Stenchikov., G., K. Pickering, A. DeCaria, W.

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 110, D14304, doi:10.1029/2004JD005582, 2005
Simulation of the fine structure of the 12 July 1996 StratosphereTroposphere Experiment: Radiation, Aerosols and Ozone (STERAOA) storm accounting for effects of terrain and interaction with
mesoscale flow
Georgiy Stenchikov,1 Kenneth Pickering,2 Alex DeCaria,2,3 W.-K. Tao,4 John Scala,5
Lesley Ott,2 Diana Bartels,6,7 and Thomas Matejka6,8
Received 5 November 2004; revised 11 March 2005; accepted 8 April 2005; published 22 July 2005.
[1] Vertical mixing of chemical tracers and optically active constituents by deep
convection affects regional and global chemical balances in the troposphere and lower
stratosphere. This important process is not explicitly resolved in global and regional
models and has to be parameterized. However, mixing depends strongly on the spatial
structure, strength, and temporal evolution of the particular storm, complicating
parameterization of this important effect in the large-scale models. To better quantify
dynamic fields and associated mixing processes, we simulate a thunderstorm observed on
12 July 1996 during the STERAO-A (Stratosphere-Troposphere Experiment: Radiation,
Aerosols, and Ozone) Deep Convection field project using the Goddard Cloud Ensemble
(GCE) model. The 12 July STERAO-A storm had very complex temporal and spatial
structure. The meteorological environment and evolution of the storm were significantly
different than those of the 10 July STERAO-A storm extensively discussed in previous
studies. Our 2-D and 3-D GCE model runs with uniform one-sounding initialization were
unable to reproduce the full life cycle of the 12 July storm observed by the CHILL
radar system. To describe the storm evolution, we modified the 3-D GCE model to include
the effects of terrain and the capability of using nonuniform initial fields. We conducted a
series of numerical experiments and reproduced the observed life cycle and fine spatial
structure of the storm. The main characteristics of the 3-D simulation of the 12 July storm
were compared with observations, with 2-D simulations of the same storm, and with the
evolution of the 10 July storm. The simulated 3-D convection appears to be stronger and
more realistic than in our 2-D simulations. Having developed in a less unstable
environment than the 10 July 1996 STERAO-A storm, our simulation of the 12 July storm
produced weaker but sustainable convection that was significantly fed by wind shear
instability in the lower troposphere. The time evolution, direction, and speed of
propagation of the storm were determined by interaction with the nonuniform background
mesoscale flow. For example, storm intensity decreased drastically when the storm left the
region with large convective available potential energy. The model appears to be successful
in reproducing the rectangular four-cell structure of the convection. The distributions of
convergence, vertical vorticity, and position of the inflow level in the later single-cell
regime compare favorably with the airborne Doppler radar observations. This analysis
allowed us to better understand the role of terrain and mesoscale circulation in the
development of a midlatitude deep convective system and associated convective mixing.
Wind, temperature, hydrometeor, and turbulent diffusion coefficient data from the cloud
model simulations were provided for off-line 3-D cloud-scale chemical transport
simulations discussed in the companion paper by DeCaria et al. (2005).
1
Department of Environmental Sciences, Rutgers University, New
Brunswick, New Jersey, USA.
2
Department of Meteorology, University of Maryland, College Park,
Maryland, USA.
3
Now at Department of Earth Sciences, Millersville University,
Millersville, Pennsylvania, USA.
Copyright 2005 by the American Geophysical Union.
0148-0227/05/2004JD005582$09.00
4
NASA Goddard Space Flight Center, Greenbelt, Maryland, USA.
Department of Earth Sciences, Millersville University, Millersville,
Pennsylvania, USA.
6
National Severe Storms Laboratory, NOAA, NCAR, Boulder,
Colorado, USA.
7
Now at NOAA Forecast Systems Laboratory, Boulder, Colorado,
USA.
8
Retired.
D14304
5
1 of 14
D14304
STENCHIKOV ET AL.: 12 JULY STERAO STORM
D14304
Citation: Stenchikov, G., K. Pickering, A. DeCaria, W.-K. Tao, J. Scala, L. Ott, D. Bartels, and T. Matejka (2005), Simulation of the
fine structure of the 12 July 1996 Stratosphere-Troposphere Experiment: Radiation, Aerosols and Ozone (STERAO-A) storm
accounting for effects of terrain and interaction with mesoscale flow, J. Geophys. Res., 110, D14304, doi:10.1029/2004JD005582.
1. Introduction
[2] It is well known that convective activity, which
currently is not explicitly resolved in general circulation
models (GCMs), plays an important role in atmospheric
circulation contributing to upward transport of water vapor,
momentum, and stratosphere-troposphere exchange. Strong
convective storms generate gravity waves that perturb the
tropopause and penetrate the stratosphere and mesosphere
where they break and dissipate, depositing momentum and
energy [Holton, 1982] that play an important role in the
circulation of the middle atmosphere. Deep convection
redistributes trace gases and aerosols from the boundary
layer to the free troposphere [Chatfield and Crutzen, 1984;
Dickerson et al., 1987; Stenchikov et al., 1996]. Chemical
species detrained in the middle and the upper troposphere
are advected for long distances, experiencing chemical
transformation, and affecting regional and global chemical
budgets [Pickering et al., 1990, 1992a, 1992b, 1996],
eventually altering Earth’s radiative balance.
[3] Some of the important effects of convection are
implemented in GCMs and chemical transport models
(CTMs) in the form of parameterizations. However, the
complexity of convective processes, the dependence of
these processes on numerous factors, and nonlinear interaction with the mesoscale environment [Donner et al.,
1999] limit the accuracy of these parameterizations.
[4] The Stratosphere-Troposphere Experiment: Radiation,
Aerosols, and Ozone, Part A (STERAO-A) campaign was
specifically designed to clarify the effects of midlatitude
convection on tropospheric ozone, accounting for a comprehensive set of physical processes including lightning
[Dye et al., 2000]. The most impressive thunderstorm in
this campaign was observed on 10 July 1996 and has been
discussed in detail by Dye et al. [2000], Skamarock et al.
[2000, 2003], and Barth et al. [2001].
[5] To better understand the variability of the effects of
convection and their dependence on the mesoscale environment we studied a convective storm observed over southeastern Wyoming, northern Colorado, and southwestern
Nebraska on 12 July 1996, during STERAO-A. We discuss
our results in this and a companion paper [DeCaria et al.,
2005]. This paper deals with the reconstruction of dynamic
fields that are used by DeCaria et al. [2005] for transport
and chemistry experiments. In the companion paper we
present results of tracer transport calculations for CO and
other gases, which are reasonably well conserved over the
lifetime of the storm. Accurate representation of convective
transport in the model, along with observations of lightning
flash rates and anvil NOx (NO + NO2), allowed DeCaria et
al. to estimate the amount of NO production per flash. The
dynamical fields were also used to drive a cloud-scale
photochemical model to compute the effects of convective
transport and lightning on tropospheric ozone, an important
radiatively active trace gas in the upper troposphere.
[6] The 12 July STERAO storm is significantly different
than the strongest STERAO storm, which occurred 2 days
earlier on 10 July. The convection on 12 July developed in a
significantly more stable environment than on 10 July.
Therefore it was extremely difficult to produce the 12 July
convection in the model simulations. It seems that both
elevated terrain and a strongly nonuniform distribution of
Convective Available Potential Energy (CAPE) played
important roles in the 12 July storm transition from a
multicellular line, to a rectangular four-cell structure, and
then to a single intense cell that may have acquired supercell
characteristics for a brief period. It is important to reproduce
this fine structure of the storm in simulations because it
significantly affects vertical convective mixing of tracers.
We employed both 2-D and 3-D versions of the Goddard
Cloud Ensemble (GCE) model [Tao and Simpson, 1989a,
1989b; Tao et al., 1991, 1993, 2003] to simulate different
(but overlapping) phases of the storm. Our 2-D results have
been discussed by DeCaria et al. [2000].
