Landslide Hazard - CERM
Transcription
Landslide Hazard - CERM
POLITECNICO DI MILANO CIVIL ENGINEERING FOR RISK MITIGATION Valmalenco Geological Report Course: Emergency Plans for Hydrogeological Risk INTRIERI, Sarah - matr. 765913 LEGNANI, Lucia - matr. 764559 MEMISOGLU, Gokay Caglar - matr.763120 INDEX Introduction ..............................................................................................................................................................3 1. Hazard Scenarios in Valmalenco...........................................................................................................................4 1.1 From Rain to Sediment Transport ..................................................................................................................5 1.2 Spriana Landslide ............................................................................................................................................6 2. Widespread Slope Erosion ....................................................................................................................................8 2.1 Values of Parameters Characterizing The Widespread Erosion .....................................................................9 2.1.1 Vegetation Cover Factor [C] and [Σ] ..................................................................................................... 11 2.1.2 Composition of Superficial Soil ............................................................................................................. 11 2.1.3 Coefficient of Soil Resistance................................................................................................................ 12 2.1.4 Coefficient Type and Extent of Erosion ................................................................................................ 12 2.1.5 Temperature – Mean Temperature [t] ................................................................................................. 12 2.1.6 Precipitation – yearly rainfall [H] .......................................................................................................... 13 2.1.7 Precipitation - rainfall depth for duration of 6 hours and return period 2 years ................................. 13 2.1.8 Water volume of rain for severe event of 100 years return period ..................................................... 14 2.1.9 Peak flow for severe event of 100 years return period........................................................................ 14 2.2 Models Used to Study the Widespread Erosion .......................................................................................... 15 2.2.1 USLE Method ........................................................................................................................................ 15 2.2.2 RUSLE Method ...................................................................................................................................... 16 2.2.3 Gavrilovic Method ................................................................................................................................ 17 2.2.4 MUSLE Method ..................................................................................................................................... 17 2.3 Results of the Model for One Single Intense Event - MUSLE ...................................................................... 18 2.4 Estimation of Material Mobilized During the Event of July 1987 ................................................................ 18 2.5 Estimation of Material Mobilized in One Year ............................................................................................ 22 2.6 Observations about Results and Uncertainties ........................................................................................... 23 2.7 Definition of Two Possible Scenarios........................................................................................................... 26 2.7.1 Normal Rainfall ..................................................................................................................................... 26 2.7.2 Intense Event (Tr=100years and Duration 24hr) .................................................................................. 27 2.7.3 Summary of the two scenarios ............................................................................................................. 29 3. Spriana Landslide ............................................................................................................................................... 30 3.1 History of Landslide ..................................................................................................................................... 30 3.2 Monitoring system ...................................................................................................................................... 30 1 3.2.1 Real-time Warning Instrumentation .................................................................................................... 32 3.2.2 Understanding the Phenomena ........................................................................................................... 34 3.3 Geometry of the Landslide .......................................................................................................................... 35 3.3.1 3D Model .............................................................................................................................................. 35 3.4 Lithology of the Landslide............................................................................................................................ 37 3.5 Behavior of the Landslide ............................................................................................................................ 38 3.6 Modelling of Spriana Landslide ................................................................................................................... 40 3.6.1 Physical Models .................................................................................................................................... 40 3.6.2 Geomechanical Model .......................................................................................................................... 40 3.6.3 Centrifugate Model .............................................................................................................................. 42 3.6.4 Mathematical Models........................................................................................................................... 44 3.6.5 Models to Study the Distribution of Fallen Material ............................................................................ 45 3.7 Definition of Scenario .................................................................................................................................. 46 4. Early Warning System ........................................................................................................................................ 49 4.1 Definitions and Functions ............................................................................................................................ 49 4.2 Monitoring and Forecasting System ............................................................................................................ 51 4.3 Definition of the Levels of Alert................................................................................................................... 53 4.3.1 Monitoring Parameters for Spriana landslide ...................................................................................... 55 4.3.2 Definition of the Levels of Alert ............................................................................................................ 58 4.4 Integration of the EWS for widespread events ........................................................................................... 59 4.5 Case study: Spriana Landslide ..................................................................................................................... 66 4.5.1 Monitoring Parameters ........................................................................................................................ 66 4.5.2 Definition of the Level of Alert ............................................................................................................. 71 Bibliografy .............................................................................................................................................................. 72 2 Introduction The present report analyzes the main hydro-geological hazard scenarios in Valmalenco. These hazards directly or indirectly threaten the city of Sondrio and this analysis has been made with the final aim of giving the technical and theoretical basis for the implementation of an emergency plan for the city. Valmalenco is a lateral valley of Valtellina, in northern Italy, which presents a large variety of slides and instability phenomena currently ongoing. The river that flows in Valmalenco is the Mallero creek; during its path it is threatened by various active landslides that may supply a very large amount of debris to the river itself and therefore to the confluence with the much larger Adda river, where the city of Sondrio is situated. The most dangerous landslide threatening the city is Spriana landslide, the most important rock avalanche in Valmalenco. Moreover, the whole valley is interested by diffused erosion and accelerated sediment yield. According to these features of the valley, two analyses have been performed: the computation of sediments due to widespread erosion and the investigation of the possible reactivation of Spriana landslide; in the next chapter the two cases of study will be explained with more details. 3 1. Hazard Scenarios in Valmalenco For the present analysis, two main probable hazard scenarios for the city of Sondrio have been considered. The first one involves the widespread erosion phenomenon in the whole valley which can be translated in a huge amount of sediment transported to the city by the Mallero creek, causing possible flooding depending on the severity of the hazard. In particular two events have been considered: erosion due to a single intense rain event and annual erosion during an year with not exceptional precipitation. The second one involves the activation of Spriana landslide, the complete collapse of the slope with the consequent formation of an earth dam in the valley and, therefore, of a lake in the upstream part that will over-top the dam causing a severe flood wave with huge sediment transport to threaten the city of Sondrio. In the next paragraphs, the two scenarios will be analyzed in detail on the geological side. 4 1.1 From Rain to Sediment Transport In this report the geological problem related to the production of debris and sediment in Valmalenco will be analyzed; therefore the attention will focus on the red first part of the following scheme (Figure 1), which describes the complete event scenario in case of water and sediment flooding in the city of Sondrio. In this particular case the widespread phenomenon of slope erosion will be taken into account considering different methods and hypotheses for the computation of the total sediment volume. The input parameters needed for the calculations are: The morphology of the area (mean slope, characteristics of the hydrological basin, type of soil coverage, etc.); The rainfall data (event based or yearly based depending on the purposes of the analysis); The already existing mitigation and prevention measures. The final result of the analysis will be given in terms of volumes of sediment for different sub- basins in case of different events. 5 Figure 1: Scheme for Hydrogeologic Problem 1 1.2 Spriana Landslide Another aspect that will be analyzed in this report regards the possible collapse of Spriana landslide and the consequent formation of an earth dam in Valmalenco that may obstruct the whole 6 valley. The attention will be therefore focused on the first part of the following scheme (red box), which describes the complete event scenario in case of Spriana landslide collapse due to heavy rainfall and the consequent water and sediment flooding in the city of Sondrio. Figure 2: Scheme for Hydrogeologic Problem 2 In this particular case, the possible occurrence of the Spriana landslide collapse has been analyzed through different models: 7 Geometric model; Geomechanical model / Centrifuge model; Empirical model. The input data required by the models can be briefly summarized as: Geometry of the landslide (including a 3D model of the landslide body); Stratification layers and lithology; Triggering factors (that may be the loss of cohesion due to heavy rainfall or the loss of stability due to an earthquake): in this particular case only rainfall has been considered as triggering factor since the probability of occurrence of a heavy rainfall event in this area is much larger than to have an earthquake phenomenon. The final result of the implementation will be given in terms of volume of the displaced material of the landslide and in terms of possible shapes of the fallen mass and, consequently, of the earth dam that may obstruct the valley. 2. Widespread Slope Erosion The phenomenon of widespread erosion is important in Valmalenco because it’s the source of many types of sediment that flow in the rivers decreasing the conveyance and increasing the possibility of flood in the valley. Two different kind of analysis has been performed to evaluate the sediment volume due both to the annual erosion for normal precipitation and to the erosion for one single intense event. The first analysis has been done using three different methods: USLE, RUSLE and Gavrilovic. The analysis of the intense event, instead, has been performed using the MUSLE method. Finally an effort has been done in order to try to evaluate which kind of minor rainfall event can cause a smaller sediments’ volume that can be considered as a critical threshold value. In the following paragraphs first the values of the parameters characterizing the widespread erosion are reported and then the four methods and the results will be explained. A comparison will be also done between the results of MUSLE method and the ones of a survey made to estimate the mobilized volume of sediments during the big event of July 1987. This comparison is useful to perform 8 a back-analysis and check the validity of the implemented MUSLE method. 