AERMOD: Model Formulation and Evaluation
Results
99-476
Robert Paine
ENSR Corporation, 35 Nagog Park, Acton, MA 01720
Russell Lee
17 Cobbleridge Court, Durham, NC 27709
Roger Brode
Pacific Environmental Services, 5001 S. Miami Blvd., Research Triangle Park, NC 27709
Robert Wilson
U.S. Environmental Protection Agency Region 10, 1200 6th Avenue, Seattle, WA 98101
Alan Cimorelli
U.S. Environmental Protection Agency Region 3, 841 Chestnut Street, Philadelphia, PA 19107
Steven Perry1
Atmospheric Sciences Modeling Division, Air Resources Laboratory/NOAA (MD-80), U.S.
Environmental Protection Agency, Research Triangle Park, NC 27711
Jeffrey Weil
CIRES, University of Colorado, Boulder, CO 80309
Akula Venkatram
College of Engineering, University of California at Riverside, Riverside, CA 92521
Warren Peters
Office of Air Quality Planning and Standards, U.S. Environmental Protection Agency, Research
Triangle Park, NC 27711
On assignment to the National Exposure Research Laboratory, U.S. Environmental Protection Agency.
-------
ABSTRACT
AERMOD is an advanced plume model that incorporates updated treatments of the boundary
layer theory, understanding of turbulence and dispersion, and includes handling of terrain
interactions. This paper presents an overview of AERMOD's features relative to ISCST3.
AERMOD has been evaluated on 10 databases, which include flat and hilly terrain areas, urban
and rural sites, and a mixture of tracer experiments as well as routine monitoring networks with a
limited number of fixed monitoring sites. This paper presents a summary of the evaluation
results of AERMOD with these diverse databases.
INTRODUCTION
In 1991, the United States Environmental Protection Agency (US EPA), in conjunction with the
American Meteorological Society (AMS), formed the AMS/EPA Regulatory Model
Improvement Committee (AERMIC). AERMIC's charter was to build upon earlier modeling
developments to provide a state-of-the-art dispersion model. The resulting model, AERMOD1, is
the subject of this paper.
AERMOD represents an advance in the formulation of a steady-state, Gaussian plume model. It
is apparent that AERMOD has an advantage over ISCST3 when the various scientific
components are compared. However, AERMOD would be expected to perform at least as well
as or better than the existing modeling techniques.
The performance evaluation of AERMOD involved four short-term tracer studies and six
conventional long-term SO2 monitoring databases in a variety of settings. The purpose of these
studies is to be sure that AERMOD has been tested in the various types of environments for which
it will be used. Compared with other widely used models, AERMOD has been subjected to a large
degree of testing with these evaluation databases.
AERMOD FORMULATION
The focus of the AERMIC group has been on applied models designed for estimating near-field
impacts from industrial source types. The primary products of the ongoing AERMIC
development work are the AERMOD (AERMIC Model) dispersion model, the AERMET
meteorological preprocessor, and the AERMAP terrain preprocessor.
The development of a new model is generally dependent not only on published research in
atmospheric diffusion, but also on model development work that has gone on before. This is
certainly true with AERMOD. A "new generation plume model" is not simply a variation on the
traditional Gaussian plume model, but, instead, takes advantage of more recent research on
turbulence and diffusion in the atmosphere. Other models in this category include PPSP2,
HPDM3, TUPOS", CTDMPLUS5, and, more recently, ADMS6 (developed in the United
Kingdom) and OLM7 (developed in Denmark). AERMIC members were involved in the
development of three of these models, PPSP, CTDMPLUS and HPDM. As with most
technological developments, AERMOD was built on the knowledge and experience gained from
the development of these earlier models.
The AERMOD modeling system is composed of one main model (AERMOD) and two
preprocessors—a meteorological preprocessor (AERMET) and a terrain preprocessor
-------
(AERMAP). AERMET calculates hourly boundary layer parameters for use by AERMOD,
including friction velocity, Monin-Obukhov length, convective velocity scale, temperature scale,
convective boundary layer (CBL) height, stable boundary layer (SBL) height, and surface heat
flux. In addition, AERMET passes all observed meteorological parameters to AERMOD
including wind direction and speed (at multiple heights, if available), temperature, and, if
available, measured turbulence. AERMOD uses this information to calculate concentrations in a
manner that accounts for changes in dispersion rate with height, allows for a non-Gaussian plume
in convective conditions, and accounts for a dispersion rate that is a continuous function of
meteorology. In contrast, ISCST3 assumes that the dispersion rate is constant with height, that
the plume is always Gaussian in form, and is based on discrete dispersion (stability) categories
that were developed in the 1960's and can result in jumps in calculated concentrations with small
changes in meteorology. AERMAP prepares terrain data for use by AERMOD in complex
terrain situations. This allows AERMOD to account for terrain using a simplification of the
procedure used in the CTDMPLUS model5. Table 1 summarizes the differences between
AERMOD and ISCST3 (space limitations prevent the inclusion of contrasts between AERMOD
and other models such as CTDMPLUS). Detailed descriptions of the formulations are presented
by Cimorelli, et al1.
MODEL EVALUATION DESIGN
The evaluation of AERMOD has been accomplished in two phases. The first phase, the
"developmental evaluation," was performed concurrently with the development of the model. As
each feature of the model was added, a relevant portion of the developmental evaluation was
repeated with five databases to identify any problems that might have been introduced at that
stage of the model's development. Because of the possibility that the model may have been
inadvertently biased to fit particular characteristics of the developmental databases used, a
second phase, the "independent evaluation," was conducted using three additional data sets. This
second evaluation was conducted with a minimum of model changes (only those required to fix
run-time errors or to correctly implement the model formulation).
Developmental Evaluation
AERMOD is intended to handle a variety of pollutant source types (including surface and
buoyant elevated sources) in a wide variety of modeling situations (including rural, urban, flat
terrain and complex terrain). With this in mind, data from five diverse field studies were selected
for the developmental evaluation. Due to space limitations, maps of the various sites are not
provided in this report, but can be found in Paine et al8.
The Prairie Grass study (Barad9, Haugen10) used a near-surface, non-buoyant tracer release in a
flat rural area. The Prairie Grass study involved a tracer of SO2 released at 0.46 m above the
surface. Surface sampling arrays (arcs) were positioned from 50 m to 800 m downwind.
Meteorological data included 2-m wind speed, sigma-theta, and delta T (2 m - 16 m). Other
surface parameters, including friction velocity, Monin-Obukhov length, and ay were estimated. A
total of 44 10-minute sampling periods were used, including both convective and stable
conditions.
The Kincaid SF6 study (Liu and Moore"; Bowne, et al.12) consisted of an elevated, buoyant
tracer release in a flat rural area. Six weeks of intensive study was conducted during the spring
-------
and summer of 1980 and 1981. During this study, approximately 200 monitors providing 1-hour
averaged samples were placed in arcs from about 500 m to 50 km downwind of the single 187-m
stack. Meteorological data included wind speed and direction, u-v-w winds, delta T from a 100-
m instrumented tower, delta T from a 10-m instrumented tower, and nearby National Weather
Service (NWS) data. Estimates of lateral plume spread (ay) are available from the sampling arcs.