[7] The 3-D and 2-D GCE models are based on the same
physics. The 2-D GCE model has been used extensively for
storm-scale research and convective transport of trace gas
studies in the tropics and midlatitudes [e.g., Tao et al., 1991,
1993, 1996; Pickering et al., 1991, 1992a, 1992b, 1993,
1998; Scala et al., 1990]. It has been tested in numerous
environments and agreed well with both remote (such as
radar and passive microwave) and in situ observations from
aircraft penetrations. Stenchikov et al. [1996] have applied
the 2-D GCE model for a case study of a strong convective
storm over the Great Plains in North Dakota on 28– 29 June
1989, which was observed in detail during the North Dakota
Thunderstorm Project (NDTP) [Boe et al., 1992].
[8] The 3-D GCE model, which requires significantly
more computer resources than the 2-D version, is actively
used in studies of interactions of clouds with each other
[Tao and Simpson, 1989a] and with their surroundings. It
was successfully applied to better quantify heat, moisture,
momentum, mass, and water budgets associated with deep
convection [e.g., Tao and Simpson, 1989b, 1993, Tao et al.,
1993] accounting for surface fluxes and their impact on
precipitation, CAPE distribution, and boundary layer structure [Lynn et al., 2001; Baker et al., 2001; Lynn and Tao,
2001; Wang et al., 2003; Johnson et al., 2002]. Cloud-scale
chemical tracer transport based on the 3-D GCE simulations
is discussed by Pickering et al. [1996] and DeCaria et al.
[2005]. A review of recent applications of the GCE model
to the understanding of precipitation processes is given by
Tao et al. [2003] and Tao [2003].
[9] We found that we had to modify both the 2-D and 3-D
GCE models to use spatially nonuniform initial conditions
and account for terrain effects in order to simulate the
12 July STERAO storm. The period of linear organization
from about 2030 UTC to 2300 UTC was simulated by
DeCaria et al. [2000] using the 2-D GCE model. The 2-D
simulations of the 12 July storm were conducted along the
41N cross section from 105W to 100W, accounting for
terrain and nonuniform initial fields extracted from the
National Centers for Environmental Prediction (NCEP)
Eta forecast fields. Horizontal resolution was 1 km. The
2 of 14
D14304
STENCHIKOV ET AL.: 12 JULY STERAO STORM
vertical grid was nonuniform with 31 levels and finest
resolution (220 m) was located near the surface. Our 2-D
simulation of the 12 July thunderstorm produced a convective cell with characteristics similar to observed at the initial
stage of the storm development, but it was difficult to expect
that a chosen cross section would be suitable for an
extended simulation period in such a region with strong
nonhomogeneities in the initial and boundary conditions.
[10] In the 2-D simulations the surface-based convection
began as two isolated cells forced nearly simultaneously by
the upslope flow. The storm intensity maximized in about
1.5 hours when the top of the convective cell reached 13 km.
Further evolution is characterized by a developing singlecell structure and a rapid decay of the storm when it
moved into a region with low CAPE along 41N. The
speed of the simulated 2-D storm was 2 times faster than
observed.
[11] On the basis of these findings, we concluded that
using the 2-D approach we can simulate only the initial
evolution of the 12 July thunderstorm (i.e., the development
of a multicellular line along the Wyoming-Colorado border
as simulated by DeCaria et al. [2000]). Here we use the 3-D
GCE model in a 5-hour simulation to capture the further
evolution of the storm, including the transition from linear
multicell convection to a highly 3-D single cell system.
The 3-D meteorological fields were provided for the offline transport experiments conducted in the companion
paper.
[12] The main objectives of this study are as follows:
(1) reconstruct the complex fine structure of the July 12
STERAO storm; (2) account for interaction of convection
with the terrain and mesoscale circulation; (3) produce 3-D
dynamic fields for chemistry modeling; (4) compare 2-D and
3-D simulations to clarify limitations of the 2-D approach;
and (5) compare simulations with observations and with the
July 10 STERAO storm.
2. GCE Model
[13] The GCE model hydrodynamics is based on a
complete set of compressible, nonhydrostatic equations in
a Cartesian coordinate system. A second-order finite difference scheme in the vertical direction and the positive definite
non-oscillatory horizontal advection schemes with small
implicit diffusion [Smolarkiewicz, 1983; Smolarkiewicz and
Grabowscki, 1990] are employed for spatial approximation.
[14] Open boundary conditions of Klemp and Wilhelmson
[1978b] are used at the lateral boundaries. Newtonian
damping is applied to the potential temperature and components of horizontal velocity at the top of the domain at
about 25 km. At the bottom the GCE model is coupled with
the land surface/vegetation model of Boone and Wetzel
[1999]. However, this scheme was not employed in the
current simulation.
[15] A parameterization of sub-grid turbulent mixing is
based on the prognostic equation for turbulent kinetic
energy [Deardorff, 1975; Klemp and Wilhelmson, 1978a;
Soong and Ogura, 1980]. Turbulent mixing is handled in
the cloud model using a turbulent diffusion approximation.
[16] A Kessler-type scheme [Kessler, 1969; Houze, 1993]
for liquid hydrometeors (cloud water and rain) and the
three-category scheme of Lin et al. [1983] for solid hydro-
D14304
meteors (ice, snow, and hail) are employed to parameterize
cloud microphysics. The hydrometeors are assumed to be
spherical with exponential size distributions except for
cloud water and cloud ice, which are monodisperse. The
hydrometeors are advected with the air motion, mixed by
turbulent processes, and affected by deposition and microphysical transformations. A noniterative saturation adjustment scheme accounting for water and ice was implemented
by Tao et al. [1989].
[17] Short-wave and long-wave radiative transport is
implemented following Chou and Kouvaris [1991], Chou
and Suarez [1999], Chou et al. [1995, 1999], and Kratz et
al. [1998]. The spectral approximation is conducted on 16
spectral intervals in the solar and thermal infrared (IR). The
solar radiation scheme includes absorption due to water
vapor, CO2, O3, and O2, aerosols and cloud hydrometeors,
as well as Rayleigh scattering and scattering by clouds and
aerosols. The long-wave scheme accounts for the effects of
both gaseous (CO2, H2O, O3, N2O, CH4, CFCs) and
hydrometeor absorption.
[18] The GCE model uses a Cartesian grid with the
altitude over a flat surface as the vertical coordinate. It
originally did not account for terrain and assumes uniform
initialization from a single sounding. To apply the 3-D GCE
model for a case-study in a region with strong spatial
inhomogeneities, we have modified it to use nonuniform
initial distributions of the meteorological fields and account
for terrain effects, as we did previously with the 2-D GCE
model [DeCaria et al., 2000]. We account for the dynamical
effect of terrain by keeping all components of the velocity
equal to zero on and under terrain during the course of the
calculations. This effectively obeys a ‘‘no slip’’ boundary
condition at the surface and prevents any hydrodynamic
fluxes of energy, momentum, or mass through the terrain.
This approach is similar to the immersed boundary method
first proposed by Peskin [1981] that is now widely used for
calculating hydrodynamic flows with complex boundaries
using finite difference methods in Cartesian coordinates
[Iaccarino and Verzicco, 2003; Mittal et al., 2004].
[19] In this study we integrated the model in a domain of
360 km by 328 km in the x and y directions, respectively.
Positive x direction is from west to east, and positive y
direction is from south to north. The horizontal grid spacing
was 2 km in both horizontal directions, and 0.5 km in the
vertical. The grid dimensions were 180 164 50 in the x,
y, and z directions, respectively. For this spatial resolution,
the time step that allowed numerically stable calculations
was 3 s. We also applied additional damping at the lateral
boundaries to prevent growth of the convective disturbances
near the boundaries especially at the southern border of the
domain.
