United States
Environmental Protection
Agency
Great Lakes National
Program Office,, ° -
536 South Clark Street
Chicago, Illinois 60605
EPA-905/9-84-001
February 1984
4>EPA
An Experimental Study
Of Lake Loading by
Aerosol Transport and
Dry Deposition in the
Lake Erie Basin
-------
EPA-905/2-84-002
June 1984
AN EXPERIMENTAL STUDY OF LAKE LOADING
BY AEROSOL TRANSPORT AND DRY DEPOSITION IN
THE LAKE ERIE BASIN
H. Sievering, D.A. Dolske, V. Jensen, R.L. Huges
Environmental Science Program
College of Applied Sciences
Governors State University
University Park, Illinois 60466
Project Officer: C. Risley, Chief
Surveillance and Research Staff
U.S. Environmental Protection Agency
GREAT LAKES NATIONAL PROGRAM OFFICE
REGION V
U.S. ENVIRONMENTAL PROTECTION AGENCY
536 South Clark Street
Chicago, Illinois 60605
-------
FOREWORD
This study was supported by a Great Lakes National Program grant to
Governors State University for investigating the rate of deposition
of atmospheric transported pollutants to Lake Erie. The R/V Carson,
a laboratory and lake water quality sampling ship, operated by this
office, was used to support this project. Data in this report covers
a period from June 1979 to July 1982.
-------
DISCLAIMER
This report has been reviewed by the Great Lakes National Program
Office, Region V, U.S. Environmental Protection Agency, and approved far
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
endorsement or recommendation for use.
-------
EXECUTIVE SUMMARY
A Lake Erie experimental program to assess the contribution of
atmospheric dry loading of aerosol trace elements and nutrients has
been completed. Trace elemental concentrations were determined by
atomic emission spectroscopy for seventeen elements. Nutrient species
phosphate and sulfate were determined by standard USEPA techniques.
The Lake Erie sampling program was successful, despite generally
poor weather conditions. However, any attempt to calculate atmospheric
dry loading must be viewed with skepticism. The very fact that unusual
weather conditions prevailed during at least two of the three Lake Erie
ship outings means that the data base is not climatologically
representative. Neither the aerosol elemental concentrations nor
the wind speed and direction data are sufficiently representative
to allow use of the Lake Michigan "binning" procedure (EPA-905/4-79-
016). For example, no high wind speeds (i.e., greater than 10 m s~ )
or thermally unstable conditions were encountered. Southwesterly
flow from the Cleveland source region was severely under-represented.
Also confounding, though, is the fact that rains (often times heavy)
and, therefore, wet soil/surface conditions prevailed during the Lake
Erie sampling program. These conditions caused the aerosol concentrations
for soil-derived elements (e.g., Iron) to be more than one order-of-
magnitude lower over Lake Erie than over Lake Michigan. These "rainy"
conditions must also have affected anthropogenic source elemental
(Pb, Zn, Cd, Cu, Cr, Ni--see Table next page for explanation of
symbols) concentrations. Yet the observed Zn, Cd, Cu, Cr, and Ni
ambient air concentrations are comparable to previous measurements
in the northeastern U.S. This, and the fact that Zn, Cd, Cu, Cr,
and Ni are suspected to be near to toxic levels in Lake Erie waters,
provoked the calculation of atmospheric loadings below--presented in
3 -1
metric tons per year (10 kg yr~ ). Results are shown along with
revised estimates for the southern basin of Lake .Michigan (EPA-905/4-
79-016).
m
-------
The range in values shown may be considered to be the ^25% a^d
confidence limits in the estimation of mean loadings. The full range
of uncertainty may be more than an order of magnitude. Thus, these
loading numbers must be "viewed with skenticisn" as stated above.
ATMOSPHERIC DRY LOADING, 103 kg yr"1
Element
Iron
Lead
Zinc
Cadmium
Copper
Chromium
Nickel
Sulfate
Phosphorus
(Tnt.?l-P)
(Fe)
(Pb)
(Zn)
(Cd)
(Cu)
(Cr)
(Ni)
(soj)
Lake Erie
170-330**
10-25
120-300
8-20
15-50
8-18
5-15
20-50x1 O3
< ]o**
Southern Basin
Lake Michigan
350-850
200-500
70-200
—
(10-20)*
—
—
3.5-llxlO3
75-150
Percentage Anthropogenic
Lake Erie Lake Michigan
>75 >65
>95 95
>97 >75
>99
%n5 (?)*
>80
^90
__
A
Parens indicate dubious nature of Lake Michigan Cu data because of
sampling contamination.
**
Low due to depleted soil source during Lake Erie sampling program.
A first point to notice is that numbers for Cd, Cu, Cr, and Ni are
available for Lake Erie but not for Lake Michigan. This is the result
of improved analytical detection limits as well as the fact that average
sampling times over Lake Erie were nearly twice as long as over Lake
Michigan. The Fe and total-P Lake Erie loadings are three- to ten-fold
lower than for Lake Michigan because of the markedly depleted soil source,
IV
-------
These two loading figures are certainly unrepresentatively low estimates.
The Pb, Cr, and Ni loading estimates may also be unrepresentatively low.
3
(A previous Lake Erie dry loading estimate [IJC, 1977] for Pb is 600x10
kg yr~ ; this value may be an over-estimate.) Clearly Pb loading, primarily
from automotive and other combustion sources, is reduced relative to that
for the southern basin of Lake Michigan, even considering the order-of-
magnitude potential uncertainty. In light of this, the estimated Zn
loading to Lake Erie is surprisingly large. Considering the large
uncertainty prevalent, the atmospheric dry loading of Zn to Lake Erie
and to the southern basin of Lake Michigan are about the same. In the
case of Lake Michigan, atmospheric Zn dry loading constitutes about one-third
of the total loading from dry and wet deposition and from runoff combined.
One may calculate the relative contribution of the soil- and lake-
derived natural aerosol sources versus anthropogenic aerosol sources,
on the assumption that all sources other than soil and lakes are
anthropogenic. Results of this calculation are shown in the last
two columns of the table. Since the percentage of anthropogenic Pb,
Cd, and Cu is comparable to that for Zn, it is suggested that a
substantial percentage—at least 10% to as much as 50%--of the total
Lake Erie loading for these elements may be due to dry deposition.
However, before any comparisons with wet deposition and runoff are
considered valid, a larger Lake Erie aerosol data base must be obtained.
For Zn and Cu alone it is estimated that 30 to 50% of their atmospheric
dry loading may occur during high wind speeds (i.e., greater than
10 m s~^). Yet no data were obtained during such conditions nor during
thermally unstable conditions (which prevail in winter months). It
is also impossible to use the combined Lakes Erie and Michigan data
base to further estimate loadings for others of the Great Lakes.
Given the difficulty in obtaining loading estimates for Lake Erie
a second look at the Lake Michigan data base was undertaken. In particular,
the chemical element balance (CEB) approach was used to evaluate the
percentage contribution to over-Lake Michigan aerosol mass by each of
seven source types: the lake itself, soil, surface excavation (e.g., rock
quarrying), iron/steel manufacturing, residential oil-burning, coal-
burning and automobile combustion. It was found that the lake may be
a large contributor to the total aerosol mass observed overlake although
-------
neither of the natural lake or soil sources is a major contributor to
the trace elements Cd, Pb, or Zn mentioned above.
The final area of research during this Lake Erie Grant program
was an effort to better parameterize aerosol deposition at the air/water
interface. During the Lake Michigan Grant program a simple estimate of
the deposition velocity (v,) at the air/water interface was made by
assuming it to be a product of wind speed and the drag coefficient.
One way to obtain a better parameterization is by estimating v, as a
function of smooth, moderate, and rough flow conditions at the air/water
interface. Consideration of these three distinct flow regimes results
in an average v, about half the value previously used. As a consequence,
the range in previously reported Lake Michigan loading values is nearly
one-half of that reported in EPA-905/4-79-016 . This approximately
twofold reduction is reflected in the Lake Michigan southern basin
loadings shown in the table in this Executive Summary.
VI
-------
CONTENTS
Page
Executive Summary ^^
List of Figures vjii
List of Tables .jx
1 . Introduction ]
2. Conclusions and Recommendations 5
3. Description of Experimental Program 7
3.1 Theory and Design 7
3.2 Execution of the Sampling Program 15
4. Data, Analysis, and Basic Results - ig
4.1 Aerosol Elemental Data 19
4.2 Deposition Velocity Parameterization
5. Interpretation of Results 35
5.1 Ship/Buoy Meteorological Data Intercomparison
and the Representativeness of Ship Sampling 35
5.? Aerosol Characteristics and the Estimation of
Atmospheric Dry Loading 40
5.3 Natural and Anthropogenic Source Contributions to
Great Lakes Atmospheric Dry Loading and the Future of
Atmospheric Loading Estimates 43
References 51
Appendices
A. Lake Breeze Effects on Particle Size Spectra and Sulfate
Concentration over Lake Michigan 53
B. Chemical Uniformity of Atmospheric Aerosol : Its Violation at a
Mid-Lake Erie Site on
C. Technical Note: Trace-Element Pass-Through for Cellulose
Impactor Substrates and Filters When Used for Aerosol Collection . 96
D. Chemical Elements in Atmospheric Aerosol Over Southern Lake
Michigan: The Contribution of the Lake Source -IQO
E. Some Effects of Wind-Shift on Over-Lake Turbulence and Aerosol
Deposition 138
vii
-------
LIST OF FIGURES
Number Page
1. Data Collection Areas for GALE, 1979-1980 3
2. Location of R/V Rachel Carson and CCIW Buoys 9
3. R/V Simons and Sampling Boom 11
4. Aerosol Trajectory Plot, Source Region Overlake 23
5. Aerosol Trajectory Plot, Cleveland Area Source Region 24
6. Summary of 1979 Plotted Trajectories, End-Point is Most Likely
Location 24-hr Back in Time 25
7. Aerosol Volume Distribution Plot for Overlake Trajectory Case . £8
8. Aerosol Volume Distribution Plot for Cleveland Source Region
Trajectory Case 29
9. Diagrammatic Representation of the Surface Layer Above the
Air-Water Interface: (a) Smooth flow, (b) Rough flow 31
10. Cumulative Frequency Plot of Ship-sampled and Lakes Erie and
Michigan C1imatological Wind Speed Conditions 39
VI
-------
LIST OF TABLES
Number Page
1. ICAP Blanks, Detection Limits and Typical Sample (yg 1~ ) . . . 20
2. Mid-Lake Erie Aerosol, GALE 1979 Sets 21
_3
3. Geometric Mean Concentrations, GALE 1979 Sets, ng m 26
4. Parameterization of Deposition Velocity, v , 34
5. ^ass Balance Estimate for Lake Erie 47
-------
SECTION 1
INTRODUCTION
The U.S.-Canada Agreement on the Great Lakes has, in the past
few years, stimulated research into the relationship between regional
air pollution and the lakes' water quality. A part of the USEPA
effort in this regard has been grant funding, through the Great Lakes
National Program Office (GLNPO), of studies addressing various aspects
of atmospheric-route loading. Projected future shoreline and inland
population gains, industrial development, and energy producing facilities
make the atmospheric route a critically important element in the
overall problem of Great Lakes pollution loading. The region
encompasses several intense urban-industrial air pollution source
areas. Prevailing winds often carry anthropogenic particles and gases
out over the lakes. A real potential exists for serious detrimental
effects on Great Lakes water quality.
In order to achieve accurate estimates of atmospheric-route
loadings, the processes of wet and dry deposition are probably best
considered separately. Bulk (wet plus dry) sampling presently gives
no better than an order-of-magnitude accuracy in the dry loading
estimation. Regarding small particles (the vast majority of anthropogenic
particles are small particles), sampling techniques that simulate
natural-surface particle collection efficiency do not yet exist. Wet
deposition loadings have been reliably known for years. And, dry
deposition loadings had been considered negligible until early modelling
estimates suggested dry loadings could be substantial (Winchester and
Nifong, 1971; Gatz, 1975; Sievering, 1975). A lack of knowledge
regarding temporal variability in ambient aerosol concentration and
deposition rate pointed out the need for environmental (field)
sampling as a means of decreasing the uncertainty in estimates of the
dry deposition loading rate. During a 1976-1979 study by Governors
State University (GSU) aerosol and meteoological data were collected
over mid-Lake Michigan (Sievering et al., 1979). Aerosol concentrations
and deposition rates were found to be strongly dependent on surface
layer meteorological conditions. The uncertainty in estimates of dry *
-------
2
deposition loadings for Lake Michigan was reduced; however, the
applicability of the conclusions of that study to the other Great
Lakes remained in question. This present work was funded in 1979
by USEPA-GLNPO, to complement and extend the results of the earlier
Lake Michigan work by GSU. A sampling program using identical or
compatible methods was carried out during the summer of 1979 on
Lake Erie. Ultimately, the conclusions of this work, in conjunction
with Lake Michigan results and those obtained from the Great Lakes
Atmospheric Deposition (GLAD) network, should lead to a more generalized
understanding of aerosol transport and deposition in the Great Lakes
region, as well as a better estimation of atmospheric-route Great
Lakes pollution loading.
