OCR error (C:\Conversion\JobRoot\000002QN\tiff\200073BV.tif): Unspecified error
-------
EPA-905/4-87-001
GLNPO No. 87-03
March 1987
DESIGN OF A GREAT LAKES ATMOSPHERIC INPUTS AND
SOURCES (GLAIS) NETWORK
Thomas J. Murphy
DePaul University
Edward Klappenbach
Project Officer
Grant Mo. R005818-01
U.S. Environmental Protection Agency
Great Lakes National Program Office
230 South Dearborn Street
Chicago, Illinois 60604
-------
DISCLAIMER
This report has been reviewed by the Great Lakes National Program Office,
U.S. Environmental Protection Agency, and approved for publication.
Approval does not signify that the contents necessarily reflect the
views and policies of the U.S. Environmental Protection Agency nor
does mention of trade names or commercial products constitute endorse-
ment or recommendation for use.
-------
FOREWORD
The Great Lakes National Program Office (GLNPU) of the United States
Environmental Protection Agency was established in Region V, Chicago
to focus attention on the significant and complex natural resource
represented by the Great Lakes.
GLNPO implements a multi-media environmental management program drawing
on a wide range of expertise represented by Universities, private firms,
State, Federal and Canadian Governmental Agencies and the International
Joint Commission. The goal of the GLNPO program is to develop programs,
practices and technology necessary for a better understanding of the
Great Lakes system and to eliminate or reduce to the naximuni extent
practicable the discharge of pollutants into the Great Lakes system.
The Office also coordinates U.S. actions in fulfillment of the Agree-
ment between Canada and the United States of American on Great Lakes
Water Quality of 1978.
-------
TABLE OF CONTENTS
Page
I. CONCLUSIONS
II. INTRODUCTION 1
III. SAMPLING 2
General 2
Quality Assurance 3
Precipition Gauging 4
Rain Gauging 4
Snow Gauging 6
Precipitation Sampling 6
Rain Sampling 6
Wet Loadings 8
Snow Sampling-Collectors 10
-Cores 11
Dry Deposition 12
Deposition Velocity 14
Deposition Collectors 15
Micrometeorological Methods 16
Vapor Exchange 17
Macro Methods 18
Surrogate Compounds 19
Accumulation Methods 19
Source Determination 20
identification of Source Regions 21
Emission Data 21
Regional Air Monitoring 22
IV. SITING 22
Climatology 22
Over Water Measurements 25
Location of Sampling Sites 27
Master Station 29
V. MATERIALS TO 8E DETERMINED 30
Organics 31
Metals 31
Mercury 31
Nutrients 32
VI. ACKNOWLEDGMENTS 32
VII. REFERENCES 33
-------
APPENDIX A - List of Possible Equipment for a GLAIS Site.
APPENDIX B - Summary of Siting Information from GLAD Sites.
1. Site Location Data
2. Siting Criteria Data
3. Map showing distances (km) separating GLAD and GLPN sites
List of Figures Page
1. Rain gauge catch efficiency vs. wind speed 5
2. Airflow pattern over an open cylinder 5
3. Snow collection efficiency vs. wind speed 7
4. Snow gauges 7
5. Areal distribution of water from a
convective raincell 9
6. Raincells during a rain period 9
7. Structure and details of a typical lake
breeze 23
8. Ecoregions of the contenninous U. S. 28
-------
I. CONCLUSIONS
1. Since any useful GLAIS program will of necessity, be complex and
expensive, the objectives of any such project must be well thought out
in advance, realistic and firmly based on good science and available
technology.
2. The accuracy, precision and validity of the measurements will have to
documented with a thorough quality assurance program to make their
determination worth while.
3. There are at least five distinct mechanisms for inputs of materials from
the atmosphere to bodies of water. They are: rain; snow; small
particles «~2 urn); large particles (>-2 urn); and vapor exchange. In
general, very different methods and techniques are necessary to
determine the inputs by each of these mechansims. Thus any program to
determine total atmospheric inputs will, of necessity, be complex.
4. A program to measure the atmospheric inputs of toxic materials to the
Great Lakes will be significantly different from most other atmospheric
inputs programs now in operation. The reasons are that the other networks
are focused primarily on the measurement of materials related to acid
deposition, and they are measuring the inputs to land and vegatative
surfaces. The GLAIS network will require different methods and
techniques to collect the information on toxics, and to determine the
inputs to large water surfaces.
5. In order to obtain information on the sources to the atmosphere of
the materials going into the Lakes from the atmosphere, it will
require a different set of measurement techniques than are used to
obtain atmospheric inputs from the atmosphere.
6. Because of the expense in setting-up and operating (including analyses)
a monitoring site for a variety of parameters involved with different
mechanisms and/or also obtaining information on sources, a GLAIS
network may consist of a relatively few sites.
7. A useful GLAIS network should be expected to evolve with time as the
ability to determine specific components is acquired and improved.
However, the procedures should be as unchanged as possible to permit
measurements made over several annual cycles to be compared.
-------
8. Because of the uncertainties in our capability to determine inputs
occurring by any of the five above mechansims, any project to make a
reasonable attempt to measure atmospheric inputs by one or more of the
operating mechanisms will need some research component.
9. To be successful, expertise in a number of disciplines will be needed to
develope a good network plan. These disciplines include: aerodynamics;
chemistry; climatology; cloud physics; geochemistry; hydrodynamics;
meteorology; and micrometeorology.
10. If the GLAIS network is to be designed and operated in a productive and
efficient manner, it has to be kept abreast of the state of the art. It
therefore must maintain contact with other networks in U. S., Canada and
abroad.
VI
-------
II. INTRODUCTION
This report is a document to aid in the design of a network aimed at
determining the identity, amounts, and sources to the atmosphere, of the
materials coming into the Great Lakes from the atmosphere. It is meant to
discuss what is known and what needs to be known about sampling atmospheric
inputs to lakes. It does not strive to be thorough and comprehensive as this
is not possible for such a broad subject in a short report. It is based on a
variety of articles and reports, and relies heavily on suggestions and
discussions with a variety of investigators in different disciplines. It
assumes that experts in each of the necessary disciplines will be consulted or
will participate in the final network design. Finally, it reflects the
strengths, weaknesses, biases and misspellings of the author. Discussion
among others is needed to fill it in and to round it off.
Familiarity with the literature in the area of atmospheric inputs or the
reading of this report should lead one to conclude that our present abilities
to determine atmospheric inputs to large bodies of water are limited. Some of
the information desired is not able to be obtained at this time with a useful
degree of precision and accuracy. If it is decided to implement a program to
determine atmospheric inputs into the Great Lakes and/or their sources to the
atmosphere, it must be a realistic one. The specific objectives of such a
program will have to be carefully chosen so that they have a reasonable chance
of being achieved.
The principal decision which needs to be made is the level of confidence
desired in the results. As is discussed in this report, some measurement are
able to be made with greater precision and accuracy than are others. For
instance, estimates of rain inputs to the Great Lakes might be ±50%, while
estimates of vapor exchange will probably be ±300%, and overall atmospheric
inputs could be ±100%. If such estimates turn out to be valid, it has to be
decided what level of confidence is needed for each type of measurement in
order to justify the costs involved in obtaining it. It has to be recognized
that there may be some parameters that would be most desirable to determine,
but there is no reliable method of doing so at the present time.
The determination of atmospheric inputs is still very much in the
development stage. This is particularly true for the toxic materials
proposed to be measured in a GLAIS network. This will be a pioneering
effort, with all of the hardships that suggests. There are not well proven
techniques for making many of the the measurements that are needed. Thus,
any atmospheric inputs plan that is implemented will necessarily have some
developmental and/or research component associated with it.
-------
In order to be useful, there must be a committment to operate a GLAIS
network for several years. The principal reason for this is the great
variability in the climatology. There are cold winters and warm winters, wet
summers and dry summers, early springs and late springs, etc. Several year's
data are needed to get a reliable idea of what the means and the annual
variability are in the atmospheric inputs of the materials sampled for. Longer
term sampling will be necessary to be able to discern trends in the data.
There are presently a number of other operational networks world-wide,
involved with determining inputs from the atmosphere. Also there have been a
large number of studies related to the determination of atmospheric inputs.
The experiences and expertise gained inthose studies and by the other networks
should be beneficial in planning and operating a GLAIS network. But this help
has to be sought in the design and development stage.
Two important difference have to be kept in mind, however. The first is
that the focus of a Great Lakes atmospheric inputs program is on the inputs to
a water surface. Most of the studies and programs of the other networks to
date, have dealt with inputs to land and vegetative surfaces. The meteorology
over the Lakes is often different from the surrounding land areas, the lakes
affect the regional climatology, the chemistry of the surface is different,
and the ease and details of locating sampling equipment is different.
Secondly, intercomparisons with other networks will have to be done
carefully as they all do things differently. Some of them only determine wet
inputs, some determine wet and dry and some, NADP for instance, determine only
soluble inputs. A goal of a GLAIS network, will be to determine total inputs.
For some materials, the inputs of soluble or reactive forms may also be
desired.
III. SAMPLING
General
Most of the problems in determining atmospheric inputs are related to
the wind. The cause of the problems is that the particles are suspended in
the air, they are a second phase. The movement of particles in the air
is determined by several factors: being more dense than air, gravity
attracts them to the earth's surface; air molecules collide with them,
changing their velocity; and being suspended in the air, they tend to move
with the air currents, the winds. The relative importance of these forces
on particles is dependent principally on the particle size.