3. Observed Storm Development
[20] DeCaria et al. [2000] give a detailed description of
synoptic conditions for 12 July. Continuous radar coverage
was provided by the Colorado State University CHILL
radar at Greeley, Colorado (40.45N; 104.64W). The
National Oceanic and Atmospheric Administration
(NOAA) WP-3D aircraft gathered Doppler radar data,
flying legs parallel to the line of convection. The airborne
Doppler radar was scanned 20 forward and aft of the
3 of 14
D14304
STENCHIKOV ET AL.: 12 JULY STERAO STORM
D14304
Figure 1. Colorado State University CHILL radar 4.50-km constant altitude plan position indicator
(CAPPI) radar reflectivity factor (dBZ) obtained for the period (a) 2127 UTC, (b) 2207 UTC, (c) 2258 UTC,
and (d) 0113 UTC. The ordinate and abscissa refer to distance in kilometers from the radar facility located at
Greeley, Colorado (40.45N, 104.64W). The maximum reflectivity occurred at the position of the plus
symbol. Figures 1c and 1d have the WP-3D aircraft track superimposed.
flight track, which allowed the reconstruction of the 3-D
airflow field [Jorgensen et al., 1996]. The University of
North Dakota Cessna Citation II measured CO, NO, O3,
hydrocarbons, cloud physics, and meteorological data. The
WP-3D measured these variables plus a variety of additional chemical species. Visible satellite imagery is available during the entire period of storm development.
Figure 4 of DeCaria et al. [2000] shows a satellite
photograph with the anvil produced by the storm just north
of the Colorado-Wyoming border oriented in a west-to-east
direction at 2215 UTC. Strong convective activity is also
seen to the south from a separate storm in east-central
Colorado.
[21] The Colorado State University CHILL radar imagery
(Figure 1) shows that during the early stages of growth from
2046 UTC to 2141 UTC, multiple, relatively weak, separated convective cells were located in southeastern Wyoming (Figure 1a). Between 2141 UTC and 2212 UTC,
convective cells strengthened and organized in a linear
structure with the highest reflectivity gradients generally
located along the southern flank (Figure 1b). By 2258 UTC
the storm developed a four-cell spatial structure with the
strongest convection at the eastern flank (Figure 1c). The
storm moved to the east-southeast with the leading edge
reaching the Wyoming-Colorado border by 2304 UTC. The
convection evolved further from an elongate structure into a
4 of 14
D14304
STENCHIKOV ET AL.: 12 JULY STERAO STORM
more oblate form containing a strong reflectivity signature
at the western periphery of the cloud complex between
2258 UTC and 2330 UTC. By 2330 UTC, the convection
acquired more of a southeast propagation with a speed of
about 8 m/s. In addition to this somewhat deviant motion,
the radar reflectivity pattern developed a V-signature in
the midtroposphere clearly indicating a 3-D storm-scale
circulation characteristic of severe convection. Strong
reflectivity gradients remained at the southwestern edge
of the storm with an extensive anvil blowing off to the
northeast. The northeastern-most convective cell dominated
and by 0113 UTC on 13 July only this single intense
convective cell remained (Figure 1d). The separate convective activity to the south is also clearly seen on the radar image
for 0113 UTC (Figure 1d).
4. Initialization of Convection
[22] The thermodynamic diagram of temperature, dew
point, and wind composited from the soundings at Fort
Morgan, Colorado, at 1956 UTC and aircraft data taken by
the NOAA WP-3D at 2200 UTC ahead of the storm shows
instability in the lower troposphere [see DeCaria
et al., 2000, Figure 3]. However, the CAPE was as low
as 700 m2 s2, which is almost a factor of 2.5 less than on
10 July when it reached 1850 m2 s2. The significant
portion of instability on 12 July could be attributed to the
strong wind shear that reached 35 m s1 in the 6-km
layer above the terrain in the region where the storm
initially developed. The bulk Richardson number of
approximately 7 supports development of supercell convection (if CAPE is sufficient).
[23] We were unsuccessful at simulating a sustainable
storm using uniform initial conditions based on the composite soundings from DeCaria et al. [2000] with either the
2-D or 3-D GCE model. In this study we have developed
3-D nonuniform initial fields and accounted for the effects of
terrain to simulate a storm similar to the observed one using a
modified version of the 3-D GCE model.
[24] To construct nonhomogeneous initial conditions for
our simulations we utilized routine forecast and analysis
fields from the NCEP Eta Model [Mesinger et al., 1988;
Rogers et al., 1996]. The Eta Model data were provided by
NCEP in gridded binary (GRIB) format on an irregular Etamodel horizontal grid with resolution about 32 km. The Eta
Model data included surface pressure, surface elevation, and
3-D fields of pressure, temperature, water vapor mixing
ratio, and horizontal wind components on 26 pressure levels
for the entire United States. These data were first interpolated to a regular horizontal grid with 0.5 grid spacing and
vertical grid identical to that used in the cloud model. After
choosing a 3-D domain for the cloud model simulations, the
fields are then projected onto the cloud model’s horizontal
grid.
[25] Although a finer terrain profile could have been
used, the Eta Model terrain profile was chosen in order to
keep the initial meteorological fields, which are also derived
from the Eta Model, dynamically consistent at the lower
boundary. The pressure field is adjusted to make it consistent with the GCE model formulation. This helps to decrease the generation of gravity and sound waves during an
approximately 15-min spin-up period. Applying nonuni-
D14304
form initial conditions and corresponding boundary conditions allows us to account for interaction of the storm with
the mesoscale features, (e.g., upslope flow, nonuniform
spatial distribution of CAPE, and wind shear) which developed in the region because of larger-scale circulation
processes not accounted for in our calculations. Interactions
with mesoscale features appear to play an important role in
thunderstorm life cycle, as does the mesoscale response to
the deep convection that modifies CAPE and wind shear
background distributions in the calculation domain [e.g.,
Donner et al., 1999].
[26] For our relatively short-term simulations we have
chosen initial fields from the middle of the period of storm
development to better characterize the entire period. The Eta
analysis fields were not available for this particular time.
Therefore we constructed the initial conditions using most
suitable Eta forecast fields. The most recent available data
were forecast fields from the 1200 UTC 12 July Eta Model
run. Eta Model data were extracted from the 6-hour forecast
(valid 1800 UTC 12 July) and the 12-hour forecast (valid
0000 UTC 13 July.) The 12-hour forecast valid at 0000
UTC 13 July was chosen for initial conditions over the 6-hour
forecast valid at 1800 UTC because the Eta Model was slow in
developing the post-frontal upslope flow, a critical component for convection along the mountains of the Front Range.
Observations from Laramie, Wyoming, (located near
105.6W) at 1800 UTC showed a 4 m s1 easterly wind
component, while an easterly wind component was absent in
the Eta 6-hour forecast valid at that time at Laramie’s
longitude. In contrast, the 12-hour Eta forecast fields did
contain an easterly component at Laramie’s longitude, and
were in much closer agreement with the observations at
1800 UTC. The 12-hour forecast fields also verified
better than the 6-hour fields for the zonal wind speed
above 400 hPa. Temperature and surface pressure from
the 12-hour Eta forecast are also in a good agreement
with observations. However, water vapor mixing ratio is
slightly underestimated [DeCaria et al., 2000].
[27] Figure 2 shows the Eta fields used for initial and
boundary conditions. The fields are shown in the domain
used for our 3-D GCE simulations. Figure 2a depicts the
land elevation (shown by isolines with contour interval of
100 m) and vectors of surface wind. The easterly upslope
flow tends to force convection at the initial phase of the
storm evolution. The maximum value of CAPE (Figure 2b)
in the storm region is about 700 m2 s2, which is
consistent with the composite soundings from DeCaria et
al. [2000]. CAPE reaches its maximum value in the relatively narrow region elongated along the mountain ridge.