The principal objectives of the present work on Lake Erie were
to obtain an overlake data base and to relate aerosol chemical
constituent loadings to climatologically-expected meteorological
conditions over Lake Erie.. In order to approach these objectives, a
data collection program was devised that covered a wide range of
spatial and temporal scales—see Figure 1. Aerosol monitoring, at
the smallest space scales, included Active Scattering Aerosol
Spectrometer (ASAS) aerosol number concentrations, condensation nuclei
counter (CNC) number concentrations, and integrating nephelometer (IN)
back scattering coefficient. Hi-volume samplers collected particles
during filter exposures which averaged about six hours. Cascase
impactors and several types of filter material were used to provide
appropriate media for the several chemical analyses that were done.
A separate pass-through experiment using hi-volume air sampling at
GSU and the 68th Street Crib near Chicago was performed during 1980
in order to determine the percentage of Pb, Zn, and Fe lost (passed
through) by hi-volume filter sampling. Results of the pass-through
experiment (PTE) are used to correct trace metal concentration data
obtained during Lake Erie hi-volume air sampling and, also, Lake
Michigan sampling.
A second major data collection thrust was meteorological in
character. Micrometeorological information was collected and processed
by GSU. Larger space- and time-scale meteorological data were obtained
-------
Space
Scale, meters
10
10
10
io-24
ASAS--Active Scattering Aerosol Spectrometer
CNC—Condensation nouclei counter
IN--Integrating Nephelometer
v .--Deposition velocity
Synoptic Meteorology
& Lake Erie Climatology
f1e?ometeorology
Micrometeorology
Air-Water Interface
v, dynamics
IN
CNC
AS AS
Hi-Volume Sampling
Pas's-through Experiment
10
-2
10
-1
1 10 ItT 10J
Time Scale, minutes
Figure 1. Data Collection areas for GALE, 1 979-1 980,
10
-------
from the U.S. Coast Guard (USCG) and National Weather Service (NWS)
stations around the lake, Canada Centre for Inland Waters (CCIW)
Lake Erie buoys, and Atmospheric Environment Service (AES) shore
stations. Correction and reduction of the raw meteorological data
was performed at GSU. Climatological data were obtained from the
National Oceanic and Atmospheric Administration (NOAA Summary of
Synoptic Meteorol. Obser. for Great Lakes Areas, 1975).
Information collected at each of these scales (Figure 1) is
necessary to expand our understanding of the complex aerosol transport
and deposition processes that occur over the lake. Chemical
characterization of the aerosol, construction of backtrajectory plots,
calculation of estimated deposition velocities, and the relation
of those factors to Lake Erie Climatological data ultimately approach
the objectives of this study.
-------
5
SECTION 2
CONCLUSIONS AND RECOMMENDATIONS
The loading of Great Lakes' waters by air pollutants is an
extremely difficult process to thoroughly understand on qualitative
grounds let alone to establish quantitatively. Yet, a number of
conclusions can be drawn from this Lake Erie study:
1. Single point (shipboard) over lake sampling can be made to
be representative of an entire Great Lake's meteorological
conditions to better than ± 15% accuracy.
2. Insufficient aerosol elemental and compound data were obtained
over Lake Erie to afford the detailed calculation of loading
estimates by the procedure used in the Lake Michigan study
(EPA-905/4-79-016). However, rough estimates of Cu, Zn, Cd,
Cr, and Ni loadings are warranted given their potential for
negative ecological impact.
3. Extrapolation and interpolation of Lake Michigan or Lake Erie
aerosol elemental concentrations to others of the Great Lakes
is not warranted. Long-range aerosol transport nay veil have
markedly changed the absolute as well as relative concentrations
of trace elements over Lake Erie. It would be better to monitor
the anthropogenic contribution to Great Lakes' loading at a site
not dominated by long-range transport that is, yet, representative
of overlake conditions on the Great Lakes. These conditions
were met at the midlake site on southern Lake Michigan but not
by the mid-Lake Erie nor the Lake Michigan nearshore (water
intake crib) site sampling.
4. In light of 3, the Lake Michigan data have been reconsidered to
review lake effects upon aerosol transport and, especially, trace
element concentrations. This reconsideration of Lake Michigan
concentrations has also included the determination of aerosol losses
in the sampling apparatus used over Lake Michigan to then arrive at
revised atmosphere dry "fading estimates for Lake Michigan.
-------
Anthropogenic, soil and lake source contributions to over Lake
Michigan aerosol concentrations have been preliminarily specified
by using a chemical element balance technique. Results indicate
an unexpectedly large lake source contribution to the total aerosol
mass. This was found to be true for both fine and coarse aerosol
particulate matter. However, the contribution of steel manufacturing
and surface mining industries and of coal burning and the automobile
were by no means insignificant. In fact, anthropogenic contributions
(in parens)--at the Lake Michigan nearshore site were found to be:
Pb (99%), Zn (95%), and Mn (90%). It would appear that an intensive
aerosol sampling, multi-element analysis and, especially, a
multivariate statistical interpretation (including uncertainty)--in
concert with a detailed meteorological/climatological assessment--
may best ultimately unravel anthropogenic versus natural loadings.
More important, anthropogenic contributions by source type to
overlake aerosol concentrations (with < 50% uncertainty) may result.
The accurate determination of loadings to any of the Great Lakes
must await a reasonably accurate parameterization of aerosol
deposition velocities (v.) as-a function of aerosol particle
size (or chemical element), of wind speed and direction, of
thermal stability and of temporal or spatial changes in these
parameters. Our understanding of over-water v, dependence on these
parameters is extremely limited and warrants the consideration of
research specifically designed to determine these dependencies.
-------
SECTION 3
DESCRIPTION OF EXPERIMENTAL
3.1 THEORY AND DESIGN
In order to illustrate the basic concepts and data-collection
requirements of the experiment, the simple relation
F = vd . C (1)
may be used. F is the mass flux of particles to the lake surface
(kg m"^ s ~^), v, is the deposition velocity (m s ), and C is the
-3
concentration (kg m" ) of particles in air near the surface. Note
that both v, and C are complicated functions of many environmental
factors, as discussed further in this report. However, equation 1
does show that there are two main areas to investigate when estimating F.
Particle and interface physics and micrometeorological factors
enter into the determination of v,; aerosol .chemical transformations
and the mechanics of particle transport downwind of source regions
affect the determination of C. Finally, the climatological record
is used to normalize the temporal variability in F to expected mean
annual conditions, yielding annual total dry deposition loading
estimates. Overlake climatology is thus an important -factor in the
determination of annual loading rates (see Section 5.1). In order
to relate the limited number of sampling hours (approximately two
percent of the year) to the annual total loading, data sets taken in
periods of similar surface layer conditions are computationally grouped
into "bins."a Each bin is defined by a range of values for one of
three parameters: thermal stability, wind speed and source region,
the last identified through back-trajectory analysis. For each of
the j bins the annual frequency of occurrence, f , is determined for
the range of the defining parameters. The climatological denendence
of v, and C are then calculated from: the NOAA Summary of Synoptic
Meteorological Observations for Great Lakes Areas (NOAA, 1975).
-------
Annual Dry Loading = E vd1 . C. . f . . A . -^7— (2)
j ^ j oj
where: v ,. is the mean v, for aerosol data sets in the jth
dj d
bin (m s~'); aerosol (chemical constituent)
C. is the mean aerosol (chemical constituent) concentration
1
for data sets in the jth bin (kg m~6};
f . is the annual frequency of occurrence for the defining
range of the three parameters—thermal stability, wind
speed and source region;
A is the total water surface area of the Lake Erie basin
(m2);
and t is that part of the year when no precipitation occurs
overlake (s yr~').
Further discussion of basic concepts of the experimental program may
be found in the final report of the 1976-1979 Lake Michigan study
(Sievering et al., 1979).
Meteorological data were collected from several U.S. and Canadian
shoreline stations and buoys in Lake Erie. These data were used
extensively in the construction of aerosol back-trajectories, by the
methods of Sievering et al. (1979). In conjunction with climatological
information (NOAA, 1975), an attempt was made to estimate the representa-
tiveness of the year 1979 and the limited sampling periods. The
meteorological data used were: Canadian climatological station data
for Long Point and Port Col borne, Ontario; Canadian surface data for
London and Dimcoe, Ontario; three Lake Erie CCIW buoy locations: Buoy #1
(42° 07' 42" N, 81° 28' 42" W), buoy #2 (42° 3R1 45" N, 79° 56' 15" H),
buoy #3 (41° 44' 25", N, 82° 27' 04" W)--see Figure 2; meteorological data
from ship outings qf the C.S.S. Limnos.
NWS surface meteorological data were for Erie, PA, Flint, MI, and
Cleveland, OH; airport surface meteorological data for Burke Lakefront
Tower, Cleveland, OH, Pittsburgh International Airport, Buffalo Inter-
national Airport, and Toledo Express Airport; Coast Guard Marine Weather
Log for Ashtabula Light Station, OH covering the dates of the Lake Erie
cruises, July 22-28, Sept. 4-8, and October 16-20, 1979. We also obtained
upper air soundings from the NWS stations at Buffalo, NY, Dayton, OH,
Flint, 141, and Pittsburgh, PA.
-------
Pomte-aux-Pins
CCIW #1
A
+R/V
CARSON
/ l
40°
9 I
f D / \/ D a /- ho 1 - Ta
I/ m MMT n -F Tl Q \/Q 1 a n H f/19°nn'Ni Ql ° TH ' l,| \ 3 n rl rril.l Kuni/c AT
-------
10
Micrometeorological data and aerosol samples were collected from
aboard the USEPA's R/V Rachel Carson, anchored on station at 42° 00' N,
81° 30' W (Figure 2). This station is 55 km NNE of Cleveland, OH in
the central basin of the lake, 2 to 5 km south of primary shipping
lanes, and 42 km SE of Pointe aux Pins, Ontario. Thus, the sampling
site was removed from major aerosol sources on shore, and upwind of
the local (ship traffic) sources for a large percentage of on-station
sampling time. The midlake sampling site has the advantages of:
micrometeorological conditions that are representative of a large
part of the lake's surface, long over water upwind fetches that enhance
the likelihood that steady-state surface layer conditions prevail,
and well-mixed aerosol populations in which no single anthropogenic
source or pollution event is likely to dominate. If the lake surface
appears constant to the air passing over it, the mass flux, F, within
a well-developed surface layer is very likely to be constant throughout
the vertical extent of that surface layer (Kraft, 1977).
To ensure that aerosol samples and meteorological data were not
contaminated by the presence of the ship itself, all sampling was done
in a bow-anchored mode. In that configuration, the ship faced bow-first
into the wind, ± 15°. An open-beam lightweight aluminum boom extended
6 m ahead of the bow, at approx. 5 m above the mean water level
(Figure 3). Meteorological sensors were placed at the forward
end of the boom. Studies of bluff-body turbulence effects
by Hunt and Mulhearn (1973) indicate that the 6 m boom was sufficiently
long to avoid turbulence effects due to the ship structure. Turbulence
intensity, I, = a /u, where u is the running mean of wind speed and
u u
a is the standard deviation about the mean, measured at the forward
tip of the boom, ranged from 0.05 to 0.18, with a mean value of 0.11.
This is comparable to the 0.02 to 0.14 observed by SethuRaman and
Tichler (1977) from an air-sea interaction tower. Thus, very little
increased turbulence was observed at the boom tip.
Two hi-volume air samplers were located on the boom, aft of the
meteorological sensors. The hi-vols were 3 and 4 m ahead of the ship's
bow. The filter holders and heads were silicone-rubber sealed to the
-------
FIGURE 3: R/V Simons and Sampling Boom
-------
12
pump motors. The pump exhausts were vented through 4-m long, 11-cm
diameter plastic hoses which hung below the boom. This was done to
prevent re-entrainment of the exhaust and contamination of aerosol
samples. All hi-volume sampling was conducted either bow-anchored
on station or, occasionally, while underway upwind. The possibility
of contamination by the ship's diesel generator and engine exhausts
was minimized. Whenever local wind direction (WD) shifted quickly
to the point where the ship did not remain bow-first to the wind, hi-vol
sampling was avoided. Continuous WD traces from the boom tip vane
indicate that the boom remained ± 15° of directly upwind during all
sampling periods.
Shipboard sampling was performed around the clock. In order to
divide the data into manageable time segments, aerosol collection was
kept to within 4-to 8-h segments. Each data set, because of this limited
duration, correpsonds to a period of constant and readily characterizable
meteorological conditions. The occurrence of rain intermittently
precluded sampling during one cruise. At the beginning of each data
set, all aerosol filters and impactors were replaced and bulk water
samples were collected. Thus, the samples for chemical analyses can
be classified according to values of meteorological parameters that
prevailed during each data set. These results are then further
classified in the context of climatologically expected frequencies-
of-occurrence for those meteorological conditions.