-------
For large particles, like rain drops and particles larger than about 2
micrometers (urn), gravity is the dominant factor. However as is apparent on
windy days, the direction of fall of even very large particles like rain drops
can be affected by the wind. This usually does not greatly affect the
collection of rain, and the collection of rain is the least complicated of the
different atmospheric inputs.
With small particles, those less than 1 urn in diameter, the effect of
gravity is small with respect to aerodynamic forces. Turbulence is the factor
that most affects their deposition. Their velocity at any one time is almost
as likely to be away from the earth as towards it. This greatly complicates
their collection.
Snow flakes, while often as massive as rain drops, have a large surface
area. This permits the aerodynamic forces to play a more important role,
making snow collection much more difficult than rain and more like that of
particles.
A program to determine atmospheric inputs to the Great Lakes would
involve the collection of particles, droplets, flakes and possibly molecules
in the vapor phase from the air. The quantitative sampling of most materials
in the air, especially those associated with particles, is difficult. Some of
the specific methods which have been used, and some of the difficulties
associated with them will be discussed below.
The large variability of atmospheric inputs complicates the interpretation
of the results from all precipitation networks. Most studies have found that
the concentration and input statistics follow log normal distributions. High
concentration or loading events are infrequent, but expected. It is commonly
found that 10-15% of the annual loadings of most materials are contributed by
'\% or fewer of the events. Thus the quality assurance program for a GLAIS
network has to be structured so that these infrequent, large events are not
summarily rejected.
QUALITY ASSURANCE
Sievering et al. (undated) and other discussants indicated that a
thorough and complete quality assurance program will be REQUIRED for any GLAIS
network. Such a program will have to be integrated into the GLAIS design. It
will involve: determining the limit of detection, the quantification level,
and the precision for all the analytical methods; the use of lab standards and
reference standards to. determine the accuracy of each,of the measurements;
determining blank levels for the sampling equipment and sampling system, for
the sample containers and shipping system, and for the laboratory reagents,
-------
equipment and analytical system; defining the validity of the measurements by
using standard operating proceedures, carefully documenting them, and
identifying deviations from them (Gertler et al. 1985). With a good quality
assurance program, each measurement made should yield a value with a known
precision, accuracy and validity.
PRECIPITATION GAUGING
Knowledge of the amount of precipitation (rain + snow + sleet + hail)
which falls at a collection site is useful for checking the relative
collection efficiency of the samplers and for estimating and comparing the
annual rainfall with other years at that station. In order to obtain the
needed results, careful consideration must be given to the collection of
precipitation and the interpretation of the results obtained.
Rain Gauging
The measurment of rainfall is among the best developed sampling
techniques. "Standard" eight inch, cylindrical, manual-reading raingauges are
frequently used in deposition networks to determine the amount of rain that
falls. These are about the most accurate gauges available for rain
measurements, but are biased low by 10% percent or more in high winds (Larson
and Peck 1974, Fig 1). Recording raingauges, as the ttelfort unit used by the
GLAD network, have an additional low bias due to updrafts around the opening
caused by the sloping side near the top of the gauge (Jones 1969). The
,principal problems remaining in accurately determining rainfall amounts are in
selecting representative sites (sites which are not anomalously affected by
hills, trees, buildings, etc.) for the gauges and in properly shielding the
gauges from the winds.
A lot of work has been done on the remote sensing of rain (and snow)
using RADAR. At this time, the method suffers from the problem that there are
two independent variables, the rainfall rate and the RADAR reflectivity of the
rainfall. Thus, either the rainfall rate or the size of the droplets
(or snow flakes) needs to be known to determine the deposition rate and
amount. RADAR is most useful in determining the precipitation amounts to
localized areas which contain one or more well located and shielded gauges
that can be used to calibrate the RADAR. The accuracy of the results of
course, are still limited by the accuracy of the gauges.
An extensive one-year project to determine precipitation inputs to Lake
Ontario was carried out in conjunction with IFYGL (Wilson 1977). Two weather
RADARs were used along with an extensive rain gauge network. Precipitation
into the Lake was estimated using the RADAR data and asssuming that systematic
-------
LEE W. LARSON AND EUGENE L. PECK
AUGUST 1974
^ 5
»
a
Wind Speed
13461 MPS
2 I 6 fl 10 12 14 16 11 20 MPH
. 1 Gage catch deficiencies versus wind speed. Line 1 is for rain
(shield makes little or no difference in deficiencies), line 2 is for snow
»'ith a shielded gage, and line 3 is for snow with an unshielded gage
Wind
Direction
Fig. 2. Airflow pattern over an open cylinder (Goodison et al. 1981)
-------
differences did not exist between overlake and overland gauge/RADAR ratios.
The investigators felt that cold season precipitation over the Lake was
underestimated because the lake-effect storms had low echo tops which were not
seen by the RADAR.
Snow Gauging
Because of the problems with blowing and drifting snow, it is much more
difficult to quantitatively sample snow than rain. If it is windy when the
snow is falling, the air deflected by the sampler can lead to significant
under or over sampling of snow by collectors designed for rain collection
(Goodison et al. 1981; Fig. 2). For instance, a standard rain gauge can have
a collection efficiency for snow of only ~3Q% (Fig. 3) in 9 m/s winds.
A number of shielded gauges have been designed which minimize the effects
of wind on the collection of snow (Kurtyka 1953). The most widely used of
these is the Nipher gauge (Goodison et al. 1981; Fig. 4). It is a cylinder
surrounded by (inserted in) a parabolic shaped shield. The principal of the
design is that all of the approaching wind that touches the nipher collector
is deflected away from the interception cylinder, leaving the air flow over
the cylinder completely unobstructed (ideally this air does not know the
collector is there). Studies indicate that the Nipher gauge slightly over
collects snow at low wind speeds, and the collection efficiency begins to fall
off at wind speeds greater than 5 m/s (Fig. 3). At all speeds, however, it is
a much more efficient snow sampler than rain gauges, even gauges with Alter,
or other type wind shields, and it is a very good collector at low to moderate
wind speeds. A Nipher gauge should be used for all snow measurements in a
GLAIS network.
Finally, all snow gauging devices can "bridge" over under certain
infrequent types of snow conditions. In addition, snow can fill-up the space
between the outer shield and the inner cylinder in the Nipher-type collectors,
leading to collection errors.
PRECIPITATION SAMPLING
Rain Sampling
Several wet-only rain samplers are commercially available which use a
precipitation sensor to uncover a collection container during precipitation
events. These collectors are thought to do a very good job of rain sampling.
The collection areas of these samplers ranges from about 0.032 ra^ (Sangamo) to
0.21 ra^ (MIC). These samplers may be used to collect event samples, or they
may be left out for a time period of a week, month, etc. and accumulate
precipitation from one or more events. Some have been adapted for organic
-------
B.E. GOODISON
H.L. FERGUSON
Handbook of Snow
G.A. McKA*
10
5 08
O)
06
I
o
04
co
O)
o
o
02 -
0.0
MSC Nipher
Shielded
c/w = complete with
Universal c/w Alter
Tretyakov
Fischer andc , ,.
D c/w Alter
Porter
123456
Wind speed at gauge height (m/s)
? 1 8 • 3 Relationship between gauge catch and ground catch as a function of wind
speed for different types of gauges (Goodison, I978b).
Figure 4. Types of gauges used to measure the water equivalent of snowfall in
different countries, (a) MSC Nipher Shielded Guage (Canada), (b) Swedish
SMHI Precipitation Gauge, (c) USSR Tretyakov Precipitation Gauge.
-------
sampling by incorporating an extraction set-up in the apparatus for real-time
sample processing. Due to aerodynamic effects, they are probably slightly
less effecient rain samplers than the standard eight-inch rain gauge.
Ships with regular schedules on the Lakes, or which spend a lot of time
on the Lakes, could be very useful platforms for collecting rain samples over
the Lakes. A properly positioned sampler, operated when the ship is in the
open waters, should give samples equivalent to those obtained by shore
samplers (concentration information only). However, it might be difficult to
determine the direction of the storm because of the motion of the ship.
When event rain samples are collected, brief or low-intensity rain
events result in small sample sizes. These small samples cause problems for
analyses which require more sample than is available. There are two possible
solutions to this problem. First, a number of studies have shown that most
materials show much higher concentrations in the first few mm of rain than
their average concentration in an entire rain event. This is why it is very
necessary to collect this early rain in each event. But it also means that
while the volume is low, the amount of each material present in the sample may
be sufficient for quantification. All that needs to be done in such cases is
to add sufficient distilled water to bring the sample volume up to the minimum
volume required for analysis. This volume of water added should be recorded
and the appropriate corrections made when the results are interpreted.
Secondly, samples may be accumulated, and analyzed as a composite. This
will result in the loss of the information with respect to the individual
samples, but will result in the total contribution of these precipitation
events being part of the data base. If the precipitation data are going to be
analyzed on a directional basis, then these samples should also be composited
and analyzed on the basis of the sector from which the air mass originated.
Wet Loadings
The spatial variability of rain in the Great Lakes basin is quite high.
This is due mostly to the fact that 40-50J of the total precipitation comes
from convective raincells (thunderstorms), often associated with squall lines
and squall zones. Thunderstorms are usually composed of a number of
individual storms cells, whose characteristics differ widely. However, the
most frequent ones average 16 km long and 6 km wide, and last about 50 minutes
(Stall and Huff 1971, Fig 5; Changnon 1981, Fig 6). Thus, rain gauges only a
few km apart will typically collect quite different amounts of precipitation
from each thunderstorm. While the amounts will tend to average out over the
long term, annual variations for gauges 20 or so km apart are on the order of
a factor of two. Due to this large variability, the amount of precipitation
-------
Stall and Huff (1971)
1p.m. 1.10
1.20 1.30 1.40 1.50
TIME, JUNE 2. 1965, p.m.