[28] In the 3-D simulations convection (in accordance
with observations) has been initiated at the border of the
region with relatively high CAPE at about (41.1N,
104.5W) near the northwest corner of the model domain
(Figure 2b). CAPE in this region was only 600 m2 s2
which is barely enough to support initial storm development. This explains why the simulation was sensitive to the
position where the storm is initialized. If the position of
initial disturbances is shifted to the region where CAPE is
500 m2s2, the storm would not develop. The CAPE
distribution also plays an important role in the long-term
storm development. The southeastward propagation of the
storm follows the higher values of CAPE. As soon as
5 of 14
D14304
STENCHIKOV ET AL.: 12 JULY STERAO STORM
D14304
spatial structure of the convection. Therefore we modified
the GCE model to include the effects of terrain and the
capability to use nonuniform initial fields as discussed
above. For the 3-D simulations, we have to choose a smaller
domain than in our 2-D calculations. However, it covers an
area of 380 km 328 km which is large enough to follow
the storm evolution for about 5 hours until it dissipates. The
simulations start from the initial conditions constructed
from the Eta fields for 0000 UTC (Figure 2). Horizontal
resolution was 2 km in both the east-west and north-south
directions. Vertical spacing was 500 m with 50 vertical
levels. As in the work by Skamarock et al. [2000], we did
not include in this simulation the effects of radiation and of
latent and turbulent surface heat fluxes, as they are of
secondary importance for short-term convective processes
developing in an unstable meteorological environment, such
as that considered here.
[30] Because the simulated 12 July storm moved to the
southeast (consistent with observations), the domain was
shifted in this direction and convection was initialized in the
northwestern corner of the domain to allow convection
sufficient space to proceed to the southeast as in the
observations. A three-cell east-west-oriented linear convective structure was initialized with the potential temperature
perturbation Dq,
Dq ¼ dq 3
X
h
i
exp ð x xi Þ2 =s2x ð y yi Þ2 =s2y ð z zi Þ2 =s2z ;
i¼1
where xi, yi, zi are coordinates of the center of perturbations
measured from the south-west corner of the domain and
from sea level
x1 ¼ 104:75 W; y1 ¼ 41:1 N; zi ¼ 3500 m;
Figure 2. Eta Model fields used for initialization of the
GCE model simulations. (a) Land elevation (m) and surface
wind vectors. (b) CAPE (J kg1 m2). The domain
corresponds to the domain used in our 3-D calculations.
the storm propagates away from the region with CAPE
above 600 m2s2, it dissipates quickly. For example, in the
2-D simulation conducted along 41N the storm moved
rapidly out of the area with high CAPE very quickly and
collapsed much earlier than was observed. The regions of
highest values of CAPE farther south in the domain
correspond to the location of separate strong convective
activity seen on the satellite and radar images (see
DeCaria et al. [2000, Figure 4] and Figure 1).
5. 3-D GCE Calculations
5.1. Model Configuration
[29] We found that the 2-D GCE model failed to describe
the extended evolution of the storm when it obtained a 3-D
structure after about 2300 UTC [DeCaria et al., 2000]. The
2-D model also produced a storm that moved much faster
than the observed one. This is mainly because the 3-D
interactions of the storm with the terrain and mesoscale
circulation define direction and speed of the storm propagation, and led to the development of the observed fine
x2 ¼ 104:50 W; y2 ¼ 41:1 N; z2 ¼ 3500 m;
x3 ¼ 104:25 W; y3 ¼ 41:1 N; z3 ¼ 3500 m:
The widths and amplitude of perturbations are sx = 5 km,
sy = 5 km, sz = 1.5 km, and dq = 4 K.
[31] The position and the structure of the initial perturbation in the model corresponds to the position of the
multicellular line which evolved between 2046 and
2120 UTC (Figure 1a). In the following 5 hours the
observed storm proceeded to northeastern Colorado and
developed into a 3-D rectangular four-cell structure that
evolved into a single intense cell, and then dissipated. In
our 3-D simulations we focused on the period of storm
evolution that followed the initial stage studied by
DeCaria et al. [2000] onward to its dissipation at about
0130 UTC on 13 July 1996 (see Figure 1).
[32] Some numerical complications caused by using nonuniform initial fields were associated with the large CAPE
at the southern part of the model domain (Figure 2b).
This region is not of primary interest but was included to
allow our simulated storm sufficient space to develop and
move southeastward. As a result, a very strong convective
system developed at the southern boundary consistent with
observations (see Figure 1d). That system interfered with
the boundary conditions causing numerical instability
6 of 14
D14304
STENCHIKOV ET AL.: 12 JULY STERAO STORM
D14304
Figure 3. Simulated radar reflectivity factor (dBZ) at z = 4.5 km from 3-D GCE model run at (a) 40 min
of calculations (model time corresponds to 2200 UTC) and (b) 90 min (model time corresponds to
2250 UTC), (c) 120 min of calculations (model time corresponds to 2320 UTC), and (d) 150 min
(model time corresponds to 2350 UTC). Chill radar is located at x = 59.52 km, y = 172.05 km.
(Figure 3). We extended the domain as far south as available
computer RAM allowed, but a very large domain would be
necessary to move the southern boundary to a convectively
stable region. Therefore we applied nudging type boundary
conditions in a five-point belt along the lateral boundaries
[Davies, 1976; Robert and Yakimiw, 1986; Yakimiw and
Robert, 1990]. In this boundary belt the three velocity
components were relaxed to their initial values. That helped
to stabilize the numerical solution for the entire period of the
simulation. We believe it is important to keep this region of
convective instability in the southern part of the domain
because it allows us to more realistically calculate the time
evolution of the mesoscale flow that interacts with the 12
July STERAO storm.
5.2. Results
[33] To evaluate the simulated storm structure, we conducted comparisons of calculated radar reflectivity distributions in the middle and upper troposphere with the CHILL
radar reflectivities and in situ aircraft observations. The
atmospheric flow in the single intense cell regime has been
evaluated using Doppler radar from WP-3D measurements
that are available only at the final stage of the storm
evolution.
[34] Figure 3 shows the simulated radar reflectivity in the
middle troposphere at the 4.5-km level (for visualization
purposes only the part of the simulation domain where
the storm is developing is shown) for comparison with the
CHILL radar observations in Figure 1. On the basis of the
CHILL radar observations discussed above and lightning
observations presented by DeCaria et al. [2005], we assume
that the beginning of the run corresponds to 2030 UTC. The
very early stages of storm development proceeded more
rapidly in the model than was observed. Therefore we have
advanced the times of our comparisons with observations by
50 min. At 40 min after the beginning of the simulation
(roughly corresponds to 2200 UTC), we see in Figure 3a a
linear convective structure similar to the radar image at
2207 UTC in Figure 1b. The middle cell shows two closely
located reflectivity maxima exhibiting a minor tendency for
splitting. However, at 90 min (corresponds to 2250 UTC)
this linear structure has fully evolved into the four-cell
rectangular structure as in observations at 2258 UTC
(Figure 1c). Consistent with observations the rectangle has
7 of 14
D14304
STENCHIKOV ET AL.: 12 JULY STERAO STORM
D14304
Figure 4. Simulated radar reflectivity factor (dBZ) at z = 10 km from 3-D GCE model run at (a) 120 min
of calculations (model time corresponds to 2320 UTC), (b) 150 min (model time corresponds to
2350 UTC), (c) 190 min (0030 UTC on 13 July), and (d) 210 min (0050 UTC on 13 July). Chill radar is
located at x = 59.52 km, y = 172.05 km.
dimensions of about 40 20 km with the longer sides
slightly rotated to the southeast. At the southern border of
the domain we see disturbances that propagate from the
region of instability. The four-cell structure continues to
move southeastward (Figure 3c and 3d). The southeasternmost cell is gradually becoming the most intense. The other
cells are decaying because the southeasternmost cell first
processed the CAPE and left a more stable environment for
the other cells. In the observations, there is clear competition between the two easternmost cells with the northeasternmost cell being slightly stronger (Figure 1c).
[35] After the very initial stage of development, the
overall time evolution of the storm and the speed of
propagation are very similar to observations. For 150 min
of evolution (from 40 to 190 min of simulation) shown in
Figure 3 and 4, the southeasternmost cell that developed
into a marginal supercell moved by about 35 km to the
south and 60 km to the east, covering a distance of 70 km
with average speed of 7.8 m/s. This is in a very good
agreement with the observed storm propagation speed
derived from CHILL radar observations (Figure 1). The
3-D model captures the directionality and the speed of
propagation of the storm much better than in the 2-D
simulation where the storm moved 2 times faster than in
observations.
