Slater, L., Binley, A.M., Daily, W., Johnson, R., 2000. Cross-hole electrical imaging of a controlled saline tracer
injection. J. Appl. Geophys. 44, 85 – 102.
Telford, W.M., Geldart, L.P., Sheriff, R.E., 1990. Applied Geophysics, 2nd ed. Cambridge Univ. Press, New York.
Tooth, S., 2000. Process, form and changes in dryland rivers: a review of recent research. Earth-Sci. Rev.
51, 67 – 107.
253
Valentin, C., Bresson, L.-M., 1992. Morphology, genesis and classification of surface crusts in loamy and sandy
soils. Geoderma 55, 225 – 245.
Vuillaume, G., 1969. Analyse quantitative du rôle du milieu physico-climatique sur le ruissellement et l’érosion à
l’issue de bassins de quelques hectares en zone sahélienne (Bassin de Kounthouzout, Niger). Cah. ORSTOM,
Ser. Hydrol. VI (4), 87 – 132.
Yoon, G.L., Park, J.B., 2001. Sensitivity of leachate and fine contents on electrical resistivity variations of sandy
soils. J. Hazard. Mater. B 84, 147 – 161.
HYDROLOGICAL PROCESSES
Hydrol. Process. 22, 384– 394 (2008)
Published online 29 May 2007 in Wiley InterScience
(www.interscience.wiley.com) DOI: 10.1002/hyp.6608
Characterization of seasonal local recharge using electrical
resistivity tomography and magnetic resonance sounding
Marc Descloitres,1,2 * Laurent Ruiz,1,3 M. Sekhar,1,4 Anatoly Legchenko,2 Jean-Jacques Braun,1,5
M. S. Mohan Kumar1,4 and S. Subramanian1,6
1
3
Indo-French Cell for Water Sciences, IISc-IRD Joint laboratory, Indian Institute of Science, 560012, Bangalore, India
2 IRD, UR012-LTHE, UMR/CNRS-IRD-INPG-UJF, BP53, F-38041 Grenoble, Cedex 9, France
INRA Agrocampus Rennes, UMR, Sol-Agronomie-Spatialisation, 65 rue de St Brieuc, CS 82415, 35042 Rennes, France
4 Department of Civil Engineering, Indian Institute of Science, 560012, Bangalore, India
5 IRD, LMTG-OMP, UMR 5563 CNRS-UPS-IRD, Toulouse, France
6 Department of Metallurgy, Indian Institute of Science, 560012, Bangalore, India
Abstract:
A groundwater recharge process of heterogeneous hard rock aquifer in the Moole Hole experimental watershed, south India, is
being studied to understand the groundwater flow behaviour. Significant seasonal variations in groundwater level are observed
in boreholes located at the outlet area indicating that the recharge process is probably taking place below intermittent streams.
In order to localize groundwater recharge zones and to optimize implementation of boreholes, a geophysical survey was carried
out during and after the 2004 monsoon across the outlet zone. Magnetic resonance soundings (MRS) have been performed to
characterize the aquifer and measure groundwater level depletion. The results of MRS are consistent with the observation in
boreholes, but it suffers from degraded lateral resolution. A better resolution of the regolith/bedrock interface is achieved using
electrical resistivity tomography (ERT). ERT results are confirmed by resistivity logging in the boreholes. ERT surveys have
been carried out twice—before and during the monsoon—across the stream area. The major feature of recharge is revealed
below the stream with a decrease by 80% of the calculated resistivity. The time-lapse ERT also shows unexpected variations at
a depth of 20 m below the slopes that could have been interpreted as a consequence of a deep seasonal water flow. However,
in this area time-lapse ERT does not match with borehole data. Numerical modelling shows that in the presence of a shallow
water infiltration, an inversion artefact may take place thus limiting the reliability of time-lapse ERT. A combination of ERT
with MRS provides valuable information on structure and aquifer properties respectively, giving a clue for a conceptual model
of the recharge process: infiltration takes place in the conductive fractured-fissured part of the bedrock underlying the stream
and clayey material present on both sides slows down its lateral dissipation. Copyright  2007 John Wiley & Sons, Ltd.
KEY WORDS
groundwater recharge; hard rock aquifer; time-lapse ERT; MR sounding (MRS); Moole Hole watershed; south
India
Received 31 January 2006; Accepted 17 October 2006
INTRODUCTION
Groundwater resource is a key for agricultural and
human welfare in south India. Groundwater resource is
increasingly used all over this region especially when
the monsoon is irregularly distributed (Shivanna et al.,
2003). In some areas, groundwater is the main source for
irrigation (Rama Mohan Rao et al., 1996). The aquifers
are mainly located in weathered and fractured hard
rock. There is a need for better understanding of their
hydrogeological functioning in order to protect them from
excessive pumping and pollution, as well as helping
in artificial recharge management (Krishnamurthy et al.,
2000).
Indirect recharge from water bodies and streams can
contribute significantly to groundwater recharge (Scanlon
et al., 2002). This process can induce local and ephemeral
water table mounding. These short scale water level
variations can cast doubt on the validity of a common
monitoring of groundwater table at the watershed scale
through a piezometer network.
* Correspondence to: Marc Descloitres, IRD, UR012-LTHE, BP53,
38041 Grenoble, Cedex 9, France. E-mail: marc.descloitres@hmg.inpg.fr.
In this paper, a methodology is presented to assess
the spatial and temporal variability of water table level
combining two surface geophysical methods: electrical
resistivity tomography (ERT) and magnetic resonance
soundings (MRS). A field example is presented in a
small experimental watershed set up in a tropical climate
in the western Ghâts, south India (Braun et al., 2005).
The geophysical survey was carried out during and after
the 2004 monsoon with the objective to spatialize the
recharge below the main stream and to evaluate the role
of the slopes in the recharge process, if any. The results
are compared with the borehole data. The advantages and
limitations of both the methods are highlighted. Below
the slope, some ERT results show discrepancies with
borehole at depth, and are discussed using numerical
modelling. Finally, a conceptual model of the recharge
process below the stream is proposed.
INVESTIGATED AREA
The Moole Hole experimental watershed is situated in
the western Ghâts, in south India (Figure 1), in the
forested area of the Bandipur National Park, at 12°
385
CHARACTERIZATION OF SEASONAL RECHARGE WITH ERT AND MRS
Eastern coordinates (UTM, in m)
656200
656400
656600
INDIA
12° N
Bangalore
study area
3
76°30′ E
RT
se E
-lap
time
1 km
dl
she
ter
wa
t
imi
2
9
10
8
1297500
1
7
11
1297300
repeated
MRS
track
13
12
MRS
soundings
piezometers
Northern coordinates (UTM, in m)
1297700
5
1297100
Figure 1. Location map of the geophysical survey at the outlet of the Moole Hole experimental watershed
of latitude and 800 m elevation. Climate is sub-humid
tropical (1200 mm of yearly rain). The substratum of the
watershed belongs to the basement of Dharwar super
group (Moyen et al., 2001). It consists of a gneiss
intermingled with amphibolite and some quartz dykes.
The average strike value is N80° , with a dip angle ranging
from 75° to the vertical. The weathered thickness varies
a lot laterally (from a few metres to more than 35 m)
according to the nature and the fracturing of the gneiss
units, which are generally 5 to 25 m thick. In such hardrock context, the aquifer is generally of two types (Sekhar
et al., 1994, Maréchal et al., 2004). One is in the porous
clayey to loamy regolith with an apparent density lower
than the rock bulk density. Its hydraulic conductivity is
usually low. The other aquifer is in the fractured-fissured
protolith. Its apparent density remains close to the bulk
density of the rock. A network of fractures is present in its
upper part and the fracture density decreases with depth.
This aquifer plays a significant role for groundwater
exploitation but the amount of water is generally lower
than in the regolith. The geometry of the regolith as well
as the directions of the fractures or fissures in the protolith
can lead to an anisotropic hydraulic conductivity at the
scale of the borehole (Maréchal et al., 2004).
A set of piezometers (1, 2, 3) was implemented at the
outlet and in the slopes (Figure 1) in order to monitor
groundwater dynamics linked with the monsoon cycle.
In April 2004, using the ERT results, complementary
piezometers were implemented (7, 8, 9, 10, 11). Two
piezometers (12, 13) where drilled after the MRS survey. Groundwater electrical conductivity was also monitored because it influences the electrical resistivity of the
ground measured by geophysics. Variations of groundwater level and electrical conductivity of water observed in
2004 are shown in Figure 2 for the piezometers 1, 2, 8
and 10. The others are not shown for the sake of clarity.
Below the stream (piezometer 1, 7 and 11), groundwater level reacts very fast to the rain, and the amplitude
of variation is about 10 m. In the slope, the reaction is
delayed and less pronounced while moving away from
the stream. In the piezometer 8, 35 m away from the
stream, the level rises very progressively and the amplitude is about 3 m. In the piezometers 9 and 10, the rise
in water level starts more than 4 months after the beginning of the monsoon, and the amplitude is only 1Ð5 m. In
the piezometers 2 and 3, no increase in water level was
detected. The two events of water level rise observed in
the piezometer 2 are probably due to preferential infiltration along the piezometer pipe, due to an imperfect
watertightness around the casing and a local topography
allowing accumulation of water around the piezometer.
The groundwater electrical conductivity ranges
between 200 and 800 µS cm1 , with the piezometer
located below the stream showing significantly lower
values. Conductivities seem to decrease at the beginning
of the rainy season, and slowly increase during the dry
season. Although this seasonal trend is not very marked,
it could indicate that low conductivity new water dilutes
groundwater during the rainy season.
METHODS
ERT
The ERT method is widely used to perform surveys
where the sub-surface electrical resistivity is heterogeneous. It provides useful results on the geometry of
regolith and bedrock where aquifers take place if their
respective electrical resistivities are different. Electrical
Hydrol. Process. 22, 384– 394 (2008)
DOI: 10.1002/hyp
386
M. DESCLOITRES ET AL.
Figure 2. Piezometric levels and groundwater electrical conductivity records during the 2004 monsoon
resistivity is a parameter that depends on water content,
porosity, electrical conductivity of water, type of minerals and temperature (Telford et al., 1990; Rein et al.,
2004). Many authors used time-lapse ERT to locate and
monitor infiltration in the unsaturated zone (see Daily
et al., 1992; Barker and Moore, 1998; Binley et al., 2002;
French et al., 2002). Generally, bulk electrical resistivity
of unsaturated soils decreases if water content increases
with time. In the saturated zone changes in bulk electrical resistivity are usually linked with changes in electrical
conductivity of the groundwater.
Resistivity variations with time are useful to locate
the infiltration using apparent resistivity mapping, as
shown in an arid gully area (Descloitres et al., 2003)
or at the scale of a cultivated plot (Michot et al.,
2003). Practically, apparent resistivity is measured at the
surface using two current electrodes, A and B and two
potential electrodes, M and N (see Reynolds, 1997). For
a two-dimensional (2D) or three-dimensional (3D) data
acquisition, lots of electrodes are sequentially connected
using a multiplexer. Raw data are displayed in the form of
apparent resistivity as a function of the electrode spacing.
A longer spacing increases the depth of exploration.
An inversion scheme transforms apparent resistivity field
data into calculated resistivity. This calculated resistivity
is expected to be equal (or close to) bulk electrical
resistivity of the ground. Further details can be obtained
in the publications of Loke and Barker (1996), and Loke
(2000).
The ERT experiment consists of two data sets. The
first set is a complete survey of profile 1 (Figure 1). Its
objective is to give a distribution of resistivity in the
sub-surface with high resolution. This has been done
in March 2004, a few days before the first monsoon
rain. The second set is a survey that is focused on the
stream area. It is made several times during the monsoon.
Its objective is to delineate the infiltration and recharge
making the hypothesis that the variations of resistivity
in the vadose zone between the two dates are due to
significant variations of water content.
In this study two geometric arrays were chosen to
perform the acquisition. The first one is the Wenner array.
It is more sensitive to the vertical variations of resistivity
(Loke, 2000). Moreover, it is suitable for monitoring
purpose because this array brings a high signal-to-noise
ratio (Barker and Moore, 1998). The second array is the
dipole-dipole. It is more sensitive to the lateral variations
of electrical resistivity. It is well suited for detecting
2D or 3D objects because the two current electrodes are
adjacent and create a focused injection pattern. This array
is efficient in fractured hard rock studies as shown by
Seaton and Burbey (2002) because in such medium, the
distribution of resistivity is often 2D. In this study, a
configuration of the dipole-dipole array is used with the
distance between the electrodes A, B, M and N remaining
constant. This maintains the signal-to-noise ratio as high
as for the Wenner array. To combine the advantages of
these two different arrays, the two apparent resistivity
data sets are merged into the same inversion process
(Loke, 2000; De la Vega et al., 2003).
An in-line array of 64 electrodes was laid out and rolled
along the profile 1 crossing the stream (Figure 1). The
orientation of this profile is perpendicular to the strike
direction of the gneiss. The electrode spacing is 4 m.
This survey provides an estimated investigation depth of
25 to 30 m. Both sides of the stream (252 m long) were
monitored during the monsoon in 2004 using a Syscal
R2 resistivity-meter (Iris Instruments).
The RES2DINV inversion software was used to process the field data. The time-lapse ERT data set is interpreted using the time-lapse procedure proposed by Loke
(1999). For this procedure, a model of calculated resistivity is calculated when inverting the first data set. This
initial model is then used as a starting model to invert
the second data set in a sequential mode. The inversion
parameters were adjusted to the field conditions using the
Hydrol. Process. 22, 384–394 (2008)
DOI: 10.1002/hyp
following parameters:
ž a damping factor that increases slowly (1Ð05) with
depth,
ž a limited range of resistivity, from 10 m (clayey
soils) to 7500 m (fresh rock),
ž an option minimizing resistivity differences from one
data set to another,
ž a robust (blocky) inversion (Loke et al., 2001) had to
be used because the transition between the regolith
(weathered zone) and the fresh rock occurs in a few
metres, as observed by several resistivity logging (in
this case the robust inversion is recommended, as
proposed by Olayinka and Yaramanci (2000)),
ž a fine finite element grid (2 m width, corresponding to
the half of the electrode spacing) providing a better
accuracy in the calculations.
In addition to the ERT survey, resistivity loggings were
carried out below the water level in the piezometers 7,
8, 10 and 13 to allow a comparison with the resistivities
calculated by the 2D inversion.
MRS
The MRS method is a recently developed method for
prospecting groundwater (Legchenko and Valla, 2002;
Roy and Lubczynski, 2003). MRS differs from other geophysical methods for groundwater because it measures a
signal that is produced directly by groundwater itself. It
detects the presence of water by generating a resonance
of the protons HC of water molecules. When they are
excited by an alternating magnetic field at the Larmor frequency, they oscillate around their equilibrium position.
The Larmor frequency value depends on the intensity of
the earth magnetic field at the local survey area. In the
field, a cable is laid on the ground in a square loop of
50 ð 50 m2 at the sounding point. A current oscillating
at the Larmor frequency is injected into the transmitter loop to create a magnetic field. When the current
is abruptly turned off in the transmitter loop, this loop
acts as a receiver that records the secondary magnetic
field amplitude produced by the relaxation phenomena
when the protons go back to their original state. The secondary magnetic field is decaying with time. At present,
the method measures only the protons located in the saturated part and only if they are ‘free’. Bound-water protons
produce a signal that is too weak and too short to be
measured with available equipment. For more information on the method, see Legchenko and Valla (2002),
Legchenko et al. (2002), and Roy and Lubczynski (2003).
The method sounds deeper for an increasing intensity
of the excitation current and pulse duration. The sounding is performed using several current steps, while the
pulse duration is kept constant. The resulting sounding
curve is analysed to estimate the depth and thickness of
the aquifer, the MRS free water content and the MRS
hydraulic conductivity (see Lubczynski and Roy, 2003;
Legchenko et al., 2004; Vouillamoz et al., 2005). The
387
MRS parameters can be correlated with the aquifer characteristics through a calibration procedure using pumping
tests when available.
A detailed 2D MRS survey was carried at the outlet
of the watershed at the end of the monsoon (November
2004) using the NumisPlus equipment from Iris Instrument. This survey is presented in Legchenko et al.
(2006). The results of these studies are used in the present
paper for comparison with ERT. In addition to these
data, the MRS implemented at the centre of the stream
(Figure 1) above the recharge spot detected by ERT is
presented. This sounding was performed twice at the
same place to monitor groundwater depletion: in November 2004, when water level was at its maximum elevation,
and at the end of January 2005 when water level has
dropped to the lower level. This time-lapse MRS example
is one of the first attempts to use the MRS as a monitoring tool, a promising goal for MRS as suggested by
Lubczynski and Roy (2003).
RESULTS
ERT profile
The results of the 2D electrical resistivity survey along
profile 1 performed in March 2004 are presented in
Figure 3.
The calculated resistivity values range from 20 m to
more than 7500 m. From chemical analysis on cuttings
extracted from reference borewells in the watershed, a
correspondence is made between resistivity and the type
of rock. To highlight the main information, four intervals
of calculated resistivity that corresponds to four types of
material are displayed:
ž From 20 to 60 m: This interval corresponds to soils
(saturated or not) and clayey weathered materials. The
weathered materials are distributed in patches mainly
located at the south (between X D 64–140 m) between
the surface and 10 m deep. Some large patches are also
present between X D 256–320 m, but become scarce
below the northern slope.
ž From 60 to 150 m: This interval corresponds to
highly weathered rock, loamy to sandy materials. This
material is found mainly on the northern slope.
ž From 150 to 600 m: This interval corresponds to
weathered rock.
ž Over 600 m (and up to more than 7500 m): This
interval corresponds to the protolith. Its depth is highly
variable, from 5 to 25 m producing a jagged shape. This
may result from both the steep dip angle (more than
75° ) and the heterogeneous composition of the gneissic
bedrock that may lead to differential weathering. From
place to place (X D 150, 200, 320, 440 and 560 m) the
fresh rock is cut down by weathered formations (i.e.
electrically more conductive) that can go down to a
depth of 25 m.
The ERT results have been compared to resistivity logging performed below the water level in the piezometers
Hydrol. Process. 22, 384– 394 (2008)
DOI: 10.1002/hyp
388
S
N
Elevation (m)
860
distance from
the first electrode
576
(in m)
time-lapse ERT
eam
850
0.0 12
830
7
64.0
128
str
13
ma
in
840
192
448
9
2
320
8
3
512
384
Resistivity (Ohm.m)
10 102 103 104
0
256
2
calculated
4
6
measured in
november 2005
Depth (m)
820
810
8
10
12
measured
14
800
16
calculated resistivity (Ohm.m)
150 to 600
60 to 150
20 to 60
18
20
over 600
Piezometer 7
Figure 3. ERT results in March 2004 along profile 1. The calculated resistivity is plotted using four intervals. The RMS estimate for inversion
presented here is 5Ð9%. The circled numbers correspond to the piezometers located in Figure 1 (piezometers 2 and 3 are situated outside the profile).
Water level in March 2004 is drawn with a dashed line. Resistivity measured in boreholes with logging is presented at a depth using squares coloured
with the same colour scale as the calculated resistivity. Resistivity logging for piezometer 7 is presented versus depth by comparison with ERT
inversion results on the right side to illustrate the good agreement between calculated and measured resistivity
835
128
160
192
1
ma
7
in s
trea
m
S
Elevation(m)
224
8
256
288
distance from
the first electrode N
of profile 1 (in m)
9
352
320
825
800
resistivity ratio
(final state / initial state)
decrease
815
810
increase
0.2 to 0.4
0.4 to 0.6
0.6 to 0.8
0.8 to 1
1 to 1.3
Figure 4. Results of ERT time-lapse survey in the central part of profile 1 (see the location of the time-lapse survey in Figures 1 and 3). The ratio
between calculated resistivity values at final state (19 May 2004) to initial state (26 March 2004) is plotted using five ratio intervals. The decrease
of the calculated resistivity between the two dates (values below 1) is detailed using four intervals from black to light grey. Above 1, the calculated
resistivity has increased. Water levels measured at the same dates in the piezometers 1, 7, 8 and 9 are plotted using dotted (26 March 2004) and
continuous (19 May 2004) white lines
7, 8, 10, and 13. Resistivity logging results are shown in
Figure 3 for some representative depths, and a complete
comparison is shown for piezometer 7. They confirm the
resistivity calculated by the 2D inversion. A noticeable
result is depicted in Figure 3: in the piezometer 7 the
resistive bedrock (above 600 m) is encountered at 7 m
depth, while 35 m apart only weathered material (below
600 m) is found at 24 m depth at the bottom of the
piezometer 8. This logging result corroborates the high
lateral variability of resistivity calculated by the inversion and validates the parameters taken for the inversion
procedure.
highlight the variations of resistivity, Figure 4 shows the
resistivity ratio between final to initial state. This allows
us to identify the decrease of resistivity (values below 1),
or an increase of resistivity (above 1).
Before 26 March 2004 only 13 mm of rain was
recorded on the site, so this date corresponds to a very
dry status of the soils. The period between 26 March and
19 May 2004 was particularly rainy as 364 mm were
recorded, resulting from heavy pre-monsoon convection
storms. In the zone above an elevation of 815 m, which
corresponds to the unsaturated part at the initial state, the
major pattern is as follows:
Time-lapse ERT
Figure 4 presents the ERT time-lapse result obtained
comparing the initial state on 26 March 2004 and the
final state on 19 May 2004 when the rise of water level
below the stream had already occurred (see Figure 2). To
ž A decrease of the calculated resistivity is observed as a
quasi-continuous layer just below the surface down to
2Ð5 m depth.
ž A major decrease (more than 60%, i.e. values below
0Ð4) is located below the stream.
Hydrol. Process. 22, 384–394 (2008)
DOI: 10.1002/hyp
389
S
P9
Elevation (m)
830
P7
water levelcalculated
with MRS
P8
820
810
800
time-lapse ERT
790
measured water levels
125
175
225
275
325
MRS hydraulic conductivity (m/s)
N
MRS soundings
(loop extension)
1.2E-005
1E-005
8E-006
6E-006
4E-006
2E-006
375
Distance (m)
Figure 5. MRS hydraulic conductivity across the stream in November 2004. The centres of the MRS loops (Figure 1) are indicated with black
triangles, the loop extensions (50 m long) are attached to the symbol. The MRS performed twice (13 November and 26 January) above the main
stream is shown using a bold line. The water level calculated with MRS and the measured water level in November 2004 are indicated using a black
dashed line and black dots, respectively. The base of the section investigated with time-lapse ERT is shown using a grey dashed line
ž In the northern slope between X D 250 and 300 m,
below the uppermost layer with a decreasing resistivity,
the inversion results show a wide zone where resistivity
is almost constant (value around 1) or even increase
(above 1 and up to 1Ð3).
Below an elevation of 815 m that corresponds to the permanent water table, the calculated resistivity decreases.
This is noticeable below the stream and at X D 288 m. A
decrease of resistivity in the saturated part should be correlated with an increase of groundwater conductivity. But
groundwater conductivity is decreasing at these dates (see
Figure 2). Consequently, time-lapse ERT results below
water level are highly doubtful and this discrepancy is
investigated in the discussion.
MRS
The result of the MRS survey carried out in November
2004 and focused on the time-lapse ERT area is presented
in Figure 5. To draw this cross section, each MRS
have been interpreted using a one-dimensional (1D)
assumption and the resulting 1D models have been
interpolated along the profile to produce a pseudo-2D
image of the sub-surface (Legchenko et al., 2006).
Using numerical modelling, the MRS depth limit in
the Moole Hole has been estimated at 60 m, that is
twice the ERT investigation depth. The threshold of water
detection with MRS was estimated as 0Ð3% (Legchenko
et al., 2006). MRS water content provided by inversion
of field measurements was calibrated near a borehole.
This borehole (piezometer 13, Figure 1) was chosen as
a reference because it is located in a 1D geological
environment. It was found that the static water level
corresponds to the depth where MRS water content
reaches the half of its maximum value. The accuracy of
the water level estimation with MRS is determined to be
š1 m.
In Figure 5, the MRS water level varies from 3 m
(X D 225 m) to more than 15 m (X D 325 m). It matches
the measured water table below piezometer 7 and 9, but
overestimates it by 7 m below piezometer 8, due to the
lack of lateral resolution of the MRS method using a
50 ð 50 m2 loop as investigated in Legchenko et al.
(2006). Above the water table, the MRS hydraulic
conductivity cannot be calculated (unsaturated medium).
Below, the MRS hydraulic conductivity ranges from 2 ð
106 to 2 ð 105 m s1 and is irregularly distributed.
The MRS hydraulic conductivity is bell-shaped just
below the main stream.
A MRS measurement is repeated on two dates above
the stream (Figures 1 and 5) to monitor groundwater
depletion. The MRS loop surrounds four piezometers: 1,
7, 8 and 11. The first sounding is performed on the 13
November 2004, at the end of the monsoon. The second
sounding is done on 26 January 2005, once water has
depleted close to its pre-monsoon level. The MRS water
content and the MRS hydraulic conductivity versus depth
are presented in Figure 6 for the two dates.
Table I presents water levels measured on the dates of
the MRS survey in the piezometers 1, 7, 8 and 11.
In November 2004, water level is at its highest level
below the stream, i.e. 3 to 4Ð15 m below the surface
(piezometers 1, 7 and 11). At the end of January 2005,
water level depletion is nearly 6 m below the stream. At
the same time, the piezometer 8 shows a smaller depletion
of 1Ð5 m.
Estimated MRS water levels are 3Ð5 and 8Ð75 m on 13
November 2004 and 26 January 2005, respectively. This
water depletion (5Ð25 m, Figure 6) is close to the value
of mean depletion (6 m) given by piezometers 1, 7 and
11. Piezometer 8 is not considered for the mean water
level calculation because water depletion is much lower
(1Ð5 m) indicating a very different behaviour in this area.
Table I. Measurements of water level (in metres) at the date of
MRS measurements
Piezometer
1
7
8
11
13 November
2004
26 January
2005
Water level
decrease (m)
3Ð00
3Ð65
9Ð55
4Ð15
9Ð05
9Ð50
11Ð05
10Ð05
6Ð05
5Ð85
1Ð50
5Ð90
Hydrol. Process. 22, 384– 394 (2008)
DOI: 10.1002/hyp
390
0
mean water level on november, 13
th
-5
-5
mean water level on january, 26th
-10
-10
Depth (m)
Depth (m)
estimated
MRS water level
0
-15
-20
unreliable
MRS results
-20
-25
date of measurement
th
november, 13
january, 26
date of measurement
november, 13th
-25
th
january, 26th
-30
-30
0
(a)
-15
0.3
1
2
3
MRS water content (%)
1E-007
1E-006
1E-005
(b)
Figure 6. (a) MRS water content and (b) MRS hydraulic conductivity versus depth calculated for the MRS above the stream in November 2004 and
January 2005. For MRS water content (a), the hatched area corresponds to the results that are not possible to ascertain because of the low amplitude
of the MRS signal measured in the field. Estimated MRS water levels are given according to a calibration with a reference borehole (see text). For
the MRS hydraulic conductivity (b), the plot is limited to values above 107 m s1 for the same reason. The mean value of water level measured
in the boreholes 1, 7 and 11 is shown for both dates
Resistivity (Ohm.m)
0
75
150
225
300
375
0
-1
lower limit of infiltration
Depth (m)
The shallow aquifer seen in November (water content
2Ð7%, hydraulic conductivity 1Ð7 105 m s1 ) no longer
exists at the end of January. The result obtained in
January reveals a deeper aquifer (water content 0Ð5%,
hydraulic conductivity 1Ð4 ð 106 m s1 ) that is hidden
in November. This result shows the consequence of a
screening effect by the shallow aquifer, as investigated by
Legchenko (2005) and Legchenko et al. (2006). When a
very shallow aquifer is present (between the surface and
5 m deep) a deeper aquifer may be hidden if its water
content remains low compared to the superficial aquifer.
The MRS hydraulic conductivity of the lower aquifer is
revealed once the upper aquifer disappeared. Its value is
10 times less than in the upper part. As the main result of
this time-lapse MRS experiment, results indicate that a
significant depletion of water level occurs below the main
stream after the monsoon, in accordance with piezometric
measurements.
-2
Auger hole monitoring
moisture march, 26th
moisture may, 19th
resistivity march, 26th
resistivity may, 19th
-3
-4
DISCUSSION
Time-lapse ERT
In the vadose zone, the ERT resistivity decreases
(from 40 to 80%) between the surface and up to 5 m
down. To control this outcome, the results that were
obtained when monitoring the water infiltration carried
out on several auger holes located near the survey area
are used. Measurements included neutron probe and
resistivity logging down to 4 m through the vadose zone
every 15 days. Figure 7 shows an example of the results
obtained at the dates of ERT measurements. It shows that
the infiltration front reached a depth of 1Ð5 m only on the
19 May 2004. The soil water content increases from 20
to 30%, inducing a decrease of resistivity from a mean
value of 200 to 30 m, i.e. 85% decrease. This decrease
is consistent with the results obtained by ERT. However
the depth of the infiltration front obtained with time-lapse
10
15
20
25
30
35
Volumetric water content (%)
Figure 7. Resistivity logging and soil moisture variations measured in an
auger hole for a typical soil near the survey. The two dates considered
here are the same as the time-lapse ERT
ERT (2Ð5 m) is overestimated compared to the neutron
probe monitoring (1Ð5 m). This may be due to the large
spacing between electrodes used in this survey (4 m) that
is not adequate for very shallow investigations.
Below the stream ERT shows a major decrease of
resistivity by more than 60%. This is consistent with
piezometer data that shows water level increase of about
10 m in the piezometers 1 and 7, and of less than 1 m in
the piezometer 8 nearby. ERT results are consistent with
the water level records and allow delineating the water
table mounding below the stream (Figure 4).
Hydrol. Process. 22, 384–394 (2008)
DOI: 10.1002/hyp
391
Below the northern slope, some patches show an
increase of resistivity in the vadose zone. This result
is surprising because a decrease of water content in
the vadose zone during the monsoon is not likely.
Moreover, below water level (13 m and deeper), the
major part of the ERT section shows a decrease of
resistivity. In the saturated zone, such a decrease could be
explained only by an important increase of groundwater
conductivity. However, groundwater monitoring shows
that conductivity rather tends to decrease. Therefore,
these ERT results are questionable. To address this
question a synthetic model using a 1D layered ground
is studied. Two models are generated. The first model is
a typical resistivity arrangement of the sub-surface. From
the surface and down, four layers are considered:
ž a 1 m-thick dry soil (200 m),
ž a 9Ð7 m-thick weathered medium (100 m),
ž a 6Ð3 m-thick highly fractured rock and saprolite
(400 m),
ž a fresh rock (5000 m).
The second model is equal to the initial one but the resistivity of the first layer (1 m thick) decreases from 200
to 30 m to simulate an infiltration equivalent to the
infiltration measured with resistivity logging in the auger
hole (Figure 7). The synthetic apparent resistivities are
computed using the same time-lapse inversion algorithm
used for the interpretation of the field data. The resulting calculated resistivities are shown in Figure 8 as a
function of depth. The ratio of the initial to the final calculated resistivity is also plotted. The ratio shows first an
infiltration thicker than the simulated one (2Ð5 m instead
of 1 m). The decrease of resistivity is slightly underestimated (64% instead of 85%). This result confirms
that the ERT inversion could overestimate the depth of
infiltration. Second, an increase (C17%) and a decrease
(–33%) are noted deeper, in a zone where no model
variation was introduced. This phenomenon is damped
initial
1m
deeper. This modelling illustrates clearly that a time-lapse
inversion can exhibit artefacts (false variation at depth)
much deeper than the shallow infiltration. The reason
why the time-lapse inversion does not give satisfactory
results is an issue that cannot be addressed in detail in
this paper. A combination of different factors could be
involved. First, the characterization of the shallow infiltration in the field with an electrode spacing of 4 m is not
adequate. To characterize efficiently a shallow infiltration
(i.e. less than 2 m), smaller electrode spacing is required
in the field for recovering of the actual resistivity variations near the surface. If a shallow infiltration occurs,
which is generally the case if the soils are dry before
the first rains, care should be taken when interpreting 2D
time-lapse ERT data with a large spacing (i.e. 4 m or
more) between electrodes because the infiltration is not
well sampled. For thicker infiltration down to 5–10 m
(or recharge), the unit electrode spacing of 4 m is suitable because it provides an investigation depth similar to
the infiltration thickness.
Another reason why the time-lapse inversion is not
giving reliable results could be the non-uniqueness of
the model calculated by the inversion, due by example
to equivalence and suppression problems encountered
in electrical prospecting (Parasnis, 1997). Some recent
developments in inversion procedure could be considered
in the future to improve the reliability of ERT timelapse inversion, as proposed for example by Nguyen and
Kemna (2005) using difference inversion. The use of
external information is also a promising way to reduce
the non-uniqueness of the model and to get more reliable
time-lapse results as suggested by Loke (2000).
The modelling gives us an estimate of the uncertainty
of the ERT method in this case. For a true infiltration
of 1Ð5 m, the thickness of infiltration given by ERT is
overestimated. Deeper, resistivity variations in the range
35% to C20% should be considered as the result of
inversion inaccuracy rather than true (bulk) variations.
final
200 Ω.m
30 Ω.m
0
- 64%
-5
100 Ω.m
model
400 Ω.m
17 m
Depth (m)
-10
10.7 m
-33%
-15
-20
5000 Ω.m
+ 17%
inversion
resistivity calculated
by time-lapse 2D inversion
-25
initial
final
+ 6%
decrease increase
-30
10
100
1000
Calculated resistivity (Ohm.m)
10000 0.2 0.6 1 1.4
resistivity ratio
Figure 8. Resistivity calculated for a shallow infiltration (1 m) simulated over a 1D model using the 2D time-lapse inversion algorithm. The ratio of
the calculated resistivity (final/initial) is plotted versus depth and compared to the model resistivity ratio (right)
Hydrol. Process. 22, 384– 394 (2008)
DOI: 10.1002/hyp
392
Elevation (m)
830
P7
820
N
S
840
stream
P9
P8
?
810
800
790
1.2E-005
1E-005
8E-006
6E-006
4E-006
2E-006
780
125
165
205
245
285
325
Distance (m)
365
isocontour 600 Ohm.m
ERT
clayey material
(20-60 Ohm.m)
Figure 9. Comparison between ERT and MRS. For ERT, the resistivity interval 20–60 m is contoured and represents clayey materials. The
isocontour 600 m is the limit between the regolith (weathered) and the protolith (fissured or fresh). The black arrows indicate the possible pathway
for recharge below the stream. The shape of the water level is represented as a bold grey continuous line at the north. Because no water table level
measurements were available for the southern part at the time of the survey, the shape of the water table is suggested at the south using a dashed
line. The grey points indicate the water level measured in piezometers 7, 8 and 9 in November 2004
These uncertainties are related to the field data set (i.e.
arrays, electrode spacing, and actual resistivity values)
and may be different in other studies.
Comparison between ERT and MRS
A comparison between ERT and MRS is presented in
Figure 9. At the north, the water table level interpolated
from piezometer data is represented by a bold grey line.
At the south, the piezometer where not implemented at
the time of the ERT survey, thus the water table level is
only suggested as a possible distribution. To facilitate
the comparison, ERT results (Figure 3) are simplified
and superimposed to the MRS hydraulic conductivity
distribution.
ž The clayey materials are delineated in Figure 9 using a
resistivity ranging from 20 to 60 m. Their hydraulic
conductivity should be very low.
ž The clayey to sandy material are characterized by resistivity ranging from 60 to 600 m. They correspond to
the lower part of the regolith, i.e. a weathered rock.
These formations are usually considered as a potential
reservoir. MRS is not able to quantify their hydraulic
conductivity because those materials are mainly situated above the saturated zone, excepted below the
stream when they are temporarily saturated during the
monsoon. At this place, MRS indicates a hydraulic conductivity around 105 m s1 .
ž The fractured-fissured rock (protolith) is characterized by resistivity above 600 m. The highest values of MRS hydraulic conductivity (4 ð 106 to
1 ð 2 105 m s1 ) are mainly situated deeper than the
ERT 600 m contour. A noticeable correspondence
is found between X D 120 and 230 m. From X D 230
to 340 m, the MRS hydraulic conductivity does not
follow in detail the narrow conductive structures evidenced with ERT and boreholes. This is due to the lack
of lateral resolution of the MRS when the MRS loop is
wider than the structure (Legchenko et al., 2006). This
MRS hydraulic conductivity distribution indicates that
the fractured-fissured rock can be hydraulically conductive in accordance with the conceptual model of the
aquifer given by Maréchal et al. (2004).
ž From X D 340 m northwards, there is no more
detectable MRS signal. At this place, the bedrock
evidenced with ERT is situated above water.
Consequently, the free water content is much less in
the ground, and the MRS is no longer able to detect
it. This illustrates the lack of accuracy of current MRS
equipment for formations that contain less than 0Ð5%
water (Legchenko et al., 2006).
Finally, the investigated aquifer is highly variable at
a distance comparable with MRS loop size. MRS and
ERT have very different field set-ups (a loop and a line,
respectively). ERT gives a detailed image of distribution
of weathered part (electrically conductive) and of the
fissured-fractured part (electrically resistive) thanks to a
multiple array acquisition and a 2D inversion code that
provides an adequate lateral resolution. MRS gives an
image that integrates a volume of the ground at the scale
of the loop size, therefore with a lesser lateral resolution
than ERT. MRS clearly identifies the fissured-fractured
rock as a hydraulic conductive part of the aquifer, giving
valuable information not provided with ERT. Thus, even
though there is not perfect correspondence between the
results given by these two methods, it is considered
that results provided by both methods are giving very
complementary information on the aquifer.
Pattern of recharge inferred from geophysical results
From the comparison between ERT and MRS shown
in Figure 9, it is possible to propose a conceptual model
of the recharge process. The stream is cutting into thick
clayey materials. At this place the bedrock is close to the
surface. The upper part of the bedrock is hydraulically
Hydrol. Process. 22, 384–394 (2008)
DOI: 10.1002/hyp
conductive as observed with MRS measurements. The
recharge takes place in this fractured-fissured part of
the bedrock. The clayey materials with low hydraulic
conductivity slow down the recharge laterally. In a truly
2D geometry, these clayey materials could act locally
as hydraulic barriers. This hypothesis may explain the
high lateral variability of water level measured in the
piezometers. At the north of the stream, the shape of
the water mounding can be delineated as proposed in
Figure 9: the water level is almost flat below the stream
and deepens steeply along a clayey barrier. At the south
of the stream, another barrier is present and the water
level may also exhibit the same shape, but additional
boreholes are required to confirm this hypothesis.
CONCLUSION
At the outlet of the Moole Hole experimental watershed,
water level variations and recharge below the main stream
are studied during and after the 2004 monsoon using ERT
and MRS methods with the objective of spatializing the
phenomena.
For ERT, the bedrock and the regolith materials are
studied using the electrical resistivity distribution before
the monsoon. The results exhibit a jagged shape of the
regolith/bedrock interface due to differential weathering
of the vertically-foliated gneiss. The recharge is then
investigated during the monsoon using time-lapse ERT,
expecting resistivity variations linked with water content
variations. The time-lapse ERT results show first a shallow infiltration down to 2 m confirmed by neutron probe
measurements. Second, a recharge is marked as a major
decrease of resistivity below the stream (more than 60%),
while the piezometric level was rising at the same time.
Third, in the slopes, the calculated resistivity variations
show an increase (C30%) at intermediate depth (4–10 m)
and decrease deeper (more than 60%) below the water
table, not confirmed by water conductivity that decreases
at the same time. Modelling shows that an ERT inversion artefact occurs. This artefact may be a consequence
of the decrease of resistivity at shallow depth when infiltration begins. Consequently, it was found that time-lapse
ERT can suffer from severe interpretation artefacts. These
artefacts are troublesome to ascertain the bulk resistivity
variations at depth in the slopes. In forthcoming studies,
to design surveys or during the interpretation, a synthetic
modelling approach constrained with appropriate external
data such as time-lapse resistivity logging could be decisive to discard inversion artefact. This limitation could
be also investigated with synthetic modelling. Regarding
the recharge below the stream, it can be ascertained using
time-lapse ERT because the decrease of the bulk resistivity (more than 60%) is significant and deep enough to
make the phenomenon detectable and to avoid inversion
artefact.
A MRS survey is performed across the stream. MRS
is suffering from a lack of lateral resolution when the
water level is varying within the MRS loop. Some future
393
developments of the MRS equipment could overcome
this lack of resolution by using a smaller transmitter loop
combined with low-noise acquisition. MRS hydraulic
conductivity ranges from 2 ð 106 to 2 ð 105 m s1
and is clearly delineated, exhibiting significant variations
laterally. Preliminary slug tests carried out in some of
the piezometers give hydraulic conductivity values that
are in the same range of magnitude (Legchenko et al.,
2005). A survey including a long duration pumping test
is scheduled in the site to confirm these results. A single
sounding was repeated in the stream area once the water
level had depleted after the monsoon. This depletion is
clearly evidenced by MRS, confirming that MRS is a
very promising tool to monitor water level fluctuations.
