Lightning produced NOx - Institut für Physik der Atmosphäre

Lightning produced NOx - Institut für Physik der Atmosphäre

Lightning produced NOx (LINOX) Experimental Design and Case Study Results
Hartmut Höller, Ullrich Finke, Heidi Huntrieser, Martin Hagen, and Christian Feigl
Institut für Physik der Atmosphäre, DLR-Oberpfaffenhofen
Abstract. This paper investigates the role of lightning in the production of nitrogen oxides (NOx) and
their subsequent distribution by thunderstorms. These questions were addressed by the field
experiment LINOX (Lightning produced NOx) which was performed in southern Germany in July 96.
The structure of thunderstorms was observed by radar and satellite, the lightning activity was recorded
by a lightning detection network and airborne chemical measurements were performed aboard a jet
aircraft penetrating the storm anvils. NOx concentrations in the storm anvils were found to typically
range from 1 to 4 ppbv. The NO contribution to the total NOx was found to be dominant in narrow
peaks produced by flashes as well as near cloud boundaries probably due to increased photolysis rates
of NO2. Using CO2 as an air mass tracer, the lightning produced NOx amount was discriminated from
the contribution due to transport of air from the boundary layer. It was found from a case study of a
large storm anvil that lightning produced NOx was present in the same order of magnitude as the
amount of NOx originating from lower levels, during later stages of cloud development, the content of
the former even exceeded the latter one. A simple 2D model of advection and dispersion of the
lightning produced NOx was able to reproduce the general structure of the anvil NOx plume. Some
NOx peaks could directly be attributed to flash observations close to the aircraft track.
1
Introduction
Reactive nitrogen oxides NOx (NOx=NO+NO2) are
central to photochemistry. The ozone production rate
depends strongly and non-linearly on NOx
concentration [Ehhalt and Rohrer, 1995]. The main
inputs of NOx to the middle and upper troposphere
arise from vertical transports originating in the
atmospheric boundary layer, from lightning in
thunderstorms, due to transport from the stratosphere,
and in-situ emissions of NOx from aircraft [Levy et
al., 1996].
The global lightning induced NOx (LNOx) source
represents the largest uncertainty in the global NOx
budget, with estimates varying between 1 and 220
TgN/yr [Liaw et al., 1990]. More recent model
analysis and field experiments come up with more
narrow ranges of 2 to 5 TgN/yr [Levy et al., 1996;
Ridley et al., 1996]. The strength and distribution of
the LNOx sources must be known, in particular, for
assessments of the impact of aircraft NOx emissions
on the state of the atmosphere [Beck et al., 1992;
WMO, 1995; Brasseur et al., 1996].
Up to now few measurements relating NOx
concentrations to thunderstorm activities have been
performed. Dickerson et al. [1987] studied trace gas
transport in a midwestern thunderstorm. They
measured up to 4 ppbv of NOx in the storm’s anvil
and concluded that about 3 ppbv must have originated
from lightning. Further airborne measurements
showed peak NOx concentrations of up to 4 ppbv over
Kansas [Luke et al., 1992], up to 1.3 and 2.0 ppbv NO
within the anvil of 2 thunderstorms over New Mexico
[Ridley et al., 1996], and about 0.7 ppbv NO after 24
hours in the outflow of a thunderstorm over Spain
[Huntrieser et al., 1996].
Some of the key questions which were investigated
by the LINOX field experiment are: (1) What is the
lightning distribution within a thunderstorm? (2)
What is the NOx concentration field in the vicinity
and in the anvils of individual thunderstorms? (3) Can
the different sources of NOx be discriminated by
measurements and thus determine how much NOx can
be attributed to lightning?
In LINOX, for the first time, a combination of
measuring tools was available for looking at
2
Nürnberg
Germany
N
Switzerland Austria
Italy
DLR
Alps
Figure 1. Map of the LINOX experimental area around the DLR radar site. Two different types of flight
patterns are shown: pattern 1 illustrates the long range surveillance flights, pattern 2 was performed in and
around thunderstorm anvils preferentially in the SOA (Special Observation Area).
lightning, microphysical, and chemical phenomena in
thunderstorms; something not reported before.
Huntrieser et al. [1998] have presented an
evaluation of the LNOx production of the
thunderstorms observed during LINOX in a statistical
sense. The contribution of transport of low level air to
the total NOx content observed aloft was assumed to
be represented by measurements in cumulus
congestus clouds. For larger storms it was found that,
in general, 60-75% of the total NOx present in the
anvils were due to lightning activity.
In addition to this overall evaluation, a detailed
investigation of a thunderstorm event including radar
and lightning observations is reported in this paper.
Applying parameterizations of NOx production by
individual flashes, lightning data enable an
assessment of the LNOx field in the storm outflow
region. These results will be compared with the actual
airborne chemical measurements.
2
Experimental Design
During the LINOX field experiment new
measurements and observations of thunderstorms,
lightning and NOx were performed in July 96 in
southern Germany (see Figure 1).
The major experimental tools and supporting
observational systems involved will now be detailed.
• Research Aircraft (FALCON). Airborne
measurements were performed using DLR’s research
jet equipped with chemical probes for measuring NO,
NO2, CO2, and O3 [see Schulte et al., 1997]. The NO
and NO2 measurements made use of the
chemiluminescence technique in combination with a
photolytic NO2-to-NO converter. The NO2
concentration can be determined from the converter
produced concentration [NOc] using
[NOc] = [NO] + γ [NO2]
(1)
where γ = 0.66 is the converter efficiency. Basic
meteorological parameters like pressure, temperature,
humidity and wind were also recorded.
• Lightning Location System (LPATS). Lightning
data were provided from a 2-dimensional lightning
location system (LPATS - Lightning Position and
Tracking System [Bent and Lyons, 1984]), which is
operated by local power stations (‘Bayernwerke’ and
‘Badenwerke’). The lightning localization is based on
time-of-arrival of the signals for at least 3 stations.
3
350
355
360 CO2 (ppmv)
Stratosphere
12
12
10
NO
Height (km)
Height (km)
10
8
O3
6
CO2
4
8
6
2
0
0
0
50
100
150
200 O3 (ppbv)
10
8
Troposphere
6
0
NO/NOx
0.2 0.4 0.6 0.8 1
4
2
(a)
Airborne Data
NO2
Height (km)
345
12
Boundary
Layer
0
(b)
0.1
Ground Based Data
0.3
0.2
Mixing ratio (ppbv)
0.4
Figure 2. (a) CO2 and O3 profiles obtained during aircraft ascent on 23 July 96 in the pre-storm
environment, (b) NO and NO2 profiles obtained from aircraft measurements during the later stages of the flight
as compared to the ground based mean values from the van for the same period.
The type of discharge (cloud-to-ground - CG, intracloud - IC) is inferred from the signal shape and
polarity. In the case of a CG flash this roughly
corresponds to the location of the flash’s impact on
the ground. The detection efficiency of the system is
about 70% for CG lightning but much less
(approximately 1%) for IC discharges.
• Polarimetric and Doppler radar (POLDIRAD).