[36] Figure 4 shows the evolution of the simulated
reflectivity in the upper troposphere at 10 km (approximately
the middle of the detrainment layer) starting from 120 min of
storm simulation onward to the supercell formation. At 120
min all four cells are seen at 10 km, but the cells at the southern
flank of the rectangular system are obviously stronger. They
are propagating in the environment with larger CAPE than the
cells at the northern flank of the system. The southeasternmost cell gradually became the strongest one. This cell
dominates, producing the largest anvil with total hydrometeor
mixing ratios of more than 5 g/kg and simulated reflectivity
about 50 – 55 dBZ. At 190 min into the simulation
(corresponding to 0030 UTC on 13 July) the storm developed
into a single intense cell (or marginal supercell) consistent
with the radar observations at 0113 UTC (Figure 1d). As in
Figure 1d the simulation also shows clear interaction with the
convection at the southern border of the domain. The simulated single intense cell moves southeastward to the area with
low CAPE and transforms slightly faster than in observations.
The convective cell is still relatively strong at 210 min
(roughly corresponding to 0050 UTC) as seen in Figure 4d
8 of 14
STENCHIKOV ET AL.: 12 JULY STERAO STORM
D14304
D14304
Table 1. Maximum Upward CO Flux (g m2 s1) Calculated by DeCaria et al. [2000] Using 2-D GCE Model Compared With That
Obtained in This Study Using 3-D GCE Model
2100 UTC
2200 UTC
Altitude
2D
3D
3 km
5 km
7 km
9 km
11 km
13 km
2.33 104
4.72 104
4.55 104
9.42 105
no convective influence
no convective influence
2.85 104
6.79 104
8.24 104
6.86 104
3.00 104
no convective
influence
2300 UTC
2D
3.65
6.06
1.13
1.16
7.37
1.62
with the anvil blowing to the northeast as in the observations (Figure 1d). The simulated convection dies in about
30 min (at 0120 UTC on 13 July) slightly earlier than
observed. However, the last stage of the storm simulation
(after 210 min) is less realistic. The convective cell leaves
the region with high CAPE at about 220 min and begins to
decay and to strongly accelerate to the east, presumably
because of interaction with the mesoscale flow modified
by other convective processes developing in the domain.
This analysis shows that the initial nonuniform distribution
of CAPE and the mesoscale transformation of the flow in
the entire domain, as affected by the STERAO storm and
the strong convection in the southern part of the domain
(that modify CAPE and wind shear), are the most important factors that define the long-term evolution and the fine
structure of the storm.
[37] This analysis shows that 3-D dynamics produce a
storm structure significantly different compared with the
storm structure that DeCaria et al. [2000] obtained in the
2-D simulations. The 3-D flow has more degrees of
freedom and allows for more complex storm development. For example, the flow in the upper troposphere in
2-D simulations tends to move above and below the anvil
enhancing vertical mass exchange. In the 3-D simulations,
upper-troposphere flow tends to move around the anvil in
a horizontal plane without production of a significant
vertical wind component. The storm has more flexibility
to develop in the 3-D nonuniform background and
appears to be stronger than the 2-D storm. The mesoscale
structure introduced by initial conditions is vitally important to storm evolution because we were even unable to
produce sustainable convection using uniform initialization from soundings. In this case the dynamic instability
in the lower troposphere adds energy to sustain strong
convection. The more flexible 3-D dynamics allowed the
storm to take advantage of more of the available instability than in the 2-D simulations. The wind shear in the
upper troposphere defines the evolution of the anvil, the
horizontal transport of pollutants, and their dilution rate.
However, the processes in the anvil are also very sensitive to the microphysical parameterization.
[38] The stronger 3-D convection translates into a more
intensive vertical transport of pollutant tracers from the
boundary layer into the upper troposphere. Tables 1 and 2
show, respectively, maximum vertical flux and mixing ratio
of CO in the 2-D and 3-D simulations for three subsequent
times at various altitudes. The 3-D convection consistently
produced larger fluxes and higher concentrations of CO at
all altitudes above the boundary layer.
3D
104
104
103
103
104
104
6.58
9.76
1.22
1.29
8.75
1.74
2D
104
104
103
103
104
104
1.74
2.62
5.18
3.06
1.08
3.05
3D
104
104
104
104
104
105
5.90
9.42
1.19
1.14
8.06
2.49
104
104
103
103
104
104
[39] To analyze the fine spatial and dynamic structure of
the convection in the single cell stage of the 12 July storm,
we consider the vertical cross section (Figure 5) that goes
through the core of the cell at y = 160 km at 190 min into
the calculations (Figure 4c). We evaluate the simulation by
comparing the results with the airborne Doppler radar
observations. Because of the slightly faster development
of the simulated storm at this stage we have chosen for
comparison the airborne Doppler radar observations at
0105 UTC on 13 July 1996 that are approximately 30 min
later than the simulated time.
[40] In Figure 5a we see that the simulated radar reflectivity reaches a maximum value of 50– 55 dBZ at 4 – 5 km
which corresponds to the altitude of the approximately
60-dBZ observed maximum. However, the simulated
reflectivity (55 dBZ) at 9 –10 km is greater than in
observations. The top of the convection is between 12
and 13 km, and the axis of the convective cell is at about
60 to the horizon. The simulated vertical velocity in the
updraft (Figure 5b) exceeds 14 m s1 and reaches a
maximum at 9 km. The air in the downdrafts in the
middle and upper troposphere descends with the speed of
2 to 4 m s1. The relatively intense downward
motion in the lower troposphere is produced by an inflow
in the convective cell. We will discuss it in more detail below.
[41] The width of the updrafts in the convection is about
10 km. The cloud top altitude at the period of maximum
intensity reaches 13 km as in the observations. The maximum vertical velocity in the course of the simulated 12 July
storm evolution was about 20 m s1, which is consistent
with the WP-3B radar observations. All these characteristics
are weaker than for the 10 July storm (which produced
vertical velocities of more than 30 m s1). This result is
consistent with the less unstable meteorological environment on 12 July.
Table 2. Maximum CO Mixing Ratio (ppbv) Calculated by
DeCaria et al. [2000] Using 2-D GCE Model Compared With That
Obtained in This Study Using 3-D GCE Model
2100 UTC
Altitude
2D
3 km
5 km
7 km
9 km
11 km
128.67
124.87
119.06
104.42
no convective
influence
no convective
influence
13 km
9 of 14
2200 UTC
3D
2300 UTC
2D
3D
2D
3D
126.52
122.86
121.49
123.26
128.03
130.99
130.99
130.99
130.99
130.99
124.08
118.65
119.62
118.05
121.74
131.36
131.36
131.36
131.36
131.36
no convective 107.60 116.27
influence
96.51
128.53
130.46
130.46
130.46
130.04
122.65
D14304
STENCHIKOV ET AL.: 12 JULY STERAO STORM
D14304
Figure 5. Vertical cross section from simulations at y = 160 km at 190 min (corresponds to 0030 UTC
on 13 July). (a) Radar reflectivity factor (dBZ) and (b) vertical velocity w (m/s). Vertical cross section of
Doppler radar observations from WP-3D aircraft at 0105 UTC on 13 July. (c) Radar reflectivity (dBZ)
and (d) vertical velocity (m/s).
[42] Figures 5c and 5d show observed reflectivity and
vertical velocity from the WP-3D radar. In the observations
the updraft is offset with respect to the maximum of
reflectivity. The core of the convection is located 10 km
east of the location of maximum rainwater, which is in the
subsiding air in the lower troposphere. The intensity of the
updraft oscillates in time reaching a maximum value about
20 m s1, but at this moment it is weaker than in the
simulation. However, the maximum altitude of the reflectivity distribution moves toward the updraft current position
and reaches about 13 km ASL.
[43] The simulated horizontal wind convergence is shown
in Figure 6a. A region of large positive convergence
indicating the most intensive inflow is located in the
4 – 6 km layer. Above this level the air is moving upward
(see Figure 5b) while below it the air is moving downward and hits the ground, producing the diverging flow
along the Earth’s surface seen in the lowest kilometer.
