Photochemical Simulations of Point Source Emissions
With the Models-3 CMAQ Plume-in-Grid Approach
James M. Godowitch* and Jeffrey O. Young*
Atmospheric Sciences Modeling Division
Air Resources Laboratory
National Oceanic and Atmospheric Administration
Research Triangle Park, NC 27711
* On assignment to the National Exposure Research Laboratory,
U.S. Environmental Protection Agency
ABSTRACT
A plume-in-grid (PinG) approach has been designed to provide a realistic treatment for the
simulation of the dynamic and chemical processes impacting pollutant species in major point
source plumes during a subgrid scale phase within an Eulerian grid modeling framework. The
PinG science algorithms include a Plume Dynamics Model (PDM) processor and a Lagrangian
plume module. The PDM processor provides plume dimensions and related parameters needed
by the Lagrangian PinG module, which is an integral component of the Community Multi-scale
Air Quality (CMAQ) chemical transport model. PinG uses grid concentrations as boundary
conditions and an important feedback occurs with plume pollutants impacting the appropriate
grid cell when plume width reaches the grid cell size.
Simulations were performed with the PinG treatment applied to a group of point sources
exhibiting a wide range of NOX emission rates situated in a regional modeling domain
encompassing Nashville, Tennessee. Selected plume model results are presented from a case
study day from the Nashville/Middle Tennessee ozone study period during July 1995. Modeled
plume ozone and nitrogen species concentrations evolved in the same manner found from
observed plume data, which provides encouraging initial evidence of the capability of the PinG
technique. The ozone recovery period in the modeled plume core and NOX oxidation rates were
strongly dependent on the NOX emission rate. Additionally, the excess ozone above background
in the modeled plume was generally greater for the higher NOX emission point sources.
INTRODUCTION
There has been recognition of the need for an improved modeling methodology to deal with
subgrid scale emission features, such as pollutant plumes emanating from major point sources, in
regional Eulerian photochemical grid models. Substantial anthropogenic primary emissions of
nitrogen oxides (NOX), which are important precursors of ozone (O3) and other secondary
pollutant species, are released from the tall stacks of relatively isolated fossil-fuel power plants.
The traditional Eulerian grid modeling approach has been to instantly mix these point source
-------
emissions into the entire volume of a model grid cell. In contrast, notable aspects of point source
plumes are their small initial horizontal dimension relative to regional grid cells and their finite
expansion rate. Depending on meteorological conditions, several hours and a considerable travel
distance may be needed for plume width to reach the typical grid cell size of a regional
photochemical model. This overdilution by instantly dumping of high NOX emissions into
relatively large grid cell volumes promotes premature initiation of rapid photochemistry of
secondary species, such as O3, which strongly distorts the spatial and temporal features of the
photochemical process. Consequently, this inadequate treatment of major point source emissions
in Eulerian photochemical grid modeling has been a contributing factor to model uncertainty.
Although finer grid cell sizes are commonly used for urban-scale domains which somewhat
alleviates overdilution and improves the spatial resolution of major point source emissions, a fine
grid cell size of 4 km or less remains computationally impractical for the large regional-scale
domains. Consequently, there is a need for a subgrid scale treatment for large point source
emitters in regional grid cell domains since results of these applications are also commonly used
to provide initial and boundary conditions for subsequent finer nested grid model domains.
A limited number of subgrid plume models have been developed and implemented in Eulerian
photochemical grid models. The plume-in-grid approach described by Seigneur et al.1 was
included in a modified version of the Urban Airshed Model (UAM) and applied to an urban
domain with a 4 km grid cell size. It provided crosswind plume resolution with a contiguous
array of vertically well-mixed plume cells of different widths to maintain equal pollutant mass
among the cells. Each rectangular plume section travels in a Lagrangian manner with the mean
wind flow. An updated plume-in-grid approach composed of elliptical rings was incorporated
into the UAM-V model (Morris et al.2), and the same plume technique was embedded in the
SAQM model as documented by Myers et al.3. With this technique, when each ring attains the
grid cell size its plume material is returned to a grid cell. Kumar and Russell4 designed a plume
model for the URM grid model that treats plume puffs for up to one hour before transferring the
plume pollutants into the grid framework. They also performed simulations with a 20 km grid
cell size on a regional domain covering the northeast US. However, the common result from
these different PinG treatments was that the notable impact on ozone concentrations relative to
values generated from the traditional Eulerian modeling approach was limited to the grid cells
containing or nearby the point source emission locations.
