EPA/6Q0/A-95/1 28
9.1 INVESTIGATION OF PHOTOCHEMICAL MODELING OF POINT SOURCE POLLUTANTS
WITH EULERIAN GRID AND LAGRANGIAN PLUME APPROACHES
James M. Godowitch*
Atmospheric Sciences Modeling Division
Air Resources Laboratory
National Oceanic and Atmospheric Administration
Research Triangle Park, North Carolina
1. INTRODUCTION
Significant amounts of nitrogen oxides (NOx) and
sulfur oxides (SOx) are emitted from individual
elevated point sources. The primary emission of
NOx, in particular, are important precursors in the
photochemical generation of a variety of secondary
oxidants, such as ozone (03). Since major point
sources are distributed inhomogeneously in space and
the physical dimensions of a pollutant plume increase
at a finite rate with distance downwind, a key aspect
for realistic modeling of this notable class of
emissions is to be able to properly resolve the spatial
scale of the pollutant plume. From a photochemical
perspective, a fresh power plant plume represents a
high NOx, low HC (hydrocarbon) regime, while the
surrounding environment can often be in the exact
opposite regime. As the plume spreads horizontally
and vertically due to atmospheric dispersion
processes, it evolves chemically as background
pollutants are entrained and mixed with in-plume
pollutants. Consequently, model resolution and
plume growth rate are important factors which
strongly influence the photochemical evolution of
plumes.
The modeling approaches selected to simulate
anthropogenic emissions include the Eulerian grid and
Lagrangian plume techniques. The widely-used
treatment in many Eulerian grid models is to inject
the point source emissions into a particular grid cell.
However, the initial lateral dimension of a plume is
generally much smaller than the grid cell size. Thus,
the instant dilution of the primary emissions of the
pollutants over the entire grid cell volume can
strongly distort the non-linear photochemistry
* On assignment to the National Exposure Research
Laboratory, U.S. Environmental Protection Agency
Author Address: J. Godowitch, US EPA, MD-80,
Research Triangle Park, NC 27711
process and contribute to model uncertainty. To
compensate for overdilution of a plume, a smaller
grid cell size can be specified. In contrast, the
Lagrangian plume model approach features a set of
expanding, contiguous cells representing the overall
crosswind dimension of a plume. Each plume cell is
treated as a single vertical column extending over the
mixing layer, while an Eulerian grid model generally
contains multiple vertical layers within the boundary
layer. Nevertheless, the artificial dilution due to
dumping of the point emissions into a grid cell is
more severe in the horizontal.
In this paper, results of Eulerian grid and
Lagrangian photochemical model simulations of
emissions from a major elevated point source are
presented. A series of simulations with grid sizes
varying from 30 km to 2 km were performed with the
Urban Airshed Model, a photochemical grid model, in
order to examine the capability of the model to
resolve plume features and to emulate the chemical
evolution of the pollutant plume. Differences in
pollutant concentrations with emphasis on 03 for
various grid sizes are highlighted. The grid model
simulation results are compared to 03 concentrations
generated from a Lagrangian reactive plume model.
Both models were applied to treat the transport,
dispersion, and photochemical processes impacting a
pollutant plume. In order to promote the comparison
and interpretation of the model results, meteorological
inputs for the Lagrangian plume model were provided
through an interface program using the input data sets
employed to drive the grid model.
2. MODEL DESCRIPTION
2.1 Eulerian Grid Model
The Urban Airshed Model (UAM) is a three-
dimensional, Eulerian photochemical grid model that
treats the physical and chemical processes affecting
pollutant concentrations from both surface area and/or
elevated point sources. The regulatory model version
(UAM-IV), described by Scheffe and Morris (1993)

