COMPARISON OF SPATIAL PATTERNS OF POLLUTANT DISTRIBUTION WITH CMAQ
PREDICTIONS
Sharon B. Phillips* and Peter L. Finkelstein1
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711 EPA/600/A-04/090
1Atmospheric Sciences Modeling Division, National Oceanic and Atmospheric Administration, U.S.
Department of Commerce On assignment to the National Exposure Research Laboratory
e-mail: phillips.sharon@epa.gov
Voice (919) 541-2138 Fax (919) 541-0044
1.0 INTRODUCTION
One indication of model performance is the
comparison of spatial patterns of pollutants, either
as concentration or deposition, predicted by the
model with spatial patterns derived from
measurements. If the spatial patterns produced
by the model are similar to the observations in
shape, location, and magnitude it can add to our
confidence that the model is performing well.
However, deriving spatial patterns from measured
pollutant data is not always trivial. Modeling
networks are spatially sparse, and frequently
biased toward certain types of land use. Likewise,
frequently there is measurement bias between
monitoring networks. This study will compare the
observed spatial patterns with those predicted by
EPA's Community Multiscale Air Quality (CMAQ)
model, noting any similarities and differences. We
will consider sulfate (S04), nitrate (N03),
ammonium aerosols (NH4), and ozone (03).
2.0 CMAQ CONFIGURATION
An annual 2001 simulation consisting of a
36x36 km model resolution Lambert conformal
horizontal domain with 14 vertical layers was
conducted over the continental U.S. utilizing
CMAQ. Initial and boundary conditions were
provided by the GEOS-CHEM, a global 3-D
chemistry/transport model. Year-specific 2001
meteorological data provided by Mesoscale Model
(MM5) version 3.6.1 were used and processed by
MCIP v2.3 for model-ready inputs. Anthropogenic
emissions were provided by the 1999 National
* Corresponding author address: Sharon B.
Phillips, Air Quality Modeling Group, Emissions,
Monitoring, and Analysis Division, Office of Air
Quality Planning and Standards, U.S.
Environmental Protection Agency, 109 T.W.
Alexander Drive, MC: D243-01, RTP, NC 27711
Emissions Inventory (NEI). Biogenic emissions
were obtained from BEIS 3.12.
3.0 OBSERVATIONAL DATA
After exploring the development of reliable
spatial models for the monitoring data using a
variety of approaches, we chose the radial basis
functions interpolation procedure for the S04, NH4,
N03 fields. Natural neighbor interpolation was
used for 03 because it handles dense, irregular
networks better. Data from the CASTNet and STN
monitoring networks were combined because their
measurement techniques for S04 and NH4 are
similar. Because no two networks measure
nitrogen aerosols in the same way, we chose to
use total nitrate (N03 + HN03) reported by
CASTNet for the spatial analysis of nitrogen.
Ozone data were taken from the state and local air
monitoring networks (AIRS) and CASTNet.
STN data are collected 1 day in 3 or 6,
depending on the site. CASTNet data are
collected as weekly integrated samples. To
overcome this difference, 28 day averages (lunar
months) were constructed from all networks, with
the start date corresponding to the CASTNet
sampling schedule. Ozone comparisons were
made based upon the maximum hourly average
observed each lunar month. Annual 2001
average spatial plots of observed and predicted
S04, NH4, and N03 are considered here.
4.0	RESULTS
4.1	Sulfate
Examinations of observed and predicted
sulfate spatial patterns reflect CMAQ's predictive
skill in simulating S04 spatial patterns (Figures 1
and 2). The pattern of maximum concentration
predicted by the model is similar to the

-------
interpolated pattern from the data both in location
and in magnitude. The East to West gradient is
similar within the limited resolution of the data.
The only difference seems to be a bit higher
observed concentration in Southern California and
the Pacific Coast.
4.2	Nitrate
Annual averages of total nitrate (N03 + HN03)
observed in CASTNet and predicted by CMAQ are
given in Figures 3 and 4. The patterns are very
similar, with maxima of the same magnitude in
Indiana and Ohio, and secondary maxima in
Southeastern Pennsylvania and Southern
California. The East-West gradients are also quite
similar.
4.3	Ammonium
Predicted and observed patterns of NH4 are given
in Figures 5 and 6. They are remarkably similar,
with maxima in the upper mid-west, in southern
Pennsylvania, and in Northern Georgia. They
differ in that CMAQ also predicts a secondary
maxima in Eastern North Carolina that is not
observed by the stations in the area. Patterns in
the West are quite similar, although one observing
station measured high levels in Salt Lake City that
aren't predicted by the model.
4.4	Ozone
Maximum observed and predicted hourly level of
ozone during lunar month 8, the highest month for
ozone, are shown in Figures 7 and 8. The distinct
spatial patterns that are seen in other pollutants
are not present in ozone; however there are
similarities between the model forecasts and
observations. Both have maxima along the
Northeast coast between Washington DC and
Boston as well as along the Southern California
Coast. They also agree in showing few major
maxima, but scattered values between .08 and .1
ppm throughout the rest of the country.
5.0 SUMMARY
Spatial patterns predicted by CMAQ are very
similar to those interpolated from observations in
both location and magnitude. These similarities
suggest that CMAQ is doing a credible job of
predicting air pollutant concentrations on this
scale.
6.0 Disclaimer
The research presented here was partially 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.

-------
Fig. 1 Spatial plot of Observed 2001 Annual
Sulfate from STN and CASTNet.
Fig. 2 Spatial plot of Predicted 2001 Annual
Sulfate from CMAQ.
-125 -120 -1 1 5 -1 10 -105 -100 -95	-90	-85	-80	-75	-70
Fig. 3 Spatial plot of Observed 2001 Annual Total
Nitrate (N03 + HNQ3) from CASTNet.
-125 -120 -1 1 5 -1 10 -105 -100 -95	-90	-85	-80	-75	-70
Fig. 5 Spatial plot of Observed 2001 Annual
Ammonium from STN and CASTNet.
Fig. 4 Spatial plot of Predicted 2001 Annual Total
Nitrate (N03 + HN03) from CMAQ.
50—1—1	1	1	1	1	1	1	1	1	1	1	1	r
¦U^r. "v-"~
.Yn— % *
o
*	7	1 ,
V
-125 -120 -1 15 -1 10 -105 -1 00 -95	-90	-85	-80	-75	-70
Fig. 6 Spatial plot of Predicted 2001 Annual
Ammonium from CMAQ.

-------
Fig. 7 Spatial plot of Observed Maximum hour
Ozone during LM 8 (July 17 - August 13, 2001)
from CASTNet and AIRS.
Fig. 8 Spatial plot of Predicted Maximum hour
Ozone during LM 8 (July 17 - August 13, 2001)
from CMAQ.

-------