A research and development effort has culminated in the implementation of plume-in-grid
modeling algorithms within the Community Multi-scale Air Quality (CMAQ) modeling system5.
The CMAQ system is a collection of state-of-science meteorological, emissions, and air quality
grid modeling algorithms that reside within the Models-3 system framework which also contains
a variety of tools for the execution and analyses of the modeling components6. The PinG
approach was designed to address the need to more realistically resolve the spatial scale of
pollutant plumes emanating from a major elevated point source emitter (MEPSE) within an
Eulerian regional modeling domain. Model simulations with the PinG algorithm embedded in
the CMAQ Chemical Transport Model (CCTM) were performed on a model domain with a 36
km grid cell size encompassing the Nashville, Tennessee region. The CCTM/PinG simulations
were performed for experimental days from the Southern Oxidant Study (SOS) summer 1995
field study period in the Nashville area. A select group of five isolated point sources were
2
-------
simulated to assess the capability of the PinG technique to model the evolution of gas-phase
primary and secondary oxidant species. These MEPSEs exhibited a wide range of NOX emission
rates. The focus in this paper is on subgrid plume model results from a particular case study day
(July 7,1995) to demonstrate the evolution of O5 and nitrogen species in the point source plumes.
Qualitative comparisons of the PinG modeled results to features found from analyses of observed
plume data are also provided where applicable.
MODEL DESCRIPTION
The discussion herein is limited to an overview of the PinG technique since details of the
technical approach and model formulation can be found in Gillani and Godowitch7. The plume-
in-grid approach of the Models-3 CMAQ employs a Lagrangian framework to simulate a
continuous plume by a series of moving plume sections, which are released at hourly intervals
from a point source. The PinG approach has been designed to simulate one or multiple major
point source plumes at any time during the entire diurnal cycle within a regional Eulerian grid
modeling domain. The two key modeling algorithms include a Plume Dynamics Model (PDM)
processor and the Lagrangian PinG plume module. The PinG plume algorithm has been
integrated into the CCTM and is exercised simultaneously during the grid model simulation.
The conceptual design of the Models-3 PinG is to treat each plume section during a subgrid scale
phase until plume width attains the size of a model grid cell, which is followed by the feedback
or "handover" of the plume pollutants at the proper grid cell location. Figure 1 illustrates the
temporal behavior of plumes released during daytime and nocturnal periods and it shows the
relevant processes simulated in the PinG module. Plume rise is computed in the PDM processor
and the plume bottom and top heights are specified. Although a plume section is initially
elevated, vertical dispersion leads to plume growth, which causes it to extend from the surface to
the daytime planetary boundary layer (PEL) height. During nocturnal hours, plume sections
released above the PEL remain elevated until being fumigated by a growing PEL during the next
morning. Horizontal resolution of each plume section is provided by a crosswind array of
attached plume cells, as depicted in the bottom portion of Figure 1. All plume cells within a
plume section have equal widths and have the same vertical depth. In the current version of
PinG, each plume cell represents a single vertical layer. Turbulence and vertical wind shear
processes contribute to horizontal plume expansion from time t to t + dt as shown in Figure 1,
which is an important factor that strongly influences the chemical evolution in the plume.
Parameterization techniques described by Gillani and Godowitch7, which rely on 2D and 3D
meteorological fields generated by the Penn State/NCAR MM5 mesoscale model, are used in the
PDM simulation to determine the width of each subgrid scale plume section along its path. As a
plume section expands gradually in the horizontal and vertical dimensions, PinG simulates
dilution and the entrainment processes at the plume edges and top with CCTM grid
concentrations providing the 'background' conditions at each plume edge. The horizontal
diffusion process between adjacent plume cells is simulated with an eddy mixing coefficient
method. Additionally, PinG includes pollutant dry removal using the deposition velocity method
and surface emissions in the grid cell where the plume section is located are incorporated into
each plume cell. PinG applies the same gas-phase photochemical mechanisms and chemistry
solvers contained in the CCTM. Currently, either the Carbon Bond (CB-4) or RADM2
3
-------
Figure 1. Illustration of the plume formulation depicting relevant processes in a time-height
view (top) and cross-wind view of a plume section (bottom) at time t and t+dt.
illiiiiiiililii Gas-Phase 8!H!sHB!ijii!!iH!