Polarimetric radar measurements provide estimates of
different kinds of hydrometeors [Höller et al., 1994].
In addition, Doppler measurements with POLDIRAD
were performed showing interesting storm
developments in a range up to about 100 km from the
radar.
• Supporting Observations. There was a variety
of activities supporting the experiment. Ground based
measurements have been obtained from a van
equipped with standard meteorological instruments
(recording pressure, temperature, humidity, and wind)
as well as with chemical sensors determining
concentrations of NO2, NOx, and O3. Radar data,
lightning observations and radio soundings from
neighboring countries were made available. Satellite
data were used for monitoring the development of
convection
The airborne investigations were performed close
to the radar site in order to measure the NOx outflow
from individual thunderstorm anvils in conjunction
with the radar and lightning information. The
experimental concept was to infer the lightning
produced amount of NOx from the anvil
concentrations using, for the first time, CO2 as a
tracer for boundary layer air. During the upward
transport the low level air is mixed with the
environmental air. It is assumed that the mixing
process is similar for CO2 as well as for NOx. The
difference of the expected NOx mixing ratio and the
aircraft measurement is assumed to result from
lightning activity.
3
The Storm Environment on 23
July 96
On 23 July 1996 a cold front approached the
experimental area (to the west of the DLR site, near
Munich) during the afternoon hours. The storms
investigated here formed during the passage of the
frontal system.
In a thunderstorm a rapid vertical transport of low
level air towards the upper troposphere is
accomplished with strong updrafts. During this
transport the rising air is undergoing mixing processes
with the surrounding air mass. The NOx produced by
lightning (LNOx) in the storm is mixed with the NOxenriched air originating from the boundary layer. In
order to discriminate between the contributions from
the two different sources by tracer analysis one has to
consider the background concentrations of NOx in the
boundary layer as well as in the free troposphere.
For characterizing the storm environment, airborne
measurements, located closely in space and time to
4
Free Troposphere
Figure 3. Transport and mixing model used for
calculation of LNOx production. Boundary layer air
(1) is mixed with upper level air (2) by entrainment
processes into the storm updraft. LNOx is added
during ascent of the air mass. High NOx but low CO2
mixing ratios characterize the boundary layer air,
whereas the opposite holds for the free troposphere.
the large cloud system (discussed in section 5), were
extracted. As the O3 measurements were
contaminated by precipitation particles during the
later cloud penetrations, the ’representative’ O3 and
CO2 (for consistency) profiles are taken from the
aircraft ascent from 15:00 to 15:50 UTC in the prestorm environment mostly undisturbed by convection
(Figure 2a). The boundary layer air can clearly be
recognized by its low CO2 and high O3 content. The
CO2 minimum can be attributed to the vegetation
which takes up CO2. A distinct increase of CO2 is
observed at the top of the boundary layer at about 2.5
km height from 348-350 ppmv at low levels to 356362 ppmv in the troposphere. O3 does also show a
change in concentration but, in contrast to CO2, the
values first decrease and then increase again with
height. This holds at least for a smoothed profile not
taking into account the small scale variations with
height. Therefore only CO2 (and to a rather limited
degree also O3) can be looked upon as an air mass
tracer. Low CO2-values found aloft in storm updrafts
or anvils indicate a transport of boundary layer air
into higher levels via updrafts.
Information on ground level concentrations of O3,
NO and NO2 is provided by the chemical
measurements taken from a van. It was positioned in a
rural environment about 50 km west of the radar site.
The storms penetrated by the aircraft passed about 5080 km to the north of this site. Therefore no direct
indication of enhanced NOx concentrations due to
lightning could be detected in these data. NOx values
range from 2-5 ppbv during the time of the aircraft
operations. Most of the NOx is available as NO2 as
would be expected from photochemical equilibrium.
The mean ratio NR=[NO]/[NO2] was 0.1 for this
period. The O3 measurements of about 80-90 ppbv are
consistent with the aircraft observations in the late
afternoon hours.
Due to the instrumental design, airborne NOx
measurements are only available in the upper
troposphere at heights exceeding about 6 km. Figure
2b shows extra-cloud profiles from the later stages of
the flight around 17:50 UTC when the aircraft began
its final decent. The data shown were obtained outside
the active thunderstorms and outside the anvil region.
In the upper troposphere NOx values are relatively
low as compared to the ground-level values. Both,
[NO] and [NO2] are smaller than 0.05 ppbv below 8
km height where the ratio NR does not exceed 0.4.
Probably due to a low photolysis rate caused by the
cloud cover present at high and mid-levels, the NR
values are quite low during these early evening hours.
The NOx measurements seem to be quiet reliable as
the accuracy of the airborne NO (NO2) measurements
can be assessed to 8% (15%) under the observed
conditions [Ziereis et al., 1999] and possible
interferences are not likely to have occurred due to
the low concentrations of nitrogen compounds in the
upper troposphere. Closer to the tropopause, the
mixing ratios increase. Values in excess of 0.1 ppbv
are connected with NR ratios larger that 0.4. The
relative contribution of NO to the total NOx does
increase in accordance with an increase of the photodissociation rate with height. Possibly due to the
presence of cloud covering, this increase is not strictly
monotonic.
4
Principles of LNOx Assessment
Some principle considerations used for the
assessment and the interpretation of the NOx
production by lightning will be discussed in this
section.
(1) With respect to the environmental data
available (as discussed in the previous section) a
simple mixing analysis is applied for performing the
5
LNOx assessment.
(2) The presence of narrow peaks in the NO time
series is indicative for a recent source generated by an
individual lightning flash.
(3) Considering most of the LNOx is originally
produced as NO, the chemical composition can give
hints towards the plume age (during the first few
minutes after production). Later on high NO
concentrations can also be caused by high photolysis
rates under suitable radiational conditions sometimes
extending over large cloud areas.
For the assessment of lightning NOx production it
is assumed that the simplified profiles shown in
Figure 3 can be applied. The low level air lifted in the
storm updraft (CO2,1 ,NOx,1) is mixed with the upper
level air (CO2,2 ,NOx,2). The mixing is looked upon
as an entrainment process adding environmental air to
the cloudy air mass. Detrainment is neglected during
the growing phase (or in the main updraft) of the
cloud. The degree of mixing of low level air with
environmental air is described by the entrainment
parameter µ:
[CO2,1] + µ·[CO2,2] = (1+µ) [CO2,m]
(2)
where [CO2,m] is the CO2 mixing ratio of the mixed
air masses. The mixing parameter µ can be obtained
from Equation (2)
µ=
[CO2,m] − [CO2,1]
[CO2, 2] − [CO2,m]
(3)
and [CO2] of the mixture amounts to
[CO ] =
[CO ] + µ ⋅ [CO ]
2, m
2 ,1
2, 2
(4)
1+ µ
The same mixing procedure is also assumed for the
NOx mixing ratio
[NO x,m] =
[NO x,1] + µ [NO x,2]
1+ µ
(5)
Using µ as computed from Equation (3), the NOx
mixing ratio expected inside the cloud resulting from
transport (advection and mixing) of boundary layer air
can be computed from Equation (5). If an additional
NOx source due to lightning is present during this
transport process, its concentration [NOx]L can be
assessed from the difference of the measured NOx
mixing ratio [NOx]M and the transported amount
[NOx]T as computed from Equation (5), i.e.,
[NOx]L = [NOx]M – [NOx]T
(6)
The procedure described above critically depends on
the choice of the environmental parameters
characterizing the regimes (1) and (2). The CO2
profile can be represented by average values taken as
[CO2,1].= 349 ppmv and [CO2,2] = 357 ppmv as can
be inferred from the Figure 2. These values are
obtained from the airborne measurements. The upper
level NOx amounts are mostly smaller than about 0.1
ppbv except for heights above 9 km.