The Indianapolis study (Murray and Bowne13) consisted of an elevated, buoyant tracer (SF6)
released in an urban area. The site is a flat-terrain, urban to suburban area with a single 84-m
stack. Data are available for approximately a four- to five-week period with 177 monitors
providing 1-hour averaged samples in arcs from 250 m to 12 km downwind. Meteorological data
included wind speed and direction, a0 on a 94-meter tower; and wind speed, AT (2m - 10m) and
other supporting surface data at three other towers. Observed plume rise and estimates of plume
ay are also available from the database.
The Kincaid SO2 study (Liu and Moore" Bowne, et al.12) consisted of a buoyant, continuous
release of SO2 from a 187-m stack. The site is in a rural area in flat terrain. The study includes
about six months of data between April 1980 and June 1981. There were 30 SO2 monitoring
stations providing 1-hour averaged samples from about 2 km to 20 km downwind of the stack.
The meteorological data are the same as in the Kincaid tracer study.
The Lovctt Power Plant study (Paumier et al.14) consisted of a buoyant, continuous release of
SO2 from a 145-m tall stack. The site is located in complex terrain in a rural area. The data spans
one year from December 1987 through December 1988. Data were collected from 12 monitoring
sites (10 on terrain, 2 as background) providing 1-hour averaged samples that were located about
2 to 3 km from the plant. The important terrain features rise approximately 250 m to 330 m
above stack base. The monitors on terrain are generally about 2 to 3 km downwind from the
stack. Meteorological data include winds, turbulence, and delta T from a tower instrumented at
10 m, 50 m, and 100 m. NWS surface data were obtained from a station 45 km away.
Independent Evaluation
The independent evaluation of AERMOD initially employed the first three databases described
below. Results for two additional databases were added to respond to comments by peer
reviewers of AERMOD.
The Baldwin Power Plant15 is located in a flat terrain setting of southwestern Illinois. Three
184-meter stacks aligned approximately north-south with a horizontal spacing of about 100
meters between each stack were modeled for this evaluation. There were 10 SO2 monitors
providing hourly averages that surrounded the facility, ranging in distance from two to ten
kilometers. On-site meteorological data from the Baldwin field study covered the period from
April 1, 1982 through March 31, 1983 and consisted of hourly wind speed, wind direction, and
temperature measurements taken at 10 meters and hourly wind speed and wind direction at 100
meters.
The Clifty Creek Power Plant is located in southern Indiana on the north side of the Ohio. The
area immediately north of the facility is characterized by cliffs rising about 115 meters above the
river and intersected by creek valleys. Three 208-meter stacks meters were modeled in this
evaluation. This database was used in a major EPA-funded evaluation of rural air quality
dispersion models in the early 1980s16. There were six SO2 monitors on the surrounding terrain
-------
that provided hourly averaged concentration data. Meteorological data from the Clifty Creek
field study covered the two year period from January 1, 1975 through December 31, 1976,
although only the data from 1975 were used in this evaluation.
The Martins Creek Steam Electric Station (MCSES) is located on the Pennsylvania/New
Jersey border, approximately 30 kilometers northeast of Allentown, PA and 95 kilometers north
of Philadelphia, PA on the Delaware River. The area is characterized by complex terrain rising
above the stacks toward the southeast. The seven SO2 monitors providing hourly averages that
were used in this evaluation17 were on located Scotts Mountain, which is about 2.5 - 8 kilometers
southeast of the Martins Creek facility. On-site meteorological data for the Martins Creek station
covered the period from May 1, 1992 through May 19, 1993. Hourly temperature, wind speed,
wind direction, and aA at 10 meters were recorded from an instrumented tower located in a flat
area approximately 2.5 kilometers west of the Martins Creek power generation station. In
addition, hourly multi-level wind measurements were taken by a SODAR located approximately
three kilometers southwest of the Martins Creek station.
The Westvaco Corporation's pulp and paper mill in Luke, Maryland is located in a complex
terrain setting in the Potomac River valley in western Maryland18. A single 190-m stack was
modeled for this evaluation. There were 11 SO2 monitors surrounding the facility, with eight
monitors well above stack top on the high terrain east and south of the mill at a distance of 800 -
1500 meters. Hourly meteorological data were collected between December 1980 and November
1991 at three instrumented towers: the 100-meter Beryl tower in the river valley about 400
meters southwest of the facility; the 30-meter Luke Hill tower on a ridge 900 meters north-
northwest of the facility; and the 100-meter Met tower 900 meters east-southeast of the facility
on a ridge across the river.
The Tracy Power Plant19 is located 27 kilometers east of Reno, Nevada in the Truckee River
valley with mountainous terrain on all sides. A field tracer study was conducted at the power
plant in August 1984 with SF6 being released through the 91-m stack servicing unit 3. A total of
128 hours of data were collected over 14 experimental periods. Most of the hours were during
stable atmospheric conditions. On-site meteorological data for Tracy were collected from an
instrumented 150-m tower located 1.2 kilometers east of the power plant for the 128-hour period.
The wind measurements from the tower were extended above 150 meters using a Doppler
acoustic sounder and temperature measurements were extended with tethersonde data.
Evaluation Procedures
The model evaluation was designed to provide diagnostic as well as descriptive information
about the model performance. Highlights of the evaluation results for the current model
presented by Paine, et al.8 used selected residual plots and quantile-quantile (Q-Q) plots. The
residual plots feature box and whisker symbols that show the distribution of cases along the y-
axis domain for various "bins" or domain segments along the x-axis. Q-Q plots are simple
ranked pairings of predicted and observed concentrations, such that any given quantile of the
predicted concentration is plotted against the same quantile of the observed concentration. The
Q-Q plot is an effective method for comparing the frequency distributions of two data sets.
Cox and Tikvart20 proposed a robust test statistic that represents a smoothed estimate of the
highest concentrations, based on a tail exponential fit to the upper end of the distribution. With
-------
this procedure, the effect of extreme values on model comparison is reduced. This statistic is the
robust highest concentration (RHC) and evaluation results using the RHC are reported elsewhere
in this paper.
Comparisons between AERMOD and ISCST321 are included in the evaluation results.
Comparisons were also made with the CTDMPLUS model22 and RTDM23 for complex terrain
and with the HPDM model3-15 for selected data sets8.