In general, a minimum of 4-h elapsed time was used per data set
in order to collect enough material on the hi-vol filters to be above
the detection limits of the chemical analytical procedures. Past that
minimum, rapid and persistent changes in the 5-m height monitored wind
speed (Uc), WD, or thermal stability conditions were cause for the
U
termination of a data set. Such changes were defined as two successive
15-minute mean values for any parameter being more than 1 a removed
from the running mean of that parameter. A maximum time of ^ 8 h was
allowed so that no data set was biased by a widely differing run time.
Aerosol sample preparation and chemical analyses followed the procedures
of Sievering et al . (1979).
-------
13
Three optical aerosol sensors were located at the aft-most end
of the boom. An Active Scattering Aerosol Spectrometer (ASAS) monitored
aerosol number concentration in 60 size channels between 0.1 ym and
3.5 urn diameter. An Integrating Nephelometer (IN) gave a continuous
output of aerosol backscattering coefficient, b ., which may be
transformed into a measure of aerosol mass (M) concentration using
the formula b . = 0.3 + 0.014 M (Dave' et al ., 1979). Finally, a
scat
condensation nuclei counter (CNC) provided relatively fast-response,
large increase nuclei count indications of local aerosol sources
(e.g., passing ships), in addition to monitoring ambient nuclei
number concentrations.
In addition to the boom-mounted instruments, the forward part of
the foredeck was the site of aerosol sampling. Two hi-volume samplers
with type A glass fiber filters were run on overlapping schedules to
collect aerosol mass concentration data. A third hi-vol on the foredeck
was used with membrane filters to collect particles for optical analysis.
Surface layer thermal stability conditions were monitored by
detailed measurement of vertical temperature gradients through the use
of an array of matched and carefully intercalibrated fan-aspirated
composite thermistor probes. These probes were mounted on a light mast
at the bow, 10, 7, and 5 m above mean water level. The sensors were
referenced to duplicate probes at the 5-m level. Statistically
significant temperature gradients were observed during some data sets.
Additional points in the vertical temperature gradient were measured
by a hand-held infrared thermometer (IR) and a bucket thermometer.
The IR was used to scan an area of the lake surface below the boom
and, thereby, measure a surface temperature, T . The bucket thermometer
measured lake water temperature at Mem depth, T . The set and drift
of surface water currents were calculated by observing the movements of
drifters with an optical rangefinder, stopwatch, and hand-bearing compass,
Air movements above the near surface layer at the ship were crudely
observed during daylight hours by tracking 39-g pibal (PTE) balloons
with a prism-viewing hand-held compass.
-------
14
The pass-through experiment (PTE) was conducted during April to
August, 1980 at Governors State University and at the City of Chicago
68th Street water intake crib on Lake Michigan. The PTE was designed
to estimate, operationally, the overall particle collection efficiency
of the modified Sierra 1, 3, 6 impactor using cellulose substrates/filter,
the impactor/filter combination used during all Lake Erie and Lake Michigan
sampling. Meteorological parameters were monitored during sampling, and
continuous detailed records of temperature, relative humidity, and b ,
scat
were kept.
Collection efficiency of the cellulose filter material for Pb, Fe,
and Zn relative to the standard type A glass fiber filter was determined
by running different multiple-filter assemblies on two hi-vols. On the
first hi-vol , a type A glass fiber filter was placed 1 cm above a low-trace-
element content (spectrograde, SG) glass fiber filter. The type A was
used to gravimetrically find total aerosol mass concentration. The SG
filter was hot acid leached (EPA procedure) and analyzed for Pb, Fe, and
Zn by AA. On the second hi-vol of the simultaneously sampling pair, a
modified Sierra 1, 3, 6 stage cascade impactor was used with cellulose
impaction substrates filter. An SG filter was placed in line 1 cm
behind the impactor. All four stages were prepared for Pb, Fe, Zn
analysis.
The collection efficiencies determined here are relative to the
type A because it is assumed that type A glass fiber is 100 percent
efficient, i.e., lets no particles pass through. If this is true, or
nearly so, then the SG backup of a type A should show concentrations
near filter-blank level. The SG backup behind the less than 100 percent
-------
15
efficient cellulose filter should show concentrations at some level
inversely related to the collection efficiency of the cellulose filter.
Such multi-filter simultaneously collected pairs of filter sets were
run on co-calibrated hi-vols. Filter exposure was at a flow-controlled
rate of 1.13 m min~ (40 SCFM) for periods ranging up to 72 h. This
length of time was required to allow sufficient material to pass through
to the final SG backup for the chemical analysis.
3.2 EXECUTION OF THE SAMPLING PROGRAM
During the summer season, 1979 three cruises were conducted on Lake
Erie. The USEPA R/V Rachel Carson anchored on station at 81° 31' W,
42° 00' N for several days during each cruise. A summary of shipboard
sampling activities and description of meteorological conditions
encountered follows:
The first cruise began 23 July 1979, with R/V Carson on site at
2300 CDT. The evening of 23 July 79, a weak surface front was situated
north of Lake Erie. This was displaced to the northeast with the
movement of an associated low pressure area. The morning of 24 July 79
was 80 percent overcast. There was some clearing at mid-morning, and
by 7 PM CDT sky cover was only 50 percent. Several episodes of light
rain interrupted sampling the evening of 24 July 79 through the next
morning. By daybreak on 25 July 79, there was 90 percent cloud cover.
Again, several episodes of rain occurred during the late afternoon and
evening of 25 July. An associated low pressure system moved from
northern Lake Michigan on the morning of 25 July 79 to the Quebec area
by 26 July 79. This system trailed a cold front which passed through
Lake Erie sometime earlier on the 26th. Ahead of the cold front passage,
there was an increase of wind speed and the sky was entirely overcast
-------
16
the morning of 26 July 79. The sky began clearing 0900 CDT on 26
July 79. The night was very clear 27 July 79 and no further sampling
was conducted.
Of the 78 hours on-site, about 72 hours were devoted to actual
sampling. Four to 5 hours of the inactive time was due to periods of
rainfall, during which all sampling temporarily halted; the remaining
time was used in adjusting equipment and changing Hi-vol filters.
Results from the aerosol data analysis indicate that the cruise covered
two distinct aerosol regimes. Early in the cruise, the site was generally
downwind of the Cleveland area and moderate aerosol mass concentrations
(around 40 yg m" ) were encountered. After passage of a cold front later
in the cruise, the shift in winds brought about much lower aerosol
concentrations, by two-fold. A total of twelve data sets were collected,
each consisting of size-fractionated aerosol samples for metal and
nutrient analysis, a bulk water sample, as well as physical aerosol and
meteorological data.
This cruise period was one of generally steady, moderate winds in
the range of 3 to 6 m s . Atmospheric temperature stability throughout
the period may be characterized as neutral to slightly stable,
AT = (Tai> - TQ) = 0.1 to 1.5°C.
The second cruise began 5 Sept 79. This cruise was shortened
first by late arrival of the ship at Cleveland and later by heavy
weather encountered 7 Sept 79. The weather during the entire
second outing was influenced by the presence of hurricane David.
On the morning of 5 Sept 79, Lake Erie was under the influence of a
broad Great Lakes high pressure area. By the time sampling was started,
winds at the sampling site were from the ENE "at about 5ms". High
-------
17
cirrus provided a 90% cloud cover and the horizon was very hazy. Winds
dropped to about 3ms during the night and shifted to the north--
but as David progressed up the East Coast, the speed increased steadily
reaching 7 m s~ . The wind direction became progressively more westerly,
being almost due west at 1200 CDT on 6 Sept 79 and WSW by the early
afternoon. Some light rain was encountered during the late afternoon
of 6 Sept 79.
The windspeed began to drop again during the evening and the wind
direction moved to NNW. Winds were too light to do any sampling from
about 0000 CDT to 0100 CDT on 7 Sept 79 when the wind picked up from
the NNW. Near sunrise the sky cover was about 30 percent. This cover
increased through the remainder of the day, becoming about 80 percent
in the afternoon. Visibility was, however, exceptionally good. Near
1200 CDT on 7 Sept 79, the wind shifted to the WNW and began to increase.
By 1500 CDT, winds were over 10 m s
Of the 49 hours on-site, 45 hours were devoted to actual sampling;
there was a total of 3 hours of rain delays and low wind speeds when
no sampling could be done. Less than one hour was necessary for on-site
equipment adjustments and filter changes. A total of eleven data sets
were collected on this cruise. This period was one of prevailing winds
from the northern shore of Lake Erie, and aerosol mass concentrations
_3
of 10-25 yg m were indicated by the Integrating Nephelometer.
This cruise ended short of the time scheduled. At about 1400 CDT
on 7 Sept 79 the wind speed sharply increased from a 5-6 m s" range
to a 10-13 m s range; wave heights promptly increased from less than
1 m to 3-5 m. For reasons of personnel safety and prevention of
permanent damage to equipment the ship weighed anchor and departed
station at about 1700 CDT on 7 Sept 79.'
-------
18
The third and last cruise on Lake Erie began 16 Oct 79, with the
Carson arriving on site at about midnight. At the beginning of the
cruise, Lake Erie was Southeast of a rapidly occluding front. This
brought small amounts of scattered rain and a generally overcast sky.
Winds were 5-6 m s" from the SW. As the system passed over the lake,
winds dropped to 2 m s with a wind shift to the NW accompanied by some
clearing. By 2100 CDT on 16 Oct 79, winds had shifted to ENE and were
increasing. 17 Oct 79 was overcast except for a brief period of partial
clearing in the morning. That evening, the lakes began to feel the
influence of a well-developed low pressure system west of the lakes.
The leading warm front swept over Lake Erie that night. The pressure
dropped rapidly and winds increased dramatically to 8-10 m s . The
morning of 18 October 79, the sky was clear, but overcast again by 1100 CDT.
All of the 65 hours on-site were spent in sampling. Mo rain
delays were necessary despite scattered showers in the area; overlapping
aerosol data sets contributed to the 100% efficiency of on-site sampling
time. A total of thirteen data sets were collected. Aerosol concentration
-3
during this cruise period was moderate, in the 30-80 yg m range. Wind
speed and direction were, as previously described, quite variable.
Atmospheric temperature stability was neutral to unstable, AT = 0 to 0.2°C.
-------
19
SECTION 4
DATA, ANALYSIS AND BASIC RESULTS
4.1 AEROSOL ELEMENTAL DATA
One immediately obvious factor in examining the aerosol trace
element data is that concentrations over Lake Trie were relatively
low, compared to expectations based on Lake Michigan data. Despite
longer average filter run times during the Lake Erie cruises, almost
all samples initially fell below the inductively coupled argon plasma
atomic emission spectroscopy (ICAP) detection limits set by USEPA-CRL.
Uoon careful examination of the raw (no detection limit) ICAD data
a SEPA rate series of reagent blanks, filter blanks, field blanks,
and spike samples were calculated. This led to the determination of
detection limits (c + 3-a of lab DI water samples) for this group of
samples far below the USEPA-CRL values. That calculation resulted in
the recovery of many data points which had been thought to be lost in
the ICAP instrumental noise. The filter field blank corrections (c + a)
were also determined and analyses of replicates indicate a mean
reproducibility of about 20 percent or better for ten elements, even
at these low levels (Table 1). Typical sample values, also in Table 1,
were derived from overall averaged concentrations and run times.
Because the typical sample concentrations are not always too far above
the filter blanks, it is difficult to justify, with statistical
significance, the examination of the ICAP data in individual data set
by data set cases. It is more reasonable to aggregate several data
sets, based o.n similar values of defining parameters, and discuss mean
concentration values for the group of sets. Recovery studies shov/ that
>90/o for all except Al , for which 50-70% recoveries were determined. Al
concentrations determined by ICAP were enhanced by a factor of 60% to
compensate for its poor recovery.
Table 2 is based on all data sets collected at mid-Lake Erie
during the summer and fall of 1979. The overall geometric mean
concentrations are given for those elements which appeared above
detection limits and blank corrections in 75 percent or more of the
data sets. These mean concentrations are generally a factor of three
to eight lower than the concentrations observed over Lake Michigan
during 1977.
-------
20
Table 1. ICAP BLANKS, DETECTION LIMITS,
AND TYPICAL SAMPLE (yg I"1)
Al
Ba
Cd
Cu
Cr
Fe
Mn
Ni
Pb
Ti
In
Field Blank
(c + l»a)
85
12
2.8
44
6.6
180
4.7
0.3
n
2.9
59
Detection Limit
(c + 3« a)
0.8
0.1
0.3
2.5
0.1
2.0
3.4
3.3
7.6
0.3
9.0
Typical
Sample
180
20
22
105
18
380
15
42
32
12
330
-------
21
Table 2. ^ID-LAKE ERIE AEROSOL, GALE 1979 SETS
Al
Ba
Cd
Cu
Cr
Fe
Mn
Ni
Pb
Ti
Zn
C, ng-m
35*
1 .2
2.0
10.4
1 .8
46
0.8
3.0
3.5
1.0
37
F/C1
1.9
4.8
7.4
4.3
3.4
2.0
2.2
2.6
6.2
1 .2
6.3
M%
0.1
.004
.006
.03
.004
0.13
.003
.01
.01
.003
0.11
EF**
= 1 .0
8.0
2900
1100
3.3
2.4
2.0
140
720
- 0.8
1500
pp **
=1 .0
3.5
1800
120
n.d.
