United States
Environmental Protection
Agency
Atmospheric Sciences Research
Laboratory
Research Triangle Park NC 27711
Research and Development
EPA/600/S8-85/007 May 1985
Project Summary
SAI Airshed Model
Operations Manuals
J. Ames, T. C. Myers, L. E. Reid, D. C. Whitney, S. H. Golding, S. R. Hayes, and
S. D. Reynolds
The SAI Airshed Model Operations
Manuals present a general view of the
Systems Applications, Inc. (SAI) Air-
shed Model as well as detailed operating
instructions for the user. The User's
Manual includes discussions of all the
files needed to run the model, the data
preparation programs that produce each
file, the input formats and data prepa-
ration methods, samples of input and
output, and information flow diagrams
that illustrate the job stream control on
any computer.
The Systems Manual, the companion
to the User's Manual, describes the
system from a programmer's point of
view. Included are descriptions of all
subroutines and how they fit together,
run-time core allocation techniques,
internal methods of segment handling
by using secondary storage, and de-
tailed structure definitions of all files in
the system. The Systems Manual also
discusses procedures for implementing
the programs on different computers,
and the addition, removal, or replace-
ment of computing algorithms and data
preparation methods.
This Project Summary was developed
by SPA's Atmospheric Sciences Re-
search Laboratory, Research Triangle
Park. NC, to announce key findings of
the research project that is fully docu-
mented in a separate report of the same
title (see Project Report ordering infor-
mation at back).
Introduction
The Systems Applications, Inc. (SAI)
Airshed Model was designed to calculate
the concentrations of both reactive and
inert air pollutants through simulation of
the numerous physical and chemical
processes that take place in the atmos-
phere. Much of the current interest In the
model arises from the need to calculate
ozone concentrations as a part of various
environmental assessment studies (such
as State Implementation Plans and New
Source Reviews). The Airshed Model
provides a means for analyzing a variety
of urban-scale air quality problems.
Interactions among organic compounds
(including pure and oxygenated hydro-
carbons and hydrocarbon derivatives) and
nitrogen oxides (NOJ are usually respon-
sible for the relatively high ozone con-
centrations observed in and downwind of
urban areas. Among the main factors
affecting photochemical air quality in
urban areas are:
The spatial and temporal distribution
of emissions of NO* and organic
species.
The composition of the emitted organic
species.
The spatial and temporal variations of
the wind field.
The stability of the atmosphere and the
associated dynamics of the mixing
layer.
The chemical reactions involving
organic species and NO*.
The diurnal variations in ultraviolet
radiation.
The loss of ozone and its precursors by
surface uptake processes.
The ambient background concentra-
tions of ozone, organic species, and
NO, immediately upwind of the urban
area and within an elevated inversion
layer.
-------
Thus, development of an accurate ozone
prediction relationship requires investi-
gation of the relevant physical and
chemical atmospheric phenomena that
influence the magnitude and distribution
of both ozone and its precursors.
To limit ozone concentrations, one must
control all or part of the ozone formation
process. Practically, man has the most
influence on anthropogenic emissions. If
ozone concentrations were proportional
to emissions of organic compounds or
NOx, it would be simple to calculate the
reduction in, anthropogenic emissions
necessary to achieve a desired reduction
m ozone concentrations. However, the
relationship between ozone air quality
and emissions of NOX or organic species
is complicated and nonlinear. Further-
more, the relationship changes with
location and even with time at a given
location because of variations in emis-
sions, meteorology, and other factors.
To make available an appropriate quan-
titative treatment of this relationship, the
U.S. Environmental Protection Agency
(EPA) has, through a multiyear research
effort, supported the development of the
SAI Airshed Model.
The Airshed Model can simulate the
dynamic behavior of up to 20 pollutants.
When photochemical simulations are
carried out, 11 species must be included:
nitric oxide (NO), nitrogen dioxide (NOz),
ozone (Oa), single-bonded carbon atoms
(PAR), double-bonded carbon atoms ex-
cept ethylene (OLE), aromatic-bonded
carbon atoms (ARO), carbonyl-bonded
carbon atoms, ethylene, benzaldehyde
(BZA), peroxyacetyl nitrate (PAN), and
carbon monoxide (CO). In addition, the
model can provide predictions of SOz and
selected aerosol parameters.