Fig. 5
2.00 2.10
STANLEY A. CriANGNON, JR. AUGUST 1981
JOURNAL OF THE ATMOSPHERIC SCIENCES
** 1 8 . O Raincells dunn£ the rain penod of 2-1 Aucusi 1965 Isochroncs deptci
motion of the leading edge of the runcelK
-------
collected at any one site is a poor indication of the regional amount of
precipitation, or of the precipitation inputs for that event. (However, it
has been recently found that separate thunderstorm systems can coalesce at
night under some conditions, forming storms (Mesoscale Convective Complexes,
MCCs) which cover large areas (-2X10^ km2), have high rainfall intensities,
and long durations (Kerr 1985). To the extent that these MCCs occur, they
would decrease the variability of rainfall between sites).
Thus, most atmospheric deposition projects use the results from rain
sampling to calculate the weighted average concentration in precipitation in ;
an area. Loading are then determined by multiplying this average
concentration by the annual average rainfall amount, or by the annual amount as
determined by a number of rain gauges distributed over the area of interest.
Thus, the collection efficiency of samplers is less important then their
having a reproducable efficiency. The amount of rain determined by the rain
gauges are not needed for deposition calculations'and, in fact, probably
should not be used for them. The rain gauges are useful, however, to make a
record of the onset, duration and intensity of precipitation events, and to
serve as a monitor on the sampling efficiency of the collector.
Snow Sampling-Collectors
The commercially available wet-only rain samplers, described above, have
been widely used for snow sampling. There are no reports of their snow
sampling efficiency, but as described for snow gauges, they would be expected
to do very poorly under windy conditions. While the principles and design of
Nipher and Alter-type wind shields is reasonably well understood for
cylindrical collectors, it is not clear how a shield might be designed for a
rectangular collector.
Also, snow blown onto the precipitation sensor of wet-only collectors
when it is not snowing, can cause the sampler to open. Snow can then be blown
out of the collector. To prevent this from happening, the collector should
have a height/diameter (H/D) ratio of 4 or greater. The collectors used in the
commercially available wet-only samplers have a ratio of only about 1.
To alleviate these problems, the CAPMoN network (daily) in Canada uses a
modified Sangamo collector with a 35 cm diameter opening and with a H/D ratio
of -2.5:1. Also, the APIOS-C network in Canada uses a standard Sangamo wet-
only collector but uses a longer bucket in it during the winter (H/D = 4).
The precision of snow sample collection then is poor. However, since the
loading calculations are based on concentration data, these collectors may be
satisfactory. Also, if wind speed information is available from the site, the
10
-------
collection efficiency may be able to be estimated for each event and a
correction applied.
Recently, snow collectors with heated collection surfaces have been
designed. In these collectors, the snow which falls is melted and stored or
extracted on site as a water sample. These heated snow collectors have two
potentially serious problems. First, since snow flakes typically have only a
small mass, it takes a while for the individual droplets of melted snow to
coalesce sufficiently to run off the sloping collection surface. This is more
of a problem on surfaces like Teflon which is not wetted by water. During
this time they are on a warm surface (probably 15-35° above ambient
temperature) and can readily evaporate. The samples thus may show higher
concentrations of non-volatile materials, loss of volatile materials, and
lower amounts of water. A possible solution to this problem is to heat the
collector to melt the snow only after the event is over and the roof is
closed.
A second problem is the large heated area of the collector. It will
cause convection currents above the collector, possibly decreasing the
collection efficiency. Ironically, these induced" currents may cause the most
problem during period of low wind speeds, just the time when these collectors
would otherwise have a reasonable collection efficiency.
Finally, all snow collection devices can "bridge" over under certain
infrequent types of snow conditions. In addition, snow can fill-up the space
between the outer shield and the inner cylinder in the Nipher-type collectors,
leading to collection errors.
Snow Collection-Cores
An alternative method of measuring atmospheric inputs associated with
snow, is to collect snow cores (see Barrie and Vet 1984, for instance). This
method involves the collection of cores of snow after the snowfall, which are
representative of the snow which fell. Alternatively, cores can be collected
infrequently and the results will then serve as a measure of the bulk inputs
(wet + dry + vapor inputs - evaporation) over that period of time. This
method has most frequently been applied in areas where little thawing occurs
during the winter and cores are collected every several weeks or at the end of
the snow season. The sample then represents the net atmospheric inputs over
that time period.
Since the first few percent of melt water carries away a major fraction
of the soluble materials in the snow, the successful application of this
method requires that no melting occurs between the snowfall event and the
11
-------
sample collection. However, an area could be prepared ahead of time to retain
any melt water and serve as a collector. A sample obtained at the end of the
snow season would then represent the net deposition during that season (or
samples could be ootained at intervals and their inputs summed).
Since snow sublimes during the winter the snow pack at the end of the
season is less than what fell. This leads to a concentration of the non-
volatile components. Thus, the inputs that are calculated using this method
will have to be based on loadings and NOT on concentrations.
If this method were applied to the frozen surface of the Great Lakes in
protected areas where the ice forms early and remains throughout the winter,
or areas adjacent to the frozen surface of the Lakes, a sample collected
before breakup of the ice would oe a direct measure of the net atmospheric
inputs from the ice cover. This method gives information ONLY on the inputs
to the snow surface, to the ice covered portion of the Lakes (an average of
19% of the lake surface of Lake Michigan during an average winter). Other
methods would still be needed to determine the inputs to the open water
portions of the Lakes. Inputs of snow to the open waters of the Lake would
be expected to be higher than to the ice-covered portions since the open
waters would also serve as a sink for blowing snow.
For snow collections then, an event snow collector with a deep bucket and
an Alter-type shield should be used at most all of the GLAIS sites. In
addition, the snow core method could be used in the north to obtain some
information on bulk deposition to snow-covered surfaces.
DRY DEPOSITION
Dry deposition is defined as atmospheric inputs that occur when it is not
raining (precipitating). It is thought to be an important input mechanism for
many materials to bodies of water, but there are now no well qualified methods
of determining these inputs (Slinn 1980, 1983). In addition, most
investigations into determining dry deposition have been concerned with inputs
to land and vegetative surfaces.
There are two reasons why these results may not be applicable to inputs
to water surfaces. The first is the nature of water and its surface. The
surface is a liquid and it is uniform. Since it is liquid, it will "wet" many
materials and thus irreversibly capture them; the water at the surface will be
continually renewed with new water from within the epilimnion; and its surface
properties are quite different from those of solid surfaces.
12
-------
Secondly, because of the high heat capacity of water, and the mixing
which occurs in the epilimnion, the surface temperature will be quite constant
on a diel basis, regardless of short-term energy gains and/or losses. Thus
the temperature of the Lake surfaces, and its variability, are very different
from the adjacent land surfaces during many months of the year. These
differences will result in different dry deposition rates to the land and
water surfaces.
With respect to determining dry deposition inputs, the results of several
studies indicate that the dry deposition of small particles «1 urn) is not a
significant contributor of atmospheric inputs (Talbot and Andren 1983;
Davidson and Friedlander 1978). This is the case even for those materials,
such as lead, that are present in the atmosphere chiefly on the small
particles. Wet deposition is the principal atmospheric removal process for
these materials, and what dry deposition occurs of these materials, seems to
be due chiefly to the small percentage of these materials associated with
large particles.
At least initially then, direct determination of small particle inputs
need not be made. An estimate of their inputs could, nowever, be included in
the atmospheric inputs total. Inputs of vapor are usually included as a
component of dry deposition, but in this discussion, vapor inputs will be
considered separately.
Most of the materials to be determined in a GLAIS network will probably
be associated principally with small particles. It is interesting to note
then that the only dry deposition sampling for them suggested here is that
fraction associated with large particles!
The principal mechanisms by which large and small particles deposit from
the atmosphere are different. The collection methods for the large and small
particles are also usually different, but as mentioned above, all of the
methods suffer from significant limitations and uncertainties. The better
developed methods are briefly discussed below. Deposition collectors could oe
useful only for large particles (Wesely et al. 1985) micrometeorological
methods could be useful only for small particles; and the deposition velocity
method could be used to estimate both large and small particles. It is
suggested that the Canadian-type deposition velocity method, as discussed
below, be used initially in a GLAIS network to estimate dry deposition. It is
also recommended that the NOAA/EPA/DOE method and the deposition bucket
methods also De evaluated at the Master Station for consideration of their
wider use.
13
-------
Deposition Velocity
A commonly applied method of estimating dry deposition inputs is to assume
that each different material in the atmosphere tends to "settle out" or
"deposit" at a fixed rate, its deposition velocity, Vd. The deposition rate
(mg/(m2's)) to a particular surface then is the concentration (mg/m^) of that
material in the atmosphere times its V^ (m/s).
An advantage of the deposition velocity method is that if air monitoring
samples are being collected for the determination of sources, then those data
would be available for the calculation of inputs based on deposition-velocity
type calculations. However, information on the particle-size distribution for
each material will be needed. Much information of this type is available for
a variety of materials from other studies to serve as a guide.
