SIMULATIONS OF AEROSOLS AND PHOTOCHEMICAL SPECIES
WITH THE CMAQ PLUME-IN-GRID MODELING SYSTEM
J. M. Godowitch
Atmospheric Sciences Modeling Division, Air Resources Laboratory
National Oceanic and Atmospheric Administration
Research Triangle Park, North Carolina
email: godowitch.james@epa.gov	EPA/600/A-04/095
1.	INTRODUCTION
A plume-in-grid (PinG) treatment has been an
integral component of the Community Multiscale Air
Quality (CMAQ) modeling system. The PinG
approach was designed to provide a realistic
treatment of the dynamic and chemical processes
governing pollutant concentrations in the plumes
emitted from the selected major point source stacks
on the subgrid scale. For large regional or
continental modeling domains, in particular, grid cell
sizes are generally specified to be tens of
kilometers on a side. However, due to their small
spatial extent in the horizontal dimension, point
source plumes are a subgrid scale feature on a
typical Eulerian grid modeling framework.
Consequently, artificial dilution is considerable
when point source emissions are instantly mixed
into a large grid cell volumes, which impacts
primary pollutant concentrations and the chemical
processes governing secondary species. In
contrast, the PinG technique spatially resolves the
large concentration gradients within plumes by
simulating the gradual growth downwind using a
Lagrangian framework. Since the plumes are
modeled at the proper spatial scale, chemical
processes in the plume evolve in a more realistic
manner.
This paper contains an overview of the key
components of the current CMAQ plume-in-grid
modeling system. A description of the procedures
to perform model simulations with the PinG
treatment will also be given. In previous public
releases of the CMAQ modeling system, the PinG
module was capable of treating photochemical
processes. A notable update in the current release
version of the CMAQ/PinG is the inclusion of a
treatment for aerosol formation processes. Model
results of selected photochemical and aerosol
species from a test application are also presented
to illustrate the impact on the gridded concentration
field.
2.	MODEL OVERVIEW
The key modeling components of the CMAQ
PinG approach are the plume dynamics model
(PDM) processor program and the PinG module,
which is a Lagrangian reactive plume model. A
technical description of the relevant plume
processes treated in these PinG algorithms and
Author address: US EPA, NERL, E243-03, RTP,
NC 27711. On assignment to the EPA National
Exposure Research Laboratory.
their mathematical formulations have been
documented in Gillani and Godowitch (1999). The
PDM processor generates a data file containing
plume position and plume dimension information for
the PinG module, which is fully integrated inside the
CMAQ Chemical Transport Model (CTM). To be
consistent, the PinG module applies the same
chemical mechanisms (i.e. CB-IV, SAPRC-99 ) as
the CTM. The current version of the CMAQ/PinG
applies the GEAR chemical solver (Gipson and
Young, 1999), which provides an accurate solution
for the high NOx concentration regime found in
many point source plumes. The current CMAQ
aerosol module (Binkowski, 1999) applied in the
CTM has been incorporated into the PinG module
to provide a treatment of aerosol processes in the
subgrid plumes. The Binkowski aerosol algorithm
represents the size distribution by the superposition
of 3 lognormal subdistributions (i.e. modes).
Relevant processes are considered that impact
PM2.5, PM-io, and secondary aerosol species
including sulfate (S04"2), nitrate, ammonium, water,
and organics from precursors of anthropogenic and
biogenic sources.
	i
Pltiffw !>i Vitrnfcs
Meimmfagy-
Figure 1. Flow diagram of the key plume-in-grid
modeling components (PDM processor and PinG
module ) in the CMAQ modeling system.
The flow diagram in Figure 1 illustrates the two
plume-in-grid modeling components. The PDM
processor program shown in Figure 1 requires a
stack parameter file created by a SMOKE
emissions processor and the meteorological data
files generated by the Meteorology Chemistry
Interface Processor (MCIP). Therefore, when the
PinG treatment will be applied, the first step begins
during processing with the SMOKE emissions
modeling system. Various criteria for emissions
and/or stack parameter available to the user must
be applied when executing the SMOKE

