AIR DISPERSION MODELING:
A Review of Available Point Source Dispersion Models
Mathematical Modeling of Energy and Environmental Systems
MANE6960H01
Spring 2013
Sarah Mahon
INTRODUCTION
Air dispersion modeling is the mathematical simulation of how pollutants disperse in the
atmosphere. Dispersion modeling can be used for multiple purposes; including to model the path of
volcanic ash in the atmosphere after an eruption, to model the plume of pollutants emanating from
a coal-burning electricity producing plant, to model the impact that an increase in vehicular traffic
will have on ambient air conditions, or for establishing evacuation zones after a release of toxic
chemicals.
A wide variety of free software is available for public use. To successfully decide which software to
use, it is important to understand exactly what needs to be modeled and what output is desired.
Models exist for small local scales and larger planetary scales, for dispersion of primary pollutants
(such as NOX) or for secondary pollutants (such as ozone, which NOX can convert into), for
steady state emissions of particles (from stacks) or for non-steady state puffs (from chemical
releases). Model outputs can predict worst case pollutant concentration at a specific point, expected
concentrations based on specific atmospheric conditions at a certain point, or create a threedimensional graphic showing the concentration of a pollutant within space. For example, the
Regional Modeling System for Aerosols and Deposition (REMSAD), a program available from the
Environmental Protection Agency (EPA) used to model inert and chemically reactive pollutants
over regional scales (including processes that contribute to haze), can not be used be emergency
response personnel to quickly setup evacuation zones downwind of a hazardous pollutant release.
This report focuses on point source dispersion modeling based on the Gaussian air dispersion
equations in order to determine concentrations of pollutants at downwind receptors. The report will
discuss the mathematical basis for point source dispersion modeling, the regulatory framework
(specifically in Connecticut), and on models that were researched and used in an attempt to model
air dispersion from a point-source.
MATHEMATICAL BASIS OF AIR DISPERSION MODELING
The basic Gaussian equation, according to equation 10.3 in “Energy and the Environment” by Fay
and Golomb, is:
Where Qp is the mass rate of a pollutant in grams/second, H is the effective stack height, y is the
horizontal distance, z is the vertical distance, and the sigmas represent the standard deviation of the
bell shaped pollutant distribution in the horizontal and vertical plane at a specific downwind distance
x (Fay & Golomb). The sigmas are a function of atmospheric stability. According to D. Bruce
Turner in the Workbook of Atmospheric Dispersion modeling, there are three basic stability
categories;

Unstable: net upward heat flux, super adiabatic thermal structure, considerable horizontal
and vertical turbulence, typically mid day with clear skies and light wind

Neutral: no heat flux, near dry adiabatic thermal structure, mid-range turbulence, typically
during windy or cloudy days

Stable: net downward heat flux, near isothermal or inversion thermal structure, damped out
vertical turbulence, typically during nighttime with clear sky and light wind
If equations are done by hand, the values of sigma can be determined with charts showing
downwind distance x and the atmospheric stability class. Most computer models have this
information embedded within the programming language; users do not need to manually look this
up.
There are several assumptions associated with the Gaussian model, including

Steady state meteorological conditions (wind speed / direction remains constant) from
source to receptor

Conservation of mass (no deposition, no chemical reactions in the atmosphere, no
components absorbed by vegetation/water bodies)

Crosswind and vertical concentration distributions can be well represented by the
Gaussian distribution

