United States
Environmental Protection
Agency
Environmental Sciences
Research Laboratory
Research Triangle Park NC 27711
Research and Development
EPA-600/S3-84-092 Nov. 1984
&EPA Project Summary
Field Validation of Exposure
Assessment Models
J.C. Doran, O.B. Abbey, J.W. Buck, D.W. Glover, T.W. Horst, R.N. Lee, and
F.D. Lloyd
This report, in two volumes, describes
work done to evaluate the Point, Area,
and Line source model (PAL), a Gaussian
diffusion code modified to account for
dry deposition and settling (the modified
model is designated PAL-DS). The first
volume describes the experimental
techniques employed to dispense,
collect, and measure depositing (zinc
sulfide) and nondepositing (sulfur
hexafluoride) tracers. The measured
concentrations of the tracers form a
data set by which the PAL-DS model
may be tested. These concentrations
are given in tabular and graphic form for
five downwind distances from the
release point, ranging from 100 to
3200 m. Measurements of wind speed,
direction and temperature at seven
heights from 1 to 61 m were taken
during the tracer releases as well as
several parameters describing the
turbulence characteristics. A discussion
of particle size distributions of the
depositing tracer is given, and the
calibration and quality assurance pro-
cedures used for sample analysis are
described. Two chlorocarbon tracers
were also used in the tracer releases,
but with little success. The first volume
concludes with a discussion of the
problems encountered with these
tracers and some recommendations for
possible solutions.
The second volume contains an
analysis of the field data and an
evaluation of four atmospheric disper-
sion models, PAL-DS and three similiar
alternates. The four models are de-
scribed, and an evaluation of the perfor-
mance of each is given. The evaluation
is based on an analysis of Cd/C0, the
ratio of the crosswind-integrated con-
centrations of a depositing and non-
depositing tracer, respectively, at a
height of 1.5 m. The PAL-DS model is
found to overestimate this ratio; a
corrected source depletion model
appears to give significantly better
results. A novel method of determining
the effective deposition velocity of the
depositing tracer, based on a surface
depletion, approach is described. A
discussion of model sensitivities, ex-
perimental design, and the effects of
measurement errors on the model
evaluation is also given. Experimental
uncertainties may well affect the
performances of the models, but it is
doubtful that their relative performances
would be significantly changed. Errors
in describing the diffusion meteorology
are likely to be more important in
predicting depleted concentrations
than errors introduced by the choice of
a particular deposition model.
This Project Summary was developed
by EPA's Environmental Sciences
Research Laboratory, Research Triangle
Park, NC, 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
The basic aim of this work was to
release and sample tracers in the
atmosphere, to obtain data suitable for an
evaluation of EPA is PAL model as
modified for depositing pollutants, and to
perform such an evaluation. An additional
goal was to make some limited releases
of organic tracers and evaluate their
measurement accuracy under controlled
field conditions.
-------
The PAL model is a Gaussian plume
diffusion code for point, area, and line
sources which have been modified to
account for dry deposition and settling.
The resulting PAL-DS code uses diffusion-
deposition algorithms that are based on
analytical solutions of a gradient-transfer
model for dry deposition of gaseous and
suspended participate pollutants from a
plume.
In order to evaluate the PAL-DS model,
three kinds of data are required: the
distribution of some tracer subject to
deposition to the surface, the distribution
of some other tracer not subject to
deposition, and the prevailing meteorology
during the measurement periods. Zinc
sulfide (ZnS) was chosen for the depositing
tracer, while sulfur hexafluoride (SFe)
was used for the nondepositing tracer.
Wind and temperature sensors mounted
on several towers were used to assess
the meteorological conditions during the
tracer releases. Several organic tracers
were also tried, but with little success.
The first volume provides a description
of the experimental techniques employed
to dispense, collect, and measure the
tracers used for the releases. It gives
additional information on the character-
ization of the particle size distributions for
the particulate tracer, the types of
supporting meteorological measurements
obtained, and the calibration procedures
used for the tracer and meteorolgical
measurements. The report is specifically
concerned with various aspects of quality
assurance in the collection and analysis
of the data. Finally, the report contains
listings and figures describing the
principal features of the measured tracer
distributions and the winds and tempera-
tures during the releases.