Due to the lack of airborne NOx measurements
below 6 km height, the assessment of a representative
low level NOx concentration is rather difficult. The
ground based measurements taken from the van may
only be looked upon as an upper limit. The
temperature and humidity profiles as obtained from
aircraft and radio soundings (not shown here) indicate
that the boundary layer was not strictly well mixed
during the specific observation period. This indicates
that the NOx mixing ratios which were measured at
the ground (ranging between 2 and 5 ppbv) are
probably
overestimating
the
storm
inflow
concentrations if assumed representing low level
values.
This interpretation is further supported by the
mixing computations itself. It is required that the
computed in-cloud NOx mixing ratio due to transport
must always be smaller than the observed value. This
is only ensured for low level mixing ratios in the
order of 0.5 to 1.5 ppbv. Furthermore, as was
demonstrated by Huntrieser et al. [1998], typical NOx
mixing ratios in cumulus congestus (cu-con) clouds
did rarely exceed 0.5 ppbv on the same day. These
clouds can be assumed as not being contaminated by
lightning produced NOx. On the other hand, cu-con
clouds can be expected to have undergone a mixing
process with the environmental air during ascent of
the low level air mass. Thus, a NOx mixing ratio of 1
ppbv is assumed in the following as being
representative for the low level air, i.e. [NOx,1] = 1.0
ppbv, whereas the upper level air is characterized by
[NOx,2] = 0.1 ppbv.
6
Beside the mixing considerations discussed above,
a further principle aspect of data interpretation is the
age of a plume and the relation to its spatial width.
The NO measurements have a temporal resolution of
2 s (see also [Schulte et al., 1997] for a description of
the
chemiluminescence
instruments)
which
corresponds to a spatial resolution of about 400 m
(aircraft velocity was around 200 m/s). The original
flash containing the NO, produced during the
discharge process when the temperature has dropped
to about 3000 K and the NO gets frozen in, has a
dimension of a few centimeters (approximately 24 cm
after Hill et al. [1980]). Mixing of the channel air
with the surrounding air is responsible for the
subsequent NO dispersion.
For assessing the time scale involved a simple
Gaussian model of plume broadening is assumed. At
the beginning of the diffusion process the width of the
plume at time t can be described by
σ2 = ε t3
(7)
where σ2 is the variance and ε is turbulent energy
dissipation rate [e.g. Blackadar, 1997]. The energy
dissipation rate can be obtained from the radar and
aircraft measurements [Meischner et al., 1997]. It was
highly variable within the storm and values ranges
between 10-4 and 10-1 m2s-3. For a rough assessment
of the mixing an average value of ε = 5 10-3 m2s-3 is
assumed here. With this value a broadening of the
plume to a diameter of 400 m can be established
within 200 s (between 1 and 10 min for the extreme
values mentioned above). Thus a δ-shaped peak in
mixing ratio represented by a single data point would
be indicative of a fresh NO emission not older than
about 3-4 min. This time scale is also typical for the
photochemical equilibrium to be established among
NO and NO2. Therefore, it is not likely to detect
LNOx from the chemical composition of an air
sample within older and broader NO peaks.
The composition is determined by the amount of
ozone available and by the radiation conditions
governing the photolysis rate of NO2 (see Equation
11). The reactions dominating the nitrogen chemistry
are
NO + O3
NO2 + hν
→
→
NO2 + O2
(8)
NO +O
(9)
O + O2 (+M) →
O3 (+M),
(10)
demonstrating that radiative effects (photolysis of
NO2, Equation (9)) and oxidation (ozone reactions,
Equations (8) and (10)) are the most important
influencing factors. Reactions involving peroxy
radicals like HO2, CH3O2, and RO2 can also
contribute to a conversion of NO to NO2. But,
considering the relatively high NOx concentrations
within a thunderstorm anvil, peroxy radical
concentrations can be expected to be low enough for
these reactions not to be dominant [Ridley et al.,
1992]. Except on short time scales of the order of a
few minutes, the reactions (8) to (10) cause the
concentrations to remain in a quasi equilibrium state
which can be determined from
[ NO ]
j
=
[ NO2 ] k[O3 ]
(11)
where j is the photolysis rate and k is the reaction
constant of Equation (10). k depends on temperature
T according to
k = ae
−
b
T
,
(12)
where a and b are appropriate constants.
5
Discussion of Anvil
Penetrations
The convection in the southern part of the frontal
system intensified when it approached the observation
area at about 16 UTC. Various thunderstorms
developed in connection with the frontal passage. The
FALCON aircraft penetrated the anvils of these
storms which were in different stages of development.
In the following, measurements will be discussed
which were made in (i) a small thunderstorm cell that
had produced a few flashes prior and during the time
of the aircraft penetration and (ii) a large storm
system that had developed a widespread anvil and
which was combined with a very active zone of
lightning occurrence connected with the main storm
(radar reflectivity) center.
7
8
8000
Flashes per Minute
2
0
-2
50
7000
6000
(a)
120
-4
360
5000
100
356
80
40
30
1/min
4
Height (m)
w (m/s), q (g/kg)
6
20
352
60
20
344
Mixing Ratio (ppbv)
1340
16:30
17:00
17:30
18:00
18:30
19:00
Time (UTC)
0.8
1.5
0.6
1
0.4
0.5
0.2
(c)
20
Mixing Ratio (ppbv)
0
16:00
(b)
02
0
Figure 5. Time series of flash frequency registered
by the LPATS network for the penetrated storm
system between 16:00 and 19:00 UTC.
1.5
1
0.5
(d)
0
Figure 4. Cloud penetration between 16:15 and
16:20 UTC. (a): vertical velocity w and height h of
the aircraft; (b): O3 and CO2 mixing ratios; (c): NO,
NO2, NOx mixing ratios and the NR ratio; (d): NOx
mixing ratio from measurement, calculated NOx
mixing ratio resulting from transport [TNOx] and
from lightning [LNOx].
5.1
10
348
40
Small, isolated cell
At 16:17 UTC the aircraft penetrated a small storm
which was releasing precipitation to the ground. The
storm had already produced some (5) cloud-to-ground
(CG) flashes and 2 intra-cloud (IC) flashes had been
recorded by the LPATS system. The aircraft entered
the cell at about 7.2 km height and ascended further
up to 7.9 km where it left the cloud. Peak updraft
velocities of about 7.2 m/s were encountered (Figure
4a).