To estimate atmospheric inputs by the Vd method, the air concentration and
the Vd of each material of interest must be known. The limitation of the
method is accurately estimating the value of the Vd. The difference is that
the value of the Vd is a function of a number of variables. These include:
the particle size distribution over which the material of interest is is
present, the wind velocity, the nature of the particles, the atmospheric
stability, and other factors. Thus while it is a single number, it is quite
complex to calculate or measure, and to use (Williams 1982; Slinn 1982).
Also, a number of studies have demonstrated (Chamberlain 1976; see also
Davidson et al. 1985) that the sampling efficiencies of high volume air
samplers and cascade impactors decrease as the particle size increases. This
results in significantly underestimating the number of particles larger than
10 urn in the air, and the inputs associated with them.
There are additional complications in determining Vd's over the Great
Lakes. These include: the stability of the atmosphere is often very different
over the Lakes than over the adjacent land; the complications near shore of
upwellings and downwellings in the Lakes on the local stability; changes in
wind speed and direction as air moves from the land over one of the Lakes, or
vice versa; the complications of lake and land breezes along the shores of the
Lakes on many days of the year; and the effect of the sea state on particle
deposition.
In general, the Vd for each material or compound of interest in the
atmosphere will be different and it will vary for each material as the
meteorology and the atmospheric compositions change. Vd's have been estimated
in a variety of ways, most commonly using models. The better models attempt
14
-------
to take into account the variables mentioned above, but the uncertainties for
most materials may still be as large as an order of magnitude (Sievering 1984).
There are some recent developments that should increase the accuracy and
reliability of this method (Wesely et al. 1985; Rahn et al. (1984, 1985)
report that in the northeastern U. S., air masses from the same region tend to
have similar particle size distributions and particle-size associations of
different materials. Air masses from different regions tend to be different.
Thus it might be possible to make the effort determine the particle-size
association of each material, for each different type of air mass for each
sampling site. This information could then be used for all calculations on
air masses from those regions.
Secondly, materials whose inputs to the Lakes are known to be
predominantly atmospheric, such as lead, some of the radionuclides, and some
organics, may be used to calibrate the deposition models. Such an effort has
already attempted by Andren (unpublished) for Lake Michigan using lead. The
calculations for Lake Michigan were based on information on lead deposition
obtained from intensive studies on a small lake in northern Wis. that has no
surface water inputs.
Two Canadian networks, CAPMoN and APIOS-D, estimate dry deposition inputs
of sulfuric and nitric acids, and S02 using a V^ method. Daily samples are
collected on a filter pack consisting of a particulate filter, a nylon filter
to absorb nitric acid, and a carbonate-impregnated glass fiber filter to
capture acidic sulfur compounds. The day's meteorology, along with a
deposition model, are used to estimate the dry deposition of these materials.
Over-the-lake meteorological information (air (-2m) and water
surface temperature, humidity, wind speed and direction, water current speed
and direction, and wave heights) for all of the Great Lakes is available
during the ice-free season from NOAA bouys in the Lakes. These data can be
available in real time, and data from past years are available on tape or on
microfiche. These data should be useful for determining what goes on in mid-
lake during the time of year the bouys are present.
Deposition Collectors
Probably the most commonly used but also the most widely criticized
depostion collection method is the use of dry deposition collectors. These
are usually surfaces or containers that are exposed between precipitation
events for a measured period of time, and the material that they accumulate is
determined. It is widely believed that the shape of the collector (plate; low
aspect bucket; high aspect bucket; etc.) affects the amount of material
15
-------
collected. Two recent reports address these questions (Dolske and Gatz 1985;
Dascn 1985). The findings seem to be that teflon surfaces undercollect; that
the height of the bucket is not an important variable (the sides are not
acting as the principal collecting surface, or the collection of small
particles is not important); that a bucket with water collects more material
than does a dry bucket (one would want to use a wet collector for lake input
studies); and finally that the errors involved in the use of bucket collectors
is less than a factor of three, and probably less than a factor of two.
The reports on the collection of sulfate (Dolske and Gatz 1985; Feely et
al. 1985) indicate that the deposition rate found by bucket collectors were
higher than that found by other deposition measurement techniques, perhaps by
a factor of 1.8 or so. It is to be expected that the other methods
underestimate the deposition. Thus the buckets may over sample, but the error
involved is not large. If it is known that the buckets may slightly over
collect, this could be factored into the estimates produced using them. Since
dry deposition is not going to be an important input route for many materials,
being able to put an upper limit on it will be very useful.
The problem this method shares with all the other dry deposition
collection methods is an unknown collection efficiency. However, if water is
kept in the collector, the method is calibrated using 210Pb, lake studies or
other methods, and determination of the the small particle fraction is not
attempted, the accuracy of this method may be as good or better than other
dry deposition estimates.
An advantage of the method is that wet/dry samplers will be in use at all
of the collection sites, but only the wet side will be in use. Certainly this
method should be evaluated at the Master Station, and the results compared
with those found using other methods.
Micrometeorological Methods
For small particles «1 urn) for which gravitational settling is not
important, and which tend to move with the eddies in the air, a number of other
methods have been tried to determine their transfer rates to surfaces. These
include: eddy correlation, eddy accumulation, modified Bowen ratio, gradients,
and varience (see Hicks et al. 1980 for a description and discussion).
Despite the optimism in the 1980 report, these methods continue to have
several drawbacks. 1) They are still in the research stage or are not yet
developed as field-tested operational equipment able to be used in a routine
manner. 2) The fast-response sensors needed by some of the methods are not
available for any materials on particles. 3) The calibration of these
16
-------
techniques is difficult under high wind and sea conditions with an unstable
atmosphere, when the air to water transfer rates may be high.
A joint NOAA/EPA/DOE effort is developing a routine monitoring Vd method
for dry deposition inputs of small particles. This method uses filter packs
similar to those used in the Canadian networks, but determines V^ using
micrometeorological methods. Thus only the deposition of small particles can be
estimated, and the air sampler has a large particle "denuder" on the front to
prevent these particles from being collected. The meteorological parameters
determined are temperature, solar radiation, wind speed and the standard
deviation of the wind direction (a measure of the turbulance and therefore of
atmospheric stability), and humidity.
A weekly air sample is collected and the meteorological parameters are
determined throughout the sampling period. An average V^ is calculated for the
week from the meteorological data. This is multiplied by the average
concentration of the materials of interest from the filter sample, to yield an
estimate of the dry deposition inputs for that week.
Because of the large uncertainty associated with estimating the dry
inputs of small particles, a direct measure of their contribution would be
very useful to any GLAIS program. Thus it is recommended that this method be
evaluated at the Master Station for possible inclusion into a GLAIS network.
If air samples are already being collected for source determination, only the
micrometeorological measuring system need be added.
VAPOR EXCHANGE
Compounds present in the atmosphere as gases, such as oxygen, dissolve to
some extent in the Lakes. In the environment then, there is continuous
exchange or transport of these compounds across the air/water interface. Each
compound will dissolve or evaporate until the equilibrium condition for that
compound is satisfied. At equilibrium the fugacity (escaping tendency) of the
compound is the same in the air and water phases and, while the flux through
the water surface may be quite large, no net transfer occurs. The equilibrium
condition is determined by the relative tendency of the material to evaporate
(vapor pressure) and dissolve (solubility). The distribution coefficient
(Henry's Constant) is the ratio of the vapor pressure of the compound to its
solubility.
Net transport from the air to water, or water to air will occur until
equilibrium is attained. So, if the water is undersaturated with respect to
the air, net transfer to the water will occur. This would constitute an
atmospheric input. Conversely, if the air is undersaturated with respect to
17
-------
the water, net transfer to the atmosphere will occur. This would be a loss
mechanism for the lake.
To determine which direction net transport is occurring, the fugacities
of the material in the air and water phases, and the Henry's constant need to
be known. The amount of material transported across the air/water interface
is then calculated from the difference in fugacity between the two phases and
a mass transfer coefficient for that compound under the prevailing conditions.
The Henry's constants are known for some of the compounds of interest, but
others need to be determined.
The fugacity determinations to date are usually complicated by the
association of the compounds of interest with small particles in the air and
water phases. For some compounds, this is only a small effect, for others, it
may be significant. The use of denuders on air samplers allows the fugacities
in the vapor phase to be determined, but suitable methods are not available
for water samples. However, since the effect is probably small «25%) for
most compounds, a correction may be estimated and applied. The use of model
or surrogate compound could be very useful in making these estimates.
A more significant problem is determining reasonable values for the mass
transfer coefficient under all conditions over the Lakes. The problem in
determining these values is that they are very dependent on the wind speed,
temperature, sea state, the presence of a surface microlayer and other
factors. Again, a lot of modeling work, wind tunnel studies and ambient
measurements have been done which have improved the state of the art greatly
in recent years, including: Liss 1982; Mackay and Yuen 1983; Tucker et al. 1983.
MACRO METHODS
The term macro here means methods that obtain information on atmopheric
inputs indirectly by a method involving measurements on the Lakes or
atmosphere. The information is usually obtained from measurements made at
different times and/or different locations in the Lakes, with the input term
being related to the differences found. In general, these methods measure a
parameter that integrates the inputs over an annual or seasonal period. For
some of the methods, one or a few GOOD measurements per year is all that is
required (of course, great care and a lot of planning is necessary to obtain a
few GOOD measurements).
The appeal of these methods often is that they serve as a direct measure
of atmospheric inputs and they are independent of the effects of the winds,
precipitation, etc. In an atmospheric inputs program, besides being useful in
18
-------
their own right, they could serve the most useful function of being an
independent method of verifying the results obtained using other methods.