For the tracer databases, results for 1 -hour averages are reported (with the Prairie Grass 10-
minute averages taken as 1-hour averages). For the 1-year SO2 data sets, 3-hour, 24-hour, and
annual results are reported. All of the observed concentrations for the long-term databases are
subject to uncertainty because a background concentration is subtracted from the actual
monitored observations to obtain a "source-caused" impact. In addition, it should be realized
that SO2 monitors typically have a 6 ppb (16 ng/m3) detection limit, and baseline (zero) drifts of
up to 10 ppb (26 ng/m3) are not corrected24. Concentrations below the detection limit are
typically set to half of the limit (8 Hg/m3), even though they may actually be zero. Another factor
that could result in overestimates of "observed" concentrations is the acceptance without
correction of nonzero concentrations caused by baseline drift that should actually be reported as
zero. Therefore, the combined potential errors in SO2 measurements from the detection limit
treatment, ignored baseline drifts, and background concentration estimates can result in
significant uncertainties in "observed" annual averages. Peak short-term averages are not
affected significantly because the uncertainty is typically a small percentage of the reported
value. However, the reader should interpret evaluation results for annual averages with
considerable caution.
EVALUATION RESULTS
Due to space limitations, a limited number of figures showing Q-Q and residual plots for the
various databases provided in this report, while more extensive results can be found in Paine et
al8. In general, 1-hour average statistics are discussed below for the tracer databases, and 3-hour,
24-hour, and annual averages for the long-term databases.
Developmental Evaluation
Prairie Grass
The Q-Q plot for the Prairie Grass data set for AERMOD and ISCST3 (see Figure 1) indicates
that both models predict well within a factor of 2. The 1-hour RHC results (see Table 2),
consistent with the Q-Q plot, indicate a slight underprediction by AERMOD (0.87 ratio of
predicted to observed RHCs), and an overprediction by ISCST3 (1.50).
Kincaid SF$
Q-Q plots for all cases (see Figure 2) show that AERMOD's performance is clearly superior,
with substantial underpredictions noted for ISCST3. A separate analysis of convective
conditions for AERMOD showed very good performance. The peak unstable concentrations are
significantly higher than the peak stable concentrations. AERMOD's inability to match the
comparatively lower observed stable concentrations may be partially due to a limited sample size
in this database, and this behavior is not evident in the Kincaid SO2 results discussed below.
-------
The 1-hour RHC results (see Table 2) indicate a slight underprediction by AERMOD (0.76 ratio
of predicted to observed RHCs), and an underprediction by ISCST3 (0.68).
Indianapolis
The Indianapolis data set provides a database on which to test the behavior of the models in an
urban setting. The Q-Q plots that include the entire database (see Figure 3) show a nearly
unbiased trend for AERMOD over the entire range of concentrations, while ISCST3 exhibits an
overprediction tendency over the whole range. In convective conditions, AERMOD shows a
very slight underprediction tendency, with only a small trend with distance. The Q-Q plot for
stable conditions indicates a nearly unbiased performance for AERMOD for a large portion of
the concentration domain. Residual plots for AERMOD indicate a notable trend with distance,
with underpredictions especially evident in the near field (within 1 km). However, these
distances are generally associated with low observed concentrations (near the observation
threshold), so an underprediction ratio involving two small values is not of significant concern.
The 1-hour RHC results (see Table 2) indicate a slight overprediction by AERMOD (1.20 ratio
of predicted to observed RHCs), and a higher overprediction by ISCST3 (1.30).
KincaidSO2
The Kincaid SO2 database provides data from the same stack source as the Kincaid SF6. There
are, however, three main differences in that study: 1) The database contains several months of
continuous observations, 2) the sampler network is less dense, and 3) the pollutant being
measured is the SO2 that is emitted due to the sulfur contained in the fuel instead of the SF6
tracer. Because the samplers are not arranged in arcs, residual plots by distance are not
meaningful, and therefore have not been included. However, the database does allow for
computation of 1-hour, 3-hour, 24-hour, and "annual" average concentration statistics. For this
data set, the single highest concentration for each evaluation period was used. In each case for
the 1-hour Q-Q plots (see Figure 4), AERMOD's curve parallels the 1-1 line more closely as
ISCST3 is shown to consistently underpredict. Analyses of the convective AERMOD
predictions show good results that are consistent with those of the Kincaid SF6 results. The Q-Q
plot of the stable hours indicates reasonably good AERMOD performance, in contrast with the
poorer showing of AERMOD in the sample size-limited Kincaid SF6 database.
The 3-hour and 24-hour RHC results and the annual peak results (see Table 2) indicate a nearly
unbiased predicted to observed ratio for AERMOD for the 3-hour and 24-hour averages (1.01
and 0.97, respectively) as opposed to underpredictions by ISCST3 (ratios of 0.45 and 0.45 for the
3-hour and 24-hour averages). Both models underpredict for the annual RHC statistic (0.30 for
AERMOD and 0.14 for ISCST3). However, the low annual concentrations (near the instrument
threshold) and the uncertainties in subtracting background concentrations make the "observed"
average concentrations subject to considerable uncertainty.
Lovett
The Lovett data set provides a test on the AERMOD treatment of complex terrain. In terms of the
complexity of its theoretical formulation, AERMOD lies between the current screening models
and the CTDMPLUS refined model (Perry, 1992). Q-Q plots of AERMOD 1-hour average
results (see Figure 5) show a curve very close to the 1-1 line for each averaging time. ISCST3,
-------
on the other hand, substantially overpredicts these concentrations for all three averaging times.
(ISCST3 uses the COMPLEX-I screening model and the EPA Intermediate Terrain Procedures in
these calculations, which is inherently "conservative," that is, it tends to overpredict.) The
CTDMPLUS results show a consistent overprediction tendency, by about a factor of 2.
In convective conditions, the AERMOD Q-Q plot curve parallels the 1-1 line with very little bias
for most of the concentration domain. In stable conditions, the AERMOD curve overstates
concentrations except for the top few, which indicate a modest underprediction tendency.
The 3-hour and 24-hour RHC results and the annual peak results (see Table 2) indicate an
unbiased predicted to observed ratio for AERMOD for the 3-hour and 24-hour averages (1.00 for
both averaging times) as opposed to overpredictions by ISCST3 (ratios of 8.20 and 9.11 for the
3-hour and 24-hour averages) and overpredictions by CTDMPLUS (ratios of 2.36 and 2.02 for
the 3-hour and 24-hour averages). AERMOD shows a slight underprediction for the annual
average (ratio of 0.79), while ISCST3 continues to show a large overprediction (ratio of 7.51), as
well as CTDMPLUS (ratio of 1.71).
Independent Evaluation
Baldwin
The Baldwin site is a test of the model performance for tall stacks in flat terrain. Q-Q plots of
AERMOD 1-hour average results (see Figure 6) show nearly unbiased results for the 3-hour and
24-hour averages. In each case, ISCST3 shows nearly unbiased concentrations for the top end of
the concentration domain, but AERMOD's performance is much better for a much larger range
of the concentration domain in each case. ISCST3 underpredicts at the lower concentration
values in each case.