PI 11 mo jsiJissgiiii _.. . . lllUllU'
Plume Jjigggigggigi Chemistry ill
'
--i-fx---
Surface Emissions Deposition
JW
>.
^
UUUlfl
ni
N
i
Y
<£
to
s~^
"^
'N-
D
s\*
ffus
t
,»
on
i
?
-*«
-*"
MfMf
Plume
cross-section
at time t
^*- at (t + dt)
^^^_
D
1
L
U
T
1
M
t nil cuniMGru
from sides
HH*
mechanism can be selected to perform gas-phase chemistry. Chemistry solvers include the
quasi-steady state approximation (QSSA) and the sparse matrix vectorized Gear (SMVGEAR)
methods8.
The flow diagram in Figure 2 illustrates where the principal PinG modeling components fit into
the CMAQ system of science codes. The PDM processor is exercised with input data files which
include a stack data file generated by the Models-3 Emission Processing and Projection System
(MEPPS) and 2-D and 3-D meteorological data files generated by the Meteorology-Chemistry
Interface Processor (MCIP) program. The PDM processor is capable of simultaneously modeling
multiple MEPSE plumes for the entire diurnal cycle and it stores the plume dimensions, position,
and related parameters in a data file for use in the PinG simulation. With the integrated, coupled
modeling approach, the embedded PinG module is executed within a CCTM simulation. When
invoked from the CCTM driver routine during a CCTM time step, PinG accesses MCIP
meteorological files, surface emissions contained in the ECIP (Emission Chemistry Interface
Processor) 3-D data file, and the MEPSE emissions file generated from MEPPS processing.
During the PinG simulation, each plume section is subjected to the various processes at the
appropriate time step. The PinG module integration time interval is synchronized with the grid
model time interval. Plume pollutant species impact the gridded concentration array when the
plume size or another criterion is triggered. During a simulation, PinG stores the species
4
-------
Figure 2. Flow diagram of the PDM processor and PinG module with associated programs of
the CMAQ modeling system.
Embstons
mmmn^^^m
MEPPS
Ch&mistry- Transport
concentrations in all plume cells of each active plume section in a separate plume concentration
file.
MODEL SIMULATION RESULTS
A model domain consisting of 21 x 21 horizontal grid cells with a 36km grid cell resolution and
21 vertical layers was used for CCTM/PinG simulations. The modeling domain encompassed a
region surrounding Nashville, Tennessee and represented a subdomain of a much larger 36 km
gridded CCTM model domain covering the eastern half of the United States, Since CCTM
model simulations for the larger regional domain had been already performed, realistic initial and
boundary conditions for the smaller domain were available from the large domain outputs.
Domain-specific meteorological data sets were produced from MCIP processing of MM5 output
files and hourly 3-D emissions were generated by the ECIP processor using the area and point
emission files already produced by the MEPPS emissions system.
A set of five MEPSEs exhibiting a wide variation of NOX emissions rates were designated for the
PinG treatment and their emissions were excluded from the 3-D ECIP emission file, although
other point source emissions in the domain were simulated in the traditional way in the CCTM.
The five point sources treated by PinG included the power plant emissions from Shawnee (SH),
Paradise (PA), Cumberland (CU), Johnsonville (JV), and Gallatin (GA). The order from lowest
to highest in the total daily NOX emissions for these point sources for July 7, 1995 is GA, SH,
JV, CU, and PA. With GA as a reference (32 tons/day of NOX), the NOX emissions at SH, JV,
CU, and PA were a factor of 3.0, 3.7, 15.0, and 15.2 times greater, respectively.
-------
Simulations with the CCTM/PinG model for this mini-regional grid domain were performed on a
Sun Ultra 2 workstation. A model ran for the 24-hour period required less than 2 CPU hours
when applying the CB-4 chemical mechanism and QSSA solver. Simulations with the RADM2
mechanism and/or SMVGEAR solver require greater processing time. There were 10 plume
cells defining the crosswind extent of each plume section. Differences in the CPU time were
minor between simulations with PinG or without PinG for this set of point sources.