Convergence at 5 km is fairly large reaching 4 103 s1.
Integrating convergence over a 10 10 km horizontal region
and 1-km depth layer we computed that inflow is about 1.5 108 kg s1. The negative convergence in the upper troposphere (UT) shows that anvil exhaust from the convective cell
maximizes at 10 km. The secondary convergence regions
seen at 11 and 14 km probably correspond to generation of
gravity waves by the convection.
[44] Strong wind convergence generates cyclonic rotation. This positive angular momentum is advected upward
with the airflow and stabilizes the convective updraft. The
vertical component of vorticity is especially strong in the
middle and upper part of the convective cell at altitudes of
5– 12 km. The vertical vorticity component (Figure 6b)
reaches 8 103 s1, which is equivalent to an angular
velocity of about 4 103 s1 and a revolution time around
the center of the cell of about half an hour. In the region of
the downward flow below the level of inflow negative
vorticity is generated.
[45] Figures 6c and 6d show convergence and the vertical
component of vorticity, respectively, calculated using WP3D Doppler radar wind fields. Because of the lack of
observations, the entire picture cannot be seen. The observed convergence and vorticity are weaker than calculated
in the same proportion as observed vertical velocity is
weaker than calculated, but the spatial structure is practically identical. In Figure 6c we see positive convergence at
4 – 5 km level and negative convergence at 7 – 11 km.
10 of 14
D14304
STENCHIKOV ET AL.: 12 JULY STERAO STORM
D14304
Figure 6. Vertical cross section from simulations at y = 160 km at 190 min (corresponds to 0030 UTC
on 13 July): (a) convergence of horizontal velocity (103/s) and (b) vertical component of vorticity
(103/s). Vertical cross section of Doppler radar observations from WP-3D aircraft at 0105 UTC on
July 13: (c) convergence of horizontal velocity (103/s) and (d) vertical component of vorticity (103/s).
Positive vorticity is seen above the inflow level at 6 km and
negative vorticity below.
[46] In Figures 7a and 7b we show horizontal cross
sections at the levels of maximum inflow (5 km) and
maximum outflow (10 km). Figure 7a shows convergence
and horizontal velocity vectors. We see that most intensive
inflow is from the north and northwest. Below the convergence region the wind is blowing to the southeast. This
structure is similar to the observed shown in Figure 7c that
also shows a strong rear-inflow jet.
[47] Figures 7b and 7c show the vertical component of
vorticity and horizontal wind vectors at 10 km. We see that
the main anvil exhaust from the convective cell is to the east
both in simulations (Figure 7b) and in observations
(Figure 7c). However, the simulation also produced a
moderate northward component that is absent in observations. The structure of the observed and simulated vorticity fields is similar. It shows a south-north-oriented
dipole with positive vorticity at the core of the convective
updraft and negative vorticity to the north. Negative
vorticity is stronger in the observations.
[48] The simulated turbulent diffusion coefficient was
relatively large (reaching 600 m2/s) at the level of inflow at
6 km (Figure 8). The turbulent mixing was assumed to be
isotropic. The region of strong turbulence has dimensions
of 25 20 km, which was much larger than the cross
section of the updraft. At this level, turbulent kinetic
energy (TKE) was generated both by thermal instabilities
and wind shear. It is important to note that in the anvil at
12 km (Figure 8b) the turbulent diffusion coefficient was
still relatively large, reaching 320 m2/s, which corresponds
to a mixing length of more than 100 m. Intense turbulence
in the anvil was maintained by vertical advection of TKE
and dynamic instabilities enhanced by oscillations of the
convective cell. The large values of the coefficient in the
anvil indicate that turbulence could significantly contribute
to redistribution of chemically active compounds transported by convection to the upper troposphere. This
process must be accounted for in the transport calculations.
Turbulence at the top of the convective cell even could
affect stratosphere-troposphere exchange. However, in the
12 July storm the strength and altitude of the convective
cell is not enough (tropopause was at 15 km) to cause a
sizable stratosphere-troposphere mass exchange (contrary
to the 10 July storm). Consistently, ozone measurements
from the UND Citation aircraft did not show any intrusions
11 of 14
STENCHIKOV ET AL.: 12 JULY STERAO STORM
D14304
D14304
Figure 7. Horizontal cross section from simulations at 190 min (corresponds to 0030 UTC on 13 July):
(a) convergence of horizontal velocity (103/s) and horizontal wind vectors at 5 km and (b) vertical
component of vorticity (103/s) and horizontal wind vectors at 10 km. Horizontal cross section of
Doppler radar observations from WP-3D aircraft at 0105 UTC on 13 July: (c) convergence of horizontal
velocity (103/s) and horizontal wind vectors at 5 km and (d) vertical component of vorticity (103/s) and
horizontal wind vectors at 10 km.
of stratospheric air with high ozone mixing ratios in the
upper troposphere.
[49] In the above discussion we characterized the terminal
phase of the 12 July storm as a single intense cell or
marginal supercell regime. This is consistent with the
relatively weak CAPE observed on 12 July. However, we
have shown that the convective cell possessed a rotating
updraft and developed in a strong wind shear environment
(35 m s1 in the 6-km layer above terrain). CHILL radar
images also show that between 0058 and 0214 on 13 July
there are occasions when V-shaped indentations (‘‘notches’’)
developed in the reflectivity field. These may be indicative of
the weak echo region that is a characteristic feature of a
supercell convection. Therefore we can characterize the
12 July single cell convection as a single intense cell
marginally reaching supercell status.
6. Summary
[50 ] Upward convective transport occurs in narrow
updrafts on very fine spatial scales of atmospheric flow.
The dynamical fields required for a posteriori simulation of
this transport cannot be obtained from observations and
have to be reproduced in model simulations. The fine
structure of the convective flow and turbulent diffusion
are crucially important for correct simulation of transport
of chemical species from the boundary layer and LT into the
UT/LS.
[51] Here we used the 3-D version of the GCE model to
simulate the 12 July STERAO storm. To correctly describe
the spatial structure of this storm, we implemented terrain
and nonuniform initial conditions in the 3-D version of the
GCE model. We showed that with these modifications the
3-D GCE model is capable of simulating the transition from
the linear multicell convection to a highly 3-D single severe
cell. The archived meteorological fields from the cloud
model simulation were evaluated using available observations and used as input to the Cloud-Scale Chemical
Transport Model (CSCTM) [DeCaria et al., 2005].
[52] Mesoscale models like RAMS, MM5, ARPS, and
WRF provide capabilities for accounting for terrain and
nonuniform initial fields that allow them to better reproduce
12 of 14
D14304
STENCHIKOV ET AL.: 12 JULY STERAO STORM
D14304
[56] 3. The 2-D approach does not allow us to describe
the entire evolution of the storm, but does provide useful
estimates of the storm development during the quasi-linear
development phase of the storm for about the first 2 hours.
The 3-D convection, consistently with observations, appears
to be stronger than 2-D convection and initiated more
intensive vertical transport of tracers from the boundary
layer into the upper troposphere.
[57] 4. The 12 July STERAO storm is significantly
different than the one on 10 July since it developed in an
environment with much lower CAPE. The structure of
simulated convection for this case appears to be much more
sensitive to the detailed spatial distribution of initial meteorological fields, terrain effects, and boundary conditions.
However, the 12 July storm produced relatively strong
convection significantly affecting the chemical structure of
the upper troposphere [see DeCaria et al., 2005], although it
did not cause significant stratosphere-troposphere exchange.
This study shows that parameterization of convection in
the marginally unstable environment might need special
attention.
Figure 8. Horizontal cross section from simulation at
190 min (corresponds to 0030 UTC on 13 July): (a) turbulent
diffusion coefficient (m2/s) at 6 km and (b) turbulent diffusion
coefficient (m2/s) at 12 km.
mesoscale development interactively with the storm evolution. Here we showed that accounting for these effects
improves a GCE model simulation. We found that dynamic
fields from the GCE model output every 10 min provide a
well-balanced input into our cloud-scale chemical transport
model, not producing any spurious sources or sinks of mass.
