United States
Environmental Protection
Agency
Atmospheric Sciences x
Research Laboratory •»•
Research Triangle Park NC 2771C
Research and Development
EPA/600/S3-86/048 Dec. 1986
11 v
Project Summary
Urban Aerosol Modeling:
Incorporation of an S02
Photochemical Oxidation
Module in AROSOL
J. R. Brock
Current developments of the urban scale
K-theory aerosol model termed AROSOL
are described. Major modifications include
introduction of sulfate aerosol conversion
schemes based on the empirical model
due to Meagher as well as the Carbon
Bond Mechanism. Modules have been
developed for including the aerosol
dynamical processes such as nucleation,
condensation/evaporation and coagula-
tion. This will permit modeling of the ur-
ban aerosol composition distributions,
now that supercomputers are widely
available. Concurrently, a version of
AROSOL is being ported to the IBM PC-
RT, a relatively inexpensive workstation, so
that the code will be generally available in
a user friendly form.
This Project Summary was developed
by EPA's Atmospheric Sciences Research
Laboratory, Research Triangle Park, NC, to
announce key findings of the research pro-
ject that is fully documented in a separate
report of the same title (see Project Report
ordering information at the back).
Introduction
The adverse effects of urban air pollu-
tion are well known: increased incidence
of respiratory diseases, reduced visibility,
economic loss, and inadvertent weather
modification, to name a few. The urban
aerosol plays an important role in these
adverse effects. An understanding of this
role requires, among other factors, knowl-
edge of the aerosol dynamical processes
shaping the aerosol size and composition
distributions. These dynamical processes
include advection, dispersion, deposition,
primary and secondary source input,
nucleation, condensation/evaporation, and
coagulation. Models for the particle size
and composition distributions reflecting
these dynamical processes will ultimately
be necessary for development of realistic
aerosol air quality regulations and control
strategies. For example, the deposition
pattern of particles in the lung is a func-
tion of the particle size distribution. Ab-
sorption and scattering of light by the am-
bient aerosol is a complex function of the
composition distribution. Many of the
chemical species making up the urban aer-
osol are produced through the interaction
of complex chemical processes and the
ambient aerosol, the details of which vary
with particle size and composition. There-
fore, the effects of pollutant emissions
reduction and control are likely to be non-
linear in character and will require the
detailed knowledge cited.
This project has been concerned with
development of urban aerosol models
which include the dynamical processes
shaping the aerosol size and composition
distributions. The first part of this work
has involved evaluation of data bases and
a K-theory model, termed AROSOL, for
the super-and sub-micrometer aerosol
mass concentration. Recent work has
been devoted to development of computer
modules for accurate description of the
coagulation and condensation/evaporation
processes for single component aerosols.
Current work has proceeded along several
lines. The data bases from the EPA Phila-
delphia Aerosol Field Study (PAFS) have
been evaluated and corrected for evalua-
tion of AROSOL for sulfate aerosol model-
ing. Modules have been developed for in-
-------
elusion in AROSOL which account for
chemical conversion, nucleation, and con-
densation, evaporation, and coagulation as
a step in the extension of AROSOL to
multicomponent aerosols. Work has been
done on coupling the nucleation and con-
densation/evaporation processes to a suit-
able oxidant mechanism for sulfate forma-
tion. A parallel development is the deriva-
tion of a more highly parameterized model
from AROSOL, suitable for assessments,
which will run on a relatively inexpensive
computer workstation such as the IBM
PC-RT.
Model Description
AROSOL is an urban scale K-theory
model for the ambient atmospheric aero-
sol. It has been previously used in a
number of studies including modeling total
aerosol mass concentration in Maricopa
County, Arizona and super- and sub-
micrometer aerosol mass concentration in
St. Louis, Missouri using the Regional Air
Pollution Study (RAPS) data base. This
model has been under continuous develop-
ment. The extension of the model to in-
clude secondary sulfate aerosol sources
and aerosol dynamics is described in this
Project Summary. First, the modules of
AROSOL are listed. These modules are
briefly described, including the recent
work on inclusion of multicomponent
aerosol dynamics and oxidant chemistry
for the gas phase conversion of S02 to
sulfate.
AROSOL is arranged with a number of
options for inclusion of meteorological
data ranging from a single measurement
of wind speed and direction to time series
of data on wind speed and direction and
atmospheric stability. At present, the
model includes no numerical meteoro-
logical driver.
The structure of AROSOL is best indi-
cated by the schematic diagram shown
below.
As indicated, data input includes emis-
sions inventory data and meteorological
data. Emissions inventory data are entered
from tape or other data storage media.
Meteorological data may be entered ex-
ternally from data storage media or, in the
case of limited meteorological data, may
be entered internally through data state-
ments. The model has provision for infer-
ring complete wind field and stability con-
ditions over a study area for a range of
data availability, from measurements at a
single point to complete time series at
multiple points.