Selected plume model results are presented from July 7, 1995. Briefly, clear weather conditions
prevailed on July 7 with a steady, northwesterly wind pattern during the daytime period due to
the recent passage of a cold front. Figure 3 shows the trajectory and growth of the plume width
generated by the PDM processor for the plume section released at 1500 UTC (10:00 local
daylight time) from each point source situated in the model domain. Due to the prevailing flow
pattern, the plumes from these point sources were nearly parallel and also isolated from each
other on this day. The paths traveled by the plumes are determined from mean wind components
which are derived by averaging winds over the layers spanned by the plume depth. It is
encouraging that they closely coincide with plume positions displayed from aircraft interceptions
at various downwind distances during horizontal traverses through the plumes over the afternoon
period9. The subgrid scale PinG simulation period of a plume section would normally cease
when the plume width achieved the 36 km grid cell size, which appears to be 3 or more grid cells
downwind of their source locations in Figure 3. This distance was at least 100 km downwind in
this case. However, the PDM and PinG were exercised to allow plume sections to be simulated
longer so that the full extent of the chemical cycle would be realized.
The ozone concentrations displayed in Figure 4 are at various times for the plume section
released at 1500 UTC from CU, one of the two largest point sources. The evolution of ozone
through 3 chemical stages documented in observed plumes9 during the daytime is also
characterized by the same distinctive behavior of ozone in this modeled plume section. The first
stage exhibits a significant deficit of ozone due to titration by the high NO concentrations, which
is characteristic of a relatively fresh plume dominated by high NOX emissions closer to the
source. As the plume continues to expand horizontally downwind, a second chemical stage
reflects the rapid ozone formation occurring near the plume edges due to entrainment of
background (grid) values containing greater volatile organic compound (VOC) concentrations.
In phase 2, ozone concentrations at in cells closer to the plume edges exceed background values,
while ozone in the plume core is still recovering. Stage 3 represents the mature chemical stage as
the plume is substantially diluted with NOX concentrations dropping to low levels and ozone
across the entire plume exceeding background grid levels.
Although ozone concentrations in NOx-rich plumes have been found to chemically evolve
through this same photochemical cycle during a typical summer daytime period, the overall time
needed for plume ozone to go through this chemical cycle can be expected to be strongly
dependent on the NOX emission rate. Figure 5 illustrates the modeled plume section from
releases at the same time from three point sources with significantly different NOX emission
rates. It is interesting that after 2 hours travel time, the ozone in each of these plume sections is
in a different chemical stage. Since the widths of these plume sections are nearly the same, the
notable differences in the chemical evolution cannot be attributed to differences in plume growth.
6
-------
Figure 3. Trajectory and horizontal growth of the plume sections released in the 36 km
gridded modeling domain at 1500 UTC from the major point sources on 7 July, 1995.
-120000
-200000-
-260000-
-360CQO-
-440000-
-520000-
-600000-
-680000-
-760000-
-640000:
-920000
I ' ' ' ' ' ' ' I ' ' ' ' ' ' ' 1
-130000 -8DOOO 30000 110000 190000 270000 350000 430000 510000 S9QOOO 670000
In fact, the plume section from the smallest NOX emission source (GA) has already reached stage
3 while the ozone concentrations in the plume section from the highest NOX source (CU) remain
in stage 1. The plume section from JV with NOX emissions a factor of 3 greater than GA is
clearly in phase 2 with ozone in the plume core approaching background values and relatively
higher ozone concentrations occur near the plume sides.
The results in Figure 6 show the temporal behavior of ozone concentrations in the plume core
relative to background grid values. It provides insight into ozone evolution in the cell on each
7
-------
Figure 4. Ozone concentrations relative to background in the plume section released from CU
at 1500 UTC at different times (UTC) / downwind distances (km); o - 1600/25,
* - 1700/48 , 0 - 1830/89 , A - 2000/133, - 2100/166. The symbol at the extreme
edge represents the CTM grid concentration.