-------
has been widely applied in metropolitan areas
required to demonstrate attainment of the maximim
hourly 03 standard. The UAM is based on the
atmospheric continuity equation for a pollutant
species, which represents the mass balance of
emissions, transport, diffusion, photochemical
reactions, and dry deposition processes. Of particular
relevance, its treatment for turbulent diffusion
involves a constant horizontal eddy diffusivity
coefficient (Ky= 50m5/s). However, a significant
factor contributing to horizontal dispersion in grid
models comes from the numerical advection scheme.
In the standard UAM, the Smolarkiewicz (1983)
method is used as the horizontal advection technique.
A vertical eddy diffusivity (KJ, which is a function
of stability and turbulence parameters, maintains
strong vertical exchange between model layers below
the mixing height. Finally, the Carbon Bond IV
mechanism (CBM-IV) containing 86 reactions and 33
individual species simulates the photochemical
reaction processes. Emissions from a point source
are processed to determine plume to determine the
appropriate vertical layer for injection. Morris and
Myers (1990) also give additional details about the
techniques for the atmospheric processes in the
regulatory UAM and running the modeling system.
In this modeling study, a modified version of the
UAM was applied. Modifications were made to
allow for 3-D inputs of additional meteorological
parameters as described by Godowitch et al. (1992).
In addition, the horizontal advection scheme of Bott
(1989), which exhibits less numerical diffusion
(Odman et al, 1995), was installed as an alternative
to the method in the regulatory UAM,
2.2 Lagrangian Reactive Plume Model
The reactive plume model version IV (RPM-IV)
described in EPA (1993) is a Lagrangian
photochemical plume model that simulates the
transport, diffusion, and chemistry in an array of
adjacent cells representing a plume cross-section
perpendicular to the wind flow. The RPM-IV model
treats the entrainment of ambient pollutants into the
edge plume cells and entrainment/detrainment
between inner plume cells gradually as the horizontal
dimension of the plume increases downwind. The
transport speed, and the horizontal and vertical
dimensions of the plume section must be provided as
a function of downwind distance. The model was
designed to simulate plume sections under convective
conditions since each plume cell represents a single
well-mixed vertical column over the depth of the
mixing layer. The plume model employs the same
CBM-IV photochemical mechanism as the UAM. A
more detailed description of the development and
attributes of this model are provided by Stewart and
Liu (1981).
3. MODEL SIMULATIONS AND RESULTS
Both models were applied to simulate the
concentrations in the plume emanating from a major
elevated point source situated west of St. Louis,
Missouri for a historical case study day (9 July 1976)
when experimental measurements were also obtained
to characterize the transport, dispersion, and
transformation of the pollutant plume during the
daytime (Gillani and Wilson, 1980). The emission
rate of 1.88 kg/s of NOx for each model was
prescribed with NO representing 95% of the
emissions and the remainder as N02. The stack
parameter information consisted of a 214 m stack
height, 8.8 m stack diameter, 27.4 in/s exist velocity,
and a 441 °K exit temperature.
A modeling domain was defined to be 240 km on
a side and a series of simulations were performed
with grid cell sizes of 2, 4, 8, 16, and 30 km. While
the latter grid specifications are comparable to typical
regional scale grid cells, the 2 km grid cell size is
close to a lower limit for the grid model. The vertical
model layer structure, except for the 2 km grid size,
was typical of most applications with 5 layers; 2
layers below and 3 layers above the mixing height
For the 2 km grid simulation, one lower and one
upper layer were modeled. Gridded input data sets
for each cell size were generated by a meteorological
processor program (Godowitch et al., 1992), which
employs diagnostic/interpolative methods to derive the
hourly 3-D wind, temperature, and the 2-D mixing
height fields. Observations included the routine
surface hourly observations from Lambert STL airport
and upper air profiles obtained four times daily at
four special-study rawinsonde sites in the region.
Detailed pollutants measurements needed to define
initial and boundary conditions were unavailable.
Since this study is not intended as a model
evaluation effort, the same initial (IC) and constant
boundary (BC) conditions were specified with
reasonable background values for consistency in all
simulations. Concentrations levels for IC and BC
included; total HC of 100 ppbC partitioned into the
various organic classes defined in the CBM
mechanism, NO set to 0.25 ppb, NO, at 0.75 ppb. and
03 set at 60 ppb. Model runs were performed with
point source emissions starting at 12 LST with

