United States
Environmental Protection
Agency
Environmental Monitoring
Systems Laboratory
Las Vegas NV 89193-3478
Research and Development
EPA/600/S4-87/044 Feb. 1988
Project Summary
Evaluation of Existing Total
Human Exposure Models
Muhilan D. Pandian
Modern technology has brought a
dramatic increase in the production
and consumption of chemicals. The
public has become increasingly
aware of the ability of some
pollutants to cause unexpected
results at some point far removed
from where they were emitted. This
awareness has generated two
important questions which have and
will continue to motivate much of the
research in environmental science:
1) What is the expected environ-
mental concentration-time profile
for a pollutant at specific locations in
various media? 2) What are the
hazards to man and his environment
resulting from these pollutant
concentrations?
Due to the magnitude and
complexity of the problems
confronted in answering these
questions, researchers have resorted
to modeling techniques to help
overcome some of those problems.
In this report, a special class of
models is examined and several
existing formulations are compared.
These models utilize pollutant
concentration distributions and
human time-activity patterns,
methods of matching concentrations
and activities, the number of
pollutants which can be handled,
accommodation for short-term
(acute) or long-term (chronic) ex-
posures, treatment of uncertainties
and errors in the modeling
techniques. Models are also
compared to ascertain whether the
computer program itself is well
written, modular, and user friendly
with simple input of variables and
clear, understandable outputs.
This Project Summary was
developed by EPA's Environmental
Monitoring Systems Laboratory, Las
Vegas, NV, to announce key findings
of the research project that is fully
documented in a separate report of
the same title (see Project Report
ordering information at back).
Introduction
Modern technology has brought a
dramatic increase in the production and
consumption of chemicals. In some
cases, the benefits from the use of these
chemicals have been accompanied by
unexpected adverse effects. For
example, the mercury contamination of
fresh water (Gilbertson, 1974), the
widespread distribution of PCBs
(Gustafson, 1970), and the alleged
destruction of the ozone layer in the
stratosphere due to the release and
accumulation of chlorofluorocarbons
(Molina and Rowland, 1974) have made
the public increasingly aware of the
ability of some pollutants to cause
unexpected results at locations far
removed from the point of release.
Two important questions have been
generated as a result of this increased
awareness (Neely and Blau, 1985):
1) What is the expected
environmental concentration-
time profile for a pollutant at
specific locations in various
media?
2) What are the hazards to man
and his environment resulting
from these environmental
concentrations?
Due to the magnitude and complexity
of the problem, researchers have sought
to develop modeling methods in order to
answer these questions. This report
-------
examines a specific class of models;
those which utilize environmental
pollutant concentration distributions and
human activity patterns to estimate
human exposure. The specific models
chosen for evaluation are:
1) AERAM (Air Emission Risk
Assessment Model)
2) HEM (Human Exposure Model)
3) LIFETIMEI (Lifetime exposure
model based on SHED {S.A.I.
Human Exposure and Dosage
Model})
4) NEM (NAAQS {National Ambient
Air Quality Standards} Exposure
Model)
5) SHAPE (Simulation of Human
Air Pollution Exposure)
6) SuperSHEAR (An extension of
SHEAR {S.A.I. Human Exposure
and Risk})
Ott (1984), defines "exposure" to a
pollutant as the joint occurrence of two
events: a) person i is present at some
location L with coordinates (x,y,z) at time
t, and b) concentration c is present at the
same location L at time t:
(Person i is present at location L at
time t) (Concentration c is present at
location L at time t).
To calculate exposure in such a
fashion one needs information on the
spatial-temporal distribution of the
concentration c(x,y,z,t) and the
coordinates of person i, (x,y,z) in relation
to time t.
When person i is present at location L,
he/she is enveloped by one or more of
the pollutant-carrying media - air, water
and food. Pollutants in air can cross the
human envelope on inhalation followed
by air-exchange in the lungs and/or on
contact with the dermal layer. Pollutants
in water can be absorbed by the human
body by oral consumption and/or dermal
contact. And pollutants in food,
especially those biological in nature, can
be digested by the human body through
oral consumption.
The total human exposure models
formulated thus far by the research
community studying exposure to
environmental pollutants, have focused
primarily on airborne pollutants.
Consequently, this report has a similar
focus. Some of the models go beyond
exposure, and estimate dose and even
risk.
The evaluation process emphasizes
the manner in which the concentration
distributions of the pollutants, the
distributions of the exposed populations,
exposure to the pollutants, estimation of
dosage after exposure, and the risk
assessment techniques are handled by
the models. Since all the models
evaluated, which calculate pollutant
dispersion, do so only through the air
medium, the models were characterized
by their atmospheric modeling
capabilities. Mention is made where the
modular structures which represent the
other pollutant carrying media might
easily be added to the existing program
structure. Special consideration was
given to models which include pollutant
concentrations in indoor environments.