From the comparison between ERT and MRS, a clearer
picture of the groundwater recharge is given. The ERT
determines the regolith/bedrock interface, whereas MRS
quantifies the hydraulic conductivity of the saturated
materials. The combination of time-lapse ERT and MRS
is found efficient to detect and outline the main recharge
phenomena below the stream. In the slopes, ERT evidences a decrease of resistivity linked with a shallow
infiltration down to 2 m. Deeper, no infiltration/recharge
is detected (i.e. down to more than 5–10 m) that would
have been evidenced by the time-lapse ERT as a major
resistivity decrease. The stream has cut into clayey material, and the recharge takes place in the fractured-fissured
part of the bedrock favouring the infiltration through
hydraulically conductive materials. Laterally and both
sides of the stream, clayey materials marked as electrically conductive structures by ERT, are acting as a barrier
slowing down the lateral infiltration of water. The pattern
is confirmed by the piezometric data on one side of the
stream.
In this hard-rock aquifer, it is found that the combination of ERT and MRS methods is an efficient way for
localizing the main recharge below the stream. In this
case, care should be however taken when interpreting
time-lapse ERT in the presence of shallow infiltration,
as some artefacts may occur in the inversion deeper
than 2 m. Despite this limitation, in similar environments
with localized recharge, borehole implementation can be
more easily optimized using this combination of nondestructive surface geophysical methods.
ACKNOWLEDGEMENTS
This study was carried out thanks to research grants
provided by:
- ‘Kabini river basin project’ of ORE-BVET
(www.orebvet.fr),
- French programs ‘ECCO-PNRH’ and ‘ACI-Eau’,
- Indo-French Cell for Water Science in Bangalore,
- Embassy of France in India,
- Indo French Centre for Promotion of Advanced Research
(IFCPAR) WA 3000.
Hydrol. Process. 22, 384– 394 (2008)
DOI: 10.1002/hyp
394
The authors thank the Karnataka Forest Department
for all the facilities and support they provided. The
field and laboratory assistants of the IFCWS are greatly
acknowledged for their contribution. The first author
wishes to thank Dr M.H. Loke (University of Sciences,
Penang, Malaysia) for the fruitful discussions on timelapse ERT inversion and J. Riotte for the pre-review of
the manuscript.
REFERENCES
Barker R, Moore J. 1998. The application of time-lapse electrical
tomography in groundwater studies. The Leading Edge 1454– 1458.
Binley A, Winship P, West LJ, Pokar M, Middleton R. 2002. Seasonal
variation of moisture content in unsaturated sandstone inferred from
borehole radar and resistivity profiles. Journal of Hydrology 267(3– 4):
160– 172.
Braun JJ, Ruiz L, Riotte J, Mohan Kumar MS, Murari V, Sekhar M,
Barbiéro L, Descloitres M, Bost A, Dupré B, Lagane C. 2005.
Chemical and Physical Weathering in the Kabini River Basin, South
India. Goldschmidt Conference Abstracts, Paper number A691,
Moscow, USA.
Daily W, Ramirez A, LaBrecque D, Nitao J. 1992. Electrical resistivity
tomography of vadose water movement. Water Resources Research
28(5): 1429– 1442.
De la Vega M, Osella A, Lascano E. 2003. Joint inversion of Wenner
and dipole-dipole data to study a gasoline-contaminated soil. Journal
of Applied Geophysics 54(1– 2): 97–109.
Descloitres M, Ribolzi O, Le Troquer Y. 2003. Study of infiltration in
a Sahelian gully erosion area using time-lapse resistivity mapping.
Catena 53: 229–253.
French HK, Hardbattle C, Binley A, Winship P, Jakobsen L. 2002.
Monitoring snowmelt induced unsaturated flow and transport using
electrical resistivity tomography. Journal of Hydrology 267(3– 4):
273– 284.
Krishnamurthy J, Mani A, Jayaraman V, Manivel M. 2000. Groundwater
resources development in hard rock terrain—an approach using remote
sensing and GIS techniques. International Journal of Applied Earth
Observation and Geoinformation 2(3/4): 204– 215.
Legchenko A. 2005. Improved modeling of the magnetic resonance signal
in presence of shallow aquifers. Near Surface Geophysics 3: 121–130.
Legchenko A, Valla P. 2002. A review of the basic principles for
proton magnetic resonance sounding measurements. Journal of Applied
Geophysics 50: 3–19.
Legchenko A, Baltassat J-M, Beauce A, Bernard J. 2002. Nuclear
magnetic resonance as a geophysical tool for hydrogeologists. Journal
of Applied Geophysics 50: 21– 46.
Legchenko A, Baltassat J-M, Bobachev A, Martin C, Robain H,
Vouillamoz J-M. 2004. Magnetic resonance sounding applied to aquifer
characterization. Journal of Ground Water 42(3): 363– 373.
Legchenko A, Descloitres A, Ruiz L, Reddy M, Girard J-F, Sekhar M.
2005. Characterization of fractured rock aquifers by surface
geophysical methods. In Extended Abstracts of the 11th European
Meeting of Environmental and Engineering Geophysics, Palermo, Italy,
5–8 September 2005, paper A034, CD ROM edition.
Legchenko A, Descloitres M, Bost A, Ruiz L, Reddy M, Girard J-F,
Sekhar M, Mohan Kumar MS, Braun JJ. 2006. Resolution of MRS
applied to the characterization of hard rock aquifers. GroundWater
44(4): 547– 554.
Loke MH. 1999. Time-lapse resistivity imaging inversion. In Proceedings
of the 5th Meeting of the Environmental and Engineering Geophysics
Society, European Section, Em1, Budapest.
Loke MH. 2000. Electrical Imaging Surveys for Environmental and
Engineering Studies. Technical note of Geotomo Software. Available
on line at http://www.geoelectrical.com.
Loke MH, Barker RD. 1996. Rapid least-squares inversion of apparent
resistivity pseudosections by a quasi-Newton method. Geophysical
Prospecting 44: 131–152.
Loke MH, Acworth I, Dahlin T. 2001. A comparison of smooth and
blocky inversion methods in 2-D electrical imaging surveys. In
Extended Abstracts of the ASEG 15th Geophysical Conference and
Exhibition, Brisbane.
Lubczynski M, Roy J. 2003. Hydrogeological interpretation and potential
of the new Magnetic Resonance Sounding (MRS) method. Journal of
Hydrology 283: 19–40.
Maréchal JC, Dewandel B, Subrahmanyam K. 2004. Use of hydraulic
tests at different scales to characterize fracture network properties in
the weathered-fractured layer of a hard rock aquifer. Water Resources
Research 40: W11508.
Michot D, Benderitter Y, Dorigny A, Nicoullaud B, King D, Tabbagh A.
2003. Spatial and temporal monitoring of soil water content with
an irrigated corn crop cover using surface electrical resistivity
tomography. Water Resource Research 39(5): 1138.
Moyen JF, Martin H, Jayananda M. 2001. Multi-element geochemical
modeling of crust-mantle interactions during late-Archean crustal
growth: the Closepet Granite (South India). Precambrian Research
112(1– 2): 87–105.
Nguyen F, Kemna A. 2005. Strategies for time-lapse electrical resistivity
inversion. In Extended Abstracts of the 11th European Meeting of
Environmental and Engineering Geophysics, Palermo, Italy, 5–8
September 2005, paper A005, 4 pp., CD ROM edition.
Olayinka AI, Yaramanci U. 2000. Use of block inversion in the 2-D
interpretation of apparent resistivity data and its comparison with
smooth inversion. Journal of Applied Geophysics 45: 63– 81.
Parasnis DS. 1997. Principles of Applied Geophysics (5th edition).
Chapman & Hall: London.
Rama Mohan Rao MS, Adhikari RN, Chittaranjan S, Chandrappa M.
1996. Influence of conservation measures on groundwater regime in
a semi arid tract of South India. Agricultural Water Management 30:
301–312.
Rein A, Hoffman R, Dietrich P. 2004. Influence of natural timedependent variations of electrical conductivity on DC resistivity
measurements. Journal of Hydrology 285: 215– 232.
Reynolds JM. 1997. An Introduction to Applied and Environmental
Geophysics. John Wiley & Sons: Chichester.
Roy J, Lubczynski M. 2003. The magnetic resonance sounding technique
and its use for groundwater investigations. Hydrogeology Journal 11:
455–465.
Scanlon BR, Healy RW, Cook PG. 2002. Choosing appropriate
techniques for quantifying groundwater recharge. Hydogeology Journal
10: 18–39.
Seaton WJ, Burbey TJ. 2002. Evaluation of two-dimensional resistivity
methods in a fractured crystalline-rock terrane. Journal of Applied
Geophysics 51: 21–41.
Sekhar M, Mohan Kumar MS, Sridharan K. 1994. Parameter estimation
in an anisotropic leaky aquifer system. Journal of Hydrology 163:
373–391.
Shivanna K, Kulkarni UP, Joseph TB, Navada SV. 2003. Contribution of
storms to groundwater recharge in the semi arid region of Karnataka,
India. Hydrological Processes 18(3): 473– 485.
Telford WM, Geldart LP, Sheriff RE. 1990. Applied Geophysics (2nd
edition). Cambridge University Press: New York.
Vouillamoz JM, Descloitres M, Toe G, Legchenko A. 2005. Characterization of crystalline basement aquifers with MRS: comparison with
boreholes and pumping tests data in Burkina Faso. Near Surface Geophysics 3: 107– 111.
Hydrol. Process. 22, 384–394 (2008)
DOI: 10.1002/hyp
Catena 67 (2006) 105 – 118
Deep infiltration through a sandy alluvial fan in semiarid Niger inferred
from electrical conductivity survey, vadose zone chemistry
and hydrological modelling
Sylvain Massuel a,*, Guillaume Favreau a, Marc Descloitres b, Yann Le Troquer c,
Yves Albouy c, Bernard Cappelaere a
a
IRD, UMR HydroSciences (IRD/CNRS/UMI/UMII), MSE, B.P. 64501, 34394 Montpellier Cedex 5, France
b
IRD, UR Geovast, Bangalore, India
c
IRD, UR Geovast, 92143 Bondy cedex, France
Received 21 July 2005; received in revised form 20 January 2006; accepted 23 February 2006
Abstract
In semiarid southwestern Niger, most of the groundwater recharge is indirect and occurs through endoreic ponds. Elsewhere in the
landscape, there is no evidence of deep infiltration, with a possible exception for gullies and alluvial fans on sandy slopes. In order to verify
this hypothesis, a detailed geophysical and geochemical survey was conducted on a large, representative mid-slope fan (6 ha). At this site,
distributed hydrological modelling conducted over the encompassing endoreic catchment (190 ha) showed high losses of runoff water by
infiltration. Electromagnetic mapping and 2-D electrical imaging survey were used to investigate the 35 m deep vadose zone; in addition,
8 boreholes were drilled following the geophysical survey to constrain the interpretation. Variations in apparent electrical conductivity
measured in boreholes appear to be mainly linked with changes in the soil solution mineralization. An extrapolation throughout the area
shows that apparent electrical conductivity of the ground is systematically lower below channels; this suggests localised leaching through the
unsaturated zone. A physically-based, 2-D distributed hydrologic model was used to estimate the amount of surface water loss by infiltration
for the 1992 – 2002 period. Depending on year, infiltrated volumes range from 1000 to 24 000 m3. This represents between 5% and 16% of
the runoff that reaches the final outlet of the basin, an endoreic valley bottom pond where recharge to the aquifer has been shown to occur.
Because leaching of the vadose zone is observed down to a depth of 10 m below channels, episodic groundwater recharge through sandy
mid-slope fans is highly probable during rainy years.
D 2006 Elsevier B.V. All rights reserved.
Keywords: Niger; Semiarid area; Infiltration; Local recharge; Alluvial fan; Geophysical survey; Unsaturated zone chemistry
1. Introduction
In southwestern Niger, since the early 1990s, hydrodynamics and geochemical methods have been applied at a
regional scale (4000 km2) to estimate natural groundwater
recharge to the unconfined aquifer (Leduc et al., 1997;
Favreau et al., 2002). In arid and semiarid Niger, studying
groundwater recharge is of paramount importance for
sustainable development, as most of the population depends
* Corresponding author. Fax: +33 4 67 14 47 74.
E-mail address: sylvain.massuel@msem.univ-montp2.fr (S. Massuel).
0341-8162/$ - see front matter D 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.catena.2006.02.009
upon this single permanent water resource for its own
consumption. In this environment, most of the groundwater
recharge is indirect and occurs through endoreic ponds,
natural outlets of a mosaic of catchments of the order of a few
square kilometres (Desconnets et al., 1997; Martin-Rosales
and Leduc, 2003). Elsewhere in the landscape, infiltration
deeper than 5 m below the soil surface, estimated by neutron
probe and soil moisture surveys, has not been evidenced and
has only been suggested as possible under specific locations
such as narrow banded vegetation on the plateaux (Galle et
al., 1999) and gullies in the sandy hillslopes (Peugeot, 1995;
Peugeot et al., 1997; Esteves and Lapetite, 2003). Surpris-
106
S. Massuel et al. / Catena 67 (2006) 105 – 118
ingly, whereas rainfall decreased by about 20% since the
1950 – 60s, hydrodynamics investigations have revealed a
continuous increase in groundwater reserves of about 4 m for
the last four decades, a phenomenon explained by the intense
land clearing that has induced crusting of the top cm of the
soil; as elsewhere in the Sahel, soil crusting has enhanced
Hortonian runoff, thus increasing both the number of
endoreic ponds and the amount of surface water reaching
the ponds (Leduc et al., 2001; Seguis et al., 2004). Increased
runoff may also have enhanced deep infiltration at some
runoff collecting sites other than ponds, but those have not
been identified yet.
The main objective of this study is to investigate the
possibility of deep infiltration (i.e. typically deeper than 5 m)
below the drainage network on the sandy slopes of this area.
In semiarid areas, deep infiltration producing groundwater
recharge is very localized in time and space and difficult to
estimate; combining various methods is often the key to
obtain reliable results (Scanlon et al., 1999a; Simmers, 2003).
Our approach is based upon a combination of sub-surface and
borehole geophysics, vadose zone chemistry and physicallybased hydrological modelling.
Subsurface geophysics used in this study is aimed at
mapping differences in electrical conductivity that could be
linked to variations in water content and/or conductivity of
the pore water and/or soil texture within the unsaturated zone,
both laterally and vertically. Such differences are expected in
the study site (Fig. 1), a densely braided sandy channel area
where infiltration is supposed to occur (Cappelaere et al.,
2003). When correlated with unsaturated zone profiles of
geochemical tracers, electrical conductivity mapping can
provide reliable extrapolation of punctual estimate of
recharge; subsurface geophysics can also help to spatially
better constrain hydrological models of surface/subsurface
flows. Previous investigations in semiarid areas have shown
that geophysical methods based on electrical conductivity
measurements are often well suited to delineate electrical
properties of the subsurface. Among the methods measuring
electrical conductivity at various depths, the more suitable
are: (i) Direct Current (DC) resistivity mapping or sounding
(e.g. Descloitres et al., 2003) and 2D-DC electrical imaging
when the ground cannot be approximated by a 1D model (e.g.
Beauvais et al., 2004), (ii) Frequency-Domain Electromagnetics (FEM) mapping (e.g. Cook et al., 1989; Scanlon
et al., 1999a,b), while (iii) Time-Domain electromagnetic
method (TDEM) is also considered as a suitable tool in some
situations as deep aquifers and mineralised waters (e.g.
Guérin et al., 2001). Within the scope of this study, the main
objective was to map the heterogeneities in electrical
conductivity down to depths exceeding 30 m below a large
mid-slope alluvial fan. FEM mapping was carried out at the
site-scale; in addition, a 2D-DC electrical imaging was
performed on a representative cross-section of the fan.
Vadose zone geochemistry is a widely used approach in
semiarid areas to infer mean groundwater recharge rates and
estimates of its temporal changes (e.g. Edmunds et al.,
1991). This approach has also been frequently used as a
supplementary tool in regional groundwater balance studies
(e.g. Wood and Sanford, 1995). Because it provides only
point-scale estimates, more representative results are
obtained when it is used with complementary approaches,
including sub-surface resistivity mapping (Cook et al.,
1989; Scanlon et al., 1999a,b). In southwestern Niger, data
on the deep unsaturated zone are limited. In the study area,
Fig. 1. The Wankama watershed with zoom in on the alluvial fan area and drill holes (small inset); the thin black lines refer to the watershed Digital Elevation
Model and the white network to the main gullies recorded by GPS survey in March, 2003. Inset: AAV: location of the 2D electrical profile (cf. Fig. 4). Numbers
refer to the drill hole locations and D indicates the inlet of the alluvial fan where hydrological runoff estimations were computed. Aerial photographs of
November, 1992 (IGNN, Niamey, Niger).
previous data were limited to the first upper metre (e.g.
Wezel et al., 2000), and for a single study, to a depth of up to
few metres (Nagumo, 1992). However, tracking deep
infiltration requires getting information down to several
tens of metres (ideally, to the water table). In this study,
vadose zone chemistry is used, along with other parameters
(water potential, texture, water content), both to interpret the
measured differences in electrical conductivity and to better
estimate the solute and water balance in the studied area.
In semiarid regions, the difficulty in obtaining good
quality data records of ephemeral and episodic floods is
widely recognized (e.g. Lange et al., 1999). Physically
based, spatially distributed hydrological modelling is a way
to overcome these difficulties, and can be used to generate
data for ungaged parts of a catchment. This approach was
chosen for the catchment that includes the studied mid-slope
alluvial fan (Fig. 1; Peugeot et al., 2003; Cappelaere et al.,
2003). For the present study, the water balance of the fan
was computed at the rainfall-event scale through the 1992 –
2002 decade, thus providing consistent values of annual
surface water loss by infiltration. From this set of data, a
hydrological functioning of the deep unsaturated zone under
sandy slopes is proposed.
2. Study site
The study site is located in the Sahelian southwestern
Niger, at 60 km east of Niamey (Fig. 1). The climate is
semiarid, with a mean annual temperature of 29 -C, a mean
potential evapotranspiration near 2500 mm yr 1 and a yearly
mean precipitation of 567 mm (Niamey, 1908 –2003; Niamey
Airport, pers. com.); these values are considered to be
representative for the study site. The rainy season from June
to September (90% of the annual rainfall) consists in intense
rainfall events of convective origin. These short duration
events produce Hortonian runoff that rapidly (within 1 – 3 h)
concentrates in temporary ponds, natural outlets of endoreic
catchments of a few square kilometres. In this environment,
all hydrological data indicate that most of the unconfined
aquifer recharge is indirect and occurs by deep infiltration
below the ponds (Desconnets et al., 1997; Leduc et al., 1997;
Martin-Rosales and Leduc, 2003). The geological context is
sedimentary and shallow formations belong to the Continental terminal (Tertiary) made up of loosely cemented clays,
silts and sands of continental origin; this formation outcrops
over a surface area of 150 000 km2 in southwest Niger. Dating
from drier periods of the Quaternary, aeolian sand deposits
occur in some places and can reach a few metres in thickness.
The water table elevation exhibits a classical pattern for
semiarid areas: a continuous, smooth surface (hydraulic
gradients < 1), with transient potentiometric fluctuations of
up to few metres below temporary ponds during the rainy
season (Leduc et al., 1997; Favreau et al., 2002). Depending
on the topography, the depth to the water table varies between
75 m below the lateritic plateaux to less than 10 m below the
107
dry valleys. The natural vegetation of the region is a wooded
savannah but under increasing clearing much of the area is
now a patchwork of fallow and millet fields.
The Wankama catchment (Fig. 1) has been intensively
studied since 1992; details about the hydrological survey
and data analysis are available elsewhere (Desconnets et al.,
1997; Peugeot et al., 2003). To summarize, the catchment
area is of 190 ha, with a mean slope gradient of 1.5% from
west to east. At the lower end, the endoreic, elongated
temporary pond acts as the natural outlet of water runoff of
the basin; the gully reported in Fig. 1 represents its main
tributary. According to runoff simulations for the 1992–
2002 period, surface water reaching the pond varies between
23 000 and 149 000 m3 yr 1 (Table 1). Most of this water
(about 90%) infiltrates and creates a temporary mound
below the pond. At mid-slope a large sandy alluvial fan
(‘‘spreading zone’’) acts as a natural collector of most of the
surface runoff from the upstream basin (Cappelaere et al.,
2003). Such large alluvial fans are a common feature in the
landscape (D’Herbes and Valentin, 1997). Hillslope soils of
the catchment are mainly sandy, weakly structured and can
be classified, according to Soil Taxonomy, as a sandy
siliceous isohyperthermic psammentic Haplustalf (Bielders
et al., 2000). Organic carbon content is less than 0.5%, with
fine particle content typically within the range of 5– 20%
(Nagumo, 1992; this study).
Within the catchment, this study focused on the alluvial
fan of about 6 ha (3% of the catchment area) occurring at
mid-slope; this fan represents the main outlet of the upper
part of the drainage basin (Fig. 1). Its main characteristics
are as follows: mean slope of 1.6% (close to the one of 1.5%
for the whole catchment); water table depth between 32 to
41 m; land surface occupied by shrub fallow (mainly Guiera
senegalensis), millet fields and sandy channels (17% of the
area in 2002). Whereas the main gully is narrow and reaches
few metres in depth in the upper part of the catchment, the
braided channels are typically large and shallow (< 0.5 m)
within the alluvial fan. Consequently, this results in possible
changes of the channel patterns after exceptionally high
flooding years.
Table 1
Computed runoff at the point of inflow for the alluvial fan (V D) and at the
pond (V p) for the 1992 – 2002 period (rainfall is reported as the sum of the
recorded events used for hydrological modelling)
Year
Rainfall (mm)
V D (103 m3)
V p (103 m3)
V D / V p (%)
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
Mean
485
474
541
513
537
353
510
489
433
247
291
443
17
21
8
24
10
13
18
12
16
1
4
13
117
129
75
149
91
84
127
84
107
23
36
93
15
16
10
16
11
15
14
14
15
5
10
13
108
3. Methods
3.1. Electrical conductivity
Ground electrical conductivity (ECg) is a complex
function of the soil characteristics (mineralogy, texture,
and structure) and of its water and solute contents. The wellknown Archie’s law (Archie, 1942) originally expressed for
saturated formations can be transformed for the unsaturated
zone as follows (Keller, 1988):
EC g ¼
1
IEC w ISwn I/m
a
ð1Þ
where ECg is the ground electrical conductivity (S m 1),
ECw is the conductivity of the pore water (S m 1), U is the
porosity (dimensionless), S w is the pore space saturation
(dimensionless, L3/L3), a is the saturation coefficient
(dimensionless), m is the cementation factor (dimensionless), and n the saturation exponent. For the sandy
formation, Keller (1988) proposes the values of 0.88, 1.40
and 2 for a, m and n respectively. This empirical law is valid
for sandy formations; when present, clayey particles could
play a role in increasing the value of the ground electrical
conductivity, because of their possible high cation exchange
capacity (CEC). As a consequence, the ground electrical
conductivity ECg can vary over a wide scale of values,
ranging from more than 1000 AS cm 1 for clayey saturated
material to less than 10 or even 1 AS cm 1 for dry sand. The
ECw and S w variables are difficult to obtain in the field. In
this study, these parameters are estimated by surrogates
obtained in the laboratory, respectively the experimental
conductivity ECwe produced with the usual lixiviation
protocol, and the gravimetric water content h w (moisture
weight / total weight). From Eq. (1) it is shown that the
ground electrical conductivity ECg given by geophysical
methods is highly dependent on the saturation, the porosity
and the electrical conductivity of the water in the soil.
Electromagnetic (EM) mapping was performed using a
Geonics EM-34 electromagnetic device to survey the
watershed with three intercoil spacings, 10, 20 and 40 m.
The operating frequencies are respectively 6400, 1600 and
400 Hz. For practical reasons, the coils were aligned
vertically (horizontal dipole mode), providing a stable
reading of the ground electrical conductivity at three depths
of investigation. This survey design provides a good
sensitivity to the upper surface layer conductivity, and an
investigation depth that can be roughly comparable to the
intercoil spacing. The ratio of secondary to primary
magnetic field over a uniform earth is directly proportional
to the ground electrical conductivity ECg (Mc Neill, 1980).
In the case of an electrically layered ground (1D case), the
reading is given as an apparent electrical conductivity ECa,
which is a function of the respective conductivities of each
layer. Two measurement campaigns were performed. In
August, 2002 the entire catchment was covered using the
40 m intercoil spacing (Fig. 2). Then the survey was
dedicated to a preliminary mapping of the fan area using
the intercoil spacings 10 and 20 m, with measurement
every 40 m (Fig. 3a and b). In March, 2003, a map of the
whole alluvial fan (425 400 m) was performed using the
20 m intercoil spacing, with measurement every 10 m. For
each campaign, a base station was monitored every 2 h to
overcome any problem due to instrumental drift.
A 2D electrical imaging survey was conducted in March
2003 along the profile AAV (Fig. 1), using a Syscal R2
resistivity-meter with 64 electrodes (IRIS instruments). A
couple of electrodes (A and B) was used for current
injection and the resulting potential difference was measured with a second couple of electrodes (M and N). The
basic field procedures, electrode arrays and interpretation
technique are described in Loke (2000). For our survey, the
electrodes were laid out every 4 m allowing a spacing of
252 m, that was repeated once to perform a profile of 508 m
(Fig. 4). Due to the very dry sandy surface, the contact
resistance was decreased by digging 20 cm deep pits, filled
with a salty clayey mud. The acquisition was performed
combining 2 arrays, the Wenner and Dipole – Dipole, taking
advantage of their different sensitivity to 2D distribution of
the ground resistivity. The Wenner and Dipole –Dipole data
sets have been interpreted jointly using the RES2DINV
inversion software (Loke, 2000).
Electrical conductivity logging was performed in the
vadose zone using an inflatable logging tool (Descloitres
and Le Troquer, 2004) in each of the 8 drilled auger holes
(Fig. 1); the acquisition was done using the ‘‘normal’’ pole –
pole array. This quadripole involves two inner electrodes A
Fig. 2. EM-34 mapping at the catchment scale, intercoil spacing 40 m
(August, 2002); measurement locations are indicated by black dots. The
white network refers to the main gullies. Inset: zoom in on the apparent
electrical conductivity changes at the fan scale; drill holes 1 and 2 are
located on high and low conductivity anomalies respectively.
109
Fig. 3. EM-34 mapping in the lower part of the fan area, W to E direction, intercoil spacing 10 m (a) and 20 m (b), August 2002. c): EM-34 mapping, intercoil
spacing 20 m, N to S direction, March, 2003. Measurement locations are indicated by black dots. The white network refers to the main gullies. Background:
microlight aircraft photograph of the fan area, August, 1998 (J.L. Rajot, IRD, Niamey, Niger).
and M and two remote surface electrodes B and N at 150 m
away from the drill hole. The AM spacing was 0.25 m. The
measurements were done every 0.5 m down the hole. The
short spacing between electrodes A and M allows measurements of the ground electrical conductivity ECg within an
estimated radius of 20 cm around the sampling point.
3.2. Vadose zone chemistry
For this study, 8 boreholes of 50 mm of diameter were
drilled without any fluid to depths between 5 to 25 m in
August, 2002 (drill holes 1 and 2) and March, 2003 (drill
holes 3 to 8) with a power engine drillmite auger (locations
are shown in Fig. 1). At surface, soil samples were collected
each 0.5 m and rapidly poured using plastic gloves into 335
cm3 aluminium tins to preserve samples from evaporation
and contamination. For this study, gravimetric water
content, water potential measurement, particle-size analyses,
experimental conductivity of the pore water, major ion
chemistry and pH were measured. Analyses were performed
in Montpellier, France, within a few months of sampling.
Random duplicates showed good reproducibility. On selected samples, X-ray diffractions were also performed to
determine the soil mineralogy.
Gravimetric water content (h w) was measured after
drying an aliquot of about 100 g of each sample in an oven
for 24 h at 105 -C. Water potential was estimated for some
duplicate samples by the filter-paper method described in
Hamblin (1981), using Whatman-42 filter paper, with an
uncertainty of about 20%. Solute content was obtained after
elutriation of 20 g of dry sediments in 50 ml of doubledeionised water (< 1 AS cm 1) during 30 min; experimental
conductivity of the pore water (ECwe) was subsequently
measured on a 0.45 Am filtered aliquot with a commercial
conductimeter (WTW, Tetracon). On an unfiltered aliquot,
pHH2O and pHKCl (1 mol L 1 KCl) were measured with a
commercial pH-metre (WTW, Sentix). Major ions were
analysed on 0.45 Am filtered aliquots by capillary ion
analyser (precision of about 5%). Particle-size was analyzed
by sedimentation on 25 selected samples from drill holes 1
and 2 using the pipette-method with an automatic particlesize analyser.
3.3. Hydrological model
The physically based, 2D-distributed hydrologic model
of Cappelaere et al. (2003) was used for the present study.
This model was built using the abc-rwf generic model
developed by these authors from the original r.water.fea
model of Vieux and Gaur (1994). In this model, time and
space are discretized consistently and finely enough to
represent the water flow dynamics of individual storm
events over the whole catchment (grid resolution of 20 m).
Infiltration, runon/runoff production and routing functions
(kinematic-wave with Green– Ampt and Manning equations) are fully coupled, and solved concurrently using finite
110
elements in space and finite differences in time. The model
was calibrated and validated for the Wankama catchment
based on the rainfall events that occurred from 1992 to 2000
and reproduced the observed catchment behaviour satisfactorily (Cappelaere et al., 2003). The alluvial fan is
represented in the model by a 7.6 ha area with the normal
DEM slope.
4. Results
4.1. Electromagnetic mapping
Electromagnetic mapping was used to delineate relative
differences in vadose zone conductivity. In Fig. 2 are
presented the EM34 40 m-spacing mapping results at the
catchment scale. Apparent electrical conductivity values
range from 10 to 200 AS cm 1 (1000 to 50 V m respectively)
and show a general increase from upslope (west) to
downslope (east). This trend is explained by a decreasing
thickness of the vadose zone with decreasing elevation: when
going downward, the thickness of the resistive (unsaturated)
ground decreased from more than 60 m down to less than 20
m, thus raising the measured apparent electrical conductivity
value. On the sandy fan area (Fig. 2), the values lay between
36 at the north and 83 AS cm 1 at the centre (between 280 and
120 V m respectively). The two deeper drill holes (1 and 2,
Fig. 2) were installed to explain this contrast: drill hole 1 was
located at a higher apparent electrical conductivity anomaly
near a large channel, whereas drill hole 2 was located in a low
apparent electrical conductivity spot, corresponding to a
small-slope fallow plot (Fig. 2).
Results from the shallow sub-surface were obtained
using shorter intercoil spacings. Fig. 3 presents the results of
the EM mapping focusing on the sandy fan area, using
intercoil spacings of respectively 10 (Fig. 3a) and 20 m (Fig.
3b, c). The 10 m spacing map shows apparent conductivities
lying between 11 and 50 AS cm 1 (from 900 to 200 V m
respectively). The distribution of the poorly conductive
zones appears complex: in the centre, it could be linked with
the dense channel distribution. Except for the middle part of
the northern gully, large spots of higher apparent electrical
conductivity occur away from the main gullies. The 20 m
spacing map shows the same range of values, from 11 to 50
AS cm 1 (Fig. 3b). The less conductive spots (below 17 AS
cm 1) are distributed at the centre and in the northeastern
part of the area, and higher apparent conductivities are
observed in the southern and nortwestern parts. In details,
significant differences appear with the 10 m intercoil
spacing map; this may be due to the time-lag between the
two field measurements (¨ 1 month) and to subsequent
surface water infiltration (see ‘4.4) and/or to locally
heterogeneous distribution of apparent conductivities with
depth.
In March, 2003 a larger EM-34 survey of the fan (18 ha,
intercoil spacing of 20 m, north –south tracking, measurement each 10 m) confirmed the observations obtained in the
lower part of the fan (Fig. 3a and b); in particular, (i) though
the measurements took place by the end of the dry season, the
same range of values was observed and (ii) large spots of
higher apparent electrical conductivity occurred around the
fan, with, in details, a more complex zonation (Fig. 3c).
In order to compare methods, EM-34 measurements with
intercoil spacing of 20 m were performed simultaneously to
the 2D electrical imaging on a single profile (AAV in Fig. 1;
Fig. 4). In accordance with EM-34 mapping results,
relatively lower apparent electrical conductivity was observed below the main gullies.
4.2. 2D electrical imaging
A two dimensional (2D), 508 m-long electrical imaging
profile was performed perpendicularly to the fan area (Fig.
1). In Fig. 4 is reported the calculated ground electrical
conductivity versus depth obtained by joint inversion of
the Wenner and Dipole – Dipole 2D data sets. The number
of iterations was limited to three because there was no
significant decrease of the RMS criteria for further
inversions. As the inversion had to comply with two sets
of data, the corresponding RMS is relatively high (19%).
Fig. 4. Joint analysis of Wenner a and b profiles (mutual inversion) by Res2Dinv; unit electrode spacing 4 m, iteration 3, RMS error 19.1%. A higher
conductivity layer is displayed (blue colours) between 5 and 10 m depth; below most sandy channels this conductive layer is interrupted. Upper part of the
figure: apparent electrical conductivity measured by EM-34 survey (intercoil spacing 20 m, measurement each 4 m). (For interpretation of the references to
colour in this figure legend, the reader is referred to the web version of this article.)
111
Fig. 5. Physical parameters measured in drill holes 1 and 2; a) ground electrical conductivity ECg, b) matric suction, c) water content and d), e) grain size
distribution for drill holes 1 and 2, respectively. The soil surface is respectively at 226.84 and 227.05 m a.m.s.l for drill holes 1 and 2.
The conductivities range from 1.25 to 330 AS cm 1 (8000
to 30 V m respectively). From the surface down to 2 – 3 m
a resistive layer is noted, and corresponds to a dry sandy
layer (March 2003, dry season); from 3 to 10 m, a
conductive layer is observed. Its conductivity ranges from
60 to more than 300 AS.cm 1 in a discontinuous way,
forming patches with higher conductivity separated by
lower conductivity ones. Below this level, from 10 down
to 35 m (maximum depth of investigation), the vadose
zone is mostly resistive. Its conductivity mostly ranges
from 1.25 to 3.3 AS cm 1 with at some places, some more
conductive patches.
4.3. Electrical conductivity logging and vadose zone
analysis
Results of electrical conductivity logging are shown on
Figs. 5a and 6a for the two deepest drill holes (1 and 2) and
on Fig. 7a for the others. Each of the two drill holes 1 and
2 represents a distinct pattern of electrical conductivity
change with depth. For the drill hole 1, ground electrical
conductivities are ranging from 0.8 to 15.3 AS cm 1
(12 500 to 650 V m). Those values are typical for an
unsaturated sandy formation, with low water content. Drill
holes 3, 6 and 8 display the same behaviour as the
drill hole 1 with ground electrical conductivity below
20 AS cm 1 (500 V m) all along the logging profile (Fig.
7a). For the drill hole 2, the range is wider, from 1.6 to
200 AS cm 1 (6250 to 50 V m). Ground electrical
conductivity rapidly increases from surface to 4 m deep.
From 5 to 10 m depth, the ground is more conductive,
values are over 150 AS cm 1 (below 65 V m) with a
maximum at 8 m in depth. These higher values typically
indicate that the formation is either more clayey, contains
more water or presents an increase in the water solute
content. Drill holes 4, 5 and 7 have the same behaviour
Fig. 6. Chemical parameters measured in drill holes 1 and 2; a) ground electrical conductivity (reported from Fig. 5), b) experimental conductivity of the pore
water, c) anions, d) cations; e) pH and pH-KCl.
112
Fig. 7. Physical and chemical parameters measured in drill holes 3 to 8. a) ground electrical conductivity ECg; b) anions; c) water content; d) experimental
conductivity of the pore water; e) cations; f) pH. Because both Ca vs. Mg, and pH-H2O vs. pH-KCl appeared to be well correlated (r 2 of 0.98 and 0.87
respectively), Ca and pH-H2O were chosen to represent their changes with depth for drill holes 3 to 8.
as the drill hole 2 with ground electrical conductivity over
100 AS cm 1 (below 100 V m) when reaching 4 m depth
(Fig. 7a).
Grain size distribution analysis shows that sedimentary
formations are homogeneous between the drill holes 1 and 2
(Fig. 5d, Fig. 5e). Grounds are essentially sandy (33% to
90%) to silty (3% to 28%) with variable content of clay (3%
to 41%); pebbles occur between 5 and 10 m in small
proportion (< 10%). Two stratums are more clayey and occur
at depths from 5 to 7 and 10 to 12 m for the two drill holes.
For these layers, X-ray diffractions confirm the abundance of
quartz (sand) and show that clay fraction is made almost
exclusively of kaolinite (goethite is also present). For the
whole profiles, such a similar grain size distribution suggests
that porosity could be the same for the two drill holes.
Consequently, the influence of porosity U in Eq. (1) may be
similar for the two drill holes. Because kaolinite is known to
have a low CEC, influence of the clay content on the
apparent electrical conductivity is expected to be low.
Matric suction measurements were performed on dedicated duplicates for drill holes 1 and 2. For both profiles,
deeper than 4 m, values are high and lie between 25 to 75
bar; around 2 to 3 m, matric suction is even higher and can
reach 150 bar (Fig. 5b). At surface, it displays a rapid
decrease, down to 0.05 bar at 0.1 –0.7 m below the soil
surface, followed by a steep rise in the top cm for drill hole
1 (Fig. 5b). Considering that sampling occurred during the
rainy season (August, 2002), such a typical ‘‘S’’ shape can
be explained by recent infiltration of rain water at shallow
depth, followed by incomplete re-evaporation. However,
though the two holes are located at various distances from
gullies (Fig. 1), very similar water potential profiles are
obtained and no noticeable difference in infiltration at the
time of sampling can be inferred. This can be explained by
the low amount of rainfall and runoff that occurred in 2002
(see below, ‘4.4), thus preventing any significant infiltration
through gullies.
Gravimetric moisture content profiles are similar in drill
holes 1 and 2 (Fig. 5c). The measured h w range from 1.8%
to 11.3% and are closely related to the grain size distribution
(Fig. 5d and e). Except for the top metre where h w partly
represents recent infiltration (as shown by matric suction
values), higher values systematically correspond to
increases in clay content, and conversely lower values to
decreases in clay content. Almost the same range of
moisture (0.6 – 10.7%) is observed for drill holes 3 to 8 that
present a single pattern of increasing moisture with depth
(for these holes, the lower moisture content near the soil
surface can be explained by sampling during the dry season;
Fig. 7c). Consequently, the influence of the saturation
parameter S w in Eq. (1) could be considered as invariant in
time, space and depth (> 2 m).
Experimental conductivity of the pore water (ECwe), pH
and ionic contents profiles are reported in Fig. 6 for drill
holes 1 and 2 and in Fig. 7 for drill holes 3 to 8 respectively.
For each profile, ECwe appears to be well correlated to
ground electrical conductivity (ECg). As for ECg profiles,
two distinct families of ECwe change with depth can be
distinguished, being respectively represented by drill holes 1
and 2 (Fig. 6b; Fig. 7d). For profiles of the first group (drill
holes 1, 3, 6, 8) ECwe is rather constant with depth (except
for the first top four metres) and ranges from 4 to 24 AS
cm 1; this implies a low ion content. For profiles of the
second group (drill holes 2, 4, 7), the maximum ECwe lies
within the range 46 to 276 AS cm 1. Drill holes 5, though
related to high ECg values, display relatively low ECwe at
depth and represents an exception (Fig. 7d); this may be due
to local small-scale heterogeneity at the sampling location,
the ECg value representing a larger ground volume.