-------
ELEVPOINT processor. Based upon input criteria,
certain major point sources are designated as PinG
sources. Emission rate criteria are generally
specified in order to select the largest point sources
in an inventory. For example, 47 point sources
were identified from the 1999 inventory as PinG
point sources since their emission rates for NOx
emission rate were greater than 75 tons/d or their
S02 emission rates were greater than 150 tons/d.
An output stack file (i.e. stack_groups file) contains
the stack parameters of the major point sources to
be treated later in the CTM/PinG simulation. This
stack_groups files is also applied during the
SMOKE merge step in order to generate the
companion PinG emissions file.
The PDM processor simulations are performed
in advance of the CTM/PinG modeling runs since
PDM generates a data file (i.e., PDM_PING_1)
needed by the PinG module. In particular, PDM is
executed for a 24-hour period and it contains
methods to compute plume rise, horizontal and
vertical dimensions of each plume section, plume
grid positions, and an important plume flag
(IPLMFLG) variable. In particular, the IPLMFLG
signals the PinG module when to initialize a new
plume section and when to end the simulation of a
particular plume section, which triggers the
feedback of plume material to the CTM grid. The
PDM code also has the capability to continue active
plumes into the next day. Consequently, for
processing cases after the first day, an environment
variable in the run script (IOLDFIL) is revised from 0
to 1, and the PDM data file from the previous day
(i.e., PDM_PING_0) is read in order to properly
initialize the PinG module with active plumes at the
start of the current simulation day. The PDM
processor is executed for each day of a simulation
period. Finally, the number of PinG sources
currently allowed must be less than 100.
As with other process modules, the PinG
algorithm is included in the CTM by the user when
building the model executable code. The PinG
emissions file (i.e., MEPSE_1), MCIP data files, 3-D
emissions file (i.e., EMIS_1), and the PDM output
file (i.e., PDM_PING_1) are used during execution
of the PinG module. The plume concentration file
generated by the PinG module contains the species
concentrations in all plume cells of each active,
subgrid scale plume section. A default set of 8
plume cells resolves the horizontal cross-section of
a plume section. A separate PinG 2-D dry
deposition output file is also generated containing
the species deposition in the grid cells where the
active plume sections are located. After completion
of a CTM/PinG simulation run, a postprocessor
program is available to merge the subgrid scale
plume concentration file (CTM_PING_1) and the
CTM gridded concentration field. The current PinG
module is not applicable for model grid resolutions
below about 12 km.
3. MODEL TEST SIMULATIONS
The model domain defined for CTM/PinG test
simulations consisted of 21 x 21 horizontal grid cells
with a 36 km grid cell size and 21 vertical layers.
The PDM processor and CTM/PinG model runs
were performed for a series of days in July 1995,
which coincides with an intensive field study period
conducted within the region. In the PinG module,
photochemical processes were treated with the CB-
IV chemical mechanism using the Gear solver and
aerosol formation was modeled with the same
aerosol algorithm (AE3) also applied in the CTM.
For these simulations, the PinG module was applied
with a set of 10 attached cells (5 on each side of the
plume centerline) to resolve the horizontal plume
cross-section.
Although the results presented herein are from a
particular case (i.e., July 7, 1995), the findings are
indicative of those found on other days from the
modeling period. Five major point sources were
selected for the plume-in-grid treatment. The point
source emissions treated in PinG were from the
Shawnee (SH), Paradise (PA), Cumberland (CU),
Johnsonville (JV), and Gallatin (GA) fossil-fuel
power plants. Total emissions from continuous
monitoring measurements (CEM) for July 7, 1995 in
Table 1 reveal a large range in the NOx and S02
emissions among these point sources. The
S02/N0x ratio also differs greatly, which should
provide an interesting variation among the
photochemical and aerosol species concentrations
for this set of sources.
Trajectories and growth of plume sections
released at 1500 UTC on July 7, 1995 from these
point sources are shown in the modeling domain in
Figure 2. Due to the steady northwesterly flow on
this day, the point source plumes were transported
in parallel to each other. Other hourly plume
releases also followed similar paths, which will
assist in the interpretation of concentration field
differences to be displayed later. The downwind
movement of each plume section in Figure 2 is in
15 minute increments for the active, subgrid scale
phase. During the PinG simulation, boundary
concentrations are provided by the CTM grid cell
concentrations where each plume section is
situated. At this grid resolution, it is evident that
these plume sections were subgrid scale for several
grid cells downwind of their source locations. Once
a plume section's width reaches the grid cell size,
the PinG simulation of it ends and a feedback to the
grid occurs.
TABLE 1.
Source
NOx/NOx (GA)
S02/NOx
GA
1.0
7.3
JV
2.0
6.0
SH
2.9
2.3
CU
14.8
0.13
PA
14.9
1.7