The 3-hour and 24-hour RHC results and the annual peak results (see Table 2) indicate a modest
overprediction tendency for AERMOD for the 3-hour average (ratio of 1.31) and a nearly
unbiased 24-hour and annual average set of ratios (1.02 for the 24-hour average and 0.97 for the
annual average). ISCST3 results indicate higher overpredictions for the 3-hour and 24-hour
averages (ratios of 1.48 and 1.13, respectively), and underpredictions for the annual average (a
predicted to observed ratio of 0.63).
Cliffy Creek
This case features a tall stack with terrain extending at least halfway to stack top. Q-Q plots of
AERMOD 1-hour average results (see Figure 7) show nearly unbiased results for the 3-hour
average, and a modest underprediction for 24-hour averages. For the 3-hour and 24-hour
averages, ISCST3 shows nearly unbiased concentrations for the top end of the concentration
domain, but AERMOD's performance is once again better for a much larger range of the
concentration domain. For the annual average, the results for AERMOD and ISCST3 are similar
at the top end of the concentration scale. ISCST3 underpredicts at the lower concentration values
in each case.
The 3-hour and 24-hour RHC results and the annual peak results (see Table 2) indicate a modest
overprediction tendency for AERMOD for the 3-hour average (ratio of 1.25) and a modest
-------
underprediction for the 24-hour average (0.72). ISCST3 results for the same averaging times are
0.98 and 0.67 for the 3-hour and 24-hour averages, respectively. Both models show
underpredictions for the annual peaks (ratios of 0.54 for AERMOD and 0.31 for ISCST3).
Martins Creek
This site represents a test of the complex terrain algorithms of AERMOD, ISCST3, and
CTDMPLUS. Q-Q plots of AERMOD 1-hour average results (see Figure 8) show a similar trend
in each case, featuring overpredictions over most of the concentration domain, but showing that
the curve approaches the 1-1 line at the top, or has two peak points below the line. On the other
hand, predictions of ISCST3 and CTDMPLUS show significant overpredictions (with turbulence
data for CTDMPLUS coming from AERMOD internally-generated profiles).
The 3-hour and 24-hour RHC results and the annual peak results (see Table 2 indicate a nearly
unbiased result for AERMOD for the 3-hour average (ratio of 1.06) and an overprediction for the
24-hour average (1.74). AERMOD shows a modest underprediction ratio for the annual average
(0.74). In contrast, the 3-hour and 24-hour ratios for ISCST3 are 7.25 and 8.88, showing
significant overprediction. The CTDMPLUS resulting ratios are 4.80 and 5.56 for the same
averaging times. For annual averages, ISCST3 and CTDMPLUS are still overpredicting, with
predicted to observed ratios of 3.37 and 2.19, respectively.
Westvaco
Westvaco is another complex terrain database. It was one of the independent evaluation data sets
for CTDMPLUS. Q-Q plots of AERMOD 1-hour average results (see Figure 9) show a nearly
unbiased trend for the upper part of the concentration domain for each averaging time. For the
short-term averages, CTDMPLUS shows a factor-of-2 overprediction trend, with a less
overprediction for the annual average.
The 3-hour and 24-hour RHC results and the annual peak results (see Table 2) indicate a nearly
unbiased result for AERMOD for the 3-hour and 24-hour averages (ratios of 1.08 and 1.14,
respectively), and an overprediction for the annual average (1.64). For the short-term averages,
CTDMPLUS shows overpredictions (ratios of 2.14 and 1.54 for the 3-hour and 24-hour
averages). The CTDMPLUS annual average ratio is 0.93.
Tracy
The Tracy Power Plant database was a developmental evaluation data set for CTDMPLUS. In
the Q-Q plot (see Figure 10), both curves parallel the 1-1 line for the entire concentration
domain, but AERMOD shows nearly unbiased results at the top end of the concentration range,
while CTDMPLUS exhibits a modest underprediction tendency. This trend is consistent with the
results of the 1-hour RHC analysis, for which the AERMOD ratio of predicted to observed
concentrations is 1.02, as opposed to a ratio of 0.77 for CTDMPLUS.
SUMMARY AND CONCLUSIONS
A summary of the results over all of the evaluation databases for the RHC statistic is presented in
Table 2. These results show that AERMOD is nearly unbiased, on average, across all averaging
times. Results for the developmental and independent evaluation subsets are qualitatively
-------
similar. For 1-hour averages (the tracer databases), the ratio of predicted to observed values for
the RHC ranges from 0.76 to 1.20, with a geometric mean of 0.96. For 3-hour averages, the
RHC predicted to observed ratio ranges from 1.00 to 1.31, with a geometric mean of 1.11. The
same ratio for 24-hour averages ranges from 0.72 to 1.72, with a geometric mean of 1.06.
Annual average statistics are less reliable because background concentrations, which are removed
from the measured values in many cases, are uncertain and approach the value of the source-
caused impact. The AERMOD RHC predicted to observed ratio for annual averages ranges from
0.30 to 1.64, with a geometric mean of 0.73. For all averaging times in general and in most
cases, AERMOD's model performance was better than that of ISCST3.
The model evaluation results consistently show AERMOD behavior on the Q-Q plots that
parallel the 1-1 line over a larger range of the concentration domain than other models tested.
The AERMOD model bias exhibited on the Q-Q plots and in the RHC statistics shows an overall
slight overprediction tendency. Apparent underpredictions for annual averages may be due in
part to artifacts of the low concentrations (close to the instrument thresholds) and the uncertainty
in determining background concentrations that need to be subtracted from the reported total
concentrations.
DISCLAIMER
This document has been reviewed in accordance with the US EPA'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
1. Cimorelli, A. J., S. G. Perry, A. Venkatram, J.C. Weil, R. J. Paine, R. B. Wilson, R. F. Lee,
and W.D. Peters, 1998: AERMOD: Description of Model, EPA web site
(http://www.epa.gov/scram001).
2. Weil, J.C. and R.P. Brower, 1984. An Updated Gaussian Plume Model for Tall Stacks. J. Air
Pollut. Control Assoc., 34, 818-827.
3. Hanna, S.R. and R.J. Paine, 1989: Hybrid Plume Dispersion Model (HPDM) Development
and Evaluation. J. Appl. Meteor., 28, 206-224.
4. Turner, D.B., T. Chico and J. Catalano, 1986. TUPOS—A Multiple Source Gaussian
Dispersion Algorithm Using On-site Turbulence Data. EPA/600/8-86/010. US EPA, Research
Triangle Park, NC, 39pp.
5. Perry, S.G., D.J. Burns, L.H. Adams, R.J. Paine, M.G. Dennis, M.T. Mills, D.J. Strimaitis,
R.J. Yamartino, and E.M. Insley, 1989. User's Guide to the Complex Terrain Dispersion Model
Plus Algorithms for Unstable Situations (CTDMPLUS) Volume I: Model Description and User
Instructions. EPA-600/8-89-041. U.S. Environmental Protection Agency, Research Triangle
Park, NC (NTIS PB89-181424).