ECwe represents an integrated value of the ionic water
composition. In order to determine the chemical composition of the solute content, major ion analysis (Ca2+,
Mg2+, Na+, K+, for cations, SO42, NO3 and Cl for
anions) were performed for each sample; pH-H2O and pHKCl measurements were also performed to determine free
and exchangeable H+ respectively. For these two parameters, values range between 4.6 and 8.8 pH units (pH-H2O)
and between 3.9 and 8.2 pH units (pH-KCl), the positive
difference ranging between 0.1 and 2.4 pH units (Fig. 6e;
Fig. 7f). Ion contents are reported graphically on Fig. 6c
and d for drill holes 1 and 2 and on Fig. 7b and e for drill
holes 3 to 8. Increases in ECwe appear to be mainly linked
with increases in NO3 for anions, and in Ca2+ for cations;
for the highest solute contents (drill hole 4), NO3 and
Ca2+ reach respectively 1.86 meq L 1 (288 ppm) and 1.05
meq L 1 (53 ppm). Mg2+ appears to be highly correlated
with Ca2+ and follows the same variations with a lower
content. Some higher levels in Ca2+ and Mg2+ correspond
with increases in pH values up to 8.6 or 8.8 pH units (drill
holes 2 and 4 respectively), thus suggesting the presence
of carbonate minerals. SO42 content is always low (nearly
2 / 3 of the analyses are below the detection threshold) and
never exceed 15% of the anion content. Cl and K+
contents are always low (< .1 meq L 1, i.e. <9 ppm) and
do not correlate with the bulk mineralization. In details, the
vadose zone chemistry changes in chemical composition
with depth, with Na+ for cations and Cl for anions being
dominant for some drill holes at discrete depths (Fig. 6;
Fig. 7). These results are in good agreement with previous
findings in the same region of an important small scale
chemical heterogeneity within the first upper metres of the
ground (Nagumo, 1992).
113
4.4. Hydrological modelling
The Wankama catchment model was run on an event basis
from 1992 to 2002 (Table 1). Rainfall input was recorded with
rain-gauges located on the basin. The hydrological balance
was computed for each cell of a 20 m resolution grid.
According to the fully distributed model, for the whole
period, all of the incoming flow was lost in the alluvial fan by
infiltration. Runoff volumes (V D) computed at point D (the
point of inflow for the alluvial fan, see Fig. 1) are compared
with runoff volumes computed at the downslope endoreic
pond (V p), where recharge has been shown to occur
(Desconnets et al., 1997; Leduc et al., 1997). V D ranges
between 5% to 16% (mean 13%) of the total surface flow
production computed in the pond; this represents between
1000 and 24 000 m3 of surface water infiltrating through a
sandy channel area estimated near 1 ha (17% of the active part
of the alluvial fan). Compared to the surface of the pond, the
infiltrating fan area appears smaller (the maximum surface of
the pond is near 9 ha). However, as reported in Table 1, the
maximum annual V D entering the fan (24 000 m3) exceeds the
minimum V p value (23 000 m3), for which groundwater
recharge was indeed observed. Therefore, all other things
being equal, it could be concluded that groundwater recharge
may have occurred through the alluvial fan for the 1992–
2002 period, at least for the highest computed yearly runoff.
Two other points inferred from the hydrological modelling approach lie (i) in the relative importance of V D vs. V p
depending on years and (ii) in the non-linear relationship
between rainfall and runoff. According to computed values
reported in Table 1 (and beyond the logical observation that
high V D are positively correlated with high V p) the relative
contribution of V D increases with total runoff (V D + V p); in
other words, the higher the runoff, the more (in relative part)
the fan area may contribute to deep infiltration. In Fig. 8a
are displayed computed V D as a function of time for
respectively a wet (1995) and dry year (2002). Fig. 8b
displays total rainfall events for the same two years. Though
rainfall in 1995 (513 mm) is 1.8 times higher than in 2002
(291 mm) both the number of runoff events (7 vs. 3) and the
runoff volumes V D reaching the fan (24 000 vs. 4000 m3)
vary in greater proportion (respectively by a factor of 2.3
and 6.0; Table 1; Fig. 8a). This further emphasizes the fact
that depending on years, larger changes in runoff and
eventually deep infiltration can be expected than simply
inferred from changes in rainfall (Table 1).
5. Discussion
5.1. Ground electrical conductivity (ECg) interpretation
In the study area, direct measurements in drill holes have
shown a good relationship between ECg and ECwe (Fig. 6a
and b; Fig. 7a and d). In our case, ECwe is relatively high
compared to the contribution expected from a solid matrix
114
Fig. 8. a) Runoff volumes computed by hydrological modelling at the point
of inflow for the alluvial fan in 1995 (wet year) and 2002 (dry year). b)
Measured event rainfall for these two years; vertical arrows (1 to 5) indicate
dates of measurements for 2002: 1: EM-34 with 40 m intercoil spacing
mapping, 2: EM-34 with 10 m intercoil spacing mapping, 3: drilling of hole
1, 4: drilling of hole 2, 5: EM-34 with 20 m intercoil spacing mapping.
made of quartz and kaolinite with low CEC (estimated about
7.5 meq/100 g; Nagumo, 1992). Other matrix terms
involved in ECg values, such as porosity U and granulometry do not seem to act significantly upon its observed
changes (Fig. 5). Assuming that the relationships between
the Archie law variables ECw and S w on one hand (Eq. (1)),
and their experimental surrogates ECwe and h w on the other
hand, can be acceptably approximated by some linear or
power functions (i.e. of the general form y = k 1 I x k2, with
constant k 1 and k 2 for a given soil), then the transformed
Archie empirical law, Eq. (1), can be reformulated as:
logEC g ¼ K þ alogEC we þ bloghw
ð2Þ
where K, a and b are new unknown constants (K
incorporates in particular the effects of Eq. (1)’s a, m and
U). These parameters in Eq. (2) can be estimated from the
drill hole data by applying linear regression of Eq. (2) from
the drill hole data, yielding K = 0.89, a = 1.3 and b = 0.87
(Fig. 9). The resulting R 2 is 0.82 with a contribution by
ECwe and h w to the expressed variance respectively of 63%
and 19%. This simple analytical model confirms that ECwe
values play a prominent part on the ECg measurements; this
observation is valid for the whole scale of ECg measurements, with no significant change in the determination
coefficient with the ECg range considered.
In the study area, the quasi-exclusive ECg / ECw relationship is in accordance with (i) the large, two order in
magnitude change in ECwe (Fig. 6b; Fig. 7d), (ii) the
kaolinic nature of the clay fraction, with consequently very
low CEC and (iii) the lack of any deep infiltration during the
2002 rainy season (Fig. 5b; Fig. 8). Elsewhere in the
landscape, such a simple correlation between ECg and ECw
may not be observed, particularly in clayey valley bottoms
(shallow water table, higher h w, smaller range of ECw and
presence of vermiculite/smectite within the clay fraction;
Nagumo, 1992), and for more humid periods of measurements (possibly high and transient h w signal). Within the
investigated alluvial fan area, ECg changes measured by
sub-surface geophysics (EM-34, DC) are interpreted in
terms of changes in ECw.
EM-34 mapping at 40, 20 and 10 m intercoil spacing
show significant changes at small scale within the studied
fan area (Figs. 2 and 3). Even if the EM34 device measures
only an apparent electrical conductivity ECa in a nonuniform ground, the apparent conductivity variations
measured with EM34 can be roughly related to ECg
calculated from 2D electrical imaging inversion along the
DC profile (Fig. 4). The EM34 apparent electrical conductivity variations are a representation of various vadose zone
leaching intensities. Because the resolution is decreasing
with depth, these differences are probably more related to
leaching of the upper part of the investigated zone
(depending on the intercoil spacing considered). Higher
leaching may be observed below the densely braided
channel area, whereas lower leaching is observed at
distance, below fallow and millet fields (Fig. 3c). Although
complex in details, the generally lower conductivity
observed within the fan area expresses its hydrological
functioning as a deep infiltration area. In the upper part of
the fan, shifting braided channels from year to year makes
difficult a detailed interpretation. In Fig. 3a and b, small
changes in ECa can be noticed within the fan and could be
linked with transient changes in h w; however, in most parts
of the fan, the general distribution of ECa remains constant
for the survey period and express a stable leaching pattern.
2D-DC electrical imaging (Fig. 4) highlighted the spatial
extent of changes in solute contents already characterized by
EM-34 and drill hole measurements. At surface, a leached
Fig. 9. ECg computed as a function of ECg measured for the 127
measurements of the 8 drill holes with ECg = 10K I ECwea I h wb and K = 0.89,
a = 1.3 and b = 0.87 (the measured ECg are reported as a function of depth
in Figs. 5a and 7a; for drill holes 1 and 2, the first two metres of
measurements were excluded from the data set, the ECg being impossible to
be correctly measured due to broadening of the upper part of the hole
during the drilling).
sandy layer of about 2 to 3 m in thickness is observed
throughout the transect, and could represent the mean
annual depth of rain water infiltration. More in depth, a
high solute content layer is mostly present between 4 to
10 m. Different hypotheses about this solute accumulation
are developed in conclusion. To the best of our knowledge, no
previous evidence of a high mineralized vadose zone layer
had been reported before in the region, as soil studies were
restricted to the first top metres of the ground. This deep
mineralized layer is interrupted at discrete places, below the
main sandy channels (Fig. 4); this denotes occasional deep
leaching, down to depth of at least 10 m (for minor channels,
this relationship is less obvious, due to lower runoff and/or
more recent functioning). Between this depth down to more
than 25 m (the maximum drilling depth, at hole 1), the vadose
zone displays lower solute contents, as reported in DC
modelling (Fig. 4). A calculation of model uncertainty (not
shown here) using RES2DINV software displays an uncertainty percentage ranging between 20% and 30% below the
depth of 20 m. This uncertainty remains probably underestimated: for the drill hole 1, the inversion displays a value of
900 V m, while the resistivity logging displays a value of
1250 V m, indicating a 38% difference. However, those
uncertainties remain relatively low and it can be concluded
that the 2D electrical imaging provides a reliable estimate of
the bulk conductivity down to 30 m.
5.2. Dynamics of deep infiltration
One of the main challenges when dealing with groundwater in semiarid areas is to determine the main process in
play for deep infiltration and eventually groundwater
recharge (Simmers, 2003). The results from this study,
using sub-surface geophysics and vadose zone chemistry,
confirm previous conclusions obtained with other methods
in southwestern Niger: deep infiltration and groundwater
recharge follow an indirect process, occurring only where
surface runoff concentrates (Leduc et al., 1997; Desconnets
et al., 1997; Favreau et al., 2002). For the studied fan area,
hydrological modelling shows that runoff and deep infiltration are largely discontinuous, both at an intra-seasonal and
inter-annual scale (Fig. 8); annual runoff and deep infiltration vary by about one order of magnitude for the
investigated decade (Table 1). This result is consistent with
previous studies (e.g. Cappelaere et al., 2003) that showed
that runoff is more dependent on rainfall events distribution
and magnitude than on annual rainfall amount.
Next to the study area, infiltration capacity of sandy
gullies was reported in Peugeot et al. (2003) at 450 mm h 1.
Considering the 1 ha surface of sandy channels in the
alluvial fan, the infiltration capacity could reach 4500
m3 h 1 and therefore easily infiltrate the mean runoff event
of 1600 m3 computed at point D for the studied decade.
A changing pattern of deep infiltration has also to be
considered for the fan area, considering its long-term
dynamics. Following land clearance for the last decades, a
115
general runoff increase by a factor close to three has been
computed at the catchment scale (Seguis et al., 2004). This
increase in runoff has led to an upslope shifting of the D
point (Fig. 1) due to the progradation of sandy deposits.
Aerial photographs from 1950, 1992 and 1998 show that it
moved westwards by about 143 m between 1950 and 1992,
and of 79 m between 1992 and 1998. In Fig. 3c, the large,
low conductive area, interpreted as being the most leached
zone of the fan, appears to be located downslope of the
densely braided gully zone, where the most active infiltration is supposed to occur. Considering the westwards
movement of the fan for the last decades, the downslope
location of the most leached zone within the study area can
be interpreted as an integrated result of past leaching and
deep infiltration in the downward part of the fan.
5.3. Solute content of the vadose zone
Chemical analyses of the vadose zone solute contents
were performed in order to decipher their possible origin. A
comparison with the dry and wet deposition reported for the
area (Ca and N dominated; Drees et al., 1993; Freydier et
al., 1998; Galy-Lacaux and Modi, 1998) show that the
chemical composition of the most mineralized part of the
vadose zone (Ca, Na and NO3 dominate, in various
proportions) could only partly be explained by a simple
rainfall infiltration– re-evaporation process. On the other
hand, the matrix mineralogy is mostly made of quartz and
kaolinite and its incongruent dissolution could not lead to
the observed vadose zone chemistry. Considering that all of
the solute content stored in the vadose zone originates from
atmospheric deposits (dust deposits and rainfall events),
calculations based on published inputs (Drees et al., 1993;
Freydier et al., 1998; Galy-Lacaux and Modi, 1998) show
large discrepancies for the time scale required for accumulation, depending on the element considered. For instance,
for the most mineralized part of drill hole 2 (the vadose zone
between 5 and 11 m, representing 75% of the solute content
of the profile; Fig. 6), the equivalent time scale for the
accumulated solute content would range from about 100
years for Cl, up to 1200 years for Na (marine constituents),
while of about 300 years for Ca (terrigenous constituent).
Obviously, other sources and processes may be involved.
Within the study region, in cultivated areas and fallows
with the same dominant shrub species (G. senegalensis)
Wezel et al. (2000) described an important small scale
variability of the chemical properties of the top 0.10 m of
the soil; they showed that the chemical composition of the
shrub litter seems to influence the degree of soil enrichment.
In southwestern Niger, another possible source of nutrients
lies in the nitrogen fixing process, either by leguminous
woody plants (Acacia sp.) or by microbial crusts at the soil
surface (Malam Issa et al., 2001). All of these sub-surface
processes can contribute to the complex, nitrogen-rich
solute content observed at depth within the unsaturated
zone. A detailed study of the deep unsaturated zone, that
116
could include isotope analysis for the biogenic constituents
(15N – NO3, 14C/13C of organic C) or transient neutron probe
measurements would be necessary to determine whether
processes having led to this deep accumulation of solute are
still active (e.g. by occasional deep infiltration followed by
transpiration through deep rooting) or represent paleoconditions dating back to the humid periods of the late
Quaternary. Though deep rooting cannot be ruled out, most
studies have shown that G. senegalensis mostly extract
water from the top two metres of the soil (Brunel et al.,
1997; Gaze et al., 1998). The well known deep rooting
Faidherbia albida is also present on the site but its today’s
density is too low to explain the observed high solute
content within the deep unsaturated zone. Further analyses
are obviously needed to better interpret the vertical
distribution and solute fluxes within the deep vadose zone.
6. Conclusion
This local scale study of an alluvial fan in southwestern
Niger combines sub-surface geophysics, vadose zone
analysis and hydrological modelling. Two main conclusions
can be outlined:
(1) Channels in the alluvial fan act as preferential pathways for deep infiltration. By exploring the deep part
of the unsaturated zone, our results confirm the
occurrence of leaching down to 10 m below sandy
channels. On the basis of hydrological modelling at
the catchment scale for the decade 1992 – 2002,
computations show that infiltration through the fan
range from 1000 to 24 000 m3, i.e. between 5% and
16% of surface water reaching the final outlet of the
basin, an endoreic pond where recharge to the aquifer
occurs annually. In the study area, deep infiltration
and eventually groundwater recharge was reported to
occur only through endoreic ponds, where surface
runoff concentrates (Desconnets et al., 1997; MartinRosales and Leduc, 2003). This study demonstrates
that deep infiltration can also occur episodically
through alluvial fans on sandy slopes, thus representing additional potential sites for groundwater recharge. This result confirms previous hydrological
investigation in nearby catchments that showed
important surface water losses through sandy gullies
for intense runoff events (Peugeot, 1995; Esteves and
Lapetite, 2003). However, our conclusion differs from
a similar study in Burkina – Faso (granitic context with
very clayey regolith), where surface water was
reported to infiltrate not deeper than 0.80 m below
main gullies (Descloitres et al., 2003). As outlined by
Poesen et al. (2003), further studies are needed to
better understand how gullies interact with hydrological processes and to determine their importance in
hydrological balances.
(2) Next to recharge areas, there is a continuous layer,
approximately located between 5 and 10 m below the
soil surface, where the vadose zone displays high
solute contents. This second conclusion is of much
interest for the hydrological and geochemical balance
of soil studies. To the best of our knowledge, the
presence of a (quasi) continuous mineralized soil layer
at depth between about 5 and 10 m below the soil
surface was unknown in the area. Buerkert and
Hiernaux (1998) have emphasized the complex pattern
of nutrient transfers in the West African Sahelian zone.
Considering the possibility for some Sahelian trees to
reach several ten metres below the soil surface (e.g.
Faidherbia albida; Canadell et al., 1996) there is
obviously the need to take into account a deeper part
of the vadose zone to balance hydrological and
nutrient cycles for the Sahelian biome.
For groundwater recharge and salinity, the existence of a
nitrate-rich layer at depth within the vadose zone appears as
a key information to explain some observed changes with
time. In southwestern Niger, some seasonal and long-term
changes in groundwater chemistry have been observed near
infiltrating ponds (Elbaz-Poulichet et al., 2002); these
changes have been explained by seasonal recharge and
leaching of the thick unsaturated zone. Our results, by
identifying an important source of solute for the hydrological cycle, confirm and clarify this interpretation. In
particular, some important increases in nitrate content that
occurred during exceptional recharge events, at distance
from any usual source of pollution (Favreau et al., 2003),
could be explained by leaching of nitrate-rich layers of the
vadose zone by massive infiltration of surface water.
From a methodological point of view, the absence of any
relationship between chloride and bulk mineralization is
another puzzling observation. In semiarid areas, the
Chloride Mass Balance (CMB) method has been widely
used to infer groundwater recharge rates, assuming that the
Cl content closely represents the bulk salinity of the vadose
zone under piston-flow recharge process (e.g. Bromley et
al., 1997). However, in our study area, considering deep
infiltration and groundwater recharge as a steady pistonflow process is probably not relevant. As for soil studies, a
better description of the deep unsaturated zone appears as a
basic prerequisite for groundwater recharge studies in
semiarid areas.
This study has shown the importance of combining
various methods to obtain reliable results on deep infiltration through a thick unsaturated zone. In our zone, a simple
relation between soil solution conductivity (deduced from
soils samples) and an apparent electrical conductivity
measured by geophysics has been evidenced. As outlined
in other semiarid areas (Cook et al., 1989; Scanlon et al.,
1999a,b) apparent electrical conductivity mapping used to
delineate changes in recharge rates and process appears as a
powerful method that should be used more systematically
for groundwater recharge studies. When adding more
sophisticated geophysical tools such as 2D electrical
imaging or vadose zone electrical logging, quantification
between electrical conductivity and other pertinent parameters becomes a definite advantage to better understand the
processes of deep infiltration and groundwater recharge.
Acknowledgements
This study was funded by IRD and partly by a PhD
grant from the University of Montpellier II. O. Ribolzi, H.
Robain, J. Touma, L. Barbiéro and L. Ruiz (IRD) are
thanked for helpful discussions that improved the data
interpretation. The collaboration of Sandrine Caquineau
(IRD-Bondy), Monique Oı̈ (Hydrosciences Montpellier),
François Monat and Abdoulaye Koné (IRD, Niamey) and
of the DRE in Niger (Direction of Hydraulic Resources,
Ministry of Water Resources, Niamey) are warmly
acknowledged.
References
Archie, G.E., 1942. The electrical resistivity log as an aid in determining
some reservoir characteristics. Am. Inst. Min., Metall. Pet. Eng. Tech.,
paper 1422.
Beauvais, A., Ritz, M., Parisot, J.C., Bantsimba, C., Dukhan, M., 2004.
Combined ERT and GPR methods for investigating two-stepped
lateritic weathering systems. Geoderma 119, 121 – 132.
Bielders, C.L., Michels, K., Rajot, J.L., 2000. On-farm evaluation of
ridging and residue management practices to reduce wind erosion in
Niger. Soil Sci. Soc. Am. J. 64, 1776 – 1785.
Bromley, J., Edmunds, W.M., Fellman, E., Brouwer, J., Gaze, S.R., Sudlow,
J., Taupin, J.D., 1997. Estimation of rainfall inputs and direct recharge
to the deep unsaturated zone of southern Niger using the chloride profile
method. J. Hydrol. 188 – 189, 139 – 154.
Brunel, J.P., Walker, G.R., Dighton, J.C., Monteny, B., 1997. Use of stable
isotopes of water to determine the origin of water used by the vegetation
and to partition evapotranspiration, a case study from HAPEX-Sahel. J.
Hydrol. 188 – 189, 466 – 481.
Buerkert, A., Hiernaux, P., 1998. Nutrients in the west African Sudano –
Sahelian zone: losses, transfers and role of external inputs. Z.
Pflanzenernähr. Bodenk. 161, 365 – 383.
Canadell, J., Jackson, R.B., Ehleringer, J.R., Mooney, H.A., Sala, O.E.,
Schulze, E.D., 1996. Maximum rooting depth of vegetation types at the
global scale. Oecologia 108, 583 – 595.
Cappelaere, B., Vieux, B.E., Peugeot, C., Maia, A., Seguis, L., 2003.
Hydrologic process simulation of a semiarid, endoreic catchment in
Sahelian west Niger, Africa: 2. Model calibration and uncertainty
characterization. J. Hydrol. 279, 244 – 261.
Cook, P.G., Walker, G.R., Jolly, I.D., 1989. Spatial variability of
groundwater recharge in a semiarid region. J. Hydrol. 111, 195 – 212.
Descloitres, M., Ribolzi, O., Le Troquer, Y., 2003. Study of infiltration in a
Sahelian gully erosion area using time-lapse resistivity mapping. Catena
53, 229 – 253.
Descloitres, M., Le Troquer, Y., 2004. Sonde de diagraphie électrique pour
la mesure de la résistivité sur la paroi d’un forage. French Patent n2845 416, INPI. Bulletin Officiel de la Propriété Industrielle.
Desconnets, J.-C., Taupin, J.-D., Lebel, T., Leduc, C., 1997. Hydrology of
the HAPEX-Sahel Central Super Site: surface water drainage and aquifer
recharge through the pool systems. J. Hydrol. 188 – 189, 155 – 178.
117
D’Herbes, J.-M., Valentin, C., 1997. Land surface conditions of the Niamey
region: ecological and hydrological implications. J. Hydrol. 188 – 189,
18 – 42.
Drees, L.R., Manu, A., Wilding, L.P., 1993. Characteristics of aeolian dusts
in Niger, west Africa. Geoderma 59, 213 – 233.
Edmunds, W.M., Gaye, C.B., Fontes, J.C., 1991. A record of climatic and
environmental change contained in interstitial waters from the unsaturated zone of northern Senegal. Isotope Techniques in Water Resources
Development. IAEA Proceedings Series, Vienna, pp. 533 – 549.
Elbaz-Poulichet, F., Favreau, G., Leduc, C., Seidel, J.-L., 2002. Major ion
chemistry of groundwaters in the continental terminal water table of
southwestern Niger (Africa). Appl. Geochem. 17, 1343 – 1349.
Esteves, M., Lapetite, J.-M., 2003. A multi-scale approach of runoff
generation in a Sahelian gully catchment: a case study in Niger. Catena
50, 255 – 271.
Favreau, G., Leduc, C., Marlin, C., Dray, M., Taupin, J.-D., Massault, M.,
Le Gal La Salle, C., Babic, M., 2002. Estimate of recharge of a rising
water table in semiarid Niger from 3H and 14C modeling. Ground Water
40, 144 – 151.
Favreau, G., Leduc, C., Seidel, J.-L., Ousmane, S.D., Mariotti, A., 2003.
Land clearance and nitrate-rich groundwater in a Sahelian aquifer,
Niger. In: Servat, E., Najem, W., Leduc, C., Ahmed, S. (Eds.), Niger.
Int. Conf. Hydrology of Mediterranean and Semiarid Regions,
Montpellier, France, 2003, IAHS Publ., vol. 278, pp. 163 – 167.
Freydier, R., Dupré, B., Lacaux, J.P., 1998. Precipitation chemistry in
intertropical Africa. Atmos. Environ. 32, 749 – 765.
Galle, S., Ehrmann, M., Peugeot, C., 1999. Water balance in a banded
vegetation pattern. A case study of tiger bush in western Niger. Catena
37, 197 – 216.
Galy-Lacaux, C., Modi, A.I., 1998. Precipitation chemistry in the Sahelian
savanna of Niger, Africa. J. Atmos. Chem. 30, 319 – 343.
Gaze, S.R., Brouwer, J., Simmonds, L.P., Bromley, J., 1998. Dry season
water use patterns under Guiera senegalensis L. shrubs in a tropical
savanna. J. Arid Environ. 40, 53 – 67.
Guérin, R., Descloitres, M., Coudrain, A., Talbi, A., Gallaire, R., 2001.
Geophysical surveys for identifying saline groundwater in the semiarid region of the central Altiplano, Bolivia. Hydrol. Process. 15,
3287 – 3301.
Hamblin, A.P., 1981. Filter-paper method for routine measurement of field
water potential. J. Hydrol. 53, 355 – 360.
Keller, G.V., 1988. Rock and mineral properties. In: Nabighian, M. (Ed.),
Electromagnetic Methods in Applied Geophysics, vol. 1. Society of
Exploration Geophysicists.
Lange, J., Leibundgut, C., Greenbaum, N., Schik, A.P., 1999. A noncalibrated rainfall-runoff model for large, arid catchments. Water
Resour. Res. 35, 2161 – 2172.
Leduc, C., Schroeter, P., Bromley, J., 1997. Water table fluctuation and
recharge in semi-arid climate: some results of the HAPEX-Sahel
hydrodynamic survey (Niger). J. Hydrol. 188 – 189, 123 – 138.
Leduc, C., Favreau, G., Schroeter, P., 2001. Long-term rise in a Sahelian
water-table: the continental terminal in south-west Niger. J. Hydrol.
243, 43 – 54.
Loke, M.H., 2000. Electrical imaging surveys for environmental and
engineering studies. Technical Note of Geotomo Software. available on
line at www.geoelectrical.com.
Malam Issa, O., Stal, L.J., Défarge, C., Couté, A., Trichet, J., 2001.
Nitrogen fixation by microbial crusts from dessicated Sahelian soils
(Niger). Soil Biol. Biochem. 33, 1425 – 1428.
Martin-Rosales, W., Leduc, C., 2003. Dynamics of emptying of a temporary
pond in the Sahel: the case study of Banizoumbou (southwestern
Niger). C. R. Geosci. 335, 461 – 468.
Mc Neill, D.J., 1980. Electromagnetic terrain conductivity measurement at
low induction number. Technical note TN 6 available from Geonics
Limited, 1745 Meyerside Drive, Unit 8, Mississauga, Ontario, Canada.
Nagumo, F., 1992. Pedological environment and agro-ecological system of
the Sudano-Sahelian zone, in Niger, West Africa. Hokkaı̈do University,
Japan. (101 pp.)
118
Peugeot, C., 1995 Influence de l’encroûtement superficiel du sol sur le
fonctionnement hydrologique d’un versant sahélien (Niger). Expérimentations in-situ et modélisation. Thèse de doctorat, Université Joseph
Fourrier, Grenoble. 356 pp.
Peugeot, C., Esteves, M., Galle, S., Rajot, J.-L., Vandervaere, J.P., 1997.
Runoff generation processes: results and analysis of field data collected
at the East Central Supersite of the HAPEX-Sahel experiment. J.
Hydrol. 188 – 189, 179 – 202.
Peugeot, C., Cappelaere, B., Vieux, B.E., Seguis, L., Maia, A., 2003.
Hydrologic process simulation of a semiarid, endoreic catchment in
Sahelian west Niger, Africa: 1. Model-aided data analysis and
screening. J. Hydrol. 279, 224 – 243.
Poesen, J., Nachtergaele, J., Verstraeten, G., Valentin, C., 2003. Gully
erosion and environmental change: importance and research needs.
Catena 50, 91 – 133.
Scanlon, B.R., Langford, R.P., Goldsmith, R.S. 1999a. Relationship
between geomorphic settings and unsaturated flow in an arid setting.
Water Resour. Res. 35, 983 – 999.
Scanlon, B.R., Paine, J.G., Goldsmith, R.S. 1999b. Evaluation of
electromagnetic induction as a reconnaissance technique to characterize
unsaturated flow in an arid setting. Ground Water 37, 296 – 304.
Seguis, L., Cappelaere, B., Milesi, G., Peugeot, C., Massuel, S., Favreau,
G., 2004. Simulated impacts of climate change and land-clearing
on runoff from a small Sahelian catchment. Hydrol. Process. 18,
3401 – 3413.
Simmers, I., 2003. Understanding water in a dry environment. Hydrological
Processes in Arid and Semi-Arid Zones, IAH Publ., vol. 23. Balkema
Publishers, the Netherlands.
Vieux, B.E., Gaur, N., 1994. Finite-element modeling of storm water runoff
using Grass GIS. Microcomput. Civ. Eng. 9 (4), 263 – 270.
Wezel, A., Rajot, J.L., Herbrig, C., 2000. Influence of shrubs on soil
characteristics and their function in Sahelian agro-ecosystems in semiarid Niger. J. Arid Environ. 44, 383 – 398.
Wood, W.W., Sanford, W.E., 1995. Chemical and isotopic methods for
quantifying ground-water recharge in a regional, semiarid environment.
Ground Water 33, 458 – 468.
Catena 70 (2007) 313 – 329
Using a structural approach to identify relationships between soil
and erosion in a semi-humid forested area, South India
Laurent Barbiéro a,b,⁎, Harshad R. Parate b , Marc Descloitres c,b , Adelphe Bost b ,
Sônia Furian d,b , M.S. Mohan Kumar b , C. Kumar b , Jean-Jacques Braun a,b
a
IRD, LMTG-OMP, UMR 5563, Laboratoire des Mécanismes de Transfert en Géologie, 14 Av. E. Belin, F-31400 Toulouse, France
b
Indo-French Cell for Water Sciences (IRD-IISc Joint Laboratory), Indian Institute of Science, 560 012, Bangalore, India
c
IRD, UR012-LTHE, UMR/CNRS-IRD-INPG-UJF, B.P. 53, 38041 Grenoble Cedex 9, France
d
Dep. Geografia, Avenida Prof. Lineu Prestes, 338, CEP 05508-000, Universidade de São Paulo-SP, Brazil
Received 5 April 2006; received in revised form 22 October 2006; accepted 27 October 2006
Abstract
Biogeochemical and hydrological cycles are currently studied on a small experimental forested watershed (4.5 km2) in the semi-humid
South India. This paper presents one of the first data referring to the distribution and dynamics of a widespread red soil (Ferralsols and
Chromic Luvisols) and black soil (Vertisols and Vertic intergrades) cover, and its possible relationship with the recent development of the
erosion process. The soil map was established from the observation of isolated soil profiles and toposequences, and surveys of soil
electromagnetic conductivity (EM31, Geonics Ltd), lithology and vegetation. The distribution of the different parts of the soil cover in
relation to each other was used to establish the dynamics and chronological order of formation. Results indicate that both topography and
lithology (gneiss and amphibolite) have influenced the distribution of the soils. At the downslope, the following parts of the soil covers were
distinguished: i) red soil system, ii) black soil system, iii) bleached horizon at the top of the black soil and iv) bleached sandy saprolite at the
base of the black soil. The red soil is currently transforming into black soil and the transformation front is moving upslope. In the bottom part
of the slope, the chronology appears to be the following: black soil N bleached horizon at the top of the black soil N streambed N bleached
horizon below the black soil. It appears that the development of the drainage network is a recent process, which was guided by the presence
of thin black soil with a vertic horizon less than 2 m deep. Three distinctive types of erosional landforms have been identified:
1. rotational slips (Type 1);
2. a seepage erosion (Type 2) at the top of the black soil profile;
3. A combination of earthflow and sliding in the non-cohesive saprolite of the gneiss occurs at midslope (Type 3).
Types 1 and 2 erosion are mainly occurring downslope and are always located at the intersection between the streambed and the red soilblack soil contact. Neutron probe monitoring, along an area vulnerable to erosion types 1 and 2, indicates that rotational slips are caused by a
temporary watertable at the base of the black soil and within the sandy bleached saprolite, which behaves as a plane of weakness. The
watertable is induced by the ephemeral watercourse. Erosion type 2 is caused by seepage of a perched watertable, which occurs after swelling
and closing of the cracks of the vertic clay horizon and within a light textured and bleached horizon at the top of black soil.
Type 3 erosion is not related to the red soil–black soil system but is caused by the seasonal seepage of saturated throughflow in the sandy
saprolite of the gneiss occurring at midslope.
© 2006 Elsevier B.V. All rights reserved.
Keywords: Erosion; Structural analysis; Electromagnetic induction; Chromic Luvisol; Vertisol; South India
⁎ Corresponding author. Indo-French Cell for Water Sciences (IRD-IISc Joint Laboratory), Indian Institute of Science, 560 012, Bangalore, India.
E-mail address: barbiero@civil.iisc.ernet.in (L. Barbiéro).
0341-8162/$ - see front matter © 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.catena.2006.10.013
314
L. Barbiéro et al. / Catena 70 (2007) 313–329
1. Introduction
In recent years, several studies are carried out to understand the bio-geochemical cycles of major and trace elements, to calculate the mechanical erosion and chemical
weathering rates, to estimate the role of the major parameters
(relief, climate, lithology, vegetation, anthropisation…) that
are likely to control the chemical weathering processes, to
quantify the effects of rock chemical weathering on the
carbon cycle and to find its potential role on climate changes
(Gaillardet et al., 1997, 1999; Oliva et al., 2003). The integrated study of small watersheds is one of the best ways to
provide direct and accurate information for the analysis of
ecosystems. Although this integrated ecosystem approach is
common for the temperate zone, it has not yet been widely
applied to the tropics and especially under semi-humid
climate (White et al., 1998; Braun et al., 2005). Such a study
is currently developed on small experimental watersheds
in South India, namely, surface and groundwater flow
(Descloitres et al., in press), regolith thickness and chemical
weathering, physical erosion, dynamic of the soil cover and
including interactions among the aforementioned aspects
(Braun et al., 2006). Although it is widely known today that
internal transformation of a soil cover can influence or even
govern landscape evolution through intensification or
decrease of physical erosion (whatever could be the parental
material or topographical gradient, Boulet et al., 1977;
Planchon et al., 1987; Filizola and Boulet, 1996; Barbiéro
et al., 1998; Furian et al., 1999), the soil cover itself is still a
black box in many studies on soil erosion. Moreover, very
little is known about natural erosion in forested areas in
South India where human activity is minimal, although
features of erosion have been observed. The aim of this
study is to identify the dynamic and to present the main
factors intrinsic to the soil cover that govern natural erosion on a forest area with a widespread red and black soil
system, by using a structural and a spatially distributed
approach.
2. Site
In South India, the Western Ghats parallel to the western
coast of the peninsula form an orographic barrier, inducing
an important climatic gradient, with annual rainfall decreasing progressively from about 5000 mm in the west, to less
than 750 mm just 80 km to the east (Pascal, 1982; Fig. 1).
The climatic sequence (climosequence) is associated with
changes in landscape geomorphology from convex hills
intermittent with flat floors to long concave glacis (Gunnell
and Bourgeon, 1997; Gunnell, 2000). In connection with the
geomorphological changes, the soil types (FAO-ISRICISSS, 1998) range from Ferralsols to thin red soils (Chromic
Luvisols) associated with black soils (Vertisol, Vertic
intergrades) in the climatic semi-humid transition area, and
in the semi-arid area we find Calcic Luvisol and Calcic
Vertisol (Murthy et al., 1982; Pal and Deshpande, 1987;
Bourgeon, 1991; Jacks and Sharma, 1995; Gunnell, 2000).
This association of red soils (Luvisols) and black soil
(Vertisols) is widespread on the semi-humid to semi-arid area
of the Deccan plateau (Bourgeon, 1991). The passage in the
clay mineralogy from the kaolinite-dominated humid area to
the smectite-dominated semi-arid area is achieved progressively via an intermediate area with 2:1 K clay such as illite
and sericite (Bourgeon and Pedro, 1992).
Fig. 1. Climatic gradient on the backslope of the Western Ghats (black lines are isohyets), main river course and location of the Mulehole studied site in southern
India (modified from Gunnell and Bourgeon, 1997).
Red soils occurring in the current seasonal semi-humid to
semi-arid conditions (b1500 mm annual rainfall and high
evapotranspiration) have been considered as Paleosols or
relict soils (non-buried Paleosols) that formed in an earlier
period with a moister climate than the present, but this
assertion is still under debate. Some authors consider that the
climatic conditions are not conducive to the soil-forming
processes of red soils such as deep weathering and kaolinite
formation (Bronger and Bruhn, 1989; Brückner and Bruhn,
1992). However, Gunnell and Bourgeon (1997) emphasize
that the presence of clay minerals yielding an X-ray diffraction peak at 7 Å in dry climatic zone does not necessarily
mean that these are inherited from a Paleosol formed in more
humid periods in the past. They suggested further analysis,
and in particular the comprehensive down-profile consideration of the entire spectrum of minerals, before reaching a
conclusion. In the prevailing semi-arid conditions (b 900 mm
annual rainfall), secondary carbonate is currently accumulating in the saprolite and lower parts of the red soil horizons
(Bronger et al., 2000). Micromorphological studies of the
calcrete in the semi-arid area reveal a multistage origin
(Durand et al., 2006) and recent dating of calcrete nodules
suggests fairly stable climatic conditions at the ≥ 200 ky time
scale (Durand et al., in press).
The climatic and pedoclimatic conditions are decisive in
the formation of red soils (Bourgeon, 1991), whereas the
formation of black soils depends mainly on the slow down of
the solution and lack of drainage usually in bottom part of the
landscape. In black soils, the presence of smectite clay
minerals causes appreciable shrink–swell, which induces
formation of cracks and distinctive structural elements such
as wedge-shaped peds with smooth or slickensided surfaces
(Bourgeon, 1991).
315
Field work was carried out on a 4.5 km2 watershed
located in Bandipur National Park, close to the Mulehole
check post at 11° 44' N and 76° 27' E (Karnataka state,
Chamrajnagar district). The watershed area is mostly
undulating with gentle slopes and the elevation of the
watershed ranges from 820 to 910 m above sea level.
Because it belonged to the hunting reserve of the Maharaja of
Mysore, the region has been preserved from agricultural
activity at least since the 17th century. Later it was incorporated into the Bandipur National Park, and today the
only human activities are limited to surveillance by the
rangers of the Forest Department.
The studied site is located in the climatic semi-humid
transition area (Fig. 1) and the mean annual rainfall
(n = 20 years) is 1120 mm. The climate is characterized by
recurrent but non-periodic droughts, depending on monsoon
flows. The mean yearly temperature is around 27 °C.
Streams are temporary flowing for a few hours to a few days
after the stormy events of the rainy season. Rainfall and
runoff measured at the outlet of the Mulehole watershed
were respectively 431 mm and 1 mm in 2003, 1216 mm and
59 mm in 2004, and 1434 mm and 181 mm in 2005.
The substratum belongs stratigraphically to the Precambrian Dharwar supergroup (Moyen et al., 2001) and consists
of gneiss with amphibolites and quartz dykes. The mean
strike value is N80°, with a dip angle ranging from 75° to the
vertical. The vegetation consists of dry deciduous forest
(Pascal, 1982; Agarwala, 1985; Pascal, 1986) where 4 different types of vegetation have been identified: 1 — a forest
with vegetation mainly dominated by three species, namely
Anogeissus latifolia, Terminalia alata and Tectona grandis,
called ‘ATT facies’. 2 — a vegetation called ‘Shorea facies’
characterized by the presence of Shorea roburghii and
Fig. 2. The studied watershed of Mulehole, topography (in metre), streams, North–South ECm measurement transects (……) and soil sequences T1 and T2.
316
Lagerstroemia microcarpa. 3 — the ‘Swamp facies’
consisting of grass-covered glades with scattered trees
(Ceristoides turgida). 4 — the discontinuous ‘riverine
facies’ along the talwegs characterized by the presence of
Syzygium cumini, Mangifera indica, Ficus recemosa and
Derris indica. The first 3 above-mentioned vegetation
‘facies’ have been identified in the Mulehole watershed,
whereas the fourth one was not clearly developed and/or
occupied very small area.
3. Method
3.1. Study at the watershed scale
A contour Digital Elevation Model (DEM) was generated
from 2780 topographical measurements on the area (Fig. 2),
with higher density close to the talwegs and to the main
topographical changes. An exhaustive GPS georeferenced
survey of the streambeds and the soil erosion pattern was
carried out for the entire watershed. The eroded soil volume
was roughly estimated by measuring the length, width and
thickness of each eroded area in relation to the surrounding
non-eroded area.
