United States
Environmental Protection
Agency
Environmental Research
Laboratory
Athens GA 30613
Research and Development
EPA-600/S3-84-052 May 1984
&EPA Project Summary
Air Land Water Analysis System
(ALWAS): A Multi-Media Model
for Toxic Substances
W.A. Tucker, A.Q. Eschenroeder, and G.C. Magill
The Air Land Water Analysis System
(ALWAS) is a multi-media environmen-
tal model for describing the atmospheric
dispersion of toxicants, the surface
runoff of deposited toxicants, and the
subsequent fate of these materials in
surface water bodies. ALWAS depicts"
the spatial and temporal distribution of
contaminant concentrations in a water-
shed and the air above it. Linked in
ALWAS are three submodels that pro-
vide for independent and partially cou-
pled usage modes — Dispersion and
Deposition of Toxics (DiDOT); a modi-
fication, called NPSDEP, of the Non-
point Source Model; and Exposure
Analysis Modeling System (EXAMS).
DiDOT quantifies both wet and dry
deposition rates over hourly, daily, and
longer averaging periods. NPSDEP de-
scribes surface runoff of contaminants
and EXAMS simulates their fate in sur-
face waters.
DiDOT, which was developed in the
study, is based on a Gaussian plume ap-
proach modified to account for deposi-
tion. Its meteorological and emission
input requirements are similar to those
of standard, short-term air quality as-
sessment models. Input parameters in-
clude settling velocity, dry deposition
velocity, and precipitation scavenging
ratio. Guidance for estimating these
parameters is provided.
ALWAS is appropriate for evaluating
multi-media water quality problems of
watersheds as large as 10* km2. The full
range of ALWAS capabilities may be ex-
ercised for hydrophobic organic chem-
icals of relatively low vapor pressure. By
careful selection and linkage of the sub-
models, ALWAS may be applied to in-
vestigate the environmental behavior of
a broad range of toxicants including
highly soluble or volatile organics and
heavy metals.
The full report incorporates detailed
software documentation and a user's
manual. The sensitivity of model results
to uncertainties in key input parameters
is investigated for a hypothetical simula-
tion of benzo(a)-pyrene behavior in a
small urban watershed.
This Project Summary was developed
by EPA's Environmental Research Labo-
ratory, Athens, GA, to announce key
findings of the research project that is
fully documented in a separate report of
the same title (see Project Report order-
ing information at back).
Introduction
Most recent progress in environmental
modeling has involved a single medium (i.e.,
air quality, surface water quality, ground-
water quality, or biological systems). Many
pollutants, however, are released to more
than one medium and are subject to inter-
media transfer. Significant effects may occur
in more than one medium or may be most
important in a medium other than the one
into which a pollutant is released. These
multi-media environmental problems are
receiving increasing attention as more of
them are recognized (e.g., acid rain,
volatilization and ground-water recharge
from hazardous waste lagoons, and pesticide
losses by volatilization). Multi-media en-
vironmental models for assessing such prob-
lems have not been available.
This project was intended to partially fill
the gap by developing a model to describe
-------
the effect of atmospheric emissions, non-
point source runoff, and point discharges of
toxic pollutants on surface water quality. The
Air Land Water Analysis System (ALWAS),
one of the first models of its type, has not
been calibrated or verified for any field situa-
tion, although its individual single-media
model components have been used fre-
quently on a stand-alone basis. For these
reasons, ALWAS is considered a "first
generation" multi-media model with room
for improvement and extension.
The work included a review and evalua-
tion of existing air quality models to identify
any that would be suitable for linking with
the Nonpoint Source Model (NPS) and the
Exposure Assessment Modeling System
(EXAMS). A suitable air model would
describe both wet and dry deposition pro-
cesses for toxic pollutants and would be ap-
plicable at a river basin scale (length scales
of up to 100 km) over short-time frames in
order to interface with NPS.