-------
Figure 1. Spatial patterns of maximum excess ozone above background of the point source plume from 12-18 LST
from grid model simulations at grid cell sizes of a) 30 km and b) 4 km with the same domain. The
Mississippi River (dotted line) is also displayed.
continuous emission rates for six hours. Surface area
emissions were prescribed to be zero in order to focus
on the modeled plume concentrations.
The maximum excess 03 field above background
over the afternoon period is displayed in Figure 1 to
illustrate the differences in the overall size of the
pollutant pattern and 03 concentration in the plume
for a coarse (30 km) and a fine (4 km) grid
simulation. The 30 km grid results exhibit a
considerably broader plume 03 pattern and the excess
plume ozone above background is less than in the 4
km grid size results. The coarse grid cell size
provides for much greater dilution of the primary NO
emissions. Of relevance to the photochemical
process, the HC/NOx ratio in the source cell for these
grid sizes differed by nearly a factor of 10 with
values near 30 for the coarse grid simulation.
Consequently, photochemical 03 production started
in the source cell and the peak 03 occurred relatively
close to the source for this coarse grid. Table 1
contains results from the model simulations for the
different grid sizes. It reveals that as the grid size
decreased the downwind location of the maximum 03
concentration in the plume increased. Another
interesting result in Table 1 is that a notable
decreasing trend in the excess 03 occurred at the
finer grid sizes of 2 and 4 km. These grid cell sizes
are much closer to actual plume dimensions,
especially at an early stage.
The 03 values relative to background for the
plume at hour 18 LST from the 2 km grid simulation
are shown in Figure 2. It reveals that important
features of chemical evolution in the plume as
described by Gillani and Wilson (1980) were
captured. Up to about 50 km downwind an ozone
deficit below background exists in the plume due to
ozone depletion by high NO concentrations from the
primary emissions. Ozone concentrations recover
gradually in the plume's core and in an intermediate
phase, 03 formation along the plume edges occurs as
mixing with higher HC concentrations contained in
surrounding grid cells leads to characteristic bulges in
03 concentrations. Further downwind for this source
strength, the maximum excess 03 in the plume is
found in the mature chemical phase.
For simulations with the Lagrangian RPM, inputs
for winds, mixing heights, temperatures, and
background concentrations were obtained from the
appropriate grid cell where the plume cross-section
was situated by applying an interface processor
designed to couple these models (EPA, 1993). An
additional grid model simulation was conducted with
zero emissions in order to generate gridded
background concentrations for use with the plume
model. Furthermore, lateral plume dimensions were
provided from an analytical curve fit to observed
plume widths (Godowitch et al„ 1995). The RPM
simulation commenced at 12 LST with the release of
a single plume cross-section using the same emission
rates as the grid model and the plume section was
modeled for 8 hours.

-------
Table 1. Grid Model Results from Point Source Emissions Simulations
Grid Cell Size (km)
Excess Plume Ozone* (ppb)
Downwind Distance (km)
30
32.1
67
16
43.6
71
8
49.4
102
4
46.0
101
2
40.0
112
* Note: Concentration values minus background for hour 17-18 LST
Figure 3 depicts the 03 values relative to
background concentration for the expanding horizontal
plume section at various downwind distances. The
notable features of the evolutional^ cycle of plume 03
are certainly evident in these selected plume cross-
sections. A deep ozone deficit occurs near the source,
and the transition toward a recovery of the 03
concentrations to near background values is seen in
the plume section at 50 km downwind, as well as
characteristic bulges of ozone along each edge of the
plume. A maximum excess ozone of about 25 ppb
was produced in the 150 km downwind distance
displayed in Figure 3. The maximum 03 generated in
the RPM was less when compared to the grid model
values, even in the 2 km grid simulation. A
preliminary explanation for the differences in the
excess 03 between these modeling approaches is
their different treatments and rates of diffusion,
however, further analyses are warranted.
4. SUMMARY
A limited modeling study of point source
emissions was performed with Eulerian grid and
Lagrangian plume models to examine the comparative
predictions of ozone concentrations in a pollutant
plume governed by transport, diffusion, and
photochemical processes. Results indicated that the
grid model was able to replicate the evolution of
ozone in the pollutant plume for this rather large
emissions source at the finest grid resolution. As
anticipated, simulation results for different grid cell
sizes also revealed the maximum 03 occurred further
downwind as the model grid size was decreased, and
the peak ozone level was lowest for the coarsest grid
size (30 km). Interestingly, maximum 03 values were
on a downward trend for the selected smaller grid cell
sizes for the large emissions source strength employed
in this study. A comparison between the modeling
approaches indicated the maximum ozone generated
from the RPM approach was lower than values from
the grid model simulations. Since attempts were
made to simulate the same processes and provide
common inputs to both models, the difference in
excess plume ozone is primarily attributed to the
different technical methods for diffusion in these
modeling approaches. Additional analysis of the
results to assess the impact on other secondary species
concentrations (ex., PAN, HN03) and model runs with
other (lower) emission rates are underway.
Considerable interest exists and efforts are
underway in plume-in-grid treatments, whereby a
plume module is embedded in an Eulerian grid
framework to treat point source plumes during a
subgrid scale phase. The plume-in-grid approach
could greatly reduce the rather high computational
burden encountered with the fine scale gridded
domains used in this study. Further modeling studies
of each modeling approach, such as this one, are
warranted since results could have important
implications for plume-in-grid treatments and can also
provide valuable information for the transfer of
pollutant plumes over to the grid system.
Disclaimer - This paper has been reviewed in
accordance with the U.S. Environmental Agency's
peer and administrative review policies and approved
for presentation and publication. Mention of trade
names or commercial products does not constitute
endorsement or recommendation for use.
REFERENCES
Bott, A., 1989: A positive definite advection
scheme obtained by nonlinear
renormalization of the advective fluxes.
Mon. Wea. Rev., 117, 1006-1015.