PinG MODEL RESULTS (July 7, 1995)
20
.u
co _1OJZ
CO
o
LLJ
D
20
-30
40
I I I I | I I I I | I I I I | I II I | I II I | I I I I | I II I | I I I I | I II I | I I I I | I I ITJT
-30 -25 -20 -15 -10 -50 5 10 15 20 25
DISTANCE ACROSS PLUME (km)
30
side of the plume centerline for the release at 1500 UTC, which remained active throughout the
afternoon period. In particular, it reveals that the recovery time of ozone in the plume core is
generally inversely related to the NOX emission rate. Clearly, ozone recovery in the GA plume
was fastest, while the two largest NOX sources (CU and PA) required the longest times for ozone
in the plume center to recover to background levels. It also followed that the peak excess ozone
in the plume core occurred at greater travel times (i.e., greater downwind distances) as the NOX
emission rate increased. Plume sections from CU and PA needed over 6 hours to reach full
chemical maturity in this case. Another interesting feature in Figure 6 is the difference in the
amount of the plume ozone excess above background values among this group. Considering
GA, JV and CU as a group with lowest to highest NOX emission rate, Figure 6 indicates that the
peak excess plume ozone was about 6, 10, and 18 ppb above the gridded background values,
respectively. Assuming other factors being the same, higher NOX emissions appear to lead to
greater plume excess ozone concentrations.
-------
Figure 5. Ozone concentrations relative to background at 2 hours after release reveal
different chemical stages in the plume sections released at 1500 UTC from CU (0),
JV (A), and GA ().
PinG MODEL RESULTS
NASHVILLE DOMAIN (DAYi 1995188 TIME: 1700 UTM)
I
B
CO
O
CO
O
101
5
0
-5
-10
-15
-20
-10 -8-6-4-20 2 4 6 8 10
DISTANCE ACROSS PLUME (km)
Although NOX emissions from PA and CU were nearly the same on this study day, and SH
emissions were somewhat less than those at JV, noticeable differences in plume ozone evolution
were apparent between these pairs of comparable emission sources. On further examination of
other possible factors, it was discovered that JV and CU are both located in a relatively high
biogenic emission area compared to the other point sources '" of comparable emission strength.
Hence, surface biogenic emissions are tentatively attributed to have played an active role in
promoting greater photochemical ozone production in the JV and CU plumes which produced
higher ozone values relative to the SH and PA plumes for this case.
Since these are significant primary NOX sources, it is of interest to examine the temporal
behavior of modeled plume NOX concentrations. The oxidation of simulated plume NOX is
depicted in Figure 7 by its ratio with NOY, where NOY is the total concentration of the sum of all
nitrogen species. It is evident that the rate of NOX oxidation differs greatly among the modeled
plumes from these emission sources. The most rapid NOX oxidation occurred in the GA plume,
while the NOX loss rate was much slower for the largest NOX sources. It is encouraging that
9
-------
Figure 6. Time variation of ozone concentrations in the plume core relative to background for
the plume sections released at 1500 UTC from SH (o), JV (A), PA (*), CU (0),
and GA(«).
a
a
CO
O
i
8
20
-10^
-20 T:
-30-E
-40^
-50-^
C
PinG MODEL RESULTS
-!$ -o
jr^74^5^SSr:
f // /^
l // ./
/ // //
I 1 />'
i &/ vy
' /f /#'
' '/ ' /
/ '! //
i ! / >
; /' ' /
, II fy
I//
^/
i i i i i i i i i i i i i i i i
112345678
TRAVEL TIME (hours)
the modeled results are in qualitative agreement with recent results from plume measurements ".
The temporal behavior of NOZ (= NOY - NOX ) is also of interest in order to assess nitrogen
species production during the course of plume travel. The NOZ/NOY ratio in Figure 8 depicts a
contrasting view from that just displayed by NOX/NOY in Figure 7. A line drawn at NOZ/NOY =
0.55 is also shown to signify a transition from a VOC-limited chemistry to a NOx-limited
chemistry " in the plume. It is apparent that the simulated NOZ/NOY ratio reaches 0,55 at about
the same time as the ozone deficit in the plume core disappeared in Figure 6. Therefore, as the
NOZ / NOY ratio increases above this level, plume ozone has completely recovered and the
mature chemical stage is underway. It is also encouraging that the results for the modeled CU
plume follow the same temporal behavior derived from observed CU plume results in Gillani et
al. l2 for 7 July 1995.
