Comparison of Features and Data Requirements among the CALPUFF, AERMOD, and ADMS Models

Modeling Software for EH&S Professionals
Comparison of Features and Data Requirements
among the CALPUFF, AERMOD, and ADMS Models
Prepared By:
Russell F. Lee
BREEZE SOFTWARE
12770 Merit Drive
Suite 900
Dallas, TX 75251
+1 (972) 661-8881
breeze-software.com

Comparison of Features and Data Requirements among the CALPUFF,
AERMOD, and ADMS Models
Introduction to plume and puff models
CALPUFF, AERMOD, and ADMS are new advanced air quality models that are either being
used or are being proposed for use for air pollution regulatory applications. The physical
processes in the atmosphere that determine the transport and dispersion of air pollutants are very
complex, and are not all well understood. Furthermore, simplifying the physical processes in a
model that is sufficiently accurate for regulatory purpose, but not excessively costly and difficult
to use, is not an insignificant endeavor. Consequently, the development of air quality models is a
continuing process.
The Interagency Workgroup on Air Quality Modeling (IWAQM) has recently recommended the
use of the CALPUFF modeling system for “use in assessing air quality associated with
prevention of significant deterioration” of air quality in Federal Class 1 and Wilderness areas.
For such purposes, the ability of the model to adequately represent long-range transport of
pollutants, effect of pollution on visibility, and certain chemical reactions is critical. Once
CALPUFF was selected for their recommendation, IWAQM was involved in enhancing that
modeling system to better suit these purposes.
In 1991, the American Meteorological Society (AMS) and the U.S. Environmental Protection
Agency (EPA) cosponsored an initiative to bring plume models up to date. The committee
formed to accomplish this is the AMS/EPA Regulatory Model Improvement Committee
(AERMIC), and the resulting model is AERMOD (AERMIC Model). AERMOD is intended to
fill the niche currently occupied by the ISCST3 model. It is intended primarily to be used for
modeling non-reactive pollutants such as sulfur dioxide, carbon monoxide, and particulate
matter. These pollutants are characteristically emitted by local sources, resulting in the highest
concentrations occurring within a few kilometers of the contributing sources. After reviewing
existing models, AERMIC elected to develop a new model that was sufficiently similar to
ISCST3 that users would not have excessive difficulty in learning to apply it.
Concurrently with the development of AERMOD in the United States, a new plume model was
being developed in United Kingdom. ADMS (more accurately, UK-ADMS—UK-Atmospheric
Dispersion Modelling System) was being developed by the CERC (Cambridge Environmental
Research Consultants, Ltd.) with from industry and government organizations.
An overview of plume and puff models
Plume models have been the mainstay of regulatory near-field modeling of nonreactive
pollutants. The current regulatory model in the U.S., ISCST3, is based on work by Frank Pasquill
in the 1950’s, with modifications by Frank Gifford and adaptations for computer modeling by D.
Bruce Turner. In that model, atmospheric states are divided into six stability categories, and
plume growth is treated as a function of stability category and downwind distance. The
concentration distribution of pollutants through the plume is assumed to be Gaussian in all cases.
This is a reasonable assumption in the horizontal direction. For neutral and stable flow, when
there is no local boundary (such as the ground) affecting the turbulence profile, this is also a
reasonable assumption in the vertical direction. However, the presence of a ground surface skews
the shape of the pollutant distribution by suppressing turbulence (and, therefore, dispersion) on