2.1 Values of Parameters Characterizing The Widespread Erosion In order to study the phenomena of the widespread erosion it’s necessary to know the values of several parameters concerning the precipitations, the morphology and the pedology of the basin. In this paragraph the values of these parameters are reported and in case of lack of data the assumptions that have been done are explained. The data used to describe the morphology and the pedology of the basin are the same both in the analysis of annual erosion and single event erosion, but the data about rainfall and liquid discharge are different. In the following the data used for all the four methods are described. The parameters that refers to the same features of the basin are listed together even if utilized in different models, however when a data is used in some specific models only, and not in others , it’s specified. The basin of Mallero can be subdivided in 3 sub-basins homogeneous from a hydrologic point of view (Figure 3): - A1: from the sources to the inlet of the Lanterna river (A=91 km2) - A2: the basin of the Lanterna river (A=115 km2) - A3: from of the inlet of Lanterna river to the outlet of the Mallero river in the Adda river (A=123km2) Figure 3: Sub-basins The erosion models will be apply separately for each sub-basin, therefore values of the parameters, when available, are reported divided for each sub-basin. Table 1: Length of river reaches [L] Sub-basin Main reach (m) Secondary reaches (m) A1 50.916 232.445 A2 67.054 172.714 A3 68.039 251.654 9 Tot (m) 283.361 239.769 319.694 The length of the rivers’ reaches is given for each sub-basin and it’s subdivided in main reach’s contribution and secondary reaches’ contribution. Table 2: Mean slope [S] Sub-basin Mean Slope (degree) Mean slope (%) 30,07 A1 63 26,51 A2 55 30,78 A3 63 The following map (Figure 4) represents the slope of the basin expressed in percentage. Figure 4: Slope of the basin expressed in percentage. Table 3: Perimeter [0] and mean height [D] Sub-basin Perimeter [m] Height [m] A1 42.560 2.321 A2 46.887 2.349 A3 59.063 1.814 10 2.1.1 Vegetation Cover Factor [C] and [Σ] The vegetation cover factor takes into account the kind of vegetation on the terrain and it represents the protection of vegetation that prevents rain drop erosion. Values of C can be evaluated using tables. The values of these coefficients have been chosen considering the different features of each sub-basin (Table 4). Table 4: Vegetation cover factor [C ] and [Σ] Sub-basin C Σ A1 0.35 0.68 A2 0.078 0.50 A3 0.050 0.40 A high value has been assigned to the sub-basin A1 because in that region the percentage of bare soil area is really high, especially in the valley of Sissone River. In the sub-basin A2 the surface is subdivided in two portions: one composed just by bare soil and another one covered by forest, however some pasture and riparian area with shrubs and grass are also present. In basin A3 the portion of bare soil area is smaller than in the two other basins and the percentage of surfaced covered by forest is bigger, few agricultural fields (3% of surface) are also present. This subdivision of the area has been done both according to the document “Piano Programma di Ricostruzione, Riconversione e Sviluppo della Valtellina – prog ASP/4430” and to the observation of some satellite photos. The two coefficients C and Σ refer to the same features of the basin but C is used in USLE, MUSLE and RUSLE methods, Σ, instead, is used in Gavrilovic method. 2.1.2 Composition of Superficial Soil No specific data about the composition of the shallow strata of soil in Valmalenco are available. Thus, it has been decided to use some values typical of the area and of glacial valleys. However it is suggested by this report that; to make some analyses and to build pedological map of the area in order to have more precise data and to be able to make more reliable analysis. Table 5: Composition of superficial soil Debris f% Silt f% Clay Lithology 8 24 f% Sand 68 msilt 0.03 mclay 0.002 msand 0.07 fi : % of soil in the i-th class i =arithmetic mean of soil particles of i-th class in mm Looking at these properties (Table 5) it’s possible to compute K, the soil erodibility factor that 11 represents the propensity of a soil to be eroded. K is used in USLE; MUSLE and RUSLE methods. 2.1.3 Coefficient of Soil Resistance According to the composition of soil previously considered, it has been chosen a value of Π corresponding to sediments, clay and other rock with little erosion resistance (Table 6). Since in the sub-basin A1 the area characterized by rocks with poor erosion resistance is bigger than in the other basins it has been decided to use a slightly higher coefficient. Sub-basin A1 1,8 A2 1,7 A3 1,7 Table 6: Coefficient of soil resistance Π is used in Gavrilovic method. 2.1.4 Coefficient Type and Extent of Erosion For all the three areas it has been chosen to assign values of Φ corresponding to basins characterized by 50-80% surface affected by erosion and landslides (Table 7). In particular slightly higher values have been assigned to the sub-basins A1 and A2 because in the areas at higher altitude a bigger portion of the surface is interested by soil movements. Sub-basin A1 0,8 A2 0,8 A3 0,7 Table 7: Φ values Φ is used in Gavrilovic method. 2.1.5 Temperature – Mean Temperature [t] The value of the mean temperature over an year couldn’t have been assessed using the values recorded at the stations located in the basin because the ARPA cannot provide the data. Subsequently reasonable values have been assigned according to the value of the mean height of each sub-basin (Table 8). Sub-basin A1 A2 A3 0 0 3 Table 8: Mean Temperature The temperature is used in Gavrilovic method only. 12 2.1.6 Precipitation – yearly rainfall [H] The values of yearly rainfall depth have been computed using the values recorded at different rain gauges stations (one located in Chiesa Valmalenco and two located in Sondrio) in the last years. The data have been taken from the website of ARPA Lombardia. An average of the three values is taken ( H=958 mm) H is used in Gavrilovic method. 2.1.7 Precipitation - rainfall depth for duration of 6 hours and return period 2 years The values of rainfall depth for precipitation with a duration of 6 hours and return period of 2 years have been extrapolated from depth-duration-frequency curves. These graphs has been built doing a statistical analysis on the rain depths measured at some sections in the basin. Each curve links the rain depth with the duration of a rainfall event and the return period (the time interval during which the given value of rain depth is reached one time). The data available refer to three rain gauges sited at Sondrio , Lanzada and Campo Moro. The rain gauges used to assess these data are different than the one used for assessing the depth of yearly rainfall because for the two groups of rain gauges different data are available. These data have been taken from the document “Piano Programma di Ricostruzione, Riconversione e Sviluppo della Valtellina – prog ASP/4430”. P [mm] Sub-basin of location of rain gauge A3 Sondrio 32,2 A2 Lanzada 29 A2 Campo Moro 28 Table 9 Since the precipitations slightly decrease from South to North it has been decided to use the value recorded in Sondrio for the sub-basin A3 and an average value between the one of Lanzada and the one of Campo Moro (28,5 mm) for the sub-basins A1 and A2. This value of rain depth has to be used to compute the coefficient R, erosive factor, in USLE and RUSLE methods. R=27,38 P 2,17 (To use this formula with SI units it’s necessary to enter the value of rain depth in inches and then to multiply the obtained value of R by 17,02) 13 2.1.8 Water volume of rain for severe event of 100 years return period The water volume of rain for severe event can be computed multiplying the value of rain depth on a specific sub-basin by the value of the area of the sub-basin. In the following table the values of rain depth for a severe event of 100 years return period are recorded, for each sub-basin different values of rain depth for different duration of rain event are reported. These values have been taken from the document “Studi propedeutici alla valutazione dell'impatto ambientale degli interventi del bacino del torrente Mallero” Duration (h) 1 3 6 12 24 A1 26 42 57 78 105 A2 26 42 57 78 105 A3 27 44 59 81 109 Table 10 These values are useful to perform the analysis of intense event with MUSLE method. It has been assumed a severe event of duration 24h to simulate a rainfall similar to the one that occurred in July 1987. 2.1.9 Peak flow for severe event of 100 years return period The values of peak flow for a sever event with return period of 100 at the outlet of the subbasin are reported in the Table 11. These values are necessary to implement the MUSLE model. Qc (m3/s) Qc (m3/s) A1 outlet 218 230 A2 outlet 287 281 A3 outlet 596 579 Table 11: Peak flow Since there aren’t discharge gauges along the river it’s not possible to have a direct measurement of these values, therefore this values have been estimated from the values of precipitation using a Nash model. The two values refer to slight different formulations of the model. The values are taken from the document “Studi propedeutici alla valutazione del’impatto ambientale degli interventi nel bacino del torrente Mallero”. The values reported in the table are close to the values of Qpeak computed for the strong rain 14 event of 1987 using both a corrivation model and Giandotti formula. The application of this two methods and the detail of the results are reported in the document “Progetto esecutivo - Idrologia e trasporto solido” of the year 1993. 2.2 Models Used to Study the Widespread Erosion In the following the different methods used to study the slope erosion phenomena are briefly described. 2.2.1 USLE Method The USLE method computes the soil loss per unitary area of basin in one year [A]. The formula used is simple: A = R K L S C P Where: A = soil loss per unitary area per year [t/ha y] R = erosive factor (computed as explained in par.2.1) K = soil erodibility factor L = parameter concerning the length of all the reaches in the basin S = slope parameter C = vegetation cover factor (evaluated as explained in par.2.1) P = conservation techniques factor The soil erodibility factor K is computed as: Where: fi : % of soil in the i-th class mi =arithmetic mean of soil particles of i-th class L is computed as: 15 Where: m empirical coefficient = 0,5 when slope is greater than 5% for small basin: A is the basin area. Lch is the sum of the length of all the rivers of the basin. S is computed as Where: θ is the slope in radians. The P factor considers the impact of erosion defeating systems. If no actions have been done a value of 1 is used; in other cases this value should be updated according to interventions done. 2.2.2 RUSLE Method RUSLE uses the same parameters of USLE with an unique difference: the two parameters L and S are combined in one parameter called LS. A = R K LS C P Where For slopes where tan θ > 0.09 is possible to apply this formula: is an empirical coefficient and comes from: Where f is calculated from: 16 2.2.3 Gavrilovic Method This method is based on the idea of treating separately the production of sediment and the routing of sediment. In the following the equations are described: G mean annual sediment crossing closing section: W sediment production due to erosion: Where: R routing coefficient: T temperature coefficient: Z erosion coefficient: H: mean yearly rainfall [mm/y] A: basin area [km2] O: basin perimeter [km] D: mean height of basin [km] l: length of the main river reach li: length of minor reaches t: mean yearly temperature I: mean slope of the basin Σ: soil cover coefficient Π: soil resistance coefficient Φ: kind and extension of erosion coefficient 2.2.4 MUSLE Method USLE and RUSLE only allow to estimate the mean sediment yield over a whole year, however a modification can be done on these two methods to consider the eroded volume for one single event. The modified method is defined MUSLE (Modified USLE) and it computes , the sediment yield for one single event, as: Where Rd is defined as; Where: V: water volume [m3] computed as where H= rain depth A= area of the basin c = 0.8 afflux coefficient Qp: peak flow at basin outlet [m3/s] 17 2.3 Results of the Model for One Single Intense Event - MUSLE MUSLE method gives the estimation of sediment yield for a single intense event of return period 100 years (Table 12). The magnitude of this considered event is comparable with the one of the strong rainfall of July 1987. Sub-basin A1 A2 A3 Total Valmalenco Volume MUSLE [m3/event] 1.282.257 350.291 308.663 1.941.211 Table 12: Results of MUSLE method - sediment yield 2.4 Estimation of Material Mobilized During the Event of July 1987 In this paragraph it will be introduced a survey done in the 1993 in order to estimate the volume of material mobilized during the event of 1987. ( ISMES(1993) “Progetto esecutivo - Idrologia e trasporto solido” ) The estimation has been done on the basis of geomorphologic evidences observed during the investigations of January-September 1989. In the survey the assessment of total mobilized volume considers both the contributions due to river bank erosion and to the slope. The transport of solid discharge along the rivers has been also studied to put in evidence the main areas of erosion and deposit along the river reaches. It has been decided to focus the attention only on the estimation of the volume due to slope erosion in order to compare it with the volume estimated with MUSLE method for an intense event of return period 100 years. This comparison can be relevant since, as previously observed, the magnitude of the event considered for MUSLE method is comparable with the one of the strong rainfall of July 1987. Along some sections of the river the survey considers as slope contribution not only the slope erosion but also the contribution of some small concentrated and shallow landslides, for this reason in some areas where the contribution due to the concentrated landslides is quite big the two contributions are analyzed separately. In particular in Table 13 the only contribution of slope erosion is reported. The survey gives the results divided for sub-basins; however the division is different than the one done applying the MUSLE method so some arrangements have to be done in order to compare the two results. 