EVALUATING OZONE PREDICTIONS FROM PHOTOCHEMICAL MODELS USING
NE-OPS 1999 OBSERVATIONS
Qing Sun, Anatharaman Chandrasekar, Panos G. Georgopoulos*
Environmental and Occupational Health Sciences Institute (EOHSI), a joint project of UMDNJ R. W. Johnson Medical School and Rutgers the State University
C. Russell Philbrick
Penn State University, Department of Electrical Engineering and Applied Research Laboratory,
University Park, PA, USA
Bruce Doddridge
University of Maryland, Department of Meteorology, College Park, MD, USA
1. INTRODUCTION
2. PROBLEM SPECIFICATIONS
The North American Research Strategy for
Tropospheric Ozone - North East Oxidant and
Particle Study (NARSTO-NE-OPS) is a multiinstitutional collaborative research program set
up by USEPA to improve current understanding
of the underlying causes for the occurrence (and
concurrence) of high ozone and fine particle
concentration levels in the North-eastern United
States. Various advanced meteorological and air
chemistry measurements were made in the
vicinity of Philadelphia, Pennsylvania during two
field campaigns conducted during the summers
of 1998 and 1999 (Philbrick, 2000).
Fast et al. (2002) evaluated an Eulerian
chemical transport model developed at PNNL
against NE-OPS data obtained during the period
July 23 to August 11, 1999. The present
investigation was primarily focused on a major
ozone episode that took place in July 15-19,
1999 over the Philadelphia region, to perform an
extensive evaluation of two widely used
regional/multiscale photochemical models,
namely, CMAQ (Byun, 1999) and CAMx
(ENVIRON, 2002), in predicting ozone
concentration by comparing model outcomes
with both surface measurement data from
USEPA’s Aerometric Information Retrieval
System (AIRS) and upper air aircraft data
available from the NE-OPS study (Doddridge,
2000).
The simulations for this study were carried
out for the July 11, 1999 00 UTC to July 25,
1999 12 UTC period. Three levels of nested
grids were used with grid resolutions of 36km,
12km and 4km (see Figure 1). The 36km grid
encompasses the Eastern United States while
the 4km grid encampasses the Philadelphia New Jersey region. In the vertical direction, the
CMAQ application used a non-hydrostatic
coordinate with 14 layers centered at 0.9975,
0.9925, 0.985, 0.9725, 0.955, 0.9325, 0.9, 0.84,
0.75, 0.65, 0.525, 0.375, 0.225 and 0.075 in
sigma-p units, while the CAMx application used
8 layers corresponding to the lowest 8 layers for
the CMAQ simulations.
* Corresponding author address: Panos G.
Georgopoulos, Environmental and Occupational
Health Sciences Institute (EOHSI), 170 Frelinghuysen
Road - Busch Campus, Piscataway, NJ 08854, Email:
panosg@fidelio.rutgers.edu, Phone (732) 445-0159,
fax (732) 445-0915
36 km Resolution Domain
12 km Resolution Domain
4 km Resolution Domain
Figure 1. Nested air quality modeling domains
with 36km, 12km and 4km horizontal grid
resolutions employed in the present study.
In order to obtain the meteorological inputs,
simulations were performed with the Fifth
Generation Pennsylvania State
University/National Center for Atmospheric
Research (NCAR) Mesoscale Model (MM5)
(Grell, 1994) model for the nested multiscale
model domain and for the duration of the
modeling period. Details of the MM5 simulations
can be found in Chandrasekar et al. (2002a,b,
2003).
The emissions data were processed from
the National Emissions Trends (NET) (USEPA,
1999) inventory using MCNC's Sparse Matrix
Operator Kernel Emissions (SMOKE) (Houyoux,
1999) modeling system.
The ozone concentration data from AIRS
and from the monitor station at Philadelphia Air
Management Services Laboratory, made
available through the NE-OPS study, were used
for comparisons with the model predicted ozone
values. The upper air ozone data used for
comparison with model prediction were taken
from the University of Maryland instrumented
flights with Cessna and Aztec aircrafts
(Doddridge, 2000).
Both CMAQ and CAMx simulations were
performed for the three levels of nested grids.
Figure 3 shows the daily maxima of ground level
ozone spatial distributions for 7/17/1999, as
predicted by CMAQ and CAMx, respectively,
over the 4km resolution domain; one can see
that CAMx predicts higher peaks while CMAQ
predicts wider extent of ozone. Figure 4
presents space-time paired plots and quantilequantile plots for ground level ozone
concentrations observed at fourteen AIRS ozone
monitor stations in New Jersey and one NEOPS station at Philadelphia during July 11-25,
1999, versus model predictions by CMAQ and
CAMx, respectively, at the locations of those
monitor stations, at the hours when those
monitor data were collected. One can see that
for the middle range of concentrations the two
models calculate similar concentration values
but for the lower concentration range the CMAQ
predictions tend to be higher than those of
CAMx while for the high concentration range
CAMx tends to overpredict ozone values.