The population characterization
function was evaluated according to the
level of divisioning of the study area.
Models with large population distributions
might include factors to accommodate
sensitive populations or population
migration patterns and population growth,
and models with individuals or small
groups of people might include
microenvironments and human time-
activity patterns. The methods used by
these models to match pollutant
concentrations with population
distributions was also evaluated.
The model evaluation process also
involved characterizing other areas as
well. These additional factors are:
1) Whether deterministic or
stochastic procedures are used
by the various functions;
2) If the exposure model can
handle more than one pollutant;
3) If the exposure model can
accommodate both short-term
(acute) and long-term (chronic)
exposure analyses;
4) If uncertainties and errors in the
modeling techniques are
identified and adequately
explained; and
5) If the computer program itself is
well written - it should be
modular in structure allowing for
the easy inclusion of external
subroutines, facilitate the simple
input of variables, output results
in a clear and understandable
fashion (using tables and graphs
as appropriate), be compatible
with mainframe and personal
computers, be user-friendly,
and carry simple and easy-to-
read documentation.
AERAM
AERAM is an exposure/risk model
formulated to estimate the effects of
hazardous pollutants from coal-fired
power plants. Flyash and gaseous
pollutants emitted from the power plant
stack, which is treated as a point source,
are analyzed for their transport through
the atmospheric environment (air
medium) and subsequent exposure
known population distributions.
The modular structure of AERA
contains four divisions. The dispersic
model ISCLT (Industrial Sourc
Complex-Long Term) is used 1
determine the fate of the pollutant
AERAM cannot handle pollutants fro
line, prototype, or area sources, neithi
can it handle complex terrain surroundir
the source. By using meteorologic
information from STAR (Stability Arra
data, the pollutant concentrations ai
obtained at user specified location
Indoor environments are not considered
The population is characterized at tf
census tract level by using the U.:
Bureau of Census data. User specific
cohort groups can also be used
estimate exposure of specialized <
sensitive populations. Exposure is n
estimated at the individual level; thu
AERAM does not require any hums
time-activity patterns in the analysi
The annual exposures estimated ai
extended for cancer risk assessment.
lifetime period is assumed for the ur
risk factor and the option for using one
the three dose-response functions - or
hit, log probit, multi hit - is availabl
Depending upon the availability of anim
health data for long-term exposures f
the pollutant of interest, the ris
assessment module estimates tr
necessary parameters and calculate
cancer risk.
Since AERAM has a moduli
structure, external algorithms can t
easily added to include air dispersic
models similar to ISCLT and eve
models representing other medi
AERAM is written in Fortran and can t
used on both mainframe and person
computers.
HEM
HEM estimates average annu
ambient pollutant concentrations by usir
a modified version of the atmospher
dispersion model CDM (Climatologic
Dispersion Model). CDM which simulati
major point source emissions
contained within the HEM program. In tr
modified version of CDM, the dispersic
algorithm simulates the dispersic
processes and reports resulting polluta
concentrations in 16 wind directions at '
radial distances from a point source. Th
model can simulate a number of sourc
types, but is limited when short-ter
analyses, complex terrain consideration
industrial source complexes, and area
prototype sources need to t
considered. There is also an option
-------
HEM where the dispersion algorithm in
the EPA's SHEAR model can be
accessed to determine pollutant
concentrations.
HEM uses U.S. Bureau of Census
data to determine the total population for
each BG/ED (Block Group/Enumeration
District) in the study area. The total
population reported for each BG/ED is
assumed to be located at the population
centroid of the BG/ED; thus each record
within the data file contains population
and latitude/longitude values associated
with the centroids.
The exposure methodology matches
pollutant concentrations with population
centroids and determines exposure to
the related population. Two matching
schemes are employed in HEM, one for
population centroids relatively near the
source (typically within 3.5 km) and and
another for those distant from the source.
Three measures of exposure are
computed: maximum individual
exposure,exposure by concentration
level, and aggregate exposure. Maximum
individual exposure represents the
highest concentration to which any
individual is exposed. Exposure by
concentration level is a measure of the
number of people exposed to a
concentration greater than or equal to a
specified level. Aggregate exposure
reflects the exposure for the entire
population within the study area and is
computed by summing the exposures by
different concentration levels.
Some of the shortcomings of the HEM
include the fact that only the air medium
is considered, the indoor environment is
totally ignored, the dosage received is
assumed to be equal to the exposure,
and only cancer risks are estimated.