Electromagnetic induction is becoming widespread for
soil survey in general, although it has been used mainly for
the monitoring of spatial and temporal changes in soil
salinity (Corwin et al., 2006). Preliminary studies on the
watershed have shown that red and black soils have a different apparent electrical conductivity (EC). Therefore, soil
distribution was first attempted by conducting an electromagnetic conductivity survey, using an EM31 portable device (Geonics Ltd, Ontario, Canada). The device measures
an apparent conductivity (ECm values) in milliSiemens per
metre (mS/m). The EM31 has a fixed 3.66 m space between
the coils (transmitter and receiver) and the measurements
were carried in the vertical dipole configuration, which affords an investigation depth of about 4 to 6 m (McNeill,
1980). The measurements were carried out along 31 North–
South oriented transects and with a space of 100 m between
the transects (Fig. 2). The measurement points were taken
and stored automatically by a data logger every 5 s. A Global
Positioning System (GPS) was coupled with the EM31
device in order to get the geographical coordinates of each
measurement point (Cannon et al., 1994). The survey was
carried out in January 2004, i.e. in the middle of the dry
season.
The ECm data underwent a geostatistical treatment before
kriging. Duplicates were removed from the data set before
treatment, on the basis of a 2 m tolerance in the X and Y
directions. These duplicates are due to local difficulties in
progressing through the vegetation or in crossing obstacles
(topographic accidents) during the survey. A chi-squared test
showed that the data might not be assumed to have a normal
distribution. Therefore, the calculation was performed on a
theoretical distribution of the data by lognormal transformation as recommended by Dowd (1982),
zðxi Þ ¼ lnðsðxi ÞÞ
ð1Þ
where s(xi) is the ECm data at xi, z(xi) is the log-transformed
data. An estimate of the sample variogram is given by the
formula:
N ðhÞ
gðhÞ ¼
1 X
ðzðxi Þ−zðxi þ hÞÞ2
2N ðhÞ i¼1
ð2Þ
Where N(h) is the number of pairs of points and z(xi) and
z(xi+h) are the logarithms of the ECm values at xi and xi+h.
Fig. 3. Plan view of the toposequence T1, red and black soils distribution, streambed and neutron probe access holes.
Raw and directional variogram were calculated to detect a
possible anisotropy in the field. The kriged map is built from
a model of variogram fitted to the sample variogram.
In order to relate the ECm map to the soil distribution,
forward modelling was carried out using the PCloop
software (Geonics, Ltd) and was based on (i) the theoretical
response of the EM31 over an horizontally layered medium
(McNeill, 1980); (ii) resistivity measurements of representative horizons in red and black soils along soil profile; and
(iii) resistivity logging down auger holes drilled into
representative soils (Ferralsols, Luvisols and Vertisols).
A geological survey was carried out from the identification of about 300 georeferenced rock outcrops within the
watershed and the extrapolation of the data was carried out
using the ECm survey.
The development of vegetation depends narrowly on the
hydric regime of the soil, and consequently on the type and
thickness of soil. Therefore, a survey of the three main types
of vegetation, namely the ATT, Shorea and Swamp facies
was carried out across the watershed in order to extrapolate
the soil data.
3.2. Comprehensive study along soil sequences
Two soil sequences were studied on the watershed (Fig. 2).
An existing spoon-shaped erosional landform (rotational slip
type) was targeted for the excavation of a 80 m long trench
(T1) in order to understand the soil morphology of the
portions of landscape vulnerable to soil erosion (Figs. 3 and
4A). The second soil sequence (T2) located where the
streambed appears to be currently incising (just a few metres
upstream from the current incision, Fig. 4E), was studied in
order to understand if any specific morphology of the soil
cover could favour the development of the talweg.
The soil pattern was studied in detail, emphasizing the
geometrical relationships between the different horizons
identified from basic field observations (colour, texture,
structure, porosity, presence of coarse elements, intensity of
biological activity…). The procedure follows routine techniques developed by Boulet et al. (1982) and Fritsch et al.
(1992).
In the first step, 2D electrical imaging was performed
around toposequence T1 to locate the red soil/black soil
contact and to characterize its morphology. Five boreholes
were drilled down to the saprolite in the red soil, black soil
and transition area, along a line parallel located at a distance
of 5 m from T1 (Fig. 3). The different horizons of the
boreholes were identified and compared to the description
from the trench T1 in order to map the layout of the red soil–
black soil system. Soil moisture was monitored during the
rainy seasons 2004 and 2005 through neutron probe measurements (soil moisture probe type I.H. II, Didcot Instrument Co. Ltd., Abingdon Oxon, England). Measurements
were carried out at every 10 cm, every 15 days, and daily
during heavy rainy periods. Between two successive neutron
probe measurement periods, the boreholes were clogged
317
with inflatable rubber tubes inserted into the holes in order to
prevent any infiltration from the topsoil runoff or along the
hole during rainfall.
For each horizon, a relationship was established between
the neutron probe measurements and the volumetric water
content. For this purpose, bulk samples were collected while
drilling holes and immediately placed and sealed in metallic
boxes. Simultaneously the neutron probe measurements
were carried out at the corresponding depth. Gravimetric
water content was determined in the laboratory by weighing
the samples before and after oven drying for 24 h at 105 °C.
This procedure was carried out at the end of the dry season
(dry state) in the monitored holes and at the end of the
monsoon season (wet state) in new holes drilled at about
50 cm away from the previous ones. The bulk density was
measured using the paraffin method on aggregates collected
along the trench T1 in each horizon (Singer, 1986). The
volumetric water content in the boreholes was estimated by
multiplying the gravimetric water content by the bulk density
of the corresponding horizon.
The calibration was established for each horizon using
linear regression between the volumetric water content
against the R/Rw ratio, where R denotes the number of
counts per second of neutron probe in soil and Rw denotes
the number of counts per second in a water standard. The
calibration was established on a volumetric water content
range of 16 to 31% in red soil (horizons 2, 3, 4 and 5), 21 to
35% in black soil (horizons 7, 8, 9, 10 and 11), 12 to 30% in
organic topsoil horizon (6), and 13% to 29% in the saprolite
(horizons 1 and 13).
4. Results
4.1. ECm measurements and ECm survey
From the 10,935 ECm measurement points, 439 duplicates were discarded before treatment. The conductivity
values range between 0.1 and 52 mS/m, with an average
value of 7.21 mS/m, and a standard deviation of 7.06 mS/m.
The variation coefficient (0.98) indicates a low dispersion of
the data around this average value.
The experimental variogram built from log-transformed
ECm data is presented in Fig. 5. A slight anisotropy is
detected by comparing the raw and directional variogram,
showing a higher dependence of the ECm values in the
direction N63.6° (East-Northeast/West-Southwest). This
anisotropy was taken into account for the kriging computation. In N63.6° direction the experimental variogram shows a
nugget effect of almost zero, a range of about 300 m and the
scale of 0.105 (log value), and therefore it fits better with an
exponential model with the following characteristics:
Scale = 0.105; range = 300 m.
The kriged map presented on Fig. 6 shows that the soil
electromagnetic conductivity is not distributed regularly
throughout the watershed. High conductivity values are
located in the flat bottom part of the watershed and on some
318
Fig. 4. A) A rotational slip erosional landform (erosion type 1) was targeted for the excavation of an 80 m long trench (T1); B) seepage erosion (type 2) at the top
of black soil profile; C) midslope erosional landform (type 3) resulting from the combination of seepage and mass movements into the non-cohesive material of
the gneiss saprolite; D) rotational slip, the material has been partially evacuated; E) linear depression where the stream is currently incising; F) detail of the profile
in the depression of T2 showing the vertic structure (defined by vertical and sub-horizontal cracks indicated by arrows) preserved in the sandy material due to
silica cementation (knife is about 25 cm); G) roots of Tectona grandis crossing the streambed at about 1 m high from the bottom, indicating recent incision.
319
where, namely in the valley bottom, along the slope or on the
crest line. Paragneiss is dominant and consist mainly of
quartz, feldspar (plagioclase and potassic) with a low
quantity of biotite. Bedrock exposures are usually poorly
weathered, except along the talweg at the higher third of the
watershed where the gneiss occurs as non-cohesive loose
saprolite.
4.3. Vegetation survey and soil–vegetation relationships
Fig. 5. Experimental variogram for electromagnetic conductivity (ECm) data
and adjusted model.
areas on the crest line, while low conductivity values are
mainly observed along the slopes.
4.2. Geology
As presented on the lithological map (Fig. 7), most of the
watershed is developed on paragneiss (peninsular gneiss)
and basic rocks (amphibolite and derived facies). The latter
cover about 17% of the watershed and are not related to any
particular topographic locations. They can be found any-
The ATT vegetation type is dominant, covering about
70% of the watershed, and has developed on both, thick red
soils and thin black soils with a vertic horizon usually less
than 1 m thick. A few isolated Ceristoides trees have developed into the ATT type although they are usually associated
with Swamp vegetation. In this case, the presence of the
Ceristoides is always associated to higher ECm values and
with the presence of a 0.5-m-thick black soil developed from
an alternation of metre-thick veins of gneiss and amphibolite
saprolite. However, Ceristoides is absent in the southeastern
part of the watershed with ATT vegetation type and high
ECm values. Observations carried out along auger holes and
a pit indicated that in this area, close to the topsoil (e.g. at
0.3 m depths), we also find the presence of a conductive
saprolite consisting of weathered amphibolites.
The Swamp vegetation type is mainly located in the low
parts of the watershed with some spots along the crest line,
and always associated with higher ECm values. Field
observations indicated that it grows exclusively on thick
black soils (N2 m) in the lower part of the watershed as well
as on the crest, covering about 5% of the area. The Shorea
vegetation type covering about 15% of the watershed occurs
Fig. 6. Kriged map of soil electromagnetic conductivity (ECm) at the Mulehole watershed.
320
Fig. 7. Lithological map of the studied area.
on very shallow red soil overlying a sandy gneiss saprolite
found very close to the topsoil (0.2–0.4 m below the topographic surface).
4.4. Soil distribution
The map presented on Fig. 8 was drawn from the overlay
of ECm, lithology and vegetation survey and crosschecked
with about 60 isolated soil observations. A clear relationship
was observed between the soil electromagnetic conductivity
and certain soil characteristics explained below. Ferralsols
are usually 2 to 3 m thick and have lower conductivity,
whereas Luvisols are thinner and have higher conductivity.
Since the EM31 response depends on both the thickness and
the apparent conductivity, no significant contrast was
detected with this method, and it does not make it possible
to discriminate the spots of Ferralsols from the surrounding
Chromic Luvisols. Therefore, both Ferralsols and Luvisols
have been grouped together into a red soil unit. The boundary between red and black soils was identified at an ECm
value of 8–10 mS/m (0.9 to 1 log value) and ECm values
above 18 mS/m (1.26 log value) indicates thick black soil,
and this boundary is strictly in agreement with the distribution of the Swamp vegetation type. However, although
the relationship between ECm and soil type was tested in
many places on the watershed, it was not valid for the southeastern part where the saprolite of an amphibolite type rocks
was observed close to the topsoil and associated to higher
ECm values (Fig. 7).
The major part of the watershed is covered by red soils
that are about 1 or 2 m thick and can reach about 4 m at
certain locations. Thin red soils (about 0.2 to 0.5 m thick)
overlying loose gneiss saprolite and associated with the
Shorea vegetation are mainly located on the central wash
divide between the two main talwegs and in a discontinuous
crescent-shaped area along the slopes. Black soils have
developed on two types of location: (i) the low-lying area,
occupying the lower part of the slope and the flat valley
bottoms. Black soil areas are about 2 m thick at the perimeter
but can reach more than 6 m at the centre. They have
developed from both gneiss and amphibolite saprolite. (ii) At
higher levels black soils are about 0.2 to 0.5 m thick,
except at the depressions (50 to 100 m in diameter) on the
crest line where the black soils can reach 2.5 m. They are
always associated with gneiss, which alternates with
amphibolites.
4.5. Streambed features and soil–streambed relationships
Although the stream is meandering in the valley bottom,
there is no undermining of the banks in the convex curves or
point-bar deposits inside meanders, and the stream seems to
have sunk on its own bed. The bottom of the streambed is
steep-sided of about 2 to 4 m and flows on the hard saprolite
on the lower 2/3 of the watershed, and on the soil cover on
the upper 1/3 part of the watershed. At several places of
lower third of the watershed, roots of trees such as Tectona
grandis are occasionally crossing the streambed at about 1 m
from the bottom (Fig. 4G). A peculiar distribution of streambed was observed in relation to the black soil developed
downslope: When the streambed enters a black soil area, it
does not flow straight through it but gets around the thick
black soil area and meanders into the thin black soil as shown
on Fig. 8.
321
Fig. 8. Soil cover on Mulehole watershed and distribution of erosion spots. 1 — rotational slip; 2 — seepage erosion at the top of black soil; 3 — seepage erosion
and mass movement in non-cohesive saprolite at midslope.
4.6. Types and distribution of erosion spots
All the erosion spots are located in the vicinity of the
talwegs (i.e. less than 30 m). Three main types of erosion
were identified (Sidle et al., 1985) and their descriptive
statistics are given in Table 1.
The first and the most widespread type is a rotational slip
(25 sites) with vertical edges, whose standard dimensions are
about 5 m wide, 25 m long and 2 m deep (Fig. 4A and D). In
most of the slips, the material has been subsequently
evacuated towards the stream. This type of erosion is well
developed in the bottom area and within the lowest third of
the watershed. All the slips are distributed at the crossing
between the streambeds and the iso-conductivity line of
11 mS/m, i.e. very close to the contact between red and black
soils. More precisely they are slightly inside the black soil
area (Fig. 8) and always develop towards the red soil domain.
The second type (14 sites) refers to superficial erosional
scars whose widths and lengths are about 2 or 3 m,
respectively, with depths of about 0.5 m thick (Fig. 4B). This
landform is provoked by occasional seepage occurring at the
top of the black soils at a textural contrast between a clay
horizon and the sandy topsoil horizons with many coarse
elements such as ferruginous nodules, quartz, etc. It was
found in several places such as at the bottom parts of the
watershed, along the slopes and close to the crest line.
The third type (7 spots) is much wider than types 1 and 2
and the average eroded soil volume reaches 3300 m3 per spot
(Fig. 4C). Although the lower part is predominantly a flow
movement, the upper part involves recurrent small sliding. It
has developed in the vicinity of the streambeds, but only at
places where the non-cohesive saprolite of the gneiss is close
to the topsoil. In other words it occurs within the uppermost
third of the watershed and at the intersection between the
streambeds and the crescent-shaped area where the Shorea
vegetation was surveyed.
4.7. Soil morphology at the red soil–black soil contact
Fig. 9a shows the soil distribution pattern along the
contact between red and black soil, which can be easily
divided into two domains. Upslope, a 3 m-thick red soil
overlies white gneiss saprolite. In the saprolite (1) the structure and the sub-vertical foliation of the rock itself is still
preserved. Five horizons (2 to 6) have been distinguished in
the red soil. From the bottom to the top of the profile, these
horizons differ mainly by their colour evolving from grey
(7.5YR6/1), brown (7.5YR4/4), reddish-brown (7.5 to
5YR4/4) then red (5YR4/3 to 4/2), and also by the structure
evolving from angular blocky to micro-aggregate and granular. Ferruginous nodules (3 to 6 mm) are observed from the
top of the gneiss saprolite to the topsoil horizon (6), with a
maximum of concentration at the top of horizon (4). The
Table 1
Descriptive statistics of the erosion area at Mulehole watershed
Erosion type
Number
of spots
Type 1: rotational slips
25
Type 2: seepage erosion
14
Type 3: recurrent combination of 7
earthflow and sliding
Mean eroded Standard
soil volume deviation of
eroded soil
m3
volume
250
5
3300
48
1.4
926
322
above-described horizons (2 to 6) are parallel to each other
and to the topography of the slope.
Downslope, lateral morphological changes occur from the
bottom of the soil to almost close to the topsoil, where clay
films coat the structural faces of the soil aggregates. The
presence of clay films defines horizon (7), which intersects
horizons (2), (3), and (4) without changing their respective
structures. The clay films become thicker in horizon (7)
while moving downslope and it progressively turns into a 10
to 20 cm-thick clay horizon (8), which is dark red–brown (5
to 7.5YR3/2 to 4/2) with a coarse angular blocky structure.
The blocks are compact, hard and separated by vertical and
sub-horizontal cracks but without slickensides. Horizons (7)
and (8) have many ferruginous nodules, and they extend the
maximum of nodules observed into horizon 4. Horizon (9)
differs from horizon (8) wherein the colour shifts to dark
brown (5 to 7.5YR3/2 to 2/2) and the structure becomes
clearly vertic (15 cm wide) with sub-horizontal slickensides.
Horizon (10) below (9) is horizontal, concordant with the flat
valley bottom occupied by black soil cover on the right side
of Fig. 9b. It is more greyish (10YR3/1 to 3/2) and the vertic
structure of about 7 to 8 cm is still dominant but with an
angular blocky sub-structure. At horizon (11) the soil
material progressively turns into saprolite of the gneiss in
which the lithologic structure is still preserved. Isolated
volumes of the saprolite preserving the orientation of the
parental material are observed up to the base of horizon (10).
In addition to the above-mentioned description, two
bleached horizons have been identified, discriminated from a
contrast in colour (lighter) and texture (more sandy) and
structure (massive). The first one (12) lies above horizons (7)
and (8), evolving downslope progressively from dark (5YR3/
2) clay–sand to light (7.5YR4.5/3) sand. In this horizon the
bleaching increases downslope and there is a higher proportion
of coarse elements (centimetre-sized angular quartz fragments,
ferruginous nodules). The presence of nodules is an extension
from the already mentioned nodules in horizons (4), (7) and
(8). At the upslope part of horizon (12), the coarse elements are
separated from each other by a sandy clay matrix whereas
downslope they are more frequently in contact with each other.
Towards the stream, horizon (12) becomes thicker and its
organisation intersects the black soil system comprising of
horizons (8) to (11). Horizon (12) is itself intersected by the
incision of the talweg, but a similar sandy bleached material,
although cemented probably by amorphous silica, was
observed in the middle of the streambed.
A second bleached horizon (13) is observed within the
saprolite of the gneiss between 11 and 1. It is about 0.8 m
thick close to the streambed, wedge-shaped and extends up
to 30 m upslope.
Fig. 9. Cross section along the toposequence T1, showing a: the morphology of the red soil–black soil system (numbers refer to the horizons described in the
text); b: relationships with the development of the stream and the erosion types 1 and 2 (letters refer to the different steps in the development of the soil cover and
streambed described in the text; cross-hatched area is part of (C) cemented by amorphous silica).
323
Fig. 10. Morphology of the soil cover along the toposequence T2 showing the development of bleached horizon 12 in the depression before the incision by the stream
(numbers refer to the horizons described in the text, and letters refer to the different steps in the development of the soil cover and streambed described in the text).
Sequence T2 is located across a depression developed
along a red soil-black soil contact (Fig. 10). The layout along
T2 is almost similar to that along T1, except that it is not
intersected by the incision of the streambed (Fig. 4E). In the
depression, a sandy horizon, with location and characteristics
similar to those of horizon (12) of T1, has developed at the
top of the vertic clay horizon. At the downslope of horizon
(12) although the texture is sandy, the vertic structure of a
previously clay horizon is preserved (Fig. 4F) probably due
to the cementation by amorphous silica. The shape and the
size of the structure are similar in every respect to that
observed in the vertic clay horizon (10).
4.8. Hydrodynamic behaviour of the red soil–black soil
system
There is a strong contrast in the evolution of the water
contents between A5 and A2 whereas along A3, A1 and A4
the evolution is intermediate. Therefore, the description here
will be limited along these two end members, i.e., A5 for the
Fig. 11. Neutron probe monitoring along red soil profile A5 (a), and black soil profile A2 (b) and (c), numbers on the curves refer to comments in the text.
324
red soil and A2 for the black soil. At the end of the dry
season, a uniform 20% volumetric water content is found
along the red soil profile A5 (Fig. 11a, curve 1). At the
beginning of the rainy season, the moisture content increases
and the moisture front lowers regularly down to the saprolite
(curve 2). During the wet season, the volumetric water
content remains almost uniform along the profile, oscillating
from about 30% immediately after the rainy events (curve 3)
to 27% after draining the gravity water (curve 4).
Along A2, at the end of the 2004 dry season (Fig. 11b,
curve 1), the water content was more contrasted. Values of
about 20% at 0.2 m depth increased progressively to about
27% in the vertic clay horizons (9) and (10) between 0.8 to
1.8 m depth, and decreased again down to 20% in the
saprolite (11) and, at further depth, reached 15 to 19% in the
bleached horizon (13). Similar to what was described along
the red soil profile, a regular progression of the moisture front
is observed at the beginning of the wet season at the topsoil
horizons (6) and (12) (curves 2 and 3). When the moisture
front reached the clay horizons (8) and (9), the water rapidly
descends to the bottom of the clay horizon (10) until the top of
the saprolite (11) and the moisture content reaches 35%
(curve 4). A strong rainfall event occurred on August 5, 2004
(Fig. 11c, curve 2). The water content increased by about 3 to
5% all along the profile whereas an abrupt increase was
observed at 3 m in horizon (13). The water content increased
up to 65% but this value is not reliable because it is beyond
the calibration range of the neutron probe (maximum water
content of 35%). A few hours later the water content had
decreased again to about 20%, indicating that the water
drained out quickly at horizon (13) (curve 3).
In 2005, the moisture conditions at the end of the dry
season (Fig. 11c, curve 1) were very close to those at the same
period in 2004 (Fig. 11b, curve 1). The 2005 rainy season
started earlier but with two main rainfall events in April and
July separated by a dry period. In the black soil, a different
behaviour of the moisture front was observed during these
two rainy periods. During the first one, similar to what was
observed during the previous rainy season of 2004, the
moisture front reached clay horizon (8) and moved quickly
along the profile A2 down to the top of the saprolite (11).
During the dry interval, the moisture content decreased along
the profile, and particularly in the topsoil horizon. At the
following rainfall event, the moisture content increased again
in the topsoil horizons with maximum water in the sandy
horizon (12) (Fig. 11c, curve 4). After a few days, the water
content increased again in the clay horizons (8), (9) and (10).
5. Discussion
measured by the EM31 device. It also indicates that the high
density of measurements is sufficient to take into account the
variations and describe the spatial structure of conductivity at
short distance. Moreover, the interspace between the transects (100 m) is below the range value of 300 m shown on
Fig. 5, indicating that the density is also sufficient for a good
assessment of ECm distribution between the transects. The
elaborated ECm map is therefore a reliable tool for the
extrapolation of soil and rock data on the studied area, and to
discuss soil distribution.
5.2. Hydrological behaviour of an area vulnerable to erosion
The neutron probe monitoring carried out along sequence
T1 shows two different types of hydrological behaviour of
the black soil during the season.
At the beginning of the wet season, the cracks of the vertic
clay horizon are open, favouring infiltration down to the
saprolite after the moisture front reached the vertic horizons
(9 and 10). This distinctive behaviour has often been
described in black soils of South India (Hodnett and Bell,
1981). Several authors have observed that the cracks are
usually open down to 2 m deep at the beginning of the wet
season, and ending at the slickenside zone which is most
strongly expressed at or just below the depth of the cracks in
thick black soils (Kalbande et al., 1992). A similar
information is given by Hodnett and Bell (1981), who
found essentially no infiltration below the black soils where
the clay horizons were deeper than 2 m whereas large
quantities of water infiltrates at places where the saprolite is
observed at less than 2 m deep and is reached by the cracks.
The effect of the swelling and closing of the cracks is
emphasized with the neutron probe monitoring in 2005.
During the first rainy event, the cracks are open and allow a
fast distribution of the water down to the saprolite, whereas
during the second one, the cracks are closed which significantly decreases the infiltration that results in the formation
of a perched watertable in the subsurface at horizon (12).
Radhakrishna and Vaidyanadhan (1994) reported that the
seasonal changes in water infiltration in some black soils can
shift from more than 70 mm h− 1 at the beginning of the rainy
season to less than 0.1 mm h− 1 after swelling and closing of
the cracks.
The monitoring along T1 also indicates a substantial but
fleeting increase in the moisture content observed below 3 m
depth during the floods (Fig. 11c, curves 2 and 3). It is
attributed to a lateral flow from the streambed into the porous
horizon (13). The water is very quickly evacuated from
horizon (13), which confirms that bleaching in this horizon
can be attributed to a currently ongoing process.
5.1. Reliability of the ECm survey
The experimental variogram built from ECm data shows
no nugget effect, which emphasizes that there is no
variability in ECm response at short distance. The absence
of a nugget effect testifies to the stability of the ECm values
5.3. Soil distribution and relative chronology in the soil
cover
On the one hand, black soils mainly occur in the valley
bottom although not exclusively, because they are also found
at certain spots along the slope and at the crest line. On the
other hand, black soils are sometimes related to the presence
of amphibolite, but they are not exclusively on amphibolites
alone since they are frequently observed on gneiss saprolite
as well. Therefore, although dry climate (b 1200 mm), downslope topography and lack of drainage are considered important factors in the development of black soils, our
observations suggest that both topography and lithology
have influenced their formation.
The soil cover morphology, comprising the layout of the
horizons in cross section and more precisely the concordances or discordances between several sectors of the soil
cover, make it possible to draw the relative chronology in the
formation of the soil–streambed system. Five different ensembles are distinguished, namely the red soil, the black soil,
the bleached horizon (12), the streambed, and the bleached
horizon (13), and they are referred as systems A to E on
Figs. 9b and 10. The red soil system A is developed along the
slopes under good drainage conditions, and consists of
horizons concordant with the slope topography. The regular
evolution from the saprolite of the gneiss to the topsoil
suggests that this material is autochthonous and has developed from the gneiss. B is the black soil system with horizons concordant to the flat valley bottom and developed
under bad drainage conditions. The following two arguments
allow us to conclude that B progresses at the expense of A. 1.
At the contact between the two systems A and B, we observed that B is developing from the base of the red soil
system, first with the clay films on the structural faces of
horizons (2), (3) and (4). The clay films are located around
and not within the aggregates, and this pattern must be
interpreted as formation and not destruction of clay material.
2. The morphology of the red soil is intersected by that of the
black soil, indicating that the latter has developed more
recently and at the expense of the red soil material. Within
horizon (11), isolated volumes of the saprolite still preserve
the orientation of the parental material, which indicate
autochthony. The same is observed at the base of the black
soil on the right side of the stream, and no morphological
discontinuity nor evidence of alluvial/colluvial deposits has
been detected towards the top of the black soil profile.
Therefore, we suppose that the black soil has developed on
autochthonous material. This point could be debated because
the vertic horizon is known to homogenise the material due
to the shrinkage–swelling effect, and could have removed a
possible discontinuity in the soil profile. The structural approach alone does not make it possible to settle the argument.
On the left side of the stream, the morphology of the black
soil B is in its turn intersected by horizon (12) (C), which has
therefore developed subsequently after B. Because C starts
occurring close to the topsoil and just downslope from the
contact between A and B, it suggests that C is induced by
lateral and sub-surface drainage at the top of B where a
perched watertable is fleetingly occurring during rainfall. It
develops downslope due to a longer duration of the
episaturation.
325
The fourth system (D) is the streambed itself, which
intersects B and C. The black soil morphology B is horizontal and observed on both sides of the stream, and the
vertic horizons (10) and (11) of the B system are in particular
intersected by the streambed D. We conclude that the soil
cover, previously continuous and almost horizontal, was
removed by the later incision of the streambed, which is also
confirmed by the presence of roots crossing the streambed.
The streambed D also intersects unit C. The chronology
of C with respect to the formation of the streambed is
debatable. On the one hand, C could have been provoked or
favoured by the drainage induced by the talweg and therefore
have developed subsequently. On the other hand, C could
have developed first and have been subsequently incised by
the streambed. The soil cover morphology along sequence
T2 shows that C is continuous and had developed before the
incision of the streambed D (Fig. 4). Therefore, and because
of the similarity in the morphology of both T1 and T2, it
suggests that the same had occurred at T1.
Eventually, the bleached horizon (13) E has developed at
the base of the black soil and within the saprolite of the
gneiss. E intersects the red soil-black soil contact and is
almost horizontal, i.e. concordant with the water level in the
stream. Therefore we attribute the bleaching in E to the fast
oscillation of the watertable induced by the stream and
highlighted by the neutron probe monitoring. E is well
developed on the left side of the stream in the gneiss
saprolite, i.e. towards the red soil system. On the right side of
the stream, however, the thickening of the vertic clay
horizons (10) and (11) obstructs its development and it is
therefore only a few decimetres wide.
5.4. Downslope landscape evolution
We previously concluded that the development of the
streambed is a recent process that took place after the
development of the soils. Moreover at the watershed scale,
we observed that the streambed lies within the thin black
soil, skirting around the thick black soil area. It suggests that
the thin black soils have favoured or guided the development
of the drainage network. Based on the above-mentioned
observations of the relative chronology and hydrological
behaviour of the soil cover we can propose a model of recent
evolution for this downslope part of the landscape described
in four steps (Fig. 12):
At stage 1, red soils occupy the slope, whereas black soils
are developed on the flat valley bottom. At the beginning of
the rainy season, the cracks are opened down to about 2 m
deep in the thick black soil area and they end within the clay
horizon, which prevents deep drainage. However, close to
the border of the black soil area, although the cracks reach
the same depth, they end at the sandy and permeable
saprolite located below that enables the infiltration of a large
quantity of water during the first events of the rainy season.
At stage 2, the chemical erosion or leaching is likely to
have provoked the formation of a depression that will
326
Fig. 12. Four-stage model showing the relative chronology in the recent formation of the soil cover at downslope. 1. Initial red soil–black soil contact (arrows
denote the water flow along cracks and within the saprolite); 2. Development of the depression in the thin black soil; 3. Bleaching in the depression and hardening
due to amorphous silica; 4. Incision of the stream and bleaching in the saprolite below the black soil (system E).
develop preferably in the thin black soils. Infiltration in the
black soils occurs only at the onset of the wet season. During
the rainy season and after the closing of the cracks, permeability of the black soils decreases and the depression
behaves as a gutter collecting the runoff water. In its turn, the
flow of water in the depression at the outer part of the black
soil will favour the soil bleaching and horizons will progressively turn sandy.
Soil bleaching leads to stage 3 that corresponds roughly to
the morphology observed along sequence T2. Observations
at sequence T2 confirm that the formation of the streambed is
preceded by the presence of the depression along the contact
between red soil and black soil where the bleached system C
has developed. The central part of the system C is cemented
and a vertic structure is observed. The vertic structure could
not have developed within the sandy material observed at
present in system C but within a swelling clay horizon of the
system B. During bleaching and textural change from B to C,
the vertic structure is supposed to disappear. The presence of
the vertic structure in C suggests that cementation and
bleaching have occurred simultaneously, which made it
possible to maintain the vertic structure in the sandy material
C. On the left and at the contact between A and B the soil
solution slows down, favouring the formation of clay coating
around aggregates through over saturation of the soil solution with respect to silicate clays that led to the development
of horizons (7), (8) and (9).
At stage 4, the streambed had developed down to the
saprolite of the gneiss into the depression and in the sandy
horizons of system C. A portion of C is preserved because of
cementation by amorphous silica. The presence of the
streambed could favour the bleaching in horizon (12) at the
top of the vertic clay. During the rainy season, the rapid
oscillations of the watertable provoke the bleaching in the
saprolite and the formation of system E.
5.5. Soil erosion relationships
The afore-described model for the formation of the soil
cover is in agreement with the observations made at several
scales on our study site. At the watershed scale, the model
explains why the streambed is passing through the thin black
soil area instead of crossing directly through the middle of
the flat bottom area covered with thick black soil. At the
scale of the sequence, the model is in agreement with the soil
morphology observed along T2 and T1, respectively before
and after the incision of the streambed. It also agrees with the
hydrological behaviour in the black soil along soil profile
A2. The development of the natural erosion in this area can
be explained through the interaction of the stream and its
distribution, with the type of soil cover and its hydrological
behaviour. We will consider landform types 1, 2, and 3
successively.
Rotational slips (type 1) and seepage erosion (type 2)
occur close to the contact between red and black soil. The
type 1 features are favoured by the presence of system E. The
neutron probe monitoring shows that a temporary watertable
occurred very fleetingly within E when the water level was
up in the stream. The thickening of the vertic clay horizon is
obstructing the development of the system E towards the
centre of the black soil area, but it developed predominantly
from the streambed towards system A into the permeable
saprolite. Because the system E is the plane of weakness for
the erosion type 1, the rotational slips are also predominantly
327
developed towards the red soil system A, and concern the
whole soil cover down to the saprolite.
Seepage erosion (type 2) develops in system C at the
upper part of the black soil. Because of the swelling in the
black soil, the cracks get closed usually in the middle of the
wet season, i.e. during the month of July. Later, heavy
rainfall and less infiltration provoke the formation of a
perched watertable within C and the water flows sub-superficially towards the streambed. Whatever the process of clay
elimination may be (leaching, ferrolysis…), the bleaching
increases downslope resulting in a relative accumulation of
the coarse elements, which become contiguous. Hence the
whole bleaching process increases the vulnerability to erosion, which probably occurs in soaked (C) material during
heavy rainfall, leading to sub-surface seepage erosion at the
contact between (B) and (C).
The third type of erosion is not related to the red soil–
black soil system, but to the non-cohesive sandy saprolite of
the gneiss when it is exposed close to the topsoil. The erosion
is due to the seasonal throughflow seepage of the watertable
occurring within the sandy saprolite over the less permeable
fractured rock during the rainy season. The combination
between earthflow and sliding (type 3) occurs mostly in
midslope positions (Fig. 8) and further studies should focus
on it as a possible regional feature. In this case, these large
erosional scars could influence the geomorphologic landscape evolution at wider scale and further study should also
focus on the agreement between midslope erosion and the
regional geomorphologic model proposed by Bourgeon and
Gunnell (1998) and Gunnell et al. (2003).
6. Future directions
The objective of this study was to understand the distribution, dynamics and the factors intrinsic to the soil cover that
are likely to influence or even govern the development of
present and recent natural erosion in the forested area of
South India. This aim was tackled through a structural
approach of the soil cover, which includes the overlay of
various types of survey at the watershed scale (electromagnetic induction, soil, geology, vegetation), and the study of
concordances–discordances between horizons along representative soil sequences.
The present erosion is not randomly distributed. Three
different types of erosion have been identified: Downslope
rotational slips are governed by a temporary watertable
within a bleached saprolite at the base of black soil–red soil
transition. Seepage erosion is caused by a perched watertable
occurring after closing of the cracks at the top of the vertic
clay horizon of black soil. At midslope, a combination of
earthflow and sliding occurred at places where the noncohesive sandy saprolite of gneiss is exposed close to the
topsoil.
Our study highlights the relative chronology in the development of the downslope soil cover and in particular we
show that the geomorphology of valley bottoms and their
328
erosion have been recently reactivated with the development
of streambeds. Further research effort should focus on the
study of the pedological processes prior or subsequent to the
development of the streambeds in order to identify their
contribution to the quality of the stream water. In particular,
because silica cemented horizons have been observed at
several points along the streambed, they are likely to be
integral components of the soil system. Therefore, further
study should focus on the identification of the soil forming
processes that provide aqueous silica to this part of the
system.
Acknowledgments
This study was supported through the research project
“Kabini river basin” of ORE-BVET (Observatoire de
Recherche en Environnement-Bassin Versant Expérimentaux Tropicaux, www.orebvet.fr), the French national
programs “ECCO-PNRH” and “ACI-Eau” funded by IRD/
INSU/CNRS, the Indo-French Centre for the Promotion of
Advanced Research (IFCPAR WA-3000), and the Embassy
of France in India. We thank Dr C. Camerlynck from UMR
7619 Sisyphe, University of Paris 6 for providing the EM31
equipment, Dr R. Wins (BRGM) for its contribution to the
geological survey, the Karnataka Forest Department for
providing the access to the site, and Dr. Vasanthi Dass for
editorial advice.
References
Agarwala, V.P., 1985. Forests in India, Environmental and Production
Frontiers. IBH Publications, New Delhi, pp. 1–344.
Barbiéro, L., Ould Mohamedou, A., Caruba, R., 1998. Influence de la
maturation des sols de mangrove sur la déflation éolienne et la formation
de dunes argileuses dans le delta du fleuve Sénégal. Comptes Rendus de
l'Académie des Sciences Paris 327, 115–120.
Boulet, R., Bocquier, G., Millot, G., 1977. Géochimie de la surface et formes
du relief. I., Déséquilibre pédobioclimatique dans les couvertures
pédologiques de l'Afrique tropicale de Ouest et son rôle dans
l'aplanissement des reliefs. Sciences Géologiques 30, 235–243.
Boulet, R., Chauvel, A., Humbel, F.X., Lucas, Y., 1982. Analyse structurale
et cartographie en pédologie. I — Les études de toposéquences et
principaux apports à la connaissance des sols. II — Une méthode
d'analyse prenant en compte l'analyse tridimensionnelle des couvertures
pédologiques. III — Passage de la phase analytique à une cartographie
générale synthétique. Cahier ORSTOM, Série Pédologie 19, 309–351.
Bourgeon, G., 1991. Les “sols rouges” de l'Inde péninsulaire méridionale:
Pédogenèse fersiallitique sur socle cristallin en milieu tropical. Doctorat
Thesis, Université Paris VI, France.
Bourgeon, G., Gunnell, Y., 1998. Rôle du regime tectonique et du taux de
denudation sur la repartition géographique et les propriétés des sols
tropicaux. Comptes Rendus de l'Académie des Sciences Paris 326,
167–172.
Bourgeon, G., Pedro, G., 1992. Rôle majeur du drainage climatique dans la
differentiation altéritique et pédologique des sols des regions chaudes.
Comptes Rendus de l'Académie des Sciences Paris 314, 717–725.
Braun, J.J., Ndam Ngoupayou, J.R., Viers, J., Dupré, B., Bedimo Bedimo,
J.P., Boeglin, J.L., Robain, H., Nyeck, B., Freydier, R., Sigha Nkamdjou,
L., Rouiller, J., Muller, J.P., 2005. Present weathering rates in a humid
tropical watershed: Nsimi, South Cameroon. Geochimica et Cosmochimica Acta 69, 357–387.
Braun, J.J., Descloitres, M., Barbiéro, L., Fleury, S., Boeglin, J.L., Ruiz, L.,
Sekhar, M., Mohan Kumar, M.S., Kumar, M.C., Dupré, B., 2006.
Regolith thickness inferred from geophysical and geochemical studies in
a tropical watershed developed on gneissic basement: Mulehole,
Western Ghats (South India). Goldschmidt Conference Abstracts 2006.
Geochimica et Cosmochimica Acta 70 (18), A65 (Supplement 1).
Bronger, A., Bruhn, N., 1989. Relict and recent features in tropical alfisols
from South India, in Bronger. In: Bronger, A., Catt, J.A. (Eds.),
Paleopedology. Catena Supplement, vol. 16, pp. 107–128.
Bronger, A., Wichmann, P., Ensling, J., 2000. Over-estimation of efficiency
of weathering in tropical “Red soil”: its importance for geoecological
problems. Catena 4, 181–197.
Brückner, H., Bruhn, N., 1992. Aspects of weathering and peneplanation in
South India. Zeitschrift für Geomorphologie 91, 43–66.
Cannon, M.E., McKenzie, R.C., Lachapelle, G.P., 1994. Soil salinity
mapping with electromagnetic induction and satellite based navigation
methods. Canadian Journal of Soil Science 74, 335–343.
Corwin, D.L., Lesch, S.M., Oster, J.D., Kaffka, S.R., 2006. Monitoring
management-induced spatio-temporal changes in soil quality through
soil sampling directed by apparent electrical conductivity. Geoderma
131, 369–387.
Descloitres, M., Ruiz, L., Sekhar, M., Legchenko, A., Braun, J.J., Mohan
Kumar, M.S., Subramanian, S., in press. Characterization of seasonal
local recharge using Electrical Resistivity Tomography and Magnetic
Resonance Sounding. Hydrological processes.
Dowd, P.A., 1982. Lognormal kriging — the general case. Mathematical
Geology 14, 475–499.
Durand, N., Gunnel, Y., Curmi, P., Ahmad, S.M., 2006. Pathways of calcrete
development on weathered silicate rocks in Tamil Nadu, India: Mineralogy,
chemistry and paleoenvironmental implications. Sedimentary Geology
192, 1–18.
Durand, N., Gunnell, Y, Curmi, P., Ahmad, S.M., in press. Pedogenic
carbonates on Precambrian silicate rocks in south India. Orgin and
paleoclimatic significance. Geological Society of America Special
Paper.
FAO-ISRIC-ISSS, 1998. World reference base for soil resources. World Soil
Resources Report 84. FAO, Rome.
Filizola, H.F., Boulet, R., 1996. Evolution and opening of closed depressions
developed in a quartz-kaolinitic sedimentary substratum at Taubaté basin
(Sao Paulo, Brazil) and analogy to the slope evolution. Geomorphology
16, 77–86.
