Diurnal cycle of air pollution in the Kathmandu Valley,
Nepal: 2. Modeling results
The MIT Faculty has made this article openly available. Please share
how this access benefits you. Your story matters.
Citation
Panday, Arnico K., Ronald G. Prinn, and Christoph Schär.
“Diurnal Cycle of Air Pollution in the Kathmandu Valley, Nepal: 2.
Modeling Results.” Journal of Geophysical Research 114, no.
D21 (2009).
As Published
http://dx.doi.org/10.1029/2008jd009808
Publisher
American Geophysical Union
Version
Final published version
Accessed
Thu May 26 20:40:21 EDT 2016
Citable Link
http://hdl.handle.net/1721.1/85645
Terms of Use
Article is made available in accordance with the publisher's policy
and may be subject to US copyright law. Please refer to the
publisher's site for terms of use.
Detailed Terms
Click
Here
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 114, D21308, doi:10.1029/2008JD009808, 2009
for
Full
Article
Diurnal cycle of air pollution in the Kathmandu Valley, Nepal:
2. Modeling results
Arnico K. Panday,1,2,3 Ronald G. Prinn,4 and Christoph Schär5
Received 10 January 2008; revised 27 May 2009; accepted 10 June 2009; published 6 November 2009.
[1] After completing a 9-month field experiment studying air pollution and meteorology
in the Kathmandu Valley, Nepal, we set up the mesoscale meteorological model MM5
to simulate the Kathmandu Valley’s meteorology with a horizontal resolution of up to
1 km. After testing the model against available data, we used it to address specific
questions to understand the factors that control the observed diurnal cycle of air pollution
in this urban basin in the Himalayas. We studied the dynamics of the basin’s nocturnal
cold air pool, its dissipation in the morning, and the subsequent growth and decay of
the mixed layer over the valley. During mornings, we found behavior common to large
basins, with upslope flows and basin-center subsidence removing the nocturnal cold air
pool. During afternoons the circulation in the Kathmandu Valley exhibited patterns
common to plateaus, with cooler denser air originating over lower regions west of
Kathmandu arriving through mountain passes and spreading across the basin floor,
thereby reducing the mixed layer depth. We also examined the pathways of pollutant
ventilation out of the valley. The bulk of the pollution ventilation takes place during the
afternoon, when strong westerly winds blow in through the western passes of the valley,
and the pollutants are rapidly carried out through passes on the east and south sides of the
valley. In the evening, pollutants first accumulate near the surface, but then are lifted
slightly when katabatic flows converge underneath. The elevated polluted layers are mixed
back down in the morning, contributing to the morning pollution peak. Later in the
morning a fraction of the valley’s pollutants travels up the slopes of the valley rim
mountains before the westerly winds begin.
Citation: Panday, A. K., R. G. Prinn, and C. Schär (2009), Diurnal cycle of air pollution in the Kathmandu Valley, Nepal: 2.
Modeling results, J. Geophys. Res., 114, D21308, doi:10.1029/2008JD009808.
1. Introduction
[2] In this paper we present results of theoretical investigations using a numerical model aimed at addressing
questions about processes that control air quality in the
Kathmandu Valley, Nepal. These questions arose from the
analysis of data from a field experiment that we conducted
in 2004 – 2005 [Panday and Prinn, 2009]. The Kathmandu
Valley is a bowl shaped basin in central Nepal, halfway
between the Ganges Plains and the Tibetan plateau. It has a
flat floor area of 340 km2, and it is located 1300 m above
sea level (masl), surrounding by mountains reaching 2000 –
1
Program in Atmospheric and Oceanic Sciences, Princeton University,
Princeton, New Jersey, USA.
2
Formerly at Center for Global Change Science, Massachusetts Institute
of Technology, Cambridge, Massachusetts, USA.
3
Now at Department of Environmental Sciences, University of Virginia,
Charlottesville, Virginia, USA.
4
Center for Global Change Science, Massachusetts Institute of
Technology, Cambridge, Massachusetts, USA.
5
Institute for Atmospheric and Climate Science, Swiss Federal Institute
of Technology, Zurich, Switzerland.
Copyright 2009 by the American Geophysical Union.
0148-0227/09/2008JD009808$09.00
2800 masl. There are five passes of 1500– 1500 masl on
the west and east sides, and one winding river outlet (the
Bagmati River) to the south. The valley is inhabited by
close to 3 million people. Its vehicle fleet exceeds
300,000 motorcycles and a hundred thousand other vehicles.
In recent years its growing air pollution problem has caught
increasing attention, especially during winters.
[3] In urban mountain basins, an important effect of the
surrounding topography is to reduce the exchange of valley
air with the free troposphere and thereby to concentrate
pollutants near the basin bottom. Ventilation of basins is
often dominated by complex valley wind systems that are
driven by differential heating and cooling of surrounding
mountain slopes and characterized by a strong diurnal cycle.
At night, basins often fill with pools of cold air. In addition
to local radiative cooling, basin bottoms experience an
additional cooling component due to the arrival of drainage
flows from colder surrounding mountain slopes [Whiteman
et al., 2004a]. Such cold air pools have been studied
extensively in the Columbia Basin [Whiteman et al.,
2001; Zhong et al., 2001] Danube Basin [Zängl, 2005],
Virginia [Allwine and Lamb, 1992], Colorado [Whiteman,
1982; Kelly, 1988; Neff and King, 1989; Whiteman et al.,
1999], Bulgaria [Savov et al., 2002], Slovenia [Rakovec et
D21308
1 of 20
D21308
PANDAY ET AL.: KATHMANDU VALLEY MODELING RESULTS
D21308
Figure 1. Kathmandu Valley, viewed from the south in Google Earth (printed with permission). The
high Himalayas and Tibetan Plateau are visible at the top edge of the image. Kathmandu city is labeled in
red. Yellow labels show important measurement sites.
al., 2002] and Japan [Kondo et al., 1989]. Cold pools in
small basins and sinkholes in the Rockies and in Austria
have been chosen for the extensive field observations
designed to understand their processes [Clements et al.,
2003; Whiteman et al., 2004a, 2004b]. Often the top of a
cold air pool is marked by a strong temperature jump
[Whiteman et al., 2001]. Within and immediately above
the cold pools, there is a strong temperature inversion, which
suppresses buoyant vertical transport of pollutants. The
daytime growth of a mixed layer over a basin thus requires
the dissipation of the basin’s cold air pool and its associated
temperature inversion.
[4] In the Kathmandu Valley, we conducted a field experiment between September 2004 and October 2005. A companion paper [Panday and Prinn, 2009], and a dissertation
[Panday, 2006], describe the results in detail. Here we
summarize some salient aspects of the observational campaign and formulate the objectives of the current paper. The
observational set up included a laboratory in Bouddha, east of
Kathmandu city, where we continuously measured carbon
monoxide (CO), ozone, fine particulates (PM10) and weather
variables. In addition, during part of the time we had a mini
sodar operating nearby, and we had four temperature loggers
on a TV tower, as well as additional temperature loggers on a
valley rim peak, at the base of the valley rim, and on a small
peak overlooking the lab site. These sites are shown on
Figure 1.
[5] One of the key results discussed in the companion
paper is the observation of two daily CO peaks in the
Kathmandu Valley. Figure 2 shows the diurnal cycle of
carbon monoxide measurements at the laboratory in Bouddha
in October 2004 and January 2005. We have chosen to show
the maximum and minimum values for each 10-min time
interval, in order to show the values below which CO never
dropped, and the values that were never exceeded. In addition,
Figure 2 shows the median for each 10-min time interval as a
measure of where 50% of the data may be located in a
possibly not normal distribution. The diurnal cycle shows a
distinct peak in the morning, and one in the evening. In part 1
[Panday and Prinn, 2009] we have shown that the two peaks
are not merely driven by the emissions but strongly affected
by the diurnal circulation and mixing in the Kathmandu
Valley. In particular, CO bag sampling on small neighboring
peaks indicated the nighttime lifting of pollutant layers, while
ozone measurements on a tall peak showed the isolation of
these sites from Kathmandu Valley air at night. In the current
paper (part 2) we will use a high-resolution numerical model
to clarify the underlying dynamical mechanisms.