18 River reaches Mallero: S.Giuseppe - inlet of Sissone Sissone Ventina Entovasco Nevasco Mallero: Chiesa Valmalenco – S.Giuseppe (*) Furla Curlo Lanterna Mallero: Ponte di Spriana – Chiesa Valmalenco (*) Sora Suello Secchione Daglia (*) Torreggio (*) Marveggio Venduletto Mallero: Sondrio – Ponte di Spriana (*) Valdone (*) Antognasco (*) Bedoglio Volume slope erosion [m3] 108.000 932.000 46.000 5.000 5.000 8.800 2.000 66.000 Tot for sub-basin A1 248.400 Tot for sub-basin A2 203.820 1.500 4.000 4.500 16.500 500 1.500 2.500 3.100 8.000 Tot for sub-basin A3 1.172.800 m3 248.400 m3 245.920 m3 Table 13: Results of the survey: contribution of slope erosion In the rows where there’s the symbol (*) it means that for that reach it has been considered the only contribution due to slope erosion, but the survey put in evidence that for the given reach there’s also a contribution due to concentrated and shallow landslides. This contribution will be considered after because it’s not relevant now for the comparison with the result of MUSLE method. Looking at the results it’s evident that the reaches that give a bigger contribution to the accumulation of eroded material are Sissone and Lanterna. The big contribution of the Lanterna river can be explained by the fact that its basin is quite big, the big contribution given by Sissone instead is due to the high erodibility of the higher part of sub-basin A1. However the Sissone sub-basin is quite small in comparison to the whole basin and, even considering an high erodibility factor, the result of the survey looks unexpected, thus it’s suggested to make additional analyses on this sub-basin. . 19 Figure 5: The results of the survey have been compared with the one computed with MUSLE method (Table 14). Sub-basin A1 A2 A3 Sum Volume slope erosion [m3] Survey MUSLE 1.172.800 1.282.257 248.400 350.291 245.920 308.663 1.667.120 1.941.211 Table 14: Comparision of the results of Survey and MUSLE Method The comparison reported in the Table 14 give satisfactory results since the contributions of the tree sub-basins computed by MUSLE method are close to the results estimated by the survey. The differences can be due to the difficulty in assigning the exact values to some coefficients as the erodibility factor and the surface coverture factor. River reaches Mallero: S.Giuseppe - inlet of Sissone Sissone Ventina Entovasco Nevasco Mallero: Chiesa Valmalenco – S.Giuseppe (slope erosion) Mallero: Chiesa Valmalenco – Volume slope contribution [m3] 108.000 932.000 46.000 5.000 5.000 88.000 27.800 20 Volume bank erosion [m3] Deposit along the reach [m3] 498.000 530.000 42.000 141.800 93.000 Balance S.Giuseppe (conc./shal. landslide) Furla Curlo Tot for sub-basin A1 Lanterna Tot for sub-basin A2 Mallero: Ponte di Spriana – Chiesa Valmalenco (slope erosion) Mallero: Ponte di Spriana – Chiesa Valmalenco (conc./shal. landslide) Sora Suello Secchione Daglia (slope erosion) Daglia (conc./shal. landslide) Torreggio (slope erosion) Torreggio (concentrated) Marveggio Venduletto Mallero: Sondrio – Ponte di Spriana (slope erosion) Mallero: Sondrio – Ponte di Spriana (conc./shal. landslide) Valdone (slope erosion) Valdone (conc./shal. landslide) Antognasco (slope erosion) Antognasco (conc./shal. landslide) Bedoglio Tot for sub-basin A3 2.000 66.000 +1.279.800 248.400 +248.400 203.820 +141.800 -1.163.000 207.400 -207.400 637.000 =258.600 =41.000 126.780 1.500 4.000 4.500 2.100 31.300 16.500 624.250 500 1.500 165.000 606.000 27.000 2.500 82.000 3.100 25.300 8.000 +1.162.500 32.000 +165.000 -1.277.100 =50.400 Tot at Spriana 350.000 Table 15: Results of the survey To conclude it can be interesting to report the complete results of the survey done about the event of July 1987. Table 15 specifies for each sub-basin, not only the contribution of the slope erosion to the solid discharge, but also the estimation of: - Volume eroded from the rivers’ banks (only along Mallero river) - Volume of sediments due to a concentrate landslide in Torreggio valley - Volume of sediments due to some little concentrated shallow landslides - Volume of sediments deposited along the reaches. % of tot sediments produced that arrives at the outlet 258.600 18,26 41.000 16,50 350.000 11,67 tot sediment produced [m3] sediment at output [m3] A1 A2 A1+A2+A3 1.421.600 248.400 2.997.500 Table 16: Sediments that arrives at the outlet of each sub-basin 21 It can be noticed that the survey put in evidence that not all the eroded material flows along the river and reaches the city of Sondrio, in fact a big part of it deposits along the reach and only a smaller volume (350000 m3) reaches the Adda River. Thus it can be considered that only the 11,7% of the total eroded material reaches the final section of the Mallero basin. The detail of the % of the total produced sediments that arrives at the outlet of each sub-basin is reported in Table 16. 2.5 Estimation of Material Mobilized in One Year The results of USLE, RUSLE and Gravilovic are summarized in the following tables. Table 17 gives the result in m3/year, Table 18 instead divide the contribution in 365 days per year and give the result in m3/s. This discharge has been computed only to give an idea of the order of magnitude; however it must taken in account that the precipitations are not equally distributed along all the months of the year so the given value cannot be considered an exact estimation. For USLE and RULE it has been used the same value of C utilized in the MUSLE method in the previous paragraph. .Sub-basin Produced in A1 Produced in A2 Produced in A3 Tot Valmalenco Volume USLE [m3/year] Volume RUSLE [m3/year] Volume Gavrilovic [m3/year] 2.326.806 1.572.067 117.343 648.469 514.284 45.893 676.607 467.928 99.675 3.651.884 2.554.281 262.912 Table 17: Results in m3/year Sub-basin Outlet A1 Outlet A2 Sondrio Qsolid USLE [m3/s 0,073 0,020 0,116 Qsolid RUSLE [m3/s] 0,049 0,016 0,081 Qsolid Gavrilovic [m3/s] 0,0037 0,0014 0,0083 Table 18: Results in m3/s The results obtained with USLE and RUSLE methods are really bigger than the ones obtained with Gavrilovic method. The Gravilovic results can be considered more reliable because they are close to other values estimated in literature (Table 19). In particular in the past the values of annual sediment yield for normal rain event has been estimated both using empirical formula and mathematical models and the results are reported in the document “Progetto esecutivo-Idrologia e trasporto solido” 22 . Computed volume at Sondrio section[m3/year] 262.912 Gavrilovic 206.446 Empirical formula (Prof.Datei) Mathematical model (Ing.Peviani) 170.000-270.000 Table 19: Comparison of Results In addition the result that has been found is coherent with the one estimated for the ValTartano. In fact in Valtartano using Gravilovic methods it can be estimated a sediment yield of 53000 m3/year, considering that Valmalenco is six times bigger than ValTartano it’s reasonable to expect a volume of sediment yield equal to 6x53000 m3/year 2.6 Observations about Results and Uncertainties Some observations about the degree of uncertainties of the results reported in this chapter need to be done. First of all it has to be notice that the USLE, RUSLE and MUSLE methods has been originally created to solve the problem of erosion on agricultural fields, they have been then adapted to mountain catchments but errors in results can be expected. Then uncertainties in the results can also arise from the difficulty in assessing the values of some parameters. In fact the values of the morphologic features as the length of the rivers, the altitude of the basins and the inclination of the slopes can be easily derived from DTMs and maps, but other coefficients that refers to the vegetation, the composition and the properties of the soil or to the mitigation measures implemented in the basin are more difficult to be assessed. In the following it will be briefly explained in which way the uncertainties on the parameters influence the final results. In USLE , MUSLE and RUSLE methods the coefficients more difficult to be assessed are: the composition of the superficial soil, due to uncertainty about the pedology of the basin; the coverture factor, due to not precise data about the percentage of area covered by different type of vegetation and to the fact that in literature there’re lots of different tables with different coefficients; the precise value of rain depth due to the availability of data only for few rain gauges over the whole basin. 23 The uncertainty about the composition of the superficial soil doesn’t affect strongly the results since a variation of 20% of the material present in each class makes the K coefficient vary only in a small range between values of 0,037-0,043. This variation doesn’t affect the final result in a significant way since the order of magnitude and the first digit of the total volume of sediments don’t change. K= 0,037 K= 0,043 USLE [m3/y] 3.108.539 3.651.884 RUSLE [m3/y] 2.174243 2554.281 MUSLE [m3/event] 1.482.433 1.741.549 For what concern the vegetation cover factor C, it can be shown that changing a little the assumptions about the type of vegetation covering the different parts of the basin or choosing the values from different tables available in literature the final value of C can change a lot. The change in the value of C affects directly the final result since in these methods the total volume of sediments is given by the product of C with the other coefficients. As example it’s shown how the results can vary assuming a reduction of surface of bare soil (-15%) and an increase of surface covered by forest (+15%). No changes -15% bare soil & +15% forest USLE [m3/y] 3.651.884 1.401.091 RUSLE [m3/y] 2.554.281 991.456 MUSLE [m3/event] 1.941.211 673.876 In USLE and RUSLE methods an error on the value of rain depth (for a precipitation of duration 6hours and return period 2 years) can make the results vary a lot, in fact a variation of the depth of only 5 mm makes the R coefficient varying of 1/3 and since the final value of sediment is given by the multiplication of R with the other coefficients also the final results is affected by the same variability of the R coefficient. In MUSLE method instead the same variation of 5mm in the input rain depth ( depth for a single rain event of 24 hours) doesn’t affect the final result. Rain depth used depht lowered of 5 mm USLE [m3/y] 3.651.884 2.438.075 RUSLE [m3/y] 2.554.281 1.697.200 MUSLE [m3/event] 1.941.211 1.694.610 24 In Gavrilovic method, for the same reasons mentioned for the other methods, the coefficients more difficult to be assessed are: the coefficient of soil resistance and the expected extension of erosion , due to uncertainty about the composition of superficial soil of the basin; the vegetation cover factor; the precise value of yearly rain depth However, since this method differs from the previous ones because of the formulae and the tables from which the coefficients are derived, the influence of the variability of the coefficients on the final result can be different. In Gavrilovic method the value of soil resistance Π can be assessed looking at a table that define it according to the composition of the soil. For the basin under investigation a value in the range of 1,7-2 could have been chosen , in the table it’s shown the variability of the final result according to the different values of Π. Π= 1,7 Π= 1,8 Π= 1,9 Π= 2 Gavrilovic [m3/y] 253.271 275.944 299 256 323.189 The expected extension of erosion is difficult to be precisely defined, however for the studied basin it can been roughly estimated that about50- 80 % of the total basin is affected by erosion. For this range of percentages the value of Φ can be assumed equal to 0.7-0.9, the deriving variability of the volume of sediment is represented in the table: Φ = 0,7 Φ = 0,8 Φ = 0,9 Gavrilovic [m3/y] 246.986 273.526 300.954 To analyze the dependency of the final result on the variability of the vegetation cover coefficient the assumptions about the type of vegetation in each sub-basin have been changed. In the table it’s shown an example in which it has been reduced the surface of bare soil (-15%) and increased surface covered by forest (+15%). The same example has been previously done for the three other models, therefore it can be notice that for Gavrilovic method the final result changes but less respect the other methods. No changes -15% bare soil & +15% forest Gavrilovic [m3/y] 262.912 186.956 25 Finally the variation of rain depth strongly influences the volume of sediments. In fact the final result is computed multiplying the precipitation depth by the other coefficients so a variation in the depth directly causes the same variation in the volume of sediments. To conclude it must be mentioned that the models implemented don’t take in consideration the possibility of reactivation of both shallow and deep concentrated landslides. In particular in the Mallero basin significant concentrated landslides can be expected in Torreggio Valley, the reactivation of these landslides would produce a consistent increase of sediment yield in the Mallero River. For the case of intense rain event, it can be taken as reference value the estimated volume of sediments due to concentrated landslides for the event of 1987, reported in survey previously mentioned. 