LIFETIME1
LIFETIME1 is an experimental model
formulated for calculating lifetime to toxic
pollutants. The model is built around the
SAI Human Exposure and Dosage model
(SHED) with the inclusion of two dynamic
factors, namely migration of people
between regions and the use of indoor
environments. The LIFETIME1 exposure
model is structured to:
1) Include census sets to establish
pollutant receptors;
2) Use actual populations
associated with these receptors;
3) Use the industrial source
complex (ISC) model to estimate
air quality at these receptors;
4) Establish a migration pattern in
the analysis; and
5) Use a simple lifetime exposure
algorithm.
The special characteristic of the
LIFETIMEI exposure model is that it
accounts for population migration into
and out of areas affected by a toxic
pollutant and calculates exposures over a
70-year period. The lengths of
residence in exposure districts and the
resulting year-to-year variations in
actual exposure are factors considered
by the model. Simple and complex
migration models have been developed.
A simple transformation is applied to
convert averaged outdoor concentrations
to concentrations in indoor environments.
More than being a tool for exposure
analyses, the LIFETIMEI is a conceptual
model which embodies valuable
information for improving existing,
working total human exposure models
and developing more accurate newer
models. The LIFETIMEI exposure model
assists in better understanding the
population characterization and exposure
estimation modules. Mechanisms to
include population migration and growth
patterns in the exposure analysis are
detailed in the formulation. The
alternative exposure indicator suggested
includes the time variable which better
explains exposure patterns in different
periods of time. The simple treatment of
indoor environments where translation of
outdoor concentrations of a pollutant to
indoor concentrations are obtained using
proportionality factors does not
adequately represent the variations in the
different indoor environments.
NEM
The NEM computer program is written
in programming language PL1 for
execution on the UNIVAC at EPA's
National Computer Center (NCC),
Research Triangle Park, North Carolina.
The program contains certain
characteristics and compiler signals that
are specific to UNIVAC's PL1 compiler.
With the necessary modifications, NEM
can be executed on any mainframe
computer with a PL1 compiler.
NEM uses pollutant concentration
data from monitoring stations that are
located in the study area. Concentration
levels in indoor environments are
obtained by determining indoor/outdoor
ratios from a linear model. The indoor
environment, being a complex
environment controlled by many
variables like air exchange, building
design, indoor sources and sinks, and
human activities cannot be easily
represented by a linear model.
The population characterization
module in NEM effectively determines
the time-activity patterns of the
individuals involved in the analysis. The
population is divided by age, occupation,
and commuting styles and their activities
characterized by an exercise level.
Exposure is estimated in terms of
(concentration) x (population).
The different versions of NEM, each
handling a different pollutant, have been
developed sequentially. This process
shows advancements in the later
models. To avoid much redundancy,
NEM might be developed modularly, with
those subroutines which are pollutant-
dependent being external to the main
program.
SHAPE
SHAPE focuses mainly on the pop-
ulation characterization, exposure, and
dosage estimation modules in the entire
total human exposure modeling process.
Carbon monoxide (CO) concentrations,
both microenvironment and background,
are user-specified inputs. Therefore,
SHAPE does not include dispersion
models to estimate CO concentrations.
Demographic variables of randomly
selected individuals are determined from
probability distributions, which are also
user-specified.
With the exception of the calculation
of carboxyhemoglobin in the blood
stream due to CO inhalation, all other
estimations in SHAPE are stochastic in
nature. Once a hypothetical individual is
randomly selected for the exposure
simulation process, his/her time- activity
pattern is obtained from statistical
considerations. The individual's daily
routine is regulated to a certain extent,
however; for example, a person who
travels to work by bus is assumed to
return back from work also by bus.
SHAPE was formulated specifically to
estimate exposure and dosage to CO.
The microenvironments were chosen to
reflect the sources and prevalent areas of
CO. The Coburn equation which
describes the phamaco-kinetic behavior
of CO across the air exchange region in
the lungs cannot be extended to explain
the process of absorption of any other
pollutant into the human body. All these
factors suggest that SHAPE cannot be
used for modeling exposure to any
pollutant other than CO, even in any
modified form. However, the concepts
involved in SHAPE could be successfully
used to similarly model other pollutants.
The FORTRAN program written in the
SHAPE formulation is modular in
structure and allows for easy additions of
external subroutines. Relevant dispersion
models, either deterministic or empirical,
can be included to represent more than
-------
just the few urban areas that have been
used in previous analyses.