Surrogate Compounds
A tracer-compound method might also be useful to determine the inputs of
a variety of organic compounds. The method requires identifing a compound
whose only source to the Lakes is the atmosphere. Other compounds whose
atmospheric distribution and deposition mechanisms are similar to the tracer
compound, then would be expected to deposit in the same manner. Thus, their
atmospheric inputs relative to the tracer compound, corrected for differences
oetween the compounds, would be assumed to be proportional to their ratio to
the tracer compound in the air. For compounds that partition between the air
and water, this method could be useful in separating the net inputs from the
total inputs.
For organic materials, care would be needed in choosing the tracer
compounds. One of the necessary properties would be that the compound be
resistent to degradation to permit accumulations in the water and sediments to
be determined. It would be most useful if several compounds could be found
with different properties. For instance, one with a low volatility could
serve as a tracer for particulate-associated organics; one present in the
vapor phase with a high Henry's Constant could help in determining washout by
rain; etc.
Accumulation Methods
There are recent reports of methods that use the epilimnion of Lake
Michigan as an integrating collector of atmospheric inputs. The idea is
that during stratification, the epilimnion is cut-off from all direct inputs
other than from the atmosphere. For those materials that do not undergo
transformations, the change in the amounts of them per unit area over the
period of measurement then is the atmospheric input loading rate (wet + dry +
net vapor). Determining the accumulation over a period of time then would
yield the net atmospheric inputs.
There are two reports of the application of this type method to the Great
Lakes in the literature. Tisue (1984) observed that the profile of trace
metals in the water column was uniform before stratification. At the end of
the stratified season, the concentrations of the metals were higher near the
surface. The amount of the metals needed to increase the concentration above
that found at the thermocline, was assumed to have come from the atmosphere
since stratification began (corrected for losses).
19
-------
Chambers et al (1983) measured the concentrations of a number of
substances in the epilimnion after stratification began and then at maximum
stratification. They also deployed sediment traps toward the top of the
thermocline during stratification. From the change in concentration of a
material in the water during stratification, and the amount of that material
found in the sediment trap, the inputs to the epilimnion for that period of
time could be calculated (the contributions due to upwellings, diffusion,
zooplankton grazing, and other possible gains and losses were not estimated or
corrected for).
It is unfortunate that these methods probably will not work well during
the winter when the stratification in the Lakes is weak, dry deposition is
probably at its maximum rate, and snow is hard to collect. It should work
best during the summer, but this is when the atmosphere is very stable and dry
deposition may not be occurring.
An additional method might use "lakes within the Lakes". This would be
an application and extension of the studies pioneered by Swain (1978) in his
studies on atmospheric inputs to Siskiwit Lake on Isle Royale in Lake
Superior. This method is an ingenious approach to solving the problem of
designing a collector for atmospheric inputs that mimics a lake surface. It
uses a lake!! The idea is that many islands in the Great Lakes contain lakes.
Studies on these lakes, similar to the detailed loadings, budget and inventory
studies conducted in recent years by Andren and co-workers on Crystal Lake in
northern Wisconsin, and by Hites and co-workers on Siskiwit Lake could provide
information directly applicable to atmospheric inputs to the Great Lakes.
A number of islands in the Great Lakes, including Isle Royale, Beaver,
and Manitoulin contain a number of lakes, and a number of other islands
contain one or more lakes. Some of these should have suitable exposure and
hydrology to yield useful information on loadings and for the calibration of
models. It is possible that some lakes on land, but close enough the Great
Lakes to be within their climatological and meteorological regime, could also
be useful for such studies.
SOURCE DETERMINATION
The program element to determine the sources to the atmosphere of the
materials coming into the Great Lakes could consist of several components.
The first should be to determine the direction from which the materials of
concern are coming. The second could be to collect emission data in those
regions found to be sources, for the different materials from point and non-
point sources. A third component could be air monitoring in regions away from
the lakes to help identify source regions for the materials of interest.
20
-------
There are a number of initiatives in the U. S., Canada and Western Europe
to determine the sources of acid forming materials in the atmosphere. Many of
the methods developed by these programs should be applicable to a program to
determine sources of materials to the Great Lakes.
Identification of Source Regions
In order to determine the source regions of materials being transported
to the Great Lakes, short terra sampling or intermittent sampling will be
required. For instance, wet deposition will have to be collected on an event
basis so that the materials and concentrations determined in the samples can
be related to the appropriate region upwind.
If it is decided to obtain information on the source regions for the dry
deposition material, then air samples should be collected such that each
sample represents air from only one air mass. Back trajectories could then
be determined to identify the region that was upwind of the each of the air
samples.
One suggested sampling scheme (Stearns 1985) would be to have three air
samplers, one to sample when the wind was out of a particular sector. The
sectors are related to the synoptic weather patterns around the Great Lakes
and would be: 0° to 180°; 180° to 270°; and 270° to 360°. Since most storms
are out of the SW or NW, this scheme separates these two source regions from
each other and from the rest of the regions. Which sampler was operated at a
particular time, and the beginning and end of each sampling period, would be
determined by tne wind direction. The time and duration of each sampling
event, and the wind speed and direction during each event would be recorded
for each sample. At the end of a week, or month, or etc., analyses of the
three samples would indicate the average air concentration for air masses from
each sector for the sampling time period.
If such a sampling scheme were to be used, care would have to be taken in
its design to minimize passive sampling by the samplers not turned on. In
addition, the quality assurance program would have to be able to quantify the
amount of passive sampling which did occur so that it could be corrected for.
Emission Data
The collection of emission data for different materials in regions
adjacent and remote from the Lakes would need to be co-ordinated with other
agencies (Rothblatt 1985). These data, in conjunction with the presence and
amounts of the different materials found in the GLAIS sampling program, could
help identify the sources of these materials to the atmosphere.
21
-------
Regional Air Monitoring
If it is determined that some source regions seem to be significant
contributors to the atmosphere of materials going into the Lakes, then efforts
could be made to determine the sources of these materials. This information
could be available from EPA Regional Air Quality Programs. It might also
involve additional air sampling in the regions thought to be the sources.
IV. SITING
If samples of precipitation and/or air are collected to determine
atmospheric inputs, it is important that the collectors be located such that
useful samples are obtained. The general requirements are that a site should
not be affected by local sources, it should be accessible to 120V electric
power, it should have a dedicated and qualified operator, and it should be
close to the lake. It is not necessarily easy to find sites in many areas of
the Great Lakes where these requirements are all met. Areas remote from
sources and urban areas tend to have few people and power lines, and the power
is often not reliable, particularly in the winter. Open areas are sometimes
hard to find in the forested northern areas, and secure, unobstructed sites
are difficult to find in and close to urban areas.
Alternate power sources such as diesel or gasoline powered generators
would remove the need for 120V electricity, but they emit organics and need
regular attention. Solar cells are not practical in many areas adjacent to
the Great Lakes in the winter due to the large number of days with dense cloud
cover. They may, however, be useful in the summer on islands or other remote
locations.
Climatology
People who live adjacent to the Great Lakes know, and the meteorologists
and climatologists who have have studied the Lakes have demonstrated, that the
Great Lakes affect the weather in the vicinity of the Lakes (Gatz and Changnon
1976; Eichenlaub 1979; Cole and Lyons 1972). There is also some good evidence
that the presence of man in the Great Lakes basin effects the weather over the
Lakes. The siting and operation of a GLAIS network must take into account the
effects of the Lakes on the mesoscale climatology.
The effects of the Lakes are due primarily to their large heat capacities
and thus their more constant temperatures. For instance, the surface
temperatures of the open waters of Lake Superior range from 0°C to about 15°C
over an annual cycle, while air temperatures range from about -40°C to +35°
over the adjacent land areas. The result is that the open waters of the Lakes
22
-------
I50C
IOOC
100
TYPICAL SMOKE AND CLOUD PATTERNS
INLAND
CUMULUS
ADVECT OVER
FRONT &
DISSIPATE
SMOKE LAYERS ALOFT
OVER LAKE
SUSPECTED TRAJECTORIES
-10
SHORELINE
20 KM
i m. i, Schematic representation of the cross-sectional structure of a lake breeze cell
ami u* ctteet on the diilusion and transport of pollutants. Heights are given in meters.
flu- hea\> solid line lound m each diagram represents the mesoscale lake breeze front
(1.\oti.s 1'JTlb). ii) Streamline patterns calculated from pibals illustrate inflow and re-
turn tlou. layers, the convergence zone updraft and subsidence over thelake. b) Thermal
structure; light solid lines are constant potential temperature. Over the lake, stability
is i ml ic a led b\ an increase in potential temperature with height. Over the land, the in-
ternal boundary layer is defined by values of potential temperature that decrease or are
constant \\ ith height. Temperature soundings are indicated by the solid dark lines. The
lake .sounding and lines of constant potential show three inversions. Over the land, a
doep adiabatic boundary Layer is overlain by a synoptic scale subsidence inversion.
e) Smoke and cloud patterns associated with typical lake breeze cell. Shown is a ship's
plume that would advect shoreward with little vertical diffusion until it crosses the shore-
line into the turbulent boundary layer. The plume from a tall stack, initially emitted into
-table air tumigates \\hen it is intercepted by the building turbulent boundary layer. The
diagram al.-o shows a "wall of smoke" in the convergence zone and concentration of
-moke aloft in the bottom of the return flow layer and in the upper portion of the inflow.