the low side of a near-ground plume. During convective (i.e., unstable) conditions, things are
even more complex. Because there are very large turbulent eddies for this case, the presence of
the ground skews even elevated plumes. In addition, since the turbulent eddies are much larger
than the plume itself, parts of the plume get caught in updrafts while other parts get caught in
downdrafts. This results in a very non-Gaussian plume. These and many other problems with the
older models have been addressed in the newer plume models AERMOD and ADMS. These
newer models no longer characterize turbulence in terms of six discrete stability categories. They
recognize that turbulence is continuously variable, depending on height above ground, amount of
heating in the daytime or heat loss at night, and wind shear. Concerning the height above ground,
virtually all airflow near the ground is neutral, becoming more turbulent (daytime) or less
turbulent (nighttime) with height.
All plume models are, however, steady state by nature. This means that the models assume
steady or constant conditions between the time of emission and the time it reaches any receptor.
For downwind distances of a few kilometers, this is a reasonable assumption. However, consider
this. A plume model applied to a pollutant release during stable conditions at six o’clock in the
morning will make concentration calculations at a distance of 50 kilometers based on that one
wind direction under stable conditions. It will ignore the fact that the plume may have traveled
for six hours or more, changing direction due to changes in wind speed, and growing at varying
rates as it encounters a whole range of stabilities. It is clear that plume models cannot, in general,
be expected to give reasonable results for long range transport of pollutants.
This is where puff models are valuable. Puff models do not require the steady state assumption.
The emission that occurs from a source during the first time step (which may be a few minutes or
an hour) will be followed downwind. As the wind direction changes, the direction of motion of
the puff changes. It can even account for different winds at different locations. Unfortunately, to
give an accurate picture of the concentration distribution in a plume, these puffs must be
sufficiently close together to overlap. Close to the source, where little dispersion has occurred
and, thus, the puffs are small, it may be necessary to model a puff a second to avoid gaps
between the puffs. This would lead to extremely large runtimes for the model. For this reason, a
puff model is generally not a practical way to model concentrations near a point source, although
CALPUFF gets around the problem by some techniques described below. At large travel
distances, where the puffs have grown to considerable size, model runtimes are less severe. In
addition, the puff model can account for changes in wind and stability as the puff moves
downwind. Thus, a puff model can provide good results when long range transport is involved.
Discussion of the CALPUFF model
While many of the entries in Tables 1–3 are self-evident, several also require some discussion.
Under item 3b of Table 1, both puff and slug are indicated as options for CALPUFF. As noted
above, puff models are generally unable to represent a continuous plume near a point source, due
to the small sizes of the puffs. The plume ends up being represented by a sequence of puffs with
space between them, with resultant errors in calculated concentrations. CALPUFF solves this
problem with the optional use of a slug for such cases. A slug is an elongated puff. This allows
two separated puffs to be connected, to better represent the plume without having to generate an
excessive number of puffs. This may make it possible for CALPUFF to be used to estimate
concentrations nearer the source than is normally possible for puff models. CALPUFF also
allows an option to use a non-Gaussian probability density function for convective (unstable)

cases, as is done in AERMOD. There are also options allowing CALPUFF to emulate ISCST3
and CTDM in the near field. These should be used with caution at the present time, since there is
limited experience using CALPUFF to estimate near-field concentrations. It will be interesting to
learn how well it emulates plume models such as AERMOD and ADMS for these cases.
Since the main purpose of CALPUFF is to provide a model for long range transport of
pollutants, the meteorological data requirements can be substantial (see Tables 2 and 3).
Fortunately, for regulatory purposes, it will probably not be required to use hourly gridded wind
fields from the MM4, MM5, or CSUMM models. Nevertheless, it is a much more difficult model
to use than the traditional Gaussian models.
In general, CALPUFF is a technically good approach for long-range transport modeling, and can
account for some chemical reactions. Because of the introduction of the use of the concept of the
slug, as well as the non-Gaussian pdf, CALPUFF has potential for near-field analyses as well,
though this needs to be tested.
Discussion of the AERMOD model
AERMOD represents an improvement over traditional Gaussian models. The theory is much
more “solid,” and evaluation studies so far show substantial improvement over the traditional
models as well. AERMOD uses a non-Gaussian pdf during convective conditions, and accounts
for variations in wind, turbulence, and temperature gradient with height. The effect of wind shear
on transport is accounted for by estimating a characteristic wind speed and direction through the
thickness of the plume. AERMOD does not, however, deform the plume in response to the shear
as CALPUFF does.
AERMOD is a steady state plume model, and as such is not appropriate for long-range transport.
Although it will probably be approved for use for distances up to 50 kilometers, it is doubtful
that any plume model will retain its accuracy much beyond 20–30 kilometers. At the present
time, AERMOD does not include the capability of accounting for deposition, though there are
plans for adding this.
Discussion of the ADMS model
[TO BE ADDED]

Table 1. Comparison of Model Features
Category Feature CALPUFF AERMOD ADMS
1. Graphical User
Interface
Point & click model setup
and data input
YES Trinity
version: YES
EPA version:
NO
YES
Enhanced error checking YES Trinity
version: YES
EPA version:
NO
YES
Online help files YES Trinity
version: YES
EPA version:
NO
YES
2. Source Types
2a. Point sources YES YES YES ≤501
Variable emissions & stack
parameters
YES YES YES
Plume rise YES YES YES
Building downwash YES YES YES
Huber-Snyder YES YES NO
Schulman-Scire YES YES NO
Selection method User option Automatic N/A
Other YES
2b. Line sources YES NO YES ≤11
Variable emissions YES N/A YES
Plume rise YES N/A YES