Simulations are performed on a three-
dimensional grid selected by the user to
cover the region of interest. The base of
an elevated inversion layer is often se-
lected as the top of the grid, but the model
includes provisions to facilitate treatment
of the inversion layer itself. The model's
computer codes can handle any number
of cells in the horizontal and vertical
directions, although practical considera-
tions of computer storage and time re-
quirements limit the extent of the model
domain.
Treatment of Atmospheric
Chemistry
Ozone is not emitted from sources;
rather, it is formed by chemical reactions
in the atmosphere. Consequently, to
calculate ozone concentrations, a physico-
chemical model must contain a kinetic
mechanism; that is, a group of chemical
reactions and rate constants intended to
represent the ozone formation process.
Because of computing time limitations, it
is not possible to include each chemical
species of interest; thus, most of these
kinetic mechanisms treat organic com-
pounds in groups, which are often based
on reactive functional components they
contain.
Because of the association of reactions
and reactivities with carbon bonds, the
range of reactions and rate constants to
be treated in a kinetic mechanism can be
narrowed considerably when each carbon
atom is treated according to its bond type.
This concept is the basis for the Carbon-
Bond II Mechanism, a 65-reaction mech-
anism developed at SAI and employed in
the SAI Airshed Model In this mecha-
nism, the carbon atoms of each organic
compound are assigned to one of the
following groups: PAR, OLE, ethylene,
ARO, carbonyl-bonded carbon atoms, and
BZA. The user can optionally include four
reactions in the mechanism describing
SOa oxidation and reaction expressions
accounting for the formation of sulfate,
organic, and nitrate aerosol products.
Simulations of smog chamber experi-
ments indicate that the Carbon-Bond II
Mechanism performs significantly better
than those previously employed in the
Airshed Model.
Treatment ofAdvective
Pollutant Transport
Pollutants are transported primarily by
advection, that is, by the mean or bulk
motion of the wind. A major difficulty in
treating advection is wind shear, the
variation of the wind with altitude. The
wind speed near the ground typically
increases with height, but the wind
several hundred meters aloft may be
faster or slowerand frequently from a
different direction. The Airshed Model
has the capability to treat wind shear
phenomena. One or more objective tech-
niques are used to prepare appropriate
three-dimensional wind inputs to the
model.
Treatment of Turbulent
Diffusion
Pollutants are transported and dis-
persed largely by the action of the wind
rather than by molecular diffusion. The
above discussion of advection does not
represent pollutant transport completely
because it ignores the influence of small-
scale features of the wind, eddies. De-
scribing these microscale features of the
wind deterministically is difficult because
available data are insufficient to charac-
terize winds on such a fine scale. Con-
sequently, it is necessary to parameterize
the microscale turbulent diffusion pro-
cesses. In the SAI Airshed Model, and in
most other grid models as well, transport
of a pollutant by turbulent diffusion is
assumed to be proportional to the rate of
change of concentration with position
(i.e., concentration gradient). The propor-
tionality factor is termed the eddy dif-
fusivity coefficient. These coefficients are
sometimes treated as constants, but they
can more accurately be treated as func-
tions of atmospheric stability, surface
roughness, and height abovethe ground.
Because of the empirical nature of the
eddy diffusivity concept, it has up to this
time been difficult to obtain precise
measured or theoretical estimates for the
diffusivity coefficients. The approach
taken in the Airshed Model uses control
theory techniques in conjunction with the
results of a sophisticated planetary bound-
ary layer model to generate "optimal"
diffusivity values.
Treatment of Surface
Removal Processes
Many types of pollutants, including
NOa, Oa, and SOj in particular, can be
removed from the atmosphere by chem-
ical reaction, adsorption, or absorption at
various surfaces found in urban areas. A
methodology for calculating the removal
of gaseous pollutants by surface sinks
has been incorporated in the model.
Surf ace removal is assumed to take place
in two steps: transport to the surface
followed by uptake by the surface. Pa-
rameterization of this two-step process is
accomplished by defining a resistance to
mass transport and a resistance to surface
removal. The transport resistance is
estimated from theoretical considerations
of turbulent transfer in the atmospheric
boundary layer, and the surface resis-
tance is obtained from experimental data
on the uptake of pollutants by various
types of surfaces.