[6] To set the scene for the simulations, Figures 3 and 4
show meteorological observations on six representative,
consecutive days in February 2005. Most days were sunny,
with low wind speeds at night, and strong westerly winds
during the afternoons. Figures 4a and 4b indicate that
saturation occurred on most mornings, and that on several
afternoons the dew point dropped, indicating the arrival of a
different air mass. Figure 4c shows the temperature recorded
on Nagarkot peak (1972 masl), and at Kharipati (1373 masl),
at the base of Nagarkot. While Kharipati was warmer than
Nagarkot during the daytimes, it was colder during most
nights, indicating a temperature inversion. Figure 4d compares the temperature at Kharipati with a valley-center
location. We see that in the evening, the temperature
dropped more rapidly in Kharipati. This is consistent with
the formation of a nocturnal cold air pool in the valley
2 of 20
D21308
PANDAY ET AL.: KATHMANDU VALLEY MODELING RESULTS
D21308
Figure 2. Diurnal cycle of CO at the main laboratory in Bouddha during the months of (a) October 2004
and (b) January 2005, averaged over 10-min intervals. The centerline represents the monthly median value
for each 10-min interval. Vertical bars show the highest and lowest 10-min averages recorded that month for
each time interval. The x axis is in Nepal Standard Time, which is UTC + 5 h 45 min.
through the arrival of katabatic winds from the valley rim
slopes. Figure 5a shows the lapse rate computed from the
temperature difference between Nagarkot and Kharipati. We
see stable conditions at night, and a midday lapse rate near
the 9.8 K/km dry adiabatic lapse rate, indicating a well-
mixed layer between the valley bottom and the altitude of
Nagarkot. Figures 5b and 5c show the lapse rate calculated
using temperatures at different levels on the TV tower. We
see stable conditions down to the lowest levels at night
(consistent with a cold air pool in the valley), and strong
Figure 3. (a) Solar radiation, (b) wind speed, and (c) wind direction measured by the weather station on
the Hyatt roof in Bouddha during a typical 6-day period in spring 2005. The period considered contains a
weekend (26 – 27 February).
3 of 20
D21308
PANDAY ET AL.: KATHMANDU VALLEY MODELING RESULTS
Figure 4. (a) Relative humidity at the Hyatt roof in Bouddha during the same 6 days as in Figure 2,
(b) temperature (black) and dew point (green) on the Hyatt roof during the same time, (c) temperature on
Nagarkot Peak (red) and Kharipati at the base of Nagarkot (blue), and (d) temperature at Kharipati (blue)
and at 5 m on the TV tower (black).
Figure 5. Lapse rates (K/km) estimate between different measurement sites: (a) between Kharipati and
Nagarkot, (b) between the bottom two loggers on the TV tower, and (c) between the 5-m and the 48-m
logger on the TV tower.
4 of 20
D21308
D21308
PANDAY ET AL.: KATHMANDU VALLEY MODELING RESULTS
thermals at midday that shrink with height. Sodar data were
not available in February, but sodar-derived mixed layer
heights in May showed that the mixed layer peaked at
midday at a depth of 400 to 800 m (see part 1). These values
are surprisingly low. It is possible that the sodar’s range
prevented us from seeing the full height of the mixed layer.
However, the fact that sodar data showed a shrinking mixed
layer height over the course of the afternoon suggests that
perhaps other processes were at play.
[7] The observational results are discussed in more detail
in part 1 [Panday and Prinn, 2009]. It explained the
observed daily twin peak pattern of CO in terms of diurnal
variations in emissions and ventilation. The morning peak
decreased at the same time that the nocturnal cold air pool
had dissipated, and strong westerly winds removed pollutants that had accumulated in the city over the course of the
morning. The evening peak began when evening rush hour
took place after rapid ventilation ceased. Parts of the
evening emissions were found to be lifted and recirculated
the next morning, adding to the morning pollution peak. Our
observations left some important questions unanswered,
however. They did not give us a complete picture of the
dynamics of the nocturnal cold air pool and the daytime
mixed layer, nor did they indicate the pathways of pollutant
ventilation out of the valley. In this paper we use numerical
modeling to address these questions, using Pennsylvania
State University (PSU)/National Center for Atmospheric
Research (NCAR) mesoscale meteorological model (also
known as MM5). We will use observations to validate the
numerical model results, and then turn attention to two
specific scientific issues. These concern the dissipation of
the nocturnal cold-air pool and its replacement by a mixed
layer in the Kathmandu valley, and the analysis of air-parcel
trajectories to elucidate the fate of polluted valley air.
[8] Section 2 describes the model set up. Section 3
compares model results with observations. Section 4 discusses possible mechanisms for the dissipation of nocturnal
cold air pools and mixed layer growth in mountain basins,
and shows modeling results for Kathmandu. Section 5
examines pathways of pollutant ventilation out of the valley
during different times of day. The study is concluded in
section 6.
2. Numerical Modeling
[9] We set up version 3.6.1 of the mesoscale meteorological model MM5 for use around the Kathmandu Valley.
Initial and boundary conditions were taken from global final
analyses (FNL), prepared every 6 h at 1° resolution by the
National Center for Environmental Prediction (NCEP), and
available online through the National Center for Atmospheric
Research (NCAR) Data Support Section webpage (http://
dss.ucar.edu/datasets/ds083.2/). The modeling domain was
set up using a Lambert conformal projection, with four
levels of two-way nesting, stepping down in resolution from
one degree NCEP FNL data at the side boundaries while
maintaining a 1:3 grid resolution ratio. The four nested
domains had horizontal grid resolutions of 27, 9, 3, and 1 km
and were all centered at 27.7°N, 85.3°E, corresponding to a
point in central Kathmandu City, close to the Dharahara
tower (Figure 6). The outer two domains had 50 by 50
horizontal grid squares, while the inner two domains had 75
D21308
by 75 grid squares. The outermost domain spanned much of
the Ganges Basin and a big part of the Tibetan Plateau,
while the innermost domain spanned approximately three
times the horizontal extent of the Kathmandu Valley, reaching into neighboring valleys.
[10] Land use and vegetation for the simulations were
taken from the public USGS 25-category 30 arc-second
database downloadable via the MM5 homepage (ftp://ftp.
ucar.edu/mesouser/MM5V3/TERRAIN_DATA/), while underlying topography was taken from the USGS GTOPO-30
(global 30 arc-second) database and smoothed appropriately
for each nest level. At the 1-km grid resolution of the
innermost domain, the model topography was able to capture
essential features of the Kathmandu Valley, including the
surrounding mountains, passes, and relatively flat basin
bottom. It did not capture small-scale features within the
valley bottom, such as Chobhar hill, Chobhar Gorge,
Swayambhu hill, the Pullahari-Kapan ridge, Changu Narayan
hill or any of the river channels. MM5 is not designed for use
at resolutions less than 1 km without modifications to the
code [Gohm et al., 2004].
[11] Our model runs were initialized with NCEP analyses
at 0600 UTC, corresponding to 1145 Nepal Standard Time.
Initialization at midday rather than midnight was chosen
because mountain valleys often contain stably stratified air
masses at night, which are disconnected from overlying
regional flows and poorly represented in the synoptic-scale
analysis that initialize and drive the model. The model runs
were allowed to spin up for 36 h before the results were
used. Experimental runs with various vertical sigma coordinate configurations led to a final choice of 40 vertical
layers extending from the surface to 100 mbar. The five
bottommost layers were each 15 m thick (corresponding to
0.006s), while above them the layer thickness was gradually increased. At the top of the model domain, boundary
conditions were chosen to allow radiation of internal gravity
waves [Klemp and Durran, 1983]. Simulation time steps
were 60 s in the outermost domain with 27 km resolution,
necessitating a 1 s time step in the innermost domain. The
lateral boundaries of the outermost domain used linear
interpolation between the 6-hourly time steps of the NCEP
analysis, with a relaxation extending four grid cells into the
domain.
[12] Cloud microphysics was parameterized using the
Simple Ice Scheme [Dudhia, 1989]. This uses explicit
microphysics calculations to simulate cloud and rainwater
as well as cloud ice. MM5 was allowed to calculate cumulus
clouds explicitly in the inner two domains (3 km and 1 km
resolution), while in the coarser resolution outer domains
cumulus clouds were parameterized using the updated KainFritsch scheme [Kain, 2004]. The boundary layer was
parameterized using the MRF scheme [Hong and Pan,
1996], which is suitable for fine resolution runs while
maintaining computational efficiency. Solar and terrestrial
radiation were calculated every 30 min, using a parameterization that included long- and short-wave radiation interacting with clear air, clouds, precipitation and the ground
[Dudhia, 1989]. For the surface energy budget, a five-layer
soil model was used [Dudhia, 1993], with soil moisture
initialized by the driving analysis.