6. Carruthers, D.J., R.J. Holroyd, J.C.R. Hunt, W.S. Weng, A.G. Robins, D.D. Apsley, D.J.
Thomson, and F.B. Smith, 1992. UK-ADMS - a New Approach to Modelling Dispersion in the
Earth's Atmospheric Boundary Layer. In Proceedings of the Workshop: Objectives for Next
10
-------
Generation of Practical Short-Range Atmospheric Dispersion Models, May 6-8, 192, Riso,
Denmark, pp. 143-146.
7. Olesen, H.R., P. Lofstrom, R. Berkowicz, and A.B. Jensen, 1991. An Improved Dispersion
Model for Regulatory Use - the OML Model. In Proceedings of the 19th International Technical
Meeting on Air Pollution Modeling and its Applications, lerapetra, Crete, Greece. 29 September
-4 October 1991.
8. Paine, R.J, R.F. Lee, R. Erode, R.B. Wilson, A.J. Cimorelli, S.G. Perry, J.C. Weil, A.
Venkatram, and W. Peters, 1998. Model Evaluation Results for AERMOD.
http://www.epa.gov/scram001/
9. Barad, M. L., (Ed.), 1958: Project Prairie Grass, A Field Program in Diffusion. Geophysical
Research Papers, No. 59, Vols. I and II, Report AFCRC-TR-58-235, Air Force Cambridge
Research Center, 439 pp.
10. Haugen, D. A. (Editor), 1959: Project Prairie Grass, A field program in diffusion.
Geophysical Research Paper, No. 59, Vol. III. Report AFCRC-TR-58-235, Air Force Cambridge
Research Center, 439 pp.
11. Liu, M. K., and G. E. Moore, 1984: Diagnostic validation of plume models at a plains site.
EPRI Report No. EA-3077, Research Project 1616-9, Electric Power Research Institute, Palo
Alto, CA.
12. Bowne, N. E., R. J. Londergan, D. R. Murray, and H. S. Borenstein, 1983: Overview, Results,
and Conclusions for the EPRI Plume Model Validation and Development Project: Plains Site.
EPRI Report EA-3074, Project 1616-1, Electric Power Research Institute, Palo Alto, CA, 234 pp.
13. Murray, D. R., and N. E. Bowne, 1988: Urban power plant plume studies. EPRI Report No.
EA-5468, Research Project 2736-1, Electric Power Research Institute, Palo Alto, CA.
14. Paumier, J. O., S. G. Perry, and D. J. Burns, 1992: CTDMPLUS: A dispersion model for
sources near complex topography. Part II: Performance characteristics. J. Appl. Meteor., 31, 646-
660.
15. Hanna, S.R. and J.C. Chang, 1993. Hybrid Plume Dispersion Model (HPDM) Improvements
and Testing at Three Field Sites. Atmos. Envir., 27A, 1491-1508.
16. TRC, 1982. Evaluation of Rural Air Quality Simulation Models. EPA-450/4-83-003 (NTIS
# PB83-182758), prepared for Environmental Protection Agency, Research Triangle Park, NC.
17. TRC, 1994: Air Quality Model Performance Evaluation and Comparison Study for Martins
Creek Steam Electric Station. TRC Project No. 14715-R61. TRC Environmental Corporation,
Windsor, CT.
18. Strimaitis, D. G., R. J. Paine, B. A. Egan and R. J. Yamartino, 1987. EPA Complex Terrain
Model Development: Final Report. Contract No. 68-02-3421, U. S. Environmental Protection
Agency, Research Triangle Park, North Carolina.
19. DiCristofaro, D. C., D. G. Strimaitis, B. R. Green, R. J. Yamartino, A. Venkatram, D. A.
Gooden, T. F. Lavery and B. A. Egan, 1985. EPA Complex Terrain Model Development: Fifth
11
-------
Milestone Report - 1985. EPA-600/3-85-069, U. S. Environmental Protection Agency, Research
Triangle Park, North Carolina.
20. Cox, W. and Tikvart, J., 1990: A Statistical Procedure for Determining the Best Performing
Air Quality Simulation Model. Atmos. Envir., 24A, 2387-2395.
21. U. S. Environmental Protection Agency, 1995: User's guide for the Industrial Source
Complex (ISC3) dispersion models. Volume II: Description of model algorithms. EPA-454/B-
95-003b, 120 pp. [NTIS PB95-222758.]
22. Perry, S. G., 1992: CTDMPLUS: A dispersion model for sources near complex topography.
Part I: Technical formulations. J. Appl. Meteor., 31, 633-645.
23. Paine, R.J. and B.A. Egan, 1987. User's Guide to the Rough Terrain Diffusion Model
(RTDM), (Rev. 3.20). NTIS # PB88-171467/REB. U.S. Environmental Protection Agency,
Research Triangle Park, NC.
24. Gendron, L., 1998. Personal Communication (ENSR Corporation).
12
-------
Table 1. Comparison of Dispersion Model Features
AERMOD vs. ISCST3
Feature
Types of sources modeled
Plume Rise
Meteorological Data Input
Profiling Meteorological Data
Use of Meteorological Data in
Plume Dispersion
Plume Dispersion: General
Treatment
Urban Treatment
Characterization of Modeling
Domain Surface Characteristics
ISCST3
Point, area, and volume sources
Uses Briggs equations with stack-top
wind speed and vertical temperature
gradient
One level of data accepted
Only wind speed is profiled
Stack-top variables for all downwind
distances
Gaussian treatment in horizontal and
vertical
Urban option either on or off; no
other specification available; all
sources must be modeled either rural
or urban
Choice of rural or urban
AERMOD
Same as ISCST3
In stable conditions, uses Briggs equations
with winds and temperature gradient at
stack top and half-way to final plume rise;
in convective conditions, plume rise is
superposed on the displacements by random
convective velocities
An arbitrarily large number of data levels
can be accommodated
AERMOD creates profiles of wind,
temperature, and turbulence, using all
available measurement levels
Variables measured throughout the plume
depth (averaged from plume centerline to
2.15 sigma-z below centerline; changes
with downwind distance)
Gaussian treatment in horizontal and in
vertical for stable conditions; non-Gaussian
probability density function in vertical for
unstable conditions
Population is specified, so treatment can
consider a variety of urban conditions;
sources can individually be modeled rural
or urban
Selection by direction and month of
roughness length, albedo, and Bowen ratio,
providing much user flexibility
Comments
Models are comparable
AERMOD is better because in stable
conditions it factors in wind and temperature
changes above stack top, and in unstable
conditions it accounts for convective updrafts
and downdrafts
AERMOD can adapt multiple levels of data to