2.1
5.9
230
3200
0.9
320
Includes 60% enhancement due to poor analytical recovery,
4-
'Corrected for pass-through filter loss.
**
Approximate due to Al recovery problem.
-------
22
It is important to interject here the results of the pass-through
experiment (PTE). It was found that 37 ± 7% of Pb, 32 ± 6% of Zn
and 12 ± 3% of Fe collected on the backup filter passes on through that
filter. A significant correction factor must, then, be applied to
the observed concentrations. This has been done in Table 2 by accounting
not only for pass-through losses but also the 5% cascade impactor wall
loss of small particles identified by the impactor manufacturer (Sierra).
Other parameters describing the composite of the aerosol also seem quite
interesting for Lake Erie. Values of the fine to coarse (F/C) aerosol
ratio are calculated as
= c (backup filter, D < 1 .0 ym)
c (1st stage impactor, D > 1 .0
The elemental mass percent, (M%),
„„/ _ c (trace element)
B (total aerosol )
values are much lower than the corresponding values for the Lake Michigan
aerosol. However, enrichment factors, EF, based on Bowen's (1966)
midwestern soils with Al as the indicator element, correspond fairly well
with the Lake Michigan aerosol data (Table 2). This may suggest some
similar source types for both sites, with added source modification or
transport-dependent processes affecting the Lake Erie data.
In order to examine the differences in aerosol composition parameters
more closely, back-trajectory plots were used to group data sets having
similar source regions. Figures 4 and 5 are examples of the trajectory
plots, showing horizontal extent and vertical depth of the most probable
source regions for each set. Figure 6 is a summary of all plotted
trajectories, showing the most likely 24-hr averaged plots. From these
data, each of the sets was grouped into one of four general source regions.
On the U.S. side of the lake, all non-Cleveland area trajectories were
labelled US-rural. All sets from the Canadian side were called Canada-rural,
while all trajectories that did not reach shore within 24-hr are "Lake"
sets. Finally, those data sets obtained while the ship was directly
downwind of Cleveland were grouped together. The approximate ship-to-
shore compass sectors for each of these four source regions is: US-Psural ,
60°-190°; Lake, 240°-260° and 20°-60°; Canada-Rural, 260°-20°-, and
Cleveland, 190°-240°. The mean concentration data (Table 3) for each
-------
LONDON
BUFFALO
ACCIW #2
A
CCIW #1
A
R/V
CARSON
42
42°
CCIW #3
0 4
Aerosol Trajectory Plot,
Source Region Overlake.
CLEVELAND
PITTSBURGH
/ I
40°
-------
LONDON
BUFFALO
ACCIW
CCIW #1
A
R/V
CARSON
42°
CCIW #3
0 A
Aerosol Trajectory Plot,
Cleveland Area Source
Region.
CLEVELA*
50 75 100
Kilometers
PITTSBURGH
O
/ i
40'
-------
1979
84
83
42C
Summary of 1979 Plotted
Trajectories, End-Point
is Most Likely Location
24-hr Back in Time.
41 <
-1 1 1
1 1
f
PITTSBL
O
RfiH ' — I I — 1 1 — > l
nun i i i i i
0 50 75
Kilometers
|
100
82
81
80°
79°
/ I
40°
-------
26
Table 3. GEOMETRIC MEAN CONCENTRATIONS,GALE 1979 SETS, ng m"3
AT
Ba
Cd
Cr
Fe
Mn
Ni
Pb
Ti
Zn
US-Rural
27
3.3
1 .9
0.6
17
1 .0
0.7
2.2
0.3
18
Lake
26
0.5
0.4
0.8
30
0.8
0.4
2.6
0.3
36
Canada-Rural
25
1 .4
0.1
4.1
59
0.3
0
0.2
0.3
13
Cleveland
55
2.4
3.6
1.5
74
0.8
0.7
11 .
1 .0
147
-------
27
of these source regions contains no surprises. That is to say, the
Cleveland source clearly dominates, especially given the prevailing
winds in this area. Note that the mean concentration of Pb and Zn in
the Cleveland source is more than twice that from all three other
source regions combined. Additional information about each source
region is obtained from the ASAS data. Figures 7 and 8 are plots of
aerosol volume distribution, from the ASAS data, which show distinct
differences between the lake and Cleveland source regions. The lake
region source has a clear bimodality suggesting that two separate source
types contribute, the small aerosol (r of 0.1-0.2 ym) source may be
similar to the Cleveland aerosol source. The large aerosol (r of 0.4-0.5 pm)
source may be the lake itself. A discussion of the Great Lakes as an aerosol
source is found in Appendix D.
4.2 DEPOSITION VELOCITY PARAMETERIZATION
The largest single uncertainty in Great Lakes atmospheric loading
estimates—despite efforts to reduce it—remains the estimation of a
deposition velocity, v.. Progress over the last two years has been made
on the modeling of the deposition process and some data on deposition
velocity as a function of aerosol diameter, v , (D) were obtained over Lake
Michigan. Yet, the uncertainty in v, (D) is still nearly an order of
magnitude for 0.02 < D < 2 ym. Indeed, the summary table in the Executive
Summary of the Lake Michigan final report (EPA 905/4 - 79 - 016) would
be better to have the columns "minimum dry deposition" and "mean dry
deposition" retitled to read "moderate estimate" and "high estimate,"
respectively. This less accurate statement of loadings is the result of
a more sophisticated understanding of the deposition process obtained by
our parameterization efforts as well as the modeling efforts by the SI inns
and by Hicks and Williams. SI inn and SI inn (1980) have assessed the
contribution of particle growth in high RH environments and Hicks and
Williams (1980) have added the effects of diffusion, sedimentation and
growth to their own analysis of whitecapping and subsequent lake capture
of airborne particles by reducing surface transfer resistance to zero
over that percentage of the surface where whitecapping exists. Both
of these modeling efforts suggest a fairly strong dependence of v. on
-------
28
Set 01130 (LAKE)
AV
jum3
cm 3
10
10
0.1
1.0
particle radius, fim
Figure 7. Aerosol Volume Distribution
Plot for Overlake Trajectory
Case.
-------
29
Set 01180 (CLEVELAND)
AV
jjm3
cm 3
10
0.1
1.0
particle radius,
Figure 8.
Aerosol Volume Distribution Plot for Cleveland Source
Region Trajectory Case.
-------
30
D in the 0.1 < D < 1 ym particle size range, as have previous modeling
efforts. Our parameterization has accounted for what meager data is
available on v. (D). Limited profile sampling while aboard the R/V
Simons on Lake Michigan in 1977 showed that v . varied by only two- to
threefold in the 0.1 < D < 1 ym particle size range. It was also found
that the ratio of v. to the aerodynamic transfer velocity (transfer
controlled by turbulence) was 1/3 to 1/2 for eleven moderate wind speed
(2.4 < u < 8.2 m s ) profile data sets.
Any mechanism which results in a small variation in aerosol v. for
0.1 < D < 2 yn is a candidate for explaining the profile results above.
Conditions near to the air/water interface are portrayed in Figure 9,
Here, the surface layer is seen as three seoarate zones: a turbulent
layer, a buffer layer and a viscous sublayer. The turbulent layer
constitutes the largest portion of the surface layer depth of several
meters to several tens of meters and here v"J = r . A continuous viscous
sublayer enveloping all or nearly all of the surface roughness elements
follows the gross observable contour of the water's surface. Mass
transfer within the viscous sublayer is dominated by molecular transport,
i.e., Brownian motion (for small particles less affected by gravitation).
The buffer layer is a transition zone in which turbulence is reduced by
the close proximity of the surface. Both turbulence-induced eddies and
Brownian motion contribute to mass transfer in the buffer layer. As
higher wind speeds and less stable air prevail within the surface layer,
a rough flow regime becomes dominate [Figure 9 (b)] in which many of the
surface roughness elements protrude outside of a now discontinuous viscous
sublayer resulting in a more efficient path for mass transfer. The wind
speed and/or temperature stability at which this rough flow regime occurs
is quite uncertain. Yet it is certain that under rough flow conditions
turbulent transfer and impaction will increase overall mass transfer
substantially.
Kondo et al. (1973) found that 30% of the roughness elements due to
high-frequency components of ocean waves protrude outside the viscous
sublayer at U,Q= 2 m s wind speeds. At u-,0 = 8 m s~ Kondo et al . (1973)
found that over 99% of the high-frequency roughness elements protrude outside
the remaining discontinuous sublayer. These percentages should not be very
-------
SMOOTH FLOW
TURBULENT LAYER
ROUGH FLOW
TURBULENT LAYER
BUFER LAYER
VISCOUS SUBLAYER
AAAAAAAAAAAAAAAAA A
BUFER LAYER
a
Figure 9.
Diagrammatic Representation of the Surface Layer Above the
Air-Water Interface: (a) Smooth flow; (b) Rough flow.
-------
32
different for the surface of Lake Michigan. From a physical viewpoint it
must be concluded that the ratio of the height of the protrusions, h ,
to the viscous sublayer thickness should be an important factor for
smooth or rough flow dominance (Schlichting, 1968). Since 307, of the
high-frequency wave components can be expected to protrude outside the
sublayer at 2 m s and the mean surface roughness height may equal the
viscous sublayer thickness at 3 m s~ (Kondo et al ., 1973), smooth flow
conditions may not prevail above 2ms and probably not above 3ms
Kondo et al. (1973) found the roughness Reynolds number range for transition
from smooth to rough flow as 6 < u*h / v> < 67, where u* is the friction
velocity and v the kinematic viscosity. SethuRaman (1979) found
20 < u*h /v < 75 for this transition region at a stabilized buoy off
Long Island. This descrepancy with Kondo et al. (1973) suggests the
transition region may not be encountered until u-, ^ reaches 4 or 5 m s
However, SethuRaman (1979) restricted his analysis to cases when the
waves propagated in the same direction as the wind. In the 2< u,g < 5ms
range, wind speed and especially wind direction variability increase.
As a result, the "aged-wave" condition which SethuRaman (1979) considered
may not prevail. Indeed, Done!an (1977) found drag coefficients (C^)
of 15 x 10" and larger in the developing wave condition and "a pronounced
minimum in this drag coefficient as the wave field approached maturity."
Boutin et al. (1977) found a direct dependence of the drag coefficient
uoon wind speed variability below D9 = 4 m s~ with resultant drag
3
coefficient mean and standard deviation of 1.8 and 1.3 x 10 , respectively.
Wind speeds were in the 1.2 < iL < 3.6 m s range.
Fully rough flow for aerodynamic transfer is usually assumed to
prevail at uin = 8 m s~ and above, whereas one may expect a transition
-1 -1
flow between 3 and 8 m s~ , and smooth flow below 3ms . It has been
stated that aerodynamic transfer can be expected to be larger (possibly
much larger) than mass transfer in the buffer layer during transition and
smooth flow because of what is known as the bluff body effect (Chamberlain,
1966). In the immediate vicinity of the surface, transfer to the surface
is controlled by skin friction and pressure forces, caused by fluid
impacting on the roughness elements of the surface. Transfer by pressure
forces at the surface is known as form drag and since there is no apparent
-------
33
analogue in mass transfer, a possibly severe limitation of the near
equality between aerodynamic and mass transfer is encountered within the
buffer layer and especially the viscous syblayer.
However, Wu (1979) has shown that the "young-wave" condition observed
by Donelan (1977) and Boutin et al. (1977) can generate gravity-capillary
waves at wind speeds of 2-3 m s" . This could explain the occurrence of
high frequency wave protrusion through the viscous sublayer at
2 < u,g < 3 m s . Further, Owen and Thomson (1963) have shown that
horseshoe eddies wrap themselves around the individual roughness elements
for those that are closely spaced. Such eddies can be expected to result
from high-frequency wave components. If we now reconsider the data of
Kondo et al. (1973), it does not seem unlikely that transfer through the
buffer layer and impaction to the roughness elements within a discontinuous
viscous layer induce a low-resistance path for particle transfer when
compared to molecular diffusion transfer within the viscous sublayer.
This path may be available to only a few of the particles in the buffer
layer but may nonetheless be a relatively efficient transfer path even
at 2 m s wind speeds. Further discussion may be found in Sievering
(1981).
The above discussion indicates that u*h A> may be a key parameter
in the estimation of aerosol vd over the Great Lakes, h , the characteristic
height of water surface roughness elements is the only unknown in this
parameter. By following the reasoning of Kitaigorodskii (1973) h can be
stated as ^ -, 2
\ • Y
_2
where g is the gravitational constant, 9.8 m s and u*, again, is the
friction velocity, u^ may itself be approximated as ^- so that we can
approximate h over Great Lakes surfaces as a function which is
proportional to the third power of the mean wind speed during any one
filter sampling set. During rough flow conditions we may still assume
vd to equal aerodynamic transfer (vrf = u • C ^, where C~D is the stability
corrected drag coefficient—see App. B of the Lake Michigan final report).