2c. Volume sources YES YES YES ≤51
Variable emissions YES YES YES
Plume rise NO NO NO
2d. Area sources YES YES YES ≤51
Variable emissions YES YES YES
Uses Emissions
Production Model (U.S.
Forest Service) output
for controlled burns and
wildfires.
YES NO NO
Plume rise YES NO YES
2e. “Jet” sources
(non-buoyant, ejected
at any angle)
NO NO YES
2f. Treatment of
variable emissions
(all source types):
Hourly file of emissions
and stack parameters
User option User option User option
Factors by hour of day User option User option User option
Factors by season NO User option User option
Factors by month User option User option User option
Factors by hour &
season
Factors by wind speed
& stability
Factors by temperature User option NO NO

3. Plume/Puff
Characteristics
3a. Plume Rise
Buoyant rise YES YES YES
Momentum rise YES YES YES (?)
Stack tip effects YES YES YES
Partial penetration YES YES (?)
Vertical wind shear YES YES
(limited–
does not
deform
plume)
YES
(limited–
does not
deform
plume) (?)
3b. Plume/puff form
Steady-state plume NO YES YES
Puff User option NO Optional
puff for
short
releases
Slug User option NO NO
Non-Gaussian pdf User option3
YES YES
3c. Dispersion
coefficients y, z
based on:
Direct measurements of v,
w
User option2
User option (?)
Estimated v, w using
similarity theory
User option2
User option YES
PG (ISC3-Rural) User option NO NO
McElroy-Pooler (ISC3-
Urban)
User option NO NO
CTDM coefficients
(neutral/stable)
User option NO NO

3d. Overwater and
coastal effects
YES NO Overland
coastal only
Overwater boundary layer
parameters
YES NO NO
Change from overland to
over water conditions
modeled
YES NO NO
Plume fumigation YES NO YES (over
land)
TIBL included in subgrid
scale modeling
User option NO YES (on-
shore wind)
3e. Spatial variability
of meteorology
affecting plume or
puff
Gridded 3-D winds and
temperature
YES NO NO
2-D fields of zi, u*, w*, L,
precipitation rate
YES NO NO
Vertical variations in
turbulence
YES YES YES
Individual puffs split User option N/A N/A
Horizontal variation in
turbulence
YES NO NO
3f. Chemical
transformation
Exponential decay NO YES YES
Pseudo first-order reaction
for SO2, SO4
=
, NOX, HNO3
& NO3
–
User
option2
(MESO-
PUFF II
method)
NO YES
Diurnal cycle of
transformation rate
User option NO NO (?)

3g. Dry deposition User option NO YES
Gas and/or particle Both
optional
NO Both
optional
Full space/time
variations of deposition
“resistance” values
Simpler diurnal cycle of
deposition “resistance”
values
3h. Wet deposition User option NO YES
Scavenging coefficient
approach
YES NO YES
Rate is function of precip.
intensity & type
YES NO YES
4. Outputs
Concentration (Max, 2nd
high, etc.)
YES YES YES
Averaging times < 1 hour YES NO YES
Running averages Requires
external
processing
Requires
external
processing
YES
Percentiles Requires
external
processing
Requires
external
processing
YES
Deposition YES NO YES
Visibility YES NO YES
Radioactive dose NO NO YES
1. Maximum number of sources or receptors permitted in the model.
2. A (or the) preferred option for this model.
3. For convective cases, CALPUFF has non-Gaussian pdf for near-source concentrations to
approximately emulate AERMOD

NOTE: CALPUFF features referenced above do not include those in the CALGRID
Photochemical Model and the KSP Particle Model. CALPUFF and KSP are part of the
CALPUFF modeling system, and allow additional model features, including complete
photochemical modeling and lagrangian particle modeling. These models, though included in the
CALPUFF system, are documented separately.