-------
EPA, 1993: Reactive Plume Model IV (RPM-IV)
User's Guide, Office of Air quality Planning
and Standards, U.S. Environmental
Protection Agency, EPA-454/B-93-012,
Research Triangle Park, NC.
Gillani, N.V. and W.E. Wilson, 1980: Formation
and transport of ozone and aerosols in power
plant plumes. Annals of New York Academy
of Sciences, 338, 276-296.
Godowitch, J.M., R.T. Tang, and J. Newsom, 1992:
Development of an improved urban airshed
modeling system, 85th Annual Meeting of
the Air and Waste Manage. Assoc., Kansas
City, MO, 92-86.01, AWMA, Pittburgh, PA.
Godowitch, J.M., J. Ching, and N.V. Gillani, 1995:
A treatment for Lagrangian transport and
diffusion of subgrid scale plumes in an
Eulerian grid framework, 11th AMS Symp.
on Boundary Layers & Turb., Charlotte, NC,
American Meteorol. Soc., Boston, MA,
p. 86-89.
Morris, R.E. and T.C. Myers, 1990: User's guide
for the urban airshed mode!: Vol. I: User's
manual for UAM-IV. Office of Air Quality
Planning & Standards, U.S. EPA, EPA-
450/4-90-007A,Research Triangle Park, NC.
 i 1 l l,„
, I , , , ,V,

-15	-10	-S	0	S	10	15
PLUME CROSS-SECTION (km)
Figure 3. Relative ozone values in the horizontal
plume section from the RPM simulation at various
downwind distances depict the chemical evolution of
ozone in the point source plume. The RPM simulated
10 subplume cells, with the value on each edge being
the background ozone specified by the grid model.

-------
TECHNICAL REPORT DATA
1. REPORT NO.
EPA/600/A-95/128
2.
3 .R
4. TITLE AND SUBTITLE
Investigation of Photochemical Modeling of Point
Source Pollutants with Eulerian Grid and Lagrangian
Plume Approaches
5.REPORT DATE
g.PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
James M, Godowitch
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
NATIONAL EXPOSURE RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NC 27711
13.TYPE OF REPORT AND PERIOD
COVERED
Proceedings, FY-95
14. SPONSORING AGENCY CODE
EPA/600/9
15. SUPPLEMENTARY NOTES
16. ABSTRACT
In this paper, results of Eulerian grid and Lagrangian photochemical model simulations of
emissions from a major elevated point source are presented. A series of simulations with grid sizes
varying from 30 km to 2 km were performed with the Urban Airshed Model, a photochemical grid model, in
order to examine the capability of the model to resolve plume features and to emulate the chemical
evolution of the pollutant plume. Differences in pollutant concentrations with emphasis on 03 for
various grid sizes are highlighted. The grid model simulation results are compared to 0,
concentrations generated from a Lagrangian reactive plume model (RPM). Both models were applied to
treat the transport, dispersion, and photochemical processes impacting a pollutant plume. In order to
promote the comparison and interpretation of the model results, meteorological inputs for the Lagrangian
plume model were provided through an interface program using the input data sets employed to drive the
grid model. Results indicated that the grid model was able to replicate the evolution of ozone in
the pollutant plume for this rather large emissions source at the finest grid resolution. As
anticipated, simulation results for different grid cell sizes also revealed the maximum 0:) occurred
further downwind as the model grid size was decreased, and the peak ozone level was lowest for the
coarsest grid size (30 km) . Interestingly, maximum 0, values were on a downward trend for the selected
smaller grid cell sizes for the large emissions source strength employed in this study. A comparison
between the modeling approaches indicated the maximum ozone generated from the RPM approach was lower
than peak values from the grid model simulations.	
17.	KEY WORDS AND DOCUMENT ANALYSIS
a. DESCRIPTORS
b. IDENTIFIERS/ OPEN ENDED
TERMS
c.COSATI



IS. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19. SECURITY CLASS {This
Report}
UNCLASSIFIED
21.NO. OF PAGES

20. SECURITY CLASS (This
Page)
UNCLASSIFIED
22. PRICE
\ ^	Xn\i q_sV .~3> m

-------