Fritsch, E., Peterschmitt, E., Herbillon, A.J., 1992. A structural approach to
the regolith: identification of structures, analysis of structural relationships and interpretations. Sciences Géologiques Bulletin 45, 77–97.
Furian, S., Barbiéro, L., Boulet, R., 1999. Organisation and dynamic of a soil
mantel in tropical southeastern Brazil (Serra do Mar). Relation with
landslides processes. Catena 38, 65–83.
Gaillardet, J., Dupré, B., Allègre, C.J., Négrel, P., 1997. Chemical and
physical denudation in the Amazon River Basin. Chemical Geology 142,
141–173.
Gaillardet, J., Dupré, B., Louvat, P., Allègre, C.J., 1999. Global silicate
weathering and CO2 consumption rates deduced from the chemistry of
large rivers. Chemical Geology 159, 3–30.
Gunnell, Y., 2000. The characterization of steady state in earth surface
systems: finding from the gradient modelling of an Indian climosequence. Geomorphology 35, 11–20.
Gunnell, Y., Bourgeon, G., 1997. Soils and climatic geomorphology on the
Karnataka Plateau, peninsular India. Catena 29, 239–262.
Gunnell, Y., Gallagher, K., Carter, A., Widdowson, M., Hurford, A.J., 2003.
Denudation history of the continental margin of western peninsular India
since the early Mesozoic—reconciling apatite fission-track data with
geomorphology. Earth and Planetary Science Letters 215, 187–201.
Hodnett, M.G., Bell, J.P., 1981. Soil physical processes of groundwater
recharge through Indian Black Cotton Soils. Report, vol. 77. Institute of
hydrology, Wellingford, UK, pp. 1–17.
Jacks, G., Sharma, V.P., 1995. Geochemistry of calcic horizons in relation to
hillslope processes, Southern India. Geoderma 67, 203–215.
Kalbande, A.R., Pal, D.K., Deshpande, S.B., 1992. b-Fabric of some
benchmark Vertisols of India in relation to their mineralogy. Journal of
Soil Science 43, 375–385.
McNeill, J.D., 1980. Electromagnetic terrain conductivity measurement at
low induction numbers. Technical Note TN-6, Geonics Ltd, Ontario,
Canada.
Moyen, J.F., Martin, H., Jayananda, M., 2001. Multi- element geochemical
modeling of crust–mantle interactions during late-Archean crustal
growth: the Closepet Granite (South India). Precambrian Research
112, 87–105.
Murthy, R.S., Hirekerur, L.R., Deshpande, S.B., Venkat Rao, B.V., 1982.
Benchmark Soils of India. National Bureau of Soil Survey and Land Use
Planning, Nagpur, India.
Oliva, P., Viers, J., Dupré, B., 2003. Chemical weathering in granitic
environments. Chemical Geology 202, 225–256.
Pal, D.K., Deshpande, S.B., 1987. Genesis of clay minerals in a red and
black complex soils of southern India. Clay Research 6, 6–13.
Pascal, J.P., 1982. Forest map of South India, 1/250 000 scale, sheet
Mercara–Mysore. Travaux Section Scientifique et Technique Institut
Français de Pondichéry, hors série, vol. 18a.
329
Pascal, J.P., 1986. Explanatory booklet on the forest map of South India.
Travaux Section Scientifique et Technique Institut Français de
Pondichéry, Hors Serie, vol. 18.
Planchon, O., Fritsch, E., Valentin, C., 1987. Rill development in a wet
savannah environment. Catena Supplement 8, 55–70.
Radhakrishna, B.P., Vaidyanadhan, R., 1994. Black soils of Karnataka. In:
Radhakrishna, B.P., Vaidyanadhan, R. (Eds.), Geology of Karnataka.
Geological Society of India, Bangalore, pp. 239–251.
Sidle, R.C., Pearce, A.J., O'Loughlin, C.L., 1985. Hillslope stability and
land use. Water Resources Monograph. Series, vol. 11. American
Geophysical Union, Washington D.C.
Singer, M.J., 1986. Bulk density-paraffin clod method. In: Singer, M.J.,
Janitzky, P. (Eds.), Field and Laboratory Procedures Used in a Soil
Chronosequence Study. U.S. Geological Survey Bulletin, vol. 1648. U.S.
Government Printing Office, Washington D. C., pp. 38–41.
White, A.T., Blum, A.E., Schulz, M.S., Vivit, D.V., Stonestrom, D.A.,
Larsen, M., Murphy, S.F., Eberl, D., 1998. Chemical weathering in a
tropical watershed, Luquillo Mountains, Puerto Rico: I. Long term
versus short term weathering fluxes. Geochimica et Cosmochimica Acta
62, 209–226.
Available online at www.sciencedirect.com
Journal of Applied Geophysics 64 (2008) 83 – 98
www.elsevier.com/locate/jappgeo
Study of water tension differences in heterogeneous sandy soils
using surface ERT
Marc Descloitres a,⁎, Olivier Ribolzi b , Yann Le Troquer b , Jean Pierre Thiébaux b
a
IRD-LTHE, UMR 012 IRD-CNRS-UJF-INPG, 38041 Grenoble, France
b
IRD-UR 176, Ban Sisangvone, BP 5992, Ventiane, Laos PDR
Received 12 October 2006; accepted 27 December 2007
Abstract
Herbaceous vegetation in the Sahel grows almost exclusively on sandy soils which preferentially retain water through infiltration and storage. The
hydrological functioning of these sandy soils during rain cycles is unknown. One way to tackle this issue is to spatialize variations in water content but
these are difficult to measure in the vadose zone. We investigated the use of Electrical Resistivity Tomography (ERT) as a technique for spatializing
resistivity in a non-destructive manner in order to improve our knowledge of relevant hydrological processes. To achieve this, two approaches were
examined. First, we focused on a possible link between water tension (which is much easier to measure in the field by point measurements than water
content), and resistivity (spatialized with ERT). Second, because ERT is affected by solution non-uniqueness and reconstruction smoothing, we
improved the accuracy of ERT inversion by comparing calculated solutions with in-situ resistivity measurements. We studied a natural microdune during
a controlled field experiment with artificial sprinkling which reproduced typical rainfall cycles. We recorded temperature, water tension and resistivity
within the microdune and applied surface ERT before and after the 3 rainfall cycles. Soil samples were collected after the experiment to determine soil
physical characteristics. An experimental relationship between water tension and water content was also investigated. Our results showed that the raw
relationship between calculated ERT resistivity and water tension measurements in sand is highly scattered because of significant spatial variations in
porosity. An improved correlation was achieved by using resistivity ratio and water tension differences. The slope of the relationship depends on the soil
solution conductivity, as predicted by Archie's law when salted water was used for the rain simulation. We found that determining the variations in
electrical resistivity is a sensitive method for spatializing the differences in water tension which are directly linked with infiltration and evaporation/
drainage processes in the vadose zone. However, three factors complicate the use of this approach. Firstly, the relation between water tension and water
content is generally non-linear and dependent on the water content range. This could limit the use of our site-specific relations for spatializing water
content with ERT through tension. Secondly, to achieve the necessary optimization of ERT inversion, we used destructive resistivity measurements in the
soil, which renders ERT less attractive. Thirdly, we found that the calculated resistivity is not always accurate because of the smoothing involved in
surface ERT data inversion. We conclude that further developments are needed into ERT image reconstruction before water tension (and water content)
can be spatialized in heterogeneous sandy soils with the accuracy needed to routinely study their hydrological functioning.
© 2008 Elsevier B.V. All rights reserved.
Keywords: Electrical resistivity Tomography (ERT); Water tension; Sandy soil; Soil moisture; Sahel
1. Introduction
The degradation of natural resources in the arid parts of the
Sahel, is currently a quite serious problem leading to
desertification, loss of biodiversity, increase of surface runoff
and soil losses. Within this degraded landscape, sandy deposits
are islets of fertility (Thiombiano, 2000) which often take the
⁎ Corresponding author.
E-mail address: marc.descloitres@ird.fr (M. Descloitres).
0926-9851/$ - see front matter © 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.jappgeo.2007.12.007
shape of microdunes (Casenave and Valentin, 1992). Microdunes are very important ecological units where significant
infiltration of water takes place (Ribolzi et al., 2000) and are
therefore potential starting points for regeneration of the
Sahelian environment. Little is known, however, about their
hydrological functioning during and after rainfall, processes
such as infiltration, evapo-transpiration and drainage which can
only be studied using destructive soil moisture and water tension
measurements. Since these fragile sandy soils are unstable, they
cannot be brought back from the field or reproduced in the
84
M. Descloitres et al. / Journal of Applied Geophysics 64 (2008) 83–98
laboratory. All experiments must be carried out in the field. To
replicate the controlled experimental conditions of the laboratory,
we used simulated rainfall to reproduce natural rain/evaporation
cycles in the field. The primary objective of our study was to
spatialize water content variation during infiltration and evaporation using surface Electrical Resistivity Tomography (ERT),
which is a technique highly suitable for groundwater studies.
To achieve this objective, two methodological problems must
be overcome. Firstly an experimental relationship between
resistivity and water content has to be established. Secondly, a
protocol for accurate ERT inversion is necessary to reconstruct
actual resistivity distribution in the subsurface. The first problem
requires simultaneous water content monitoring and ERT
acquisition at several measurements points and/or laboratory
experiments on soil samples. These measurements could not be
practically undertaken because of the lack of experimental
sensors to measure directly water content at several point measurements in a thin dune. Also the soil is too fragile for resistivity
measurements on small cores. Therefore, we investigated
possible approaches for determining a relationship between
resistivity and water tension, which is more easily measured in
the soil and a potentially interesting way to study hydrological
processes linked with infiltration and evaporation.
As with resistivity, soil water tension is strongly related to
water content. Soil water tension measurements are well suited
for determining soil water status in the field (Richards, 1931).
Water tension (or pressure head, or matrix potential) is generally
expressed as a negative value that reaches zero when the soil is
saturated. Water tension data can be considered in two ways:
first, there is a direct relationship between water tension and
water content, as studied by Van Genuchten (1980) for example.
This relationship is generally non linear. Second, the difference
between water tension at the final state and initial states,
provides important information on the movement of water: a
positive difference shows an increase in water content, whereas
a negative difference indicates a decrease. A difference of zero
suggests zero flux with drainage (below) and evaporation
(above). Hence, by spatializing water tension differences we can
get a better understanding of water movement in the vadose
zone. Therefore, the development of a spatialized image of
water tension differences during natural or artificial infiltration
(and evaporation) experiments would be of strong interest. But
the number of sensors (small ceramic cups) that can be used
remains limited because if too many are inserted the medium
will be destroyed. This makes it difficult to spatialize the results
laterally when the soil is heterogeneous. Thus, if a relationship
between water tension and resistivity can be established, ERT
offers attractive possibilities for spatializing water tension. In
this study, we investigated both aspects of tension (its actual
value and its differences from one state to another) in relation to
resistivity. This was done using experimental data collected in
the field during artificial rainfall cycles. A possible extension of
this approach was also considered: we attempted to establish an
experimental relationship between water tension and water
content with the aim of using ERT to spatialize water content.
Because of scattered results due to unexpectedly heterogeneous
soil, this final objective was not achieved.
The second issue we addressed was how to use ERT to
reconstruct spatial and temporal variations in resistivity from one
hydric state to another within a dune. In the vadose zone, electrical
resistivity mainly depends on 3 parameters: water content, water
conductivity and porosity. The empirical Archie's law (Archie,
1942), which is applicable for studies on sandy soils, integrates
these three parameters. Temperature and clay content can also
modify resistivity values (Telford et al., 1990). For saturated
sandy soils, Archie's law is convenient for monitoring variations
in water conductivity, when water content and porosity remain
constant. Singha and Gorelick (2006) investigated the use of
Archie's law for monitoring tracer concentration. They found that
water conductivity values derived from Archie's law using ERT
agreed with experimental data only if the formation factor (the
porosity-dependent parameter of Archie's law) was varied in
space and time. In the present study, the effect of spatial variations
in porosity inside the dune was taken into account when
interpreting our results. In unsaturated soils, such as those in
our study, it is more difficult to derive resistivity variations into
hydrological parameters because water content has also a
significant influence on resistivity values. On the other hand,
the sensitive dependence of resistivity on water content is an
advantage for tracking hydrologic processes, especially if we
consider that water tension is also sensitive to water content.
French et al. (2002) found that ERT is suitable for localizing
infiltration zones, even if resistivity is affected by changes in
saturation values and tracer concentrations. In this study, we tried
to minimize the extent of water conductivity variability by using
demineralized water for the 2 first experimental rainfall events.
We also investigated the use of salted rain to evaluate the effect of
water conductivity, during the third (and final) experimental
rainfall.
Despite the methodological difficulties involved, resistivity
remains an attractive parameter because it offers the advantage
of being easy to measure using non-destructive ERT surface
measurements. Moreover ERT can be applied at different scales
of investigation and is well adapted for plot studies. For example, Michot et al. (2003) used ERT and additional measurements to study water uptake by plants by successfully
monitoring corn crops growing under irrigation. In our study,
we specifically adapted miniaturized arrays to the scale of the
dune, following the concept proposed by Depountis et al. (2001)
in their ERT centrifuge modeling experiment. In addition, ERT
can be used to study 2 or 3D geometry which is a promising
technique/method for reconstructing complex resistivity distribution patterns in the subsurface. Although, in many studies,
cross bore hole ERT was considered as an efficient method for
resistivity imaging (see the pioneering studies by Daily et al.,
1992 as well as work by Slater et al., 2000, or Kemna et al.,
2002), we did not use cross bole hole imaging in this study
because our main methodological objective was to evaluate the
efficiency of non destructive surface ERT.
Several authors reported that ERT resolution is limited due to
the spatial smoothing of inversion algorithms (see Kemna et al.,
2002; Singha and Gorelick, 2006), thus in this paper we also
focus on the following issues: does ERT give a reliable estimate
of resistivity? Finally, is ERT monitoring a useful technique for
determining water tension distribution in sandy soils? To
address these questions, we performed a field experiment with
artificial rainfall. ERT was applied before and after each rainfall.
In a first step, to evaluate ERT accuracy, ERT inversion was
optimized by comparing the calculated resistivity with in-situ
resistivity. In a second step, we investigated the experimental
relationship between water tension and resistivity before and
after rain. The relationship is then discussed and enlightened
with Archie's law. In conclusion we discuss practical considerations for using surface ERT for spatializing water tension in
heterogeneous sandy soils.
2. Experimental setup
The site is located in northern Burkina Faso (Fig. 1). It is a
small degraded 82 ha catchment overgrazed by livestock. The
climate is Sahelian, with only one rainy season (June to
September). The average annual rainfall is 512 mm and mean
annual potential evapotranspiration is 2396 mm. Sandy soils are
formed by aeolian and/or runoff deposits. They are thin (0.3 to
1 m thick) and form more than one third (1/3) of the landscape
surface. The microdune (5.3 m2) shows a typical crescent form
(Fig. 1) depending on wind direction. The windward side is bare
and steeper because of wind erosion, while the leeward side has
vegetation (typically grass) that grows protected from wind. The
root system is sparse (2.6 to 6% of the total volume), consists of
fine roots (diameters 1 to 5 mm) and is not developed below a
depth of 3 to 4 cm. This same dune shape and vegetation
distribution is noted for all sizes, namely from the microdune (a
few meters square) to large sandy deposits (10–50 m long or
85
more). Such dunes generally overlay a more clayey and
compact impervious horizon, which forms the substratum and
is an obstacle to deeper infiltration.
2.1. Hydrological and soil analysis methods
The rainfall simulator used in this study (Asseline and
Valentin, 1978) consists of a sprinkler system mounted at a
height of 4 m (Fig. 1) connected to a constant flow pump that
provides a kinetic energy similar to tropical rainfall (Valentin,
1991). We performed three simulations (Rain 1, Rain 2 and Rain
3) on different dates in February 2002. The rainfall intensity was
set at 70 mm/h for 40 min, typical for this climatic zone. For Rain
1 and Rain 2, we used demineralized water with a low electrical
conductivity (3 μS/cm) identical to natural rain. For Rain 3, we
added salt (NaCl) for hydrochemistry studies so that the
electrical conductivity was 1000 μS/cm. Rain 2 was simulated
24 h after Rain 1, and Rain 3 was simulated 1 h after Rain 2. A
total of six ERT data sets were obtained, each measurement was
made just before or just after rain. The detailed sequence of ERT
experiments is described in the Results section. Before Rain 1,
the sand was too dry to accurately measure resistivity because
the electrodes could not form a good contact with the soil. Thus,
we considered only the data acquired after Rain 1 when the soil
was wet.
Following Rain 3, the microdune was destroyed for soil
samples. We took 42 samples (Fig. 2) simultaneously from a
vertical cross section along the alignment of the ERT electrodes
using a grid sampler. Later we calculated porosity, sand, silt and
clay contents in the laboratory.
Fig. 1. A) Location of the experimental site. B) Rain simulator in the field. C) Map of the microdune. ERT surface electrodes were located along the EW line. The
location of the temperature and electrical probes is also shown.
86
Fig. 2. Experimental set up. Cross section of the central section of the microdune showing the location of surface electrodes, electrical probes 1 and 2, temperature
probe and tensiometers. It was not possible to activate tensiometer 10. The soil samples were collected according to the grid. The model block arrangement used for
geophysical inversion of pole-pole data sets is shown.
For water tension measurements we used very small ceramic
cups (model SDEC 220, length: 11 mm, diameter: 2.2 mm, pore
size: ≈ 1.5 μm, air entry value: 1.5 bar, hydraulic conductivity:
5.10− 7 cm s− 1) to get very localized measurements. We set-up
10 micro-tensiometers along 2 diagonals (Fig. 2) to measure the
water tension below the leeward and windward slopes, which
may exhibit different lateral properties.
We attempted to determine an experimental relationship
between water tension and water content by sampling the sand
at different depths. For this, we drove in thin copper tubes
vertically at different periods of time at 1 m away from electrode
line. The volumetric water content was calculated by weighing
soil samples before and after drying. Water content was related
to tension measurements at the same depth.
2.2. Geophysical measurements
Surface ERT requires an in-line electrode setup. In order to get
dense resistivity data at shallow depth, we set the electrode
spacing to 4 cm. To minimize disturbance and to provide an
acceptable surface for low contact resistance, the electrodes were
small 5-cent copper coins driven vertically into the soil (plane
perpendicular to the profile). We achieved a good contact
resistance (lower than 5 kΩ) only after Rain 1 due to the rain
itself. The array set-up used in this experiment was pole–pole
array. This array has the advantage that it provides a better depth
of exploration compared to other arrays for a limited lateral
extension (Loke, 2000a), and also a good lateral coverage using a
limited number of electrodes. We primarily used 32 electrodes
along the profile shown in Fig. 1. We used another 14 electrodes
to focus on the central zone of the dune where water tension
measurements were made, (Fig. 2). We measured 91 apparent
resistivities within 6 min using a Syscal R2 resistivity-meter (IRIS
Instruments). We inverted the data using Res2Dinv software
(Loke, 2000b) to produce a 2D calculated resistivity section of the
subsurface. Fig. 2 details the block arrangement chosen for
inversion. It is well known that ERT inversion is subject to
solution non-uniqueness. Therefore to perform an optimized ERT
image reconstruction using a unique inversion parameter set, we
tested several parameters and compared the calculated resistivity
results to actual resistivity as proposed by Loke (2000b). For this
purpose, some point measurements of electrical resistivity at
depth were required. The resistivity data measured along the
probe was only used for optimizing the inversion, but not as data
in the inversion. The use of external data points was also reported
in Liu and Yeh (2004). They recommended that sparse point
measurements should be used to enhance ERT accuracy (i.e.
lowering its non-unique inverse solution). They used a stochastic
information fusion concept to invert their data sets. Their
approach also relied on point measurements of moisture content
(considered as essential in their approach) and electrical potential.
This type of data were not available for our study but our approach
was similar to that of Liu and Yeh's in that ERT inaccuracy was
reduced by using optimized inversion parameters.
We positioned electrical and temperature probes in the
ground a few months before the rainy season to avoid
preferential infiltration along the probes and so the soil had
time to reorganize around them. Because resistivity varies with
87
temperature, we also recorded temperature every 5 minutes,
vertically inside the ground using thermistors at every 2 cm
along the probe, as shown in Fig. 2. During experiments, the
temperature varied on the surface from 8 to 24 °C and at 10 cm
depth it varied from 14 to 18 °C. At 25 cm depth, it remained
around 18 °C ± 1 °C. We corrected calculated resistivity values
using the formula given by Keller and Frischknecht (1970) at a
reference of 25 °C, expressed as follows:
qd;25- C ¼ qd;Td ⁎ð1 þ 0:025⁎ðTd 25ÞÞ
ð1Þ
where ρd,25°C is the calculated resistivity at depth d (in Ω m),
corrected for a reference temperature of 25 °C, ρd,Td is the
calculated resistivity at depth d (in Ω m), at temperature Td, and
Td is the temperature at depth d during the measurement (in °C).
For a 12 °C variation on the surface, a resistivity variation of
30% is expected, showing that the correction is compulsory.
For resistivity probes, the electrodes were nickel-made rings
(diameter 1 cm), equally spaced (2 cm) along a plastic tube. We
calculated the geometric coefficient necessary to get bulk
resistivity of soil in the laboratory using a tank filled with water
of known resistivity. The location of the two electrical probes with
16 electrodes is shown in Fig. 1. They were placed within the 2
slopes as shown in Fig. 2. We performed these measurements
using pole-pole array just before and after each ERT acquisition.
From these repeated logging measurements, it was possible to
ascertain that there was no resistivity change (within 2%
deviation) during ERT acquisition. Therefore 6 min of surface
ERT acquisition provided a snapshot of resistivity distribution in
the ground.
Fig. 3. Resistivity monitored at depth with electrical probes on the windward
side for Rain 2 and Rain 3, after each ERT surface acquisition. Resistivity values
were corrected with temperatures recorded at the same depth and time. Note the
sharp variations in resistivity between 15 and 21 cm deep and invariant
resistivity below 27 cm.
27 cm and below, there were no variations in the resistivity, and
therefore, this zone was considered as invariant.
3. Results
3.2. ERT inversion optimization
3.1. Resistivity logging
An example of the results logged with the electrical probe on
the windward side is shown in Fig. 3. The trend was similar on
the leeward side. We corrected the data using temperature (at
25 °C reference). The result shown was obtained for Rain 2
(demineralized water) and Rain 3 (salted water).
Prior to Rain 2, resistivity varied between 400 Ω m (at 7–
11 cm) to more than 700 Ω m (at 27 cm, and deeper). A sharp
variation in resistivity was noted between 15 and 21 cm. Just
after Rain 2, the resistivity decreased down to a depth of 15 cm.
One hour later, just before Rain 3, the resistivity further
decreased down to a depth of 21 cm, indicating that water had
penetrated deeper. The lowest value, 320 Ω m, was measured at
9 cm (decrease of 20%). This decrease is significant for
understanding the hydrological process, hence ERT must be
sufficiently reliable to monitor such 20% resistivity changes.
For Rain 3, variations were recorded down to a depth of
27 cm. Because the salt lowered the water conductivity, larger
decreases in resistivity were measured in the upper part of the
soil compared to Rain 2, and reached 50% at 3 cm depth (400 to
200 Ω m). Resistivity variations in the lower part of the profile
(below 17 cm) did not exceed 20%. This range is almost equal
to the decrease noted for Rain 2 (demineralized rainwater). At
To optimize ERT inversion we used the Rain 3 data set
because it showed the highest resistivity variations. The initial
resistivity model for each data set is defined as a homogeneous
medium with a mean resistivity calculated from all apparent
resistivity data. We did not use the time-lapse mode proposed by
Loke (1999), which uses the final model obtained for inversion
of an initial data set as the starting solution for the second data
set inversion. Thus we avoided any inter-dependence with an
initial model.
We used a 3 step approach. Firstly, we fixed grid geometry as
shown in Fig. 2. Secondly, we adjusted several inversion
parameters to standard values that remained invariant for all
data set inversions. The initial damping factor was set to 0.15,
applicable to a low-noise data set, to optimize reconstruction
smoothing according to resistivity probe data that showed smooth
variations down to the substratum. We set the flatness ratio to 1 to
avoid resistivity patterns oriented in the vertical or horizontal
direction. Previous tests on other parameters such as the mesh
size, reduction of side effects or Jacobian calculations were not
considered because their effect was not significant (i.e. different
values gave similar results when compared to resistivity probe
data). Thirdly, by fixing the previous parameters, we focused on
other important parameters that are known to contribute
88
Fig. 4. Comparison between resistivity values logged on the leeward and windward sides and the calculated resistivity given by inversions, just after Rain 3. The effects
of different inversion parameters are shown. A) Effect of incorporating an invariant substratum (zone with a constant resistivity of 650 Ω m). B) Effect of the iteration
number. The gray arrows indicate notable discrepancies from measured data.
significantly to calculated resistivity patterns, some of which were
previously investigated in other studies. These parameters are the
number of iterations (Olayinka and Yaramanci, 2000a), the
incorporation of a priori information (i.e. regions with fixed
resistivity), the smooth or blocky inversion mode (Olayinka and
Yaramanci, 2000b; Loke et al., 2001), and the type of
topographical modeling. The results of inversions were compared
to resistivity probe data from both the leeward and windward
sides. To demonstrate inversion optimization, two examples are
presented in Fig. 4 to illustrate the effect of these parameters: a)
incorporation of a substratum of known resistivity, and b) varying
the iteration number. Both are compared to the resistivity probe
data to give an estimate of ERT accuracy. Although not included
in Fig. 4, the two other parameters also had a significant impact on
inversion and were also tested once the first two parameters were
chosen.
3.2.1. Incorporating a priori information
As shown previously in Fig. 3, resistivity did not vary below
27 cm, hence we decided to incorporate a flat substratum of
invariant resistivity. We chose a mean resistivity of 650 Ω m.
We applied the fifth iteration because it varied by less than 0.5%
from the fourth iteration. Without setting an invariant
substratum, the ERT inversion produced for the leeward side
was a homogeneous layer of 300 Ω m between 10 and 30 cm
deep. However, this geometry does not fit well with the pattern
Fig. 5. Evaluation of ERT accuracy. The ratio of measured resistivity (final/initial) from electrical logging before and after Rain 3 is represented with a continuous line.
The ratio of calculated resistivity derived from optimized inversion is represented with a dotted line. The result using 5 iterations is also represented by a gray
continuous line to highlight severe discrepancies with actual data.
Fig. 6. A) Calculated resistivity using optimized inversion for 6 ERT surface data sets. Two rains (2 and 3) are considered only. The blocks are drawn according to the block arrangement shown Fig. 2. B) Ratio of calculated
resistivity (after/before). The decrease in resistivity (values below 1) is detailed using a range of cold colors. The increase in resistivity is shown with 2 classes in order to simplify the figure. The locations of tensiometers are
shown to facilitate the location of results shown in Figs. 11 and 12.
89
90
seen with electrical probe data. By including the invariant
substratum, the agreement with electrical probe data improved.
However, this inversion remained imperfect, as can be seen for
the windward side for instance: the calculated resistivity has
shifted significantly from the logged values.
3.2.2. Effect of iteration number
As shown in Fig. 4, the best inversion was obtained for the
windward side following two iterations (RMS 3.4%). Conversely, iteration 5 (better RMS, 1.65%) generated an inversion
result significantly different from the actual data. Iterations 3
and 4 show intermediate values. Thus, if the number of
iterations is set too high, the inversion tends to lead to a decrease
in RMS by using resistivity values that are far from the actual
range determined with electrical probe data. Even though the
RMS value decreased, the inversion extended too far. Thus we
used iteration 2 even if the RMS was higher than subsequent
iterations. Of note, even iteration 2 values did not match
perfectly with the electrical probe data. For instance, some sharp
increases in resistivity measured on the leeward side at a depth
of 20 cm were not represented in the iteration 2 inversion.
Lastly, using the same approach, we tested the effect of
smooth versus blocky mode and several types of topographical
correction (not shown here).
The best results were obtained using a smooth (non-blocky)
mode and topographical correction with the Schwarz–Cristoffel
option provided by RES2DINV. This topographical method
calculates the distortion in the subsurface layers in cases with a
large topography curvature (Loke, 2000b). Finally we collated
all the parameters that led to the best agreement with the
electrical probe data into a unique parameter set for the entire
experiment.
3.2.3. ERT accuracy estimate
We analyzed the resistivity ratio calculated before and after
rain. By using this ratio we could determine ERT accuracy more
efficiently. We compared resistivity ratios measured with
logging and those calculated from optimized ERT. This
comparison is shown in Fig. 5 for data from both the leeward
and windward sides. For the leeward side, the optimized
inversion gave an acceptable result from the surface down to
23 cm, with a small deviation value of − 0.12 between 5 and
12 cm and +0.10 between 12 and 23 cm. Below 23 cm the
deviation is higher, at + 0.25, showing an increase in resistivity
that was not measured by electrical probe data. Thus our findings
demonstrate that at depth, ERT did not reproduce accurately
resistivity variations, for data from the leeward side. On the
windward side, we found deviation values of + 0.16 between 5
and 17 cm, and − 0.17 between 17 and 24 cm. In summary, the
optimized inversion produced accurate results close to the actual
resistivity variations, with deviations mostly between − 0.1
(− 10%) and +0.15 (+ 15%). On the leeward side, at depths
below 25 cm, an unrealistic increase in resistivity appears to
have occurred. Without optimization, ERT results would have
been useless. As an illustration, if we consider results obtained
with the same set of parameters, but for the fifth iteration (with an
improved RMS) as seen in Fig. 5, severe discrepancies are noted
for data regarding the leeward side (0.25 between 10 and 20 cm)
and regarding the windward side: in this area the calculated
resistivity ratio reaches 1.5 at 10 cm deep (+ 0.75 from the actual
value 0.72) which is a false increase in resistivity.
3.3. Surface ERT
We chose data from Rains 2 and 3 to illustrate ERT results in
Fig. 6. For Rain 1, incomplete ERT results were obtained before
rain 1 due to bad contact resistance for some electrodes and are
not shown. To facilitate comparisons between rain and evaporation phases, resistivity ratios between ERT acquisitions (after/
before) are shown in Fig. 6. Values below or above 1 indicate a
decrease or an increase in resistivity, respectively.
Just prior to Rain 2 the calculated resistivity ranged from 250
to more than 1000 Ω m. The highest resistivities were recorded
on top of the microdune, while at the center of the microdune
Fig. 7. Water tension variations on leeward and windward sides during and after Rain 2.
they ranged from 300 to 600 Ω m. Below 26 cm , resistivities
were very close to 650 Ω m which is the value measured for the
invariant substratum.
Just after Rain 2, the calculated resistivity values were lower
close to the surface, both on the windward and leeward sides
(ranging from 200 to 300 Ω m), and the number of highly resistive
cells was reduced at the top of the microdune. At the center,
calculated resistivities remained above 300 Ω m. The resistivity
ratios showed a general decrease, for example a 40% decrease on
the windward side was seen up to a depth of 12–15 cm.
Following 1 h of evaporation, just before Rain 3, there were
more resistive cells on top of the microdune, while at the center
values between 300 and 450 Ω m were recorded. Resistivity ratios
showed an increase from the surface down to 10 cm at the center.
A decrease of −5 to −15% was found below that. Just after Rain 3
the lowest resistivities were located on the surface (40–100 Ω m).
Resistivity ratios indicated i) a major decrease down to 8 cm at the
center and on the leeward side (−90%), ii) a small decrease
(−15%) on the windward side at a depth of 10–15 cm, and iii) a
moderate decrease (−30%) at the bottom of the microdune. We
also noted a slight increase on the leeward side at the depth where
accuracy in the ERT optimized inversion falls (see Fig. 5).
One hour after Rain 3, calculated resistivities ranged from 90
to 500 Ω m. The resistivity ratios indicated a general increase in
resistivities (up to + 75%) between the surface and down to
10 cm, below which there was no noticeable changes.
3.4. Soil water tension
Fig. 7 shows an example of soil water tension measured
during and after Rain 2. For Rain 1 and 3 the trends were
similar. We focused on the windward side using tensiometers 6,
7, 8, 4 and 5 and on the leeward side with tensiometers 1, 2, 3
91
and 9. At the initial stage before Rain 2, water tension ranged
from − 1.2 m near the surface to − 0.6 m at 15 cm deep. At the
end of Rain 2 (40 min) the water tension was close to or equal to
0 near the surface (saturation) while at 20 cm it remained the
same as at the initial stage. One hour after Rain 2 (just before
Rain 3), the water tension near the surface had started to
increase with evaporation. It decreased below 7 cm indicating
drainage. This trend was also found for the leeward side, but
evaporation started at a slightly deeper level of 10 cm. Evidence
of evaporation and drainage on the leeward and windward sides
after Rain 2 is in agreement with ERT results that showed a
resistivity increase at shallow depths, up to 12 cm, and a
decrease of resistivity below it. The slight difference between
the slopes (evaporation taking place at a deeper level on the
leeward side) is also seen with ERT in Fig. 6.
3.5. Porosity and particle size distribution
The porosity and particle size distribution obtained with soil
samples using 5 value ranges are shown in Fig. 8. Total porosity
ranged from 27% to 52%. The highest porosities were generally
encountered near the surface of the leeward side (45–52%)
where root density was higher also.
Particle size distribution showed that the percentage of sand
(50–2000 μm) was between 50 and 80%, while the percentage
of silt (2–50 μm) was between 16 and 45%. No particular
spatial distribution was noted for these two classes of particles.
The percentage of clay (particles smaller than 2 μm) ranged
from 2 to 20%. The highest percentage, 20%, was found in an
isolated section of the leeward side at a depth of 6 cm. Apart
from this value, the other values range from 2 to 9% and the
highest amounts of clay were found in samples from the bottom
of the microdune. The higher values correspond to the clayey
Fig. 8. Soil sampling results following the sampling grid shown Fig. 2. Porosity was not calculated in the top surface because of a lack of soil when the samples were
made. The percentages of sand, silt and clay material are plotted according to 5 classes.
92
function expresses volumetric soil water content θe as a
function of water potential ψ as follows:
he ¼ ½1 þ ð agwÞnm
Fig. 9. Experimental data fitted using the Van Genuchten's retention curve
function (Van Genuchten, 1980). The Van Genuchten's function using the
parameters α = 1.45, m = − 1, and n = 2.68 (see Eq. (2) selected by Carsel and
Parrish (1997) for typical sandy soils is also drawn. Water tension is expressed
here in centimeters (cm), and is negative. It has been converted into absolute
values for using a logarithmic scale.
horizon of the microdune substratum. Porosity distribution was
not clearly correlated with particle size distribution. The large
range of porosity (27 to 52%) indicates a highly heterogeneous
soil. This may have had a significant effect on the resistivity
values, which also depend on porosity.
3.6. Relationship between tension and water content
In Fig. 9 values for water tension measurements made with
tensiometers, were plotted with respect to saturation. Experimental data were fitted using the Van Genuchten's retention
curve function (Van Genuchten, 1980). Van Genuchten's
ð2Þ
where αg is conceptually the inverse ψb [m− 1], ψb being the
bubbling pressure, and n and m the fitting coefficients (with
m = 1 − (1 / n). The retention curve function has been extended
in Fig. 9 to high water tension values for demonstration
purposes only. In Fig. 9, we also added Van Genuchten's
function using parameters (α = 1.45, m = − 1; n = 2.68) selected
by Carsel and Parrish (1997) for typical sandy soils. Although
the experimental fit remains close to a typical curve for sandy
soils, our data points are scattered, and therefore this curve
could not be used to derive further conclusions.
4. Discussion
4.1. Relationship between ERT resistivity and water tension
The relationship between ERT resistivity and water tension
measurements is shown in Fig. 10, taking Rain 2 as an example.
3 points were drawn for each tensiometer. They correspond to
the values of the resistivity–water tension couple taken before
and just after Rain 2, and following 1 h of evaporation. The
results show that there is no correlation between resistivity and
water tension, with a scattered relationship. Data from
tensiometers 1, 2, 3, 6, 7 located near the surface show the
same behavior: decreases in water tension were accompanied by
decreases in resistivity. During evaporation, an inverse trend
occurred where water tension and resistivity increase. Measurements made with tensiometer 4, 5, 8 and 9 located at a deeper
level, showed an initial decrease in resistivity while water
Fig. 10. The raw relationship between resistivity calculated with inversion of surface ERT and water tension measured during the Rain 2 cycle. Two groups are evident:
tensiometers 1, 2, 3, 6 and 7 near the surface exhibit one class of similar behavior, and tensiometers 4, 5 8 and 9 at a deeper level exhibit a second class of another
behavior (dotted lines), as described in the text.
tension remained almost constant (tensiometer 5) or increased
slightly. When water tension increased significantly, resistivity
did does not show much variation. However, as shown by the
results of the optimization, resistivity is not accurately
calculated at depth.
The absence of a correlation between ERT resistivity and
water tension means that the relationship cannot be used for any
spatialization. We thus investigated the cause of the scattered
relationship between ERT resistivity and water tension by
examining it in relation to heterogeneous distribution of
porosity. The empirical Archie's law (Archie, 1942) is
expressed as follows (see Keller, 1988):
qg ¼ a:
qw :Um
Swn
ð3Þ
where ρg is the resistivity of the ground given by geophysical
measurement (in Ω m), a is the saturation coefficient, ρw is the
resistivity of the soil solution (in Ω m), Φ is the porosity of the
ground (dimensionless), m is the cementation factor (dimensionless), Sw is the saturation (volume of pore filled by water,
dimensionless), and n is the saturation exponent (dimensionless). The validity limit of this law is restricted to sandy
formations. Parameters a, m and n are taken as 1.37, 0.88 and 2
respectively for sand, from Keller (1988).
93
In our case, for a given soil solution resistivity and
saturation, Archie's law shows that when the porosity increases
by 15% the resistivity decreases by 40%. In the field, porosity
varies by at least 15% from place to place. Therefore variations
in resistivity of 40% are expected in the microdune due to
porosity differences. Hence, in order to improve the relationship
it is necessary to avoid porosity effects.
4.2. Relationship between resistivity ratio and water tension
difference
To avoid the porosity effect we used the resistivity ratio
instead of resistivity itself. Indeed, as deduced from Eq. (3), the
resistivity ratio between two hydrological states (for example
before and after rainfall) depends only on the variation of soil
solution resistivity and saturation. In order to compare water
tension between two states, we used water tension differences
rather than ratios. By determining the difference between two
hydrological states we could evaluate water movement: a
positive difference indicates an increase of water content while a
negative difference indicates evaporation.
We compared the resistivity ratio (final over initial state) to
water tension differences (final minus initial state). On the same
graph in Fig. 11 we considered both Rain 1 (evaporation phase
only) and Rain 2 with demineralized water. The water tension
Fig. 11. Comparison of resistivity ratio (final/initial) calculated from surface ERT results and water tension differences (final-initial) for Rain 1 and 2. Data from Rain 1
(evaporation phase only) and Rain 2 (rain and evaporation phases) are plotted with different symbols. For rain 1, points from tensiometers 9 and 5 (in gray) appear doubtful
and were not included for fitting the data. ERT deviation bars are deduced from ERT accuracy analysis shown in Fig. 5. Water tension deviation bars are deduced from an
inter-comparison of data between tensiometer 3 and 8 (not shown here). The doted line indicates the fitting (equation and R-squared criteria given in legend).
94
measured at the end of Rain 1 from tensiometers 5 and 9 was not
reliable due to bad readings and was omitted.
For resistivity ratios, we added deviation bars taking into
account the limited accuracy of the ERT shown in Fig. 5. The
following values were used: ± 0.05 for leeward (tensiometers 1,
2 and 9) and ± 0.075 for center and windward (tensiometers 3, 4,
5, 6, 7 and 8).
For water tension, the measurement errors were very small
(within 2% deviation) and not taken into account. But when
tensiometers 3 and 8, which were situated close together, were
simultaneously compared, a difference of ± 0.05 m (not dealt in
this paper) was found, probably due to local variations in soil
properties around the sensor. Consequently, we included a
possible variation of ± 0.05 m as a deviation bar for water
tension values.
We found that the relationship followed a logical trend: the
lower the resistivity ratio, the higher the water tension
difference. The relationship was adjusted by conversion to a
logarithmic relationship with a coefficient (R-squared) of 0.79.
The relationship obtained is well suited for tensiometers close to
the surface: ERT is able to delineate the tension differences with
confidence within the first 15 to 20 cm from the surface. We
noticed also that a few group of points deviated significantly
from the logarithmic relationship. For example, points for the
tensiometers 4, 5 and 9 for Rain 2 at depth show resistivity ratio
values that are too low to fit with the logarithmic relationship.