various stack and plume heights
AERMOD is much improved over ISCST3 in
this area
AERMOD treatment is far more advanced than
that of ISCST3; accounts for meteorological
data throughout the plume depth
AERMOD's unstable treatment of vertical
dispersion is a more accurate portrayal of
actual conditions
AERMOD provides variable urban treatment as
a function of city population, and can
selectively model sources as rural or urban
AERMOD provides the user with considerably
more options in the selection of the surface
characteristics
13
-------
Feature
ISCST3
AERMOD
Comments
Boundary Layer Parameters
Wind speed, mixing height, and
stability class
Friction velocity, Monin-Obukhov length,
convective velocity scale, mechanical and
convective mixing height, sensible heat flux
AERMOD provides parameters required for
use with up-to-date planetary boundary layer
(PEL) parameterizations; ISCST3 does not
Mixed Layer Height
Holzworth scheme; uses
interpolation based upon maximum
afternoon mixing height
Has convective and mechanical mixed layer
height; convective height based upon hourly
accumulation of sensible heat flux
AERMOD's formulation is significantly more
advanced than that of ISCST3, includes a
mechanical component, and in using hourly
input data, provides a more realistic sequence
of the diurnal mixing height changes
Terrain Depiction
Elevation at each receptor point
Controlling hill elevation and point
elevation at each receptor, obtained from
special terrain pre-processor (AERMAP)
that uses digital elevation model (DEM)
data
AERMOD's terrain pre-processor provides
information for advanced critical dividing
streamline height algorithms and uses digital
data to obtain receptor elevations
Plume Dispersion: Plume Growth
Rates
Based upon 6 discrete stability
classes only; dispersion curves
(Pasquill-Gifford) are based upon
surface release experiments (e.g.,
Prairie Grass)
Uses profiles of vertical and horizontal
turbulence (from measurements and/or PEL
theory); variable with height; uses
continuous growth functions rather than a
discrete (stability-based) formulation
Use of turbulence-based plume growth with
height dependence rather than that based upon
stability class provides AERMOD with a
substantial advancement over the ISCST3
treatment
Plume Interaction with Mixing Lid:
convective conditions
If plume centerline is above lid, a
zero ground-level concentration is
assumed
Three plume components are considered: a
"direct" plume that is advected to the
ground in a downdraft, an "indirect" plume
caught in an updraft that reaches the lid and
eventually is brought to the ground, and a
plume that penetrates the mixing lid and
disperses more slowly in the stable layer
aloft (and which can re-enter the mixed
layer and disperse to the ground)
The AERMOD treatment avoids potential
underpredictions suffered by ISCST3 due to its
"all or nothing" treatment of the plume;
AERMOD's use of convective updrafts and
downdrafts in a probability density function
approach is a significant advancement over
ISCST3
Plume Interaction with Mixing Lid:
stable conditions
The mixing lid is ignored (assumed
to be infinitely high)
A mechanically mixed layer near the
ground is considered. Plume reflection
from an elevated lid is considered.
AERMOD's use of a mechanically mixed layer
is an advancement over the very simplistic
ISCST3 approach
14
-------
Table 2. Summary of AERMOD Evaluation Results
Database
Prairie Grass (SC>2)
Flat, grassy field (Nebraska,
USA)
Kincaid (SFg)
Flat, rural (Illinois, USA)
Kincaid: (SC>2)
Flat, rural (Illinois, USA)
Baldwin (SC>2):
Flat, rural (Illinois, USA)
Indianapolis (SFg)
Flat, urban (Indiana, USA)
Clifty Creek (SO2)
Moderately hilly terrain, rural
(Indiana, USA)
Tracy (SF6):
Hilly terrain, rural (Nevada,
USA)
Ratio of Modeled/Observed Robust Highest Concentrations*
AERMOD:
ISCST3:
AERMOD:
ISCST3:
AERMOD:
ISCST3:
AERMOD:
ISCST3:
AERMOD:
ISCST3:
AERMOD:
ISCST3:
HPDM:
AERMOD:
ISCST3:
HPDM:
AERMOD:
ISCST3:
HPDM:
AERMOD:
ISCST3:
AERMOD:
ISCST3:
HPDM:
AERMOD:
ISCST3:
HPDM:
AERMOD:
ISCST3:
HPDM:
AERMOD:
CTDMPLUS:
0.87 (1-hravg)
1.50 (1-hravg)
0.76 (1-hravg)
0.68 (1-hravg)
1.01 (3-hravg)
0.56 (3-hr avg)
0.97 (24-hr avg)
0.45 (24-hr avg)
0.30 (annual peak)
0.14 (annual peak)
1.31 (3-hravg)
1.48 (3-hr avg)
1 .06 (3-hr avg)
1.02 (24-hr avg)
1.1 3 (24-hr avg)
1.02 (24-hr avg)
0.97 (annual peak)
0.63 (annual peak)
1.15 (annual peak)
1.20 (1-hravg)
1.30 (1-hravg)
1.25 (3-hr avg)
0.98 (3-hr avg)
1.33 (3-hr avg)
0.72 (24-hr avg)
0.67 (24-hr avg)
1 .46 (24-hr avg)
0.54 (annual peak)
0.31 (annual peak)
0.96 (annual peak)
1.07 (1-hravg)
0.77 (1-hravg)
15
-------
Database
Ratio of Modeled/Observed Robust Highest Concentrations*
Martins Creek (SC>2): Hilly
terrain, rural
(Pennsylvania/New Jersey,
USA)
AERMOD:
CTDMPLUS:
ISCST3:
RTDM:
1.06 (3-hr avg)
4.80 (3-hr avg)
7.25 (3-hr avg)
3.33 (3-hr avg)
AERMOD: 1.72 (24-hr avg)
CTDMPLUS: 5.56 (24-hr avg)
ISCST3: 8.88 (24-hr avg)
RTDM: 3.56 (24-hr avg)
AERMOD:
CTDMPLUS:
ISCST3:
RTDM:
0.74 (annual peak)
2.19 (annual peak)
3.37 (annual peak)
1.32 (annual peak)
Lovett (SO2)
AERMOD:
CTDMPLUS:
ISCST3:
1.00 (3-hr avg)
2.36 (3-hr avg)
8.20 (3-hr avg)
AERMOD: 1.00 (24-hr avg)
CTDMPLUS: 2.02 (24-hr avg)
ISCST3: 9.11 (24-hr avg)
AERMOD:
CTDMPLUS:
ISCST3:
0.78 (annual peak)
1.71 (annual peak)
7.49 (annual peak)
Westvaco (SO2):
Hilly terrain, rural
(Maryland/Virginia, USA)
AERMOD:
CTDMPLUS:
ISCST3:
1.08 (3-hr avg)
2.14(3-hravg_
8.50 (3-hr avg, estimated)
AERMOD: 1.14 (24-hr avg)
CTDMPLUS: 1.54 (24-hr avg)
ISCST3: N/A (24-hr avg)
AERMOD:
CTDMPLUS:
1.64 (annual peak)
0.93 (annual peak)
Notes:
The Robust Highest Concentration (RHC) is a statistical estimator for the highest concentration. It is
determined from a tail exponential fit to the high end of the frequency distribution of observed and
predicted values. The number of points used for the fit is arbitrary, but usually ranges between 10 and
25.