The use of microenvironments and
human time-activity patterns
characterizes the finest details one can
accommodate in a population distribution
module. SHAPE effectively uses this
concept and tracks down an individual's
activities during his/her daily routine. Due
to the lack of available data, only one
occupational category - working male or
female - is included in the program
structure and more groundwork will have
to be laid before additional occupational
categories are considered. Also, no
special attention is focused on sensitive
populations. The number of
microenvironments presently used can
be increased to cover more areas with
varying CO concentrations and diverse
human activities. This will be particularly
necessary when other occupational
categories are considered in the
modeling process.
SuperSHEAR
SuperSHEAR is an extension of
SHEAR (Anderson and Lundberg, 1983).
It produces estimates of the atmospheric
dispersion profiles and patterns of
residential population in the area
exposed to the hazardous pollutants of
interest. The model produces estimates,
for each source and species, or for
aggregations thereof, of concentration,
number of people exposed at each
concentration level, the mass of pollutant
respired by the population, the health
risk per person exposed at a point, and
the aggregate health risk for the exposed
population.
SuperSHEAR does not require
meteorological or population data as
input because complete national files of
formatted data are elements of the
program itself. The meteorological data
consist of matrices of long-term joint
frequencies of the occurrence of specific
values of wind speed, wind direction, and
atmospheric stability.
SuperSHEAR can particularly handle
air dispersion modeling of non-criteria
pollutants from area or prototype
sources. Multiple sources within an area
can be accumulated to determine their
additive effects or modeled individually.
Pollutant concentration levels are
determined as long-term averages for
sources located only in flat terrain. No
indoor environments are considered.
The population in the study area is
characterized by using data files from the
U.S. Bureau of Census. Exposure is
estimated after assigning outdoor
pollutant concentrations to the population
centroids of BG/EDs. The estimated
annual average exposures are used to
determine cancer risk during a lifetime.
Other Models
The report also briefly describes a few
other models that have been formulated
to estimate total human exposure to
environmental pollutants, including:
1) GEM/GAMS (Graphical Expo-
sure Modeling System/GEMS
Atmospheric Modeling Sub-
system);
2) IBM (Inhalation Exposure
Methodology);
3) PAQM (Personal Air Quality
Model);
4) RIVAD (Regional Impacts of
Visual and Acid Deposition).
Conclusions
All the existing total human exposure
models cover only certain aspects of the
entire modeling process and none of the
models conclude with an error analysis
or much-needed validations. The
exposure limits - lower and upper
estimates - predicted by a few models
are not statistically determined
uncertainties but values obtained from
different population treatments for the
same exposure levels. None of the
models consider any pathway other than
air to determine the concentration of
pollutants in and around the receptor
points. The exposure models which
consider microenvironments and human
time-activity patterns, SHAPE and the
different versions of NEM, are pollutant
specific and have been formulated in
such a way that inclusion of any new
pollutants to the exposure analysis will
require additional effort.
References
1. Gilbertson, M., (1974): Seasonal
Changes in Organochlorine
Compounds and Mercury in Common
Terms of Hamilton Harbor, Ontario,
Bull. Environ. Contam. Toxicology.,
12, 726.
2. Gustafson, C.G., (1970): PCB's
Prevalent and Persistent, Environ. Sci.
Technol., 4, 814.
3. Molina, M.J., F.S. Rowland, (1974):
Stratospheric Sink for Chlorofluoro-
methanes: Chlorine Atom Catalyzed
Destruction of Ozone, Nature
(London), 249, 810.
4. Neely, W.B., G.E. Blau, (1985):
Introduction to Environmental
Exposure from Chemicals, In
Environmental Exposure from
Chemicals, CRC Press, Inc., Boca
Raton, Florida, 2.
5. Ott, W.R., (1984): Exposure Estimate
Based on Computer Generate
Activity Patterns, J. Toxicol.--Clii
Toxicol., 21 (2), 97-128.
6. Anderson, G.E., G.W. Lundber
(Systems Applications, Inc., 1983
User's Manual for SHEAR, Prepare
for the Environmental Protectio
Agency, Research Triangle Park, NC.
-------
r i -; u
Muhilan D. Pandian is with the University of Nevada, Las Vegas, NV 89154.
Joseph V. Behar is the EPA Project Officer (see below).
The complete report, entitled "Evaluation of Existing Total Human Exposure
Models," (Order No. PB 88-146 8401 AS; Cost: $14.95, 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:
Environmental Monitoring Systems Laboratory
U.S. Environmental Protection Agency
Las Vegas, NV, 89193-3478
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
Official Business
Penalty for Private Use $300
EPA/600/S4-87/044
ps
. GOVERNMENT PRINTING OFFICE: 1988—548-013/87C
------- |