Inside the cloud the NO and NO2 mixing ratios
were distinctly higher than in the environment. Peak
[NO] ([NOx]) values of 1.7 (1.8) ppbv have been
detected. Ozone concentration was also higher inside
the clouds, whereas CO2 mixing ratios drop down to
349 ppmv, a value assumed as representing the
boundary layer conditions. These opposite tendencies
of NO, NO2 (NOx) and O3 on one hand and CO2 on
the other hand do reflect the characteristics of the
profiles as discussed above (e.g. Figures 2 and 3).
Following the tracer analysis discussed in the
previous Chapter, one can separate the advective part
of the total NOx amount from the lightning produced
part as shown in Figure 4d. The contribution from
both parts are of about equal magnitude.
In Figure 4c the ratios of NO and NOx are
displayed. Extremely large values (near unity) can be
noted upon entering the cumulus towers from the SW
side. Nearly all of the available NOx is made up of
NO. It is likely that this does not originate from fresh
NO emissions due to lightning but, moreover, is due
to an enhanced radiational forcing causing very high
photolysis rates j (see Equation 11). The cloud tower
is illuminated on its western parts by the sun. Inside
the cloud the ratio decreases gradually and finally
reaches a value of 0.4 upon leaving the cloud. These
values are typical for the eastern side of the storm at
about 8 km height with the Cb anvil extending
eastward right above the aircraft flight level.
Enhanced photolysis rates in clouds have also been
documented e.g. by Kelley et al. [1995]. They found
enhancements of up to 58% when entering the tops of
altocumulus clouds at 7.6 km height. Even if the
present experiment did not include radiation
measurements, the NO registrations suggest that the
effects on the photolysis rate might be even more
pronounced under thunderstorm conditions.
5.2
Large mature storm
The storm discussed above was part of a multiple
8
110
100
2
100
90
80
70
17:06
O3
17:05
70
8200
100
80
60
40
20
40
17:04
60
60
17:07
17:03
50
50
8180
(b)
8160
360
O3
CO2
355
350
(c)
(d)
3
1
345
NO
NO2
NOx
NO/NOx
0.8
0.6
2
0.4
17:08
1
40
40
17:09
0.2
0
4
0
NOx
TNOx
LNOx
3
30
30
(a)
130
120
110
100
90
80
70
60
50
8240
0
120-2
80
h
8220
-1
90
w
1
2
1
CO 2(ppmv)
110
(e)
0
Figure 6. Anvil penetration I from 17:03 to 17:10 UTC. (a) Composite of radar reflectivity at 3° elevation
: IC flash)
(around 7 km height at 120 km range), lightning events ( : negative CG; ↓: positive CG flash;
during 15 min prior to the radar scan and track of the aircraft in 1 min intervals (marked by +). (b): vertical
velocity w and height h of the aircraft; (c): O3 and CO2 mixing ratios; (d): NO, NO2, NOx mixing ratios and the
NR ratio; (e): NOx mixing ratio from measurement, calculated NOx mixing ratio resulting from transport TNOx
and from lightning LNOx.
line system associated with the cold front passage. It
happened in an early stage of development. A more
intense and widespread system did follow the first
line from the west. A cell complex associated with
strong lightning activity did develop a broad and
elongated anvil cloud which could be penetrated
several times at varying heights by the aircraft.
The storm was a very active lightning source as
demonstrated by Figure 5. A rapid increase in flash
frequency was observed at 16:30 UTC. Flash rate
remained at a high level but was varying in intensity
with 2 bursts of flashes at 16:45 and 17:15 UTC.
The aircraft penetrated the storm several times
between 17:00 and 17:45 UTC during the second
intense lightning period. Four penetrations will be
discussed more closely in the following. The chemical
measurements illustrate different stages of the anvil
development. Moreover, they represent different parts
of the anvil as they were performed at different
distances to the main storm center and, therefore, to
the main lightning activity.
5.2.1
Penetration I
During penetration I (Figure 6) the cloud anvil was
still relatively small and in its growing phase. This is
indicated by the radar as well as by the satellite data.
Figure 6a shows the different kinds of flashes
registered by the LPATS during a 15 min interval
starting 15 min prior to the time of the radar
measurement. Even if most of the lightning flashes
occurred within the precipitation field, yet, due to the
storm movement toward the east, a large number of
flashes, shown in Figure 6, are located outside the
reflectivity contours.
The aircraft was just penetrating the easternmost
outflow region of the anvil at about 8 km height.
Upon entering the cloud, [NO] increased rapidly up to
0.6 ppbv whereas [NO2] did not exceed 0.1 ppbv thus
causing a high value of NR at the outermost parts of
the anvil (see Figure 6d). This feature is persistent for
about 10 km along the flight track thus it is not
indicative of a fresh NOx source. This can be,
perhaps, attributed to radiative effects causing an
9
110
1
110
10400
10000
17:20
90
90
17:21
80
9600
-1
9200
8800
-2
w
h
356
-3
80
Height (m)
100
100
w (m/s), q (g/kg)
0
8400
8000
CO2
17:23
60
60
352
350
348
3
346
50
40
17:24
50
40
17:25
17:26
30
30
1
NO
NOx
2.5
NO2
NO/NOx
2
0.6
1.5
0.4
1
0.2
0.5
0
3
10
10
Distance West of Radar
0
NOx
TNOx
LNOx
2.5
Mixing Ratio (ppbv)
20
20
0.8
NO/NO x
70
CO 2(ppmv)
17:22
70
Mixing Ratio (ppbv)
Distance North of Radar
354
2
1.5
1
0.5
0
Figure 7. Anvil penetration II from 17:20 to 17:27 UTC. (a) Composite of radar reflectivity at 4.5° elevation
(around 8 km height at 100 km range), lightning events ( : negative CG; ↓: positive CG flash;
: IC flash)
during 15 min prior to the radar scan and track of the aircraft in 1 min intervals (marked by +), (b): vertical
velocity w and height h of the aircraft; (c): CO2 mixing ratio; (d): NO, NO2, NOx mixing ratios and the
NO/NOx ratio; (e): NOx mixing ratio from measurement, calculated NOx mixing ratio resulting from transport
TNOx and from lightning LNOx.
enhanced photolysis rate in regions of a relatively thin
anvil cloud.
An assessment of the amount of NOx originating
from lightning is also given in Figure 6e. As
determined by the amount of CO2 measured in the
anvil about 1/2 of the total NOx is made up of LNOx
in most parts of the penetration. An exception is the
spike encountered in the later stages of the
penetration. It is made up by 2 NO data points
exceeding the 1.5 ppbv level which corresponds to a
peak width of roughly 400 m. NO contributes most of
the total NOx whereas NO2 has a distinct minimum.
According to the tracer analysis, about 3/4 of the total
NOx is made up by LNOx for this peak. Therefore it
is concluded that a relatively recent flash was
responsible for this peak event.