Further, from the Lake Michigan profile results we can estimate v, to be
1/3 to 1/2 of u • Cg_ during moderate flow conditions. We will continue
to assume v , = 0 under smooth flow conditions. Table 4 shows, then, the
parameterization .resulting from our consideration of the available
(meager) experimental data plus theoretical arguments.
-------
34
Table 4. PARAMETERIZATION OF DEPOSITION VELOCITY,
u*h # of Lake
- - v , Erie Sets
v d
0-10 0 5
10-25 (1/8 to 1/4) . u . CDD R
25-50 (1/3 to 1/2) . u . CDD 7
50-70 (2/3 to 3/4) . u . CDD 6
>70 1 . u . CD[) 2
-------
35
u*hs
The available data on versus flow condition indicates that the
V
range of values between 10 and 25 to be indefinite as to smooth or
moderate and 50-70 indefinite as to moderate or rough. Thus, the
parameterization in Table 4.
Of the twenty-eight filter sets obtained at mid-Lake Erie (not
counting the three sets obtained while steaming back to dock), the
u*ns
number falling into each of the ranges is shown in the last column
vs _1
of the table. The overall Lake Erie average v. ~ 0.35 cm s obtained
-1
by this parameterization may be seen to be less than half the 0.75 cm s
had v . = u • Cnn been used.
d UU
To this point in the parameterization we have not dealt with the
particle size dependence of v,. Partly this is because the Lake Michigan
profile results (Sievering, 1981) suggest a small (^ twofold) correction.
Alternatively, a review of SI inn and SI inn (1980), Hicks and Williams
(1980), and, more recently, Williams (1981) on the contribution of particle
growth in 100% RH environments--as over the Great Lakes—suggests that
for particles of D > 1 ym, v, equals the aerodynamic transfer velocity
(u • Cnn)- Since our Great Lakes aerosol sampling obtained elemental
mass concentration in the D > 1 and D < 1 ym size fractions the
parameterization of Table 4 is to be applied only to the D < 1 ym
aerosol fraction. For D > 1 ym it is assumed, as part of this overall
parameterization, that v . = u • CDD.
-------
36
SECTION 5
INTERPRETATION OF RESULTS
Our approach to interpretation throughout this three-year
Lake Erie research effort has focused on the preparation of articles
for scientific peer review with hoped for publication in refereed
journals. Two articles already published are: 1) Profile Measurements
of Particle Mass Transfer at the Air-Water Interface, Atmospheric
Environment, vol. 15, No. 2, pp. 123-129 (1981); 2) Air-Sea Particle
Exchange at a Nearshore Oceanic Site, Journal of Geophysical Research,
vol. 87, No. C13, pp. 11027-37 (1982). Several articles are either
under review or still in preparation. For this reason they are
appendices to this research report. In order of appearance in the
appendices they are (with note in parens on status and journal):
A) Lake Breeze Effects on Particle Size Spectra and Sulfate Concentration
over Lake Michigan (submitted reviewed and being revised for publication
in Journal of Applied Meteorology ); B) Chemical Uniformity of Atmospheric
Aerosol: Its Violation at a Mid-Lake Erie Site (submitted to Atmospheric
Environment); C) Tech. Note: Trace Element Pass-Through for Cellulose
Impactor Substrates and Filters When Used for Aerosol Collection (not
yet submitted for review); D) Chemical Elements in Atmospheric Aerosol
over Southern Lake Michigan: The Contribution of the Lake Source
(submitted, reviewed and being revised for publication in Journal of
Great Lakes Research); and E) Some Effects of Wind-Shift on Over-Lake
Turbulence and Aerosol Deposition. All these appendices will be referenced
at various points in this Interpretative Section. A few points of major
concern to this USEP'A report will not be submitted for publication elsewhere
and are, therefore, discussed in some detail below.
5.1 SHIP/BUOY METEOROLOGICAL DATA INTEPCOMPARISON AND THE REPRESENTATIVENESS
OF SHIP SAMPLING
Is ship sampling at one point and for relatively brief time periods
representative of the whole lake's meteorology and climatology? The
Lake Erie sampling program provided a unique opportunity to assess the
question of ship versus whole lake meteorology thanks to the Canada
-------
37
Center for Inland Waters (CCIW). CCIW has deployed three buoys on Lake
Erie as shown in Figure 2. The intercomparison of buoy #1 with ship
sampled meteorological data is of interest. Yet, of greater interest
is the intercomparison of the three buoys' meteorological data during
the three ship sampling periods, since this intercomparison affords us
the opportunity to consider how meteorologically representative is the
ship's location.
Buoy-gathered data were taken at the three Lake Erie sites
previously mentioned. When looking at these data several points
should be kept in mind. The measurement height was not specified but
appears (from photos) to be from 1 to 2 m. The buoy data were taken
such that hourly-averaged values are centered on the hour whereas ship
data result in hourly averaged values centered 7.5 minuter before the
hour. Only temperature, wind speed and direction are here considered
since they are very likely the meteorological parameters most important
in lake loading estimates.
Inspection of the data reveals several periods when the buoy
intercomparison shows large discrepancies but fewer periods and smaller
discrepancies between ship and buoy #1 located less than 10 km apart.
Indeed the difference between ship and buoy #1 for wind speed and
temperature averaged across one-hour periods only very rarely results
in a more than 5% difference in the bulk Richardson number (proportional
to wind speed squared divided by temperature). This difference in
calculated Richardson number was never more than 8%. We may conclude
differences between ship and buoy #1 data are primarily related to
instrumentation measurement errors. The larger discrepancies between
ship (or buoy #1) data at central Lake Erie and those data obtained
by buoy #2 or #3 at the western and eastern ends of Lake Erie are more
likely due to synoptic or mesoscale meteorological differences.
There were a few cases where the buoys reported similar values, but
with a lag time between them indicating a westward or eastward moving
disturbance. These were for temperature: 25 July 4:00 to 6:00; 5:00 to
10:00, 6 Sept and for wind speed: 25 July 4:00 to 12:00; 28 July
8:00 to 12:00; 6 Sept 12:00 to 17:00; and 17 Oct 6:00 to 13:00. Times
are local. These disturbances are as would be expected with frontal
passages and require no further comment. However, a few cases did not
seem to be associated with synoptic disturbances. These generally had
-------
38
a large spread among the three buoys. For wind speed: 26 July 21:00 to
27 July 11:00; 6 Sept 14:00 to 7 Sept 15:00; 17 Get almost all of the
24-hr period; 18 Oct 14:00 to 24:00; and for temperature: 26 July 7:00 to
27 July 4:00; 5 Sept 2:00 to 6 Sept 10:00; 17 Oct 4:00 to 16:00. The
similarity of these periods of discrepancy with each other for both wind
speed and temperature point to other than instrumental problems. They
are most likely due to lake/mesoscale effects. Depending on whether one
considers wind speed or temperature these periods represent 13-23% of the whole
overall, though, the meteorological data taken at the shin are representtive
of the entire open water area (i.e., removed from nearshore) of Lake Erie
except, of course, when synoptic disturbances are present.
Regarding the question, is ship sampling representative of lake
climatology, a chi-square test was applied to compare the ship station
observations of wind speed, wind direction and thermal stability (i.e.,
air-water temperature difference) with the ten-year NOAA Lake Erie
Climatology (Summary of Synoptic Meteorological Observations for Great
Lakes Areas, Vol. 1, Lake Ontario and Lake Erie). Data on wind speed
and thermal stability were grouped into five bins each (thus, 4 degrees
of freedom) and for wind direction into eight bins (7 degrees of freedom).
Chi-square values are 42,38 and 11 for wind speed, thermal stability and
wind direction, respectively, indicating that we can accept, with only
0.1%, 0.2% and 25% probability, the hypothesis that there is no difference
between ship-sampled and annualized climatological data on Lake Erie wind
speed, thermal stability and wind direction, respectively. When only
June, July, and October climatological data are considered the probabilities
increase to 5%, 10% and 40%. These very low probabilities indicate that
a very poor representation of Lake Erie climatology was observed during
ship sampling during June, July, and October of 1979. This is especially
so regarding the under-represented high wind speeds (> 9 m s ) which
occur 12% of the climatologically average year and unstable thermal conditions
which occur nearly 40% of the climatologically average June, July, and
October as well as the climatologically average year. A cumulative
frequency plot of the ship-sampled wind speed regime (ship sample) versus
the lake's climatologically average wind speed regime is shown in Figure 10
(with Lake Michigan wind speed regime also shown). The Southwest (primarily
-------
CUMULATIVE FREQUENCY OF OCCURRENCE
-n
CQ
1
o
3 0
-'• c:
0 3
zr c:
CQ CU
CU c+
3 ->•
C~> (0
->• -p
3 -S
Cu fD
O C
— ' fD
0 3
CQ O
0
Cu ~O
__l ~«l
o
__i.
3 0
Q. -h
GO I/O
"O 3"
fD -••
fD -a
Q. 1
co
O Co
o 3
3 T3
Q. — '
r+ Q.
— 1.
O O)
3 3
CO Q.
r~
Cu
fD
CO
m
fD
Cu
3
CL
13
m
m
p
m
— i
m
DO
c/5
m
o
CO
b
b
en
b
p>
b
•^ 1
b
00
b
CD
b
p
b
_L
b
ro
b
CO
b
_».
•^
b
"^^
en
b
O)
b
_L
b
CO
b
6£
-------
40
5.2 AEROSOL CHARACTERISTICS AND THE ESTIMATION1 OF ATMOSPHERIC DRV LOADING
Even though meteorological conditions were not representative of
Lake Erie's climatology, aerosol characteristics may have been so. It
was, however, noted at the close of Section 4 that the Cleveland source
area (southwest wind direction sector) was severely under-represented.
This raises doubts about the representativeness of aerosol data in
Tables 1, 2, and 3. Since the major anthropogenic source area (Cleveland)
was lacking in its contribution and since no other anthropogenic sources
were within 40 km it is reasonable to expect the aerosol character to be
similar to continental background or rural conditions. Indeed, elemental
concentrations in Table 2 are quite low. Yet, the enrichment factors for
Cu and Zn are high relative to those over Lake Michigan.
Rahn (1976) has suggested that rural and remote continental aerosol
should all display the same relative elemental proportions. Peirson et al .
(1974) observed such a uniformity at eight locations throughout the United
Kingdom. Stolzenburg and Andren (1981) also observed this uniformity in
rural areas of the United States. It was concluded that a close
comparison of the Lake Erie, Lake Michigan, and other aerosol elemental
data bases would help resolve the question of Lake Erie aerosol representa-
tiveness. This effort has resulted in the Appendix B paper—Chemical
Uniformity of Atmospheric Aerosols: Its Violation at a Mid-Lake Erie Site.
Close scrutiny of Tables 1 and 2 of this Appendix (and the related text)
makes it clear that the Lake Erie elemental data base is not representative
of conditions over the lake. The violation of chemical uniformity (observed
by others: Peirson et al., 1974; Stolzenburg and Andren, 1981; and King
et al., 1976 even in Cleveland) by, especially, the low Pb and high Cu and
Zn enrichment factors over Lake Erie make this a dubious data base to
\
use in the binning analysis described in Section 3.1 (equation 2). On
the other hand, the Lake Michigan data base fits the chemical uniformity
hypothesis quite well, supporting its use in binning analyses performed
in the past (EPA-905/4-79-016). The Appendix B paper suggests that the
high percentage SO* aerosol mass contribution over Lake Erie (27%} versus
that over Lake Michigan (12%) may have played a role in this violation of
-------
41
the chemical uniformity hypothesis. In any event, a severely restricted
use of the binning analysis must be considered in arriving at Lake Erie
atmospheric dry loading estimates.
Before taking a close look at the Lake Erie loading estimates, it
should be noted that all elemental concentration data were corrected for
two types of sampling losses: a) cascade impactor loss of particulate
matter to the walls of the impactor (Specified only in the last few
years); and b) aerosol pass-through loss past the Misco backup filter.
The former has been identified by the cascade impactor manufacturer
(Sierra Inc.) as 5 to 10%. Since only two stages of the maximum five-stage
impactor were used in Lake Erie (and Lake Michigan) sampling a value of
5% loss, equally shared by the two stages' aerosol mass has been applied
to the data in Tables 2 and 3. The second correction—aerosol pass-through
loss—was determined as a part of this Lake Erie Grant research. Results
are fully described in Appendix C. Overall mean percentage pass-through
loss was found to be 38,32 and 11% for Pb, Zn, and Fe, respectively.
Given the fine/coarse ratio (F/C) of these three elements and the F/C
of other elements (Table 2) a correction of 10-40% (dependent upon
element) was applied to the fine fraction mass determined by ICAP analysis.
One final correction — to reiterate—has been applied to the comparative
presentation of Lakes Erie and Michigan dry loading estimates in the
Executive Summary; the deposition velocity parameterization at the end
of Section 4. Thus, Lake Michigan loading estimates presented in this
document are about one-half those presented in the Lake Michigan final
report (EPA-905/4-79-016).