.1) 1 stimated trajectories of particles with different sizes and sources of origin. The ef-
fect ot si/e sorting is discussed in the text.
-------
are warmer than the atmosphere in the fall and winter, and colder in the
spring and summer. To the extent that the Lakes affect the air above them,
they make the stability of the atmosphere over the Lakes different from
adjacent land areas. Also, while the atmosphere is frequently stable over the
land at night and unstable during the day, during some seasons there often is
no diel change over the Lakes. This has a profound effect on the local
meteorology, and is the principal factor which complicates siting collectors
to determine inputs of materials from the atmosphere and their sources.
The warmer Lake surface in the fall and winter with respect to the air
leads to unstable atmospheric conditions over the Lakes, to high evaporation
rates, increased cloud formation and lake-effect snows, and probably to
increased dry deposition rates. The colder lake surface in the spring and
summer with respect to the air leads to stable atmospheric conditions above
the lake, the formation of a stable inversion layer, decreased evaporation,
cloud formation and precipitation, the formation of Lake-breeze conditions
about 5Q% of the time, and probably to decreased dry deposition rates as
compared to the adjacent land (see following figure). The effects of the lake
climatology on wet and dry deposition rates need to be verified and quantified.
Snowfall along the shores of the Lakes is usually more frequent (lake-
effect snows), but lower in amount than areas 30 km or more inland. If the
shore snowfall reflects the snowfall over the Lakes, then the snow gauging
must take place in the near-shore zone. The location of collection sites for
snow is probably less critical than for rain, since most of the snow in the
Great Lakes is associated with synoptic storm events and the materials being
scavenged have typically come a long distance. Thus, local inputs or effects
are probably less important.
Urban areas occupy a significant fraction of the Great Lakes shoreline,
and their inputs to the Lakes thru the atmosphere were known to be significant
(Gatz 1975). Also, Sievering et al. (undated) reported that the samples
collected at the GLAD site on Beaver Island was significantly affected by the
town of St. James. This GLAD sites was located at the Mi. DNR fire control
station on Beaver Island, about 1.2 km east and across the bay from St. James.
St. James has a low density, permanent population of about 200, with several
times that many additional residents and visitors during July and August. If
one of the objectives of a GLAIS network is to quantify the atmospheric inputs
to the Great Lakes, then the contributions to these inputs from urban areas
can not be ignored.
It will be difficult to sample these inputs representatively. It could
involve siting samplers in the plumes from these areas, along the shore or on
24
-------
towers or islands offshore. Comparisons with inputs from unaffected regions
should then permit the contribution from the urban areas to be estimated.
Over Water Measurements
Host of the atmospheric inputs sampling that has been done to date has
been from the shores of the Lakes. The assumption has been that deposition
along the shore is the same (within experimental limits) as the deposition
into the Lakes. If shore sampling is to be a major component of an
atmospherics inputs program, the errors in shore sampling will have to be
quantified, and any necessary changes made in siting, sampling or
interpretation.
Almost without exception, those who made recommendations concerning a
GLAIS program, recommended an over the Lakes sampling component in order to
understand the mesoscale meteorology of the Great Lakes with respect to
atmospheric inputs to the Lakes. The most discussed problem was the ratio of
the amount of precipitation that occurs over the Lakes compared to that which
occurs at the shore. Several studies have been conducted on Lakes Michigan
and Ontario to quantify these differences. These studies all concluded that
there is more precipitation over the Lake during the winter, but less rain
over the LaKeTcturing the summer than on shore. This is in accord with what
would be expected Dased on air vs. water temperatures.
One interesting observation related to this question was made with the
weather RADAR used in the IFYGL projects over Lake Ontario, On several
occasions, stationary snow storms were observed over the Lake (Wilson 1977).
On those occasions, the air was being scavenged over the Lake, but no
observable or measurable precipitation was occurring on land.
However, in a careful consideration of all the studies on Lake Michigan,
Bolsegna (1979) concluded that the differences found in land vs. lake
precipiation were smaller than the errors associated with the precipitation
gauges and the gauge networks. Thus it does not seem that a program to
attempt to quantify differences in the on-shore vs. off-shore precipitation
would be fruitful at this time.
However, inquiries into the effects of land- and lake-breeze cells on the
samples collected by shore-based air and precipitation samplers; differences
in wind speeds and direction over the Lakes compared to adjacent shore sites;
differences in the atmospheric stability over the Lakes with respect to
adjacent shore areas during the different seasons, and its variation over the
Lakes due to winds, sunshine, ice cover, upwellings and downwellings; and the
effect of urban areas on over-the-lake precipitation frequency, intensity and
25
-------
amount, need to be carried out. Observations over the Lakes to understand
these effects could be made from islands, ships, bouys and towers. The people
who made suggestions for a GLAIS design had conflicting views on ships, bouys
and towers (Philbert 1985; Wesely et al. 1985).
Islands are probably ideal, IF they are small and in the right place,
because they are rigid and permanent (Philbert et al. 1985). Specific
considerations would be the size, ease of access, topography, habitation and
vegetative cover on the island. Islands possibly of use include Isle Royale
the Apostles and Caribou in Lake Superior; Beaver, N. £ S. Fox, S. Manitou,
St. Martin and Poverty in Lake Michigan; Great Duck, Charity and the North
Channel islands in Lake Huron; South, Middle and North Bass, Pelee and Long
Point in Lake Erie; and Galloo, Stony and the Thousand Islands in Lake
Ontario. Since many of these islands are unoccupied, the equipment sited on
the islands would have to be designed for remote operation. It could include
recording rain gauges and meteorological instruments, as well as some
samplers. Data could be recorded on site or telemetered to shore.
Ships could be useful, but probably only if they had a REGULAR
and frequent schedule in the desired area. In Lake Michigan, the Ludington-
Manitowoc and the Charlevoix-Beaver Island ferries could be useful. In Lake
Superior the ferries to Isle Royale from Copper Harbor and Grand Portage,
might be used. Research ships on the Great Lakes are attractive sampling
platforms. Their advantages are that they are equipped for sample
collections, could hold a variety of instruments and collectors, have
technically qualified personnel on board and often spend significant periods
of time in the open waters of the Lakes. Ships are perhaps the only sampling
platform that will be available in deep-water areas of the Lakes. Finally,
there is a lot of experience on the Great Lakes and the Oceans in collecting
meteorological, air and precipitation samples from ships. This experience
should be useful in designing useful projects within the limitations of ship-
board sampling.
If information is needed in relatively shallow water, guyed towers may
be the platform of choice. The advantage is that they are stable, can be
built relatively tall and can hold a variety of instruments and collectors.
A number of these have been, or are being, operated in the Great Lakes, and
so there is already a body of experience with them. A problem with this and
other sampling platforms in the lakes is that of insects and birds. During
the warm months, insects and birds are attracted to any type of structure out
in the lakes. Their presence could lead to sample contamination and to
possible interference with the operation of some of the equipment.
26
-------
NOAA presently operates 8 moored bouys (3 & 6 meter) in the upper four
Great Lakes during the ice-free season. Also, they are testing a 40' diameter
all-season bouy in Lake Superior during the winter of 1985-86. Many of these
bouys have been operating for 4 to 8 years collecting air temperature and
humidity, wind speed and direction, water temperature and sea state, and the
wave spectrum. This data can be available from real-time telemetry (see
Hartmann 1985) and on microfiche and magnetic tape. It could be a useful
resource. Problems with the NOAA bouys are that they are designed to follow
the waves, and the fact that instruments can not be placed very high above the
water (effectively ruling out precipitation sampling). The motion of these
bouys complicates the detemination of the wind speed and direction. An
instrument package compatible with the present operations of the bouys could
possibly be placed on the bouys for a minimal cost (W. Shepard, NOAA).
Locations of Sampling Sites
One of the more important decisions in an atmospheric sampling network is
choosing the locations of the sampling sites. A number of the general factors
involved were discussed above. A number of different criteria have to be met
and each site should meet as many of them as is possible. Some of these
criteria are: some sites should be in mid-lake; some sites should be upwind of
the Lakes; some sites should be downwind of the Lakes; sa ^ sites should be in
remote areas; some sites should be downwind of urban areas; all the different
"regions" of the Great Lakes basin should be represented; and all of the
sites should be well good sites. Easier said than done.
The data collected by the GLAD network should be useful in locating the
GLAIS network sites. The data could indicate areas where the GLAD network was
redundant and they could indicate that some sites were affected adversly by
local sources. The GLAD locations that have been found to be excellent sites
(siting criteria and data evaluation) should be given a high priority for
inclusion into a GLAIS network. The GLAD sites that are in large open areas,
away from major sources, have good access to power, and are adjacent or close
to one of the lakes are Hovland, Cornucopia, Green Bay (perhaps too close to
the town of Green Bay) and Hammond Bay. A summary of some of the siting
information for the GLAD sites is shown in App. B. These data were collected
during the summer of 1985.
There are other GLAD sites which do not meet all of the above criteria.
But they have open areas adjacent to the present location of the samplers
which could possibly be used. These include: Benton Harbor (urban), Mount
Clemens (urban), Put-in-Bay (urban??), Bay City (urban), Conneaut and Olcott
(POTW). In addition, the site at Empire is open, but is about 2.8 km inland
behind a 400' dune; while the GLAD site at Milwaukee is on a building at a
27
-------
AoO
^
-------
CAMP site, the water filtration plant at the shore has a large open field;
Escanaba has a open site available at the end of a point, about one km east of
the present site in town. New sites may have to be developed along the
eastern end of Lake Superior; southwestern Lake Michigan; southern Lake Huron;
western Lake Erie and western Lake Ontario.