2.7 Definition of Two Possible Scenarios Considering the problem of widespread erosion, two possible scenario must be considered in the Mallero basin: one for a condition of normal rainfall and another for an intense event (Tr=100 years). 2.7.1 Normal Rainfall In this scenario the sediment volumes and the solid discharges that can be expected for a normal year, during which not exceptional rainfall occurs, are considered. The weather conditions that have been considered are: Average yearly temperature t=0◦C in sub-basin A1 and A2 and t=3◦C in sub-basin A3; Yearly rainfall depth h=958mm. Sub-basin Produced in A1 Produced in A2 Produced in A3 Tot produced in Valmalenco Volume [m3/year] 117.343 45.893 99.675 262.912 Qsolid [m3/s] 0,0037 Outlet A1 0,0014 Outlet A2 0,0083 Sondrio * considering Q constant during the year Table 20 The results given in the table refer to the total volume of sediments eroded from the slope that flows to the Mallero river every year. In order to know if this volume can affect in a relevant way the conveyance of the river and facilitate the occurrence of flood it should be necessary to know if and 26 where the sediments deposit along the reach of the Mallero river. For this purpose an analysis of the transport of solid discharge and of morphologic evolution of the basin should be performed. 2.7.2 Intense Event (Tr=100years and Duration 24hr) For one single severe rain event with return period 100 years and duration 24h, the cumulated values of rain in each sub-basin are shown in Table 21. H [mm] A1 105 A2 105 A3 109 Table 21 The values of peak flow for a sever event with return period of 100 at the outlet of the sub-basin are reported in Table 22. These values have been estimated from the values of precipitation using a Nash model. Qc (m3/s) A1 outlet 218 A2 outlet 287 A3 outlet 596 Table 22 Using the previous data it has been assessed that the sediment volumes and the solid discharges that can be expected for the single severe rain event are as reported in Table 23 . Sub-basin Produced in A1 Produced in A2 Produced in A3 Tot produced in Valmalenco Volume [m3/event] 1.282.257 350.291 308.663 1.941.211 Qsolid [m3/s] * 11,88 Outlet A1 4,09 Outlet A2 20,16 Sondrio * considering Q constant during the 24 hours Table 23: Sediment volumes and solid dischargesfor slope erosion As previously discussed, the values refer to total volume of sediments eroded from the slope without taking in consideration that part of this material deposits along the rivers. In addition it should be considered that also the erosion of the rivers’ banks and shallow and deep concentrated landslides give a contribution to the total solid discharge. Taking as reference the assessed values of sediments for the event of 1987, volumes of sediments are estimated in Table 24. 27 A1 A2 A3 shallow/deep concentrated landslide [m3] 27.800 bank contribution [m3] 141.800 916.630 165.000 Table 24: Volumes of sediments for concentrated landslides and bank erosion Using as reference the same survey, it can also be assessed the percentage of total eroded materials that arrive at the outlet of each sub-basin. Outlet A1 Outlet A2 Outlet A3 % of tot sediments produced in the upstream basin that arrives at the outlet section 18,26 16,50 11,67 Table 25: Percentage of total sediments produced in the upstream basin Finally it’s reported the estimation of the granulometric distribution of the sediments at the section of Spriana for the event of 1987; this information can be relevant to study the morphologic evolution of the river sections due to solid transport. Figure 6: Granulometric distribution at Spriana for the event of 1987 from “ISMES(1993) – Idrologia e trasporto solido” The potential risk given by this scenario is the deposition of a consistent volume of sediments along the Mallero reach and the consequent reduction of conveyance capacity that can increase the possibility of flooding at some sections and in the city of Sondrio as well. In order to study this phenomena it’s 28 necessary to perform and hydraulic analysis that takes in account a morphologic evolution of the river. It has been decided not to produce a third scenario that considers the combination of normal year rainfall and single intense event because the production of sediments during the intense event is really big and the contribution of normal year rainfall can be considered as negligible. 2.7.3 Summary of the two scenarios In the following table the volumes of sediments mobilized in the two scenarios previously described are summarized. Intense event Produced in A1 Produced in A2 Produced in A3 Produced in the whole basin Slope erosion [m3] 1.282.257 350.291 308.663 1.941.211 Yearly precipitation Bank erosion [m3] 141.800 Concentrated landslide [m3] 27.800 165.000 306.800 916.630 944.430 Table 26: description of the two scenarios 29 Slope erosion [m3] 117.343 45.893 99.675 262.912 3. Spriana Landslide 3.1 History of Landslide The history of the Spriana landslide has been characterized by periodic movements of varying intensity and duration, alternating with periods of quiescence. The dates of major historical events are: 1878: First movement. 1911: Significant movement due to heavy rainfall. 1916: Other movement made to stop construction of ENEL tunnel. 1919: Reconstruction of ENEL tunnel started again in the rock. 1927: Some displacement after the flood. 1960: Significant movement on the slope above the Erta and after above Cucchi. 1977–1978: Reactivation of the landslide occurred, the crown was visible. 1983: Final movement which had been seen. 3.2 Monitoring system The Spriana Landslide area is largely furnished with different instrumentation networks for the real-time and periodic control of the landslide movements and of the hydrological parameters. The system has been installed at the end of the 1989 and is basically constituted by three different types of network that give the possibility to measure the following parameters: Shallow movements (extensometric and topographic networks); Deep deformations (inclinometric network); Water table levels (piezometric network); Meteorological parameters (thermometers, barometers, rain and snow gauges). The instrumentation that has been set up can be basically divided into two categories, on the basis of the use that can be made out of the registered data: Networks for the real-time warning; Networks for the achievement of a better medium and long term understanding of the phenomena. The real-time warning instrumentation has been set up in order to be able to forecast the events 30 that may originate as a consequence of the evolution of instability phenomena in the landslide area and of the formation of natural flood waves. The nowadays configuration, typology, number and position of the monitoring instruments are the result of design analysis carried out in the very first phase of the approach to the landslide and they are based on the knowledge of the phenomenon that was available at that time. The acquisition of new information through Figure 7: Monitoring Network the interpretation of the most recent data and the production of further studies have highlighted a substantial deficiency in the available monitoring instrumentation. Another weak point that can be highlighted is the fact that, starting from 1993 until 1998, no normal maintenance program has been activated in the framework of the monitoring system management. This means that the measurement carried out cannot be considered completely reliable since there is no evidence of the correct functioning of the whole system. A maintenance and renewal program should be included in the emergency plan that might consider the possibility of the implementation of an early warning system for the city of Sondrio. 31 3.2.1 Real-time Warning Instrumentation 3.2.1.1 Automatic Extensometric Network The opening width of the principal fractures on the landslide is monitored through a network of automatic measurement reading control bases that are connected with the Local Operative Unit (USP) through radio stations. In the whole landslide area 12 electric transducers have been placed: 6 of them (E107 to E112) have been installed inside the exploring tunnel, forming a whole chain of measurements; the remaining 6 (E101 to E106) have been installed on the two principal crowns. The scan of all the measurement channels in normal conditions is carried out with a frequency of one reading every 30 minutes. 3.2.1.2 Manual Extensometric Network The opening width of the principal fractures on the landslide is also monitored through extensometric measurements that are manually carried out by in-situ operators. The measuring system consists in the determination of the variation of the distance between previously defined fixed points (couples of threaded pins anchored to extensometric bases). These measurements are carried out on a monthly basis in order to control the automatic extensometric network and, just in case of long unavailability of the automatic network; the number of the manual measurements can be increased. The precision of the measurements is of about 0.1 mm. 3.2.1.3 Piezometric Network The measurement of the water table level inside the landslide body is performed using three different piezometric sensors with automatic reading system: 2 open tube, 6 automatic Casagrande and 1 electric device. In particular, the stations have been placed in 6 boreholes made specifically for the registration and acquisition of the data regarding the variations of the water table level. Some of the piezometers installed in the late 1989 are now out of service and cannot be repaired for different reasons; in particular, some of them have been permanently deformed by the landslide movements. The functioning sensors are all cable-connected to the remote units (UR) which are able to ensure the acquisition and transmission of the data to the Local Operative Unit (USP). The scan of all the automatic measurement channels is carried out every 30 minutes. In case of long unavailability of the automatic acquisition network, manual readings can be performed. 32 The type and code of the different devices employed can be found in the Table 26. Open tube Casagrande Electric PZ101 PZ 102 PZ 107 PZ118 PZ 106 PZ 108 PZ 110 PZ 117 Table 27: The type and code of the different devices 3.2.1.4 Topographic Network The topographic network has been installed in order to control the movement of some significant points located on the landslide body. The measurement system is made of 28 stations: 4 stations are located on the slope opposite to the landslide (Master network) and 24 stations are distributed on the sliding slope (fixed points). Measurements are performed manually starting from the 4 Master stations using a teodolite or other traditional techniques or, more rarely, with GPS (global positioning system) measuring methods. These last measurements involve just 3 fixed points of the topographic network (C8, C12, C22), that are representative of three different altimetric positions on the monitored slope, and 2 other fixed points placed outside the landslide area. The readings give the possibility to obtain the values in terms of N and E coordinates for every fixed point. The comparison between two different measurements gives the possibility to individuate the reactivation of controlled movements and the presence of new instability phenomena on the slope. 33 3.2.1.5 Hydro-Meteorological Network The control system for the hydro-meteorological conditions requires the acquisition and the transmission to the Local Operative Unit (USP) of the data registered by the automatic stations distributed all across the Mallero hydrological basin. In particular, rain gauges are especially relevant for the control of the hydro-meteorological parameters and of the hydrological mode. The devices employed in the system are: 12 rain gauges for the rainfall depth measurement; 3 hydrometers for the automatic measurement of the water depth in the Mallero creek (in Curlo, Ganda di Lanzada and Torre Santa Maria); 10 thermometers for the automatic measurement of the air temperature; 7 ultrasonic snow gauges for the automatic measurement of the snow depth; 2 barometers for the automatic measurement of the atmospheric pressure; 1 wind gauge for the automatic measurement of the wind speed and direction; 13 remote units for the real-time collection and transmission of the hydro-meteorological data. The scan of all the automatic measurement channels is performed every 30 minutes in normal conditions. The data collected through this network represent the computational basis for the implementation of the PREVIS hydrological forecasting model, which simulates the discharges of the Mallero creek at the Eiffel bridge section in Sondrio. 3.2.2 Understanding the Phenomena 3.2.2.1 Inclinometric Network An automatic inclinometric network has been installed in order to measure the deep deformations in the landslide body. This network is constituted by sensors placed at standardized depths. In particular, 6 inclinometric tubes have been installed: I103, I109, I111, I113, I114, and I115. Furthermore, manual readings are scheduled on a yearly base; these manual readings require the extraction of the entire inclinometric column. The methods followed in order to obtain the final data and the time required to perform them make it impossible to implement a warning procedure using these instruments. 34 3.2.2.2 Piezometric Network Manual readings are performed on three different types of piezometers: open tube, Casagrande and 3 Westbay Multipoint (W119, W120, and W121) devices. Anyway, the reading frequency, the methods followed in order to obtain the final data and the time required to perform them make it impossible either to implement a warning procedure using these instruments or to use them in the geotechnical model. In normal conditions, the reading frequency is usually equal to one reading per month but, in case significant variations are registered in the pore water pressure values, it is possible to increase it within the automatic system. 