Results from AROSOL simulations may
be displayed in various forms including
tabular data, isopleth plots, and full three
dimensional plots. Modules are available
for analysis of simulation results and
measurement data including statistical
analysis options for model evaluation as
recommended by EPA as well as spectral
analysis. New analytical modules will be
included as available.
The various rate modules of AROSOL
are indicated in the diagram above. The
Eulerian grid consists of horizontal x,y
UTM coordinates and a vertical sigma, s,
coordinate for simulation of complex ter-
rain. The horizontal grid is usually em-
ployed with a variable mesh with highest
density in the regions of large sources.
Usually, no more than the order of 14 ver-
tical levels of variable thickness are em-
ployed. Time splitting is used based on the
process. The steps are ordered as first, ad-
vection, then diffusion, and finally the aer-
osol dynamical and chemical processes.
The x,y boundary conditions are speci-
fied depending on whether the wind is
flowing into or out of the modeling region.
For wind flowing out of the modeling
region a constant flux condition in x,y is
imposed. For wind flowing into the mod-
eling region, the appropriate pollutant
variable must be imposed; these values
must be set from background assumptions
or actual upstream measurements. The
boundary condition at the surface is dis-
cussed below under the section on dry de-
position. The upper level boundary condi-
tion is usually specified by a zero flux con-
dition at the mixing height which is speci-
fied from field data. It is possible to allow
for entrainment through the upper bound-
ary but this must be done empirically.
The rate modules indicated in the dia-
gram above will now be discussed in order.
Rate Modules
Advection
The advection term in the conservation
equation is evaluated numerically using
the iterated upstream finite difference
scheme proposed by Smolarkiewicz. This
scheme is used because it maintains the
conservation and positiveness of trans-
ported conservative species reasonably
well. The scheme is also relatively fast
compared to other methods with similar
conservation properties. In addition the
scheme is local in that the advective time
change at a point is determined by the
concentrations and velocities at a small
number of neighboring points.
Dispersion
The vertical and horizontal diffusivity
terms are handled by orthogonal colloca-
tion on finite elements. This scheme has
been found to provide reasonable ac-
curacy and efficiency in dispersion simula-
tions. It is widely recognized that the eddy
diffusivities are not fluid properties but are
functions of the dynamical state of the at-
mosphere. Therefore the eddy diffusivities
are empirical constructs which must be
specified for various atmospheric states.
Empirical parameterizations of horizontal
and vertical diffusivities for stable, neutral
and unstable conditions are employed.
Dry deposition
At the surface, the boundary condition
is specified by parameterizations of depo-
sition velocities. For gases, surface resis-
tances are scaled from field data for S02.
For submicrometer and supermicrometer
particles the most recent empirical param-
eterizations are used.
Chemistry
AROSOL offers two options in the
calculation of the homogeneous gas phase
conversion of S02 to sulfate. One of
these is an empirical first-order decay
model due to Meagher. The other is a com-
plex photochemical mechanism known as
the carbon bond mechanism (CBM), suit-
ably modified for sulfate conversion.
The empirical model of Meagher (EMM)
is based on field measurements. Its sim-
plicity and computational efficiency make
it appealing for first-order estimates of sul-
fate formation rates when the amount of
reliable input data is limited. EMM is based
loosely on the diurnal and annual variation
in clear-sky solar intensity. The basic
assumption is that the rate constant for
sulfate conversion can be separated into
a constant value and a component that is
allowed to vary diurnally and annually. The
parameters in the rate equation are derived
from measurements from a number of field
studies of sulfate conversion.
The carbon bond mechanism (CBM) ap-
proach appears to be an attractive method
for modeling atmospheric organic chemis-
try and therefore, suitably modified, ap-
pears to be useful in modeling sulfate con-
version. For the CBM, organic species are
grouped according to type and number of
carbon bonds in the molecule irrespective
of the particular chemical species in which
the bond is found. The structural classes
have been determined through the use of
structural reactivity analysis.
The carbon bond mechanism has sev-
eral advantages over other lumped mech-
anisms. To name a few, it conserves car-
bon atoms, it has a narrow range of reac-
tivities for a group of lumped species, and
the reactivity of the organic mixture
changes in a natural way. The mechanism
-------
follows the reactions of the lumped
species. The rate constants for these
species are obtained by averaging the rate
constants for the individual species that
comprise the lumped identity. The reactiv-
ity of an organic mixture changes accord-
ing to the concentrations of the different
bond types in the carbon mechanism.
CBM uses a hierarchy of chemical species
in which each species involved in oxidant
formation is assigned a hierarchical level
on the basis of the number of systems in
which it occurs. AROSOL includes CBM-
IV, which is a condensed version of the
larger CBM-X scheme. This condensed
version is suitable for use in a computa-
tionally demanding model such as
AROSOL in an urban or regional area. On
the urban scale the problem of the correla-
tion terms in a turbulent reacting system
is now under study.