Sites in the National Trends Network (NTN) were located based on
bioregions as defined by Bailey (Robertson and Wilson 1985). Since
climatology is the major differentiation between bioregions, these regions
(Fig. 8) should also serve as a rational for locating GLAIS sites, at least on
the upwind sides of the lake.
Finally, some idea of the number of sites, or the number of different
types of sites will need to be known before specific sites can be chosen.
Locations o£_ Collectors at_ o_ Site
A final important concern is how to locate samplers at a site. At the 36
GLAD sites, 18 of the Aerochera samplers were on roofs while the others were on
the ground. For accuracy in collection, the ideal location for the opening of
precipitation samplers is at ground level with some provision to prevent
incorporation of splash and run-off (Kurtyka 1953). In practice of course,
this is not practical and the standard rain gauge is 3' tall. But the higher
a collector is placed above the ground, the more variable its collection
efficiency is due to increased wind speeds. Putting a precipitation sampler
on a roof also adds the problems caused by the increase in turbulance due to
the building blocking the winds.
Roof sites can have several advantages over ground sites, however, for
rain samplers. Often roof sites are less shielded by adjacent structures and
vegetation; there are few problems with samplers in the middle of roofs due to
bird droppings, and roof samplers are usually much more secure. If an
accurate measure of precipitation amounts are not needed, roof locations may
often be suitable sites for the samplers (with the gauges on the ground).
Roof sites are an advantage for air samplers as the collection efficiency
for all but the larger particles is not affected, and the problems due to the
incorporation of resuspended materials into the samples is minimized.
MASTER STATION
If a GLAIS network is planned which contains a variety of different
types of samplers, it would probably be very beneficial to have one of the
collection sites be a "master site" (Elder 1985). The site should provide a
well-equipped location where development and research projects associated with
29
-------
the network could be carried out. A major purpose would be to perform all of
the involved and detailed calibration and verification checks on the different
sampling systems used in the network that are needed to verify their
operation. Also, new types of samplers and collection and measurement methods
could be evaluated and checked-out in the field before being used in the
network. It would be a good place to site duplicate samplers, and samplers of
new and old designs for inter-comparisons. This site would probably have a
full-time operator who could also service other sampling sites, and it could
serve a variety of purposes Finally, the operator could conduct a thorough and
diverse set of quality assurance procedures on samples, the collectors and
sampling procedures.
A master station would serve as a good place to co-locate samplers with
the GLPN and other Canadian networks, and it might be advantagous to equip,
certify and operate the site as a member of the NADP network.
A component of such a site could be a large (~5m2) manually operated,
wet-only precipitation sampler. The large samples collected from such a
collector (50 I/cm ppt) would allow tne quantification of materials which are
normally below the detection limits for GLAIS samples. It would also serve as
a source of large volumes of natural precipitation for quality control
purposes.
If such a site was planned, finding a good location for it would be
critical. The site should be away from the influence of point or diffuse
sources; it probably should be downwide from one of the Great Lakes, on or
close to the shore; the site should be reasonably flat for several hundred
meters or so; it would be very useful to have one or more islands off-shore to
obtain up-wind samples and meteorological information; and it should be
reasonably accessible in all seasons. Unfortunately, most of the areas around
the Great Lakes that could meet the above criteria are in Canada. One area
which could meet the criteria would be the north eastern shore of Lake
Michigan, somewhere east of the Manitou, Fox or Beaver Islands.
V. MATERIALS TO BE DETERMINED
At this point, it is not necessary that the specific materials and
compounds to be determined are specified. It is important, however, to
consider the types of materials which have different sampling requirements
that may be determined. Sites equipped to measure these materials, should be
able to also measure a wide variety of others. The materials to be sampled
will probably fall into four categories, depending on the material required
for the sampler and on the preservation techniques (if any). The four
categories are: organics, metals, mercury, and nutrients.
30
-------
The Great Lakes Environmental Administrators met in August 1984 and
recommended that Polynuclear Aromatic Hydrocarbons, PCBs, Toxaphene, Chlorinated
Dibenzodioxins, Chlorinated Dibenzofurans, routine Organochlorine Pesticides,
Mercury, Cadmium and Lead be determined (Su 1984). They recommended using 10
sampling sites in the U. S. Some of those who made recommendations concerning
GLAIS also suggested materials to be determined (Bidleman 1985; Eadie 1985).
Organics
Organic materials need to be collected and handled in equipment which
contains no organic materials. Glass, metals and teflon are satisfactory
collector materials. In addition, because of the small concentrations of many
organic materials in precipitation, a sampler with a relatively large
interception area is required to collect sufficient sample. If the organic
collector to be used will process and discard the water on-site (adsorption or
extraction), the portions of this unpreserved sample will not be available for
other measurements (pH, conductivity, etc.).
Metals
Metals need to be collected and handled in equipment which contains no
metals. Polyethylene or teflon are frequently used. Nitric acid is
sometimes added to the sample collector to minimize loss of the metals to
the collector surfaces. Many metals are present in air and precipiation at
quite low levels. Thus, before efforts are made to collect a particular metal,
analytical methods must be available with sufficient sensitivity to reliably
quantify the metal at the levels expected.
Mercury
Mercury is present in the atmosphere chiefly in the elemental form
which has little tendency to partition into water. However some is oxidized
to an ionic form which quickly associates with particles and then can be
removed from the air by wet or dry deposition. However, since all the
oxidized and reduced forms of mercury are readily interconverted, a
preservative (oxidant) must be added to precipitation samples to prevent the
reduction and loss of the mercury in the collected sample. The concentrations
of mercury in precipitation samples are usually high enough, with respect to
analytical detection limits, that a collector with an interception area of .02
m2 should be sufficient. Glass et al. (1985) have used beakers placed within
an Aerochem #301 (.09 m2) to collect samples. If such an arrangement is used,
the collection containers should have straight sides to prevent splashing and
to be able to precisely define the interception area. A wet-only collector
need not be dedicated only to ionic mercury collection, as ther6 should be
31
-------
sufficient space in the covered bucket for several such cylindrical
collectors, only one of which needs to be for the ionic forms of mercury.
Nutrients
A preservative (usually sulfuric acid) is often added to precipitation
samples for nutrient analyses to prevent loss of ammonia and slow the
interconversion of the forms of phosphorus and nitrogen present. If event
precipitation samples are collected, these collections may need a collector
with the area of .09 m^ or more to obtain sufficient sample. Otherwise, it
could be possible to share space in the mercury sampler.
VI. ACKNOWLEDGMENTS
In writing this paper, 1 corresponded with, and/or had useful discussions
with a number of people. They included Anders Andren, Len Barrie, Terry
Bidleman, C. H. Chan, Walter Chan, Julian Chazin, Stanley Changnon, Jean
Dasch, Brian Eadie, Gary Eaton, Steven Eisenreich, Floyd Elder, Peter
Finkelstein, Donald Gatz, Gary Glass, Lennert Granat, Holly Hartrnann, Bruce
Hicks, Douglas Jones, Paul Kapinos, Barry Lesht, James Pankow, Francis
Philbert, Stephen Rothblatt, Herman Sievering, Doug Sisterson, Gary Stensland,
Charles Stearns, William Strachan, Malcolm Still, Tom Tisue, Robert Vet, Eva
Voldner, Tony Wagner, Dave Warry, and darvin Wesely. I thank all of them for
the time and help they gave me.
32
-------
VII. REFERENCES
Barrie, L. A. and R. J. Vet 1984. The Concentration and Deposition of
Acidity, Major Ions and Trace Metals in the Snowpack of the Eastern Canadian
Shield during the Winter of 1980-81. Attnos. Envir., 18, 1459-69-
Bidleman, T. F. 1985. Personnal communication.
Bolsenga, S. J. 1979. Determining overwater precipitation from overland data:
the methodological controversy analyzed. J. Great Lakes Res., 5,301-11.
Chamberlain, A. C. 1976. Response to "Approaches to Evaluating Dry Deposition
of Atmospehric Aerosols Pollutants onto lake Surfaces", by J. W. Winchester,
J. Great Lakes Res., 2, Suppl. 1, 38-41.
Changnon, S. A. 1981. Convective Raincells. J. Atm. Sci., 38, 1793-97.
Cole, H. S. and Lyons, W. C. The Impact of the Great Lakes on the Air
Quality of Urban Shoreline Areas: some practical applications with regard to
air pollution control policy and environmental decision-making. Proc. 15th
Conf. on Great Lakes Res., 436-63 (1972).
Davidson, C. I., S. E. Lindberg, J. A. Schmidt, L. G. Cartwright and L. A.
Lands 1985. Dry Deposition of Sulfate onto Surrogate Surfaces. J. Geophy.
Res., 90, D1, 2123-30.
Davidson, C. I. and S. K. Friedlander. 1978. A Filtration Model for Aerosol
Dry Deposition: Application to Dry Deposition from the Atmosphere. J. Geophy.
Res., 83, 2343-52.
Dasch, J. M. 1985 Direct Measurement of Dry Deposition to a Polyethylese
Bucket and Various Surrogate Surfaces. Env. Sol. & Tech., 19> 721-25.
Dolske, D. A. and D. F. Gatz 1985 A Field Intercomparison of Methods for the
Measurement of Particle and Gas Dry Deposition. J. Geophy. Res., 90, D1,
2076-84.