[13] We carried out simulations for February and May
2005, and used the MM5-customized interpolation and
5 of 20
PANDAY ET AL.: KATHMANDU VALLEY MODELING RESULTS
D21308
D21308
Figure 6. The four nested domains used in our MM5 simulations. Domain 4 (labeled) is centered on the
Kathmandu Valley.
plotting package RIP4 (ftp://ftp.ucar.edu/mesouser/MM5V3/
RIP4.TAR.gz) to generate plots and to calculate forward and
backward trajectories. The trajectories were calculated at
1-min intervals using horizontal and vertical vectors output
by MM5 every half hour.
3. Model Validation
[14] Next we look at MM5’s ability to reproduce the
Kathmandu Valley’s meteorology. To this end we use a
simulation initialized on 8 February 2005, 0600 UTC (1145
NST), and analyze a representative 2-day period. Figure 7
shows that MM5 accurately captured the nighttime temperature, but underestimated the daytime temperature at our
Bouddha station. It also shows that MM5 accurately captured the daytime relative humidity minimum, but that it
overestimated relative humidity for most of the simulation,
including an overestimate of the duration of saturation. The
underestimate of the diurnal temperature cycle, in combination with the overestimate of the diurnal humidity cycle,
suggests that this bias is at least partly due to an incorrect
representation of the Bowen ratio (overestimated evapotranspiration at the expense of underestimated sensible heat flux).
Such an error could derive from too high a soil moisture
content in the initial conditions. In addition, part of the bias
seen here might have explanations other than model error.
First, it is plausible that during the daytime the temperature
at the rooftop weather station was slightly contaminated by
local infrared radiation from the tile roof underneath it,
while the MM5 results represent an average over a grid cell
of 1 km by 1 km. Second, the weather station was located
22 m above ground, while MM5’s bottom layer center was
7.5 m above ground. Fog tends to develop from the ground
up, so it is indeed possible that there was fog near the ground
earlier in the morning than the time when 100% relative
humidity was reached at the weather station.
[15] Figures 8a and 8b show that MM5 captured the
diurnal temperature cycle at the TV tower and at Nagarkot
Peak remarkably well. At Kharipati MM5 slightly underestimated the temperature during the day and overestimated
it at night. Part of that may be due to topographic smoothing
by MM5, which lifted the location of Kharipati to a higher
elevation. Figure 9 shows the temperature in the lowest
model layer (between s = 1 and s = 0.994) over the
Kathmandu Valley at dawn (Figure 9a) and in the afternoon
(Figure 9b). At dawn the entire basin bottom was filled with
6 of 20
D21308
PANDAY ET AL.: KATHMANDU VALLEY MODELING RESULTS
D21308
Figure 7. Intercomparison of measurements by the weather station on the Hyatt roof (blue) on 9 and 10
February 2005 with MM5 simulations for the same time period at the grid cell containing the Hyatt roof
and main laboratory (red). (a) Air temperature and (b) relative humidity.
a pool of cold air whose temperature was similar to that
of the highest peaks on the valley rim. We also see that
surrounding river valleys (for example, in the upper left
quadrant of the plots) did not contain cold pools. In contrast
to the Kathmandu Valley’s confined basin shape, these river
valleys allows drainage flows to travel unimpeded. In the
afternoon the Kathmandu Valley’s center was warmer than
its edges, and the cold pool was gone.
Figure 8. Intercomparison of air temperature measurements by Hobo loggers (blue) at three locations
on 9 and 10 February 2005 with MM5 simulations (red) for the same time period at grid cells containing
the logger sites. (a) TV tower bottom (5 m above ground), (b) Nagarkot, and (c) Kharipati.
7 of 20
D21308
PANDAY ET AL.: KATHMANDU VALLEY MODELING RESULTS
D21308
Figure 9. (a) Air temperature in the lowest MM5 layer (color shading) at 0545 NST on 10 February
2005. The underlying topography is indicated by contour lines with a contour interval of 100 m. (b) Same
as in Figure 9a but for air temperature in the lowest MM5 layer at 1445 on 10 February 2005. The black
line indicates the location of the cross section in Figures 12, 14, 15, 16, and 17.
[16] Figure 10 shows the relative humidity in the surface
layer at dawn and shortly before noon on 10 February 2005
and compares it to photographs from Hattiban taken at the
same times. At dawn a large part of the valley center had
100% relative humidity, while the rest of the valley floor
had relative humidity exceeding 90%. In fact, during the
early morning hours the simulation showed frequent fluc-
tuations in the locations of the area reaching the 100%
threshold, indicating that many of the areas shown in dark
blue actually had relative humidity close to 100% as well.
Indeed at dawn that morning we photographed a thin fog
layer covering much of the valley. Time-lapse photography
showed that over the course of the morning the fog layer
thickened, and then disappeared from the surface by late
8 of 20
D21308
PANDAY ET AL.: KATHMANDU VALLEY MODELING RESULTS
D21308
Figure 10. (left) Photographs of the valley from Hattiban Ridge and (right) MM5 output of relative
humidity in the bottommost layer on 10 February 2005. The color scale goes from white to blue to
indicate increasing relative humidity; red indicates where the threshold value of 100% is achieved. The
glow in the early morning photograph results from a thin sheet of fog.
morning. Consistent with this, the MM5 simulation shows
that by 1145 NST the relative humidity near the surface was
below 40% in the entire valley.
[17] Figure 11 shows wind vectors computed by MM5 in
the lowest model layer before dawn and in the afternoon of
10 February 2005. The modeled patterns fit with what we
observed. At dawn the valley center had almost zero surface
winds. At the edges of the valley, katabatic winds descended
from mountain ridges, while drainage flows traveled down
both the Bagmati river valley (coordinates 20 km NS, 32 km
EW) and the river valleys outside of the Kathmandu Valley.
We also see easterly winds crossing the passes on the western
edge of the Kathmandu Valley, such as at Bhimdhunga Pass
(42 km NS, 31 km EW). During the afternoon, strong winds
traveled up the river valleys, while the entire Kathmandu
Valley bottom was swept by westerly/northwesterly winds
entering through the western passes. Afternoon wind speeds
on the order of 5 – 6 m/s over the valley were close to the
average afternoon wind speeds measured at the Bouddha
station. We also see that the up-valley winds from the
Bagmati River valley were able to enter the southern parts
of the valley, but were pushed eastward by the stronger
westerlies entering from the western passes.
[18] Figure 12 shows west– east cross sections through
the innermost domain of the MM5 simulation, with contours of potential temperature as well as wind vectors. At
night there was a strongly stratified cold air pool in the
valley (indeed its top corresponded to the height of the
valley’s passes). By 1145 NST the cold air pool was gone,
and the mixed layer extended up beyond the height of the
valley rim peaks while westerly winds travel through the
valley. This is consistent with our observation of winds in
Figure 3, observations of the lapse rate between Kharipati
and Nagarkot in Figure 5, and observations of mixed layer
height by the sodar shown in part 1.
4. Dynamics of the Nocturnal Cold Air Pool and
the Daytime Mixed Layer
[19] Having verified MM5’s ability to reproduce observed
meteorology in the Kathmandu Valley, we now proceed to
address the first of our two questions: What mechanisms are
responsible for the dissipation of the nocturnal cold air pool
in the Kathmandu Valley, and how does the mixed layer
behave during the day? We discuss four possible mechanisms for cold-pool dissipation before examining model
results.
[20] The first mechanism is often observed over flat
plains, where inversion breakups are driven by surface
heating [Stull, 1988; Young et al., 2000; Angevine et al.,
2001; Garcia et al., 2002]. A convective mixed layer grows
upward as the surface warms up after sunrise. Thermals
generated by the heated bottom rise to increasingly greater
heights, chipping away at the inversion above. As the mixed
layer grows, it entrains air that was previously isolated from
the surface by the inversion. Entrainment of ozone-rich air
can cause a rapid rise in ozone observed at the surface in the
morning. The mixed layer above a flat plain often grows
until the late afternoon, breaking through the uppermost
remnants of the inversion, and then entraining air from the
less stable residual layer that had spent the night above the
inversion. The mixed layer has been observed to grow faster
over cities owing to enhanced convection driven by urban
heat islands [Kallistratova and Coulter, 2004] and to be
9 of 20
D21308
PANDAY ET AL.: KATHMANDU VALLEY MODELING RESULTS
D21308
Figure 11. Surface wind vectors (yellow) from MM5 superimposed upon gray scale shaded topography
(meters above sea level) at (a) 0545 NST and (b) 1615 NST on 10 February 2005. The Kathmandu Valley
is the flat area at the center of the frame. For visual clarity, wind vectors have only been plotted for every
second grid cell. Their length is proportional to the wind speed; a vector length of two grid cells
corresponds to a speed of 10 m/s.
affected by reduced soil moisture content [Santanello et al.,
2005]. The Kathmandu Valley has a ratio of vertical relief
(altitude difference from the bottom to the rim) to horizontal
expanse on the order of 1:30. The basin bottom thus expe-
riences very little shading by surrounding mountains, making
this mechanism plausible.