The review of available air models revealed
that no accessible and documented models
that fit these criteria were available. Most air
quality models do not account for deposi-
tion processes adequately. Some that do
treat deposition are specifically designed for
the acid rain problem and are neither perti-
nent to toxic pollutants nor to the river basin
spatial scale. Models designed to simulate
acid rain must follow pollutants up to 1000
km. Consequently, they use multiple-station
weather data to define plume trajectories. A
local weather station is likely to represent
adequately the winds and mixing conditions
within a river basin unless the terrain is
mountainous. The average separation dis-
tance of National Weather Service Stations
is greater than the model application scale
so model domain will rarely include more
than one surface weather station.
From this review, it was apparent that an
air quality model had to be developed
specifically to link with NPS and EXAMS.
The result was the Dispersion and Deposi-
tion of Toxics model (DiDOT). The review
also showed that minor modifications of
NPS and EXAMS were required in order to
facilitate their linkage into ALWAS. The NPS
modification is called NPSDEP.
ALWAS can simulate the effects on sur-
face water quality of multi-media toxicant
releases to the environment. It is most
suitable for persistent organic chemicals that
tend to adsorb to paniculate matter, but
ALWAS, or various combinations of its sub-
models, may also provide valuable multi-
media information for metals and more sol-
uble organics, given care in its application.
A technical discussion of the assumptions
and rationale for the software development
is provided. A User's Guide for ALWAS is
presented that emphasizes portions of the
software developed specifically for ALWAS.
The User's Guide incorporates methods for
estimating some of the critical ALWAS in-
put parameters such as the deposition veloc-
ity, settling velocity, scavenging ratio, and
air degradation half-life (assuming quasi-first
order reactions).
The purpose of ALWAS is to assist in the
assessment of water quality problems as-
sociated with toxic chemicals released to the
atmosphere. The combined effect on surface
water quality of toxic releases to several
media — including air, nonpoint source
runoff, direct discharge, and ground water
— may also be evaluated. Analysis of
ground-water impacts is limited because the
user must know the rate and quality of
ground-water recharge to the surface water
body. The model will be particularly useful
in determining the most effective strategies
for controlling water quality effects of toxic
chemicals regardless of the ultimate source
and in evaluating the potential impacts of
new chemicals entering commerce.
The basic flow of the model is from air to
land surface to surface water body, with the
direct transfer from air to surface water body
also accounted for. Atmospheric point
sources (smokestacks) and area sources
(arising from vehicle use, residential heating,
etc.) can be simulated. Wet deposition,
caused by the scavenging of airborne con-
taminants by precipitation, is simulated as
is dry deposition resulting from such pro-
cesses as gravitational settling, impaction,
or dissolution of gases. Gaseous and par-
ticulate contaminants are treated uniformly
by the model. This treatment is consistent
with recent findings that many toxic organic
contaminants that exist as gases at atmo-
spheric pressure and temperature nonethe-
less will behave as particulates with respect
to deposition processes, since the fraction
of the airborne mass that is adsorbed to am-
bient aerosols is primarily responsible for at-
mospheric deposition. Distinction between
contaminants that are primarily participate,
as opposed to those that exist primarily as
a vapor, is specified through the gravitational
settling velocity, which would be zero for
vapors. The dry deposition velocity and the
scavenging ratio apply both to vapors and
particles.
The model is applicable at a wide range
of spatial scales from a calibrated NPS
watershed (limited to about 5 km2) to a major
river basin covering 104 km2. Scaling up to
10* km2 presents technical problems in both
the atmospheric dispersion model (DiDOT)
and NPS. DiDOTs applicability to large spa-
tial scales is limited because one of its
assumptions is questionable at downwind
distances approaching 100 km. The assump-
tion is that the wind is uniform, allowing for
no meandering within the application area.
Even under ideal terrain and meteorological
conditions, the assumption of uniform winds
may only be valid within about 20 km of the
measurement point. If the application area
consists of mountainous terrain or a river
valley, the scale over which the uniform wind
assumption applies may be reduced even fur-
ther. The alternatives to the uniform wind
assumption — multiple station wind data or
solution of the Navier-Stokes equations —
are impractical, however. Both approaches
significantly increase computing expenses
and would require additional field data not
generally available.
ALWAS is a time-dependent model
whose fundamental time step is one hour.