3. RESULTS
Figure 2. Air quality monitoring stations and
flight tracks in the vicinity of NE-OPS. The
circles indicated in the figure correspond to a
radius of 50 Km and 100 Km with the Baxter NEOPS site at the center.
Figure 2 shows the air quality monitoring
stations and flight tracks in the vicinity of NEOPS Baxter site. Also depicted in the same
figure (in purple) is the area representing urban
Philadelphia, the region representing the focus
of a population exposure study to Ozone and
fine Particulate Matter during the summer period
of 1999 (Georgopoulos et al., 2003).
Figure 3. Daily maxima of ground level ozone
spatial distributions for 7/17/1999 predicted by
CMAQ (upper) and CAMx (lower) for the 4km
resolution grid.
Figure 6 shows comparisons of upper air
predictions of ozone concentrations from both
CMAQ and CAMx with aircraft data collected
during the NE-OPS study on a flight on July 18,
1999. One can see for the July 18 evening flight,
that both model predictions agree with the
measurement data for altitudes in the range of
200-600 m, and both under-predict, to a different
degree, for altitudes higher than 700m and also
at the surface.
Figure 5 shows comparisons of 4km
resolution CMAQ and CAMx ozone time series
predictions with observed data from monitors
located in Middlesex, NJ and Philadelphia, PA.
The observation data for Philadelphia were
obtained from the NE-OPS study, while those for
NJ are from USEPA's AIRS database. One can
see an over-prediction of ozone by CAMx on
certain high ozone days for Middlesex.
40
0
40
80
120
Observed Concentrations (ppb)
120
80
40
0
160
0
40
80
120
Observed Concentrations (ppb)
160
CAMx−4km Predicted Concentrations (ppb)
80
CMAQ−4km Predicted Concentrations (ppb)
120
0
160
160
CAMx−4km Predicted Concentrations (ppb)
CMAQ−4km Predicted Concentrations (ppb)
160
120
80
40
0
160
0
40
80
120
Observed Concentrations (ppb)
120
80
40
0
160
0
40
80
120
Observed Concentrations (ppb)
160
Figure 4. Space-time paired comparisons and quantile-quantile plots of ground level ozone
concentrations during July 11-25, 1999 over the 4 km resolution domain: the observation data were
measured at 14 AIRS stations located in New Jersey plus the NE-OPS station located in Philadelphia: (a)
observed versus CMAQ predicted, space-time pairs; (b) observed versus CAMx predicted, space-time
pairs; (c) observed versus CMAQ predicted, quantile-quantile; (d) observed versus CAMx predicted,
quantile-quantile.
PHILADELPHIA, PENNSYLVANIA, Monitor ID: Baxter Phil-AMS ELM
Ozone Concentration (ppb)
Ozone Concentration (ppb)
MIDDLESEX, NEW JERSEY, Monitor ID: 340230011442011
180
monitor
160
CMAQ-cb4-4km
140
CAMx-chem3-4km
120
100
80
60
40
20
0
07/11/99
07/13/99
07/15/99
07/17/99
07/19/99
07/21/99
07/23/99
07/25/99
180
monitor
160
CMAQ-cb4-4km
140
CAMx-chem3-4km
120
100
80
60
40
20
0
07/11/99
07/13/99
07/15/99
07/17/99
time (EDT)
07/19/99
07/21/99
07/23/99
07/25/99
time (EDT)
Figure 5. Ozone time series comparisons between CMAQ and CAMx 4km resolution model predictions
and observation data for Middlesex, NJ and Philadelphia, PA. The observation data for Philadelphia were
obtained from the NE-OPS study, while those for Middlesex, NJ are from USEPA's AIRS database.
19a spiral up O3 7/18/1999 20:46-22:36 EDT
Aircraft
CAMx-4km
CMAQ-CB4-4km
1400
1200
1200
1000
1000
Height (m)
Height (m)
19a spiral down O3 7/18/1999 20:46-22:36 EDT
Aircraft
CAMx-4km
CMAQ-CB4-4km
1400
800
600
400
400
200
200
0
4. CONCLUSIONS
800
600
0
0
50
100
150
O3 Concentration (ppb)
200
measurements on 7/18/1999 20:46-22:36 EDT.