[21] The second mechanism usually dominates in narrow
mountain valleys. Higher slopes receive morning sunlight
10 of 20
D21308
PANDAY ET AL.: KATHMANDU VALLEY MODELING RESULTS
D21308
Figure 12. West to east vertical cross-section A through the Kathmandu Valley on 10 February 2005,
showing the Trisuli River (TR) low tributary valleys west of Kathmandu (WV), the western valley rim
peaks (Nagarjun, NJ), the elevated flat basin of the Kathmandu Valley (KV), the eastern valley rim peaks
(Nagarkot, NK), and the lower valleys to the east. Color shading and contours in the atmosphere indicate
potential temperature; vectors are proportional to wind velocity in the plane of the plot. Vectors displayed
in the bottom right of each plot correspond to maximum wind velocity (horizontal, vertical) in each plot:
20 m/s, 23.4dPa/s at 0245 NST; 15.2 m/s, 22.4 dPa/s at 0945 NST, 15.6 m/s, 29.1 dPa/s at 1045 NST, and
16.8 m/s, 37 dPa/s at 1145 NST.
first. As the slopes warm up, upslope flows are generated
which carry air up from the valley bottom. The air departing
from the valley bottom is replenished by the generation of
a wind blowing up the river valley. In many places strong
winds are found blowing up river valleys during the
daytime, while a shallow layer of air blows up sidewalls
perpendicular to the valley axes [Atkinson, 1981; Whiteman,
1990; Prévôt, 1994; Li and Atkinson, 1999; Prévôt et al.,
2000; Whiteman, 2000]. Up-valley winds driven by the
warming of the Mustang Basin have been studied in detail
in the Kali Gandaki Valley in northern Nepal [Egger et al.,
2000, 2002].
[22] A third mechanism is found in basins and wide
valleys [Whiteman, 1982]. Here warming sidewalls generate
upslope flows as in the second mechanism. But, rather than
generating a replacement flow up the river valley, the air
mass leaving the basin bottom is replenished by subsidence
overhead. Subsidence heating over the basin center provides
additional erosion of the inversion from the top. This
mechanism has been observed in basins ranging in size
from 1 km [Whiteman et al., 2004b] to several hundred
kilometers across [Banta and Cotton, 1981; Kondo et al.,
1989; Whiteman et al., 1999; Zhong et al., 2001]. If a basin
is wide enough, then the upslope flows and inversion
erosion might begin first on the east-facing west side of
the basin [Kelly, 1988]. There have also been observations
of a hybrid mechanism that combines aspects of the first and
the third, whereby the breakdown of the cold air pool
11 of 20
D21308
PANDAY ET AL.: KATHMANDU VALLEY MODELING RESULTS
happens through a combination of thermals rising from a
warming surface, as well as subsidence descent over the
cold air pool to compensate for the fraction of the air mass
traveling up the slopes [Whiteman, 1982]. In complex
valleys (and probably also valleys of intermediate size), a
combination of the second and third mechanism are also
possible [Rotach and Zardi, 2007].
[23] In the fourth mechanism the cold pool is dissipated
by turbulent erosion caused by vertical wind shear at the top
of the basin, leading to shear-generated turbulence and the
penetration of the upper-level wind downward through the
cold air pool [Whiteman and Doran, 1993]. This mechanism has been studied in Ljubljana, Slovenia, using MM5
mesoscale model simulations [Rakovec et al., 2002]. The
study found that wind speeds aloft must exceed a certain
threshold to begin eroding the cold air pool, and must
increase over time to be able to completely remove the
pool. In some basins, deep persistent temperature inversions
can be removed rapidly when cold air advects overhead,
reducing the inversion strength [Whiteman et al., 2001;
Zängl, 2005]. The authors of a field study in the South
Park Basin in Colorado [Banta and Cotton, 1981] describe
three wind regimes in the basin: downslope, upslope (both
of which take place in narrow valleys as well), and
afternoon winds in the same the direction as the wind above
the basin (something usually not found in narrow valleys). In
the Kathmandu Valley, during the dry season, the prevailing
upper-level winds are westerlies. During afternoons the
surface winds within the valley are westerlies as well.
[24] Figure 12 shows the time evolution of potential
temperature and wind vectors over the Kathmandu Valley
on a west-to-east cross section during the morning of 10
February 2005. We see evidence for the dominance of the
third mechanism described above in the episodes that we
modeled: namely, that the inversion breaks up through the
combined effect of upslope flows along the valley edge, and
subsidence over the valley center. The potential temperature
contours in Figure 12 show that on the morning of 10
February 2005, a mixed layer characterized by constant
potential temperatures first grew at the valley edge, while
the valley center remained stably stratified. A zoomed-in
view of the high-resolution model output confirms that the
winds near the surface were indeed blowing upslope. The
downward dip in potential contours over the valley center
seen at 0945 and 1045 NST suggests that wind vectors over
the valley center had a downward component. However, to
confirm that subsidence took place, we carry out a back-ofthe-envelope calculation. If there were no subsidence, then
the air leaving the valley would need to be balanced
by inflows through the Bagmati Valley. Assuming a valley
circumference of 90 km, an average upslope flow of 3 m/s,
and an upslope flow depth of 200 m, we get an outflow of
5.4 107 m3/s. For that to be balanced by inflows through
the roughly 2 km2 cross-sectional area of the Bagmati
Valley, we would need wind speeds of 27 m/s or 97 km/h!
Such winds speeds were not found in our simulations, nor
did local residents of the Bagmati Valley ever mention such
high wind speeds. A more realistic upper bound on the
inflow through the Bagmati Valley of 10 m/s would only
provide 2 107 m3/s, leaving 3.4 107 m3/s for subsidence. Over a valley-bottom area of 340 km3 that would
give a subsidence rate of 0.1 m/s or about 360 m/h.
D21308
[25] By 1145 NST, there was a uniform mixed layer over
the valley (the valley-center stratification was gone), but
there continued to be higher potential temperatures over the
peaks, driving upslope flows, which allowed air parcels
reaching the peaks to mix to higher elevations than ones
rising straight up from the basin floor.
[26] We can conclude that in February in the Kathmandu
Valley, the cold pool dissipation is more strongly influenced
by the third mechanism described above than by the second
mechanism; it appears that during the time studied, winds
up the Bagmati Valley played a smaller role in basin air
replenishment during morning upslope flows compared to
the role played by subsidence over the basin center.
[27] Further evidence for upslope flow driven mechanisms was found during time-lapse photography from
Hattiban ridge (Figure 13). We found that the fog along
the edges of the basin dissipated rapidly, hours before the
fog over the center of the basin dissipated. At the time that
the fog along the edges was dissipating, strong upslope
winds were felt at the camera location.
[28] For comparison to our winter findings, Figure 14
shows a cross section at 0945 NST on 11 May 2005. We
again see evidence of stronger vertical mixing over the
peaks in comparison to the basin center as well as upslope
flows; these were found, that day, to initiate the mixed layer
growth around the basin perimeter as early as 0700 NST,
with a deep mixed layer already existing by 0945 NST. In
addition, though, we see in Figure 14 that the land surface
over the basin bottom was heating as well, with an
unstable near-surface region of higher potential temperature than in the air above. However, this was only seen
briefly; by 1015 NST the air over the entire valley was
well mixed. The MM5 simulations suggest that in May a
hybrid of the first and the third mechanisms controls the
growth of the mixed layer. Shortly after dawn, upslope
flows begin, at least on sunlit slopes, eroding the nocturnal
stable stratification as described in the third mechanism.
However, as the Sun rises higher, the basin bottom generates thermals that let the mixed layer grow from below
(first mechanism). A likely reason that this mechanism was
seen in the May simulation but not in the February simulation is the fact that in February the basin bottom was
covered by fog, which reduced the surface heating. A delay
in the breakup of a valley inversion in winter was also found
in a lidar study in Bulgaria [Savov et al., 2002]. Note that
the surface thermals seen in the TV tower data in February
(Figure 5) only appear after the morning fog has dissipated.