Continuous, steady state emissions
Due to these assumptions, Gaussian dispersion models are limited to a short-term modeling for
primary pollutants (pollutants that do not convert to secondary pollutants in the atmosphere) from
single point sources. For industries trying to demonstrate compliance with regulatory emission
standards from steady state point source emissions, models using Gaussian dispersion models are a
good fit.
REGULATORY BACKGROUND
The Clean Air Act (CAA) is a federal law that regulates air emissions form stationary and mobile
sources. The EPA has delegated the authority to enforce the Clean Air Act regulations to some state
and local authorities, including the State of Connecticut. Regulations found at RCSA 22a-1743a(d)(3)(B) and (C) require that the owner for any source for which an application for an air permit
is submitted demonstrate that the operation will not cause or contribute significantly to a violation
of any federal or state air quality standard or prevention of significant deterioration increment.
Modeled compliance is demonstrated by adding background levels of pollutants to worst-case
scenario modeled levels and demonstrating compliance with the National Ambient Air Quality
Standards (NAAQS), Prevention of Significant Deterioration (PSD) increments, and the statespecific Connecticut Ambient Air Quality Standards (CAAQS) for dioxin. Facilities which want to
construct a new source of air pollution, modify an existing source of air pollution, or need to bring
an existing source into compliance with standards are required to demonstrate compliance with
these standards using dispersion modeling.
Due to the complexity of air dispersion modeling, the State of Connecticut recommends meeting
with the Department of Environmental Protection (CT DEP) prior to starting any modeling project.
The CT DEP Ambient Impact Analysis Guidance document, prepared in July 2009, recommends
SCREEN3 as a screening level air dispersion modeling program and AERMOD as the refined
modeling program. If a facility can demonstrate compliance using the worst-case scenarios using the
screening models, refined modeling using AERMOD is not required.
The CT DEP provides multiple useful resources for air modeling purposes, such as pre-processed
meteorological data for use in AERMOD, criteria air pollutant background levels for the most
recent three years from six CT DEP monitored sites, and links to multiple state and EPA guidance
documents for successfully completing dispersion modeling analyses. A discussion of the
Connecticut recommended models, as well as other models researched drug this project, are
discussed below.
DISPERSON MODELING PROGRAMS USED
Gaussianplume.m
The first semi-successful modeling attempt created a three-dimensional visual output of a Gaussian
plume. The model, GaussianPlume, was created by Harold Bien and available from
http://www.mathworks.com/matlabcentral/fileexchange/13279-gaussianplume . This is a Matlab based model
that calculates the dispersion of a continuous point source and outputs a three-dimensional matrix
that can be plotted using Matlabs volume visualization function. Although it can be a useful tool to
visualize a three-dimensional plume, the author states on his webpage that the program was
developed for a homework assignment, that the software was based on outdated documentation,
that he had not tested many of the options embedded in his software for accuracy, and that the EPA
has other preferred models available on-line that can be downloaded for free. In addition, the
program did not make units clear (such as Celsius, Kelvin, or Fahrenheit for temperature), which
could potentially lead to inaccurate calculations. Due to the potential for errors, this attempt was
considered “semi-successful” and additional air dispersion models researched.
L1023
One failed attempt at air dispersion modeling involved the program L1023; the failure may be due to
the fact that the program was originally written in the 1970’s and is not compatible with current
technology.
The program L1023 was made available with the book “Workbook of Atmospheric Dispersion
Estimates” by D. Bruce Turner. The book was originally published in 1970 and was an invaluable
reference to EPA in creating portions of the Clean Air Act. The book is cited in multiple EPA
reference guides, and has been made available by EPA for public use through the National Service
Center for Environmental Publications (http://nepis.epa.gov); a website that makes EPA
documents available for free download. The program was originally provided on a “floppy
diskette” and contained 26 examples for use in obtaining a basic understanding of dispersion
modeling using a variety of inputs and combination of parameter values.
Although the book no longer comes with “floppy diskette”, the program is available for download
from the CRC Press website ( http://www.crcpress.com/product/isbn/9781566700238). However,
the program does not work. After asking what initial action the user wants to take, the program will
only allow the answer to be typed within the middle of the question. Since no answer is provided in
the space the program is expecting an answer, the program will either continue to ask the same
question or completely shut itself down. A screenshot of where the program gets stuck is below; the
white circle indicates where the program allows you to type a response.
The program was installed on two separate computers; on both computer desktops, on both of the
local C: drives, and on one network drive with the same result each time. The CRC Press website
states that the program is an all windows version using an OS platform that was updated on
September 14, 2006; however the unzipped program files indicates the last update was in 1994. The
instructions for uploading the program refer to use of the floppy diskette; some of the steps outlined
for successful upload are no longer applicable with current computer technology. It is unclear if
there is a bug in the program, if the program was not uploaded properly due to the outdated
instructions, or if the program (written in the 1970’s) is incompatible with current computer
technology. Therefore, additional models were investigated for use in point source dispersion
modeling.
AERMOD/AERSCREEN
The AERMOD system is the model preferred by EPA and CT DEP in estimating pollutant
concentrations for regulatory uses. It is available for download from the EPA Support Center for
Regulatory Atmospheric Modeling (SCRAM) website. The AERMOD system is extremely complex;
it involves several preprocessors including AERMET for meteorological data, AERMAP for terrain
that incorporates USGS digital elevation data, AERSURFACE for surface characteristics, and
BPIPRIME for incorporating the effects of multi-building facilities. Since the program can not
currently calculate design values for lead, the program also has a post-processing tool, LEADPOST,
which can be used to calculate design values from the AERMOD output. The users guide and users
guide addendums for each of these modeling programs are hundreds of pages each; and each
individual program requires that the user understand the inputs/outputs for successful
implementation of AERMOD. Due to the complexity of AERMOD, the EPA recommends that
screening models are used prior to AERMOD. If a user can demonstrate compliance using the
worst-case screening scenario, the complex AERMOD model does not need to be used.
The AERSCREEN model is the screening version of AERMOD. To operate AERSCREEN, a user
needs to have the latest AERMOD executable download as well as the AERMAP terrain
preprocessor and BPIPRIME executable. The model will produce worst-case 1-hour average
concentrations from a single source without the need for hourly meteorological data input. It is
intended to produce concentration estimates that are equal to or greater than AERMOD, but the
‘conservative’ aspect of the program will vary on a case-by-case basis.
The EPA provides several test cases for operating the AERSCREEN model, including test cases for
area sources, circle sources, flare sources, point sources, point sources with a cap, horizontal point
sources, and volume sources. The point source test case is extremely easy to operate; however the
intent of the test case appears to be assisting new users with operating models within a DOS
window. The test case does not demonstrate how the preprocessors (such as AERMAP) are
incorporated into the program. Therefore, although familiarity with operating AERSCREEN was
developed, a full understanding of how the program should be used was not gained. Therefore, the
SCREEN3 modeling program (recommended by the state of CT as a screening program) was used
researched for use as a screening level model.
SCREEN3
The SCREEN3 model was successfully run; examples of a successful run for particulate matter from
a wood burning stove are attached in Appendix A. This model is recommended for screening level
assessments by the CT DEP in the Ambient Impact Analysis Guidelines published in July 2009 and
is available for download on the EPA SCRAM website. The program can be used to estimate
ambient impacts from point, area, volume, and flares to a distance of 50 kilometers.
For point sources, the program requires the following inputs:

Emission rate (g/s)

Stack height (m)

Stack inside diameter (m)

Gas exit velocity (m/s)

Gas temperature (K)

Ambient temperature (K)

Receptor height (ground level or flagpole, in m)

Urban/rural terrain

Worst-case building dimensions (if building downwash is applicable).
The receptor needs to be selected at both horizontal and vertical distances from the source being
analyzed. The model may need to be run multiple times for receptors in different directions from
the original point source. In addition, if the model shows that concentrations are still increasing at
the specified receptor area, the receptor area needs to be increased until a decrease in pollutant
emissions are observed.
The SCREEN3 model is an interactive model, with self-explanatory prompts. The program has the
ability incorporate building downwash effects, estimate concentrations in the cavity recirculation
zone, estimate concentrations due to shoreline fumigation, analyze dispersion over both flat and
complex terrain, and determine plume rise from flare releases. Worst-case meteorological conditions
are built in to the program; the user does not need to look up worst case scenarios for different
downwind distances. The effective stack height is automatically calculated by the program as well.
The SCREEN3 model is specifically for one point-source, it can not model the maximum
concentration that can result from multiple sources. The model can also estimate maximum 1-hour
concentrations; it cannot model longer period averages. Models specifically for longer averaging
periods are recommended for 24-hour or seasonal/annual averaging periods. Due to some of the
assumptions made in the SCREEN3 model, the hand calculations may differ since the program
quickly runs through a range of scenarios to determine worst case whereas people need to make
assumptions in the hand calculations. In addition, the SCREEN3 model also includes the effects of
buoyancy-induced dispersion, which are not accounted for in hand calculations.
An example of a successful run using the SCREEN3 model is included in Appendix A.
CONCLUSIONS
A large amount of models are available for the use in air dispersion modeling. It is important to
understand what the purpose of the model is and what output is desired in order to chose the right
program to use. For facilities in Connecticut who are required to demonstrate compliance with state
and federal air regulations, the SCREEN3 model available on the EPA SCRAM website can provide
a screening level assessment of potential worst-case scenarios. Depending on the screening level
outcome, facilities may need to use the more refined AERMOD model and/or consider air
pollution control equipment to reduce emitted pollutant concentrations.
APPENDIX A:
POINT SOURCE EXAMPLE: Wood burning stove