DiDOT and NPSDEP respond at that time
scale whereas EXAMS interfaces with those
two models after temporal averaging over a
user-determined number of days. The time
dependence of DiDOT is different mathe-
matically from that exhibited by NPSDEP
and EXAMS. DiDOT uses a standard air
dispersion approach that assumes that
steady state conditions exist for each hourly
time step. This is possible because the at-
mosphere responds quickly and the con-
taminants are resident in the air domain for
a short period of time. The approximation
is inaccurate under low wind speeds of
variable direction and at very large distances
from the source. The approximation is ac-
curate only if pollutants are emitted to the
model's air domain and transported through
it within the hour and do not return because
of a reversal in wind direction.
On the other hand, both NPSDEP and
EXAMS solve differential equations that
follow the contaminant through the domain
from time step to time step such that cur-
rent conditions depend on the past history
of loading, transport, and transformation.
NPSDEP follows a daily time step on days
without rain and an hourly time step on rainy
days. EXAMS response times are expected
to vary widely, depending primarily on the
size and flushing time, and the user may
specify an appropriate averaging time at in-
tegral multiples of one day for loading
provided by DiDOT and NPSDEP. The ap-
propriate averaging time would be less than
or approximately equal to the contaminant
half-life in the aquatic ecosystem as deter-
mined by an independent run of the steady
state version of EXAMS.
ALWAS is principally designed for organic
pollutants that tend to adsorb to paniculate
matter and are relatively persistent on the
land surface. ALWAS, or specific combina-
tions of its component sub-models, may be
used to assess multi-media effects of a
broader class of pollutants, however, in-
-------
eluding heavy metals or relatively hydrophilic
organics. NPS runoff and subsurface water
quality algorithms do not simulate soluble
constituents. Both DiDOT and EXAMS,
however, account for both soluble and ad-
sorbed contaminants so that these two sub-
models can be used to determine the effects
of direct deposition of hydrophilic organics
in water bodies. This feature is especially
useful for a lake whose surface occupies a
large fraction of its watershed — Lake
Superior, for example.
EXAMS is not presently designed to'simu-
late the chemistry of metals in aquatic
ecosystems. Air dispersion and deposition of
metals, however, is expected to be faithfully
simulated by DiDOT, and NPSDEP is suit-
able for describing land surface processes for
relatively insoluble metals. Thus, DiDOT and
NPSDEP could be used to study metals
loadings to surface water bodies (from air
deposition and runoff) and the relative con-
tribution of various source types to the
overall loading, but ALWAS could not be
used to describe the resulting surface water
quality effects.
Other limitations of ALWAS are principally
related to the memory and operating system
limitations of the POP 11770 computer sys-
tem for which the model has been designed
and the costs associated with handling the
intermediate outputs of DiDOT that must be
transferred to NPSDEP and EXAMS. These
affect the spatial resolution of deposition
rates and area source emissions and the total
number of point sources that may be simu-
lated. These limitations may be overcome
very simply within the inherent structure of
the model by increasing the capacity of
various storage arrays if the model were to
be implemented on a larger computer or one
with virtual addressing capability. The cost
of storing and handling the larger arrays,
however, may limit such a scale-up for many
applications. The resolution limitations of the
software, as it currently exists, may not be
a serious drawback for analysis of many
deposition-related water quality problems.
An inherent limitation of ALWAS is its in-
ability to appropriately account for volatiliza-
tion of pollutants from the land or water back
to the air. Although EXAMS quantifies the
rate of volatilization from the water surface,
this contaminant is not fed back to DiDOT
as a pollutant air emission; it simply disap-
pears from the model. On the other hand,
NPSDEP does not account for volatilization
of pollutants from the land surface. The
reasons for not accounting for volatilization
are several-fold, but principally relate to the
technical problems of incorporating feedback
into an already complex software structure,
particularly in light of the fact that this is a
"first-generation" multi-media model. The
first-order effect on water quality of deposi-
tion of airborne contaminants was of primary
interest in this development effort, rather
than the second-order effect of volatilization
followed by partial redeposition. The limita-
tions suggest that the model be used prin-
cipally for chemicals having relatively low
vapor pressures, although we have not iden-
tified a precise range of vapor pressures over
which the model is appropriate.