(left panel: spiral up; right panel: spiral down)
0
50
100
150
O3 Concentration (ppb)
200
Figure 6. Comparison of 4 km resolution CMAQ
and CAMx ozone predictions with flight
The comparisons of CMAQ and CAMx model
predictions with surface measurement data from
AIRS and NE-OPS show relatively reasonable
agreements for ozone predictions. The model
predictions capture the general trends of change
in the time series plots. Considerable
discrepancies can be seen from the comparison of
upper air model predictions and aircraft
measurement data. It is therefore necessary to
further examine the options and assumptions
underlying the application of these models in order
to identify the causes of this discrepancy; though it
may be reasonable to assume that the use of a
higher number of layers in the CMAQ application
(fourteen layers) versus that in the CAMx
application (eight layers), which was made to
reflect typical practice in the application of the two
models, explains the better agreement of CMAQ
results with upper air observation, further
investigation of this issue is necessary.
Atlantic region. In Proceedings of the PM2000:
Particulate Matter and Health Conference,
Charleston, SC, January 25-28, 2000; A&WMA,
2000; pp 4-5.
ENVIRON, 2002: User’s Guide, Comprehensive
Air Quality Model with Extensions (CAMx), version
3.10;
http://www.camx.com/pdf/CAMx3.UsersGuide.020
410.pdf.
ACKNOWLEDGMENTS
Support for this work was provided by the
State of New Jersey Department of Environmental
Protection (NJDEP) funded Ozone Research
Center (ORC); the U.S. EPA Center for Exposure
and Risk Modeling (CERM) (EPAR-827033); and
the U.S. EPA funded NorthEast Oxidant and
Particle Study (NE-OPS) (EPA-TPSU-UMDNJ826373-14). Appreciation is extended to all the
members of the NE-OPS team for their
contributions to the work presented here.
Fast, J.D.; Zaveri, R.A.; Bian, X.; Chapman, E.G.;
Easter, R.C., 2002: Effect of Regional-scale
Transport on Oxidants in the Vicinity of
Philadelphia during the 1999 NE-OPS Field
Campaign; J. Geo. Res. 2002, Vol 107, No. D16.
Georgopoulos, P.G.; Wang, S.W.; Vyas, V.M.;
Sun, Q.; Burke, J.; Vedantham, R.; McCurdy, T.;
Özkaynak, H., 2003: A Source-to-Dose
Assessment of Population Exposures to Fine PM
and Ozone in Philadelphia, submitted to Journal of
Exposure Analysis & Environmental Epidemiology.
REFERENCES
Byun, D. W., 1999: Ching, J. K. S. Science
Algorithms of the EPA Models-3 Community
Multiscale Air Quality (CMAQ) Modeling System;
EPA/600/R-99/030; USEPA Report.
Chandrasekar, A.; Doddridge, B.; Philbrick, C.R.;
Clark, R.; Georgopoulos, P.G., 2002a: A
Comparative Study of Prognostic MM5
Meteorological Model with Aircraft, Wind Profiler,
Lidar, Tethered balloon and RASS data over
Philadelphia during a 1999 Summer Episode,
Journal of Environmental Fluid Dynamics,
accepted for publication.
Chandrasekar, A.; Doddridge, B.; Philbrick, C. R.;
Clark, R.; Georgopoulos, P. G., 2002b: An
evaluation study of RAMS simulations with
Aircraft, Wind Profiler, Lidar, Tethered balloon and
RASS data over Philadelphia during a 1999
Summer Episode, submitted to Environmental
Fluid Mechanics, accepted for publication.
Chandrasekar, A.; Philbrick, C. R.; Doddridge, B.;
Clark, R.; Georgopoulos, P. G., 2003: Evaluating
the performance of a computationally efficient
MM5/CALMET system for developing
meteorological inputs to air quality models,
Atmospheric Environment 37(23): 3267-3276.
Doddridge, B.G., 2000: An airborne study of
chemistry and fine particles over the U.S. Mid-
Grell, G.A.; Dudhia, J.; Stauffer, D.R., 1994: A
description of the Fifth-Generation Penn State
NCAR Mesoscale Model (MM5); NCAR TN-398;
NCAR Technical Note; National Center for
Atmospheric Research, Boulder, Colorado; 138
pp.
Houyoux, M. R; Vukovich, J.M., 1999: Updates to
the Sparse Matrix Operator Kernel Emissions
(SMOKE) Modeling System and Integration with
Models-3. Presented at The Emission Inventory:
Regional Strategies for the Future, 26-28 October,
Raleigh, NC, Air & Waste Management
Association.
Philbrick, C.R.; Clark, R.D.; Koutrakis, P.; Munger,
J.W.; Doddridge, B.G.; Miller, W.C.; Rao, S.T.;
Georgopoulos, P.; Newman, L., 2000:
Investigations of Ozone and Particulate Matter Air
Pollution in the Northeast. In Proceedings of the
PM2000: Particulate Matter and Health
Conference, Charleston, SC, January 25-28,
2000; A&WMA, pp 1-3.
United States Environmental Protection Agency,
National Emissions Trends (NET), 1999:
ftp://ftp.epa.gov/EmisInventory