Wood burning stove installed in center of my property
SOURCE TYPE: point
EMISSION RATE: 0.002083 g/s (worst case scenario for non-catalytic wood burning stove;
EPA requires manufacturers to certify particulate emissions will not exceed 7.5 g/hour)
STACK HT: 8.53 meters (Assume 25 foot two story house with 3 feet of chimney clearance)
STACK DIAMETER: 0.0762 meters (assume 3” pipe diameter)
GAS EXIT VELOCITY FLOW RATE: 15 ACFM (makeup air requirements 10-15 cfm*)
STACK GAS TEMP: 449.8K (Assume 350-600F typical, 1000F maximum, use 350K as
worst case scenario)
AMBIENT AIR: 293 K (Assume 67F)
RECEPTOR HEIGHT: 1.7 meters (avg. height of person)
URBAN/RURAL: rural
TERRAIN: Simple
METEOROLOGY: All stabilities and wind speeds
TERRAIN HT ABOVE STACK BASE: 0 (flat terrain)
MIN/MAX DISTANCES: 63 meters (property line) – 55,000 meters (model max.)
NAAQS: 150 ug/m3 for PM10 for 24 hour period, 12 ug/m3 for PM2.5 (annual mean
averaged over 3 years)
* http://www.greenbuildingadvisor.com/blogs/dept/qa-spotlight/how-provide-makeup-air-wood-stove
Initial Output: Wood Burning Stove Example
Wood Stove – Concentrations at Discrete Distances
REFERENCES
AERMOD Implementation Guide, AERMOD Implementation Workgroup, US Environmental
Protection Agency, last revised March 19, 2009.
AERSCREEN User’s Guide, US Environmental Protection Agency, March 2011.
Ambient Impact Analysis Guideline: A Guideline for Performing Stationary Source Air Quality
Modeling in Connecticut. State of Connecticut Department of Environmental Protection, July 2009.
Bien, Harold. GaussianPlume Model, available from
http://www.mathworks.com/matlabcentral/fileexchange/13279-gaussianplume accessed April 2013.
Data Sources Required for Refined Modeling: Meteorological Data and Background Air Quality for
State of Connecticut, available at
http://www.ct.gov/DEEP/cwp/view.asp?a=2684&q=461156&deepNav_GID=1997, accessed
April – May 2013.
EPA Support Center for Regulatory Atmospheric Modeling, available at
http://www.epa.gov/ttn/scram/dispersionindex.htm, accessed April-May, 2013.
Fay, James A. and Golomb, Dan S. Energy and the Environment: Scientific and Technological
Principles. New York, Oxford University Press, 2012.
L1023 Model, download available from http://www.crcpress.com/product/isbn/9781566700238,
accessed April 2013.
National Service Center for Environmental Publications, http://nepis.epa.gov, accessed April-May
2013.
SCREEN3 Model User’s Guide. US Environmental Protection Agency, September 1995.
Turner, Bruce D. Workbook of Atmospheric Dispersion Estimates: Second Edition. Boca Raton,
CRC Press LLC, 1994.