The second volume presents an analysis
of the data that enables effective deposi-
tion velocities for the depositing tracer to
be determined, and uses these and other
estimates in an evaluation of the principal
features of the PAL-DS model and three
other models. It also presents a statistical
summary of the performance of these
models, in simulating the observed
behavior, gives some suggestions for
interpreting this behavior, and recom-
mends possible additional measurements
and modeling studies that would further
clarify the behavior of depositing materials.
Experimental Procedures
The Hanford diffusion grid, where the
tracer measurements were obtained, is
located in a semiarid region of south-
eastern Washington on generally flat
terrain. The vegetation consists primarily
of desert grasses and 1 to 2 m high
sagebrush.
Five sampling arcs, located at distances
of 100,200, 800,1600, and 3200 m from
the tracer release point, were used for
tracer collection. A122-m meteorological
tower was located approximately 100m
to the north of the release area. It
contained wind speed, wind direction and
temperature sensors at various elevations.
These data were used to monitor condi-
tions during tracer releases in an effort to
ensure that material was released during
periods with appropriate wind and
stability conditions. Near-neutral to
stable conditions were selected to
maximize differences in concentrations
of depositing and nondepositing tracers.
For subsequent, detailed analysis of
the meteorology, data were collected at a
second tower located in the center of the
1600-m arc. That tower was 61-m high,
and was equipped with 3-component
propeller anemometers and aspirated
bead thermistors at seven heights. All
anemometers were located on booms
extending at least 2 tower diameters west
(generally upwind) of the tower. Data
from all sensors were recorded on
magnetic tape approximately once every
2.3s.
A 3-m tower was located about 25 m to
the southwest of the 61-m tower, and
held a 3-component sonic anemometer
and fast-response platinum resistance
thermometer on a boom extending from
its top.
The ZnS was collected on membrane
filters mounted 1.5m above the surface.
The vacuum pumps used to draw air
through the filters were driven with
gasoline engines, and flow through the
filters was regulated by critical flow
orifices inserted just downstream of each
filter holder assembly. Filters were
located at 2° increments on the 100-,
200- and 800-m arcs, and at 1 ° incre-
ments on the 1600- and 3200-m arcs.
Each arc encompassed a sector of approxi-
mately 90°. The SFe tracer was sampled
at intervals of 8° at the 100-m arc, 4° at
the 200- and 800-m arcs, 2° at the 1600-
m arc and 3° at the 3200-m arc.
Multilayer bags were deployed at intervals
of 12° at the 800-m and 3200-m arcs and
8° at the 1600-m arc. Some effort was
also made to measure vertical profiles of
the tracer concentrations, and two
towers at the 1600-m arc were equipped
with filters and pumps to a height of 25 m.
The ZnS particulate tracer was dis-
pensed with an aerosol generator. The
ZnS was mixed with a 1%, by weight,
fluidizing agent. Prior to tracer release.
the mixture was loaded into a hopper on
the generator. Material from the hoppe
was dropped, via a notched wheel, into <
blower assembly, which ejected thi
particles horizontally into the atmosphere
During the release, the ZnS was contin
uously stirred to ensure smooth feedini
of the particles into the blower.
The SFs was dispensed from a 91 -lite
tank that had been previously fi lied from <
SF6 cylinder to a moderate pressure. Th<
flow from the tank was regulated via <
needle valve and rotometer, and the SF,
was released through a length of poly
ethylene tubing. Typical release rates
were on the order of 0.3 g/s.
The organic tracers were releasec
using a pressurized tank and generating
gun assembly. The gun consisted of ar
outer aluminum tube, two ultrasonic
nozzles, and a perforated tube behind the
nozzles that provided a sheath of aii
isolating the organic material from the
outer tube.