As stated in the Executive Summary of. this document, use of the
binning analysis with Lake Erie data results in very large uncertainty
loading estimates—at least an order of magnitude. The lack of representa-
tiveness in aerosol elemental concentrations and meteorological data in
combination with the still quite ill-determined deposition velocity causes
one to view the Lake Erie loading estimates "with some skepticism."
Given the poor representation of high wind speeds and certain wind directions,
the comparison of Lake Erie loading figures with Lake Michigan loading
figures is not very meaningful; and it is meaningless to attempt any
extrapolation of the two lakes' loadings to others of the Great Lakes.
-------
42
Cleveland source area) wind direction sector was very severely under-
represented (11% of ship sampling time as opposed to 20% of the
climatologically average year.) The same statistics when applied to
earlier Lake Michigan data base give a much higher concordance between
that ship samoling data base and the Lake Michigan climatology—except
that > 9 m s wind speeds are still under-represented. (A cautionary
remark about probabilities calculated by the chi-square test: too few
filter sets (28) are available to properly apply the chi-square test. The
probabilities stated above would likely increase had a larger data base
been obtained. Nonetheless, the stated probabilities, being so very low,
are a strong indication of the poor representation.)
The mesoscale lake breeze effect on meteorological conditions during
ship sampling was much less pronounced than during Lake Michigan sampling.
Only during four of 28 data sets were back-trajectories not drawn due to-
complexities caused primarily by the lake breeze effect. As a result,
further analysis of the Lake Michigan lake breeze periods culminated in
the Appendix A Paper, Lake Breeze Effects on Particle Size Snectra and
Sulfate Concentration Over Lake Michigan, which shows that aerosol sulfate
may increase as a percentage of total aerosol mass while air parcels
traverse the lake breeze cell.
A "funnelling effect" over Lake Erie was not previously observed over
Lake Michigan. This effect over Lake Erie is seen as a funnelling of the
wind flow over the western end of the Lake causing a turning of the wind
in a direction in accordance with the shape of the Lake. Since the Lake
is shaped such that the west end faces the northwest direction and curves
around towards the northeast, the prevailing wind flow is observed as
following this shape. This over water flow is often seen as being much
stronger (about 2-3 times -faster) than the overland surface velocities.
It has been noted during the funnelling effect that the wind velocities
were stronger on the windward side of the Lake than on the leeward side.
The above was observed to occur predominantly with southerly flows. The
speeding up of winds over the lake and longer traverse over water caused
by the funnelling effect should enhance pollutant fluxes to Lake Erie.
-------
43
Yet, there is meaning in comparing the relative loadings—one element
to another. Or, still better, there is meaning in considering the
relative contribution of natural and anthropogenic sources to the
aerosol elemental concentrations observed over Lake Erie and Lake Michigan.
5.3 NATURAL AND ANTHROPOGENIC SOURCE CONTRIBUTIONS TO GREAT LAKES
ATMOSPHERIC DRY LOADING AND THE FUTURE OF ATMOSPHERIC LOADING ESTIMATES
Given the difficulties encountered in the estimation of atmospheric
dry loading to date it may be more fruitful to pursue a different
interpretive path. Recent developments in receptor-oriented multivariate
statistical analysis (Gordon, 1980; Cooper and Watson, 1980) have
resulted in a powerful technique to quantify the contribution of
several aerosol source types to the aerosol elemental concentrations
observed at the midlake receptor sites. This technique may be labelled
chemical mass balance if a sufficiently large percentage of the aerosol
mass has been chemically characterized. In the case of the Lake Erie
and Lake Michigan data bases less than one-third of the aerosol mass has
been characterized by elemental, sulfate (SO.) and nitrate (Nn ) analyses.
Thus, a chemical element balance (CEB) approach was considered. Appendix D,
Chemical Elements in Atmospheric Aerosol over Southern Lake Michigan: the
Contribution of the Lake Source, is the result of elemental balance across
seven source types as applied to the 1979 crib site data base. The crib
site data base was chosen for CEB application for two reasons: 1) this
data base is smaller than the mid-Lake Michigan data base, making this
first CEB application easier from a computational standpoint; 2) the
highest quality chemical analyses appear to have been performed on these
'79 crib samples. From a long-term Great Lakes research perspective the
crib site CEB- results must be considered preliminary. Nonetheless, some
very interesting (preliminary) outcomes may be found in the Appendix D
paper. Of greatest interest is the somewhat tentative indication of a
substantial lake aerosol source — both fine fraction (diameter, D < 1 ym) and
coarse fraction (D > 1 ym). See Tables VII and VIII as well as related
text of Appendix D.
A preliminary analysis of the USEPA Great Lakes Atmospheric Deposition
network data suggests the presence of a lake aerosol source (Lueck and
Sievering, 1981). Figure 1, an example ASAS number distribution plot, had
shown a bimodality suggesting the presence of a lake aerosol source with a mean
-------
44
diameter of ^ 1 ym. This supports the CEB analysis result above. A
fine/coarse fraction comparison of midlake to crib site data on Lake "ichigan
(Table IV of Appendix D) does not violate this outcome, although it does
show an enhanced Mg and Ca concentration in midlake large aerosol compared
to the crib site. (Note: the SO^ and Pb fine/coarse fraction comparison
[again see Table IV of Appendix D] indicates an enhanced midlake fine
fraction. This is probably due to the coagulation of aerosol less than
0.1 ym which cause the crib site fine/coarse fractions to increase at
nidlake.)
The magnitude of the lake source—both fine and coarse fraction--
indicated by CEB analysis is from one-half to twice the soil source. The
fine fraction lake source was found to about equal the sum of anthropogenic
source contributions to total fine fraction aerosol mass. Of course, the
lake source does not contribute significantly to specific anthropogenic
pollutants such as Pb, Zn, and Cu. Further, it should be noted that
fugitive dust sources were not considered in the CEB analysis. In the
Chicago/Gary area, fugitive dust area sources contribute about equally
with anthropogenic point sources to the suspended particulate matter
(B. Bolka, 1982). However, these'fugitive dust sources (primarily highway
aerosol and road dust) should not be of a similar chemical constituency as
the lake source.
The exercise of using multivariate statistical analysis (here, chemical
element balance approach) has proven to be a fruitful one. It appears to
have enough merit that its application to the larger mid-Lake Michigan data
base is planned among future research activities. In particular, the
determination of the relative contribution of several anthropogenic source
types to over Great Lakes aerosol mass or of their relative contribution to
a particular element's overlake mass concentration would be of special interest
to the USEPA. Another look at Table VII of Appendix D shows that a relative
contribution of cement manufacturing, steel production, oil burning,
automobile exhaust and coal burning to crib site aerosol mass concentration
has been obtained. If CEB analysis were applied to the larger and isolated
(i.e., removed from the immediate vicinity of source types) mid-Lake Michigan
data base, the delineation of major anthropogenic source type contributions--
-------
45
at least one type relative to others--could be specified. This could prove
immensely helpful to the USEPA in Great Lakes enforcement strategies.
There is yet another reason for taking this interpretive path: the
determination of over water deposition velocities for better than order-
of-magnitude Great Lakes loading estimation appears to be unreachable within
the next few years. It may be better to apnly multivariate statistics to
a few Great Lakes pollution data bases in that more fruitful outcomes nay
be achieved from interpretation of these data bases.
Many difficulties have been encountered with Lake Erie atmospheric
dry loading estimation. Yet, it was also noted in the Executive Summary
that the single most uncertain factor (overwhelming all others combined)
in the loading equation—equation 2 in Section 3.1--still remains the
deposition velocity, v,. As long as this situation remains so, the
estimation of atmospheric loadings to the Great Lakes by a product of
overlake concentration measurements with v, must remain at order-of-magnitude
uncertainty. The remainder of this section 5.3 will first digress into
the alternative method for lake loading estimation—lake mass balance
analysis—and then close with a brief discussion of some future needs in
v, measurement research.
d
Dolan and Bierman (1982) have recently published a good example
study of mass balance analysis with regard to Saginaw Ray, Lake Huron.
This is one of the few mass balance studies which have included sensitivity
analysis of the free parameters in the assumed mass balance model. Of
the five free parameters, suspended solids and total metals were most
sensitive to variations in the settling velocity within the water column.
Changes in this parameter result in solids' (and total metals) leaving the
water column too soon (increased N) or staying in the water column too long
(decreased N). A 50% change in settling velocity caused from 24 to 110%
change in Cu, Pb, and Zn Saginaw Bay total metal concentrations. This
24-110% range may be compared to uncertainties in the mass balance model
results caused by the atmospheric loading input (fixed) parameter,
Dolan and Bierman (1982) used the metals analysis from six bulk samplers
(collection on a monthly basis) located at several sites around Saginaw
Bay. These bulk samples are known to under-estimate the contribution of
dry deposition by 200% to as much as 1000% (Clough, 1973; Ton, Eastman,- and
Sievering, 1983). Assuming the actual dry deposition loading to Saginaw
Bay to be 1/4 to 1/2 that of wet deposition, the uncertainty in the atmospheric
-------
46
loading input parameter is from 50% (200% x 1/4) to 500% (1000% x 1/2).
This 50-500% uncertainty in atmospheric loading is to be compared with
the 24-110% changes in modeled metals concentrations introduced by the
most sensitive mass balance model free parameter. Clearly, the uncertainty
in the atmospheric loading component overwhelms all others in the Saginaw
Bay metals mass balance model.
A similar—but less detailed—mass balance calculation for Lake
Erie was provided by Dr. J.A. Schmidt (private communique, 1983). Estimates
of atmospheric, tributary and shoreline erosion inputs (fixed parameters)
were compared with sedimentation and outflow to Lake Ontario. Table 5
shows the range in inputs and outputs for Cr, Cu, Pb, and Zn. These ranges
are: atmospheric, 7 to 17-fold; tributary, 3 to 8-fold; shoreline erosion,
30% to 2 1/2-fold; sedimentation, 40% to 2-fold; and outflow, less than
2-fold. These ranges, the result of a literature/data review, clearly
show--again--that our knowledge of atmospheric inputs is the weakest.
The order-of-magnitude uncertainty in atmospheric inputs, is many-fold larger
than all others except the tributary input uncertainty, which is about
one-half as large as the atmospheric uncertainty.
Having assured ourselves that a Great Lakes mass balance approach
does not significantly increase our understanding of nor reduce uncertainty
in the contribution of atmospheric deposition, we can now focus on the
last interpretation topic of this report: Future Improvements in Atmospheric
Loading Estimates. The last portion of Section 4 on deposition velocity
(v.) parameterization suggested that for aerosol of diameter, D < 1 ym,
vd % (1/3 to 1/2) • u ' CDp and for 0 > 1 ym, vrf £ u • CpD. The assumed
relative difference by size of 50 to 65% is the result of aerosol profiling
field studies on Lake Michigan (Sievering, 1981). This small difference
may be contrasted with the at least ten-fold difference in vrf postulated
by SI inn and SI inn (1980) for D = 0.1 ym versus D = 1 ym aerosol. This
discrepancy is at the order-of-magnitude level which has plagued the
atmospheric dry loading estimates throughout this report. Slinn and Slinn
(1980) obtain this very large vd difference for D = 0.1 ym versus D = 1 ym
aerosol as a (theoretical) result of their consideration of aerosol growth
in the high humidity environment near the water's surface. Sievering (1981)
observed the only two-fold difference for aerosol of D = 0.2 ym versus D = 2 ym
-------
47
Table 5. MASS BALANCE ESTIMATE FOR LAKE ERIE
Inputs
TOO O A C C
Element Atmospheric ' ' Tributary ' ' Shoreline Erosion
Cr 15-180 500-1800 120-220
Cu 50-350 1000-3000 130-320
Pb 150-2200 800-4800 240-310
Zn 300-5200 1800-14000 390-700
This report
2
International Joint Commission, 1969. "Pollution of Lake Erie,
Lake Ontario and the International section of the St. Lawrence River:
Volume 12--Lake Erie," 316 pp.
Schmidt, J.A. & A.M. Andren, 1983. "Atmospheric Trace Metal Loading
of the Great Lakes," in: Evnironmental Quality of the Great Lakes, Jerome
Nriage, editor, in press.
4
Blake, H.D., unpublished data. Heidelberg University.
5Chawla, V.K., Y.K. Chan, 1969. Trace Elements in Lake Erie.
Proceedings, 12th Conference on Great Lakes Res., pp. 760-765.
6Monteith, T.J. & W.C. Sonzogni, 1976. U.S. Great Lakes Shoreline
Erosion. Task D Report, PLUARG.
Outputs
Element Sedimentation7'8'9 Outflow10
Cr 2300-3600 2500
Cu 1700-2700 1000-1800*
Pb 4500-9600 700
Zn 5100-8100 1800
Kemp,. A.L.W., 1975. "Sources, Sinks and Dispersion of Fine-Grained
Sediment in Lake Erie." Proceedings of the Second Federal Conference on
the Great Lakes, pp. 369-377.
p
Kemp, A.L.W. and R.L. Thomas, 1976. "Impact of Man's Activities
on the Chemical Composition in the Sediments of Lakes Ontario, Erie, and
Huron." W.A.S.P., 5; 469-490.