ALWAS may be used to study the en-
vironmental impacts of aerial pesticide ap-
plications so long as the drift losses, droplet
size, effective height of release, and area of
application are known. Given this informa-
tion, ALWAS can be used to evaluate hu-
man exposure by inhalation, deposition of
pesticide on non-target areas, and the com-
bined impacts from drift and deposition. Use
for pesticide evaluation may encounter some
problems, however. First, DiDOT describes
phenomena having characteristic length
scales of 103 to 105 meters (1-104km2). On
the other hand, pesticide applications may
only extend over 102 - 103 meters (0.01 - 1
km2). DiDOT cannot resolve the actual resul-
tant spatial variability in pesticide concentra-
tions. ALWAS may be used to simulate
intermittent sources; however, the pro-
cedure would be cumbersome if many inter-
mittent sources were included. This factor,
coupled with the mismatch of length scales,
suggests a reasonable approach to simu-
lating the effects of pesticide application in
a large river basin. Individual pesticide spray-
ing events would not be simulated; rather,
large portions of the basin would be assumed
to be treated on specific days. The model
would not resolve fine spatial variations.
Other, perhaps more serious, drawbacks
to the use of ALWAS for pesticides are in-
herent in NPS, which does not account for
volatilization and does not simulate the sub-
surface movement of soluble contaminants.
Methods are available for estimating drift
losses, droplet size, and effective height of
release for pesticidal input to ALWAS. Sev-
eral investigators have performed field
studies and developed empirical models for
estimating the parameters of droplet release,
and these may be useful for guidance in
developing ALWAS input parameters for ap-
plication to pesticide spray drift and
deposition.
ALWAS has been applied in a hypo-
thetical simulation of benzo(a)pyrene
behavior in a small urban watershed.
Features
DiDOT
DiDOT can handle two contaminants in a
single run. To interface with NPSDEP, one
of these contaminants must be total sus-
pended particulates and the other one may
be any toxic pollutant, either vapor or par-
ticulate. Under the recommended mode of
operation, the model simulates the behavior
of the total airborne contaminant, which may
be partially vapor and partially adsorbed to
ambient aerosol. DiDOT does not explicitly
account for the particle size distribution of
paniculate contaminant, so the deposition
parameters should represent the mass
weighted average of the distribution of the
parameters. Methods for estimating these
parameters are based on the mass median
diameter of the particulate contaminant.
DiDOT's use of meteorological and source
characteristics data is consistent with stan-
dard procedures used in most EPA-sup-
ported air quality models. This feature is
intended to make the model easy to use for
those familiar with standard air modeling pro-
cedures. DiDOT is also consistent with stan-
dard air modeling procedures in its use of
Gaussian dispersion algorithms, plume rise
formulas, ASME dispersion parameters, and
stability classes based on wind speed and
solar radiation. Its method for accounting for
limited mixing under an inversion lid is also
consistent with standard approaches.
DiDOT diverges from standard air quality
models in its deposition algorithms. The
dispersion algorithms used in DiDOT repre-
sent a synthesis and application of current
research and mathematical analysis of the
deposition process. To describe the dry
deposition process from point sources,
DiDOT applies a rigorous analytical solution
that accounts for both gravitational settling
and surface depletion.
The surface depletion model correctly ac-
counts for the fact that pollutants are
deposited on the ground, thus reducing
ground-level concentrations relative to the
values that would be observed if deposition
were not occurring, leading to a vertical gra-
dient of concentration at the surface. The
source depletion model, on the other hand,
preserves the Gaussian shape for the vertical
concentration profile. It accounts for upwind
deposition by adjusting the source emission
rate, i.e., subtracting the surface integral of
the upwind deposition flux from the emis-
sion rate. This approximation adjusts the
source strength artifically by adjusting the
volume of material in the plume in propor-
tion to the concentration, rather than a sur-
face depletion model in which deposition
occurs at the ground. The assumptions of
no deposition or source depletion result in
overestimation of ground level concentra-
tions. For the source depletion model, this
leads to an over-estimate of the deposition
rate as well.