All tracers were released from a heighl
of 2 m. The separation between the SFs
and ZnS release points was less than 1 m,
while the distance between the ZnS and
organic tracer release points was about 5
m.
Concentrations of ZnS on the filters
were determined by using a device
developed at Hanford and known as a
Rankin counter. Its essential elements
consist of a weak plutonium source of
alpha particles, which are used to induce
fluorescence in the ZnS particles, a
photomultiplier tube used to detect this
fluorescence, and a sealer used to count
light flashes.
The SFe and organic samples were
analyzed by means of gas chromato-
graphy. Calibrations of these instruments
were obtained using National Bureau of
Standards (NBS) traceable standards.
The background detection limit for SF6
was approximately 15 parts per trillion
(ppt), and for the organic tracers it was
about 0.1 parts per billion (ppb). Levels
below these values were assumed to be
due to background signals alone, and
were not included in the data analysis.
Mean wind speeds, directions and
temperatures are given for each release
period and for an equal period after the
releases so that one may obtain an idea of
the stationarity of the conditions. As
might be expected under stable conditions
with low wind speeds, shifts in speed and
direction were common, and the decision
on the time to release tracers was made
partially on the basis of an educated
guess.
Winds speeds from a three component
sonic anemometer and temperaturesi
from a fast-response platinum resistance
-------
probe were recorded during parts or all of
six of the releases. From these data we
determined values of the friction velocity,
temperature flux, and the Monin-Obukhov
length.
Information is given on the standard
deviations of the horizontal and vertical
wind direction fluctuations, determined
from the propeller anemometers. For a
steady wind direction, the horizontal
standard deviation is useful for predictions
of the lateral plume spread, while the
vertical standard deviation may be related
to the vertical plume spread.
The report contains a brief, narrative
description of conditions under which
each of the six tracer experiments was
conducted. The prevailing meteorology
during the release period, the performance
of the sampling equipment and some
general descriptions of the resulting tracer
distributions are given for the experiments
which occurred during May and June of
1983.
Model Evaluation
The dual-tracer field measurements
have been used to evaluate four Gaussian
plume-depletion models: 1) the commonly
used source depletion model, 2) a
modified source depletion model that
accounts for the alteration of the vertical
concentration profile by deposition, 3) the
PAL-DS model, which is a diffusion-
deposition model based on a constant
eddy-diffusivity solution of the advection-
diffusion equation, and 4) a mass-
conserving version of the PAL-DS model.
These models are briefly described, and
the general characteristics of their
predictions are discussed in relation to
those of an exact model of plume
depletion.
The output of the models is a three-
dimensional field of tracer concentrations.
Measuring such a field in its entirety is
clearly impossible, and some compromise
is required. Moreover, the model output
depends on a number of factors, which
may be roughly divided into two classes:
those describing the diffusion meteorology
and those describing the deposition. In
reality the two sets of factors are related,
but the diffusion aspects of various
models have been evaluated extensively
in a host of studies. There appears to be
little point in emphasizing such analyses
for our limited data set; instead, we focus
on the deposition part of the model. To
evaluate the model's performance, we
assume that the diffusion meteorology is
known or can be determined. What
remains, then, is to evaluate how well the
model describes the depletion in the
plume arising from deposition and
settling. As noted earlier, we chose to
employ a dual tracer technique for this
exercise. The nondepositing tracer was
SF6, while ZnS was the depositing tracer.
Differences in SFe and ZnS concentrations
may be assumed to arise from the
removal of depositing tracer at the
surface, and the model description of this
process may thereby be evaluated. Thus,
the basic quantity to be discussed is the
ratio Cd/Co, where Ca is the crosswind-
integrated concentration of depositing
tracer 1.5 m above the surface and C0 is a
similar crosswind-integrated concentra-
tion of non-depositing tracer. In practice it
was most convenientto measure concen-
trations of both tracers at some height
close to the surface (1.5 m), where Cd-C0
is largest, rather than over an extended
vertical extent.