A further indication towards a lightning induced
source is the lack of any distinct updraft structure
associated with the spike as can be seen from the
aircraft measurements. Moreover, the radar data do
not show growing cells in this part of the storms
(even though most of the flight path is not covered by
radar observations).
There was no lightning flash recorded in the
LPATS data at a favorable time and location in order
to be detected by the aircraft. Due to the limited
detection efficiency of the LPATS system (about 70%
for CG and only 1% for IC lightning) some fraction
of the lightning flashes remains undetected. Hence,
the flash which produced the NOx spike was either
not recorded by the LPATS or it was an intra-cloud
discharge which is not monitored by the system.
Moreover, the exact position of a flash as a 3dimensional event cannot be discovered by the
system. The LPATS is designed to locate the striking
point of the flash at the ground, i.e. the lowermost
part of the lightning bolt that is mostly vertically
oriented.
5.2.2
Penetration II
15 minutes later the aircraft made a second
10
110
3
110
10120
h
w
2
10080
90
0
10040
-1
10000
Height (m)
100
90
1
w (m/s), q (g/kg)
100
-2
-3
354
9960
80
17:29
70
60
70
352
CO 2(ppmv)
17:30
17:31
350
348
5
17:32
60
4
17:34
40
50
40
0.6
2
0.4
1
0.2
4
Distance West of Radar
Mixing Ratio (ppbv)
30
0.8
3
5
0
17:35
30
NO2
NO/NOx
x
17:33
50
1
NO
NOx
3
NO/NO
80
Mixing Ratio (ppbv)
Distance North of Radar
CO2
0
NOx
TNOx
LNOx
2
1
0
Figure 8. Anvil penetration III from 17:28 to 17:35 UTC. (a) Composite of radar reflectivity at 4.8°
elevation (around 9 km height at 100 km range), lightning events ( : negative CG; ↓: positive CG flash; : IC
flash) during 15 min prior to the radar scan and track of the aircraft in 1 min intervals (marked by +), (b):
vertical velocity w and height h of the aircraft; (c): CO2 mixing ratio; (d): NO, NO2, NOx mixing ratios and the
NO/NOx ratio; (e): NOx mixing ratio from measurement, calculated NOx mixing ratio resulting from transport
TNOx and from lightning LNOx.
penetration of the storm’s anvil (penetration II shown
in Figure 7). The anvil had grown considerably in size
as a consequence of the heavy convective activity
associated with the first flash maximum at 16:45 UTC
(Figure 5). The intensification of the storm was also
documented
by
the
METEOSAT
satellite
observations. During the time between the subsequent
scans 17:00 UTC and 17:30 UTC, the area of the cold
cloud shield (< -50°C) increased from 1000 km² up to
3000 km².
Upon entering the anvil the aircraft encountered a
narrow peak in NO of nearly 3 ppbv. The horizontal
dimensions of 1.6 km (for a mixing ratio in excess of
1.5 ppbv) were larger than for the peak value
registered in penetration I. The NR ratio of 1 indicates
a rather recent source probably due to a lightning
event. The width may be explained either by a larger
turbulent mixing of the NO plume or by the
orientation of the flash relative to the aircraft track. If,
in an extreme case, both would be parallel to each
other a relatively elongated signal would be
measurable even if the flash was rather young.
Unfortunately, there was no direct confirmation of a
lightning event close to the aircraft track from the
LPATS data. As in penetration I, the peak was not
correlated with any updraft signature.
During penetration II, the aircraft made an ascent
within the anvil cloud from about 9 km ASL at the
beginning to 10 km ASL upon leaving the anvil. Thus
the measurements originate from the top region of the
storm. Therefore, a relatively high NR ratio was
observed throughout the traverse increasing slightly
from 0.5 to 0.6. Corresponding to the low CO2 values
inside the cloud an LNOx contribution is computed
which amounts to about 1/2 of the total NOx in
regions of high mixing ratios.
5.2.3
Penetration III
Penetration III was made 3 min later (Figure 8).
NOx mixing ratios are generally higher (up to 3 ppbv
with peak values of 4.3 ppbv) than in the previous
11
110
90
80
80
70
70
17:39
17:40
17:38
60
60
17:41
50
50
17:37
40
40
17:36
17:35
30
30
(a)
20
100
90
0
15
80
30
45
70
60
50
40
30
Distance West of Radar
23 July 1996
Elevation 4.5
Reflectivity (dBZ)
20
10
20
17:37
UTC
w (m/s)
Distance North of Radar
90
CO 2(ppmv)
100
8
4
356
Mixing Ratio (ppbv) Mixing Ratio (ppbv)
100
12
w
h
(b)
12
8
4
0
0
CO2
Height h (km)
110
352
348
1
0.8
0.6
0.4
0.2
0
(c)
NO/NOx
(d)
NO
NO2
NOx
3
2
1
(e)
04
NOx
TNOx
LNOx
3
2
1
(f)
17:28
17:29
17:30
17:31
17:32
17:33
17:34
UTC
Figure 9. Anvil penetration IV from 17:35 to 17:42 UTC. (a) Composite of radar reflectivity at 4.5°
elevation (around 8 km height at 100 km range), lightning events ( : negative CG; ↓: positive CG flash; : IC
flash) during 15 min prior to the radar scan and track of the aircraft in 1 min intervals (marked by +), (b):
vertical velocity w and height h of the aircraft; (c): CO2 mixing ratio; (d): NO/NOx ratio, (e): NO, NO2, NOx
mixing ratios and; (f): NOx mixing ratio from measurement, calculated NOx mixing ratio resulting from
transport TNOx and from lightning LNOx.
penetrations and the cloud anvil has grown further in
size. This is consistent with the occurrence of the
second maximum of lightning activity about 15 min
earlier. From the radar scan it can be recognized that
the storm complex was made up of 4 major cells: two
cells at the western boundary which are connected
with a high lightning activity and two weaker cells in
the NE part of the system which produced few
flashes. The latter cells were embedded in the anvil
cloud mass originating from the westmost cells. The
aircraft passed over these weaker cells and penetrated
the top of the southern one. Updraft velocity just
exceeded 2 m/s.
The passage of these cells is coinciding with two
distinct peaks in NOx mixing ratio which had quite a
different chemical composition. The first peak did
contain mostly NO indicating LNOx emission by
lightning. It was observed in a region of weak
downdraft just before entering the cell. It was located
between the cell penetrated a few seconds later and a
weaker cumulus tower growing at the SE side of the
complex. Thus it may be speculated that the NOx
peak was caused by a cloud-to-cloud discharge which
occurred between these two towers. There was no
confirmation of a discharge from the LPATS system,
possibly due to the low detection efficiency with
regard to IC (intra-cloud) or CC (cloud to cloud)
lightning.
The NO mixing ratio in the second peak was
considerably less and the NR ratio was similar to that
observed in the other parts of the cloud (0.5 to 0.6).