[29] In the episodes modeled, no evidence was found in
support of the fourth mechanism (turbulent erosion by
winds flowing across the top of the cold air pool) playing
a dominant role in the removal of the Kathmandu Valley’s
nocturnal inversion. The mechanism would have required
us to find the cold pool remaining intact in the basin bottom,
while its top was slowly eroded away by winds.
[30] The dissipation of the nocturnal cold air pool is not
the only factor that affects the growth of the Kathmandu
Valley’s daytime mixed layer. If a basin sits at a higher
elevation compared to regions outside the surrounding
mountains, it can generate thermal circulations similar to
what is found on raised plateaus. Plateaus often serve as
elevated heat sources [Egger, 1987; Molnar and Emanuel,
1999]. The daytime heating expansion of air columns above
12 of 20
D21308
PANDAY ET AL.: KATHMANDU VALLEY MODELING RESULTS
D21308
Figure 13. Kathmandu Valley from Hattiban on the morning of 8 December 2005 at (a) 0752 NST and
(b) 0830 NST.
a plateau and above nearby lowlands results in higher
pressure and lower temperature above the lowlands than
above the plateau [Gaertner and Fernandés, 1993]. Plateau
circulations have been studied in theory by several researchers,
including the effects of valleys leading to the plateau [Egger,
1987; Zängl et al., 2001], and the effect of plateau height
and the existence of a mountain rim around the plateau
[Zängl and Gonzales Chico, 2006]. If there are mountain
passes connecting a basin to the low-lying outside regions,
as is the case in the Kathmandu Valley, these passes are
found to play crucial roles in the circulation. Detailed field
and modeling studies of the gap flows taking places at the
edges of plateaus and elevated basins have been carried out
for the Tibetan Plateau [Egger et al., 2000; Zängl et al.,
2001; Egger et al., 2002], the Bolivian Altiplano [Egger et
al., 2005; Zängl and Egger, 2005; Reuder and Egger, 2006],
and the Mexico City Basin [Doran and Zhong, 2000; de Foy
et al., 2006].
13 of 20
D21308
PANDAY ET AL.: KATHMANDU VALLEY MODELING RESULTS
D21308
Figure 14. Potential temperature and wind vectors on the same cross section as Figure 17, but at
0945 NST on 11 May 2005. Vectors displayed in the bottom right of the plot correspond to maximum
wind velocity: 14.2 m/s horizontal, 99.3 dPa/s vertical.
[31] Figure 15 shows a continuation of the time series
of MM5 cross sections from Figure 12. We see the arrival
into the Kathmandu Valley through the western passes of
lower potential temperature air originating above the lower
neighboring valleys to the west. Starting from the valley’s
western rim, this air spreads across the entire bottom of the
Kathmandu Valley in less than an hour. This creates a new
stratification that confines the mixing to much lower heights
than was possible moments earlier. We thus have an
explanation of sodar observations (see part 1), which show
that the depth of the mixed layer decreased over the course
of the afternoon.
[32] A modeling study of diurnal flows through passes
leading from the Pacific lowlands to the Bolivian Altiplano
also found a density current of cooler air spreading from the
pass across the Altiplano as a cold front [Zängl and Egger,
2005]. The winds arriving into the Kathmandu Valley from
the west during the afternoon appear to be part of a larger
plateau circulation. Earlier researchers calculated a temperature difference of 2.5 K at noon time between air at the
valley bottom, and the cooler air at the same elevation
above lowlands 60 km to the southwest of Kathmandu
[Regmi et al., 2003]. The Iberian Peninsula also experiences
a plateau circulation, with an afternoon heat low, and here
too a front of cooler air travels in from the plateau edge,
pushing underneath the existing boundary layer [Gaertner
and Fernandés, 1993]. Similar gravity driven flows have
been found in idealized simulations [de Wekker et al., 1998]
and in the lee of the Sierra Nevada [Zhong et al., 2008].
Researchers of the plateau circulation at Zugspitzplatt,
Germany [Gantner et al., 2003], found air arriving on the
plateau from higher elevations, marked by a drop in the dew
point. Indeed, our data showed an afternoon drop in dew
point in Kathmandu (Figure 4b). Following the isentropes
from left to right on Figure 15, we see that the air sweeping
through the Kathmandu Valley bottom during the afternoon
originated at a higher elevation than the valley.
5. Ventilation Pathways
[33] The timing and evolution of the mixed layer in the
Kathmandu Valley has an effect on when and how pollutants can be ventilated from the valley. Our second question
thus concerns the pathways of the pollutants when they
leave the valley. Before looking at model simulations, let us
consider some hypotheses of possible pollutant ventilation
pathways.
[34] The first hypothesis is that the pollutants are ventilated from the Kathmandu Valley because the daytime
mixed layer rises above the height of the surrounding
mountains, to altitudes where polluted air can flow over
the rim mountains. Such behavior has been observed and
modeled in Mexico City [Bossert, 1997; Fast and Zhong,
1998]. During the TRACT experiment over the Black Forest
in southern Germany, it was found that the altitude of the
top of the mixed layer is higher over mountains than over
surrounding flat land, and that the mixed layer top rises and
falls with the underlying topography [Kalthoff et al., 1998;
Kossmann et al., 1998]. In other words, the top of the fully
formed mixed layer may be flat over the Kathmandu Valley
center, and then rise as one approaches the valley rim
mountains, reaching levels higher than the mountain tops.
The sodar data indicated a mixed layer height approaching
the height of the surrounding mountains even at the valley
center, while the temperature logger data indicated well
mixed conditions at least up to the height of Nagarkot near
the valley’s edge. Once mixed to heights above the surrounding mountains, pollutants can easily be carried eastward by the strong westerly winds that blow parallel to the
Himalayas during the dry season [Moore, 2004].
14 of 20
D21308
PANDAY ET AL.: KATHMANDU VALLEY MODELING RESULTS
D21308
Figure 15. Continuation of Figure 12, showing potential temperature and wind vectors during the
afternoon of 10 February 2005. Vectors displayed in the bottom right of each plot correspond to
maximum wind velocity (horizontal, vertical) in each plot: 17.7 m/s, 71.0 dPa/s at 1515 NST; 16.8 m/s,
56. dPa/s at 1545 NST; 15.7 m/s, 73.0 dPa/s at 1615 NST; 14.5 m/s, 90.5 dPa/s at 1645 NST.
[35] The second hypothesis is that the pollutants are
removed from the valley by flows through mountain passes,
with relatively clean air blowing in through the western
valley rim passes and polluted air spilling out through the
eastern passes. A field study in the South Park basin of
Colorado [Banta and Cotton, 1981], found that an inversion
in the morning prevented upper-level westerly winds from
penetrating to the surface. However, once the inversion was
gone the westerlies could reach down to the surface. An
idealized modeling study of flows through valleys that
terminated at the edge of a plateau [Egger, 1987] (similar
to the valleys to the west of Kathmandu Valley that terminate
at its western passes) shows that the existence of such valleys
enhances the wind intensity above the plateau. Egger [1987]
also shows that the up-valley wind intensity in these valleys
is determined by the difference in elevation of the plateau
heat source (1300 m in the Kathmandu Valley’s case), and
the lowland heat source (500– 600 m in the valleys west
of Kathmandu). Tracers passing from a lowland area over a
mountain ridge into a basin have been studied before [Kimura
and Kuwagata, 1993], as have flows through passes and
mountain gaps in Athens [Asimakopoulos et al., 1992],
Mexico City [de Foy et al., 2006] and the Brenner Pass in
the Austrian Alps [Mayr et al., 2002]. In winter, Mexico
City regularly experiences a low-level jet exceeding 10 m/s
entering the basin through a pass to the southeast [Doran
and Zhong, 2000]. Tracer transport modeling has shown
that the flow through this pass plays an important role in
flushing pollutants out of Mexico City [Bossert, 1997]. A
2003 field campaign studying the diurnal variations in
circulation over the Bolivian Altiplano (an elevated plateau
surrounded by mountains) carried out vertical soundings at
the in-flow passes that found inflow to start a few hours
after sunrise, at the time when the stable nighttime layer over
the Altiplano had heated sufficiently to become neutrally
buoyant [Egger et al., 2005].