Therefore, the accurate spatialization of water tension differences remains difficult for the deeper parts of the microdune.
We have shown that a porosity-independent relationship can
be established between the calculated resistivity ratio and water
tension difference. But limitations to ERT reconstruction can
lead to errors that remain troublesome for reliably recovering
detailed water tension differences, especially at depth and this
problem was also reported by Day-Lewis et al. (2005). They
showed how core-scale relationships between geophysical
properties and hydrologic parameters are altered by the
inversion, which produces smoothly varying pixel-scale
estimates. Using synthetic examples, they proposed an
approach using cross-bore well ERT and radar tomograms to
delineate the patterns of correlation loss in the inversion. For
future work on sandy soils, this approach could be used if the
scale of the experiment allows radar antennae to be used. In our
case the microdune was too small to use high frequency radar
because the size of the transmitter-receiver system is too large
(20–30 cm) compared with the size of the dune (50 cm).
4.3. Effect of soil solution resistivity
When looking at the relationship between resistivity ratio
and water tension difference for Rain 3, salted with NaCl, there
was a wide range of variation for resistivity ratios (0.1 to 1.8) as
shown in Fig. 12. The range of differences in water tension was
Fig. 12. Comparison of resistivity ratio (final/initial) calculated from surface ERT results and water tension differences (final-initial) for the salted Rain 3 (rain and
evaporation phases). See also description of Fig. 11.
low (− 0.1 to + 0.2 m) because of the short periods of
evaporation considered in the experiment. A logarithmic
relationship was hardly achieved with a coefficient (R-squared)
of 0.30 that improved slightly (0.45) if we remove tensiometers
4 and 5. The value obtained from tensiometers 4 and 5, and to
some extent tensiometers 8 and 9, located deeper, deviate from
the logarithmic relationship. We also observed that points
obtained with tensiometers 4 and 5 are situated closer to the
previous curve determined for Rains 1 and 2 with demineralized
water. This suggests that perhaps at the end of rain 3, salted
water had not yet reached the deeper part of the windward side,
and therefore the resistivity ratio remained low.
To highlight the effect of soil solution resistivity in the
experiment, we investigated its influence using Archie's law.
We plotted soil solution resistivity versus soil resistivity in Fig.
13 taking into account a constant porosity of 40% and various
saturation states from 0.25 to 1 (25 to 100%). This illustrates
how variations in water resistivity can be tracked using
variations in soil resistivity. Starting an experiment at 25%
saturation, with a soil solution resistivity of 40 Ω m, the
corresponding soil resistivity is 2000 Ω m (point number 1 in
Fig. 13). If we proceed until saturation state, 3 scenarios are
possible:
• Scenario A: the water used in the experiment is demineralised, which reproduces a typical scenario of rain infiltrating
soil. The soil solution is diluted during infiltration leading to
increased resistivity. Once the soil gets saturated, point 2 is
95
reached (Fig. 13). The corresponding soil resistivity ratio
(final/initial) is 0.3 for scenario A.
• Scenario B: water resistivity in the experiment is close to the
actual soil solution resistivity. If we assume that there are no
chemical exchanges in the soil then the soil solution should
remain constant. Once the soil becomes saturated, point
number 3 is reached, and the soil resistivity ratio reaches
0.06.
• Scenario C: water used for the experiment is more salty than
the actual soil solution. It leads to a concentration (decrease
of soil solution resistivity), and the soil resistivity ratio is
0.015. A huge contrast in soil resistivity between initial and
final states is created.
If a direct link between water tension and saturation is
assumed, then dilution (scenario A) will produce a highly
sloped relationship between resistivity ratio and tension
difference (small ratio of resistivities for a given saturation
difference). On the contrary, concentration (scenario C)
generates a relationship with a low slope (higher ratio of
resistivities for the same saturation difference). Moreover in
case C soil resistivity would drastically decrease and that could
be troublesome for calculations using ERT.
Our experimental results showed that the relationship
obtained between resistivity ratio and water tension difference
depends on soil solution resistivity as predicted by Archie's law.
The slope of the relationship depends on the situation created,
i.e. dilution or concentration of the soil solution. In the case of
Fig. 13. Example of Archie's law for a sandy soil with a homogeneous sand with 40% porosity and parameters a, m and n fixed at 1.37, 0.88 and 2 respectively. Soil
solution resistivity was plotted versus soil resistivity for different saturation states. Points 1 to 4 correspond to situations described in the text.
96
Rain 3, the addition of the salt modified the slope obtained from
Rain 1 and Rain 2.
Recent studies have also used Archie's law in their
investigations. Singha and Gorelick (2005, 2006) illustrated,
with both synthetic and field cases, that ERT response is
difficult to match with measured fluid conductivities in
saturated soil. This difficulty is due to the variability in the
effects of ERT regularization, which change in both space and
time. For this reason, they suggest that Archie's law cannot be
used to directly scale ERT conductivities to fluid conductivities.
Our findings agree with their conclusion. We can add that for
unsaturated soils, the scaling of ERT conductivities to water
tension could be difficult if the soil solution conductivity is not
constant in time. In other words, if several sprinkling tests are
made with water of different conductivity, a clear relationship
between resistivity ratio and water tension differences could not
be identified. Therefore, we advocate the use of constant water
conductivity for repeated infiltration tests monitored with ERT.
Water conductivity changes are maintained in the same value
range, minimizing the slope change in the relationship. When
salted water was used in our study, the range of variation in the
resistivity values was found to be too great compared to the
range of tension differences (the time period of evaporation was
too short). Therefore the relationship was not reliable. For
further experiments with salted water, longer evaporation
periods should be allowed for, to generate a wider range of
tension differences and thus enhance the reliability of the
relationship.
4.4. Spatialization of water content using ERT
Spatializing water content using ERT remains a tenuous
objective in our study. This important goal was not considered
essentially because the experimental relationship between
tension and water content, sketched in Fig. 9, is not linear if
we fit it using a conventional Van Genuchten's model well
suited for sands. This model is non linear and therefore, an
identical difference in water tension leads to different water
tension differences depending on the range of water content
considered. In the middle water content range, between 20%
and 85%, the relationship should be approximately linear. In the
upper range, between 85% and 100% (saturation), the same
linear trend was observed in the model with a slightly different
slope, but was not seen for our data because we had too few data
points for this water content range. Thus for an identical water
tension difference, say 0.1 m (10 cm), the difference in water
content is 5% in the middle range and 3.3% in the upper range.
We conclude that even if ERT could be useful for accurately
spatializing the water tension difference, the conversion of
results obtained with ERT to determine water content would
only lead to approximations of the water content restitution
which is dependent on the water content range, not known from
ERT results. Nevertheless, for soils, or a specific water content
range, where a linear behavior can be ascertained, the
conversion of ERT to determine water content (through an
experimental relationship derived from tension measurements)
would be feasible.
A second difficulty could arise when using ERT to estimate
the water content distribution. Several studies reported a
hysteresis behavior in the experimental relationship between
resistivity and water content obtained on soil samples in
laboratory experiments. In other words, the resistivity was not
the same for a known identical water content depending on
whether the sample was being wet or dried. This case was also
encountered in our field study: water was infiltrating the upper
part of the dune at the same time as drainage was occurring in
the lower section. Knight (1991) as well as Roberts (2002)
suggest that hysteresis should be taken into account when
measuring vadose zone resistivity. The saturation history of the
soil is therefore a factor that complicates the use of ERT. It was
not addressed in this study essentially because we found that
inaccuracies in reconstructing the ERT were by far the most
important sources of error which must be resolved prior to
considering resistivity- water content hysteresis, if indeed it
exists here.
5. Conclusion
In this field study, we compared the soil electrical resistivity
with water tension measurements under a series of 3 simulated
rainfall events, with the aim of using electrical resistivity as a
parameter to facilitate spatialization of hydrological processes
(infiltration, evaporation) within a typical sahelian microdune.
For this, the usefulness of surface ERT for examining the
distribution of soil resistivity in the microdune, with heterogeneous sandy soils, was evaluated.
Our first set of conclusions stem from attempts to optimize
ERT inversion.
• ERT inversions cannot be conducted routinely and confidently using default parameters, especially if the RMS is
considered as the ultimate parameter for deciding if the
inversion is accurate enough or not. ERT results were found
to vary considerably depending on the different sets of
inversion parameters and this can result in an under- or overestimate of the resistivity.
• We generated an optimized inversion using a comparison
between inverted and in-situ resistivity. During the optimization process, we highlighted that the comparison is strongly
influenced by the number of iterations and the incorporation
of a priori information. The best results were obtained with
only 2 iterations. After the third iteration, the calculations
lead to inaccurate calculated resistivity. Our results show that
an invariant zone of constant resistivity (the substratum in
our case) is instrumental in obtaining more reliable results
because the inversion is confined within strict geometrical
limits that are otherwise overtaken.
• Optimized ERT inversion results can be compared to actual
resistivity data to provide an estimate of their accuracy. We
found that the variations in actual resistivity were small (20%)
during the first two rainfall events with demineralized water,
but higher (up to 50%) during the last salted rainfall event. The
calculated resistivity values generated with the optimized ERT
were within the same value range , but a few discrepancies still
occurred for data collected at depth. We estimated the ERT
accuracy using the resistivity ratio (final state/initial state), and
found it to be 10 to 15% of the actual ratio, but it may be higher
(25%) at depth on the leeward side. We attribute this
inaccuracy to inversion smoothing. Consequently, localized
variations could not be precisely determined, especially where
there were sharp contrasts in resistivity.
Thus, in this study, for the ERT reconstruction to generate
more reliable estimates of the resistivity, in-situ measurements
are needed to facilitate comparisons between calculated and
actual resistivity data. We found that this difficulty limits the
interest of using non-destructive surface ERT measurements.
Our second set of conclusions derives from comparisons
between ERT and water tension measurements at a few
locations within the microdune.
• There was no correlation between ERT calculated resistivity
and water tension during rain and evaporation. We explained
this lack of correlation using soil-sampling results which
showed that porosity was very heterogeneously distributed.
• We found a clear correlation between the resistivity ratio
(final state over initial state) and the difference in water
tension (final state minus initial state), using the data for Rain
1 and 2 (both with demineralized water). As predicted by
Archie's law, this relationship is porosity-independent.
When the resistivity ratio decreases, the water tension
difference increases.
• Time lapse ERT is able to confidently track water tension
differences within the first 20 cm from the surface. However,
below 20 cm, we noticed a lack of ERT accuracy that made it
difficult to recover the distribution of water tension
differences in some parts of the microdune because sharp
contrasts in resistivity have been smoothed out.
• We used salted rain to examine how the relationship
(resistivity ratio–tension difference) may vary with soil
solution conductivity. The slope of the relationship changed
as predicted by Archie's law. The dependent relationship
with soil solution resistivity was clearly seen in our field
experiment but could not be reliably determined.
• Lastly, we could not derive water content using the resistivity
ratio–water tension difference experimental relationship
because of the non linear behavior between tension and
water content. Moreover, we found that this experimental
relationship was scattered, but further studies may take this
into account.
We can conclude from our findings that the relationship
between resistivity ratio and tension differences should be
investigated in more detail under various experimental conditions. To obtain less scattered results future work should
concentrate on point resistivity measurements coupled with
tensiometers at the same place. Nevertheless even if the
relationship may be improved, the effect of ERT smoothing
could still compromise efforts to reconstruct a detailed picture
of localized processes at depth. The inclusion of a priori
information is essential for optimizing ERT inversion. This was
97
also reported in others recent studies, for example by Yeh et al.
(2002) who used a sequential, geo-statistical inverse approach
with inclusion of a priori knowledge and point electrical
resistivity measurements, that also permitted sequential inclusion of different data sets. In our study we found that even if a
relation between the resistivity ratio and water tension
differences could be established, a more reliable ERT
reconstruction is also needed to foresee a valuable spatialization
of hydrological functioning of heterogeneous sandy soils using
resistivity.
Acknowledgements
This work was funded and conducted by Unités de
Recherche 027 and 049 of Institut de Recherche pour le
Développement (IRD), and INSU Programme National SolErosion (PNSE) project no. 99/44. We greatly acknowledge
Institut National de l'Environnement et de la Recherche
Agricole (INERA) of Burkina Faso for providing access to
the site. We also thank the team of the hydrological laboratory
of IRD at Ouagadougou, with a special mention of Moussa
Barry, Yves Dzouali, Harouna Karambiri, Dial Niang, Boureima Tou and Maxime Wubda for their help in the field.
References
Archie, E., 1942. The electrical resistivity log as an aid in determining some
reservoir characteristics. Amer. Inst. Min. Met. Eng., Tech. Publ. 1422.
Petroleum Technology. 8 pp.
Asseline, J., Valentin, C., 1978. Construction et mise au point d'un infiltromètre
à aspersion. Cahiers de l'ORSTOM. Série Hydrologie 15, 321–343.
Carsel, R.F., Parrish, R.S., 1997. Developing joint probability distributions of
soil water retention characteristics. Water Resources Research 24, 755–769.
Casenave, A., Valentin, C., 1992. A runoff capability classification system based
on surface features criteria in semi-arid areas of West Africa. Journal of
Hydrology 130, 231–249.
Daily, W., Ramirez, A., LaBrecque, D., Nitao, J., 1992. Electrical resistivity
tomography of vadose water movement. Water Resources Research 28 (5),
1429–1442.
Day-Lewis, F.D., Singha, K., Binley, A.M., 2005. Applying petrophysical
models to radar travel time and electrical resistivity tomograms: resolutiondependent limitations. Journal of Geophysical Research 110, B08206.
doi:10.1029/2004JB003569, 2005.
Depountis, N., Harris, C., Davies, M.C.R., 2001. An assessment of miniaturised
electrical imaging equipment to monitor pollution plume evolution in scaled
centrifuge modelling. Engineering Geology 60, 83–94.
French, H.K., Hardbattle, C., Binley, A., Winship, P., Jakobsen, L., 2002.
Monitoring snowmelt induced unsaturated flow and transport using
electrical resistivity tomography. Journal of Hydrology 267 (3–4), 273–284.
Keller, G.V., 1988. Rock and mineral properties. In: Nabighian, M.N. (Ed.),
Electromagnetic methods in applied geophysics, 1. SEG Publ., pp. 13–51.
Theory, Chap. 2.
Keller, G.V., Frischknecht, F.C., 1970. Electrical methods in geophysical
prospecting. International series of monographs in electromagnetic waves,
vol. 10. Pergamon Press.
Kemna, A., Kulessa, B., Vereecken, H., 2002. Imaging and characterisation of
sub-surface solute transport using electrical resistivity tomography (ERT)
and equivalent transport model. Journal of Hydrology vol. 267 (3–4),
125–146.
Knight, R., 1991. Hysteresis in the electrical resistivity of partially saturated
sandstones. Geophysics vol. 56 (12), 2139–2147.
Liu, S., Yeh, T.C.J., 2004. An integrative approach for monitoring water
movement in the vadose zone. Vadose Zone Journal vol. 3, 681–692.
98
Loke, M.H., 1999. Time-lapse resistivity imaging inversion. Proceedings of the
5th Meeting of the Environmental and Engineering Geophysics Society,
European Section, Em1, Budapest, Hungary.
Loke, M.H., 2000a. Electrical imaging surveys for environmental and
engineering studies. Technical note of Geotomo Software. Available on
Loke, M.H., 2000b. RES2DINV, Rapid 2-D Resistivity and IP inversion using
the least-squares method. Technical note of Geotomo Software. Available on
Loke, M.H., Acworth, I., Dahlin, T., 2001. A comparison of smooth and blocky
inversion methods in 2D electrical imaging surveys. Proceedings of the 15th
ASEG Geophysical Conference and Exhibition, Brisbane. Available on line
at www.geoelectrical.com.
Michot, D., Benderitter, Y., Dorigny, A., Nicoullaud, B., King, D., Tabbagh, A.,
2003. Spatial and temporal monitoring of soil water content with an irrigated
corn crop cover using surface electrical resistivity tomography. Water
Resour. Res. 39 (5), 1138. doi:10.1029/2002WR001581.
Olayinka, A.I., Yaramanci, U., 2000a. Assessment of the reliability of 2D
inversion of apparent resistivity data. Geophysical Prospecting 48, 293–316.
Olayinka, A.I., Yaramanci, U., 2000b. Use of block inversion in the 2-D
interpretation of apparent resistivity data and its comparison with smooth
inversion. Journal of Applied Geophysics 45, 63–81.
Ribolzi, O., Auque, L., Bariac, T., Casenave, A., Delhoume, J.P., Gathelier, R.,
Pot, V., 2000. Ecoulements hypodermiques et transferts de solutés dans les
placages éoliens du Sahel: étude par traçage isotopique et chimique sous
pluies simulées. Comptes rendus de l'AcadeÂmie des sciences Paris 330,
1–8.
Richards, L.A., 1931. Capillary conduction of liquids in porous medium.
Physics 1, 318–333.
Roberts, J.J., 2002. Electrical properties of microporous rock as a function of
saturation and temperature. Journal of Applied Physics 91 (3), 1687–1694.
Singha, K., Gorelick, S.M., 2005. Saline tracer visualized with threedimensional electrical resistivity tomography: field-scale spatial moment
analysis. Water Resources Research 41, W05023. doi:10.1029/
2004WR003460.
Singha, K., Gorelick, S.M., 2006. Effect of spatially variable resolution on fieldscale estimates of tracer concentration from electrical inversions using
Archie's law. Geophysics 71 (3), G83–G89.
Slater, L., Binley, A.M., Daily, W., Johnson, R., 2000. Cross-hole electrical
imaging of a controlled saline tracer injection. Journal of Applied
Geophysics 44, 85–102.
Telford, W.M., Geldart, L.P., Sheriff, R.E., 1990. Applied Geophysics, 2nd ed.
Cambridge University Press.
Thiombiano, L. 2000. Etude de l'importance des facteurs édaphiques et
pédopaysagiques dans le développement de la désertification en zone
sahélienne du Burkina Faso. Thèse d'Etat de l'Université de Cocody,
République de Côte d'Ivoire.
Valentin, C., 1991. Surface crusting in two alluvial soils of northern Niger.
Geoderma 48, 201–222.
Van Genuchten, M. Th., 1980. A closed-form equation for predicting the
hydraulic conductivity of unsaturated soils. Soil Science Society of America
Journal 44, 892–898.
Yeh, J.C., Liu, S., Glass, J., Baker, K., Brainard, J.R., Alumbaugh, D.,
LaBrecque, D., 2002. A geostatistically based inverse model for electrical
resistivity surveys and its applications to vadose zone hydrology. Water
Resources Research 38 (12), 1278. doi:10.1029/2001WR001204, 2002.
Author's personal copy
C. R. Geoscience 341 (2009) 886–898
Internal geophysics (Applied geophysics)
Influence of shallow infiltration on time-lapse ERT:
Experience of advanced interpretation
Rémi Clément a,*, Marc Descloitres a, Thomas Günther b,
Olivier Ribolzi c, Anatoli Legchenko a
a
Laboratoire d’étude des transferts en hydrologie et environnement (LTHE), UMR 5564, CNRS, INPG, IRD,
UJF, université de Grenoble, BP 53, 38041 Grenoble cedex 9, France
b
Leibniz Institute for Applied Geophysics, Stilleweg 2, 30655 Hannover, Germany
c
Laboratoire des mécanismes de transfert en géologie (LMTG), IRD, CNRS, UPS, OMP,
14, avenue Edouard-Belin, 31400 Toulouse, France
Received 17 June 2008; accepted after revision 1 July 2009
Available online 26 September 2009
Written on invitation of the Editorial Board
Abstract
Previous time-lapse Electrical Resistivity Tomography (ERT) studies have experienced difficulties in reconstructing reliable
calculated resistivity changes in the subsurface. Increases or decreases of resistivity appear in the calculated ERT image where no
changes were noted in the subsurface, leading to erroneous hydrological interpretations of the geophysical results. In this article, we
investigate how a variation of actual resistivity with time and at shallow depth can influence time-lapse ERT results and produce
resistivity artefacts at depth. We use 1 and 2-D numerical modelling to simulate infiltration scenarios. Using a standard time-lapse
inversion, we demonstrate the resistivity artefact production according to the electrode spacing parameter. We used an advanced
inversion methodology with a decoupling line at shallow depth to attenuate or remove resistivity artefacts. We also applied this
methodology to a field data set obtained in a semi-arid environment in Burkina Faso, West Africa. Here, time-lapse ERT shows
several resistivity artefacts of calculated resistivity if a standard inversion is used. We demonstrate the importance of a dense
sampling of shallow resistivity variations at shallow depth. Advanced interpretation allows us to significantly attenuate or remove
the resistivity artefact production at intermediate depth and produce reliable interpretation of hydrological processes. To cite this
article: R. Clément et al., C. R. Geoscience 341 (2009).
# 2009 Académie des sciences. Published by Elsevier Masson SAS. All rights reserved.
Résumé
Influence des infiltrations superficielles sur le suivi temporel en tomographie de résistivité électrique : expérience
d’interprétation améliorée. Certaines études de suivi temporel par Tomographie de Résistivité Electrique (ERT) ont montré des
augmentations ou des diminutions de résistivité bien identifiées dans les images de résistivité calculée dans des zones où aucun
changement hydrologique n’a eu lieu. Nous montrons comment une variation réelle de la résistivité dans le temps et dans la proche
surface peut influencer les résultats de suivi temporel ERT et produire des resistivity artefacts. Nous utilisons des modèles
synthétiques 1-D et 2-D pour simuler des scénarios d’infiltration. L’utilisation d’une approche standard d’inversion en suivi
temporel montre la production de resistivity artefacts en fonction de l’écartement inter-électrode unitaire. Nous utilisons ensuite une
* Corresponding author.
E-mail address: remi.clement@hmg.inpg.fr (R. Clément).
1631-0713/$ – see front matter # 2009 Académie des sciences. Published by Elsevier Masson SAS. All rights reserved.
doi:10.1016/j.crte.2009.07.005
R. Clément et al. / C. R. Geoscience 341 (2009) 886–898
887
méthodologie d’inversion avancée qui apporte une information a priori en introduisant une ligne de découplage à faible profondeur
pour atténuer ou enlever les resistivity artefacts. Nous expérimentons cette méthodologie sur des données de terrain obtenues en
milieu semi-aride au Burkina Faso, Afrique de l’Ouest. À cet endroit, le suivi temporel ERT montre des resistivity artefacts
importants de variations de la résistivité calculée lorsqu’une inversion standard est utilisée. Nous mettons en avant l’importance
d’un échantillonnage dense de la variation et aussi que l’inversion avancée réduit de façon significative et même élimine les
resistivity artefacts à profondeur intermédiaire, pour aboutir à une meilleure description des processus hydrologiques. Pour citer cet
article : R. Clément et al., C. R. Geoscience 341 (2009).
# 2009 Académie des sciences. Publié par Elsevier Masson SAS. Tous droits réservés.
Keywords: Electrical resistivity tomography; Shallow infiltration; Resistivity artefact
Mots clés : Tomographie de résistivité électrique ; Infiltration superficielle ; Artefact de résistivité
1. Introduction
Thanks to their specialization and quantification
capacities and non-destructive character, geophysical
methods are often considered to help in implementing
point measurements to study hydrological processes.
Among them, Electrical Resistivity Tomography (ERT)
is a recent but mature geophysical method increasingly
popular in environmental and hydrogeological studies
[1–3]. ERT is well suited to 2-D and 3-D field data
acquisition and interpretation, and can be adapted to
various scales. Time-lapse ERT can also be used to
monitor changes in electrical resistivity linked to
groundwater flows, because they create variations in
water content and/or water conductivity. Time-lapse
ERT consists in performing an identical ERT survey
several times in the same place, before, during, and after
the hydrological process under study. In an unsaturated
zone, time-lapse ERT is primarily sensitive to water
content variations. Most of the time, a decrease of
resistivity indicates an infiltration, and an increase
indicates an evaporation. In a saturated zone, time-lapse
ERT is sensitive to changes in water conductivity. A
decrease of electrical resistivity measured by ERT
corresponds to an increase in ionic concentration of the
groundwater. An increase of electrical resistivity
corresponds to a dilution of groundwater. Controlled
experiments in tanks [10,20] demonstrated the potential
of time-lapse ERT. In the laboratory or in-situ, timelapse ERT works best with strong contrasts in resistivity
values if salt tracers are used or if pollution plumes are
monitored [4,19,22]. In natural conditions in the field,
resistivity contrasts are often weaker [17] (i.e. variations
from 10 to few tens of percent) and obtaining reliable
time-lapse ERT results could be a challenge when trying
to locate deep infiltration or recharge zones [6].
Although noticeable improvements have occurred in
time-lapse ERT, some recent studies also report image
reconstruction difficulties, due to the smoothing effect
of the algorithm [10,21]. Some time-lapse ERT surveys
fail to recover reliable actual resistivity changes
because the calculated resistivity model displays
resistivity artefacts (increase or decrease of calculated
resistivity) where no changes are expected or measured
[6]. Severe misinterpretations of time-lapse ERT
surveys can occur, leading to erroneous hydrological
understanding of pollution plumes, of groundwater
recharge or erroneous modelling. Previous authors [6]
have suggested that if a shallow surface infiltration or
evaporation occurs during an ERT survey, it could be
misinterpreted during ERT inversion. These authors
[11] have already demonstrated that a variation of actual
resistivity in shallow layers can lead to an opposite
variation of apparent resistivity at intermediate electrode spacing. This situation could be particularly acute
when the ground is composed of a resistive first layer
above a more conductive layer, and when shallow rain
infiltration (or evaporation) occurs between two
measurements in the field. In the example given by
Kunetz [11] with a 2-layer ground, a decrease of actual
resistivity within the uppermost part (first quarter,
thickness h/4) of the first layer of thickness h can
produce an increase of apparent resistivity at intermediate electrode spacing distances between 3 h and
20 h. Then, the easiest model obtained by inversion is
one that produces an unexpected increase of calculated
resistivity.
This article investigates how a variation of actual
resistivity with time and at shallow depth can influence
time-lapse ERT results and produce resistivity artefacts
at depth. In addition, it presents an advanced time-lapse
interpretation to reduce and remove those resistivity
artefacts. We used numerical modelling, standard and
advanced time-lapse inversions based on a classical
addition of a priori information. Then we used a field
data set exhibiting typical resistivity artefacts obtained
with a standard inversion to show how these resistivity
artefacts can be removed.
888
2. Material and methods
To investigate the effect produced by a shallow
infiltration on the ERT method, we adopted a classical
method with three stages. The first stage is the
construction of two scenarios of shallow infiltration
and their translation into experimental apparent
resistivity synthetic data sets. The second stage is to
use a standard inversion procedure for the time-lapse
inversion. The last stage is to introduce a priori
information to constrain the inversion of apparent
resistivity data. Here, this is referred to as ‘‘advanced
interpretation’’.
2.1. Synthetic models
Fig. 1 presents the synthetic models: a background
(initial) model and the two superficial infiltration
scenarios. One represents a 1-D resistivity model and
the second 2-D resistivity model. From the surface
down, the background model has three geological
layers:
variations of resistivity. The second one is the dipole–
dipole that is sensitive to the lateral variations of
resistivity. As proposed by De La Vega et al. [22] and
Loke [13], the data sets were combined to form a joint
data set for inversion. The apparent resistivity for the
three different unit electrode spacings of 4, 1 and 0.5 m
was calculated and 1.5% of Gaussian noise was added.
Fig. 1 presents also an example of an apparent resistivity
data set for three different unit electrode spacings and a
Wenner alpha array. We also plotted the ratio of the final
apparent resistivity after infiltration to the background
initial model. The ratio of apparent resistivity shows:
with 4 m spacing, an increase of apparent resistivity at
intermediate and shallow acquisition levels;
with 1 m spacing, the apparent resistivity decreases
for data close to the surface and increases at the
intermediate acquisition level;
with 0.5 m spacing, the apparent resistivity decreases
significantly at low level and increases at the
intermediate acquisition level.
2.2. Standard time-lapse inversion
the superficial layer has a thickness of 2.5 m and a
resistivity of 500 Ohm m in dry periods, similar to a
sandy loam layer;
the second layer has a thickness of 3 m and a
resistivity of 30 Ohm m, similar to clay;
the third layer has a resistivity of 500 Ohm m and
represents the substratum.
The first scenario represents 1-D vertical infiltration,
which can occur during a rain event (A, left). This
model is the same as the background model at initial
time but the resistivity of the first layer decreases in the
subsurface (0.40 m thick) from 500 to 50 Ohm m. This
shallow infiltration simulation is similar to the average
infiltration thickness measured in the field data set.
The second scenario represents vertical infiltration
but with a slight 2-D geometry that represents deeper
infiltration under gullies, 0.80 m and 5 m wide (A,
right). Topography was not introduced into the synthetic
models, in order to focus only on the shallow surface
phenomena effects and avoid topographical effects. The
resistivity of the first layer decreases in the subsurface
from 500 to 50 Ohm m.
Apparent resistivities were calculated with the
software package DC2DinvRes [8]. A finite difference
method was used to simulate the synthetic apparent
resistivities. Two arrays were chosen to calculate the
synthetic apparent resistivity. The first one is the
Wenner array because it is more sensitive to vertical
Inversion of the synthetics data set was performed
with the DC2DInvRes software package, with standard
parameters (inversion type Gauss-Newton, Z-weight
factor = 1, fixed regularisation, medium smooth constraint l=30). For a detailed description of these factors,
see [8]. This software allows the introduction of a priori
information into the time-lapse inversion procedure. For
the inversion, we defined a fine mesh introducing: (i)
two cells between every electrode; and (ii) a userdefined thickness for the cells. The thickness of the cells
is constant for all data sets. We used a standard timelapse inversion following the approach by Loke [12].
First, the initial background model without infiltration
was computed. Second, we used it as a reference model
in the time-lapse inversion of the two infiltrations
models. Finally, we compared the resulting calculated
models using the ratio of calculated resistivity (the final
calculated resistivity model divided by the initial
calculated resistivity model).
2.3. Advanced time-lapse inversion
The third stage consists in incorporating a priori
information into the time-lapse inversion. In this study,
we tested the possibility of decoupling shallow cells
from the rest of the model. This approach has already
been investigated for bedrock determination by
incorporating a seismic line at depth [9]. During
889
Fig. 1. Forward Modelling. (A). Synthetic model. (B). Apparent resistivity model obtained for a Wenner array and three electrode spacings. (C).
Ratio of apparent resistivity (final stage divided by initial stage). Note the increase of apparent resistivity at intermediate acquisition levels. The 4meter spacing data set contains fewer data points (84) than the 0.5-meter spacing data (6048) with a Wenner array.
Fig. 1. Modélisation directe. (A). Modèle synthétique. (B). Modèle de résistivité apparente obtenu pour un dispositif Wenner et trois écartements
d’électrodes différents. (C). Rapport des résistivités apparentes (état final/état initial). On note l’augmentation de la résistivité apparente aux niveaux
d’acquisition intermédiaires. Le jeu de données avec un écartement de 4 m contient moins de points (84) que celui avec un écartement de 0,5 m
(6048) avec le dispositif Wenner.
inversion, individual model cell boundaries can be
weighted by using a blocky model option. In the
presence of a known boundary, the weight can be set to
zero resulting in sharp gradients at this point. Knowledge may be derived from borehole information,
seismic or GPR surveys or observations on the surface
[8,9]. We considered that (i) the infiltration front
information is known, and (ii) this front is not the only
scope of the time-lapse ERT survey that focuses
preferably on deep infiltration or deeper changes in
resistivity. Hence, we introduced the knowledge of the
infiltration front position as a priori information.
890
3. Results
We present in Fig. 2 the results using the ratio
of resistivity after infiltration and before infiltration.
A ratio below 1.0 therefore indicates a decrease of
resistivity and above one an increase of resistivity.
3.1. Synthetic models
3.1.1. 1-D case
In the area between 0 and 0.4 m (thickness of the
simulated infiltration), the ratio of calculated resistivity
ranges between 0.6 and 0.8 for a standard inversion, for a
unit electrode spacing of 4 m. For unit electrode spacing
of 1 m, it ranges from 0.2 to 0.6, which is closer to the
expected value of 0.1. Finally, with the smallest unit
electrode spacing of 0.5 m, the ratio of the calculated
resistivity is between 0.1 and 0.3, close to the required
theoretical value. Using advanced inversion, the ratio of
calculated resistivity follows the same trend for all the
spacings. A slight improvement was noted for 0.5 m
spacing data: the ratio reaches the ideal value of 0.1.
In the area between 0.4 and 2.5 m, the actual
resistivity does not change; consequently, the calculated
ratio should be 1.0. With standard inversion, all unit
electrode spacings show an increase of the calculated
resistivity model, the ratios have values ranging
between 1.2 and 6 (Fig. 2, 1-D red arrow). When the
advanced inversion is used, a clear improvement is
obvious: the ratio is limited to the range between 1 and
1.2 only.
In the area between 2.5 and 5.3 m (clayey layer), the
actual resistivity does not change; consequently the
calculated ratio should also be 1. With standard
inversion, an increase of 1.1 to 2.5 between 2.5 and
3.5 m is still found. With spacing of 0.5 m and standard
inversion, the variation is limited to a value of ratio
ranging between 1.1 and 1.7. It remains between 1 and
1.3 with advanced inversion. Deeper, between 3.5 and
5.3 m, the ratio of calculated resistivity is close to the
expected value of one whatever standard or advanced
inversion is used. In conclusion, it seems that the depth
interval affected by resistivity artefacts is reduced with
smaller unit electrode spacing.
Fig. 2. Result of time-lapse inversion of synthetic data sets (combined Wenner and dipole–dipole arrays). Ratio of calculated resistivity using
standard and advanced inversion. Red arrows represent increases of resistivity, and blue arrows represent a decrease. The ratios of the calculated
resistivity after infiltration to the initial calculated resistivity before infiltration are attached to the arrows.
Fig. 2. Inversion en mode suivi temporel des jeux de données synthétiques (dispositifs Wenner et dipôle–dipôle combinés). Rapport des résistivités
calculées en utilisant les modes d’inversion standard et amélioré. Les flèches rouges représentent des augmentations de résistivité calculée, les
flèches bleues des diminutions. Le rapport de la résistivité calculée après l’infiltration sur la résistivité calculée initiale avant l’infiltration est indiqué
à côté des flèches.
Below 5.3 m in the sandy substratum, with a standard
inversion, the resistivity decreases (ratio between 0.7
and 1 for all units of electrode spacing). With advanced
inversion, the ratio remains between 0.9 and 1.1.
We drew two major conclusions. First, the resistivity
variations at shallow depth and the infiltration depth
are logically better resolved with shorter unit electrode
spacing (0.5 m in our example). Second, the use of
advanced time-lapse inversion with a decoupling line
limits the resistivity artefacts. For example, the false
increase of resistivity below the infiltration zone is limited
to 1.3, while with standard inversion, it is greater than 5.
3.1.2. 2-D case
From the surface down to 0.8 m at the centre of the
model, the results are similar to the results obtained
with the 1-D model. With standard inversion, the
decrease of electrode spacing improves the delineation of the bulb. The ratio of calculated resistivity
approaches the theoretical value of 0.1. Using
advanced inversion and 4 m spacing, the bulb is
poorly defined. The resistivity ratio lies between 0.5
and 0.24, quite far from the required value of 0.1. For
unit electrode spacing of 1 and 0.5 m, the advanced
inversion shows a homogeneous ratio with a value of
less than 0.2.
In the zone 0.4 to 2.5 m, all spacings show that the
ratio of the calculated resistivity model increases with
both standard and advanced inversion as in the 1-D
case. The calculated resistivity ratio reaches very high
values (up to 19) with the standard inversion. For
advanced inversion, the increase remains much
smaller (around 4) with 1 m spacing.
Between 2.5 and 5.3 m, the calculated models are
similar to what we obtained for the 1-D case.
For the substratum zone, the calculated variations are
more noticeable. With both standard and advanced
inversions and 4 m spacing, the ratio of the calculated
resistivity model remains between 0.9 and 1.1, an
acceptable result. With shorter spacing, the ratio of
calculated resistivity varies between 0.5 and 0.8 for
standard inversion, and between 0.8 and 1 for
advanced inversion. However, even if the advanced
inversion seems to give better results, the patterns of
the resistivity ratio distribution appears complicated
by the 2-D geometry of the infiltration. Some
resistivity artefacts (increases) are visible in the
lower left and right corners. They are considered to be
boundary effects and are not analysed in this article.
The numerical modelling shows that at shallow
depth, the ratio of calculated resistivity and the
891
geometry of the infiltration are better resolved using
the smallest unit electrode spacing. The false increase in
the apparent resistivity during infiltration is reduced
when the advanced inversion introducing a decoupling
line is used. In the advanced approach, the calculated
ratio is limited to 1.5 in 1-D (50%) and to 1.7 (70%) in
2-D, while with the standard approach, ratios of 2.5
(250%) or even 8 (800%) with 1 and 2-D cases are
obtained, respectively.
At depth, the numerical modelling shows that the
reduction of unit electrode spacing could generate
several symmetrical zones on the cross-section with a
decrease or increase of calculated resistivity. The
contrast is greater in the 2-D case. Because we focused
our work primarily on the removal of the most severe
resistivity artefacts (increase of calculated resistivity)
below the infiltration zone, the origin of smooth
oscillations at depth is not investigated in this article.
Effects of the regularization parameter, array used, or
even data density might explain this phenomenon.
Finally, we showed that using the advanced timelapse inversion, the calculated resistivity ratio is
significantly closer to the resistivity model ratio, and
is generally limited to 0.2 (i.e. 20% of resistivity
variations).
3.2. Field data example
The field data set is a typical case showing resistivity
artefact production after time-lapse inversion. This
survey was not dedicated to shallow infiltration
monitoring but rather to study recharge processes
under an ephemeral gully in Burkina Faso, West Africa
[5]. In regions with a low rainfall index and a monsoon
climate, there is an increasing need for sustainable
groundwater resources. This requires a better understanding of groundwater recharge zones. Recharge
processes in semi-arid climates (rain < 600 mm) are
mainly located below seasonal ponds [14,15], alluvial
sandy fans [16] and intermittent (ephemeral) streams
during monsoon events [7]. Quantification of infiltration
rates and groundwater recharge relies generally on field
measurements in boreholes by means of neutron probes,
tensiometers, capacitive probes and piezometer networks. These point measurements need an optimized
implementation with geophysical surveys.
The study area, in northern Burkina Faso, is a typical
(1 ha) gully erosion area located at the outlet of an 82 ha
catchment with a crystalline basement (Fig. 3). The
surface conditions in the area are favourable to
infiltration due to: (i) a fractured quartz vein; and (ii)
sandy or pebble surfaces. Taking advantage of a long
892
resistivity for short spacing (< 1 m) varied by less than
5% in the morning thus keeping temperature effects at
an acceptable level. The infiltration pattern was also
monitored with neutron probe measurements in six
auger holes shown in Fig. 3.
The results obtained with both standard and
advanced inversions are presented in Fig. 4. We
positioned the decoupling line at a constant depth of
0.25 m corresponding to the average value given by the
infiltration front derived from neutron probe measurements.
Fig. 3. Location of the experimental site, geophysical survey and
neutron probe measurements.
Fig. 3. Localisation du site expérimental, des mesures géophysiques
et des tubes d’accès de sonde à neutron.
dry season followed by a short rainy one, we used the
time-lapse ERT approach to carry out electrical
resistivity monitoring during the rainy season, between
June and September. We used two apparent resistivity
data sets obtained just before (June) and just after the
rainy season (September) to obtain a significant
infiltration phenomenon. The stainless steel electrodes
were left in the soil for the duration of the experiment.
The cables were laid out each time.
To monitor expected infiltration down to depths of
5 m or more, we laid out a Wenner array profile along a
line crossing the gully. A first acquisition was made with
1 m spacing along the entire length of the profile. The
data set with 2 m was extracted from the 1 m data set for
demonstration purposes in this paper. Then, three panels
of apparent resistivity with the 0.5 m spacing data set
were acquired by a classical roll-along technique, with
three successive acquisitions involving 64 electrodes
each. The data with 1 m spacing were added at depth to
the 0.5 m panels. This avoids inversion distortions due
to the lack of data at depth.