Eadie, B. J. 1985. Personnal communication.
Eadie, B. J., R. L. Chambers, W. S. Gardner and G. L. Bell 1984. Sediment
Trap Studies in Lake Michigan: Resuspension and Chemical Fluxes in the
Southern Basin. J. Great Lakes Res., 10, 307-21.
Eichenlaub, V. 1979. Weather and Climate of the Great Lakes Region. Univ.
of Notre Dame Press.
Elder, F. C. 1985. Personnal communication.
Feely, H. W., D, C. Bogen, S. J. Nagourney and C. C. Torquato. 1985. Rates
of Dry Deposition Determined using Wet/Dry Collectors. J. Geophy. Res., 90>
D1, 2161-65.
Gatz, D. F. Pollutant Aerosol Deposition into Southern Lake Michigan. Water,
Air and Soil Poll., 5, 239-51 (1975).
33
-------
Gatz, D. F. and S. A. Changnon 1976. Atmospheric Environment of the Lake
Michigan Drainage Basin. In: Environmental Status of the Lake Michigan
Region. Argonne National Laboratory, ANL/ES-40 Vol. 8
Gertler, A. W., J. G. Watson and C. I. Lin 1985. Methods to Estimate
Precision, Accuracy and Validity of Wet and Dry Deposition Measurements.
Abstracts, NADP Tech. Comm. Meeting, Fort Collins, CO, Oct. 811.
Glass, G. E; Leonard, E. N.; Chan, W. H.; and Orr, D. B. Airborne Mercury in
Precipitation in the Lake Superior Region. J._ Great Lakes Res., In Press.
Graustein, W. C. and Turkekian, K. K. 210pb as a Bracer of the Deposition of
Sub-Micron Aerosols. In: Precipitation Scavenging, Dry Deposition and
Resuspension, Pruppacher, H. R.; Semonin, R. G.; and Slinn, W. G. N. Eds.,
Elsevier, 1983, p. 1315-24.
Goodison, B. E., H. L. Ferguson and G. A. Kay 1981 Snow Measurement and Data
Analysis. In: Snow Handbook, D. M. Gray and D. H. Male Eds., Pergamon, pp
141-175.
Hicks, B. B. 1985. Results of the First Six Months of a Trial Dry Deposition
Network Operation. Abstracts, NADP Tech. Comra. Meeting, Fort Collins, CO,
Oct. 811.
Jones, D. M. A. 1969. Effect of Housing Shape on the Catch of Recording
Gages. Monthly Weather Rev., 97, 604-6.
Kerr, R. A. 1985. Tracking a Stormy Beast in the Night. Sci., 229, 848-9.
Kurtyka, J. C. 1953. Precipitation Measurement Study. 111. State Water
Survey, Report of Investigation #20, 178 pp.
Larson, L. W. and E. L. Peck 1974. Accuracy of Precipitation Measurements
for Hydrologic Modeling. Water Res. Res., 10, 857-63.
Liss, P. S. 1982 Gas Transfer: Experiments and Geochemical Implications.
In: Air-Sea Exchange of Gases and Particles, Liss, P. S. and Slinn, W. G. N.
Eds., Reidel, pp. 241-298.
Mackay, D. and A. T. K. Yuen 1983 Mass Transfer Coefficient Correlations for
Volatilization of Organic Solutes from Water. Envir. Sci. and Tech., 17, 211-17.
Rahn, K. A. and Lowenthal, D. H. Pollution Aerosol in the Northwest:
Northeastern-Midwestern Contributions. Science, 228, 275-284 (1985).
Rahn, K. A. and Lowenthal, D. H. Elementary Tracers of Distant Regional
Pollution Aerosols. Science, 223, 132-39 (1984).
Robertson, J. K. and J. W. Wilson 1985. Design of the National Trends
Network for Minitoring the Chemistry of Atmospheric Precipitation. U. S.
Geological Survey Circular 964, 46 pp.
Rothblatt, S. 1985. Personnal communication.
-------
Sievering, H. Small Particle Dry Deposition on Natural Waters: How large the
Uncertainly? Atmos. Envir., 18, 2271-2 (1984)
Sievering, H.; Crawley, J.; Ton., N. Undated. Final Report for GLNPO Grant
#R005697-01. GLAD Network Data Review and GLAD Bulk Sampler Considerations.
Slinn, W. G. N. Air-to-Sea Transfer of Particles. 1982 In: Air-Sea Exchange
of Gases and Particles, Liss, P. S. and Slinn, W. G. N. Eds., Reidel, pp. 299-
405.
Slinn, W. G. N. 1983. A Potpourri of Deposition and Resuspension Questions.
In: Precipitation Scavenging, Dry Deposition and Resuspension, Pruppacher, H.
R.; Semonin, R. G.; and Slinn, W. G. N. Eds., Elsevier, 1983, p. 1361-1416.
Slinn, S. A. and W. G. N. Slinn (1980). Predictions for Particle Deposition
on Natural Waters. Atmos. Envir. 14, 1013-1016.
Smith, J. H., D. C. Bomberger, Jr., and D. L. Haynes (1980) Prediction of
the Volatilization Rates of High-Volatility Chemicals from Natural Water
bodies. Envir. Sci. Technol., 14, 1332-1337.
Stall, J. B. and F. A. Huff 1971. The Structure of Thunderstorm Rainfall.
Meeting Preprt., ASCE Nat. Water Res. Eng. Meet., Phoenix, AZ, Jan 1115. 30 pp.
Stearns, C. R. 1985. Personnal communication.
Su, G. 1984. A Proposal for a Program to Study Atmospheric Loading of loxic
Chemicals to the^Great Lakes. Report to Great Lakes Environmental
Administrators. August. 6 pp.
Swain, W. R. (1978). Chlorinated Organic Residues in fish, water and
precipitation from the vicinity of Isle Royale, Lake Superior. J. of Great
Lakes Research, 4, 398-407.
Talbot, R. W. and A. W. Andren. 1983. Relationship Between Pb and 210Pb in
Aerosol and Precipitation at a Semiremote Site in Northern Wisconsin. J_._
Geophy. Res., 88, C11, 6752-60.
I'isue, T. and D. Fingleton 1983. Atmospheric Inputs and the Dynamics of
Trace Elements in Lake Michigan. In: M. Simmons and J. Nriagu, Eds., Toxic
Contaminants in the Great Lakes, Adv. Envir. Sci. and Tech. J. Wiley and Sons.
P. 105-126.
Tucker, W. A., Lyman, W. J. and Preston, A. L. 1983. Estimation of the Dry
Deposition Velocity and Scavenging Ratio for Organic Chemicals. In:
Precipitation Scavenging, Dry Deposition and Resuspension, Pruppacher, H. R.;
Semonin, R. G.; and Slinn, W. G. N. Eds., Elsevier, 1983, p. 1243-57.
Wesely, iyl. L., B. M. Lesht and D. L. Sisterson. 1985. Personnal communication,
Williams, R. M. 1982. A Model for the Dry Deposition of Particles to Natural
Water Surfaces. Atmos. Envir., 16, 1933-38.
Wilson, J. W. 1977. Effect of Lake Ontario on precipitation. Mon. Weather
Rev., 105, 207-2
35
-------
TECHNICAL REPORT DATA
(Please read lituruenons on the rri'rot before completing/
REPORT NO.
EPA-905/4-87-001
3. RECIPIENT'S ACCESSION" NO.
THTtE AND SUBTITLE
Design of a Great Lakes Atmospheric Inputs and Sources
(GLAIS) Network
&. REPORT DATE
1987
6. PERFORMING ORGANIZATION CODE
5GL
AUTHOR(S)
Thomas J. Murphy - DePaul University
PERFORMING ORGANIZATION NAME AND ADDRESS
DePaul University
2323 North Seminary Avenue
Chicago, Illinois 60614
8 PERFORMING ORGANIZATION REPORT NO.
GLNPO No. 87-03
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
R005818-01
12. SPONSORING AGENCY NAME AND ADDRESS
U.S. Environmental Protection Agency
Great Lakes National Program Office
230 South Dearborn Street
hicago, Illinois 60604
13. TYPE OF REPORT AND PERIOD COVERED
GLAIS 1985
14. SPONSORING AGENCY COD.E
Great Lakes National Progra:
Office, USEPA, Region V
m
15. SUPPLEMENTARY NOTES
Edward Klappenbach
Project Officer
16. ABSTRACT
This report is a document to aid in the design of a network aimed at determining
the identity, amounts, and sources of the materials coming into the Great Lakes
from the atmosphere. It is meant to discuss what is knoxm and what needs to be
known about sampling atmospheric inputs to lakes. It does not strive to be -thorough
and comprehensive as this is not possible for such a broad subject in a short
report. It is based on a variety of articles and reports, and relies heavily
on suggestions and discussions with a variety of investigators in different
disciplines. It assumes that experts in each of the necessary disciplines will
be consulted or will participate in the final netxrork design. Finally, it
reflects the strengths, weaknesses, biases and misspellings of the author.
Discussion among others is needed to fill it in and to round it off.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b. IDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
Atmospheric Inputs
Rain Gauges
Snow Gauges
Wet/Dry Samplers
Air Monitoring
Dry Deposition
18. DISTRIBUTION STATEMENT
Document is available to public through the
National Technical Information Service (NTIS
Springfield, VA 22161
19. SECURITY CLASS (This Report)
21. NO. OF PAGES
44
70 SECURITY CLASS (Thispage)
22 PRICE
EPA Form 2220-1 (9-73)
------- |