3.3 Geometry of the Landslide The landslide is located at the left side (eastern part) of Mallero River in the Valmalenco Valley where it is bounded from northwest to Ballone di Badoglio and from southwest to Val Calchera. The foot of the landslide is estimated at 700 m ASML, since there both underground water sources and reliable bed rock on this section. Two visible crowns are observed at 1100 m and 1400 m; however, by the help of the analysis of morphological evidence and site investigations as well as collection of movement data, occurrence of a new crown at the level of 1700 m should be considered (Figure 8). Figure 8 : Boundary and visible crown of the Spriana Landslide 3.3.1 3D Model Hypothesis of a surface movement in rock is taking place on two different planes. The most likely scenario regarding the slide in rock is rock avalanche and the potential of having two separate surfaces of movement. There are two different surfaces of movements with different characteristics, which are divided the landslide vertically into two parts. Table 27 summarizes the characteristics of the two sides. 35 Left Part Right Part Glacier Fracture Fault Creep with slow velocity respect to right part Snow melt Right part movement Creep sliding with more velocity respect to left part Rainfall Snow melt Glacier History Geomorphological Evidence Geomorphology Velocity of the movement Triggering Factors Predisposing Factors Table 268: characteristics of the two sides The surface of rupture in the right part slides on the fault with a slip direction of 260°/35°-25°. It is known that the fault is quite dangerous since its inclination is lower than the slope and the direction is the same. Furthermore, the fault intercepts the slope at an altitude of 700 m which is the same level of the landslide foot. Some outcrop fractured materials were observed in Val Calchera which could indeed imply the existence of an outcropping fault. The drill hole S112, which is on the external right side of the landslide at an elevation of 1000 m, leads to the hypothesis that the drill hole has intercepted the fault. The reason behind this situation is; the existence of a medium or high Figure 9: Definition of the surface of movement for the Spriana landslide fractured rock in the first meters of the hole. All these evidences could be explicated by the presence of a fault with a slip direction of 260°/35°-25° that outcrops in the Val Calchera area and the gain depth towards the centre of the landslide body. The plane representing this fault is marked in Figure 9 with the red color. The left side rupture surface is less obvious due to lack of data on this section of landslide. Furthermore, it is not a clear rupture surface like the other one. Glacial origins of the valley explicate the weakness surface. The plane representing this fault is marked in Figure 9 with the yellow color. Two different ways of movements have characterized those two parts. Right part movement was characterized by the rainfall which is the most critical triggering factor for the movement where the sliding creep is expected with high velocity. On the other hand, the left part movement would be triggered by the right part movement. However, the slide speed of the left part would be considerably lower than the one observed for the right part because of the unclear rupture surface. 36 Regarding the speed and the acceleration of the slide, the role of the rock bridges that tight the stable rock with the unstable part should be considered. Some continuity is available in the rupture surface inside the rock, which can be named as bridge rock. As the slide begins, the bridge rocks starts to break up which increases the slide speed to a critical value. This situation leads the breakup of more rock bridges and then a faster slide which will continue by the help of a constantly increasing speed. Figure 10: The slide of the Spriana landslide Consequently, this cascade effects are the reason behind for the rock avalanches to be the quickest type of landslide. Figure 10 represents all those movements step by step. 3.4 Lithology of the Landslide Lithology is divided into three main parts as debris, fractured rock and hard bedrock; according to field investigation reports, drill holes results, geophysics part, exploration tunnel and monitoring system. The materials involved in the landslide mainly belong to the colluvial and moraine deposits. The debris and the rock are composed by mica schists of the Monte Canale gneiss formation. Although the depth of debris material varies from 20 m to 40 m with continuity up to the elevation of 1100 m, the presence of depth is lower starting from this elevation. Furthermore, some rock boulders are visible at the elevations between 1100 m and 1400 m. The foot of the slope has the maximum thickness of debris. Presence of blocks and shingle - as well as limited amount of gravel, sand and silt with different percentages according to depth- exist in the debris material. Pebbles, gravel and rare large blocks, immersed in full matrix of natural sandy-silt, and some amount clay are composed in the deeper part of this layer. The instability of debris increases since this part of debris layer has a high percentage of fine material. (PAPINI, MANNUCCI, LONGONI, CAIMI 2005). Composition of the second layer which is located under the debris layer is fractured and altered rock. This layer is characterized by a pronounced weakness shown by RQD results near zero. 37 Thickness of the layer varies between 60 m and 120 m. Finally, the third layer contains good quality rock with high RQD results. The results of monitoring tests about surface movements are coherent with the lithology. As regards to the deep movements, the inclinometer network confirms the presence of a slow and gradual process of deformation of slope up to a depth of 80 m. The altitudes range between 700 m and 1400 m. These movements are along the discontinuity between the layer of fractured and weathered rocks and the layer of good quality rocks. The measurements indicate there is not any movement below the altitude of 700 m. Figure 11: The result of the monioring tunnel Furthermore, it is feasible to use the monitoring tests to obtain information about the water level table. The level of the groundwater is estimated between an altitude of 900 m and 700 m where the piezometers are located. It varies between 3 m and 4 m. Groundwater can sustain at a depth of 100 m from the surface due to some infiltration in the upper part of the landslide slope. Finally, Figure 11 shows the results of monitoring tunnel and it can be taken as an example of the lithology of the slope. 3.5 Behavior of the Landslide There are two different hypothesis scenarios; debris flow and avalanche rock. The surface movement in debris flow is the contact with debris and the rock below. Although the debris material is present up to 1400 m, debris flow is expected only up to the altitude of 1100 m, since the material beyond this limit is not continuous and it is deposited in a thin layer. In addition, the slope beyond 1100 m is gentler than the one below. Performing drill holes, seismic refraction surveys and by the use of an 38 exploration tunnel it has been possible to assess that the surface of movement between the debris layer and rock layer is continuous. Therefore, the movement of material is constant through the whole depth of the debris layer. The predisposing causes of this case can be classified as; high gradient of the slope and the high percentage of fine material in the debris layer near the contact debris-rock. Reasons behind expecting a landslide are; the movement in the rock, the high degree of fracturing in the place and the foot of the landslide which does not reach the valley floor (HENDRON and PATTON 1989). The surface of movement is in the fractured rock with an average depth varying between 70-80 m from the ground surface. Moreover, it can reach 120 m at peak movements. However, it is not known if the surface of movement is continuous surface or not. Some peak displacements, which indicate the presence of a high fractured rock layer with more shift possibilities, can be found in the rock layer. The possibility of a surface of movement in the fractured rock layer is lead by the presence of movements in the weak fractured rock. Furthermore, two different intensities of movements have been recorded in both left and right parts of the landslide. Two different surfaces of movements in fractured rock can be assumed as a consequence of this fact. Figure 12: Spriana Landslide The main triggering factor in the first studies was the variation of the ground water level. 39 Increase of this level in the lower part of the landslide causes a change in the deformation field in the foot. As a consequence of comparing the results of the studies between 1977 and 2001, it is revealed that the landslide has slowed down. This is not justified by this model behavior. The authors Papini, Manucci, Longoni and Caimi responded the need to comprehend in a better way the rupture model of the landslide in 2005 by making the hypothesis of a surface of movement in rock taking place on two different planes. The planes divided the landslide vertically; right and left parts. The right part of the landslide slides on a fault while the left part slides on a glacier rupture surface. These two parts have different velocity of movements. 3.6 Modelling of Spriana Landslide The models used to analysis Spriana landslide have two different aims: check the stability of the slope computing FS and study, in case of slope failure, the size of the detached volume and its distribution. Taking in account a volume of soil on a slope, when the equilibrium between driving forces and resistance forces is reached the safety factor FS assumes value equal to 1 and the mass of soil flows downhill. According to geometry of the analyzed slope and of the valley and the geotechnical properties of the materials the mass can accelerate and lift up on the opposite flank of the valley until the situation of FS>1 is reached again. To reproduce this behavior by means of numerical models only leads to big approximations, that’s why physical models can be implemented to simulate the landslide movements. The physical models are used to calibrate mathematical models that can evaluate the value of FS. 3.6.1 Physical Models In the physical methods the landslide behavior is reproduced in scale, in particular in the analysis of Spriana landslide two physical models have been used: the geomechanical model and the centrifugal model. 3.6.2 Geomechanical Model The geomechanical model of Sprina landslide has been built in order to analyze the falling behavior of an unstable mass of soil, in particular the specific goals of the model are: a) Reproduce the fall of the entire mass of the landslide; b) Give an estimation of the heights of the cumulate volume of the fallen material in the valley; 40 c) Determine the lateral distribution of the fallen volume. It has been analyzed the behavior of a landslide with crown at altitude 1400m and foot at altitude 700m. The sliding surface is set at depth 50-70 m with inclination 35-37 degree. The area that contains the sliding movements has a length of 2000m and a width of 800m. Figure 13 41 Two models have been built in scale 1:250. A two dimensional model has been built to determine some parameters, such the angle of sliding and the characteristic of failure of the materials, in order to calibrate the three dimensional model. The three-dimensional model, then, analyses the global stability of the landslide and the accumulation of material at the foot of the slope in case of partial or global sliding. The models are built on a mechanism that can make varying the inclination of the slope, during the test the inclination can be gradually increased until the limit equilibrium condition is reached ( angle=35-37degree). The test is video recorded and the images are digitalized and memorized in a computer, then, during the elaboration phase, the coordinates of different points at different stages can be compared to evaluate the deformations and the velocities. The input data required are: - Topography of the area in scale 1:2000 - Shape of sliding surface - Physical and mechanical properties of the materials. The model has put in evidence that as soon as the value of FS becomes smaller than one the entire mass moves and this movement create cracks similar to the ones observed on site. Increasing more the inclination, bigger movements can be detected, in particular it can be observed that: - The material that deposits in the Mallero River tends to flow downhill thanks to the natural slope of the reach; - There’s a formation of a natural dam having a height of 80m in the area of Badoglio and about 40 m in the remaining parts. Due to the almost static conditions used to perform this kind of the test, the results refer to a partial sliding of the material (less than half of the volume that could potentially fall), if the falling velocity was greater the dimension of the dam would have been greater as well. 3.6.3 Centrifugate Model In order to overcome the limits of the geomechanical model that considers almost static condition a centrifugate model has been performed in order to: a) Reproduce the falling behavior of the total mass due to different falling velocities. The different velocities reproduce the decreasing of the FS for different causes that can be determined analysisng geognostic and geologic survey. 