The ratio of ozone to NOZ has been employed as a measure of ozone production efficiency
10
-------
(OPE). The focus here again is on the plume core. Background O, and NOZ values were
removed from plume concentrations before computing the ratios displayed in Figure 9. There is
a notable variation among these different NOX point sources. The highest O3/NOZ ratios existed
for the lowest NOX source (GA), and the lowest values occurred in the simulated plumes from
Figure 7. Temporal variation of the NOx/NOy ratio in the plume core from releases at
1500 UTC on 7 July 1995. Symbols are the same as in Figure 6.
i
z
-^
§
z
PInG MODEL RESULTS
1.0
0.9
0.8
0.7
0.6
*""""
0.5
0.4
0.3
0.2
0.1
Ofl
*U
"V^
V\^\-^
\ ^S ^Vlt^X
\ % \\
\ \\ X\
\ \\ \\
\ \ A \ t
\ 4 x\
\ \ \ \ \
\ \\ %
\ ^ ^x
% XV
v \ \
\ \ \
\ \ \
s ^:\ c--.
v ss \\
^ ^^ ^3k
* x\\ $*x
s\ xfe-- "^-^
^^-^^^=i=^^
1 1 1 1 1 1 1 1 1 1 1 1 I 1 1
01 2345678
TRAVEL TIME (hours)
CU and PA, the two highest NOX emission sources, NOZ production was more rapid than ozone
formation during the initial stage with large negative values due to the O3 deficit in the plume
core. Although the GA plume exhibited the highest OPE based on Figure 9, it also generated the
lowest peak O3 value and also the lowest NOZ concentrations among this set of sources. In
contrast, the modeled CU plume generated the highest peak excess O3 concentration and the
highest NOZ concentration compared to the other sources.
Another method for assessing OPE is to apply the concentration ratio approach used by Ryerson
et al.9. With this method, the slope of O, versus SO2 derived from the modeled plume cell
concentrations is divided by the NOX / SO2 emission ratio to obtain an estimate of OPE. A linear
regression was applied to obtain the best-fit slope with the O, and SO2 concentrations from all
11
-------
plume cells in the mature plume section at 2100 UTC. Hourly NOX and SO2 emission rates at
1500 UTC were used in the determination since it was the time of release of this plume section.
The results for GA, JV, SH, CU, and PA were found to be 15.4, 8.1, 7.5, 2.6, and 1.5,
respectively, from the plume section at 2100 UTC. The values computed from aircraft traverses
through mature plumes on the afternoon of July 7 for JV, CU, and PA are 7, 1.7, and 2.1,
respectively9. The modeled results are certainly comparable to these values from observed data.
Figure 8. Variation of the NOz/NOy ratio in the plume core from releases at 1500 UTC
on 7 July 1995. Symbols are the same as in Figure 6.
1.0
.9
0.8
0.7
£ 0.6^
^ 0.5
Z 0.4
0.3
0.2
0.1
Of»
.u
c
PinG MODEL RESULTS
/ ,/ ./'
/ f /
/ / /
f / //
1 /* s xx
f'i.C-'-"^^'
I I I I | I | I I [
312345678
TRAVEL TIME (hours)
SUMMARY AND CONCLUSIONS
Plume-in-grid simulations for a group of large, elevated NOX point sources were performed in a
36 km gridded mini-domain encompassing the greater Nashville region. The results presented
herein with the integrated PinG algorithm being exercised within the CMAQ chemical transport
model were generated with the CB-4 chemical mechanism. The PinG simulated the evolutionary
behavior of plume ozone in a realistic manner. The travel time required for O3 in the plume core
to recover to background values increased with higher NOX emission rates. Although ozone
12
-------
production efficiency was found to decrease with increasing NOX emission rate, the peak excess
plume ozone value above background was generally greater as the NOX emission rate increased.
In addition, qualitative comparisons of modeled and observed results from field study
measurements revealed reasonable agreement which provides evidence that the PinG module is
capturing the relevant plume processes impacting the behavior of pollutant concentrations.
However, quantitative results of PinG performance are anticipated from an upcoming diagnostic
evaluation using plume data from the Nashville SOS summer 1995 data base.
Figure 9, Variation of the ozone to NOz ratio in the plume core from plume section releases at
1500 UTC on 7 July 1995. Symbols are the same as in Figure 6.
n
a
1
i
8
i
w
O
I
PinG MODEL RESULTS
10-, ,_^
8
6
4
2
C* _:L ,
-8
(
/ """^
/
,&*>
/ A'"' ^ ^'^ _-*''''
fX / /0"" /<^'^*-''^t""
/ / / v7 <
-------
3. Myers, T., et al, Systems Applications Intl., 1996, Report SYSAPP-96/06, San Rafael,
CA.