Other Technical Features
The numerical procedure used to solve
the conservation-of-mass equations for
each pollutant is an important component
of a grid model. To minimize the propaga-
tion of numerical errors, techniques for
treating horizontal (advective) pollutant
transport (an element of the model par-
ticularly difficult to handle numerically)
were examined. The results of these
-------
studies indicated that, of the techniques
examined, the SHASTA integration
scheme provided the best balance be-
tween accuracy of prediction and compu-
tational speed.
The SAI Airshed Model contains fea-
tures designed to reduce cost by minimi-
zing labor and computing expenses. For
example, computerized data preparation
programs are available for automating
the conversion of various types of data
into the proper input format. These fea-
tures have no effect on the model's
treatment of atmospheric processes, but
they do aid the user in preparing the
inputs required by the Airshed Model.
The model output consists of the pre-
dicted pollutant concentrations for every
grid cell. These predictions are generally
averaged over a period of one hour and
are saved on a computer file for subse-
quent display and analysis. Gridded
census data can be used in conjunction
with the model predictions to estimate
the dosages and exposures that are
experienced by the human population
within the modeling region. Other dis-
plays and analyses can be prepared
depending on the needs of the user.
Overview of the System
The SAI Airshed Model System con-
tains at its core the Airshed Simulation
Program, the input data that consist of 10
to 14 files depending on the program
options chosen. Each file is created by a
separate data preparation program. There
are 17 programs in the entire Airshed
Model System. The output from the
Airshed Simulation Program consists of
three data results files and an execution
trace report. The output data files can be
further processed by the display and
analysis programs and can also be used
for restarting the simulation.
The Airshed Model System can be de-
scribed according to the five major tasks
or functions that the user and the pro-
grams must perform:
M.E.B. (meteorology, emissions, and
initial and boundary conditions) file
preparation.
M.E.B. file segmentation.
Control data preparation.
Airshed simulation.
Display and analysis.
This classification of programs and tasks
is used as the organizational basis for
both the User's Manual and the Systems
Manual.
Of the 14 files input to the Airshed
Simulation Program, two are classified as
"control files."The remaining 12 files are
the M.E.B. files. The Airshed Simulation
Program requires appropriate meteoro-
logical, emissions, and initial conditions
data for each cell of the region's rectangu-
lar grid. In addition, the concentration of
each pollutant must be specified at each
point on the boundary where the wind is
flowing into the region. These data enter
the program through the M.E.B. files.
For each file, there is a data preparation
program that takes either observations or
estimates at specific locations in the
region, creates a fully gridded data field,
and writes the data to the file in the
appropriate format. The 12 M.E.B. data
preparation programs are designed to be
used together as a unified, user-oriented
package. The input formats to all the
programs are standardized and reasona-
bly self-documenting. The interpolation
methods are intended to accommodate
input data of any complexity, from varia-
bles for which little or no data are availa-
ble to those for which there are many
observations and well-known interactions.
Also, the units of measure are standard-
ized, a flexible unit conversion scheme is
built into the programs.
The M.E.B. data preparation function
consists of the following tasks for the
user:
Determination of the appropriate files
for a given simulation.
Detailed examination of the data re-
quirements and optional methods for
each file.
Translation of available data into a
form acceptable to the model.
Selection of the proper interpolation
methods.
Establishment of the size and location
of the region.
Establishment of the vertical distribu-
tion of cells within the region.
Exercise of the data preparation pro-
grams, examination of the results, and
modifications of the inputs or algo-
rithms until a reasonable set of data is
produced.
The preparation of the M.E.B. data files
is certainly the most difficult and time-
consuming task associated with use of
the SAI Airshed Model System. It is also
the most important task, and careful,
intelligent creation of the input is required
to yield the best simulation results.
The SAI Airshed Model System has
been designed to enable simulation on a
rectangular grid of any size and shape;
that is, the dimensions of the region are
not fixed within the programs but are
specified at run time. If a specified region
does not fit within the core limitations of
the host computer, that region can be
divided into any number of contiguous
subrectangles called "segments"; in this
case, the size of each segment, rather
than the size of the entire region, is
limited by the amount of core available.
The Airshed Simulation Program proces-
ses each segment in turn, and informa-
tion at the interfaces between segments
is maintained.