Wet deposition processes are simulated
via a scavenging ratio approach. The basic
-------
assumption is that the rainfall concentration
is proportional to the ground level air con-
centration. Theoretically, several precipita-
tion scavenging processes should not follow
this pattern. These include "in-cloud" scav-
enging or rainout and the irreversible scav-
enging of an elevated plume. Nonetheless,
for many contaminants under diverse me-
teorological conditions, the assumption pro-
ves relatively accurate. The scavenging ratio
for toxic organic pollutants can be estimated
to within an order of magnitude. During rain-
fall, DiDOT does not account for upwind wet
deposition in determining the ground level
air concentration.
Although DiDOT is primarily designed to
simulate dispersion from continuous
sources, it can be adapted to simulation of
an intermittent source such as a chemical
spill, or a pesticide application. This is ac-
complished by performing multiple runs with
a selected subset of meteorological data.
IMPSDEP
NPSDEP is the result of relatively minor
modifications to NPS. NPS is a continuous
simulation model that represents the genera-
tion of nonpoint source pollutants from the
land surface. The model simulates surface
and subsurface hydrologic processes, snow
accumulation and melt, sediment genera-
tion, pollutant deposition, and pollutant
transport for any specified period of me-
teorological data.
NPS inputs include parameters that allow
the user to adjust the model to a specific
watershed. Most of these parameters are
specified by physical (observable) charac-
teristics of the watershed; however, several
of the parameters cannot be determined
from observations, and the model must be
calibrated before application to any
watershed.
NPS permits simultaneous simulation of
five user-specified contaminants. These
contaminants are assumed to migrate with
sediment entrained in surface runoff. Con-
sequently, subsurface water is assumed to
be clean. This constraint limits the validity
of the model to pollutants that are strongly
associated with sediments including hydro-
phobic organics and heavy metals. The con-
taminants are assumed to be conservative
on the land surface, which implies that the
model is most useful for non-volatile, per-
sistent pollutants. For volatile, degradable
contaminants, NPS loadings outputs may be
useful as an upper bound.
NPS allows monthly variations in land
cover, pollutant deposition (accumulation)
and pollutant removal (e.g., by street
cleaning).
The only significant modifications to NPS
that are incorporated in NPSDEP relate to
the deposition processes simulated by
DiDOT. Where NPS allowed monthly varia-
tions in deposition rates, NPSDEP accepts
daily variations. NPS permitted land-use-
dependent ratios between contaminant and
sediment (potency factors) that were as-
sumed to be the same for eroded sediment
and deposited sediment, but NPSDEP allows
two separate potency factors per land use
— one for eroded sediment from the land
surface and one for the deposited sediment
from DiDOT. Monthly variations in the
potency factors are permitted. Thus,
NPSDEP can simulate the combined impacts
of deposited contaminant and some in-
dependent source of nonpoint pollution.
EXAMS
EXAMS describes the transport and
chemical transformation of organic chem-
icals in surface water bodies. It is a multiple
compartment model in which the cells are
linked by advection and turbulent exchange.
Cells may contain mostly water (littoral,
epilimnion, hypolimnion, etc.) or mostly sedi-
ment (benthic), with different input require-
ments. Each cell consists of water, sediment,
and biomass in variable proportions.
Chemical processes are simulated on the
basis of either equilibrium partitioning or first-
order kinetic processes. Equilibrium partition-
ing among a (5 x 3) matrix of possible forms
is established for up to five ionic forms
(valences of -2, -1, neutral, +1, +2) and
between dissolved, sediment adsorbed, and
biosorbed phases. Processes simulated by
first-order kinetics include hydrolysis, oxida-
tion, volatilization, microbial transformation,
and photolysis.
The dynamic version of EXAMS used by
ALWAS determines the dynamic response
to time-varying loading rates.
Inputs
ALWAS requires six basic categories of
inputs — meteorological data, pollutant
source data, pollutant fate properties,
calibration data for NPSDEP, characteristics
of the physical environment (primarily for
NPSDEP and EXAMS), and interfacing data.
Meteorological Data
Each sub-model of ALWAS requires dif-
ferent kinds of meteorological data. Both
DiDOT and NPSDEP require hourly me-
teorological data in formats compatible with
standard reporting formats of the National
Weather Service.