The predictions of four models were
evaluated. Comparisons of the predicted
and measured values of Cd/C0 were
made. The calculated ratios of Cd/C0
were determined by using the effective
deposition velocities in each of the four
models, for all six dual-tracer releases.
The observed ratios are those actually
measured in the field tests. The PAL-DS
model showed the largest differences
between observed and calculated values,
while the corrected source depletion
model shows the best agreement. The
uncorrected source depletion model
distributes the effect of deposition
uniformly through the vertical extent of
the plume, rather than preferentially
depleting the plume at ground level, and
thus overestimates the near-surface
concentration ratio. Much better results
are given by the corrected source
depletion model, which accounts for the
change in the vertical distribution caused
by deposition.
Conclusions and
Recommendations
The PAL-DS model was found to
overestimate consistently the amount of
depositing pollutant at a height of 1.5 m.
A corrected version of this model, which
incorporates a modification that ensures
conservation of mass, performed better,
while a corrected source depletion model
gave the best results of the four models
tested. In all cases uncertainties in
describing the diffusion meteorology are
apt to be more important for estimating
depleted air concentrations than differ-
ences arising from the choice of a
particular deposition model. However, of
the four models tested, the use of the
corrected source depletion model appears
to offer the best chance of avoiding
systematic errors in the description of
plume loss arising from deposition.
Because a number of uncertainties in
the tracer concentration measurements
and their analyses may have adversely
affected the model evaluations, statistical
comparisons of model performance
should be viewed with some caution.
However, the relative performances of
the four models seem unlikely to be
seriously affected by these factors.
Three additional tasks are suggested
that would provide greatly increased
confidence in model evaluations such as
those just described and could result in
better quantitative comparisions of
models with each other or with theory.
While relatively simple to state, each task
is a major research effort in itself.
The first task is the development of a
tracer generator capable of producing
mono-dispersed or nearly mono-dispersed
particles in the size range 1-5 um
diameter. The generator should be able to
produce a quantity of material sufficient
for detection at a range of 10 km or more.
The tracer itself should be easily collected
and analyzed. Such a tracer's size
distribution could be more easily deter-
mined and non-isokinetic sampling
effects could be estimated with better
accuracy. Sample loss close to the source
caused by the preferential depletion of
larger size particles would be lessened,
and the range of meteorological conditions
in which the surface depletion method
could be profitably used to determine
deposition velocities would be increased.
The second task is a rigorous test of the
surface depletion model of dry deposition,
using this newly developed tracer tech-
nique. This task would involve a direct
determination of the deposition velocity
of a depositing tracer, e.g., by collecting
material actually deposited on the
ground, and the measurement of airborne
concentrations of the depositing material
close to the surface at a number of
downwind arcs. These measurements
could be done at relatively short distances
from the tracer release point, on the order
of 1 km or less.
The third task would be a more
extensive evaluation of models such as
the PAL-DS or corrected source depletion
model. In particular, the distance to
which concentration measurements are
made should be increased to 10 km or
more, in order to increase the range of
values of plume loss that the models are
to simulate. If a tracer generating technique
such as that described above is used,
substantial improvements in model
evaluations should be possible.
-------
J. C. Doran, 0. B. Abbey. J. W. Buck. D. W. Glover, T. W. Horst, R. N. Lee, and
F. D. Lloyd are with Pacific Northwest Laboratory, Richland, WA 99352.
Jack S href tier is the EPA Project Officer (see below).
The complete report consists of two volumes, entitled "Field Validation of
Exposure Assessment Models:"
"Volume 1. Data,"(Order No. PB 85-107 209; Cost: $17.50)
"Volume 2. Analysis,"(Order No. PB 85-107 217; Cost: $10.00)
The above reports will be available only from: (costs subject to change)
National Technical Information Service
5285 Port Royal Road
Springfield. VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Environmental Sciences Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
-tt US GOVERNMENT PRINTING OFFICE. 1984—559-016/7844
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
BULK RATE
POSTAGE & FEES PA
EPA
PERMIT No. G-35
Official Business
Penalty for Private Use $300
IST
EHICAGU
------- |