The Airshed Simulation Program re-
quires two control files: the Chemistry
Parameters file (CHEMPARAM), and the
Simulation Controls file (SIMCONTROL),
each created by its own data preparation
program.
For the CHEMPARAM file, all chemical
species to be simulated are named and
their properties are specified. This list of
species, rather than the lists that appear
on any of the other data files, determines
which species will be simulated. If there
are any reactive species, the reaction rate
constants and their other properties (such
as the activation energy and reference
temperature) are also specified. The
user's tasks in making the CHEMPARAM
file are:
Preparation of a card deck describing
the species and the chemical mecha-
nism to be used.
Running the program that creates the
CHEMPARAM file.
The SIMCONTROL file includes values
for the simulation options, time span,
integration controls, and output options,
which are the data likely to be changed
most frequently. The user's tasks in
generating this file are:
Proper specification of all input files to
be used.
Arranging to save the output files that
the program creates.
The final function of the SAI Airshed
Model system is the display and analysis
of the simulation results. This function
has been removed from the Airshed Sim-
ulation Program because often, no matter
which output options are selected for a
simulation run, different options are
needed ata later time. By writing the con-
centration predictions of all species in all
-------
dimensions to output files, one can select
any available output mode at any later
time.
The current display program capabili-
ties include printout of either instantane-
ous or average concentation grid maps
for selected species at selected vertical
levels at selected times. Instantaneous or
average vertical concentration profiles
can also be printed at user-specified sta-
tions. The user's tasks in running the dis-
play program are:
Proper specification of all input files
required (including the desired output
file from the Airshed Simulation Pro-
gram).
Preparation of an input deck contain-
ing parameters for the selection of
times, species, and vertical levels to be
printed, and station locations where
vertical concentration profiles will be
printed.
In addition, the design of the display
program provides for the inclusion of
additional analysis capabilities without
extensive modification of the program.
Specifically, the structure permits the
comparison of two concentration data
files. The two data files could be the out-
put from two simulation runs (comparing
different control strategies, for example)
or one file could be simulation predictions
and the other could be a gridded repre-
sentation of station observation data.
All SAI Airshed Model programs are
written in FORTRAN IVfusing FORTRAN-
66 compiler features) and were originally
implemented on the Lawrence Berkeley
Laboratory CDC 7600 computer system.
The programs were subsequently modi-
fied to run on the U.S. Environmental Pro-
tection Agency's UNIVAC 1100 compu-
ter. The following are some features of
the code for this system:
There are no multiple entry points or
returns.
No arrays are greater than three-
dimensional.
There are no complex or double-
precision variables.
There is no rereading of input data
cards.
All alphanumeric variables are repre-
sented by vectors containing one
character per word, left justified.
Output format statements use Ho
rith counts rather than special delin
ers for alphanumeric text fields.
Routines accessing secondary store
files or large core memory are isolai
modules that can be easily changei
All main programs begin with co
ments naming all files that the p
gram uses and, if necessary, providi
equivalences between local file narr
and FORTRAN unit numbers.
A magnetic computer tape contain
the model code and sample input a
output data will be available from t
National Technical Information Serv
(NTIS), Springfield, VA.
J, Ames. T. C. Myers, L. E. Reid, D. C. Whitney, S. H. Golding, S. R. Hayes, andS. D.
Reynolds are with Systems Applications, Inc., San Rafael, CA 94903.
Kenneth L. Demerjian and Kenneth L. Schere are the EPA Project Officers (see
below).
The complete report consists of two volumes, entitled "SAI Airshed Model
Operations Manuals:"
"Volume I. User's Manual,"(Order No. PB 85-191 567'/AS; Cost: $35.50)
"Volume II. Systems Manual."(Order No. PB 85-191 575/AS; Cost: $20.50)
The above reports will ba available only from: (costs subject to change)
National Technical Information Service
5285 Port Royal Road
Springfield, VA22161
Telephone: 703-487-4650
The EPA Project Officers can be contacted at:
Atmospheric Sciences Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
ft U.S GOVERNMENT PRINTING OFFICE. 1985-559-016/270
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
JUN 6 35 !>Hhvv~ !
Official Business
Penalty for Private Use $300
OCOG329 PS
U S 5NVIR PROTECTION AGENCY
REGION 5 LIBRARY
230 S DEARBORN STREFT
CHICAGO IL 60604
------- |