4. Kumar N. and A.G. Russell, J. Geophys. Res., 1996, 101, 16737-16744.
5. Byun, et al., 10lh AMS/AWMA Joint Conference on Applications of Air Pollution
Meteorology, 1998, Jan 11-16, Phoenix, AZ
6, Novak, J., et al., 10* AMS/AWMA Joint Conference on Applications of Air Pollution
Meteorology, 1998, Jan. 11-16, Phoenix, AZ.
7. Gillani, N.V. and J.M. Godowitch, Chapter 9, EPA Technical Report, 1999, EPA/600/R-
99/030, National Exposure Research Laboratory, Research Triangle Park, NC
8. Gipson, G., and J. Young, Chapter 8, EPA Technical Report, 1999, EPA/600/R-99/030,
National Exposure Research Laboratory, Research Triangle Park, NC.
9. Ryerson, T.B. et al., J. of Geophys. Res., 1998, 103, D17, 22569-22583.
10. Williams, J., et al., Geophys. Research Letters, 1997, Vol. 24, 9, 1099-1102.
11. Ryerson, T. et al., Poster paper, American Geophysical Meeting, Dec. 1999.
12. Gillani, N.V. etal., J. of Geophys. Res., 1998, 103, D17, 22593-22615.
DISCLAIMER
This paper has been reviewed in accordance with the U.S. Environmental
Protection Agency's peer review and administrative review policies and approved
for publication. Mention of trade names or commercial products does not
constitute endorsement or recommendation for use.
KEY WORDS
Lagrangian reactive plume modeling, plume-in-grid modeling, regional ozone
photochemistry, point source modeling, ozone modeling
14
-------
NEKL-RTt-AMD-00-063
TECHNICAL REPORT DATA
1. REPORT NO.
EPA/600/A-00/016
2.
3.RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE
Photochemical simulations of point source emissions
with the Models-3 CMAQ plume-in-grid approach
5.REPORT DATE
6.PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
James M. Godowitch and Jeffrey 0. Young
8.PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Same as Block 12
10.PROGRAM ELEMENT NO,
11. CONTRACT/GRANT NO.
12. SPONSORING AGENCY NAME AND ADDRESS
U.S. Environmental Protection Agency
Office of Research and Development
National Exposure Research Laboratory
Research Triangle Park, NC 27711
13.TYPE OF REPORT AND PERIOD COVERED
Proceedings, FY-00
14. SPONSORING AGENCY CODE
EPA/600/9
15. SUPPLEMENTARY NOTES
16. ABSTRACT
A plume-in-grid (PinO) approach has been designed to provide a realistic treatment for the simulation (he dynamic and chemical processes impacting
pollutant species in major point source plumes during a subgrid scale phase within an Eulerian grid modeling framework. The PinO science algorithms
include a Plume Dynamics Model (PDM) processor and a Lagrangian plume module. The PDM processor generates plume dimensions and related
parameters needed by the Lagrangian PinO module, which is an integral component of the Community Multiscale Air Quality (CMAQ) chemical transport
model. PinO uses grid concentrations as boundary conditions and it provides an important feedback of plume pollutants to the grid in the appropriate grid
cell when plume width reaches the grid cell size. Simulations were performed with the PinG treatment applied to point sources exhibiting a wide range of
NOx emission rates situated in a regional modeling domain encompassing Nashville, Tennessee. Selected plume model results are presented from a case
study day from the Nashville/Middle Tennessee field study in June 1995. Modeled plume ozone and nitrogen species concentrations evolved in the same
manner found from observed plume data, which provides encouraging initial evidence of the capability of the PinG technique. The ozone recovery period
the modeled plume core and NOx oxidation were strongly dependent on the NOx emission rate, The excess ozone above background in the modeled plume
was greater for point sources with higher NOx emissions.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/ OPEN ENDED
TERMS
c.COSATI
18. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19. SECURITY CLASS (This
Report)
UNCLASSIFIED
21,NO. OF PAGES
20. SECURITY CLASS (This
Page)
UNCLASSIFIED
22. PRICE
------- |