Measurements were made before noon to avoid high
temperature variations. In addition, apparent resistivity
variations were also monitored with time on a test site
during the day to evaluate the effect of temperature
on resistivity variations. We found that the apparent
At shallow depth between 0 and 0.4 m, with a large
unit electrode spacing of 2 m, the ratios of calculated
resistivity are 1.3 and 4 using standard and advanced
options respectively, indicating that the infiltration is
not visible. For smaller spacing (1 m) the infiltration is
still not detected with the standard inversion. With
advanced inversion, the infiltration is clearly seen
with a ratio below 0.5 and 1.With the smallest spacing
of 0.5 m, the ratio of resistivity is lower than 0.5
whatever type of inversion is used;
between 0.5 and 3 m, for all spacings, the standard
inversion shows a calculated resistivity ratio, which
ranges between 1.2 and 5. When advanced inversion
is used, the increase is limited to a ratio ranging
between 1 and 1.5;
below 3 m, for all inversion and with a unit-electrode
spacing of 2 m, the ratio of resistivity remains
between 0.9 and 1.2. With unit electrode spacing of 1
m, the ratio of calculated resistivity is in the range of
1–1.3 for standard and advanced inversion. For a
spacing of 0.5 m, results show noticeable variations
between 1 and 1.3 marked by a red arrow in Fig. 4. At
the right of the cross-section below the position of
44 m, the ratio of calculated resistivity ranges
between 0.5 and 0.8 as shown by a blue arrow.
4. Discussion
4.1. Comparison with neutron probe data
Fig. 5 presents the comparison between standard and
advanced time-lapse inversion of ERT with the smaller
electrode spacing (0.5 m) versus neutron probe data.
The infiltration front is drawn according to the
measurements of the six neutron tubes (TN 22, 23,
24, 25, 26 and 27). Only TN 23, 24, 25 and 26 are shown
for clarity. All tubes show infiltration down to less than
0.4 m except TN24 where the infiltration deepens to
0.80 m. In addition, below TN24, a very localized water
invasion was recorded at a depth of 4 m during the rainy
893
Fig. 4. Standard inversion and advanced inversion results for field data with three different unit electrode spacings (2, 1 and 0.5 m).
Fig. 4. Résultats des inversions en mode standard et amélioré pour le jeu de données de terrain, avec trois écartements unitaires d’électrodes
(2, 1 et 0,5 m).
season. We attribute this phenomenon to a local lateral
invasion due to the proximity of the fractured quartz
vein. Five major conclusions were drawn from the
comparison of neutron probe data and ERT:
by neutron probe data. They could also be the result of
geometrical oscillations in the inversion, as already
noted at depth with our numerical modelling of a 2-D
infiltration object.
first, for the standard time-lapse inversion, we note
that if we draw the contour line of ratio 0.8 near the
surface, the shape of this line is in agreement with the
neutron probe variation;
second, the increase of calculated resistivity just
below the infiltration was not corroborated by neutron
probe measurements as expected from our numerical
modelling. We confirm here the resistivity artefact
creation using standard inversion. In the deeper part
of the section, the variations of the ratio are high
(range 1 to 1.7);
third, for the advanced inversion using a constant
thickness of decoupling (0.25 m), the decrease of
calculated resistivity is strictly limited inside the
decoupling zone;
fourth, the increase of calculated resistivity below the
infiltration is clearly reduced, not only with the
reduction of the area involved, but the ratio also
remains limited to less than 2. In addition, in the deeper
part of the section, the variations of the ratio are not
only lower (range 0.9 to 1.1 with some local values
reaching 1.3), but affect a smaller area of the section;
fifth, the water invasion noted for tube TN24 at 4 m
depth is noticed by both inversions. It is, however,
comparable to other variations calculated laterally at
the same depth. These variations are not corroborated
Finally, we noted that using standard inversion,
severe resistivity artefacts of increasing resistivity were
produced below the infiltration front, as predicted by the
numerical modelling. The only benefit obtained from
the standard inversion is that the irregular shape of the
infiltration front fits the neutron probe data. With the
advanced inversion, we noted a clear improvement in
resistivity artefact removal. We used a constant
thickness of decoupling line. Zones with an increase
in calculated resistivity at depth are still present, but
within a smaller variation range. This is not entirely
satisfactory. We investigated further in the decoupling.
4.2. Influence of the geometry of the decoupling
line
Considering that the infiltration geometry could not
be well known in the field due to a lack of boreholes or
other methods, we investigated the effect of three
different geometries of the decoupling line. The results
are presented in Fig. 6. Three cases are discussed: (i) no
knowledge of the depth of the infiltration front
(decoupling line at a constant depth all along the
ERT profile); (ii) a precise but punctual knowledge of
the depth of the infiltration front; (iii) a complete
knowledge of the depth of the infiltration front.
894
Fig. 5. Comparison of neutron probe data with standard (top) or advanced (bottom) time-lapse inversions. For standard inversion, the contour line of
0.8 is marked by a continuous grey line to show the good accordance with neutron probe data (the infiltration front is shown by short red lines). For
advanced inversion, the position of the decoupling line is marked by a black dotted line. The blue arrow shows the localized water invasion at 4 m
depth below neutron probe tube TN 24.
Fig. 5. Comparaison des résultats obtenus avec la sonde à neutron et les inversions en mode de suivi temporel pour le mode standard (en haut) et
amélioré (en bas). Pour le mode standard, la ligne d’isocontour de rapport 0,8 est marquée avec une ligne grise continue, pour montrer la bonne
correspondance avec les données de sonde à neutron (le front d’infiltration est montré avec de courts traits horizontaux rouges). Pour le mode
d’inversion amélioré, la position de la ligne de découplage est marquée par une ligne noire pointillée. La flèche bleue montre une invasion d’eau très
localisée à 4 m en dessous du tube neutronique TN24.
The first case corresponds to the one where the
interpreter gives only an estimate of the thickness of the
infiltration front as we did when interpreting our field
data. As shown in Fig. 6a, and b, for two different
decoupling depths, the time-lapse ERT gave different
results: for a decoupling depth of 0.1 m, the increase of
calculated resistivity remains acceptable and lower than
1.25 just below the infiltration. This result is comparable, or slightly better, than what we obtained with a
decoupling depth of 0.25 m (as shown also in Fig. 4).
Using a much higher infiltration depth as decoupling
line, for example 0.75 m (Fig. 6b), ERT time-lapse
inversion no longer fits the neutron probe data. ERT
exhibits a significant increase of resistivity (ratio of
more than 3) at the north of the section for example, not
corroborated by neutron probe data.
The second case corresponds to a precise but
punctual knowledge of the depth of the infiltration
front. We introduced six decoupling lines at six constant
depths indicated by the six neutron probe data. Each line
895
Fig. 6. Effect of the geometry of the decoupling line. (a) and (b). Decoupling line with a constant depth of 0.1 and 0.75 m, respectively. (c).
Decoupling line using information obtained with neutron probe data. (d). Decoupling line with irregular shape deduced from contour line of ratio 0.8
obtained with standard time-lapse inversion with smallest unit-electrode spacing of 0.5 m (see Fig. 4).
Fig. 6. Effet de la géométrie de la ligne de découplage. (a) et (b). Lignes de découplage placées à 0,1 et 0,75 m de profondeur respectivement. (c).
Ligne de découplage placée selon l’information obtenue avec les tubes neutroniques. (d). Ligne de découplage avec une forme irrégulière déduite de
l’isocontour de rapport 0,8, obtenu avec le mode d’inversion standard et le plus petit écartement d’électrodes de 0.5 m (voir Fig. 4).
896
is centred with respect to the tube, its length is
arbitrarily limited laterally to the mid point between two
tubes. The results shown in Fig. 6c exhibit promising
improvements in resistivity artefact removal, especially
in the northern part. However, at the centre of the gully,
an increase of calculated resistivity is magnified.
The third case considers a complete knowledge of the
infiltration front as a continuous line. This information
could be extracted from other data in the field (dense
TDR measurements or ground penetrating radar profiling). For our study, we took advantage of the good
agreement noted between the shape of the ERT contour
line produced with the standard time-lapse inversion and
the neutron probe. We thus generated a decoupling line
that respects exactly the shape of the calculated contour
line. By comparison with neutron probe data, we choose
the contour line of 0.8. The results are shown in Fig. 6d. A
general improvement is noted. The increase of the
calculated resistivity is significantly reduced or even
removed just below the infiltration front. The oscillations
of the resistivity ratio at depth are still present but their
amplitude stays within a limited range (between 0.85 and
1.25). The decrease of the calculated resistivity at 4 m
depth below the neutron tube TN 24 appears magnified
(slight decrease of the ratio).
We demonstrate here that the position and the
geometry of the decoupling line are of great importance.
Acceptable results are obtained with our field data using a
small thickness of decoupling (0.1 m). In addition, and
for other surveys, the approach considering a continuous
knowledge of the depth of the infiltration front is by far
the best, even if some resistivity artefacts are still present
but limited to a range between 0.85 and 1.25. Then one
can use the shape of the infiltration given by standard
time-lapse inversion as the decoupling geometry, but it is
in any case essential to have external data at some points
along the profile. Moreover, small unit electrode spacings
are required during data acquisition. For further studies,
additional improvements could be made in time-lapse
inversion by using other a priori information such as
invariant zones (for example the knowledge of the
groundwater conductivity with time). This approach has
already been tried by Vesnaver et al. [23] for seismic
inversion and by Nguyen and Kemna [18] for ERT
inversion, but it was not tested in this study, because the
field data did not allow us to fix an invariant zone at depth.
4.3. Discrimination between resistivity artefact and
true hydrological processes
We examine here the capacity of the advanced
interpretation to discriminate between a resistivity
artefact and a true hydrological process. We chose a
common but important case for soil and agronomical
sciences: the characterisation of the zone where the
plants are taking up water within the root zone and
where resistivity is likely to increase. Therefore, as we
have seen from the modelling and field data, the
resistivity artefact of increasing resistivity at intermediate depth might be wrongly interpreted as a drying
zone (root-zone). Finally, the question arises: if a true
drying zone exists below the infiltration in the same
place as resistivity artefacts, what is the efficiency of the
advanced inversion? Does it display correctly the true
phenomenon of an increasing resistivity? A scenario
that includes a shallow infiltration and a drying zone
below was simulated using the 2D model presented in
Fig. 1. Fig. 7 presents the model that includes the drying
zone and the results obtained with standard and
advanced inversions. The standard inversion displays
Fig. 7. Comparison between standard and advanced time-lapse inversion using a scenario with a drying zone below the infiltration as
shown on the model. The true increase of resistivity is 2 and it is
satisfactorily reconstructed with advanced inversion using a decoupling line (below).
Fig. 7. Comparaison entre les modes d’inversion en suivi temporel
standard et amélioré en utilisant un scénario de dessèchement, dans
une zone située juste au-dessous du front d’infiltration, comme le
montre le modèle synthétique (en haut de l’image). La véritable
augmentation de résistivité d’un facteur 2 est reconstruite de façon
satisfaisante, avec le mode d’inversion amélioré qui utilise une ligne
de découplage (en bas de l’image).
a strong increase of resistivity, ratio more than 8,
between 2 and 5 m, and a strong decrease below (ratio
less than 0.4). When one looks at the advanced
inversion, the increase of resistivity below the infiltration is also seen and its value (ratio near 2.5) agrees well
with the expected value of 2. Below, the variation of
resistivity remains within the range 0.7 to 1. As a
conclusion, if a true increase of resistivity is present in
the soil at intermediate depth, it can be identified and
correctly quantified by the advanced time-lapse inversion. Using standard inversion, unreliable values are
obtained as resistivity artefact and the true phenomenon
add their effects.
5. Conclusion
Time-lapse ERT inversion can produce resistivity
artefacts in certain circumstances already pointed out
in previous studies. For example, when the actual
resistivity decreases at shallow depth, a typical
resistivity artefact is an increase of calculated resistivity
at intermediate depths, whereas the actual resistivity
does not change. Therefore, results of time-lapse ERT
could lead to false interpretations and ERT may not be
reliable for studying changes in resistivity at depth. We
investigated the effect of a shallow variation of
resistivity within the first decimetres of the soil on
time-lapse ERT inversion using numerical modelling to
show a typical ERT resistivity artefact. We show that 2D infiltration geometry enhances the resistivity artefact
production by creating additional oscillations of
calculated resistivity variation at depth. We used an
advanced time-lapse inversion introducing a shallow
decoupling line as a priori information corresponding to
a constant thickness of the infiltration front, supposed to
be known from external data. Using this advanced
inversion, the resistivity artefact production is significantly reduced. The wrong increase of calculated
resistivity is limited to a ratio of less than 1.3 whereas it
grows to 3 or even more when standard time-lapse
inversion is used.
The advanced time-lapse inversion was tested on
field data and the results corroborate the conclusions
derived from the numerical modelling:
data sets using short unit-electrode spacing are
required to provide a convenient base for time-lapse
ERT in case shallow infiltration (or evaporation) is
present;
using a standard (non-decoupling) approach, the
resistivity artefact creation (i.e. increase of calculated
resistivity at intermediate depth) is confirmed;
897
using standard inversion, the infiltration front can be
delineated if short electrode spacing is used. In this
case, a comparison with neutron probe data is
necessary to identify the correct calculated resistivity
isocontour and thus delineate the position of the
infiltration front in the ERT image. Then, the
infiltration front positioned with ERT can be used
for advanced inversion;
when advanced inversion that incorporates a decoupling line of constant thickness at shallow depth is
used, the resistivity artefacts noted at intermediate
depth are significantly reduced. We increased the
resistivity artefact reduction by using a continuous
line of variable thickness. The position of this line was
deduced from the comparison between neutron probe
data and standard inversion data. This allowed us to
remove almost completely the resistivity artefact of
increasing resistivity at intermediate depths. However, some oscillations at depth within a range of ratio
0.8 to 1.2 (i.e. 20%) are still present and could be
smoothed by tuning other inversion parameters such
as regularisation factors.
Finally, when performing time-lapse ERT surveys in
the presence of shallow infiltration or evaporation, we
advocate measuring dense apparent resistivity data at
shallow depth using small unit-electrode spacing (or
shallow electromagnetic profiling). Even with short
electrode spacing, a standard time-lapse inversion may
exhibit false resistivity variations below the infiltration
or evaporation front. To remove those unwanted
resistivity artefacts, we need to incorporate a shallow
continuous decoupling line into the inversion. In case of
infiltration, this decoupling line is the infiltration front.
The position and the shape of this line need to be defined
and controlled with external information such as
neutron probe data (or any other method available) as
well as deduced from the ERT survey itself. With this
approach, more reliable time-lapse ERT results are
obtained, not only for shallow depths, but also on deeper
changes in resistivity in the pseudo-section, leading to a
better characterization of hydrological processes.
Acknowledgments
We wish to thank French EC2CO project ONDINE
for funding part of this research. The INERA Institute in
Burkina Faso provided access to the experimental site.
Yann Le Troquer and Burkinabese staff are warmly
thanked for field data acquisition. We are very grateful
to Dr Thomas Ingeman-Nielsen for his helpful
comments on the first version of the manuscript.
898
References
[1] R. Barker, The application of time-lapse electrical tomography
in groundwater studies, The Leading Edge 17 (10) (1998) 1454–
1458.
[2] Y. Benderitter, Short time variation of the resistivity in an
unsaturated soil: The relationship with rainfall, European Journal of Environmental and Engineering Geophysics 4 (1999)
37–49.
[3] A. Binley, Cross borehole radar and resistivity tomography: a
comparison of techniques in unsaturated sandstone, in : Symp.
Applications of Geophysics to Engineering and Environmental
Problems (SAGEEP), Denver, 2001.
[4] G. Cassiani, V. Bruno, A. Villa, N. Fusi, A.M. Binley, A saline
tracer test monitored via time-lapse surface electrical resistivity
tomography, Journal of Applied Geophysics 59 (2006) 244–
259.
[5] M. Descloitres, O. Ribolzi, Y. Le Troquer, Study of infiltration in
a Sahelian gully erosion area using time-lapse resistivity mapping, Catena 53 (2003) 229–253.
[6] M. Descloitres, L. Ruiz, M. Sekhar, A. Legchenko, J.J. Braun,
M.S.M. Kumar, S. Subramanian, Characterization of seasonal
local recharge using electrical resistivity tomography and
magnetic resonance sounding, Hydrological Processes 22
(2008) 384–394.
[7] M. Esteves, J.M. Lapetite, A multi-scale approach of runoff
generation in a Sahelian gully catchment: a case study in Niger,
Catena 50 (2003) 255–271.
[8] T. Günther, Inversion Methods and Resolution Analysis for the
2D/3D Reconstruction of Resistivity Structures from DC
measurements, University of Mining and Technology, Freiberg,
Germany, 2004.
[9] T. Günther, C. Rücker, A General Approach for Introducing
Information into Inversion and Examples from dc Resistivity
Inversion, EAGE Near Surface Geophysics, Helsinki, 2006.
[10] A. Kemna, J. Vanderborght, B. Kulessa, H. Vereecken, Imaging
and characterisation of subsurface solute transport using electrical resistivity tomography (ERT) and equivalent transport
models, Journal of Hydrology 267 (2002) 125–146.
[11] G. Kunetz, Principles of direct current resistivity prospecting,
in : B. Trondheim (Ed.), Principles of Direct Current Resistivity
Prospecting, Geopublication Associates, Berlin, Germany,
1966 , pp. 70–71.
[12] M.H. Loke, Time-lapse resistivity imaging inversion, in : 5th
Meeting of the Environnemental and Engineering Society European Section, Budapest, Hungary, 1999.
[13] M.H. Loke, Tutorial: 2-D and 3-D electrical imaging surveys,
2004.
[14] W. Martin-Rosales, C. Leduc, Dynamics of emptying of a
temporary pond in the Sahel: the case study of Banizoumbou
(southwestern Niger), C. R. Geoscience 335 (2003) 461–468.
[15] G. de Marsily, Importance of the maintenance of temporary
ponds in arid climates for the recharge of groundwater, C. R.
Geoscience 335 (2003) 933–934.
[16] S. Massuel, G. Favreau, M. Descloitres, Y. Le Troquer, Y.
Albouy, B. Cappelaere, Deep infiltration through a sandy alluvial
fan in semi-arid Niger inferred from electrical conductivity
survey, vadose zone chemistry and hydrological modelling,
Catena 67 (2006) 105–118.
[17] D. Michot, Y. Benderitter, A. Dorigny, B. Nicoullaud, D. King,
A. Tabbagh, Spatial and temporal monitoring of soil water
content with an irrigated corn crop cover using surface electrical
resistivity tomography, Water Resources Research 39 (5) (2003)
1138, doi:10.1029/2002WR001581.
[18] F. Nguyen, A. Kemna, Strategies for time-lapse electrical resistivity inversion, in : 11th European Mtg. of Environmental and
Engineering Geophysics, European Association Geoscientists
and Engineers, A005, 2005.
[19] R.E. Nimmer, J.L. Osiensky, A.M. Binley, K.F. Sprenke, B.C.
Williams, Electrical resistivity imaging of conductive plume
dilution in fractured rock, Hydrogeology Journal 15 (2007)
877–890.
[20] L. Slater, A. Binley, R. Versteeg, G. Cassiani, R. Birken, S.
Sandberg, A 3-D ERT study of solute transport in a large
experimental tank, Journal of Applied Geophysics 49 (2002)
211–229.
[21] J. Vanderborght, A. Kemna, H. Hardelauf, H. Vereecken, Potential of electrical resistivity tomography to infer aquifer transport
characteristics from tracer studies: a synthetic case study, Water
Resources Research 41 (2005) W06013.
[22] M. de la Vega, A. Osella, E. Lascano, Joint inversion of Wenner
and dipole-dipole data to study a gasoline-contaminated soil,
Journal of Applied Geophysics 54 (2003) 97–109.
[23] A.L. Vesnaver, F. Accaino, G. Bohm, G. Madrussani, J. Pajchel,
G. Rossi, G. Dal Moro, Time-lapse tomography, Geophysics 68
(2003) 815–823.
ARTICLE IN PRESS
Waste Management xxx (2009) xxx–xxx
Contents lists available at ScienceDirect
Waste Management
journal homepage: www.elsevier.com/locate/wasman
Improvement of electrical resistivity tomography for leachate injection monitoring
R. Clément a,*, M. Descloitres a, T. Günther b, L. Oxarango a, C. Morra c, J.-P. Laurent a, J.-P. Gourc a
a
Laboratoire d’Etude des Transferts en Hydrologie et Environnement, LTHE, UMR 5564, CNRS, INPG, IRD, UJF, B.P. 53, 38041, Grenoble Cedex 9, France
Leibniz Institute for Applied Geophysics, Stilleweg 2, D-30655 Hannover, Germany
c
PROKHEM/Floralis, 6 allée de Bethléem, 38610 Gières, France
b
a r t i c l e
i n f o
Article history:
Accepted 2 October 2009
Available online xxxx
a b s t r a c t
Leachate recirculation is a key process in the scope of operating municipal waste landfills as bioreactors,
which aims to increase the moisture content to optimize the biodegradation in landfills. Given that liquid
flows exhibit a complex behaviour in very heterogeneous porous media, in situ monitoring methods are
required. Surface time-lapse electrical resistivity tomography (ERT) is usually proposed. Using numerical
modelling with typical 2D and 3D injection plume patterns and 2D and 3D inversion codes, we show that
wrong changes of resistivity can be calculated at depth if standard parameters are used for time-lapse
ERT inversion. Major artefacts typically exhibit significant increases of resistivity (more than +30%) which
can be misinterpreted as gas migration within the waste. In order to eliminate these artefacts, we tested
an advanced time-lapse ERT procedure that includes (i) two advanced inversion tools and (ii) two alternative array geometries. The first advanced tool uses invariant regions in the model. The second advanced
tool uses an inversion with a ‘‘minimum length” constraint. The alternative arrays focus on (i) a pole–
dipole array (2D case), and (ii) a star array (3D case). The results show that these two advanced inversion
tools and the two alternative arrays remove almost completely the artefacts within +/5% both for 2D
and 3D situations. As a field application, time-lapse ERT is applied using the star array during a 3D leachate injection in a non-hazardous municipal waste landfill. To evaluate the robustness of the two advanced
tools, a synthetic model including both true decrease and increase of resistivity is built. The advanced
time-lapse ERT procedure eliminates unwanted artefacts, while keeping a satisfactory image of true resistivity variations. This study demonstrates that significant and robust improvements can be obtained for
time-lapse ERT monitoring of leachate recirculation in waste landfills.
Ó 2009 Elsevier Ltd. All rights reserved.
1. Introduction
The concept of bioreactor landfill has been studied and tested
since 1970 in the United States of America (USA) and for more than
a decade in Europe. This technology aims at enhancing the waste
biodegradation in landfills. Many studies have pointed out the potential benefits of the bioreactor approach, namely:
A quicker stabilisation of organic content can be achieved (10–
15 years compared to 30–100 years with a classical land filling
operation) (Pacey et al., 1999).
The biogas production can be improved (Hossain et al., 2003)
providing a significant improvement of the efficiency of biogas
power plant.
The environmental hazard is reduced because bioreactor
requires a better monitoring (Reinhart et al., 2002).
* Corresponding author. Tel.: +33 (0)4 76 63 58 67; fax: +33 (0)4 76 82 50 14.
E-mail addresses: remi.clement@hmg.inpg.fr (R. Clément), Thomas.Guenther@
liag-hannover.de (T. Günther), christophemorra@yahoo.fr (C. Morra).
If a leachate recirculation system is used, the volume of leachate
to be treated is reduced as a part of the liquid retained by the
waste matrix (Pohland, 1980; Warith, 2002).
In situ operation of a landfill as a bioreactor requires a careful
monitoring and control of the operating parameters. The moisture
content has a major influence on the efficiency of the methanogen
bacteria (Reinhart and Townsend, 1998). The anaerobic methanogenesis is enhanced by a high moisture content that can only be
reached by adding water to the waste. Indeed, under temperate climate, the waste disposed in landfill is generally too dry to ensure
an optimal biodegradation. The leachate recirculation appears to
be a very favourable process since it could increase the moisture
content. Moreover, the leachate recirculation tends to uniform
the spatial distribution of adapted micro flora. As far as an efficient
monitoring of the bioreactor is concerned, measuring the water in
landfills is a key issue (Imhoff et al., 2007). In particular, the optimisation of leachate injection systems remains a challenging and
ongoing problem for bioreactor landfill operators. Addressing this
issue requires monitoring of these systems during long term field
0956-053X/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.wasman.2009.10.002
Please cite this article in press as: Clément, R., et al. Improvement of electrical resistivity tomography for leachate injection monitoring. Waste Management (2009), doi:10.1016/j.wasman.2009.10.002
ARTICLE IN PRESS
2
R. Clément et al. / Waste Management xxx (2009) xxx–xxx
situations. Geophysical methods applied to landfill may be of
assistance.
Over the past ten years, various geophysical studies have shown
that it is possible to use resistivity methods (mainly Direct Current
-DC- and Electromagnetic -EM- methods) to:
Characterise the waste landfill structure (Bernstone et al., 2000;
Cossu et al., 2005; Meju, 2000);
study the contamination of groundwater by leachate leaking
from a landfill (Mondelli et al., 2007; Olofsson et al., 2006; Radulescu et al., 2007; Santos et al., 2006; Soupios et al., 2007; Zume
et al., 2006);
map the plume geometry and monitor the movement of the
plume segments (Acworth and Jorstad, 2006; Guérin et al.,
2004);
Evaluate the spatial and temporal water variation in waste
(Acworth and Jorstad, 2006; Frohlich et al., 1994; Guérin et al.,
2004; Jolly et al., 2007; Mondelli et al., 2007; Moreau et al.,
2003).
Measure Induced Polarization (IP) effect during gas migration in
landfill (Cossu et al., 1990).
Most of these studies have shown that surface electrical resistivity tomography (ERT) can be a suitable method to study resistivity distribution (2D and 3D) at a large scale (ten to hundreds of
meters wide and down to 30 m deep). ERT is becoming a common
tool to study recirculation experiments in landfills. During the
recirculation process, if a leachate content variation or gas migration creates resistivity variations, ERT can be considered using a
time-lapse approach (i.e. repeating an ERT survey several times
during the injection). Time-lapse ERT has been widely considered
in areas other than landfill such as studying environmental processes as it focuses on electrical resistivity changes in the subsurface produced by groundwater flows. The main potential
applications are pollution plume monitoring (Benson, 1995; Benson et al., 1997; Day-Lewis et al., 2003; deLima et al., 1995), and
the location of shallow or deep infiltration or recharge zones (Deiana et al., 2007; Descloitres et al., 2003, 2008a,b; Frohlich et al.,
1994). Delineation of leachate plume in landfills can be studied
with time-lapse ERT (Grellier et al., 2008; Guerin et al., 2004; Guérin et al., 2004; Rosqvist et al., 2003, 2005).
Several recent studies have however shown that some timelapse surveys are not easy to interpret. They show unexpected
variations of calculated resistivity (Descloitres et al., 2008b; Guérin
et al., 2004; Jolly et al., 2007); however several explanations could
be provided to explain those ambiguous results. First, the results
are mainly attributed to the regularisation process that is necessary due to the non-uniqueness of the solution, i.e. for the same
data set of apparent resistivity there are different solutions of
inversion. Second, they could be the result of regularisation in
inversion which produces a smooth reconstructed image. Some
authors explain there are abnormal variations of calculated resistivity in areas near the injection (Guerin et al., 2004). In most cases,
these changes lead to unexpected increases in resistivity with time.
On the one hand, some authors suggest these changes could be
linked to a desaturation of medium due to gas migration in some
areas (Grellier et al., 2008). Indeed, a leachate injection could push
the gas away from the injection point (Guerin et al., 2004; Moreau
et al., 2003; Rosqvist et al., 2003, 2005). Consequently, the decrease
of water content in the waste results in an increase in electrical
resistivity.
On the other hand, some authors stress that these variations are
questionable as they may appear in reverse resistivity anomalies
that can lead to ambiguous interpretations (Guerin et al., 2004;
Jolly et al., 2007). When using methods in other areas than leachate
injections, similar problems have been encountered. Some timelapse ERT surveys have failed to detect reliable actual resistivity
changes due to the calculated resistivity model displaying artefacts
(increases or decreases of calculated resistivity) where no changes
are expected or detected (al Hagrey, 2007; Descloitres et al., 2003,
2008b; Nimmer et al., 2007). The reconstruction algorithm can
produce a significant increase in resistivity (Singha and Gorelick,
2005).
The aim of this paper is to show that false variations of calculated resistivity (artefacts) can be obtained with time-lapse ERT
inversion in some situations if standard parameters are used for
ERT inversion. We propose a classical numerical modelling approach to test typical scenarios of infiltration in landfill waste, with
2D and 3D geometries. In order to achieve this, we build numerical
models to generate synthetic ERT data set with symmetrical and
asymmetrical electrode arrays (2D case) and parallel and star array
(3D case). This study demonstrates that it is possible to obtain artefacts of increasing resistivity with standard time-lapse ERT inversion. Then, advanced procedure is tested using two inversion
tools to determine whether it is possible to limit or eliminate these
artefacts. Based on the conclusion derived from the numerical results, the star array was applied on a real data set obtained in
the field during leachate recirculation experiment, comparing the
time-lapse ERT image to independent data. Finally, we evaluate
the reliability of the advanced procedure using a synthetic modelling simulating both an injection and a biogas migration respectively, corresponding to both a real decrease and a real increase
of resistivity around the infiltration.
2. Materials and methods
2.1. Methodology
To show artefact creation and their remediation, this study uses
a classical methodology applied in several papers (Clément et al.,
2009; Radulescu et al., 2007; Yang, 2005) using ERT numerical
modelling. The methodology applied in this paper is based on three
steps. The first step is the creation of resistivity models corresponding to two realistic scenarios of leachate recirculation. The
second step produces synthetic apparent resistivity data sets using
a forward calculation with several electrode arrays. The third step
is the inversion of the synthetic data set using (a) common inversion parameters used for time-lapse ERT, referred to as ‘‘standard
inversion” in this paper and (b) advanced inversion tools which
produce significant improvement for artefact removal. Both standard and advanced calculated models are then compared to the
initial synthetic models to evaluate the efficiency of arrays and advanced inversion tools on artefact removal.
2.2. Synthetic 2D and 3D models
Realistic scenarios of leachate recirculation have been considered. Leachate injection systems commonly used on waste landfills
are either systems which create an elongated injection close to the
surface, frequently assimilated with horizontal trenches filled with
highly porous materials (Haydar and Khire, 2005; Khire and Haydar, 2003), or a punctual injection in wells or pits as discussed in
some articles (Khire and Mukherjee, 2007; Morris et al., 2003).
They represent typical 2D or 3D infiltration geometries, respectively. The hypothesis is that the model blocks are isotropic and
homogeneous. The injected leachate is considered to be more conductive than the waste. It is further hypothesised that there is no
rubber or plastic liner covering the site which could limit the electrical current flow. The structure of the initial model before infiltration is composed of two layers of soil (Fig. 1a, center). The surface
ARTICLE IN PRESS
3
Fig. 1. (a) 2D and 3D infiltration synthetic models and (b) variation of real resistivity between initial and final state of the synthetic models (longitudinal cross section). The
variations are show using Eq. (1) (see text).
layer is a loamy-sandy soil; 1.5 m thick with a resistivity of
100 ohm m. This resistivity corresponds to a resistivity observed
in July 2008 on our experimental site (see field results). The second
layer, the waste, is 1.5 m thick. Its resistivity is 15 ohm m. The
model space is 71 m long, 40 m wide and has a thickness of
15 m. The mesh size of the synthetic model is 1 m2 or 1 m3 (2D
and 3D models, respectively). Injection geometries are as follows:
2D: A trench 2 m wide and 1.5 m deep which creates a cylindrical infiltration 2.5 m high with a radius of 8 m (Fig. 1a, left).
3D: A square pit 2 m wide and 1 m deep which creates a 3D oval
infiltration with a radius of 4 m and a height of 2.5 m (Fig. 1a
right).
To build a realistic resistivity variation due to the leachate infiltration, several examples of variations of resistivity associated with
leachate injections can be found in the literature. Several studies
have shown that resistivity decreases by 60% to 70% when a
highly conductive leachate is injected in the top soil layer (Yoon
and Park, 2001). Inside the waste layer, other studies show that
resistivity decreases by 30% to 60% (Guerin et al., 2004; Moreau
et al., 2003; Rosqvist et al., 2005). Taking into account these studies, the resistivity variations were set as follows, for each layer
resistivity:
In the first layer (top soil), resistivity decreases by 70%: the
resistivity is 100 ohm m before injection and 30 ohm m after
injection.
In the waste, leachate resistivity decreases by 60%: the resistivity is 15 ohm m before injection and 6 ohm m after injection.
Fig. 1b shows the synthetic resistivity ratio (i.e. what should be
ideally obtained with time-lapse ERT). We present the relative variation of bulk resistivity between initial and final synthetic models.
In this study, whatever the resistivity being considered (apparent,
calculated, or bulk) the variation of resistivity is expressed as a per-
centage change. If the resistivity decreases, the percentage is negative. If the resistivity increases the percentage is positive, based
on Eq. (1)
Dq% ¼ ½ðqf =qi Þ 1 100
where Dq% is the percentage variation of the resistivity, qf is the
apparent resistivity at final stage (in ohm m) and qi is the resistivity
apparent at initial stage (in ohm m). In Fig. 1, we present only the image for the 2D case. The image for the 3D case is indeed similar in a
vertical plane, thanks to the axy-symmetrical pattern of the model.
2.3. Tested electrode arrays
Software DC3DInvRes was used to calculate synthetic apparent
resistivity data sets (Günther, 2004). This software uses a finite difference forward calculation. To obtain more realistic apparent
resistivity data, random noise of 1.5% and voltage dependent noise
have been added to simulate a low-noise acquisition. This study focuses on several electrode arrays, some of which are well known
and widely used in geophysical surveys, both for 2D and 3D acquisitions. For 2D, we tested the Wenner-Schlumberger, the dipole–
dipole and the pole–dipole arrays, whilst for 3D; we focused only
on the electrode configuration using only the dipole–dipole array
to limit the length of the paper. Firstly, a classical parallel line array
was tested, and subsequently a star array. Details of these arrays
are outlined below, and in Fig. 2.
2.3.1. 2D arrays
All 2D arrays use an electrode acquisition line that is perpendicularly oriented to the infiltration trench (Fig. 1a, left). This line has
72 electrodes, with a unit spacing of one metre, small enough to
monitor shallow infiltrations, whilst the total length of 71 m is long
enough to investigate an infiltration bulb which was spread down
to 4 m deep.
We chose the Wenner-Schlumberger array (Fig. 2a), as it is
more sensitive to the vertical variation of resistivity. Secondly,
ARTICLE IN PRESS
4
Fig. 2. Array geometries tested for numerical modelling: (a) Wenner-Schlumberger, (b) dipole–dipole, (c) forward pole–dipole, (d) reverse pole–dipole, (e) 3D parallel line
arrays, (f) star array.
we choose the dipole–dipole array (Fig. 2b), due to its sensitivity to
the lateral variations of resistivity. The Wenner-Schlumberger is
frequently used by several authors (Seaton and Burbey, 2002).
Anticipating the results obtained in numerical modelling, it is
hypothesised that the artefact production, in addition to inversion
process, could be generated by the symmetrical geometry of Wenner-Schlumberger and dipole–dipole. Indeed, the two arrays inject
and measure currents and voltages with a symmetrical pattern
regarding the injection bulb similarly as artefact pattern. To investigate the possible effect of array symmetry on artefact production,
we chose the pole–dipole forward and reverse arrays (Fig. 2c and
d), a typical asymmetrical array described in many publications
(Telford et al., 1991). This array is becoming popular due to providing better penetration depth, lateral coverage, and sensitivity to
both lateral and vertical variations of resistivity (Loke, 2004). For
those reasons Grellier et al. (2008) have used this array to monitor
leachate injection in landfills. However, the main disadvantages of
this array are (i) the need of an electrode to be located ‘‘at infinity”
(at a distance more than five times the maximum spacing used in
measurement sequence) and (ii) the need to obtain a double data
set (called forward and reverse) which double the acquisition time,
and this can be troublesome in time-lapse ERT for monitoring fast
phenomena. We generated two data sets with forward and reverse
pole–dipole array (Fig. 2c and d), merged into the same inversion.
sampled if it is faster than the acquisition time. We also tested another 3D electrodes array with four-line layouts with a star pattern
(Fig. 2d). Again, there were no quadrupole connections between
each line.
2.3.2. 3D arrays
For the 3D case, we used only the dipole-dipole array to limit
the length of the paper. First an electrode set up with five parallel
lines was used, with 48 electrodes only to limit the calculation
time required for calculations with Gauss–Newton inversion. The
unit electrode spacing is still 1 m and the lines were equally separated by three metres (Fig. 2e). For this 3D array with parallel lines,
we did not simulate any current injection between adjacent lines.
Indeed such a connection protocol would require more time to be
created in the field. The phenomena under study could be under-
2.4.1. Inversion with standard parameters
The following standard time-lapse inversion parameters were
used in this study namely an isotropic smoothness constraint,
Gauss–Newton minimization, and a fixed regularization parameter
(regularization parameter k = 30), see Günther (2004). First, the
initial model without infiltration is inverted. Second, we used the
resulting calculated model of the initial state as a reference model
in the time-lapse inversion of the two final infiltration models. Finally, we calculated the ratio of calculated resistivity (final calculated resistivity model divided by initial calculated resistivity
2.4. Time-lapse inversion procedures
The third step is the inversion of the synthetic data set using (a)
inversion parameters used commonly used in previous studies for
time-lapse ERT, referred to as ‘‘standard inversion” is this paper
and (b) advanced inversion tools that produce significant improvement for artefact removal. Both standard and advanced calculated
models are then compared to the initial synthetic models to evaluate the efficiency of advanced inversion tools on artefact removal.
The inversions were performed using DC2DInvRes and DC3DInvRes
software packages (Günther, 2004), which allowed the introduction of a priori information into the time-lapse inversion procedure. For the 2D case, Wenner-Schlumberger and dipole–dipole
arrays are inverted independently in a first step, and combined
in the same inversion in a second step, as proposed by de la Vega
et al. (2003) or Loke (2004). The authors also used forward and reverse pole–dipole in the same inversion. This data set was not
combined with other array in order to test only the effect of the
asymmetry.
ARTICLE IN PRESS
model). This time-lapse image was then compared to the timelapse image shown in Fig. 1. This procedure is common for most
time-lapse surveys as proposed by Loke (1999). In this study, the
‘‘blocky model” option was used in order to yield sharp resistivity
contrasts. The grid is chosen with two cells between neighbouring
electrodes. We used a user-defined logarithmic thickness for the
cells. Günther (2004) provides a detailed description of these
parameters.
2.4.2. Inversion with advanced procedures
We tested a first procedure fixing invariant resistivity regions in
2D and 3D inversions. The resistivity value of the selected regions
does not change with time. Such an approach was already experimented by Descloitres et al. (2008b) but for a fixed substratum at
depth. In the leachate injection context, we assumed knowledge of
an invariant region not only at depth (where the leachate does not
flow) but also on both sides of the injection point, a few metres
away from leachate injection influence. We considered that the
fixed region can be known by external methods such as neutron
probe monitoring, or electromagnetic profiling or soundings as
shown in our field example. Such regions are then considered as
a priori information that can be incorporated into the inversion
procedure. For synthetic modelling, fixed regions geometry is arbitrarily fixed.
The second procedure was applied directly into the inversion
process. We used an alternative constraint method that minimizes the variation from one calculated model (initial model)
to another model (subsequent or final model). Basically, the first
model (initial state) is calculated using a smoothness constraint
commonly used in ERT inversion. Then, the second data set is inverted using a minimum length constraint (Günther, 2004; Loke,
1999). In doing so, the inversion tries to minimise changes from
the initial model regardless of the neighbouring relations such
that merely the L2 norm of the model vector difference is
minimised.
5
3. Numerical modelling results
3.1. Example of synthetic data
Fig. 3a, presents an example of apparent resistivity data sets for
symmetrical arrays (Wenner-Schlumberger and dipole–dipole).
The pole–dipole array is omitted to lengthen Fig. 3. We plotted
the percentage of variation of the final apparent resistivity in relation to the background initial model after infiltration (Fig. 3b).
The apparent resistivity decreases for data close to the infiltration point (pointed with a blue arrow). The apparent resistivity increases at intermediate acquisition levels (red arrow). All profiles
show a decrease of apparent resistivity in the central section,
which corresponds to leachate infiltration. The apparent resistivities on both sides of the infiltration bulb increase at intermediate
levels close to the injection (Fig. 3b). Synthetic apparent resistivity
obtained with the Wenner-Schlumberger array increase by +20%
on both sides of the injection. With 3D injection, apparent resistivity increases by +38%. These increases are symmetrical on both
sides of the injection. The percentages of increase of apparent
resistivity are highest with a 3D injection. The dipole–dipole array
shows an increase of apparent resistivity under the infiltration
with an increase by +40% with the 2D infiltration. In regards to
the 3D infiltration, apparent resistivity increases by more than
+60%. The dipole–dipole array generates two diagonals where the
apparent resistivity decreases by 62% both for 2D and 3D. These
variations of apparent resistivities can be however far from real
variations in the ground. Data sets need to be inverted to reconstruct real model using an inversion procedure (Loke, 2004) to be
able to reconstruct the leachate injection geometry.