This peak was connected to a region of weak updraft
corresponding to the top of a storm cell as described
above. The LNOx amount is about the same for both
peaks (around 3 ppbv; see Figure 8e). These
measurements indicate that NOx, originating either
from lightning sources or from the boundary layer, is
accumulated in the top of the cell. The NOx was not
yet distributed throughout the anvil. The LPATS
network did register a lightning discharge in this cell
about 8 minutes prior to the aircraft penetration of the
cloud. There is no direct confirmation for the first
peak in the LPATS data.
As a result of the increased lightning activity of the
12
storm, not only a relatively high NOx level but also a
high LNOx level can be noted from Figure 7e. LNOx
mixing ratio generally exceeds the transported
component (by a factor of 2).
5.2.4
Penetration IV
During penetration IV (Figure 9) the aircraft just
entered the edge of the main storm core. Updraft
velocity up to 13 m/s was encountered. While
performing the flight pattern, the aircraft was
descending from its initial height of 10 km to about 7
km at the end of the flight track shown in Figure 9.
This change in height does explain the slow decrease
of the CO2 concentration between 17:36 and 17:39
UTC. When entering the strong updraft, [CO2]
dropped markedly down to 346 ppmv, a value slightly
lower than found from the vertical profile taken
during the pre-storm aircraft ascent (Figure 2a). This
air-mass probably originates from the very lowest
layers close to the ground (where no CO2
measurements are available) and was lifted within the
storm’s core.
The NO registration shows a very spiky structure.
Narrow NO peaks are superimposed on a low
background concentration. But in these gaps, NO2 can
be found in appreciable amounts thus causing the
NOx curve to be less noisy. The NO peaks in excess
of 1.5 ppbv consist of only one measurement,
whereas the updraft or the CO2 structure exhibits
much broader extremes. This is indicative of lightning
produced NO not older than about 3 minutes.
The occurence of 3 of the 4 major NO peaks can be
traced back to individual lightning flashes observed
by the LPATS (see Figure 10). Considering the
advection of the air in the lightning bolt, using the
aircraft observed wind velocities, 3 multiple-stroke
events are most likely to represent the origin of the
peaks. All flashes were negative cloud-to-ground
flashes. The major peak is attributed to a 2-fold
stroke, the following ones by a 3 and a 5-fold stroke,
respectively. Not all of the NO peaks could be
attributed to flashes observed by the LPATS. This
especially holds at the beginning of penetration IV.
Due to the lower resolution of the NO2 (NOc)
measurements as compared to the NO data, a realistic
assessment of the NOx amount within the narrow
peaks is not possible. The response times are 1 s for
the NO detector and 5 s for the NOc converter [see
Schulte et al., 1997]. Thus, in an NO maximum, the
NO2 mixing ratio as computed from Equation (1) will
be very low due to the retarded response of the
converter. Therefore, NO maxima cannot be expected
to coincide with NO2 minima as suggested by Figure
9.
6
Simulation of LNOX Fields
In order to substantiate the results discussed above
a simple two-dimensional (x,y) model is applied for
simulating the emission and the subsequent horizontal
transport and diffusion of the NOx produced by
individual lightning flashes. The computed
concentration field is then compared to the airborne
measurements.
The model assumes that each flash locally releases
a constant number of NO molecules. The resulting
NOx concentration is advected within a constant twodimensional wind field which is inferred from the
actual airborne wind measurements. Furthermore, the
original NOx plume caused by the individual flashes
is broadened due to turbulent diffusion along its path.
In contrast to Equation (7), which refers to young
emissions not older than about 10 min, the spreading
of the plume in the later stages is assumed to be
proportional to the square root of the NOx transport
time.
The concentration of LNOx at time t and location r
is calculated as the accumulation of all prior
emissions
c(r , t ) = ∑ ci (r , t )
(13)
i
where the concentration contribution from the i’th
flash, ci , is simulated by a Gaussian distribution
shifted with constant linear drift vector v and the
time dependent standard deviation
with diffusion parameter α:
[α ⋅ (t − t )]
 [(r − r ) − v ⋅ (t − t )]2 
i
i
ci (r , t ) = c0 exp −

2α ⋅ (t − ti )
î

for t > ti
1/ 2
i
(14)
The square in the argument of the exponential
13
4.0
4
LNO x Measurement
2
1
0
17:37
17:38
17:39
17:40
17:41
UTC
LNOx Simulation
3.0
2.0
1.0
(a) 0.0
80
Distance North of Radar (km)
Mixing Ratio [ppbV]
NO (ppbv)
3
17:20
17:25
17:30
17:35
UTC
17:40
49.0°
17:20
70
1.5
17:40
0.1
90
80
70
60
Distance East of Radar (km)
0
Reflectivity (dBZ)
15
30
45
positive CG
negative CG
IC
23. July 1996
Elevation 4.5°
Figure 10. Relating individual flashes (lower panel)
to the NO maxima (upper panel) during penetration
IV. Flashes are categorized (group 1 to 4) according
to the time of their occurrence. Arrows show the
advection
(assessed
from
aircraft
wind
measurements) during the time necessary to intersect
the aircraft track.
function represents the dot product of the term in
squared brackets. ri and ti are location and time of the
i’th flash as detected by the LPATS.
Since the altitude of the aircraft was changing
during the three anvil crossings, the drift velocity v
was chosen according to the mean wind in the
corresponding flight level (speed 30 ± 10 m/s,
direction 270°±15°). The parameter α was set as
constant in a model consistent way in order to
guarantee the proper dimension of the LNOx plume as
derived from comparison with the airborne
observations. Effects of vertical motions as well as
details of the lightning flash geometry are not
considered by the present model. The simulation is
aimed at finding a set of reasonable parameter values
for the simulation that fit the measured concentration
time series in a satisfactory way.
The result of the simulation is an LNOx
17:25
17:35
30 km
(b)
Falcon
track
1.0
0.5
48.5°
60
17:30
10°
11°
12°
Figure 11. 2D advection-dispersion simulation of
LNOx inferred from the LPATS observations: (a)
time series of simulated LNOx mixing ratio compared
to the LNOx determined from the in-situ
measurements; (b) simulated LNOx mixing ratio field
(in ppbv) for 17:41 UTC. The solid line indicates the
flight path of the FALCON aircraft.
concentration field containing all the accumulated
contributions from the cloud’s lightning history. This
concentration field is plotted in Figure 11 for 17:41
UTC. The area of enhanced LNOx concentration
corresponds to the anvil cloud which extends
eastward from the active cloud core. Recent lightning
in the western part produces local concentration
maxima.
In order to compare the simulated model data with
the airborne observations presented in the previous
section, the simulated NOx field is sampled according
to the FALCON flight track. The sampling is done
from the evolving NOx plume during the 20 min
interval shown in Figure 11. The resulting time series
is displayed in Figure 11a and compared to the LNOx
data derived from the measurements.
The overall structures of the individual anvil
penetrations are well represented by the LNOx plume.
Spatial sizes of the regions with high LNOx
concentration agree well, but the concentration
change at the boundaries of the anvil is smoother in
the simulated field. This field is generally less spiky,
14
but the difference between the characteristic features
of the last two anvil penetrations is reproduced. The
time series for the last penetration yields strong
fluctuations due the close vicinity to the storm core
where lightning occurred closely in time.