The estimated ISCST3 result for Westvaco is derived from the EPA Complex Terrain Model
Development study (Strimatis et al., 1987) in which several models, including CTDMPLUS and
COMPLEX-I (now part of ISCST3), were evaluated.
16
-------
Figure J&t 1-hour Quantile-Quantile plot for all hours evaluated in
the Prairie Grass database, AERMOD prodiotiorio
PRAIRIE GRASS 1-HR Q-Q PLOT (CONC)
Tl
m
03
*
H»
ID
ID
ID
o
UJ
o
Q
ut
cc
a.
10000000
1000000
100000
10000
1000
10
100
1000
10000
OBSERVED
100000
tOOOODO
10000000
m
n
n
m
m
en
\A
ID
H-^
CO
CSD
-------
Figure 4H$r 1-hour Quantile-Quantile plot for all hours evaluated in
the Kincaid SF6 database, AERMOD predictions
KINCAID SF6 1-HR Q-Q PLOT (CONG)
rn
m
r\j
cr>
>-»
ID
ID
UD
o
III
U)
te.
o.
Ul
0.01
0.01
0.1
OBSERVED
10
rn
t/i
n
o
n
rn
rn
;o
en
cs
CO
en
u
en
ID
TD
CS
ro
-------
Figure
. 1-hour Quantile-Quantile plot for all hours evaluated in
the Indianapolis SFe database, AERMODpredictions
INDIANAPOLIS SF6 1-HR Q-Q PLOT (CONG)
Tl
m
m
*
t-L
ID
ID
ID
Q
tit
O
en
0.1
OBSERVED
10
m
n
o
m
en
C3
CO
O)
l-»
CO
cs
T3
CD
\
ru
-------
o
HI
Ul
o
o
10,000
1,000
100
Figure
. 1-hour Quantile-Quantile plot for all hours in the
Kincaid SO2 database, AERMOD and ISCST3
predictions
KINCAID SO2 (Version 98314)
1-HOUR Q-Q PLOTS (CONC.)
10
10
100
1,000
OBSERVED
10,000
m
D3
r\j
en
i-*
UD
UD
UD
cn
m
m
in
70
o
o
n
§
m
c/i
CD
m
m
CD
T3
CD
f\J
-------
Q
Ul
-I
OJ
o
o
10,000
1,000
100
10
10
Figure Jtffo 3-hour Quantile-Quantile plot for all hours in the
Kincaid SO2 database, AERMOD and ISCST3
predictions
KINCAID SO2 (Version 98314)
3-HOUR Q-Q PLOTS (CONC.)
100
1,000
10,000
OBSERVED
m
ru
en
i
i-»
ID
ID
ID
CO
m
I
8
TJ
fi
m
CO
en
u
en
ID
»-^
CO
cs
cs
en
rvj
ui
-------
1,000
Q
LLJ
LU
Q
O
100
10
1
1
Figure XJfefc 24-hour Quanlile-Quantile plot for all hours in the
Kincaid SO2 database, AERMOD and ISCST3
predictions
KINCAID SO2 (Version 98314)
24-HOUR Q-Q PLOTS (CONC.)
10
100
OBSERVED
1,000
m
m
UD
UD
UD
cn
CO
m
in
n
o
Tl
m
en
CSD
CO
a
en
UD
I-*
CO
CSD
Tl
CD
\
-------
10,000
Q
HI
LLJ
Q
O
1,000
100
10
1
Figure ^£24: 1 -hour Quantile-Quantile plot for all hours in the Lovett
SO2 data base - AERMOD, CTDMPLUS and ISCST3
predictions
LOVETT SO2 (Version 98314)
1-HOUR Q-Q PLOTS (CONC.)
1
10
100
1,000
OBSERVED
Tl
m
CO
r\j
en
(-*
UD
a
en
m
m
I
n
§
n
en
cs
m
en
u
en
UD
h-*
CO
CD
ro
01
-------
10,000
Si
Figure rf£$: 3-hour Quantile-Quantile plot for all hours in the Lovett
SO2 database - AERMOD, CTDMPLUS and ISCST3
predictions
LOVETT SO2 (Version 98314)
3-HOUR Q-Q PLOTS (CONC.)
o
LU
-I
LU
Q
O
1,000
100
10
1
1
* AERMOD
D CTDMPLUS
XISCST3
10
100
1,000
OBSERVED
m
03
ru
en
H»
U3
03
UD
Ul
03
m
5
n
o
n
m
m
en
fl
UD
i-»
CO
G>
UD
ru
u
-------
Q
UJ
j
LU
0
o
1,000
100
10
1
1
Figure 4£fe 24-hour Quanlile-Quantile plot for all hours in the
Lovett SO2 database - AERMOD, CTDMPLUS and
ISCST3 predicrions
LOVETT SO2 (Version 98314)
24-HOUR Q-Q PLOTS (CONC.)
10
OBSERVED
100
m
ru
ID
U3
U3
cn
ID
m
I
n
n
§
si
cn
Q
CO
cn
u
cn
ID
h-*
CO
ru
-------
10000
1000
TJ
,0)
15
TJ
O
10
Figure J&&. 1-hour Quantile-Quantile plot for all hours in the
Baldwin SO2 database - AERMOD, HPDM, and ISCST3
predictions
Baldwin SO2
1-hr Q-Q plot (Cone.), Version 98314
100
1000
10000
m
m
r\j
en
i-'
VD
en
m
(ft
n
n
m
m
CO
en
u
ui
Observed
ru
OJ
-------
10000
1000
•o
o
"o
•o
0
100
10
Figure 4C50: 3-hour Quantilc-Quantile plot for all hours in the
Baldwin SO2 database - AERMOD, HPDM, and ISCST3
predictions
Baldwin SO2
3-hr Q-Q plot (Cone.), Version 98314
100
1000
10000
Observed
m
DO
ru
en
i-k
UD
in
n
o
n
m
m
73
en
o
CO
en
u
en
CD
CS
ro
IV)
OJ
-------
1000
•O
J£
"5
•o
o
Figure
: 24-hour Quantile-Quantile plot for all hours in the
Baldwin SO2 database - AERMOD, HPDM, and ISCST3
predictions
Baldwin SO2
24-hr Q-Q plot (Cone.), Version 98314
10
100
1000
m
tn
r\j
en
UD
U)
CS
CS
m
n
o
TI
Fl
X)
Observed
en
cs
CO
en
w
en
UD
ru
-------
10000
1000
TJ
0
"3
•o
o
10
10
7*-
Figure JMfl: 1-hour Quantile-Quantile plot for all hours in the Clifty
Creek SO2 database - AERMOD, HPDM, and ISCST3
predictions
Clifty Creek SO2
1-hr Q-Q plot (Cone.), Version 98314
100
1000
Observed
10000
to
G)
G)
m
I
s
~0
R
m
70
en
u
en
CO
KJ
UJ
-------
10000
1000
•o
o
"5
•u
o
104.