42 b) Give an estimation of the heights of the cumulate volume of the fallen material in the valley; c) Determine the lateral distribution of the fallen volume; d) Study particular behavior of the potential falling mass in case of solifluction, produced by an excess of pore pressure, or earthquake. The geometrical features of the analyzed landslide are the same considered in the geomechanical model. The input data required are: - Geometry of landslide body - Geometry of the river bed and of the opposite flank of the valley - Geometry and depth of the failure surface - Granulometric composition and mechanical properties of the landslide material. The geometry of the failure surface must be set as input since the models analyze only the postfailure behavior of the landslide. Two different three-dimensional models have been built to make two simulations that differ for level of detail and portion of the area involved in the sliding. Performing the centrifugate test its’ possible to vary two parameters: the centrifugate’s velocity and the inclination of the slope. The test is video recorded and the images are digitalized and memorized in a computer in order to analyze the errors in the falling trajectory due to Coriolis acceleration. At the end of the test the geometry of the fallen volume is measured by means of laboratory instruments. The test gives results as: - The falling velocity of the landslide in function of the Figure 14 friction angle - The geometry of the fallen material in function of the falling velocity. 43 The first result can be used to calibrate the mathematical model; the second result is used to study the behavior the wave of submersion. 3.6.4 Mathematical Models The aim of mathematical methods is to define an algorithm that can describe the “model of behavior of a landslide”. Once the algorithm is found, the “model of behavior of a landslide” can be connected to the monitoring network to compute the safety factor in real time and to indicate conditions of possible risk. The mathematical models are calibrated using the results of the physical models. Different kind of numerical analysis can be carried out: - Limit equilibrium - Finite elements FE - Statistical methods for data processing The input data that can be required by the models are: - Topography - Movement of a set of control points - Data measured by pieziometers and inclinometers - Stratigraphy - Data from geognostic analysis and laboratory test The results of the models are: -Safety factors and falling velocities for limit equilibrium methods -State of stress (from FEM) - Link cause/effect between the environmental parameters, piezometric levels and velocities of movement - Definition of interpretative model of the landslide’s behavior for real time forecast The limit equilibrium analysis has been applied on the case of Spriana landslide by a group of authors. Using the failure criterion of Mohr-Coulomb, they have analyzed different surfaces of rupture: the surface of contact between debris and fractured rocks and two surfaces of discontinuities within the fractured rocks. 44 The results show that along the surface between debris material and rocks the situation is very near to the critical one, along the two other surfaces, instead, the safety factor is less close to one than in the first scenario but the situation remains critical. 3.6.5 Models to Study the Distribution of Fallen Material In literature both empirical and numerical methods to study the distribution of fallen material are available. The empirical methods are based on formulae and diagrams derived from real case that correlates different parameters (Volume, height of falling, gravity acceleration…) involved in the landslide phenomena. The numerical methods usually are one-dimensional models that give information about the distribution of the landslide only along the direction of motion of the falling material. In these models it’s possible to define the geometry of the slope, but huge uncertainties of the results are due to the difficulty of defining the values of the constitutive properties of the materials. Both the empirical models and the numerical models don’t give accurate results, however to have an idea of the size of the distribution of fallen material for Spriana landslide in the following two empirical formula will be applied. The formula of Scheidegger allows estimating the axial extension X of accumulated material according to the volume of the landslide and the gravity acceleration: Figure 15 Another empirical formula allows to find out the value of lateral spread S of the landslide in the valley: Figure 16 45 3.7 Definition of Scenario According to the information obtained from the monitoring networks and the maps it has been decided to consider three possible scenario of hazard. It’s assumed that potential future landslides will occur under the same geological, geomorphological and climate condition as in the past. The “Scenario1” takes in account a debris flow of depth 20-30m. It has been chosen to set the crown of this landslide at 1100m (visible lower crown) because, also if debris material is present at higher altitude, above 1100m its distribution is not constant and the thickness is small. In addition the slope above 1100m is less steep than below. Both “Scenario2” and “Scenario3” describe a possible rock avalanche, however the two scenarios differ from each other because of the different elevations of the crowns and different depths. In “Scenario2” the crown is set at the elevation of 1550 m and the average depth is at 70-80 m from the ground surface. This elevation has been chosen according to the longitudinal section B-B that shows that a visible crown exists at the elevation of 1400 m, detachment of material around 1400m can cause instability in the upper part made of outcrop rocks (between 1400 m to 1550m), for this reason it has been decided to set the crown at 1550 m ASML. The “Scenario 3” considers a catastrophic rock avalanche with the crown at the elevation 1700 m and average depth of 90 m. Figure 17 46 In all the three scenarios the landslide can cause an earth dam in Valmalenco stopping the river Mallero; the difference between the three scenarios is the height and extension of the dam in the valley. Consequently to the formation of the dam,the water coming from the river can accumulate and create a lake. Thus a dam break can occur due either to water pressure or to erosion on top of the natural dam. The fall down of the landslide involved different municipality with different risks. The municipalities of Spriana, Cagnoletti e Arquino, that are close to the instable slope, can be affected by landslide material during the fall. Instead the municipalities upstream the dam as Spriana, Torre di Santa Maria, Marveggia and Sant Anna can be flooded in case of lake formation. The road SP 15, the only one along Valmalenco, can be interrupted by the landslide material, if it happens the valley would be divided in two parts and it would be impossible for the rescue teams to reach the upper one. Finally the risk for the city of Sondrio can be a flash flood due to the dam break, this scenario would be really catastrophic due to the high number of people involved and the velocity of the flood. In Table 28, the characteristics of each scenario and the possible triggering factors and predisposing causes are summarized. Type of failure Geometry Triggering factors Predisposing Causes Scenario 1 Debris flow Scenario 2 Rock avalanche Scenario 3 Rock avalanche Crown: 1100m Foot: 700m Depht: 20-30m -Cumulative rainfall -Snow melt -Seismic movement -Water flow -Erosion or debris active part (under the foot) -Gravity Crown: 1550m Foot: 700m Depht: 70-80m -Cumulative rainfall -Snow melt -Seismic movement -Fault creeps -Movements due to scenario1 -Gravity Crown: 1700m Foot: 700m Depht: 90m -Cumulative rainfall -Snow melt -Seismic movement -Fault creeps -Movements due to scenario1or scenario2 -Gravity - Geomorphological evidence (steep slope) - Fine material near the rupture surface - Presence of two different surface of movement (3Dmodel) - Geomorphological evidence (steep slope, deep seated landslide) - Same orientation of slope and rock layer -Presence of two different surface of movement (3Dmodel) - Geomorphological evidence (steep slope, deep seated landslide) - Same orientation of slope and rock layer Table 279: Characteristics of each scenario and the possible triggering factors and predisposing The magnitude of the landslide for the different scenario can be computed defining its volume; 47 this can be done using a surfer program. The volume of a landslide is defined as the net volume between the current topographic surface and surface of rupture. The shape and dimension of the dam can be estimate using the formulae introduced in the previous paragraph. In particular the formula of Scheidegger allows to estimate the axial extension X of accumulated material according to the volume of the landslide and the gravity acceleration: The lateral spread S of the landslide in the valley can also be evaluated as: The values of volume, axial extension and lateral spread for each scenario are reported in Table29. Scenario 1 Scenario2 Scenario 3 Volume [10^6 m3] 10 40 100 Axial extension X [m] 50 400 600 Lateral spread S [m] 1000 1500 2700 Table 3028: The values of volume, axial extension and lateral spread for each scenario The geomorphologic model can give the measure of the possible height of the dam. In particular the model has been implemented for a landslide with crown at altitude 1400m, foot at altitude 700m and sliding surface at depth 50-70. Under these conditions, considering a partial failure of the slope (less than half of the volume), the resulting height of the dam is 80m in the area of Badoglio and about 40 m in the remaining parts. This result can give an idea of the order of magnitude of the height of the dam also for different conditions of the landslide. 48 4. Early Warning System Since this is a purely technical report regarding the geological hazards in Valmalenco, the aim of this paragraph is simply to give some specific information that can be useful for the future implementation of a complete Early Warning System. In particular, the hazardous events have been analyzed in their technical characteristics and some thresholds have been suggested for the monitoring procedures. 4.1 Definitions and Functions When dealing with hazardous natural events and hydro-geological risks, it is always important to keep in mind that, no matter how many risk reduction and prevention measures are taken, there will always be a so called residual risk that cannot be dealt with using a mitigation approach. On this side, the implementation of an Early Warning System (EWS) may help in managing this residual risk within a wider and more comprehensive risk policy. There are many ideas and definitions of what an EWS should comprehend and of what should its functions be. By combining all these aspects and definitions it is possible to obtain a multilevel structure system based on four main sub-systems: MONITORING and WARNING → hazards monitoring and forecasting to produce information about impending hazardous events; RISK KNOWLEDGE → risk scenarios development to figure out the potential impact of an event (focusing on specific vulnerable groups and sectors of the society); RESPONSE CAPABILITY → development of strategies and actions required to reduce the losses and the damage expected from an impending hazardous event; DISSEMINATION and COMMUNICATION → communication of timely information on an impending event, potential risk scenarios and preparedness strategies. 49 Figure 18 In this report, only the monitoring and warning sub-system will be taken into account. The basic function of this sub-system is to generate adequate information on a given impending hazard. This can be achieved in two ways: MONITORING → it consists in the observation of all those triggering and alarming phenomena which usually come before an hazardous event and which may be its cause or the best conditions for its occurrence; FORECASTING → it is the process of predicting the possibility that a hazardous event will occur with certain characteristics (e.g. time, intensity, place) on the basis of observations and by means of suitable models, 50 Figure 19 4.2 Monitoring and Forecasting System As regards the monitoring system for the case under analysis, it has already been described in the related chapter. For what concerns forecasting, the only instruments currently available are the rainfall forecasts coupled with hydrological-hydraulic models (for example the CERM and the PREVIS models) that give the possibility to forecast the expected values of discharges in the Mallero creek at given sections. No landslide evolution forecast method is available at the present state and therefore the only possibility to assess the evolution of the landslide is to analyze the interaction between the water table variation due to certain rainfall events and the landslide movements. In particular, the hydro-meteorologic monitoring network is connected to the Functional Centre of the Lombardia Region through modem and radio transmission systems. The Functional Centre provides different services: collecting the climatic data; 51 elaborating the collected data; making them available to the local authorities for civil protection purposes in emergency time. The Centre currently collects real-time data from the rain gauge stations widely spread on the territory even in case these are managed by different local authorities; for what concerns the area coverage in terms of rain gauge stations, it can be considered to be satisfactory and properly distributed. Using a decision-making support system, the collected data are constantly analyzed and elaborated by the Centre through mathematical meteorological and hydrological models, in order to make the experts able to produce previsions and forecasts regarding the evolution of the ongoing events. The data collected by the system are: meteorological data for now-casting and predicting; meteo-radar rainfall surveys; meteosat images. The decision-making support systems comprehend: the