The disagreements between observation and
simulation are due to the simplifying assumptions
used in model. The diffusion model applied leads to
an overestimation of dilution at the boundaries. A
more realistic simulation should apply a time
dependent diffusion law or utilize the measured
turbulence in the cloud [Meischner et al., 1997].
Details of the cloud shape were not modelled, as they
depend on the 2D variability of the wind and
turbulence fields which has been neglected for the
present assessment.
As the aircraft descended inside the storm during
penetration IV, the structural differences noted here
do also arise from the three-dimensionality of the
transport processes. Penetrations II and III were
performed in the anvil and thus the cumulative effect
of the flashes in all lower levels is included in the
NOx concentrations measured. This also holds for the
simulated fields which are two-dimensional and
therefore cannot account for the vertical flash
structure. Thus the largest differences can be expected
for penetration IV, as is noted from Figure 11.
The LNOx emission per flash (c0) which gave the
best agreement with the measurement was c0 = 0.32
g[N]m-1 per flash. The comparison was done in terms
of averages taken over the cloud penetrations shown
in Figure 11, except for the poorly simulated
structures between 17:36 and 17:37 UTC. Comparing
the total LNOx amounts as determined from the
measurements with the values calculated from the
flash positions as observed by the LPATS implies that
the emission of NOx by each flash has to be
interpreted as an ’effective’ emission including
processes like (i) NOx production from different types
of flashes (CG or IC), (ii) different frequency of
occurrence of these flash types, or (iii) the
distribution of the LNOx among updrafts and
downdrafts in the storm. A simple break down of the
effective LNOx production into these different
contributions can be done in order to obtain a
production rate due to a CG flash. Assuming that
there are 3 IC flashes occurring per CG flash [e.g.
Liaw et al., 1990], producing NOx 10-times less
efficiently than a CG flash, results in a 30% increase
of NOx production per LPATS flash. Furthermore, it
is assumed that the IC-produced NOx is completely
transported into the anvil while 50% of the CGproduced NOx encounters a downdraft and thus can
not be detected in the anvil (if re-circulations are not
considered). Further reduction of the flash-related
production rate results from the limited detection
efficiency of the LPATS for CG flashes (around
70%). These reductions roughly balance the increase
which was due to the explicit consideration of the IC
flashes. Thus it can be concluded that the ’effective’
NOx production rate c0 is also representing the NOx
production of a CG flash.
The accuracy of the LNOx determination described
in this paper can only be assessed roughly from the
errors inherent in the measurements. The error in the
LNOx concentration is dominated by the errors in the
low-level NOx concentration which was assumed to
range around 50% ([NOx,1] = 1.0 ± 0.5 ppbv) as
discussed in section 4. The upper level NOx
concentration and the CO2 values as discussed in
section 3 have minor influence on the LNOx values as
computed from Equation (6). It turns out that the
resulting error in LNOx is also about 50%. As the
assumptions made in the mixing model are not
perfectly fulfilled, this assessment should be looked
upon as a lower limit of the error in the LNOx and c0values.
The c0-value obtained above is in reasonable
agreement with production rates reported in the
literature. For reasons of comparison with previous
studies one can extrapolate this production rate to the
global scale. Using a global lightning frequency of
100 s-1 [e.g. Lawrence et al., 1995] and a mean flash
length of 5000 m yields an annual global production
of 5 Tg[N]. This result is within the limits of
previously reported values typically ranging from 2
TgN yr-1 [Lawrence et al., 1994] to 12 TgN yr-1
[Price et al., 1997]. It points towards the lower limit
of the range.
This simulation demonstrates the possibility to
derive the observed LNOx fields from the detected
lightning flashes using reasonable assumptions of perflash production and transport from the emission
source. The results from the model calculations give
confidence to the interpretation of the aircraft data as
discussed above. It was demonstrated that the
observed flash distribution could generate the general
structure of the observed fields.
15
7
Summary and Conclusions
The present study investigates the production of
NOx by lightning. A field experiment was performed
during July 1996 in southern Germany. Airborne
measurements of NO, NO2, CO2, and O3 were
performed in thunderstorm anvils. The radar structure
as well as the lightning activity of the storms were
also observed.
Typically, enhanced NOx contents (compared to
the cloud environment) were found in the storm anvils
as well as in smaller cumulus congestus clouds. The
CO2 mixing ratio decreased inside the clouds. This
behavior could be attributed to the transport of low
level air (high NOx and low CO2 content) originating
from the boundary layer and to the additional
production of NOx by lightning. These two main
source mechanisms for the introduction of NOx into
storm anvils (and subsequently into the upper
troposphere) could be discriminated by means of a
tracer (CO2) for low level air. The total LNOx amount
in the large storm system was assessed to be about
equal to the amount caused by transport from the
boundary layer during the initial stages of the
development. Later on, when the anvil had increased
in size, the lightning produced NOx amount exceeded
the contribution due to transport by a factor of 2.
In a small (young) isolated storm cell, LNOx was
also found to contribute about the same order of
magnitude to the total NOx amount as was inferred
for the transport part TNOx. A large NR ratio was
found at the cloud edge which was illuminated by the
sun. Even though such high ratios are also typical for
fresh lightning sources, the large spatial extent of the
region also supports the hypothesis of a high
photolysis rate of NO2 causing this phenomenon.
Similarly high ratios NR were also found at the
edges and in the upper parts of the anvil of a large
storm system. Moreover, narrow NO peaks were
found, some of which could be attributed to
individual lightning strokes as registered by the
LPATS network. Considering the turbulent diffusion
in the cloud as well as the NR values, the sources
were not older than about 2-4 minutes. Further mixing
with the anvil air and the superposition of different
sources caused the structure observed during the insitu measurements. Most of the peaks in the NOx
measurements, however, could not be assigned to
lightning flashes located by the LPATS, probably due
to its limited detection efficiency. In the present study
the LPATS was mainly utilized for monitoring the CG
lightning activity of the storms.
A simple two-dimensional advection-dispersion
model could give confidence into the interpretation of
the airborne measurements. The overall structures of
the individual anvil penetrations are well represented
by the simulated LNOx plume. Details of the cloud
shape cannot be represented as they depend on the 2D
variability of the wind and turbulence fields which
have been neglected for the present assessment. But
occasionally, even single LNOx peaks could be
reproduced by the simulation, provided that suitable
advective conditions were given.
Best agreement with the observed values was found
for NO production rates of about 0.3 g per meter of
flash length. In order to compare the results of the
present study with previous investigations, the
production rate can be expressed in terms of a global
production rate. Applying a widely used assumption
of L=5 km for the mean flash length L and a mean
global flash frequency of 100 per second results in a
global emission of 5·1012 g N per year. This result is
within the limits of previously reported values.
The results presented here are not necessarily
representative for other storms, especially in different
geographical regions. Further case studies should be
performed in order to judge the representativeness of
the present case study results. The extrapolation to
global scale was only done for reasons of comparison
with previous investigations. Huntrieser et al. [1998]
recently investigated the storms observed during
LINOX with respect to LNOx production in a
statistical sense.