No ‘‘mass-fixing’’ routine, that we have found necessary
when using output from mesoscale models, was needed in
this case.
[53] The major results of this study can be formulated as
follows.
[54] 1. The GCE model with terrain and nonuniform
initialization was able to reproduce the correct observed
development of the 12 July 1996 STERAO storm. The
simulation produced storm spatial structure and temporal
evolution that compare favorably with observations. Direction and speed of storm propagation are in good agreement
with those derived from radar observations.
[55] 2. The 3-D simulation allowed reproduction of the
entire life cycle of the storm from linear multicell structure
to highly 3-D intense single cell (or marginal supercell
convection). However, it seems in our simulation that the
southeastern (rather than the northeastern as observed)
convective cell became the most intense. This may have
resulted because interaction between the cells is highly
unstable and small sporadic perturbations can cause intensification of one cell at the expense of another. It suggests
that some assimilation of observed information could be
beneficial.
[58] Acknowledgments. This research was supported under National
Science Foundation grants ATM0004120 and ATM 9912336. We thank Jim
Dye of the National Center for Atmospheric Research (NCAR), Adrian
Tuck of the NOAA Aeronomy Laboratory, Steve Rutledge of Colorado
State University (CSU) for organizing and coordinating the STERAO-A
field project, Steve Lang of NASA Goddard for help in transferring the
GCE model to different computer platforms, and Roald Akberov for help
with data processing. R. Kakar at NASA headquarters is acknowledged for
his support of Goddard Cumulus Ensemble model improvements and
applications. W.-K. Tao is mainly supported by the NASA Headquarters
Physical Climate Program and the NASA Tropical Rainfall Measuring
Mission (TRMM).
References
Baker, R. D., B. H. Lynn, A. Boone, W.-K. Tao, and J. Simpson (2001),
The influence of soil moisture, coastline curvature, and land-breeze
circulation on sea-breeze-initiated precipitation, J. Hydrometeorol., 2,
193 – 211.
Barth, M. C., A. L. Stuart, and W. C. Skamarock (2001), Numerical simulations of the July 10, 1996, Stratosphere-Troposphere Experiment:
Radiation, Aerosols, and Ozone (STERAO)-Deep Convection experiment storm: Redistribution of soluble tracers, J. Geophys. Res., 106,
12,381 – 12,400.
Boe, B., et al. (1992), The North Dakota Thunderstorm Project: A cooperative study of high plains thunderstorms, Bull. Am. Meteorol. Soc., 73,
145 – 160.
Boone, A., and P. J. Wetzel (1999), A simple scheme for modeling subgrid soil texture variability for use in an atmospheric climate model,
J. Meteorol. Soc. Jpn., 77, 317 – 333.
Chatfield, R. B., and P. J. Crutzen (1984), Sulfur dioxide in remote oceanic
air: Cloud transport of reactive precursors, J. Geophys. Res., 89, 7111 –
7132.
Chou, M.-D., and L. Kouvaris (1991), Calculations of transmission functions in the IR CO2 and O3 bands, J. Geophys. Res., 96, 9003 – 9012.
Chou, M.-D., and M. J. Suarez (1999), A shortwave radiation parameterization for atmospheric studies, 15, NASA/TM-104606, 40 pp., Natl.
Aeronaut. and Space Admin., Washington, D. C.
Chou, M.-D., W. Ridgway, and M.-H. Yan (1995), Parameterizations for
water vapor IR radiative transfer in both the middle and lower atmospheres, J. Atmos. Sci., 52, 1159 – 1167.
Chou, M.-D., K.-T. Lee, S.-C. Tsay, and Q. Fu (1999), Parameterization for
cloud longwave scattering for use in atmospheric models, J. Clim., 12,
159 – 169.
Davies, H. C. (1976), A lateral boundary formulation for multi-level prediction models, Q. J. Meteorol. Soc., 102, 405 – 418.
Deardorff, J. W. (1975), The development of boundary layer turbulence
models for use in studying the severe storm environment, paper presented
at SESAME Opening Meeting, Natl. Oceanic and Atmos. Admin., Silver
Spring, Md.
DeCaria, A. J., K. E. Pickering, G. L. Stenchikov, J. R. Scala, J. L. Stith,
J. E. Dye, B. A. Ridley, and P. Laroche (2000), A cloud-scale model
13 of 14
D14304
STENCHIKOV ET AL.: 12 JULY STERAO STORM
study of lightning-generated NOx in an individual thunderstorm during
STERAO-A, J. Geophys. Res., 105, 11,601 – 11,616.
DeCaria, A. J., K. E. Pickering, G. L. Stenchikov, and L. E. Ott (2005),
Lightning-generated NOx and its impact on tropospheric ozone production: A three-dimensional modeling study of a Stratosphere-Troposphere
Experiment: Radiation, Aerosols and Ozone (STERAO-A) thunderstorm,
J. Geophys. Res., 110, D14303, doi:10.1029/2004JD005556.
Dickerson, R. R., et al. (1987), Thunderstorms: An important mechanism in
the transport of air pollutants, Science, 235, 460 – 465.
Donner, L. J., C. J. Seman, and R. S. Hemler (1999), Three-dimensional
cloud-system modeling of GATE convection, J. Atmos. Sci., 56, 1885 –
1912.
Dye, J. E., et al. (2000), An overview of the Stratospheric-Tropospheric
Experiment: Radiation, Aerosols, and Ozone (STERAO)-Deep Convection experiment with results for the July 10, 1996 storm, J. Geophys.
Res., 10,023 – 10,045.
Holton, J. R. (1982), The role pf gravity wave induced drag and diffusion in
the momentum budget of the mesosphere, J. Atmos. Sci., 39, 791 – 799.
Houze, R. A., Jr. (1993), Cloud Dynamics, 573 pp., Elsevier, New York.
Iaccarino, G., and R. Verzicco (2003), Immersed boundary technique for
turbulent flow simulations, Appl. Mech. Rev., 56, 331 – 347.
Johnson, D., W.-K. Tao, J. Simpson, and C.-H. Sui (2002), A study of the
response of deep tropical clouds to mesoscale processes: Part I. Modeling
strategy and simulation of TOGA COARE convective systems, J. Atmos.
Sci., 59, 3492 – 3518.
Jorgensen, D. P., T. Matejka, and J. D. DuGranrut (1996), Multi-beam
techniques for deriving wind fields from airborne Doppler radars,
Meteorol. Atmos. Phys., 59, 83 – 104.
Kessler, E. (1969), On the distribution and continuity of water substance in
atmospheric circulations, Meteorol. Monogr., 10(32), 84 pp.
Klemp, J. B., and R. B. Wilhelmson (1978a), Simulations of right- and leftmoving storms produced through storm splitting, J. Atmos. Sci., 35,
1097 – 1110.
Klemp, J. B., and R. B. Wilhelmson (1978b), The simulation of threedimensional convective storm dynamics, J. Atmos. Sci., 35, 1070 – 1096.
Kratz, D. P., M.-D. Chou, M.-H. Yan, and C.-H. Ho (1998), Minor trace gas
radiative forcing calculations using the k-distribution method with oneparameter scaling, J. Geophys. Res., 103, 31,647 – 31,656.
Lin, Y.-L., R. D. Farley, and H. D. Orville (1983), Bulk parameterization of
the snow field in a cloud model, J. Clim. Appl. Meteorol., 22, 1065 –
1092.
Lynn, B. H., and W.-K. Tao (2001), A parameterization for triggering of
landscape generated moist convection: Part II. Zero order and first order
closure, J. Atmos. Sci., 58, 593 – 607.
Lynn, B. H., W.-K. Tao, and F. Abramopoulos (2001), A parameterization
for the triggering of landscape moist convection: Part I. Analysis of high
resolution model results, J. Atmos. Sci., 58, 575 – 592.