[36] The third hypothesis is that pollutants are vented out
of the Kathmandu Valley by upslope flows at the valley rim
15 of 20
D21308
PANDAY ET AL.: KATHMANDU VALLEY MODELING RESULTS
D21308
Figure 16. Lagrangian trajectories released at 0745 NST on 10 February 2005 on a line between 38 km EW
26 km NS and 38 km EW, 46 km NS, (top) shown on a contour map (masl) and (bottom) projected onto a
west to east cross section. Green denotes times from 0745 to 1145 NST, red is from 1145 to 1545 NST,
orange is from 1545 to 1945 NST, and brown is from 1945 to 2345 NST.
mountains, which transport the pollutants into more rapid
flows aloft. Thermally driven upslope and up-valley winds
have been studied in many places and have been found to
be efficient ways of transporting air pollutants from lower
altitudes and venting them into fast-flowing upper-level
winds. At the Jungfraujoch mountain observatory in Switzerland, pollutants are seen arriving from the valleys during
warm summer afternoons [Baltensperger et al., 1997].
Winds blowing up the valleys of the Black Forest mountains
have been found to transport tracers out of Freiburg [Kalthoff
et al., 2000]. Polluted air has been observed traveling up
the Yahagi River valley in Japan [Kitada et al., 1986]. A
modeling study of Santiago de Chile found pollutant export
above mountain peaks through the convergence of upslope
flows [Schmitz, 2005]. Mountains near Phoenix, Arizona,
provide sufficient ventilation to the city that there is no
multiple-day buildup of ozone from local emission sources
[Fast et al., 2000]. Up-valley flows in the Ticino area of
16 of 20
D21308
PANDAY ET AL.: KATHMANDU VALLEY MODELING RESULTS
D21308
Figure 17. Same as Figure 16, but for a release time of 1545 NST on 10 February 2005.
southern Switzerland have been found to carry pollutants
upward from industrial northern Italy to the crest of the Alps
[Prévôt et al., 2000]. These flows can transport several times
the valley volume over the course of a day [Henne et al.,
2004].
[37] We examined ventilation pathways from the
Kathmandu Valley using the results of Lagrangian model
trajectories for 10 February 2005. The trajectories in
Figures 16 –18 were released in the surface layer, along
a 20-km north-south line running through the center of the
innermost domain (26 km NS, 38 km EW, to 46 km NS,
38 km EW). For visual clarity, release points are only shown
at 2-km intervals; these are a small selection of hundreds of
trajectories that were examined to draw conclusions.
[38] Figure 16 shows trajectories with a release time of
0745 NST. We see that, over the course of the morning,
most of the trajectories traveled toward the nearest valley
rim slope, before traveling up the slopes to a height of 500–
1000 m above the basin bottom. Then, before 1145 NST
they started to travel eastward, crossing the Kathmandu
17 of 20
D21308
PANDAY ET AL.: KATHMANDU VALLEY MODELING RESULTS
D21308
passes during the night, consistent with the strong easterly
winds observed at Bhimdhunga Pass during the night
[Panday and Prinn, 2009]. It is likely that the drainage
winds in the neighboring valleys were also responsible for
driving the light easterly winds that were found along the
top of the Kathmandu Valley’s cold pool at night and that
also were observed in the nighttime sodar soundings.
6. Conclusion
Figure 18. Trajectories released at 2345 NST on the same
line as in Figures 11 and 12, shown on a contour map
(masl). Black denotes time from 2345 to 0345 NST, blue is
from 0345 to 0745 NST, green is from 0745 to 1145 NST,
red is from 1145 to 1545 NST, orange is from 1545 NST to
1945 NST, and brown is from 1945 to 2345 NST.
Valley at high elevation. A closer look at individual tracers
shows that upon arriving at the eastern valley rim, trajectories
which passed over passes or low ridge regions maintained
their altitude, whereas trajectories passing over the eastern
mountain peaks experienced a step-like gain in altitude that
was driven by the convergence of upslope flows over the
mountain tops east of the Kathmandu Valley.
[39] Figure 17 shows the trajectories that were released
along the north – south line on the valley floor at 1545 NST
on 9 February 2005. Most trajectories converged to flow
through the Sanga and Nala passes on the eastern valley
rim, and a higher pass on the southern valley rim, above the
town of Tika Bhairav. Most of the trajectories reached
outside of the valley within less than 2 h of their release.
They no longer rose to the height of the valley rim peaks.
They could not rise as high during the afternoon because
the mixed layer had shrunk as a result of the arrival of
cooler air from neighboring valleys that formed a lowaltitude afternoon inversion over the Kathmandu Valley, as
we saw in section 4.
[40] Finally, Figure 18 shows trajectories released at
2345 NST along the same 20 km north – south line. The
trajectories originating at the southern end of the Kathmandu
Valley traveled down the Bagmati river valley. Most of the
trajectories released within the central Kathmandu Valley
lingered close to their points of origin. Trajectories released
near the western valley-rim passes rose and crossed over the
passes, and traveled down the western valleys until the
morning, when upslope and up-valley winds carried them
back up. In Figure 11a we saw strong nocturnal drainage
winds in these western valleys before dawn. The strong
drainage winds indicate a sufficient pressure gradient to
force some Kathmandu Valley air up and over the western
[41] In our field observations of air pollution in the
Kathmandu Valley during the dry season of 2004 – 2005,
we noted that carbon monoxide, as a tracer of anthropogenic
emissions, had a distinct diurnal pattern of morning and
evening peaks. In this paper we have described simulations
using the MM5 mesoscale meteorological model nested up
to 1-km resolution, and verified its ability to capture the local
meteorology in the valley as measured during our campaigns.
We studied the dynamics of the dissipation of the nighttime
cold-air pool and temperature inversion, and the subsequent
evolution of the daytime mixed layer. We concluded that the
cold pool dissipation was dominated by the mechanism of
upslope flows along the valley rim mountains, accompanied
by subsidence over the basin center. In addition, during fogfree days there was erosion of the inversion from below by
the growth of thermals forming over a warming surface. The
Kathmandu Valley’s mixed layer height peaked around noon,
and then decreased again over the course of the afternoon.
This pattern was explained using MM5 simulations that
showed gravity-driven flows penetrating into the basin bottom through the western mountain passes. These brought air
that originated above lower-lying terrain to the west of the
valley that was cooler and denser than the air found over
the Kathmandu Valley in the late morning. The Kathmandu
Valley thus exhibited an afternoon circulation pattern similar to what had been found on plateaus elsewhere in the
world.
[42] Using forward trajectories released in the MM5 flow
fields, we found that the dominant pathway of pollutant
ventilation depends upon the time of day. On the February
day investigated, we found that during the morning, tracers
were advected out of the valley by transport up the slopes of
the valley rim mountains, and from there they were carried
east by the upper-level westerly winds. During the afternoons, tracers were found to be removed from the valley at
a faster rate through the mountain passes, both to the east
and to the south. During the night, some tracers near the
Bagmati River were advected out down the river valley, and
near the western passes some tracers were able to escape
from the Kathmandu Valley through drainage flows.
[43] Our extensive new observations and numerical
modeling add significantly to the current understanding of
the valley’s atmospheric dynamics and air pollution meteorology. The Kathmandu Valley’s air quality is strongly
affected by the wind systems generated by the surrounding
topography. Our study has several implications for air
quality policy in the Kathmandu Valley. It points to the fact
that the timing of emissions is key. If the evening rush hour
takes place before sunset, as it does in summer, its emissions
are ventilated out of the valley before the cold pool sets in.
If the rush hour takes place after sunset (as it does in
winter), it is responsible for much higher pollution levels
18 of 20
D21308
PANDAY ET AL.: KATHMANDU VALLEY MODELING RESULTS
in the evening. As discussed in detail in part 1 [Panday and
Prinn, 2009], these pollutants are lifted up, but recirculated
in the morning, adding to the morning pollution. The
Kathmandu Valley also appears to be a rather unsuitable
place for combustions sources that operate 24 h a day, such
as the brick kilns on the outskirts of the city. Emissions
that take place outside of the well ventilated midday and
afternoon hours contribute much more to air pollution in
Kathmandu than would emissions that are restricted to wellventilated time periods. Designing an effective air pollution
control strategy for the Kathmandu Valley thus requires
careful consideration of the pollution diurnal cycle, and
time-dependent restrictions on polluting activities.
[44] Acknowledgments. The research described in this paper was
made possible through generous support from an MIT Presidential Fellowship, the Martin Fellowship for Sustainability, a pilot research project grant
from the Alliance for Global Sustainability, the PAOC Houghton Fund, the
TEPCO Chair account, NSF grant ATM-0120468 to MIT, NASA grants
NAG5-12099 and NAG5-12669 to MIT, and the federal and industrial
sponsors of the MIT Joint Program on the Science and Policy of Global
Change.