-------
-13000D -60000 30000 110000 19CQ00 270000 350000 430000 510000 S9C000 670000
X
Figure 2. Trajectory and growth of the plume
sections released at 1500 UTC on 7 July 1995 on
the modeling domain with a 36 km grid cell size.
Figures 3 and 4 display selected species
concentrations in the plume sections from a high
NOx, low S02 source (CU), and a lower NOx, high
S02 source (JV), respectively, at 5 hours after
initialization (release time was 1500 UTC). Note
the plume widths at this time are around 20 km,
which is still much less than the grid cell size. The
chemistry in the high NOx plume in Figure 3
strongly favored HN03 formation and 03 in the
plume core continues to recover from a huge initial
deficit due to the high NO emissions. Fine aerosol
sulfate (S042) is only slightly above background
and the hydroxyl radical (OH) values are lower in
the middle of the plume compared to the plume
edges. In contrast, higher sulfate values were
found in the JV plume, which also exhibited higher
S02 concentrations. Due to lower NOx emissions,
considerably less HN03 was formed in the JV
plume. In addition, excess 03 above background
existed in the JV plume core, while more time was
needed for an 03 excess to develop in the CU
plume. Modeled gaseous species were found to be
comparable to airborne plume measurements for
this case ( Godowitch, 2001).
Results illustrating aerosol sulfate in the center
of plume sections for the releases at 1500 UTC
from these sources are shown in Figure 5.
Interestingly, the two highest NOx sources with
differing S02 emissions yielded the lowest sulfate
concentrations above background values supplied
by CTM. It is believed the more rapid reaction of
N02 with OH caused less OH to be available for the
slower S02 reaction with OH. As a result, sulfate
formation is depressed in the higher NOx plumes.
Indeed, higher S04 levels were found in the point
source plumes exhibiting higher S02 emissions
coupled with lower NOx emissions. These results
are in agreement with emerging observational
aerosol data from plumes (Brock et al, 2002).
PinG MODEL RESULTS - CU Plume
( July 7, S96 : 2030 UTC i AGE = 5 Ins )
I I ' I I I I II I I I
"TT~TTTTTT~r
-12
—9—8—3 0 3 6 9
DISTANCE ACROSS PLUME (km)
Figure 3. Selected species concentrations in the
plume section from a high NOx / low S02 emission
source (CU) at 5 hours after release. Ozone (03) is
the solid red line. Units: S04 (ug/m3), OH (ppt),
•S02 and HN03 (ppb).
PinG MODEL RESULTS - JV Plume
( July 7. 11996 : 2000 UTC : AGE = S hrs )
12-
1
I 10
S02
S04
OH
HM03
11 ii I n i u h ' i
¦15 —12 —9—6—3 0
1 1 I I ' 1 r
-9-6—303 6912
DISTANCE ACROSS PLUME (km)
00

70

60

SO
2*
40
&
30
8
20

t>

0

Figure 4. Selected species concentration in the
plume section from a high S02 emission source
(JV) at 5 hours after release, otherwise, same as
Fig. 3.
PinG Model Results
{ July 7, 1995 : PLUME RELEASE = 1500 UTC )
15 _
1.0
"V,
'sX v

TRAVEL TIME (hours)
Figure 5. Aerosol sulfate concentrations relative to
background in the center of plume sections with
travel time from releases at 1500 UTC on July 7.