Pollutant Source Data
DiDOT and EXAMS require data
characterizing sources of the pollutant to the
environment. DiDOT treats three kinds of
pollutant sources: point sources, traditional
area sources, and distant city-sized area
sources. Traditional area sources represent
air emissions from transportation, small in-
stitutional and commercial sources, residen-
tial heating or solvent use, land use activities
— in short, all small sources that are too
numerous or otherwise inconvenient to in-
ventory individually. Distant cities are treated
by a virtual point source approach. The data
required to characterize sources in DiDOT
are typical of those required by other air
quality models, and are available in standard
state-maintained emissions inventories.
Pollutant Fate Properties
Chemical specific inputs are required
primarily by DiDOT and EXAMS. DiDOT
uses the settling velocity, deposition veloc-
ity, scavenging ratio, and atmospheric
degradation half-life. These, in turn, may be
estimated using procedures presented in the
report and knowledge of particle size,
Henry's law constant, vapor pressure, and
molecular weight, or other fundamental pro-
perties. EXAMS requires similar inputs,
solubility in water, octanol/water partition
coefficient, rates of hydrolysis, photolysis,
and oxidation.
Calibration Data for NPSDEP
NPSDEP requires calibration data — time
series of water flow and quality at the dis-
charge point of an upland watershed. The
availability of such data is critical to the site-
specific application of ALWAS but may not
be needed for application to generic, "typ-
ical" environments.
Characteristics of the
Physical Environment
NPSDEP requires as input various physical
features of the watershed — >area, length of
overland flow path, slopes, elevation, land
use characteristics, portions of watershed
that are pervious and impervious, soil char-
acteristics, etc.
EXAMS relies extensively on a user-speci-
fied description of the water body, since it
does not solve equations that describe its
hydraulics. If multiple cells are specified, the
user must quantify the advective and tur-
bulent exchanges between cells. This
demands an in-depth knowledge of flow pat-
terns in the water body, which may be
achieved via field studies and/or indepen-
dent numerical modeling using models that
solve the equations of motion. In lieu of such
information, the user may rely on the de-
scriptions of typical aquatic ecosystems con-
tained as a default input data file within the
EXAMS data base.
4
-------
Interfacing Data
The geometrical relationships between the
DiDOT receptor locations (points at which
deposition rates are calculated), NPSDEP
catchments, and EXAMS cells are within this
category. Each of the models has a unique
way of describing specific locations within
its domain, and these are interrelated in a
straightforward fashion.
As should be clear from the foregoing
discussion, the data requirements are exten-
sive, leading to significant costs in data
gathering, formatting, and handling prior to
full implementation of ALWAS. Most of
these inputs, however, are routinely available
from environmental data banks, including
those maintained by EPA, the National
Climatic Center, and the U.S. Geological
Survey.
Outputs
The fundamental ALWAS outputs are the
surface water quality outputs presented by
EXAMS. These include the total contami-
nant concentrations, as a function of time,
in each EXAMS cell, and the partitioning of
that total contaminant among various ionic,
dissolved, sediment adsorbed, and biosorb-
ed forms. DiDOT also presents summary air
quality outputs including long-term average
concentrations over the watershed and the
extreme hourly and daily average concentra-
tions. NPSDEP provides flow, sediment
transport, and toxicant loading outputs at
monthly and annual intervals, as well as short
term output for major storms.
William A. Tucker, Alan Q. Eschenroeder, and Gary C. Magil are with Arthur D.
Little. Inc., Cambridge, MA 02140.
Kenneth F. Hedden is the EPA Project Officer (see below).
The complete report, entitled "Air Land Water Analysis System (ALWAS): A
Multi-Media Model for Toxic Substances," (Order No. PB 84-17J 743; Cost:
$34.00, 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:
Environmental Research Laboratory
U.S. Environmental Protection Agency
Athens, GA 30613
-------
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
BULK RATE
Official Business
Penalty for Private Use $300
Ho UUUU
u o LWI* KK'U [c)- f 1UlV
K I- b 1 U M S L i -i W A K Y
i^iU O L'h AKdUf'''•;• olKti"
Lr'lLAUU 1 L '')U^'J'4
U.S. GOVERNMENT PRINTING OFFICE: 1984-759-102/964
------- |