9
Walters, L.J., T.J. Wolery, and R.D. Myser, 1974. In: Proceedings,
of the 17th Conference on Great Lakes Res., pp. 219-234.
International Joint Commission, 1978. "Environmental Management
Strategy for the Great Lakes System," final report from PLUARG, 115 pp.
*
Estimated from Lake Michigan trace metal data and Lake Erie river
data.
-------
48
aerosol considering a small number--!3--profiles obtained during moderate
wind speed conditions. Thus, a question lingers as to whether the two-fold
difference is truly representative (Hicks and Williams, 1980). Yet, the
mechanism of surface roughness (and its variation) with roughness element
protrusions acting as aerosol collection sites was not considered by Slinn
and Slinn (1980). It was considered by Williams (1981) in "A Model for
the Dry Deposition of Particles to Natural Water Surfaces." The model is
one with a broken surface resistance applied to a percentage of the water's
surface area based on oceanic observations by Wu (1979). The equivalent
deposition velocity (v ,)--at the small percentage of water's surface which
is estimated to be broken — is assumed to be a constant value, either 1, 10,
100, or 1000 cm s" . Williams suggests the 10 cm s~ value as most probable,
but acknowledges that we only know that "broken surface transfer is more
efficient than turbulent transfer (vrf ^ 1 cm s~ )" and that the broken
surface deposition velocity is not a constant across the broken surface
area. Further, Williams (1981) notes, "In the future, careful measurements
of deposition velocity under a wide range of wind speeds and particle
sizes should yield estimates of the value of this parameter [broken surface
v ,]. Until then we must make the assumption that the broken surface transfer
is a constant. Therefore, the effect on deposition by the broken surface
is solely contained in the fractional surface area that becomes broken."
A major problem with the Williams model is its steady state character.
The assumptions of a constant flux layer as well as constant broken surface
deposition velocity allows one to mathemically model air-water aerosol
transfer—but too simplistically. Observation of meteorological and wave
dynamics at the air-water interface of the Great Lakes should cause any
careful observer to denounce steady state modeling as too simplistic and
unrealistic. Models such as that of Williams serve a very useful purpose
in pointing out areas for further research as he states above. However,
the dynamics at the real air-water interface suggest that the actual aerosol
deposition is greater than modeling results would have us believe. Enhanced
deposition due to the "young wave" phenomenon referred to at the end of
Section 4 is to be expected. That discussion noted the likelihood that
the developing wave condition causes an increased turbulent transfer.
-------
49
Thus, a greater effect of the "young wave" phenomenon on small aerosol is
to be expected--possibly much greater. Indeed, the Appendix E naper, Some
Effects of Wind-Shift on Over-Lake Turbulence and Aerosol Deposition,
shows that wind shifts are associated with depletion of near-surface
aerosol mass and with departures from isotropy. Turbulence in the high
frequency range is known to normally have a ratio of vertical velocity
spectra to longitudinal velocity spectra of 4/3 in the isotrooic, inertia!
subrange. Departures from isotropy caused by enhanced energy in the
vertical velocity spectra were strongly correlated with a "windshiftiness"
parameter (see Table 2 of Appendix E), the rank correlation coefficient
being 0.925 (allowing one to reject the hypothesis of zero correlation at
better than the one percent level). The fact that aircraft measured
near-surface aerosol mass decreased in conjunction with these isotropy
departures as well is supporting evidence for the hypothesis that the
non-steady state dynamics at the air-water interface may enhance the
deposition of, especially, small aerosol. The above discrepancy between
modeled ten-fold and field observed two-fold differences in large
(D > 1 ym) versus small (D < 1 ym) aerosol deoosition may, then, be
attributed to models (to date) not accounting for the non-steady state
mechanisms of aerosol deposition at the air-water interface.
A countervailing argument may, however, also be presented. The
mechanism of aerosol resuspension from the water's surface is known to
be a contributor to large aerosol mass above the ocean's surface (see,
for example, Cipriano and Blanchard, 1981). Recently, Sievering, Eastman
and Schmidt (1982) have observed that resuspension may also contribute
to small aerosol mass above the ocean's surface (at 6-m height). If large
aerosol resuspension is more significant over Great Lakes' waters—as it
is generally considered to be over the oceans,then the large aerosol
downflux gradients observed over Lake Michigan would have been more reduced
by resuspension than would those for small aerosol. Resuspension
introduces an upflux lake source gradient which was, in fact, observed on
more than one occasion during the profiling measurements on Lake Michigan
(Sievering, 1981). The large aerosol mode in Figure 7 may again be
thought to be that elusive lake source aerosol. The important point here
-------
50
is that the only two-fold difference observed in larqe (D < 1 ym) versus
small (D < 1 vim) aerosol deposition may be attributed more to resuspension
of large lake-derived aerosol. Thus, it is quite possible that neither
the field observed two-fold nor modeled ten-fold difference in large vs.
small aerosol deposition are to be accepted. The answer may well lie in
between these two extremes. Knowledge of the correct answer is extremely
important since the magnitude of pollutant aerosol deposition is highly
dependent upon this answer.
Recently El-Shoboksky (1983) found, for rough versus smooth pipe flow,
an eight- to thirty-fold increase in v, for 1 vim diameter particles.
Experimental conditions included average roughness element heights of 7 ym
4
(pipe diameter was 8 mm), a Reynolds number of 10 and friction velocity
of 120 cm s . Although turbulence characteristics were unrealistically
high (for ambient atmosphere conditions) the small roughness elements
and steady state character were also somewhat unrealistic. These laboratory
results certainly support the parameterization of small difference between
large and small aerosol deposition.
It must be concluded that "careful measurements of deposition velocity
under a winde range of wind speeds and particle sizes: and at the (actual)
Great Lakes air-water interface is a future research necessity. For the
purposes of Great Lakes research, species specific measurements are also
a necessity. Pollutant species, e.g., Cd, Cu, Pb, and Zn, may be observed
to be depositing into the Great Lakes at the same time that Ca, Mg, and
other potentially lake-derived elements are emanating from the lakes'
surfaces (see Section 4.1 and 5.2 on the evidence for a lake source). Of
the presently available deposition velocity measurement methods, i.e., eddy
flux, eddy accumulation, variance and gradient techniques, the most promising.
for species specific measurement is the gradient technique. Although
difficult to utilize over water this technique is no more difficult,
logistically, than the others mentioned; however, a less direct measurement
of v is obtained and both vertical temperature and wind speed gradients
d
should be monitored in conjunction with the aerosol gradient.
-------
51
REFERENCES
Bolka, B. USEPA Environmental Measurements Branch, private communique
(1982).
Boutin, C., B. Boullery, J.P. Albignat, and H. Isaka. Boundary-Layer
Met., 12, 301-403 (1977)
Bowen, H.J.M. Trace Elements in Biochemistry (Academic Press, New
York, 1966^
Chamberlain, A.C. Proc. R. Met. Soc., A296, 45-52 (1966).
Cipriano, R.J. and D.C. Blanchard. J. Geonhys. Pes.. 86, 8085-8092
(1981).
Clough, U.S. Aerosol Science, 4-, 227-234 (1973).
Coooer, A. and J.G. Watson. J. Air Poll . Cont. Assoc., 30^, 1116-
1122 (1980).
Dave, M., D. Dolske, and H. Sievering. Atmospheric Environment, 13, 1497-
1600 (1979).
Dolan, D.M. and V.J. Bierman. J. Great Lakes Pes., ^, 676-694 (1982).
Donelan, M.A. Dynamic vs Steady-State Momentum Drag Coefficients. In
Symposium on Modeling of Transport Mechanisms in Oceans and Lakes,
pp. 46-54. Dept. Fisheries and the Environment, Ottawa, Can. (1977).
El-Shobokshy, M.S. Atmospheric Environment, 1_7_, 639-644 (1983).
Gatz, D.F. Water, Air, and Soil Poll ., 5_, 239-250 (1975).
Gordon, G.E. Environ. Sci. & Tech., 14, 792-799 (1980).
Hunt. J.C.R. and P.O. Mulhearn. J. Fluid Hech., 6]_, 245-254 (1973).
Jensen, V.E. An Evaluation of Several Methods of Atmospheric Trajectories
(Dept. of Geog., Northern Illinois Univ., Dekalb, IL(1981).
King, R.B., J.S. Fordyce, A.C. Antoine, H.F. Leibecki, H.E. Neustadter,
and S.M. Sidik. J. Air Poll . Cont. Assoc., 26, 1073-1078 (1976).
Kitaigorodskii, S.A. The Physics of Air-Sea Interaction tr. from Russian
by A. Baruch, Israel Scientific Translations, Inc., Jerusalem, 237 pp.
Kondo J., Y. Fujinawa, and G. Naito. J. Phys. Oceanogr., _3, 197-202 (1973).
Kraft, R.L. Predictions of Mass Transfer to Air-Land and Air-Water
Interfaces, Penn. St. Univ., CAES Publ., #472-77.
Lueck, D.M. and H. Sievering. Great Lakes Atmospheric Deposition Network:
Preliminary Results, Amer. Inst. of Aeronautics and Astronautics
Meeting, Orlando, FL, January, 1982.
-------
National Oceanic and Atmospheric Admin. (NOAA). Summary of Synoptic
Meteorological Obs. for Great Lakes Areas, Vol. 3, Lake Michigan.
Nat'l Climatic Center, Asheville, NC (1975).
Owen, P.R. and W.R. Thompson. J. Fluid Mech., 1_5_, 231-334 (1963)
Peirson, D.H., P.A. Cawse, and R.S. Cambray. Nature 251, 675-679 (1974).
Rahn, K.A. The Chemical Composition of Atmospheric Aerosol (Grad. School
of Oceanog., Univ. of Rhode Island, Kingston, RI (1976).
Schlichting, H. Boundary-Layer Theory. McGraw-Hill, New York (1968).
SethuRaman, S. Boundary-Layer Met. , 1_6, 279-291 (1979).
SethuRaman S. and J. Tichler. J. Appl . Met., 1_6, 455-461 (1977).
Sievering, H. Atmospheric Environment, 15, 343-351 (1976).
Sievering, H., J. Eastman, and J.A. Schmidt. J. Geophysical Research,
87-C13, 1127-1137 (1982).
Sievering, H., M. Dave, D.A. Dolske, R.L. Hughes, and P. McCoy. An_
experimental study of lake loading by aerosol transport and dry
deposition in the Southern Lake Michigan Basin. EPA-905/4-79-016
71979).
Slinn, S.A. and W.G.N. Slinn. Atmos. Environ., 1_4, 1013-1016 (1980).
Stolzenburg, T.R. and A.W. Andren. Mater, Air, and Soil Poll., 15,
263-270 (1981).
Ton, N., J. Eastman, and H. Sievering. Dry Loading Collection Efficiency
of Sulfate and Soil Aerosol by a Passive Sampler, Amer. Geophysical
Union Spring Meeting, Baltimore, MA, June, 1983.
Williams, R.M. A Model for the Dry Deposition of Particles to Natural
Water Surfaces, RER Division #8106, Argonne National Lab, Argonne,
111. ANL/ERC-81-06 (1981).
Winchester, J.W. and G.D. Nifong. Water, Air, Soil Poll 'n., 1_, 50-61 (1971),
Wu, J. J. Phys. Oceanogr., 9, 802-814 (1979).
-------
53
APPENDIX A
LAKE BREEZE EFFECTS ON
PARTICLE SIZE SPECTRA AND SULFATE CONCENTRATION
OVER LAKE MICHIGAN
by
Richard L. Hughes and Herman Si evenng
Environmental Science Program
College of Applied Sciences
Governors State University
Park Forest South, IL 60466
-------
54
ABSTRACT
Aircraft flights over southern Lake Michigan in September
of 1977 and May of 1978 conducted during which meterological and
size-distributed particle data were gathered. An anomalous local
maximum of the particle distribution in the 0.3 - 0.4 ym diameter
range occurred on those days during which a lake breeze took place.
This effect contributes more than 20 percent increase in the
particle volume across the 0.2 - 0.5 ym diameter range. Evidence
is presented suggesting that SOp gas to sulfate particle conversion
may be responsible.
-------
55
1. Introduction
The distribution by size of atmospheric particles is an
important variable in the various aspects of air pollution, particularly
in the lowest layers of the atmosphere (Meszaros, 1977). The size
distribution of sulfate particles is particularly important in the
investigation of acid rain in the vicinity of and over the Great
Lakes. It has only recently been realized, beginning with the work
of Winchester and Nifong (1972), that dry deposition of atmospheric
particles constitutes a major source of certain pollutants to large
water bodies such as Lake Michigan. Thus, until the last decade,
there had been few in situ investigations of particle characteristics
over water.
Early measurements of size spectra over natural water
surfaces were reported by Junge (1972), who observed the typical
-3 power law distribution. Bridgeman (1979) has measured particle
size spectra over the midwest. Measurements over Lake Michigan
were not the primary focus of that study, but one over-lake flight
was made using instruments very similar to those used in this study.
Extinction coefficients indicated the presence of extremely clean
air and bore some resemblance to rural air extinction coefficients.