statistical and geo-statistical analysis of actual and historical data; the extraction of descriptive, anagraphic and managerial information; risk evaluation and prevision models; instruments for the internal and external diffusion of descriptive and synthetic information. The collection of the data required for the territorial control and for the risk evaluation is of paramount importance since it makes the understanding of the evolution of the phenomena simpler and more reliable. The activities of the Functional Centre of the Lombardia Region can be also found in details on the website http://www.protezionecivile.regione.lombardia.it/. For the geological hazard analysis, object of this technical report, an integration between the monitoring system (already established in the area) and the forecasting system will be implemented in order to give as outputs different levels of alert, each of which corresponds to a certain operational procedure to be followed (whose definition will be the object of the emergency plan development). In this framework, it is important to establish some 'standards' for an implemented monitoring and forecasting system to give an effective warning. In particular it is of paramount importance to 52 name: reliability of the instruments → they should always work before and during the event in order to be able to constantly control the situation; redundancy may be a safe approach; suitability of the instruments and of their connection network → they must provide the type and quality of data required by the model and, therefore, if the model needs to be updated or even changed, there is the necessity to update or change the instruments; quality of the collected data → it is of great importance to verify the data before inserting them in the model (rubbish in, rubbish out); uncertainties → they are unavoidable due to the fact that it is not possible to collect deterministic data and by the fact that the data need to be processed to run the model. As regards the first point (reliability of the instruments), a further comment must be made: it is of paramount importance to underline once more the fact that the present state of the monitoring system seems to be inadequate and it should be improved. As it has already been said, a maintenance program should be also included in the management plan of the system. In any case, the present analysis will be dealt with considering the available instruments and networks. 4.3 Definition of the Levels of Alert The definition of the levels of alert and the possibility of passing from one level to another cannot be based on data and parameters rigidly defined. It requires, instead, good interpretative skills regarding the ongoing phenomena and their possible evolution. A proper level of flexibility can be achieved thanks to: an increasing development of the various competences and of the information-sharing skills among the different local authorities; a faster, more codified and complete flow of information between the territorial stations and the Operational Control Rooms; a real-time online sharing of the available monitoring data coupled with an improvement program of the already existing hydro-meteorological stations. What has been done for the present case is an integration among the different aspects that characterize the hydro-geological risk for this specific area, in order to have a complete and 53 comprehensive approach for the evaluation of the levels of alert. As a first step, it has been decided to pose the geological problem, and in particular the landslide hazard, as core example for the definition of a framework useful to set up the thresholds between the different levels. This analysis leads, then, to the identification of different cases that can be grouped into four main alert levels: WHITE, GREEN, YELLOW and RED. The different levels are separated by three alarm thresholds: PRE-ALARM, ALARM and EMERGENCY. 54 4.3.1 Monitoring Parameters for Spriana landslide It is necessary to implement a succession of thresholds regarding the different parameters that can be monitored, thus to be able to set up a framework for the identification of the levels of alert related to the given event. The present scheme has been produced on the basis of the relevant parameters and of the current monitoring system related to the Spriana landslide but they can be easily connected and converted to other events and settings by simply changing the available parameters and the relative thresholds. Figure 20 55 Rainfall As it is well known, rainfall represents one of the two main triggering factors for landslides. In this work, the thresholds regarding the adverse meteorological conditions have been taken from the Regione Lombardia levels: Level Rainfall depth (mm/24h) Normal < 50 Pre-alarm 50 – 80 Alarm >80 Table 29 It is anyway important to remember that: the rainfall values are expected (forecasted) values and not necessarily happening; the forecasts are referred to a particularly wide area (almost all the Bergamo mountain area). Water table level As it is well known, the position and shape of the water table level in a certain slope exerts a great influence on the stability of the landslide. In particular it can be established a correlation between the water table level and the factor of safety (Fs) of a given landslide. Common software like Geoslope can compute the influence of this parameter on the stability of the slope by means of a simple limit equilibrium analysis. In particular, different water table levels have been imposed in the simulation and their relative factors of safety have been computed for Spriana landslide. The details concerning the followed procedure can be found in paragraph 4.4.4. The water table level is, furthermore, related to the rainfall intensity. A hydrological model can be set up in order to understand the effect of a 24h rainfall on the water table level by analyzing the infiltration phenomenon and the underground water flow mechanisms. On the basis of these considerations, three threshold levels have been identified for the factor of safety: 56 Level Factor of safety (Fs) Water table depht [m] Water table depht [m] Pz A Pz B PIEZO1 > 1,3 > 53 > 55 PIEZO2 1,1 – 1,3 36 - 53 22 - 55 PIEZO3 < 1,1 < 36 < 22 Table 30 Displacements Another parameter that is very important to take into account is the magnitude of the displacements that have already taken place within the phenomenon. It is of great importance to underline that at the present state of the art it is not possible to implement a forecasting model for the evolution of the displacement field in a landslide body. What it is possible to do is to take into account the measurements related to the previous months in order to understand something more about the general stability of the slope. On this side, each phenomenon has its own displacement thresholds that need to be evaluated asking to an expert who is able to study the failure mechanisms in detail and to give some limit values for the levels of alert. In any case, three classes have been identified for the present EWS implementation. Level Recent activity EXT0 0 mm (Stable) EXT1 Active EXT2 Very active Table 31 57 4.3.2 Definition of the Levels of Alert The procedure followed till now is able to assess the level of alert whose particular procedure (which will be defined in the Emergency Planning phase) must be implemented in case of an event happening. In this framework it has been decided to consider four different levels separated by three different thresholds. The levels are hereafter described: WHITE → no impending critical conditions, normality → classes 0 – 1 – 2; ◦ PRE-ALARM threshold: ▪ adverse meteorological conditions forecasted; ▪ pre-alarm thresholds reached in the monitoring instrumentation; GREEN → watch out, moderate level of alert → classes 3 – 4 – 5; ◦ ALARM threshold: ▪ adverse meteorological conditions ongoing; ▪ diffused small instability events in a large area; ▪ premonitory events related to landslides and instabilities; 58 ▪ alarm thresholds reached in the monitoring instrumentation; YELLOW → high level of alert → classes 6 – 7 – 8; ◦ EMERGENCY threshold: ▪ landslides activation, urban centers and infrastructures involved; ▪ deformation limits reached in the monitoring instrumentation. RED → very high level of alert → classes 9 – 10 – 11. 4.4 Integration of the EWS for widespread events The early warning system that has been so far presented is suitable for concentrated events (such as the Spriana landslide). Anyway, the possibility to integrate this framework in order to make it available also for the analysis of widespread events, such as erosion or diffused minor landslides, has been taken into account and is hereafter presented. Diffused minor landslides As regards the diffused minor landslides that threaten the greater part of the Mallero basin, in most cases there is no specific monitoring system available and therefore it can be useful to integrate the same EWS that has been designed for the particular case of Spriana so that it is possible to give a certain level of alert to the local authorities on the basis of the available measurements. What is generally available in every basin is certainly a rain gauges network; therefore in this analysis only this type of data will be used. In order to try to make provisions regarding the possibility of occurrence of a landslides it is therefore necessary to first obtain all those data that can be helpful to understand the evolution of the phenomenon in time. Studies about landslides hazard usually make previsions in terms of yearly probability of occurrence or, alternatively, they consider nominal scales (very probable, probable and less probable events). Anyway, for what concerns the Emergency Planning activity, it is necessary to go deeper in the understanding of the phenomena in order to gather all those series of information that can be helpful for the alerting procedures. This can be achieved either by analyzing the temporal series of the events or by analyzing the temporal series of the triggering factors. Since it is very difficult to gather proper historical information regarding the instability phenomena, it is usually simpler to 59 proceed on a much more approximated basis that takes into account the triggering causes. The most significant triggering factors can be briefly summarized as follows: RAINFALL → it increases the volume weight of soils, the porewater pressure and the level of saturation; it might also cause piping and filtration phenomena. All these factors result in a loss of cohesion of the material. EROSION → it modifies the equilibrium profile of the slope (erosion at the foot, loading of the slope, etc). EARTHQUAKES → seismic phenomena can trigger instabilities by directly moving the already unstable slopes but they may also provoke the liquefaction of soils. As regards the particular case study this work is dealing with, it can be stated that the most valid and proper approach for the definition of the previsions of occurrence is based on the evaluation of the triggering thresholds related to rainfall events. Of course this method presents a series of limitations. The experimental equations are referred to correlations computed for shallow landslides and in particular soil slips, debris flows, mud flows, mass sediment transport, debris torrents. Although some graphs individuate some specific portions related to the reactivation of paleolandslides and large-scale landslides, it must be specified that this particular type of prevision can be applied only to shallow flow-like phenomena. The model does not take into account the geotechical properties of the materials involves, in particular it disregards the presence and the type of soil coverage, the granulometry, the thickness of the layers and the geometrical characteristics of the slope. Possible mistakes or generalizations can be therefore due to this lack of input data. The model doesn't give any kind of indication regarding the localization of the single phenomena but it considers only a critical rainfall regime for the triggering of shallow landslides and debris flows in a widespread area where soil movements are possible. The following graph shows the relationship that can be highlighted between the rainfall regime and the activation of instability events. 60 Figure 21 The use of climatic data for these purposes, at a local scale and at the present state of the art, gives the possibility to individuate the rainfall thresholds for the triggering of shallow landslides and debris flows. For the Lombardia region (where the specific case study of this work is located) the limit values are computed applying proper equations to the mean annual values. These equation have been developed by the Servizio Geologico della Regione Lombardia, in particular by Ceriani, Lauzi & Padovan (1992,1994). One of the most used methods is the one developed by the Servizio Geologico della Regione Lombardia (Ceriani, 1992) for the individuation of the triggering curves for shallow landslides. The normalized rainfall intensity curve that sets up the initial stage (less that 10 landslides per km2) has been computed using the equation , the intermediate stage curve (10-20 landslides per km2) has been computed using the equation and the catastrophic stage curve (more than 20 landslides per km2) has been computed using the equation . The normalized hourly intensity (IN) can be obtained by multiplying by 100 the hourly rainfall and mean yearly rainfall ratio. D is the duration of the event (in hours). The general procedure that must be activated even during an ongoing emergency can be developed in the following way: (a) Acquisition of rainfall data from the nearest rain gauge (from Centro Funzionale Regionale). 