Table 2. Comparison of Data Requirements among the Meteorological Preprocessors
Category Feature CALMET1
(CALPUFF)
AERMET
(AERMOD)
ADMS
(Meteoro-
logical
input
module)
Surface data
Hourly observations of:
wind speed,
wind direction
temperature
cloud cover
ceiling height
surface pressure
relative humidity
precipitation type
precipitation rate
YES—
multiple sites
(Precipitation
type required
only for wet
deposition)
YES—single
site
(Precipitation
type, relative
humidity,
surface pressure
not needed)
YES—
single site
(Precipitation
type, relative
humidity,
surface
pressure not
needed)

(CALPUFF)
AERMET
(AERMOD)
ADMS
Upper air data
Daily maximum mixing
heights
NO NO YES
RAWINSONDES:
observed vertical profiles
of:
 wind speed
 wind direction
 temperature
 pressure
 elevation
YES–twice
daily
soundings or
more
YES–one
early
morning
sounding
Requires
hourly
mixing
heights
Hourly gridded wind fields
from MM4/MM5
User option NO NO
Hourly gridded wind fields
from CSUMM
User option NO NO
Overwater observations
 air-sea temperature
difference
 air temperature
 relative humidity
 overwater mixing height
 wind speed
 wind direction
 overwater temperature
gradients above and
below mixing height
User option NO NO

(CALPUFF)
AERMET
(AERMOD)
ADMS
Geophysical
data
Terrain elevations (gridded) YES YES YES
Land use categories YES-gridded NO (?)
Surface roughness User option-
gridded
YES-by
direction
YES
Albedo User option-
gridded
YES-by
direction
YES
Bowen ratio User option-
gridded
YES-by
direction
YES (?)
Soil heat flux constant User option-
gridded
NO NO (?)
anthropogenic heat flux User option-
gridded
Derived from
city size
NO (?)
leaf area indices YES NO NO
1. CALPUFF can optionally be run, with degradation of accuracy, using meteorological data
preprocessed for ISCST3. It requires the addition of hourly (not gridded) friction velocity,
Monin-Obukhov length, surface roughness, precipitation code and rate, potential temperature
gradient, wind speed profile power-law exponent, short-wave solar radiation, and relative
humidity.
An additional option allows the use of the surface and profile meteorological data files which
are used in CTDMPLUS and AERMOD, with the addition of precipitation code and rate,
short-wave solar radiation, and relative humidity to the surface file. The accuracy is expected
to be degraded from the normal CALPUFF mode, but should be better than that obtained
using ISCST3 meteorological data.

Table 3. Comparison of Other Input Data Requirements among the Models
Emissions data
Point source, constant or diurnal
emission pattern
YES (with
plume rise)
YES (with
plume rise)
YES (with
plume rise)
Point source, arbitrarily
varying emission pattern (i.e.,
a file of hour by hour
emission parameters)
YES (with
plume rise)
YES (with
plume rise)
YES (with
plume rise)
Line source, constant or diurnal
emission pattern
YES (with
plume rise)
NO YES (with
plume rise)
Line source, arbitrarily
varying emission pattern
YES (with
plume rise)
NO YES (with
plume rise)
Area source, constant or diurnal
emission pattern
YES (with
plume rise)
YES (no
plume rise)
YES (with
plume rise)
Area source, arbitrarily
YES (with
plume rise)
YES (no
plume rise)
YES (with
plume rise)
Volume source, constant or
diurnal emission pattern
YES (no
plume rise)
YES (no
plume rise)
YES (no
plume rise)
Volume source, arbitrarily
YES (no
plume rise)
YES (no
plume rise)
YES (no
plume rise)
Other data
Deposition velocity data YES NO YES (?)
Ozone monitoring data YES NO YES
Chemical transformation data YES NO (?)
Terrain data YES YES YES
USGS DEM data, or
equivalent
YES YES YES (UK
data)
Hill shape and height
parameters
YES (as in
CTDM+)
Derived in
model from
terrain data
Derived in
model from
terrain data
Receptor locations with
associated hill ID
YES Derived in
model from
terrain data
Derived in
model from
terrain data
Coastal boundary data YES NO YES

Comparison of Features and Data Requirements among the CALPUFF, AERMOD, and ADMS Models

Comparison of Features and Data Requirements among the CALPUFF, AERMOD, and ADMS Models

More Related Content

Similar to Comparison of Features and Data Requirements among the CALPUFF, AERMOD, and ADMS Models (20)

More from BREEZE Software (20)

Recently uploaded (20)

Comparison of Features and Data Requirements among the CALPUFF, AERMOD, and ADMS Models