A further point to be considered in future studies is
the role of intra-cloud lightning in comparison with
cloud-to-ground lightning which has nearly
exclusively been available from the LPATS system
for the present study. It was suggested by the
measurement presented above that at least some of the
young and narrow NOx peaks measured in the anvil
might have originated from IC flashes. This gives
indication that the IC component might be also
important for NOx production, a contribution which
has previously been assumed as being of minor
importance. This point needs further clarification
which can be achieved from more detailed lightning
measurements covering the three-dimensional flash
16
structure as well as the IC flash history.
Acknowledgments. We gratefully acknowledge
‘Bayernwerke AG and Badenwerke AG’ for making
the lightning data available. Special thanks are due to
the flight facility of DLR for performing the aircraft
operations. Analysis of the data was part of the
EULINOX project sponsored by the Commission of
the European Communities (CEC), Contract No.
ENV4-CT97-0409.
References
Beck, J. P., C.E. Reeves, F.A.A.M. de Leeuw, and
S.A. Penkett, The effect of air traffic emissions on
tropospheric ozone in the northern hemisphere,
Atmos. Environ., 26A, 17-29, 1992.
Bent, R. B., and W. A. Lyons, Theoretical evaluations
and initial operational experiences of LPATS
(lightning position and tracking system) to monitor
lightning ground strikes using a time-of-arrival
(TOA) technique, VII’th International Conference
on Atmospheric Electricity, June 3-8, Albany,
N.Y., 317-328, 1984.
Blackadar, A.K., Turbulence and Diffusion in the
Atmosphere, 185pp, Springer, Berlin, 1997.
Brasseur, G.P., J.-F. Müller, and C. Granier,
Atmospheric impact of NOx emissions by subsonic
aircraft: A three-dimensional model study, J.
Geophys. Res., 101, 1423-1428, 1996.
Dickerson, R.R., G.J. Huffman, W.T. Luke, L.J.
Nunnermacker, K.E. Pickering, A.C.D. Leslie, C.G.
Lindsey, W.G.N. Slinn, T.J. Kelly, P.H. Daum,
A.C. Delany, J.P. Greenberg, P.R. Zimmerman, J.F.
Boatman, J.D. Ray, and D.H. Stedman,
Thunderstorms: An important mechanism in the
transport of air pollutants, Science, 235, 460-464,
1987.
Ehhalt, D.H., and F. Rohrer, The impact of
commercial aircraft on tropospheric ozone. The
Chemistry of the Atmosphere - Oxidants and
Oxidation in the Earth's Atmosphere. 7th BOC
Priestley Conference, Lewisburg, Pennsylvania,
1994 [edited by Bandy A. R.], The Royal Society
of Chemistry, Special Publication No. 170, 105120, 1995.
Hill, R. D., R. G. Rinker, and H. D. Wilson,
Atmospheric nitrogen fixation by lightning, J.
Atmos. Sci., 37, 179-192, 1980.
Höller, H, V.N. Bringi, J. Hubbert, M. Hagen, and
P.F. Meischner, Life cycle and precipitation
formation in a hybrid-type hailstorm revealed by
polarimetric and Doppler radar measurements, J.
Atmos. Sci., 51, 2500-2522, 1994.
Huntrieser, H., H. Schlager, P. Van Velthoven, P.
Schulte, H. Ziereis, U. Schumann, F. Arnold, and J.
Ovarlez, In-situ trace gas observations in
dissipating thunderclouds during POLINAT, 12th
Int. Conf. on Clouds and Precipitation, Zurich,
Switzerland, 19-23 Aug. 1996, 1058-1061, 1996.
Huntrieser, H., H. Schlager, C. Feigl, and H. Höller,
Transport and production of NOx in electrified
thunderstorms: Survey of previous studies and new
observations at midlatitudes, J. Geophys. Res., 103,
28,247-28,264, 1998.
Kelley, P., R. R. Dickerson, W. T. Luke, and G. L.
Kok, Rate of NO2 photolysis from the surface to
7.6 km altitude in clear-sky and clouds, Geophys.
Res. Letters,22, 2621-2624, 1995.
Lawrence, M. G., W. L. Chameides, P. S. Kasibhatla,
H. Levy, and W. Moxim, Lightning and
Atmospheric Chemistry: The Rate of Atmospheric
NO Production, in: Handbook of Atmospheric
Electrodynamics, CRC Press, 189-202, 1994.
Levy, II, H., W.J. Moxim, and P.S. Kasibhatla, A
global three-dimensional time-dependent lightning
source of tropospheric NOx, J. Geophys. Res., 101,
22,911-22,922, 1996.
Liaw, Y.P., D.L. Sisterson, and N.L. Miller,
Comparison of field, laboratory and theoretical
estimates of global nitrogen fixation by lightning,
J. Geophys. Res., 95, 22,489-22,494, 1990.
Luke, W.T., R.R. Dickerson, W.F. Ryan, K.E.
Pickering, and L.J. Nunnermacker, Tropospheric
chemistry over the lower great plains of the United
States. 2. Trace gas profiles and distributions, J.
Geophys. Res., 97, 20,647-20,670, 1992.
Meischner, P.F., T. Jank, and R. Baumann, Eddy
dissipation rates in thunderstorms estimated by
Doppler radar compared with aircraft in situ
measurements, 28th Conference on Radar
Meteorology, Austin, Texas, 7-12 September 1997,
438-439, 1997.
Price, C., J. Penner, and M. Prather, NOx from
Lightning, 1. Global distribution based on lightning
physics, J. Geophys. Res., 102, 5929-5941, 1997.
Ridley, B.A., S. Madronich, R.B. Chatfield, J.G.
Walega, R.E. Shetter, M.A. Carroll, and D.D.
17
Montzka, Measurements and model simulations of
the photostationary state during the Mauna Loa
observatory
photochemistry
experiment:
Implications for radical concentrations and ozone
production and loss rates, J. Geophys. Res., 97,
10,375-10,388, 1992
Ridley, B.A., J.E. Dye, J.G. Walega, J. Zheng, F.E.
Grahek, and W. Rison, On the production of active
nitrogen by thunderstorms over New Mexico, J.
Geophys. Res., 101, 20,985-21,005, 1996.
Schulte, P., H. Schlager, H. Ziereis, U. Schumann, S.
L. Baughcum, and F. Deidewig, NOx emission
indices of subsonic long-range jet aircraft at cruise
altitude: In situ measurements and predictions, J.
Geophys. Res., 102, 21,431-21,442, 1997.
WMO (World Meteorological Organization),
Scientific assessment of ozone depletion: 1994,
Global Ozone Research and Monitoring Project,
Rep. 37, Geneva, 1995.
Ziereis, H., H. Schlager, P. Schulte, I. Köhler, R.
Marquardt, and C. Feigl, In situ measurements of
the NOx distribution and variability over the
eastern North Atlantic, submitted to J. Geophys.
Res., 1999.