-------

0.006 21
0X04
0.002
-0J001
-0J003
-0J005
-0.007
-OJ010
1
I -OJ0O5
I -0.007
* -0J010
-I -0JI12
ppmV
S bJ

or
i i
\ '
PAVE
by
MCNC

1J0 21
0.4
-U
-oi
-14
-2J0
-2S
1	21
July 7,199520:00:00
Min= -0D17 at (9,14), Max= 0J003 at (8,15)
Figure 6. Ozone difference field determined by
subtracting the CTM/NoPinG from the CTM/PinG
concentration results.
S02 emission strength of the source.
Nevertheless, slightly lower S04 concentrations
are found near the point source locations as well
as further downwind in the areas where plume
section handovers to the grid model had
occurred.
4. SUMMARY AND ONGOING WORK
Photochemical and aerosol test simulation
results with the updated CTM/PinG model are
encouraging. The PinG treatment has already
demonstrated a capability of capturing the
observed photochemical behavior of 03 and other
gas species. The aerosol sulfate formation in the
S02-rich model plumes also appears to be similar
to observed plume findings. However, more
analysis is warranted and comparisons with
observed plume aerosol data are planned. Grid-
scale concentration differences are apparent from
model runs with/without the PinG approach.
CTM/PinG simulations on a continental domain
with speciated PM point source emissions are
underway and results are anticipated at a future
meeting.
DISCLAIMER
The research presented here was performed
under the Memorandum of Understanding
between the U.S. Environmental Protection
Agency (EPA) and the U.S. Department of
Commerce's National Oceanic and Atmospheric
Administration (NOAA) and under agreement
number DW13921548. Although it has been
reviewed by EPA and NOAA and approved for
publication, it does not necessarily reflect their
policies or views.
REFERENCES
Two sets of model runs were performed; one
series applied the PinG treatment for the above
group of point sources and another series without
the PinG module. The results for ozone from the
CTM/NoPinG simulation (not shown) revealed an
03 maximum within and immediately downwind of
the source locations due to the notable dilution of
the NOx emissions, which accelerated
photochemical formation of ozone. The 03
differences between the two runs shown in Figure
6 reveal lower 03 concentrations in the
CTM/PinG results, particularly in the grid cells in
the vicinity of the two largest NOx sources located
in Kentucky and central Tennessee. Further
downwind, ozone differences are smaller. The
model results for aerosol sulfate in Figure 7 are
somewhat similar, but strongly dependent on the
Binkowski, F.S., 1999: Aerosols in Models-3
CMAQ. Chap. 10 of Science Algorithm of the
EPA Models-3 Community Multiscale Air Quality
Modeling System. EPA/600.R-99.030, Research
Triangle Park, NC.
Brock, C. A. et al., 2002: Particle growth in the
plumes of coal-fired power plants. J. Geo. Res.,
107, D12,10.1029/2001D001062.
Gillani, N.V. and J.M. Godowitch, 1999: Plume-in-
grid treatment of major point source emissions.
Chap. 9, Science Algorithms of the EPA Models-
3 CMAQ Modeling System. EPA/600/R-99/030,
Research Triangle Park, NC.
Godowitch, J.M., 2001: Results of photochemical
simulations of subgrid scale point source
emissions with the Models-3 CMAQ modeling
system. Preprints, AMS Millennium Symposium
on Atmos. Chem., 14-19 Jan. 2001, Albuquerque,
NM, p. 43-49.
^ -32 !
ug/m**3	1
July 7,199520:00:00
Min= -2J6 at (12,15) Mai= OJG at (12,10)
Figure 7. Aerosol sulfate (S04) difference field
determined by subtracting the CTM/NoPinG from
the CTM/PinG concentrations at the same
afternoon hour in Figure 6.

-------