References
Allwine, K. J., and B. K. Lamb (1992), Wintertime dispersion in a mountainous
basin at Roanoke, Virginia: Tracer study, J. Appl. Meteorol., 31, 1295 –
1311, doi:10.1175/1520-0450(1992)031<1295:WDIAMB>2.0.CO;2.
Angevine, W. M., et al. (2001), Observations of the morning transition of the
convective boundary layer, Boundary Layer Meteorol., 101, 209 – 227,
doi:10.1023/A:1019264716195.
Asimakopoulos, D., et al. (1992), An experimental study of nighttime air
pollutant transport over complex terrain in Athens, Atmos. Environ., Part
B, 26(1), 59 – 71.
Atkinson, B. W. (1981), Meso-scale Atmospheric Circulations, Academic
Press, London.
Baltensperger, U., H. W. Gäggeler, D. T. Jost, M. Lugauer, M. Schwikowski,
P. Seibert, and E. Weingartner (1997), Aerosol climatology at the highalpine site Jungfraujoch, Switzerland, J. Geophys. Res., 102(D16),
19,707 – 19,715, doi:10.1029/97JD00928.
Banta, R., and W. R. Cotton (1981), An analysis of the structure of local wind
systems in a broad mountain basin, J. Appl. Meteorol., 20, 1255 – 1266,
doi:10.1175/1520-0450(1981)020<1255:AAOTSO>2.0.CO;2.
Bossert, J. E. (1997), An investigation of flow regimes affecting the Mexico
City region, J. Appl. Meteorol., 36, 119 – 140, doi:10.1175/15200450(1997)036<0119:AIOFRA>2.0.CO;2.
Clements, C. B., et al. (2003), Cold-air-pool structure and evolution in a
mountain basin: Peter Sinks, Utah, J. Appl. Meteorol., 42, 752 – 768,
doi:10.1175/1520-0450(2003)042<0752:CSAEIA>2.0.CO;2.
de Foy, B., et al. (2006), Distinct wind convergence patterns in the Mexico
City basin due to the interaction of the gap winds with the synoptic flow,
Atmos. Chem. Phys., 6, 1249 – 1265.
de Wekker, S. F. J., et al. (1998), A numerical study of the thermally driven
plain-to-basin wind over idealized basin topographies, J. Appl. Meteorol.,
47, 339 – 350.
Doran, J. C., and S. Zhong (2000), Thermally driven gap winds into the
Mexico City Basin, J. Appl. Meteorol., 39, 1330 – 1340, doi:10.1175/
1520-0450(2000)039<1330:TDGWIT>2.0.CO;2.
Dudhia, J. (1989), Numerical study of convection observed during the
winter monsoon experiment using a mesoscale two-dimensional model,
J. Atmos. Sci., 46(20), 3077 – 3107, doi:10.1175/1520-0469(1989)046<
3077:NSOCOD>2.0.CO;2.
Dudhia, J. (1993), A nonhydrostatic version of the Penn-State-NCAR mesoscale model: Validation tests and simulation of an Atlantic cyclone and
cold front, Mon. Weather Rev., 121, 1493 – 1513, doi:10.1175/15200493(1993)121<1493:ANVOTP>2.0.CO;2.
Egger, J. (1987), Valley winds and diurnal circulation over plateaus, Mon.
Weather Rev., 115, 2177 – 2185, doi:10.1175/1520-0493(1987)115<
2177:VWATDC>2.0.CO;2.
Egger, J., et al. (2000), Diurnal winds in the Himalayan Kali Gandaki
Valley. Part I: Observations, Mon. Weather Rev., 128, 1106 – 1122,
doi:10.1175/1520-0493(2000)128<1106:DWITHK>2.0.CO;2.
Egger, J., et al. (2002), Diurnal winds in the Himalayan Kali Gandaki
Valley. Part III: Remotely piloted aircraft soundings, Mon. Weather
Rev., 130, 2042 – 2058, doi:10.1175/1520-0493(2002)130<
2042:DWITHK>2.0.CO;2.
D21308
Egger, J., et al. (2005), Diurnal circulation of the Bolivian Altiplano. Part I:
Observations, Mon. Weather Rev., 133, 911 – 924, doi:10.1175/
MWR2894.1.
Fast, J. D., and S. Zhong (1998), Meteorological factors associated with
inhomogeneous ozone concentrations within the Mexico City basin,
J. Geophys. Res., 103(D15), 18,927 – 18,946, doi:10.1029/98JD01725.
Fast, J. D., J. C. Doran, W. J. Shaw, R. L. Coulter, and T. J. Martin (2000),
The evolution of the boundary layer and its effect on air chemistry in the
Phoenix area, J. Geophys. Res., 105(D18), 22,833 – 22,848, doi:10.1029/
2000JD900289.
Gaertner, M. A., and C. Fernandés (1993), A two-dimensional simulation of
the Iberian summer thermal low, Mon. Weather Rev., 121, 2740 – 2756,
doi:10.1175/1520-0493(1993)121<2740:ATDSOT>2.0.CO;2.
Gantner, L., et al. (2003), The diurnal circulation of Zugspitzplatt: observations and modeling, Meteorol. Z., 12(2), 95 – 102, doi:10.1127/09412948/2003/0012-0095.
Garcia, J. A., et al. (2002), A case study of the morning evolution of the
convective boundary layer depth, J. Appl. Meteorol., 41, 1053 – 1059.
Gohm, A., et al. (2004), South Foehn in the Wipp Valley on 24 October
1999 (MAP IOP 10): Verification of high-resolution numerical simulations with observations, Mon. Weather Rev., 132, 78 – 102, doi:10.1175/
1520-0493(2004)132<0078:SFITWV>2.0.CO;2.
Henne, S., et al. (2004), Quantification of topographic venting of boundary
layer air to the free troposphere, Atmos. Chem. Phys., 4, 497 – 509.
Hong, S.-Y., and H.-L. Pan (1996), Nonlocal boundary layer vertical diffusion
in a medium-range forecast model, Mon. Weather Rev., 124, 2322 – 2339,
doi:10.1175/1520-0493(1996)124<2322:NBLVDI>2.0.CO;2.
Kain, J. S. (2004), The Kain-Fritsch convective parameterization: An update,
J. Appl. Meteorol., 43, 170 – 181, doi:10.1175/1520-0450(2004)043<
0170:TKCPAU>2.0.CO;2.
Kallistratova, M. A., and R. L. Coulter (2004), Application of sodars in the
study and monitoring of the environment, Meteorol. Atmos. Phys., 85,
21 – 37, doi:10.1007/s00703-003-0031-1.
Kalthoff, N., et al. (1998), Temporal evolution and spatial variation of the
boundary layer over complex terrain, Atmos. Environ., 32(7), 1179 –
1194, doi:10.1016/S1352-2310(97)00193-3.
Kalthoff, N., V. Horlacher, U. Corsmeier, A. Volz-Thomas, B. Kolahgar,
H. Geiß, M. Möllmann-Coers, and A. Knaps (2000), Influence of
valley winds on transport and dispersion of airborne pollutants in the
Freiburg-Schauinsland area, J. Geophys. Res., 105(D1), 1585 – 1597,
doi:10.1029/1999JD900999.
Kelly, R. D. (1988), Asymmetric removal of temperature inversions in a
high mountain valley, J. Appl. Meteorol., 27, 664 – 673, doi:10.1175/
1520-0450(1988)027<0664:AROTII>2.0.CO;2.
Kimura, F., and T. Kuwagata (1993), Thermally induced wind passing from
plain to basin over a mountain range, J. Appl. Meteorol., 32, 1538 – 1547,
doi:10.1175/1520-0450(1993)032<1538:TIWPFP>2.0.CO;2.
Kitada, T., et al. (1986), Numerical analysis of air pollution in a combined
field of land/sea breeze and mountain/valley wind, J. Clim. Appl. Meteorol., 25, 767 – 784, doi:10.1175/1520-0450(1986)025<0767:NAOAPI>
2.0.CO;2.
Klemp, J. B., and D. R. Durran (1983), An upper boundary condition
permitting internal gravity wave radiation in numerical mesoscale models,
Mon. Weather Rev., 111, 430 – 444, doi:10.1175/1520-0493(1983)111<
0430:AUBCPI>2.0.CO;2.
Kondo, J., et al. (1989), Heat budget analysis of nocturnal cooling and
daytime heating in a basin, J. Atmos. Sci., 46(19), 2917 – 2933,
doi:10.1175/1520-0469(1989)046<2917:HBAONC>2.0.CO;2.