Aerosol dynamics
The term aerosol dynamics in the pres-
ent context refers to the processes of nu-
cleation, condensation, evaporation, and
coagulation. These processes alter the
size and composition distributions of the
ambient atmospheric aerosol. There are
two limits in the approach to modeling
secondary aerosol formation in the atmos-
phere. An illustration will serve to make
these limiting approaches clear.
For the case of sulfate conversion, most
current regional models regard sulfate as
essentially another chemical species and
take no account of its distribution in size
and composition; in this approach, termed
here the lumped aerosol model (LAM), sul-
fate is emitted from primary sources, cre-
ated by atmospheric reactions, advects
and disperses, and undergoes deposition
processes in an analogous manner to gas-
eous species. The deficiencies of the LAM
approach can be illustrated by a few ex-
amples. Deposition processes for sulfate
particles are size and possibly composition
dependent; therefore, in order to be gen-
erally applicable, empirical formulations of
such deposition rates must erroneously
assume a universal distribution of sulfate.
In addition, in LAM the distinction be-
tween condensation and nucleation pro-
cesses is ignored. Also if heterogeneous
conversion processes are neglected, the
rate of formation of sulfate mass must be
assumed to equal the rate of production
by gas phase chemical reaction; this as-
sumption may be valid for sulfate, but
would not be expected to be valid for more
volatile components such as nitrates and
some of the organic fractions. In this case,
LAM would not be a useful approach.
At the other extreme from LAM is the
model that deals with particle size and
composition distributions and attempts to
account for all the aerosol dynamics pro-
cesses — an approach termed here the
aerosol dynamics model (ADM). This may
be styled as an ab initio approach, which
at least in principle is not required to in-
voke any empirical rate information. In ad-
dition to very accurate numerical tech-
niques for the aerosol dynamical proces-
ses, available as modules, computationally
efficient moment techniques have been
developed and are also available as
modules.
AROSOL has the option of invoking
either the LAM or ADM. The LAM option
would be suitable for some assessment
purposes, where, for example, sulfate
mass concentration would be the only
interest. It is one of the aims of this pro-
ject to compare the LAM and ADM op-
tions in urban simulations with AROSOL.
This is now possible with the University
of Texas CRAY-XMP.
Recommendations
With increasing speed and memory of
computers, it is feasible to carry out evalu-
ation studies of models such as AROSOL
for the aerosol size and composition
distributions. As a first step, planning of
field studies should be carried out for
obtaining the needed field data, including
emissions inventory and measurements of
either the aerosol distributions or their
moments. The moment of a distribution is
the integral, over the range of the size
parameter, of the product of the distribu-
tion and some power of the size param-
eter. For example, the first moment of a
mass distribution of particles would be the
total mass concentration of particles. Plan-
ning of such field studies implies that
suitable measurement systems would be
available. Therefore, while development of
AROSOL continues, a parallel program for
development of experimental systems for
large area measurements of • ambient
aerosol size and composition distributions,
or their moments, would be desirable.
Modern developments in laser spectros-
copy and remote sensing may make it
feasible to carry out field studies at
relatively low cost that would gather
information on gas concentrations and
aerosol distributions on spatial scales
commensurate with the urban scale simu-
lation grids (typically, now from
0.1kmx0.1km to 1kmx1km). Point source
measurements are always subject to local
spatial and temporal fluctuations (of
atmospheric, process and anthropogenic
origins) that cannot be included in any
reasonable manner in numerical simula-
tions. Even from a regulatory standpoint,
the always restricted number of point
measurement stations are well known to
be subject to anomalous local effects
which can lead to ineffective and costly
regulatory actions.
The Gaussian plume model, although
widely employed today for regulatory pur-
poses throughout the United States, has
well recognized and serious deficiencies.
A substantial advance over the Gaussian
plume model is afforded by the conserva-
tion equation with first-order closure— the
K-theory model. For both aerosol and gas-
eous species modeling, it is recommended
that K-theory models be developed that
would run with reasonable efficiency on
newly available and relatively inexpensive
work stations such as the IBM PC-RT. Cur-
rent activity in this project includes port-
ing AROSOL to the IBM PC-RT.
-------
J. R. Brock is with the University of Texas at Austin, Austin, TX 78712.
H. M. Barnes is the EPA Project Officer (see below).
The complete report, entitled "Urban Aerosol Modeling: Incorporation of an SOz
Photochemical Oxidation Module in A ROSOL," (Order No. PB 86-239 829/A S;
Cost: $9.95, subject to change) will be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Atmospheric Sciences Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park. NC 27711
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
Official Business
Penalty for Private Use S300
EPA/600/S3-86/048
0000329
60604
------- |