Mesinger, F., Z. I. Janjic, S. Nickovic, D. Gavrilov, and D. G. Deaven
(1988), The step-mountain coordinate: Model description and performance for cases of Alpine lee cyclogenesis and for a case of Appalachian
redevelopment, Mon. Weather Rev., 116, 1493 – 1518.
Mittal, R., V. Seshardri, and H. S. Udaykumar (2004), Flutter, tumble and
vortex induced autorotation, Theor. Comput. Fluid. Dyn., 17, 165 – 170.
Peskin, C. S. (1981), The fluid dynamics of heart valves: Experimental,
theoretical and computational methods, Annu. Rev. Fluid Mech., 14,
235 – 259.
Pickering, K. E., A. M. Thompson, R. R. Dickerson, W. T. Luke, D. P.
McNamara, J. Greenberg, and P. R. Zimmerman (1990), Model calculations of tropospheric ozone production potential following observed convective events, J. Geophys. Res., 95, 14,049 – 14,062.
Pickering, K. E., A. M. Thompson, J. R. Scala, W.-K. Tao, J. Simpson, and
M. Garstang (1991), Photochemical ozone production in tropical squall
line convection during NASA/GTE/ABLE 2A, J. Geophys. Res., 96,
3099 – 3114.
Pickering, K. E., A. M. Thompson, J. Scala, W.-K. Tao, R. R. Dickerson,
and J. Simpson (1992a), Free tropospheric ozone production following
entrainment of urban plumes into deep convection, J. Geophys Res., 97,
17,985 – 18,000.
Pickering, K. E., A. M. Thompson, J. R. Scala, W.-K. Tao, and J. Simpson
(1992b), Ozone production potential following convective redistribution
of biomass emissions, J. Atmos. Chem., 14, 297 – 313.
Pickering, K. E., A. M. Thompson, W.-K. Tao, and T. L. Kucsera (1993),
Upper tropospheric ozone production following mesoscale convection
during STEP/EMEX, J. Geophys. Res., 98, 8737 – 8749.
Pickering, K. E., et al. (1996), Convective transport of biomass burning
emissions over Brazil during TRACE-A, J. Geophys. Res., 101, 23,993 –
24,012.
Pickering, K. E., Y. Wang, W.-K. Tao, C. Price, and J.-F. Müller (1998),
Vertical distributions of lightning NOx for use in regional and global
chemical transport models, J. Geophys. Res., 103, 31,203 – 31,216.
D14304
Robert, A., and E. Yakimiw (1986), Identification and elimination of an
inflow boundary computational solution in limited area model integrations, Atmos. Ocean, 24, 369 – 385.
Rogers, E., T. L. Black, D. G. Deaven, and G. J. DiMego (1996), Changes
to the operational ‘‘Early’’ Eta analysis/forecast system at the national
Center for Environmental Prediction, Weather Forecast., 11, 391 – 413.
Scala, J. R., et al. (1990), Cloud draft structure and trace gas transport,
J. Geophys. Res., 95, 17,015 – 17,030.
Skamarock, W. C., J. G. Powers, M. Barth, J. E. Dye, T. Matejka,
D. Bartels, K. Baumann, J. Stith, D. D. Parrish, and G. Hubler (2000),
Numerical simulations of the July 10 Stratospheric-Tropospheric Experiment: Radiation, Aerosols, and Ozone/Deep Convection Experiment
convective system: Kinematics and transport, J. Geophys. Res., 105,
19,973 – 19,990.
Skamarock, W. C., J. E. Dye, E. Defer, M. C. Barth, J. L. Stith, B. A.
Ridley, and K. Baumann (2003), Observational- and modeling-based
budget of lightning-produced NOx in a continental thunderstorm, J. Geophys. Res., 108(D10), 4305, doi:10.1029/2002JD002163.
Smolarkiewicz, P. K. (1983), A simple positive definite advection scheme
with small implicit diffusion, Mon. Weather Rev., 111, 479 – 486.
Smolarkiewicz, P. K., and W. W. Grabowscki (1990), The multidimensional
positive advection transport algorithm: Nonoscillatory option, J. Comp.
Phys., 86, 355 – 375.
Soong, S.-T., and Y. Ogura (1980), Response of trade wind cumuli to largescale processes, J. Atmos. Sci., 37, 2035 – 2050.
Stenchikov, G. L., R. R. Dickerson, K. E. Pickering, W. G. Ellis Jr.,
B. Doddridge, S. Kondragunta, and O. Poulida (1996), Stratospheretroposphere exchange in a midlatitude mesoscale convective complex:
2. Numerical simulations, J. Geophys. Res., 101, 6837 – 6852.
Tao, W.-K. (2003), Goddard Cumulus Ensemble (GCE) model: Application
for understanding precipitation processes, Monographs, Cloud Systems,
Hurricanes and TRMM, AMS Monogr. Ser., pp. 103 – 138, Am. Meteorol.
Soc., Boston, Mass.
Tao, W.-K., and J. Simpson (1989a), Modeling study of a tropical squalltype convective line, J. Atmos. Sci., 46, 177 – 202.
Tao, W.-K., and J. Simpson (1989b), A further study of cumulus interaction
and mergers: Three-dimensional simulations with trajectory analyses,
J. Atmos. Sci., 46, 2974 – 3004.
Tao, W.-K., and J. Simpson (1993), Goddard Cumulus Ensemble model:
Part I. Model description, Terr. Atmos. Oceanic Sci., 4, 35 – 72.
Tao, W.-K., J. Simpson, and M. McCumber (1989), An ice-water saturation
adjustment, Mon. Weather Rev., 117, 231 – 235.
Tao, W.-K., J. Simpson, and S.-T. Soong (1991), Numerical simulation of a
subtropical squall line over Taiwan Strait, Mon. Weather Rev., 119,
2699 – 2723.
Tao, W.-K., J. Simpson, C.-H. Sui, B. Ferrier, S. Lang, J. Scala, M.-D.
Chou, and K. Pickering (1993), Heating, moisture and water budgets
of tropical and midlatitude squall lines: Comparisons and sensitivity to
longwave radiation, J. Atmos. Sci., 50, 673 – 690.
Tao, W.-K., S. Lang, J. Simpson, C.-H. Sui, B. Ferrier, and M.-D. Chou
(1996), Cloud-radiative mechanisms associated with a tropical and a
midlatitude squall line, J. Atmos. Sci., 53, 2624 – 2651.
Tao, W.-K., et al. (2003), Miccrophysics, radiation and surface processes in
the Goddard Cumulus Ensemble (GCE) model, Meteorol. Atmos. Phys.,
82, 97 – 137.
Wang, Y., W.-K. Tao, J. Simpson, and S. Lang (2003), The sensitivity of
tropical squall lines (GATE and TOGA COARE) to surface fluxes: 3-D
Cloud resolving model simulations, Q. J. R. Meteorol. Soc., 129, 987 –
1007.
Yakimiw, E., and A. Robert (1990), Validation experiments for a nested
grid-point regional forecast model, Atmos. Ocean, 28, 466 – 472.
D. Bartels, NOAA Forecast Systems Laboratory, 325 Broadway, Boulder,
CO 80305, USA. (diana.bartels@noaa.gov)
A. DeCaria and J. Scala, Department of Earth Sciences, Millersville
University, P.O. Box 1002, Millersville, PA 17551, USA. (alex.decaria@
millersville.edu; john.scala@millersville.edu)
T. Matejka, National Severe Storms Laboratory, NOAA, NCAR, P.O.
Box 3000, Boulder, CO 80307, USA. (matejka@ucar.edu)
L. Ott and K. Pickering, Department of Meteorology, University of
Maryland, College Park, College Park, MD 20742, USA. (leo@atmos.umd.
edu; pickerin@atmos.umd.edu)
G. Stenchikov, Department of Environmental Sciences, Rutgers University, 14 College Farm Road, New Brunswick, NJ 08901-8551, USA.
(gera@envsci.rutgers.edu)
W.-K. Tao, NASA Goddard Space Flight Center, Code 912, Greenbelt,
MD 20771, USA. (tao@agnes.gsfc.nasa.gov)
14 of 14