3.2. Standard inversion of synthetic data
Prior to analysing the results, we decided to set the limits of
detection of calculated resistivity variations between 5% and
Fig. 3. Forward modelling: (a) synthetic apparent resistivity data obtained with DC3DInvRes software. A Gaussian noise of 1.5% and voltage dependent noise are added to the
synthetic data. Two arrays are used, Wenner-Schlumberger and dipole–dipole with a unit electrode spacing of 1 m. The apparent resistivity pseudo section is presented along
a line of 71 m. (b) Variation of apparent resistivity (ratio) for 2D and 3D models.
ARTICLE IN PRESS
6
+5%. This range is close to variation of apparent resistivity obtained
in the field with very noisy data sets. Inside this range, the resistivity variations could not be accurately described. Therefore any
time-lapse ERT survey targeting such a small variation of resistivity
with time is not considered in this study.
Also, to clarify the description of the results we have identified
four model areas corresponding to different types of calculated
resistivity variations. The guidelines are as follows:
AIs areas correspond to shallow Artefacts of Increase (AI) of calculated resistivity that could exist around the infiltration trench
in the soil cover (i.e. between 0 and 1.5 m deep).
AId areas correspond to deeper Artefacts of Increase of calculated
resistivity, between 1.5 m and 15 m.
AD areas correspond to Artefacts of Decrease (AD) of calculated
resistivity.
RV area corresponds to Real Variations (RV) of resistivity corresponding to the real infiltration zone.
3.2.1. 2D case
Fig. 4 outlines the results obtained over 2D infiltration pattern
using (a) symmetrical arrays: the Wenner-Schlumberger and the
dipole–dipole and combining these two arrays (i.e. Wenner-Schlumberger and dipole–dipole) into the same inversion and (b)
the asymmetrical array, the pole–dipole array, using an inversion
combining both pole–dipole forward and reverse data sets.
For the symmetrical arrays, Wenner-Schlumberger and dipole–
dipole arrays, and their combination, we noted the following:
In the central zone (Fig. 4a–c, RV area) the two arrays and their
combination reconstructed correctly the decrease of the resistivity within the area of infiltration, with a calculated resistivity
that decreases by 60% to 70%, close to the required model
value of 60%.
On both sides of infiltration, there are unexpected variations of
the calculated resistivity for Wenner-Schlumberger only
(+50%), noted AIs in Fig. 4a.
Deeper, in areas noted AId located around and sometime below
the infiltration; there are unrealistic increases of calculated
resistivity with a significant value of +30% to +40%. These areas
are noted whatever the symmetrical array used or their combination. They are typical artefacts of increase of calculated resistivity. This false resistivity increase could be considered at a first
glance as drying phenomenon (such as biogas driven deep down
by the piston effect of the bulb).
Under the infiltration we noted for the Wenner-Schlumberger
only a decrease in resistivity between 4 m depth to 14 m. This
decrease is about 40% with the standard inversion, noted as
AD area. Again, this false decrease of resistivity could be considered as false infiltration phenomena if not recognised.
The results obtained using the pole–dipole arrays are presented
in Fig. 4d. The standard inversion shows a decrease of resistivity by
60% at the centre of the profile, according to the synthetic infiltration pattern. At shallow depth, below two metres, weak artefacts of
increase of resistivity are noticed with the infiltration bulb being
slightly larger than the true bulb. The major result is the removal
of strong artefacts originally obtained with symmetrical arrays
and their combination.
3.2.2. 3D case
To facilitate the presentation of 3D model results, this study presents only a selected cross-section located under electrode line
(drawn in blue in Fig. 2 for both parallel line and star array). Identical
results are obtained in other directions. Fig. 5 presents the results obtained with a 3D standard inversion using only the dipole–dipole
configuration. We outline the result obtained with both parallel line
Fig. 4. Standard time-lapse inversion of 2D synthetic data, with: (a) Wenner-Schlumberger array, (b) dipole–dipole array, and (c) combining Wenner-Schlumberger and
dipole–dipole arrays and (d) pole–dipole (combining forward and reverse data sets). The synthetic data were inverted with smoothness constraints. Areas are indicated as
follows: AIs = Artefact of Increase of resistivity at shallow depth; AId = Artefact of Increase of resistivity deeper; AD = Artefact of Decrease of resistivity; RV = Real variation of
resistivity.
ARTICLE IN PRESS
7
Fig. 5. Standard time-lapse inversion of 3D synthetics models using dipole–dipole data set. Tested electrode set up: (a) parallel line array; (b) star array.
array and star array in Fig. 2 and performed the same 3D inversion
with a dipole–dipole array. The results obtained with Wenner-Schlumberger array are not significantly different from the dipole–dipole array. Consequently, they are not presented in this article.
With the parallel lines array (Fig. 5a), the resistivity increases
inside the superficial layer between 0 and 1.5 m on both sides of
the trench infiltration. These variations are between +10% and
+20%. The calculated resistivity variation is 68% in the injection
pit. This variation is in-line with the real resistivity decrease
(60%). In the waste between 1.5 and 4 m deep, at horizontal distance between 21 and 28 m, the variation of calculated resistivity
ranges from 60% to 50%. This variation represents the leachate
infiltration. In the lower part of the profile (between 5 and
9 m), the resistivity decreases by 50% at the centre. This is not
the real infiltration. Therefore, significant artefacts of increase of
resistivity (more than +30%) are also seen with 3D case around
the infiltration. These increases are comparable to the increases
generated with a 2D code. Finally, we demonstrate that 3D standard inversion using parallel lines (without connecting the lines
between each other) does not remove the artefacts.
With the star array, the results are presented in Fig. 5b. With
standard inversion, the result shows a single area of decreased
resistivity at the centre (50%) corresponding to the real infiltration. On both sides the resistivity variation ranges from 5% to
+5%. We show that the star array is much more adequate to reconstruct the real infiltration pattern than parallel line array, even if
the lines are not connected to each other.
3.3. Advanced inversion of synthetic data
3.3.1. 2D case
Similarly to the results presented for standard inversion, Fig. 6
shows the results obtained over 2D infiltration pattern using symmetrical arrays: the Wenner-Schlumberger, the dipole–dipole and
combining these two arrays into the same inversion. The advanced
inversion procedure is not applied to the pole–dipole array in this
paper because the pole–dipole with a standard inversion provides
satisfactory results by itself.
For Wenner-Schlumberger and dipole–dipole arrays, we noted
the following:
In the central zone (Fig. 6a and b, RV area), the ‘‘fixing region”
tool, the calculated resistivity decreases by 60% to 70%. It corresponds well to variations of real resistivity (60%). With the
‘‘minimum length” tool, or combining ‘‘minimum length” and
‘‘fixing region”, the variation of calculated resistivity in the RV
zone is between 45% and 55%. Thus the real decrease of resistivity is slightly minimized by the inversion.
On both sides of infiltration at very shallow depth, there are
unexpected variations of the calculated resistivity by +20% to
+48%, noted AIs in Fig. 6a, but only for the Wenner-Schlumberger
array.
Deeper around the infiltration, there is a significant improvement in artefact removal (seen previously with standard inversion, see Fig. 4a). With the ‘‘fixing region” tool, ‘‘minimum
length” tool and combining them, there are no longer unexplained increases of calculated resistivity, excepted a slight artefact for the dipole–dipole array with ‘‘fixing region” (10% to
+6%). Changes in calculated resistivity are in the region of +5
or 5% considered as out of the detectability limit.
A false decrease in resistivity was noted just under the infiltration (AD area) but their extensions are significantly limited,
remaining between 4 and 6 m deep, and exhibiting variations
between +5% and 20%.
If significant improvements are obtained when considering
Wenner-Schlumberger and dipole–dipole array independently,
ARTICLE IN PRESS
8
Fig. 6. Advanced inversion of 2D synthetic data with: (a) Wenner-Schlumberger array, (b) dipole–dipole array, and (c) combining Wenner-Schlumberger and dipole–dipole,
arrays. The synthetic data were inverted with: (1) smoothness constraints for both initial and final data sets and fixing an invariant region around the injection; (2) using a
smoothness constraint for the initial data set and a minimum length constraint for final data set and (3) combining fixing regions and minimum length tools. Areas are
indicated as follows: AIs = Artefact of Increase of resistivity at shallow depth; AId = Artefact of Increase of resistivity deeper; AD = Artefact of Decrease of resistivity; RV = Real
variation of resistivity.
the combination of the two data sets into the same inversion and
applying advanced inversion tools does not improve the result.
As seen in Fig. 6c, on both sides of the infiltration, there is a persistence of areas AIs and AId, whatever the advanced inversion tool
used (increase of calculated resistivities from +10% to +40%). Several tests (not shown) have been made to lower the artefact production using different values of regularization factor. We
conclude that some improvement could be achieved, but without
eliminating artefacts. Moreover, the choice of an optimised regularisation factor was not considered in this study, in an attempt
to keep a constant regularisation factor for all inversions, allowing
their inter-comparison.
3.3.2. 3D case
We applied the two advanced inversion tools (‘‘fixing region”
and ‘‘minimum length”) to the parallel line array only, that exhibited significant artefact as shown previously in Fig. 7 using standard inversion parameters. Advanced tools are not tested on the
star array, due to providing satisfactory results for artefact removal
(see Fig. 5).
For the ‘‘fixing region” tool, the inversion results (Fig. 7a) show
that resistivity variations are negligible at shallow depth between
0 and 1.5 m. The calculated resistivity variation remains between
+5% and 5%. Infiltration is well detected at the center of the profile. The variation of resistivity is from 40% to 60%. In the waste
layer between 1.5 and 4 m, the calculated resistivity variation remains within the range of 50% to 70% at the centre of the profile, as expected. Apart from the infiltration bulb, the calculated
resistivity variation remains weak (5% to +5%). In this area arte-
facts are totally removed. Below 1.5 m, the variation of resistivity
remains between +10% and 10%.
For the ‘‘minimum length” tool, the results show that in the first
layer between 0 and 1.5 m, there are limited variations between
5% and +5% (Fig. 7b). But some isolated blocks show increases
of resistivity (+20 to +40%). This is an effect of the minimum length
tool that results sometimes in scattered resistivity values for
blocks close to the surface. This situation is not troublesome when
looking for large patches of resistivity changes. In other studies, if
very small patches of resistivity are under consideration, care
should be taken when using this tool for interpreting shallow variations. At the centre of the waste between 1.5 and 4 m, calculated
resistivity decreases by 50% at the correct location. On both sides,
resistivity variation range is limited from 5% to +8%. Below 4 m,
the variation of resistivity remains weak within +/2%. There are
no changes deeper, in agreement with the model.
3.4. Summarizing numerical modelling results
With standard time-lapse inversion, using both symmetrical array for 2D case and a parallel line array for the 3D case, we demonstrated that variations of calculated resistivity can be greater than
+40% in unexpected areas. Further, an artefact production could
lead to severe misinterpretation. The advanced inversion tools significantly limits (‘‘fixing region”) or remove totally (‘‘minimum
length”) the increase of calculated resistivities whatever the array
used, if resistivity variations within +/5% are ignored. The real
variation of resistivity remains very close to the model value (but
ARTICLE IN PRESS
9
Fig. 7. Advanced inversion of 3D synthetic data using the Wenner-Schlumberger array and an electrode set up using five parallel lines (see Fig. 2c). The synthetic data were
inverted with: (a) smoothness constraints for both initial and final data set and fixing an invariant region around the injection (grey area); (b) using smoothness constraints
for the initial data set and a minimum length constraint for final data set without fixing invariant regions.
slightly underestimated with ‘‘minimum length”). Both advanced
inversion tools reconstruct the expected geometry properly.
When combining the two advanced tools in the same inversion,
it improves even the results obtained if Wenner-Schlumberger and
dipole–dipole data are taken separately. If two arrays are combined into the same inversion in addition to the combination of
advanced tools, it does not improve the ERT reconstruction model
at all. This is contrary to what was expected using these combinations as one of the most appropriate solution. Furthermore, it generates wrong oscillations with calculated resistivity increases of
about +20%. Further improvement at this stage was not carried
out, however for future modelling, a regularisation tuning could
be considered.
Regarding the effect of asymmetrical array for 2D case, the
study showed that the pole–dipole array could reduce or even
eliminate the artefact production even if standard inversion
parameters are used. For 3D case, we found that using a star array
instead of parallel lines, it is possible to eliminate the artefacts,
even if standard inversion parameters are used, at least when the
lines are not interconnected.
4. Field experiment results
4.1. Experimental setup
The experimental site is located in southern France. It is a class
2 landfill for non-hazardous municipal waste. The layering of the
deposit consists of a 1 m thick soil cover (Fig. 8). This soil is separated from the waste with a geotextile GCL (Geosynthetic Clay
Liner) with very low permeability. Below the geotextile, the waste
thickness is from 15 to 25 m. The landfill cell is equipped with a
biogas extraction system.
Fig. 8. Location and description of the experimental site and geophysical survey. Location of the ERT lines, EM31 and neutron probe measurements.
ARTICLE IN PRESS
10
A leachate injection was carried out using a pit located in the
centre of the star shape array set up. The pit was dug so that it
reached the top of the waste deposit. The geotextile GCL has been
removed from the bottom of the pit. The pit was 2 2 m with a
depth of 1 m. The injection lasted 72 h. About 10 m3 of leachate
were injected, maintaining a constant hydraulic head H of 0.6 m
at the bottom of the pit. During the experiment, the leachate was
stored in a tank located 35 m away from the pit, with the leachate
electrical conductivity and temperature remaining almost constant
(12700 ls cm1 at 20 °C).
We took advantage of this experiment to monitor the infiltration plume migration using time-lapse ERT. A Syscal PRO resistivity
meter was used, combined with a Switch Pro unit (IRIS instruments, Orleans, France). 168 electrodes were placed on the soil
cover and the electrodes remained on the site during the experiment. Measurements were taken of the contact resistance of electrodes before each measurement. For all dipoles, the contact
resistance was always less than 4 k ohm. During the experiment,
the weather remained dry, avoiding shallow resistivity changes
that could have produced shallow resistivity artefact as evidenced
by Clément et al. (2009). Two arrays were applied (Wenner-Schlumberger and dipole–dipole). Both arrays allow us to operate with
a fast (10-channels) acquisition mode. To get benefits from the
synthetic modelling presented above, we used a star array as
shown in Fig. 5. Four independent electrode lines were used with
1 m unit electrode spacing. The star is built with one line of 72
electrodes, a perpendicular line of 24 electrodes. Two other lines
of 36 electrodes are oriented at 45° from the previous ones.
We also used in-line acquisition sequences. No inter-line measurements were used as it was necessary to collect the data as fast
as possible in case of fast leachate migration. Every hour, four data
sets were acquired with dipole–dipole array, resulting in a total of
30 data sets being collected. This article will present only two data
sets: the initial data and the data taken 40 h after the beginning of
the injection process, which depicts representative phenomena. To
compare time-lapse ERT field results, additional geophysical surface measurements and neutron probe loggings were carried out
before, during, and after the injection (Fig. 8). First, we conducted
electromagnetic measurement profiling at the surface using frequency domain electromagnetic (FDEM) profiling system EM31
device (Geonics Ltd.). The vertical dipole configuration was used,
allowing the deepest investigation. FDEM profiling is a popular
geophysical method widely used for soil surveying, which has been
outlined in McNeill (1980). Some studies report FDEM monitoring
of spatial and temporal changes in soil salinity (Corwin et al.,
2006). For waste, Guerin et al. (2004) reported a successful mapping of the waste cell using EM31. The main advantages of using
electromagnetic profiling in this study are the following: firstly,
the EM31 with vertical dipole mode provides a suitable investigation depth to focus on the main infiltration phenomena without
being too sensitive to very shallow variations of resistivity (McNeill, 1980). Secondly, EM is very sensitive to conductive ground, as
waste. The EM31 measures an apparent electrical conductivity
(ECm) in mS/m. To show the lateral infiltration extension (and consequently invariable zones around), an initial and a final profile
were achieved. ERT cables were removed to avoid any disturbance
with FEM measurements (EM induction into the electrode cables).
We used neutron probe logging performed in some borehole
drilled around the injection point. Neutron logging is a well known
method for the detection of water content variation in soils and
rocks. In this study, a lack of calibration (technically difficult to
complete in waste) did not allow to derive water content variations. However, the neutron signal variation can clearly be interpreted when the leachate penetrates the volume of influence of
the probe. We drilled two bore holes before the experiment
equipped with access tube for neutron probe logging.
Fig. 9. Field results using the star array. Time-lapse ERT inversion results 40 h after
the starting of the leachate injection. The longest line (72 electrodes) is presented as
a 2D cross-section to lighten the figure. Comparison of time-lapse ERT results with
EM31 (above) and neutron probe logging (below).
4.2. Time-lapse ERT inversion results
The results are presented in Fig. 9. The results show that the calculated resistivity variation is close to 0% in the shallow soil layer
between 0 and 1.5 m. At a horizontal position X between 0 and
30 m and between X = 40–72 m, variations of calculated resistivity
are within the range 5% to +8%. These variations are unexpected
(no rain producing shallow infiltration during experiment). Those
variations are not confirmed with electromagnetic profile, where
EM31 shows stable values. At the center of the profile at
X = 36 m, the resistivity decreases by 60%. According to resistivity
variation measured in situ inside the injection chamber (not
shown), this result was confirmed. On both sides of the infiltration
pit at very shallow depth, we noted the presence of small patches
of false increase in resistivity of +20% to +30%. These increases are
similar to small shallow AIs area evidenced on the synthetic models
in 2D and 3D. In the waste layer between 1.5 and 6 m deep and between X = 30–38 m, the calculated resistivity decrease by 60%,
which should correspond to the infiltration plume. On both sides,
ERT shows a lower decrease in resistivity: 30% to 60%. This decrease should correspond to the lateral extension of the plume. The
decrease of resistivity extends to 4 m depth, excepted at the centre
of the profile where it is limited to 2.5 m.
The deeper area below 5 m does not show any significant resistivity variation between 5% and +5%.
5. Discussion
5.1. Comparison of time-lapse ERT field results with external data
Fig. 8 shows EM31 measurements performed along a profile
parallel to the longest electrode line (72 m) the results are shown
ARTICLE IN PRESS
in Fig. 9. The EM31 conductivity did not vary between X = 0 and
20 m during infiltration. There was no variation in this area in
the range between 0 and 6 m deep, which is in agreement with
the ERT results that did not show any variations at that area.
Between X = 16–50 m, the conductivity increases from about
110 mS/m (initial state) to 140 mS/m (after injection) (see Fig. 9).
This is in perfect agreement with the decrease of calculated resistivity seen in the ERT profile. In the zone between 50 and 72 m,
there is no major change in conductivity. The EM31 shows the
same slight asymmetry of infiltration (the plume is more extended
to the left of the injection pit rather than to the right). Thus EM31 is
able to confirm the limited lateral extension of the injection delineated with time-lapse ERT.
A comparison with data obtained with neutron logging is
shown in Fig. 9. Along the ERT profile, two neutron loggings were
implemented respectively at X = 20 m (depth to 4 m) and at
X = 36 m (depth of 8 m). At X = 20 m, the neutron probe result
shows a counting ratio that is identical between the initial state
and the final state. Thus there were no changes in water content.
This result is in agreement with time-lapse ERT, which shows no
variation of resistivity in this section. At X = 36 m, the counting rate
increases between 0 and 2.5 m. The infiltration reached 2.5 m
deep. ERT data show the same geometry with a decrease of resistivity of up to 3 m, slightly deeper than neutron probe logging.
We demonstrated here the good agreement between the data obtained with EM31 and neutron probe with the results of the
time-lapse ERT, as predicted with numerical modelling.
11
5.2. Robustness of advanced inversion tools: differentiation between
artefacts and true biogas migration
Standard ERT time-lapse inversion can produce false resistivity
increase artefact as shown in Fig. 4. In the modelling result part, we
have shown that using advanced tools on the 2D infiltration scenario, we were able to fairly evaluate the geometry of the infiltration phenomenon while avoiding the incidence of unwanted
increases of calculated resistivity. This then begs the question: if
a true increase of resistivity occurs within the subsurface during
the injection process, are the advanced tools still able to image
not only the injection plume (resistivity decrease) but also the drying phenomenon or biogas migration (resistivity increase) that can
also occur? In other words, are the advanced tools robust enough
to remove artefacts but also to reconstruct both true increases
and decreases of resistivity at the same time? To answer this question, we used the same numerical modelling approach. A more
complex geometry was used and four 2D infiltration models were
chosen, with true increase of resistivity located in different areas at
depth. They are presented in Fig. 10. From scenario A to scenario D
(Fig. 10A–D, respectively) the true increase in resistivity is +100%,
and is located successively at four different areas, the first close to
the surface at the left of infiltration (A), the last one just below the
infiltration plume (D). The apparent resistivities data sets were
generated for a dipole–dipole array. The apparent resistivities were
inverted using two advanced inversion tools proposed in the results part, i.e. fixing invariant regions at depth and laterally and
Fig. 10. Simulation of a true biogas migration during a 2D leachate injection. Four scenarios of increase of resistivity are presented built (A, B, C, and D). A real increase of
resistivity corresponding to a desaturation of the waste with biogas migration at different positions around the infiltration is display with black colour. A decrease of
resistivity corresponding to leachate plume migration is display with blue colour. The inversions of dipole–dipole 2D synthetic data are done using ‘‘fixing region” (middle)
and ‘‘minimum length” (right). Areas are indicated as follows: AIs = Artefact of Increase of resistivity at shallow depth; AId = Artefact of Increase of resistivity deeper;
AD = Artefact of Decrease of resistivity; RV = Real variation of resistivity. (For interpretation of the references to colour in this figure legend, the reader is referred to the web
version of this article.)
ARTICLE IN PRESS
12
using a minimum length constraint when inverting the second
data set.
For scenario A with the ‘‘fixing region” tool, in the range 0–
1.5 m, resistivity decreases by 60% at the center. Between 1.5
and 4 m deep, resistivity decreases by 60% at the center, which
is in agreement with the infiltration bulb. To the left of the infiltration there was an increase of calculated resistivity of +80% to
+100% in agreement with the synthetic model. Elsewhere there is
no significant variation greater than +5% or below 5%. Therefore,
this advanced tool is considered as efficient to reconstruct both decreases and increases of resistivity.
Scenarios B and C show similar results as scenario A, but with
slight differences: the infiltration bulb and the areas with increasing resistivity are fairly delineated. However, it is noted that the
real increase of resistivity distorts slightly the shape of the infiltration bulb. With the ‘‘fixing region” tool, there is persistence of false
increase or decrease of resistivity around the gas bulb, especially
close to the edge of the fixed regions. Therefore care should be taken when using the ‘‘fixing region” tool when interpreting small
anomalies. To the contrary, better results are obtained for scenarios B and C using the ‘‘minimum length” tool as demonstrated in
Fig. 10: it is noted that only weak artefact is still present at shallow
depth for scenario B.
For scenario D the increase of calculated resistivity is correctly
located but its value is only +30% instead of +100%. The same results were obtained using the ‘‘fixing region” tool or ‘‘minimum
length” tools. The calculated resistivity variation is clearly minimized by the inversion. This result is however in-line with the
physics of the electrical resistivity method. Indeed, there is a significant loss of resolution with depth when using surface electrical
methods, such as ERT (see Telford et al., 1991). Therefore it is
important to note that the advanced inversion tools proposed in
this paper to reduce or even eliminate artefacts cannot overcome
this classical limitation.
6. Conclusion
Electrical resistivity tomography is becoming popular to monitor leachate injection plume within waste during recirculation in
bioreactors. The starting point of this study was the evidence of
artefact in time-lapse ERT images in some previous studies dedicated to monitor natural hydrological processes such as infiltration
below streams. Those artefacts were obtained after a time-lapse
ERT inversion using standard parameters. They typically result in
increases in calculated resistivity in areas where the resistivity remained actually the same. Such a situation is troublesome to
reconstruct reliable hydrological processes: indeed, an increase in
resistivity could be interpreted wrongly as a loss of water. The
occurrence of such artefacts has been investigated in this study
for two leachate recirculation scenarios. We used a classical approach using numerical modelling of typical injection scenario
with 2D and 3D geometry. These scenarios correspond to injection
in shallow trench or pit respectively. Three well known arrays were
used (two symmetrical and one asymmetrical) for calculating the
synthetic apparent resistivity data sets for the 2D acquisition,
and two different electrodes arrays (parallel lines and a star array)
for 3D acquisition.
For the 2D case, the numerical results showed that when standard time-lapse inversion parameters are used, typical artefacts result in an increase of at least +30 to +50% if the symmetrical arrays
are used. They are located around the true infiltration where the
resistivity decreases combining these different symmetrical arrays
into the same inversion could lead to worse results. The asymmetrical array (i.e. the pole–dipole) was tested successfully to remove
the artefact, even if standard time-lapse inversion parameters are
used. This result gives promising perspectives for future time-lapse
ERT surveys. However, this array could not be easily applied in
some survey conditions on waste landfills as it requires longer distances to locate the electrode at infinity. We conclude that further
studies should be done in the future to explore the advantages and
limitations of asymmetrical arrays for time-lapse ERT, like multigradient array as proposed by Dahlin and Zhou (2004).
For the 3D case, we have shown that artefacts are also present
when using parallel lines (without inter-connection between the
lines). On the contrary, the star array is efficient in removing artefacts, even using standard inversion parameters.
For symmetrical arrays and for parallel line array (for which
artefacts are persistent with standard inversion parameters) two
advanced inversion procedures were tested to remove the artefacts. These procedures involve two inversion tools used alone or
jointly. The first advanced inversion tool use invariant regions into
the inversion. The second one uses a minimum length constraint
instead of a smoothness constraint for the inversion of the second
data set. The two advanced tools were tested both for 2D and 3D
geometries. The ‘‘fixing region” tool removes almost completely
the artefact whatever the symmetrical array used (Wenner-Schlumberger or dipole–dipole). This option requires however an a
priori knowledge of the invariant regions, that can be achieved
using external methods. For ‘‘minimum length” tool, we have
shown that it removes almost completely the artefacts. However,
the decrease of resistivity inside the infiltration plume is slightly
underestimated: in our numerical modelling, we have shown that
an expected decrease of 60% is reconstructed only with a 40%
value. Therefore, care should be taken when trying to interpret
the resistivity ratio in terms of water content (if such a relationship
can be obtained in the field).
Following the results obtained with the numerical modelling,
we tested the star array during a 3D leachate injection field experiment. Time-lapse ERT results were compared to external data obtained in the field during injection. This comparison with (i)
electromagnetic profiling and (ii) borehole neutron probe data,
are in accordance with ERT time-lapse imaging.
At last, the robustness of the advanced tools is tested using a
more complicated infiltration model that not only shows a true decrease of resistivity, but also a true increase of resistivity that could
be due to biogas migration around the infiltration plume. The advanced tools are able to reconstruct satisfactorily the biogas migration, but with a decreasing resolution in depth, as expected with
ERT.
From the results obtained both with numerical modelling and
the field data, we foresee significant improvement in leachate
recirculation imaging. Due to the complicated process of ERT inversion, we advocate for a numerical approach on simulated injection
scenario before building the field ERT setup. This numerical modelling is used to evaluate both ERT layout and interpretation strategy, as well as for an early recognition of possible artefacts. The
advanced procedures proposed in this study can be useful to many
other geological or hydrological situations where time-lapse ERT is
considered, or for Induced Polarization (IP) data sets. This study
also demonstrated that information obtained with external geophysical methods can be of significant advantage to obtain more
reliable ERT time-lapse results.
Acknowledgements
This work was funded and conducted by LTHE (Laboratoire
d’étude des Transferts en Hydrologie et Environnement) and the
ANR PRECODD project ‘‘Bioréacteur”. We greatly acknowledge
VEOLIA Property for providing access to the pilot landfill and to
very convenient facilities for leachate injection. Mustapha Hidra
is warmly acknowledged for his support in this project. We also
ARTICLE IN PRESS
thank the LTHE teams ‘‘HydroGeophysics” and ‘‘pôle expérimentation”, with a special mention to Konstantinos Chalikakis, Hélène
Guyard, Etienne Maury, Henri Morra, Lisa-Maria Mic, Lucas Muller
and Truong Tran Xuan.
References
Acworth, R.I., Jorstad, L.B., 2006. Integration of multi-channel piezometry and
electrical tomography to better define chemical heterogeneity in a landfill
leachate plume within a sand aquifer. Journal of Contaminant Hydrology 83 (3–
4), 200–220.
al Hagrey, S.A., 2007. Geophysical imaging of root-zone, trunk, and moisture
heterogeneity. Journal of Experimental Botany 58 (4), 839–854.
Benson, A.K., 1995. An integration of geophysical methods and geochemical analysis
to map acid mine drainage – a case study. Exploration and Mining Geology 4
(4), 411–419.
Benson, A.K., Payne, K.L., Stubben, M.A., 1997. Mapping groundwater contamination
using dc resistivity and VLF geophysical methods – a case study. Geophysics 62
(1), 80–86.
Bernstone, C., Dahlin, T., Ohlsson, T., Hogland, W., 2000. DC-resistivity mapping of
internal landfill structures: two pre-excavation surveys. Environmental
Geology 39 (3–4), 360–371.
Clément, R., Descloitres, M., Günther, T., Ribolzi, O., Legchenko, A., 2009. Influence of
shallow infiltration on time-lapse ERT. Experience of advanced interpretation.
Comptes Rendus de l’Académie des Sciences Earth and Plantary Sciences,
doi:10.1016/j.crte.2009.07.005.
Cossu, R., Ranieri, G., Margisio, M., Sambuelli, L., Godio, A., Motzo, G.M., 1990.
Geophysical methods in surveying old landfills. In: Arendt, F., Hinsenvald, M.,
Van den Brink, W.J. (Eds.), Proceedings Third International KfK/TNO Conference
on Contaminated Soil, December 1990.
Corwin, D.L., Lesch, S.M., Oster, J.D., Kaffka, S.R., 2006. Monitoring managementinduced spatio-temporal changes in soil quality through soil sampling directed
by apparent electrical conductivity. Geoderma 131 (3–4), 369–387.
Cossu, R., Di Maio, R., Fais, S., Fraghi, A., Ligas, P., Menghini, A., 2005. Physical and
structural characterisation of an old landfill site by a multimethodological
geophysical approach. In: 10th International Waste Management and Landfill
Symposium, Sardinia 2005, S. Margherita di Pula, Cagliari, Italy.
Dahlin, T., Zhou, B., 2004. A numerical comparison of 2D resistivity imaging with 10
electrode arrays. Geophysical Prospecting 52 (5), 379–398.
Day-Lewis, F.D., Lane, J.W., Harris, J.M., Gorelick, S.M., 2003. Time-lapse imaging of
saline-tracer transport in fractured rock using difference-attenuation radar
tomography. Water Resources Research 39 (10), 1740–1752.
de la Vega, M., Osella, A., Lascano, E., 2003. Joint inversion of Wenner and dipole–
dipole data to study a gasoline-contaminated soil. Journal of Applied
Geophysics 54 (1–2), 97–109.
Deiana, R., Cassiani, G., Kemna, A., Villa, A., Bruno, V., Bagliani, A., 2007. An
experiment of non-invasive characterization of the vadose zone via water
injection and cross-hole time-lapse geophysical monitoring. Near Surface
Geophysics 5 (3), 183–194.
deLima, O.A.L., Sato, H.K., Porsani, M.J., 1995. Imaging industrial contaminant
plumes with resistivity techniques. Journal of Applied Geophysics 34 (2), 93–
108.
Descloitres, M., Ribolzi, O., Le Troquer, Y., 2003. Study of infiltration in a Sahelian
gully erosion area using time-lapse resistivity mapping. Catena 53 (3), 229–253.
Descloitres, M., Ribolzi, O., Le Troquer, Y., Thiebaux, J.P., 2008a. Study of water
tension differences in heterogeneous sandy soils using surface ERT. Journal of
Applied Geophysics 64 (3–4), 83–98.
Descloitres, M., Ruiz, L., Sekhar, M., Legchenko, A., Braun, J.J., Kumar, M.S.M.,
Subramanian, S., 2008b. Characterization of seasonal local recharge using
electrical resistivity tomography and magnetic resonance sounding.
Hydrological Processes 22 (3), 384–394.
Frohlich, R.K., Urish, D.W., Fuller, J., Oreilly, M., 1994. Use of geoelectrical methods
in groundwater pollution survey in a coastal environment. Journal of Applied
Geophysics 32 (2–3), 139–154.
Grellier, S., Guerin, R., Robain, H., Bobachev, A., Vermeersch, F., Tabbagh, A., 2008.
Monitoring of leachate recirculation in a bioreactor landfill by 2-D electrical
resistivity imaging. Journal of Environmental and Engineering Geophysics 13
(4), 351–359.
Guerin, R., Munoz, M.L., Aran, C., Laperrelle, C., Hidra, M., Drouart, E., Grellier, S.,
2004. Leachate recirculation: moisture content assessment by means of a
geophysical technique. Waste Management 24 (8), 785–794.
Guérin, R., Grellier, S., Robain, H., Bobachiev, A., Vermeersch, F., 2004. Monitoring of
leachate recirculation in a bioreactor by electrical resistivity tomography. EAGE
66th Conference and Exhibition, Paris, France.
Günther, T., 2004. Inversion methods and resolution analysis for the 2D/3D
reconstruction of resistivity structures from DC measurements PhD theses.
University of Mining and Technology, Freiberg, Germany.
Haydar, M.M., Khire, M.V., 2005. Leachate recirculation using horizontal trenches in
bioreactor landfills. Journal of Geotechnical and Geoenvironmental Engineering
131 (7), 837–847.
Hossain, M.S., Gabr, M.A., Barlaz, M.A., 2003. Relationship of compressibility
parameters to municipal solid waste decomposition. Journal of Geotechnical
and Geoenvironmental Engineering 129 (12), 1151–1158.
13
Imhoff, P.T., Reinhart, D.R., Englund, M., Guerin, R., Gawande, N., Han, B.,
Jonnalagadda, S., Townsend, T.G., Yazdani, R., 2007. Review of state of the art
methods for measuring water in landfills. Waste Management 27 (6), 729–
745.
Jolly, J., Barker, R., Beaven, R.P., Herbert, A.W., 2007. Time-lapse electrical imaging to
study fluid movement within a landfill. In: Eleventh International Waste
Management and landfill Symposium, Calgliari, Italy.
Khire, M.V., Haydar, M.M., 2003. Numerical evaluation of granular blankets for
leachate recirculation in MSW landfills. In: Proceedings of Sardinia 2003, Ninth
International Waste Management and Landfill Symposium, CISA,
Environmental Sanitary Engineering Centre, Cagliari, Italy.
Khire, M.V., Mukherjee, M., 2007. Leachate injection using vertical wells in
bioreactor landfills. Waste Management 27 (9), 1233–1247.
Loke, M.H., 1999. Time-lapse resistivity imaging inversion. In: 5th Meeting of the
Environmental and Engineering Society European Section, Budapest, Hungary.
Loke, M.H., 2004. Tutorial: 2-D and 3-D Electrical Imaging Surveys. Available online
at <http://www.geoelectrical.com>.
McNeill, D.J., 1980. Electromagnetic Terrain Conductivity Measurements at Low
Induction Numbers. GEONICS, Technical Note TN-6. Available online at <http://
www.geonics.com>.
Meju, M.A., 2000. Geoelectrical investigation of old/abandoned, covered landfill
sites in urban areas: model development with a genetic diagnosis approach.
Journal of Applied Geophysics 44 (2–3), 115–150.
Mondelli, G., Giacheti, H.L., Boscov, M.E.G., Elis, V.R., Hamada, J., 2007.
Geoenvironmental site investigation using different techniques in a
municipal solid waste disposal site in Brazil. Environmental Geology 52 (5),
871–887.
Moreau, S., Bouye, J.M., Barina, G., Oberty, O., 2003. Electrical resistivity survey to
investigate the influence of leachate recirculation in a MSW landfill. In: Nineth
International Waste Management and landfill Symposium, Calgliari, Italy.
Morris, J.W.F., Vasuki, N.C., Baker, J.A., Pendleton, C.H., 2003. Findings from longterm monitoring studies at MSW landfill facilities with leachate recirculation.
Waste Management 23 (7), 653–666.
Nimmer, R.E., Osiensky, J.L., Binley, A.M., Sprenke, K.F., Williams, B.C., 2007.
Electrical resistivity imaging of conductive plume dilution in fractured rock.
Hydrogeology Journal 15 (5), 877–890.
Olofsson, B., Jernberg, H., Rosenqvist, A., 2006. Tracing leachates at waste sites using
geophysical and geochemical modelling. Environmental Geology 49 (5), 720–
732.
Pacey, J., Reinhart, D.R., Hansen, D.L., Townsend, G.T., Johnson, W.H., 1999. Landfill
bioreactor – an innovation in solid waste management. In: 22nd Annual LFG
Symposium, Lake Buena Vista, Florida.
Pohland, F.G., 1980. Leachate recycle as landfill management option. Journal of the
Environmental Engineering Division-Asce 106 (6), 1057–1069.
Radulescu, M., Valerian, C., Yang, J.W., 2007. Time-lapse electrical resistivity
anomalies due to contaminant transport around landfills. Annals of
Geophysics 50 (3), 453–468.
Reinhart, D.R., McCreanor, P.T., Townsend, T., 2002. The bioreactor landfill: its status
and future. Waste Management and Research 20 (2), 172–186.
Reinhart, D.R., Townsend, T.G., 1998. Landfill Bioreactor Design and Operation.
Lewis Publishers, New York, USA.
Rosqvist, H., Dahlin, T., Fourie, A., Röhrs, L., Bengtsson, A., Larsson, M., 2003.
Mapping of leachate plumes at two landfill sites in South Africa using
geoelectrical imaging techniques. In: Nineth International Waste Management
and Landfill Symposium, Calgliari, Italy.
Rosqvist, H., Dahlin, T., Lenhé, C., 2005. Investigation of water flow in a bioreactor
landfill using geoelectrical imaging techniques. In: Tenth International Waste
Management and landfill Symposium, Calgliari, Italy.
Santos, F.A.M., Mateus, A., Figueiras, J., Goncalves, M.A., 2006. Mapping
groundwater contamination around a landfill facility using the VLF-EM
method – a case study. Journal of Applied Geophysics 60 (2), 115–125.
Seaton, W.J., Burbey, T.J., 2002. Evaluation of two-dimensional resistivity methods
in a fractured crystalline-rock terrane. Journal of Applied Geophysics 51(1), PII
S0926-9851(02)00212-4.
Singha, K., Gorelick, S.M., 2005. Saline tracer visualized with three-dimensional
electrical resistivity tomography: field-scale spatial moment analysis. Water
Resources Research 41 (5).
Soupios, P., Papadopoulos, N., Papadopoulos, I., Kouli, M., Vallianatos, F., Sarris, A.,
Manios, T., 2007. Application of integrated methods in mapping waste disposal
areas. Environmental Geology 53 (3), 661–675.
Telford, W.M., Geldart, L.P., Sheriff, R.E., 1991. Applied Geophysics, second ed.,
Cambridge University Press, 790p.
Warith, M., 2002. Bioreactor landfills: experimental and field results. Waste
Management 22 (1), 7–17.
Yang, J., 2005. Geo-electrical responses associated with hydrothermal fluid
circulation in oceanic crust: feasibility of magnetometric and electrical
resistivity methods in mapping off-axis convection cells. Exploration
Geophysics 36 (3), 281–286.
Yoon, G.L., Park, J.B., 2001. Sensitivity of leachate and fine contents on electrical
resistivity variations of sandy soils. Journal of Hazardous Materials 84 (2–3),
147–161.
Zume, J.T., Tarhule, A., Christenson, S., 2006. Subsurface imaging of an abandoned
solid waste landfill site in Norman, Oklahoma. Ground Water Monitoring and
Remediation 26 (2), 62–69.

Author`s personal copy - Laboratoire d`étude des Transferts en

Transcription

Similar documents

Mémoire de maîtrise

draft programme - Suez environnement

manuel installation - Boutique Aqwa Expert

DARCOM-741 BLOC SANITAIRE

menu novembre décembre

TL993SE

TP d`électrotechnique : onduleur triphasé 24 volts

installatiOn instruCtiOns guide d`installatiOn guía para

M. Dominique Genty 47 years old Orsay, France dominique.genty

DARCOM-720 BLOC SANITAIRE