Kossmann, M., et al. (1998), Aspects of the convective boundary layer
structure over complex terrain, Atmos. Environ., 32(7), 1323 – 1348,
doi:10.1016/S1352-2310(97)00271-9.
Li, J.-G., and B. W. Atkinson (1999), Transition regimes in valley airflows,
Boundary Layer Meteorol., 91, 385 – 411, doi:10.1023/A:1001846005338.
Mayr, G. J., et al. (2002), An automobile platform for the measurement of
foehn and gap flows, J. Atmos. Oceanic Technol., 19, 1545 – 1556,
doi:10.1175/1520-0426(2002)019<1545:AAPFTM>2.0.CO;2.
Molnar, P., and K. A. Emanuel (1999), Temperature profiles in radiativeconvective equilibrium above surfaces at different heights, J. Geophys.
Res., 104(D20), 24,265 – 24,271, doi:10.1029/1999JD900485.
Moore, G. W. K. (2004), Mount Everest snow plume: A case study, Geophys.
Res. Lett., 31, L22102, doi:10.1029/2004GL021046.
Neff, W. D., and C. W. King (1989), The accumulation and pooling of
drainage flows in a large basin, J. Appl. Meteorol., 28, 518 – 529,
doi:10.1175/1520-0450(1989)028<0518:TAAPOD>2.0.CO;2.
Panday, A. K. (2006), The diurnal cycle of air pollution in the Kathmandu
Valley, Nepal, Doctor of Science thesis, Mass. Inst. of Technol., Cambridge,
Mass.
Panday, A. K., and R. G. Prinn (2009), Diurnal cycle of air pollution in the
Kathmandu Valley, Nepal: Observations, J. Geophys. Res., 114, D09305,
doi:10.1029/2008JD009777.
19 of 20
D21308
PANDAY ET AL.: KATHMANDU VALLEY MODELING RESULTS
Prévôt, A. S. H. (1994), Photooxidantien und Primärluftschadstoffe in der
planetaren grenzschicht in der Schweiz nördlich und südlich der Alpen,
Ph.D. thesis, 125 pp., ETH Zurich, Zurich, Switzerland.
Prévôt, A. S. H., et al. (2000), Diurnal variation of volatile organic compounds and local circulation systems in an Alpine valley, Atmos. Environ.,
34, 1413 – 1423, doi:10.1016/S1352-2310(99)00440-9.
Rakovec, J., J. Merse, S. Jernej, and B. Paradiz (2002), Turbulent dissipation of the cold-air pool in a basin: Comparison of observed and simulated development, Meteorol. Atmos. Phys., 79, 195 – 213, doi:10.1007/
s007030200003.
Regmi, R. P., T. Kitada, and G. Kurata (2003), Numerical simulation of late
wintertime local flows in Kathmandu Valley, Nepal: Implication for air
pollution transport, J. Appl. Meteorol., 42, 389 – 403, doi:10.1175/15200450(2003)042<0389:NSOLWL>2.0.CO;2.
Reuder, J., and J. Egger (2006), Diurnal circulation of the South American Altiplano: Observations in a valley and at a pass, Tellus, Ser. A, 58,
254 – 262.
Rotach, M. W., and D. Zardi (2007), On the boundary-layer structure over
highly complex terrain: Key findings from MAP, Q. J. R. Meteorol. Soc.,
133, 937 – 948, doi:10.1002/qj.71.
Santanello, J. A., et al. (2005), An empirical investigation of convective
planetary boundary layer evolution and its relationship with the land
surface, J. Appl. Meteorol., 44, 917 – 932, doi:10.1175/JAM2240.1.
Savov, P. B., et al. (2002), Lidar investigation of the temporal and spatial
distribution of atmospheric aerosols in mountain valleys, J. Appl. Meteorol., 41, 528 – 541, doi:10.1175/1520-0450(2002)041<0528:LIOTTA>
2.0.CO;2.
Schmitz, R. (2005), Modelling of air pollution dispersion in Santiago
de Chile, Atmos. Environ., 39, 2035 – 2047, doi:10.1016/j.atmosenv.
2004.12.033.
Stull, R. B. (1988), An Introduction to Boundary Layer Meteorology,
Kluwer Acad., Dordrecht, Netherlands.
Whiteman, C. D. (1982), Breakup of temperature inversions in deep mountain valleys: Part I. Observations, J. Appl. Meteorol., 21, 270 – 289,
doi:10.1175/1520-0450(1982)021<0270:BOTIID>2.0.CO;2.
Whiteman, C. D. (1990), Observations of thermally developed wind systems in mountainous terrain, Meteorol. Monogr., 23, 5 – 42.
Whiteman, C. D. (2000), Mountain Meteorology: Fundamentals and Applications, Oxford Univ. Press, New York.
Whiteman, C. D., and J. C. Doran (1993), The relationship between overlying synoptic-scale flows and winds within a valley, J. Appl. Meteorol.,
32, 1669 – 1682, doi:10.1175/1520-0450(1993)032<1669:TRBOSS>2.0.
CO;2.
D21308
Whiteman, C. D., et al. (1999), Wintertime evolution of the temperature
inversion in the Colorado Plateau basin, J. Appl. Meteorol., 38, 1103 – 1117,
doi:10.1175/1520-0450(1999)038<1103:WEOTTI>2.0.CO;2.
Whiteman, C. D., S. Zhong, W. J. Shaw, J. M. Hubbe, and X. Bian (2001),
Cold pools in the Columbia Basin, Weather Forecast., 16, 432 – 447,
doi:10.1175/1520-0434(2001)016<0432:CPITCB>2.0.CO;2.
Whiteman, C. D., S. Eisenbach, B. Pospichal, and R. Steinacker (2004a),
Comparison of vertical soundings and sidewall air temperature measurements in a small Alpine basin, J. Appl. Meteorol., 43, 1635 – 1647,
doi:10.1175/JAM2168.1.
Whiteman, C. D., et al. (2004b), Inversion breakup in small Rocky Mountain and Alpine basins, J. Appl. Meteorol., 43, 1069 – 1082, doi:10.1175/
1520-0450(2004)043<1069:IBISRM>2.0.CO;2.
Young, G. S., et al. (2000), Observations of the entrainment zone in a
rapidly entraining boundary layer, J. Atmos. Sci., 57(18), 3145 – 3160,
doi:10.1175/1520-0469(2000)057<3145:OOTEZI>2.0.CO;2.
Zängl, G. (2005), Wintertime cold-air pools in the Bavarian Danube Valley
basin: Data analysis and idealized numerical simulations, J. Appl. Meteorol.,
44, 1950 – 1971, doi:10.1175/JAM2321.1.
Zängl, G., and J. Egger (2005), Diurnal circulation of the Bolivian Altiplano. Part II: Theoretical and model investigations, Mon. Weather Rev.,
133, 3624 – 3643, doi:10.1175/MWR3033.1.
Zängl, G., and S. Gonzales Chico (2006), The thermal circulation of a
grand plateau: Sensitivity to the height, width, and shape of the plateau,
Mon. Weather Rev., 134, 2581 – 2600, doi:10.1175/MWR3207.1.
Zängl, G., et al. (2001), Diurnal winds in the Himalayan Kali Gandaki
Valley. Part II: Modeling, Mon. Weather Rev., 129, 1062 – 1080,
doi:10.1175/1520-0493(2001)129<1062:DWITHK>2.0.CO;2.
Zhong, S., et al. (2001), Meteorological processes affecting the evolution of
a wintertime cold air pool in the Columbia basin, Mon. Weather Rev.,
129, 2600 – 2613, doi:10.1175/1520-0493(2001)129<2600:MPATEO>
2.0.CO;2.
Zhong, S., et al. (2008), Forcing mechanisms for Washoe Zephyr – A daytime downslope wind system in the lee of the Sierra Nevada, J. Appl.
Meteorol. Climatol., 47, 339 – 350, doi:10.1175/2007JAMC1576.1.
A. K. Panday, Department of Environmental Sciences, University of
Virginia, Clark Hall, 291 McCormick Rd., Charlottesville, VA 22904, USA.
(arnico@virginia.edu)
R. G. Prinn, Center for Global Change Science, Massachusetts Institute
of Technology, Cambridge, MA 02139, USA.
C. Schär, Institute for Atmospheric and Climate Science, Swiss Federal
Institute of Technology, CH-8057 Zürich, Switzerland.
20 of 20