10
Figure jt&: 3-hour Quantile-Quantile plot for all hours in the Cliffy
Creek SO2 database - AERMOD, HPDM, and ISCST3
predictions
Cliffy Creek SO2
3-hr Q-Q plot (Cone.), Version 98314
100
1000
10000
m
CD
ru
en
a
ru
70
n
o
m
70
Observed
ui
Q
CO
en
u
ui
CO
LH
ru
-------
1000
•D
o
"5
•o
o
E
Figure/^34: 24-hour Quantile-Quantile plot for all hours in the Clifty
Creek SO2 database - AERMOD, HPDM, and ISCST3
predictions
Clifty Creek SO2
24-hr Q-Q plot (Cone.), Version 98314
10
100
1000
Observed
3
m
UD
UD
ID
r\j
C3
m
in
8
13
O
m
m
70
en
C3
m
en
u
en
UD
H*
m
C3
en
ru
u
-------
10000
1000
TJ
o
o
•o
o
100--
10
Figure Jt&R: 1-hour Quantile-Quantilc plot for all hours in the
Martins Creek SO2 database - AERMOD, CTDMPLUS,
RTDM, and ISCST3 predictions
Martin's Creek SO2
1-hr Q-Q plot (Cone.), Version 98314
100
1000
Observed
10000
m
M
UD
UD
UD
ru
CD
m
I
n
n
m
m
70
en
CD
CO
tn
w
en
UD
i-k
CO
CD
ru
-------
10000
1000
o
"3
•o
o
Figure 4&3fc 3-hour Quantile-Quantile plot for all hours in the
Martins Creek SO2 database - AERMOD, CTDMPLUS,
RTDM, and ISCST3 predictions
Martin's Creek SO2
3-hr Q-Q plot (Cone.), Version 98314
10
10
100
1000
Observed
10000
m
M
I\J
m
UD
UD
UD
I\J
G3
m
01
n
a
m
u
ui
UD
i-»
CO
CO
IXJ
u
-------
1000
•o
a
*o
TO
O
Figure .£&& 24-hour Quanlile-Quantile plot for all hours in the
Martins Creek SO2 database - AERMOD, CTDMPLUS,
RTDM, and ISCST3 predictions
Martin's Creek SO2
24-hr Q-Q plot (Cone.), Version 98314
10
100
1000
Tl
m
m
IV)
en
(-»
UD
UD
UD
IV)
IV)
m
n
n
m
Observed
en
G)
CO
en
OJ
en
(D
I-*
CD
UD
IV)
OJ
-------
100000
•o
o
10000
1000
Figure 4&J£ 1-hour Quantile-Quantile plot for all hours in the
Westvaco S02 database - AERMOD and CTDMPLUS
predictions
WESTVACO SO2
1-hr Q-Q Plots (Cone.), Version 98314
10
100 1000
Observed
10000
100000
m
03
i
w
en
•->
ID
m
n
o
70
CO
en
u
01
CO
CD
W
CD
W
U
-------
10000
1000
o
T3
O
0.1
A
Figure fcfo: 3 -hour Quantile-Quantile plot for all hours in the
Westvaco S02 database - AERMOD and CTDMPLUS
predictions
WESTVACO S02
3-hr Q-Q Plots (cone.), Version 98314
10 100
Observed
1000
10000
3
m
ru
en
ru
n
n
en
o
CO
en
w
01
-------
1000
100
•o
o
10
Figure |{49: 1-hour Quantile-Quantile plot far all hours in the
Westvaco SO2 database - AERMOD and CTDMPLUS
predictions
WESTVACO SO2
24-hr Q-Q Plots (Cone.), Version 98314
a AERMOD
xCTDMPLUS
10
100
1000
Observed
m
DD
FV)
cn
h-*
VD
VD
VD
ru
a
u
m
01
n
o
~
R
cn
ra
CO
cn
u
CD
ra
ru
ru
ru
u
-------
100
1
1
D
D
T>
X)
Figure J&0r. \ -hour Quantile-Quantile plot for all hours in the Tracy
SF6 database - AERMOD, CTDMPLUS, and ISCST3
predictions
Tracy SF6 1 -Hr Q-Q Plot (Cone.)
Using Version 98314 of AERMOD and 100m Data for ISCST3
Observed
m
CO
i\j
en
ID
I\J
o
o
I
m
CO
en
u
en
ID
i-'
CO
C3
OJ
-------
NERL-RTP-0-602 TECHNICAL REPORT DATA
1. REPORT NO. 2.
600/A-99/019
4. TITLE AND SUBTITLE
AERMOD: Model Formulation and Evaluation Results
7. AUTHOR (S)
'R. Paine, 2R. Lee, 3R. Brode, "R. Wilson, 5A. Cimorelli, 6S. Perry, 7J. Weil,
8A. Venkatram, and 9W. Peters
9. PERFORMING ORGANIZATION NAME AND ADDRESS
'ENSR Consulting and Engineering, 35 Nagog Park, Acton, MA 01720
2Trinity Consultants, 79 T.W. Alexander Drive, RTF, NC 27709
'Pacific Environmental Services, 5001 S. Miami Blvd., RTP, NC 27709
4 USEPA, Region 10, 1200 6th Avenue, Seattle, WA 98101
5USEPA, Region 3, 841 Chestnut Street, Philadelphia, PA 19107
'Same as Block 12
7CIRES, University of Colorado, Boulder, CO 80309
"College of Engineering, University of California at Riverside, Riverside, CA 92521
»USEPA/OAQPS, Research Triangle Park, NC 27711
12. SPONSORING AGENCY NAME AND ADDRESS
National Exposure Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, NC 2771 1
3. RECIPIENT'S ACCESSION NO.
5. REPORT DATE
6. PERFORMING ORGANIZATION CODE
8. PERFORMING ORGANIZATION REPORT
NO.
10.PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
13.TYPE OF REPORT AND PERIOD COVERED
14. SPONSORING AGENCY CODE
EPA/600/9
15. SUPPLEMENTARY NOTES
16. ABSTRACT
AERMOD is an advanced plume model that incorporates updated treatments of the boundary layer theory, understanding of
turbulence and dispersion, and includes handling of terrain interactions. This pafer presents an overview of AERMOD's
features relative to ISCST3.
AERMOD has been evaluated on 10 databases, which include flat and hilly terrain areas, urban and rural sites, and a
mixture of tracer experiments as well as routine monitoring networks with a limited number of fixed monitoring sites. This
paper presents a summary of the evaluation results of AERMOD with these diverse databases.
17. KEY WORDS AND DOCUMENT ANALYSIS
a. DESCRIPTORS b.IDENTIFIERS/ OPEN ENDED c.COSATl
TERMS
18. DISTRIBUTION STATEMENT 19. SECURITY CLASS (This Report) 21 .NO. OF PAGES
23
20. SECURITY CLASS (This Page) 22. PRICE
------- |