61 (b) Elaboration of the rainfall intensity normalized with respect to the mean annual values for the time intervals of the most severe events (or ongoing events). (c) Superposition of the values found at points (b) and (d) and individuation of the intersections between the considered curves and the triggering thresholds. (d) Analysis of the triggering thresholds and hazard assessment. Erosion As regards the problem of erosion, it must be highlighted that the spatial scale considered is much different from the one that characterizes the problems of the major concentrated landslides and of the minor diffused landslides. The area that must be taken into account is much wider and it generally coincides with the whole hydrological basin of the considered river or creek. Much more integrated and specifically set up monitoring system can be used; in order to better forecast the actual and future risk scenarios, for this purpose it would be necessary an effective monitoring system characterized by real time processing, remote observation capabilities and an high grade of flexibility. Ground-based SAR (GB-SAR) interferometric systems, are able to provide multitemporal deformation maps of areas prone to erosion at time intervals that range from few hours up to many days, using pure remote observations. Anyway, there is no evidence of the presence of such instrumentation in the considered area and, in any case, it seems too expensive to be installed for the given purposes. Since the main triggering factor of erosion is the rain in order to monitor the production of sediment an effort can be done to determine which rainfall event can cause a sediments’ volume that can be considered as a critical threshold value. Definition of thresholds for erosion In the previous chapter it has been computed the volume of sediments Ws87 eroded from the slope for an event of Tr=100 years and duration 24 hours, this event is comparable to the one that occurred in July 1987. In that occasion the huge amount of sediments that flowed through the Mallero river caused terrible consequences in term of flooding in the city of Sondrio. In order to develop an emergency plan for Sondrio, it has been thought that it can be interesting to 62 evaluate which kind of minor rainfall event can cause a smaller sediments’ volume that can be considered as a critical threshold value. In particular it has been decided to assume such threshold value equal to and it has been evaluated the kind of precipitation that can generate it. Then the procedure has been repeated to determine the rainfall events that can produce a volume of Ws87 to determine an intermediate threshold between dangerous event and really dangerous event. For this purpose the MUSLE method, previously described, has been applied in a reverse way: the volume of sediments has been defined and the amount of rain that generates it has been computed. In particular the MUSLE method computes the amount of eroded sediments in a catchment on the basis of morphologic data, information about the shallow strata of soil and the vegetation covering it, cumulated rainfall depth and peak flow liquid discharge at the outlet section of the considered basin. The data concerning the basin have been kept equal to the one used to run the MUSLE model for the 1987 event, the rainfall depth has been made varied to obtain the target sediments’ volume and the peak flow discharge has been estimated using the corrivation model (or linear kinematic model). That model is based on the concept of “corrivation time” of the net rainfall from the point where the raindrop reaches the soil to the downstream section of the basin. (For the previous analysis of the event of 1987 it hasn’t been necessary to use such formula because the values of peak flow were available in literature). The corrivation time To has been computed as: in hours Where: L= length of the main reach in Km H=mean height of the basin in m Z= height of the outlet section S= area of the basin Solving the previous formula it results: To= 3,8 hr for the sub-basin A1 To= 4,7 hr for the sub-basin A2 Since the corrivation times are different for the two sub-basins the analysis of the thresholds has been carried on separately for A1 and A2. 63 The basins have been considered linear thus the area-time curve is represented by a line The discharge at the outlet sections of the two sub-basins A1 and A2 can be computed as , where is the intensity of rain equal to the ratio between rain depth(h) and rain duration (θ). If the duration of the rain is longer than the corrivation time (θ>To) the Qpeak is reached at time equal to , if , instead, the duration of the rain is shorter than the the corrivation time To and it results corrivation time (θ<To) the Qpeak is reached at time θ and it results . In particular four rainfall events of different duration (3hr , 6hr ,12 hr and 24 hr) have been considered and for each of them it has been computed the depth of rain that generates the volume of sediments equal to Ws87 , Ws87 and Ws87. The results of the analysis are: sed. of 1987 event h [mm] in 3 hours h [mm] in 6 hours h [mm] in 12 hours h [mm] in 24 hours Volume sediments [m3/event] 72 90 127 180 1282257 66 75 106 149 350291 2/3 sed. of 1987 event h [mm] in 3 hours h [mm] in 6 hours h [mm] in 12 hours h [mm] in 24 hours Volume sediments [m3/event] 50 63 88 125 854838 46 52 73 104 233527 1/3 sed. of 1987 event h [mm] in 3 hours h [mm] in 6 hours h [mm] in 12 hours h [mm] in 24 hours Volume sediments [m3/event] 27 34 48 67 427419 25 28 40 56 116764 The thresholds can also be plotted in depth-duration curves and intensity-duration curves. 64 It must be noticed that the values of volume of sediments refers to the total volume of sediments produced in each basin from the erosion of the slope but it doesn’t take in account the river bank erosion, the shallow/concentrated landslides and the fact that part of these sediments deposit along the river before reaching the final section. As discussed in the paragraph 2.4 , for the event of 1987 the amount of the contribution due to bank erosion and shallow/concentrated landslides and the quantity of deposit can be found in literature. For the two other scenario characterized by lower rainfall the values reported in the previous tables cannot be evaluated, anyway it can be chosen to consider the same percentages as estimation. However that hypothesis is a really strong assumption because if the precipitation varies, even the liquid discharge vary and thus the amount of erosion/deposition along the river banks varies as well. To conclude it’s worthy to notice that as mentioned in the paragraph 2.6 the uncertainties linked to MUSLE methods are high. With the purpose of reducing this uncertainty the parameters of the model have been calibrated with observed data of 1987, thus the results can be considered quite precise for the 1987 event but the uncertainty increases a lot for the two other scenarios. In order to try to decrease this uncertainty it can be suggested to set some flow gauges in the river and to increase the 65 number of rain gauges in order to have a more precise correlation between rain and Q in some sections of the river. 4.5 Case study: Spriana Landslide The core phenomenon that has been considered in this work can also be analyzed in order to understand the effects of the 1987 event on the stability of the Spriana landslide body. As it is known, this event caused several floods and debris transport problems in the whole Torreggio and Mallero basins up to the confluence with the Adda River. The event did not trigger any failure in the Spriana landslide and therefore it is interesting to analyze how a rainfall event, that caused so many harmful consequences in terms of erosion and sediment transport, could possibly influence the stability of a large landslide like Spriana. The particular scenario that has been considered is the worst case regarding the major failure of the slope with surface of rupture at 90 m of depth and almost 108 m3 of rocks and debris in movement. The same procedure that has been described until now, will be used for the evaluation of the level of alert for the Spriana case during an event with the same characteristics of the one that took place in July 1987. 4.5.1 Monitoring Parameters Rainfall As it has already been said, the rainfall event that has been considered is the July 1987 one that was characterized by an average rainfall depth of 106 mm in 24 hours over the whole considered basin. Therefore, the class that has been applied is the Alarm one. Water table level As regards the water table level, as it has already been said, the limit thresholds should be individually analyzed for each particular event. For this case study, the software Geoslope has been used in order to set up the water table levels referred to the threshold values of the factor of safety. The software uses the limit analysis in order to establish the Fs of the slope once the surface of rupture, the slope profile, the geometry of the layers, the soil properties and the water table surface have been 66 defined. For what concerns the geometry of the slope profile and of the different layers, the model that has been implemented can be found in the following picture with the position of the piezometers that we suggest to place in order to collect the measurements useful for the implementation of the EWS. Figure 22 It has been decided to define only two layers, one made of debris and the other one made of fractured rock. As regards the geometry of the layers it has been set up as follows. Figure 23 67 For both layers, the properties that have been used are: Layer Type of rock/soil RMR Q Debris Limestone, Marl, Dolomites 3 A B T 0.01 0.042 0.534 0 Fractured rock Amphilobites, Gneiss, Granites 23 0.1 0.203 0.686 - 0.0001 Table 32 And the results found for different water table surfaces are: Fs = 1.52 Figure 24: Fs = 1.52 68 Fs = 1.34 Figure 25: Fs = 1.34 Fs = 1.2 Figure 26: Fs = 1.2 69 Fs = 1.12 Figure 27: Fs = 1.12 Fs = 1.00 Figure 28: Fs = 1.00 70 In the particular case study of 1987 that has been considered in this study, even if it is difficult to recover precise data regarding the water table depth, the level can be set as not critical and so lower than the one that gives a factor of safety equal to 1.1 , therefore, the slope can be considered stable. The class that has been applied is therefore the PIEZO2. Displacement Since there is no evidence of any relevant displacement during the event of July 1987, it has been decided to apply the EXT0 class to the phenomenon (stable slope). Final monitoring parameters assessment Intersecting these three values, it has been possible to evaluate the monitoring parameter class for the considered event. In particular, a class 7 has been applied to the phenomenon. 4.5.2 Definition of the Level of Alert Using the framework that has been implemented in this analysis, it has been possible to assess the level of alert that should have been issued in case the present EWS had been activated during the 1987 event. In particular, it has been assessed that a class 7 should have been applied, which corresponds to a YELLOW level of alert. This means that even if there have been clear evidences of adverse meteorological conditions ongoing, of diffused small instability phenomena in a large area (erosion as well) and of monitoring thresholds trespassing, the stability of the landslide is not completely compromised. In fact the slope did not collapse during the 1987 event and this can be considered as a confirmation of the validity of the framework that has been implemented in this work for the design of a proper early warning system working on the Spriana area. 71 Bibliografy ISMES (1989) – Piano Programma di Ricostruzione,Riconversione e Sviluppo della Valtellina e delle zone adiacenti delle provincie di Como, Bergamo e Brescia colpite dalle avversità atmosferiche dei mesi di Luglio ed Agosto 1987. Prog Asp/4430 ISMES (1990) – Modelli comportamento frana. Modello geomeccanico. Prog Asp/4914 ISMES (1990) – Modelli comportamento frana. Modello geomeccanico. Rapporto finale. Prog Asp/4914 ISMES (1990) – Modelli comportamento frana. Modello in centrifuga. Prog Asp/4914 ISMES – Modelli comportamento frana. Modello in centrifuga. Disegni ISMES (1990) – Modelli comportamento frana. Modello matematico. Prog Asp/4914 ISMES (1990) – Modelli comportamento frana. Modello matematico. Disegni. Prog Asp/4914 ISMES (1990) – Indagine geofisica mediante sismica a rifrazione eseguita in località Case Cucchi nel comune di Spriana (So) ISMES (1990) – Indagini ed ulteriori studi nel versante in frana. Cunicolo esplorativo. Prog. Asp4914. L.NOE, R. LANCELLOTTA, F. LIMONETTI, A. MARCELLO - Val Malneco : inquadramento geologico, problemi di stabilità , criteri di intervento , Milano,1990 ISMES (1991) – Modelli matematici di comportamento della frana. Aggiornamento al 30.04.1991. Centro di monitoraggio, Sondrio ISMES (1992) – Idrologia e trasporto solido. Prog Asp/4914 ISMES (1993) – Frana di Spriana. Piano protezione civile Rapporto finale.. Prog Asp/4914 ISMES– Rapporti complementari; Aree di invasione della frana. Rapporto finale.. Prog Asp/4914 ISMES – Disegni. Centro di monitoraggio, Sondrio RADAELLI, CASTELLOTTI – Indagini geognostiche. Centro di monitoraggio, Sondrio ISMES – Schemi interpretativi del rischio geologico e idrogeologico, definizione dei livelli di soglia e delle procedure di analisi dei dati strumentali, nell’area della frana di Spriana. Prog. Asp/4914. ISMES – Considerazioni sull’evoluzione del versante durante il periodo di monitoraggio:1990-1992 CENTRO MONITORAGGIO GEOLOGICO – Sintesi dei dati registrati nel periodo 1990-1998 QUADRO DI RIFERIMENTO AMBIENTALE - Studi propedeutici alla valutazione dell'impatto ambientale degli interventi del bacino del torrente Mallero Piano comunale per la protezione civile (2000) Piano di emergenza intercomunale Piano comunale per il coordinamento delle operazioni di evacuazione e di soccorso della popolazione in caso di eventi calamitosi concernenti il torrente Mallero (1987) Sito web ARPA Lombardia: www.arpalombardia.it Sito web ARPA Lombardia – centro monitoraggio geologico: Http://89.118.97.243/webcmgfrontend/default.asp 72