Most evidence suggests that in the range 0.1 < diameter (d) < 10 ym,
the number distribution is bimodal with maxima around d = 0.1 - 0.3 vm
and d = 5.0 - 8.0 ym (Whitby, 1973). But, as pointed out by Whitby
(1978), this modality is only marginally visible in the number
distribution, and is most evident in the surface area and volume
distributions. In connection with the Great Lakes Atmospheric
Loading Experiment (GALE), measurements of size spectra, along with
' other data, were gathered aboard an airborne platform in 1977 and
1978 over the southern basin of Lake Michigan.
-------
56
Instrumentation
Measurements were taken aboard a Beechcraft Queen Air aircraft
provided by the Research Aviation Facility of the National Center for
Atmospheric Research. The aircraft departed from Midway airport and,
upon reaching the lake, proceeded at a low altitude (usually 30 m) to
the midlake sampling site, 87° OCf W, 42° OCf N. Vertical soundings
were taken in a square spiral "box" pattern over this site, over a
point halfway to shore, 87° 15 W, 41° 52' N (referred to as the halfway
point) and, in 1978, over a near-shore site, a City of Chicago water
intake crib, 87° 32" W, 41° 4?' N (see Figure 1.). The lowest data
were taken at an altitude of 15 m and the soundings extended through,
about 2000 m. Flight times were approximately 0700-0900 CDT, 1100-1400
CDT, and occasionally 1600-1800 CDT during September 26-30 of 1977 and
May 17, 19, 22-24, and 26 of 1978.
Particle size spectra were obtained from the Active Scattering
Aerosol Spectrometer (ASAS) manufactured by Particle Measuring Systems,
Boulder, Colorado. The instrument measures light scattering internal
to the cavity of a He-Ne laser. By a system of precision optics and
photomultipliers, the absolute number of particles in each of 60 size
channels is recorded. It is described by Schuster and Knollenberg
(1972). The instrument used in 1977 gathered data in the range
0.23 < d < 30.0 ym. Ambient air was drawn through an approximately
isokinetic sampler (see de Pena, et al., 1975 for description) and
transmitted along a length of copper tubing to the intake port of the
ASAS. Intake was flow-controlled using aircraft Venturis to produce
o 1
an inflight sampling rate of 0.28 ± 0.20 cm sec . The number of
-------
57
counts in each of the 60 size channels, along with various meteorological
data were recorded each second onto magnetic tape by the Aircraft
Recording Instrument System (Glaser, 1973). Location was determined
by an inertia! navigation system. Height above 500 m was determined
by a pressure altimeter, and below 300 m by a radio altimeter. A
weighted average of the values from these two instruments was used
between 300 m and 500 m.
During 1978 a slightly different configuration was used. The
ASAS used during this period gathered data in the range 0.11 < d < 3.0 ym
and the copper tubing was somewhat longer and was heated to significantly
reduce water vapor contribution to ASAS number counts. Interpretation
was aided by National Weather Service surface and upper air data.
Surface high volume filter, bulk water and .meteorological data were
also gathered in connection with the GALE at mid-Lake Michigan in
1977 and at the near-shore site in 1978 (Sievering, et al., 1979).
3. Calculation of Particle Size Spectra
The instrumentation used and the experimental setup produced
several consequences which made the calculation of the size spectra
less than straightforward. It was feared that the length of copper
tubing used to transmit the air sample would cause some loss of measured
aerosol counts. To investigate this, the original tubing from the
two aircraft configurations was used to sample air which was drawn
through the approximately isokinetic sampler in a wind tunnel at
Particle Measuring Systems in Boulder, Colorado. The spectra thus
obtained were compared with spectra obtained without the copper tubing
and the loss of aerosol as a function of size was calculated. In the
spectra that follow, the raw number count was multiplied by the factor
-------
58
required to compensate for this estimated loss. These correction
factors are shown in Figure 2. Since the ambient air in Boulder,
Colorado was used to obtain these curves their applicability is
uncertain, but it should be noted that the correction factor for
particles with diameters less than 0.5 ym is quite small. Our attention
was principally directed to this size range.
The 60 sampling channels are grouped into four slightly over-
lapping "size ranges" of 15 channels each. Because of the manner of
operation of the ASAS, it is possible for a size range to systematically
measure number counts which are higher or lower than neighboring or
overlapping size ranges. This was observed to occur in the larger of-
the two size ranges in 1977 and the second smallest of the four size
ranges in 1978. The signal-to-noise ratio is higher for the larger-radius
channels in any of these size ranges, so the groups with anomalously
high readings were empirically matched with the overlapping larger-
radius channels of a neighboring size range. As a result, number
counts in the anomalous range were multiplied by 0.08 for the 1977
data and 0.33 for the 1978 data to bring it roughly in line with
neighboring ranges. Because of this, while relative differences
between individual samples and relative numbers within a group retain
their validity, absolute numbers are suspect.
4. Description of Data
Five flights in September of 1977 and seven in May of 1978 were
chosen for concentrated study. These flights were selected so as to
obtain data from air which had a direct trajectory from an urban area
(September) and from air which either clearly underwent lake breeze
-------
59
circulations or from that which did not (May). Table 1 presents
certain meteorological parameters measured at the lake's surface
during these flights. Temperature difference between 5 m height and
the surface was obtained using an infrared thermometer. Distributions
of aerosol number were in general unremarkable, following a typical
-3 power curve described by Junge. In the range above 2.0 ym diameter,
particle counts were too few to accurately locate the large particle
size maxima in the bimodal surface and volume distributions. With
an exception yet to be discussed the small particle size maxima were
near d = 0.2 ym. Total particle volume in the range 0.11 < d < 2.0 ym
-12 3 -3
near the surface was in the range 10 to 20 x 10 cm cm in
September, 1977 and 20 to 30 x 10"12 cm3 cm"3 in May, 1978. Total
_3
particle number counts were 15 to 90 cm in September and 200 to 600
cm~ in May.
Three of the May flights (18, 22, and 25) had a spectral anomaly
in the form of a local maximum located within the range 0.3 < d < 0.4 ym.
The appearance in the number spectrum is that of a minor "bump" (see
Figure 3). This bump represents an increase in the particle number count
_3
on the order of 100 cm ; this is an increase of about 30 percent in the
0.2 < d < 0.5 ym range over the number of particles counted in this
range when the bump is absent (see Figure 3). As mentioned above, this
type of local maximum is more obvious in the volume distribution plots.
Figure 4 shows the volume distribution corresponding to the number
spectrum plot in Figure 3. The Whitby (1978) grand average continental
distribution has been forced to fit the small and large particle ends
of the distribution for the 1242 - 1245 CDT, 60 m altitude, near-shore
data on May 18, 1978. For this case, then, the bump referred to above
-------
60
is evident in the 0.25 < d < 0.5 ym range. Clearly in this case, and
quite often otherwise, the d ^ 0.4 ym maximum exceeds the average
continental distribution volume maximum at 0.3 ym. The bump invariably
represents an increase in particle volume of more than 20 percent in
this range. The bump is often quite narrow in range (< 0.2 ym),
suggesting a single source for these particles. Since surface area
and volume affect the reactivity and the air/water exchange of particles,
the anomaly may be of singificance to sulfate formation and particle
loading of the lake.
The three flights on which these bumps were observed (May 18, 22,
and 25) took place during lake breeze episodes. The aircraft made it
possible to directly observe the lake breeze cells (see Figure 5).
Backward-in-time trajectories were calculated for all experimental
periods by a process described in Sievering, et al. (1980). These
trajectories, shown in Figure 6, also support the existence of a lake
breeze on May 18, 22, and 25, 1978.
There were some differences in the appearance of the bumps on
these three days. On May 22 and May 25, the bumps usually appeared
in the range 0.35 < d < 0.45 ym, but on May 18 the bump also extended
below 0.35 ym. The bump was always found in air samples at altitudes,
of 30 to 60 m above the surface on the three lake breeze days. This
seems to indicate that the domain of this anomalous distribution was
spatially continuous in near-surface layers. On May 22 the bump was
found up to altitudes of 600 m while on May 18 and May 25 it was
found at altitudes near to 110 m and a second discrete region was
found between 220 and 800 m. The data at higher altitudes is more
-------
61
sparse because of both the flight track and the greater length of
constant altitude data needed to obtain statistical significance in
particle counts at higher altitudes.
The September flights, in contrast, showed only a half-dozen
isolated instances of a local maximum in the 0.35 < d < 0.45 ym range.
This was true despite the fact that all of these September flights
occurred when there was a wind directly from the heavily populated
area to the west and southwest of the sampling area. That the urban
plume was in fact sampled during the September flights is indicated
by a time plot of the total aerosol volume detected over the lake
(Figure 7). The abrupt increase, if extrapolated back in time using
aircraft-measured mean wind speeds, occurred in air that would have
left the industrialized western shore of the lake around 0745 to
0815 CDT, just as the rush hour was occurring. The aircraft measurements
indicate that the wind direction was steady during the 0730 to 0930
CDT period; i.e., the flow was direct rather than meandering. It
appears that advection from an urban area is not sufficient to produce
the spectral anomaly by itself.
Surface high1 volume air samplers equipped with three-stage
cascade impactors were in place at a 13m height above the lake surface
at the near-shore crib site during May of 1978. Reliable data were
available during only four of the six May flights discussed above, but
for these days, the ratio of fine (d < 1 urn) to coarse (d > 2 ym)
particles was larger during the lake breeze events perhaps resulting
from greater coarse particle deposition during the long over-lake
return fetch. Additionally, it was found that concentrations of Pb, V,
Zn, and Cu (metals especially associated with anthropogenic sources)
-------
62
were enhanced on the lake-breeze days (see Sievering and Dolske, 1982).
The high volume filters were analyzed for sulfates and phosphates as
well as trace metals. The ratio of sulfate to certain of these trace
metals and to phosphate is shown in Table 2. These ratios were averaged
for the two lake breeze and the two non-lake breeze days. It is evident
that in the comparison of at least these four periods, sulfates were
strongly enhanced with respect to Pb, Fe, Mn, and P-P04 and somewhat
enhanced with respect to Zn.
5. Discussion
The appearance of a local maximum of restricted size range in
an otherwise ordinary particle spectrum suggests a single source type.
Two of these anomalous cases occurred when there was no direct surface
trajectory from an urban area, as determined by the back-trajectory
analyses referred to above. But on those non-lake breeze days when
the analyses indicated either urban or non-urban source regions the
anomalies were absent. Thus, the lake breeze is sufficient in some
cases to cause an otherwise normal particle distribution to be perturbed.
Lyons and Olsson (1973), among others, have pointed out the manner in
which a spiral-shaped lake breeze can cause a re-circulation of
pollutants carried over the lake by a return flow. These pollutants
are then brought back to shore by the surface lake breeze. They
mention that a size-sorting of the aerosol was detected, which they
attributed to gravitational fallout of larger particles. In the case
they reported, this caused an augmentation of the smaller aerosol near
the surface, similar to that suggested by the high volume air sampler
data of Table 2.
-------
63
Lyons and Olsson (1972) found that particles in the lake
breeze return flow penetrated more than 20 km over-lake in a well-
developed lake breeze cell. This would be consistent with the occurrence
of the spectral bump found 30 - 50 km from shore.
The fact that the bump was not present after the short (< 2 hr)
direct fetch from the Chicago area suggests that the lake breeze itself
contains a mechanism for the sulfate enhancement noted above. Particles
the size of those in the bump (0.2 < d < 0.5 ym) are often associated
with the contribution of sulfate particles to the accumulation particle
mode (Whitby, 1978). Recent work by McMurry and Wilson (1982) in
high humidity environments such as that over Lake Michigan shows that
particulate sulfate predominates diameters of 0.3 to 0.5 ym whereas
low humidity environments produce sulfate particles less than 0.2 ym.
Airborne measurements of particle size distributions in power plant
plumes conducted by Hobbs, et al. (1979), indicate that an increase
o
in 0.3 to 0.5^».m diameter particles on the order of 10 to 20 cm above
ambient concentrations could be expected after a few hours travel time,
presumably from gas-to-particle conversion mechanisms. If the percentage
increase is applied to the concentration of particles normally present
at mid-Lake Michigan in the 0.3 < d < 0.5 ym range, an increase of 100
_3
cm is found which is consistent with the magnitude of the bump
described above. It should be noted that an urban plume would afford
more particulate surface area for heterogeneous, catalyzed reactions
than particles from an isolated power plant plume. Wilson, et al. (1977),
have shown that the transformation rate of gaseous sulfur to particulate
sulfur was less than 2 percent per hr in the Labadie power plant plume
-------
64
until mixing with the St. Louis urban plume, at which point
transformation rates increased to as much as 5 percent per hr.
Lyons (1975) has described the expected range of wind speeds
and the extent of a Lake Michigan lake breeze cycle. Using these
estimates (which were consistent with the aircraft observations for
May 18, 22, and 25) and assuming subsidence to take place about 30 km
from shore, a parcel will take 6 to 10 hr to complete a lake breeze
cycle. Assuming a mean travel time of 8 hr and a mean SOp to SO,
transformation rate of 2 percent per hr, well over 10 percent of the
SOp in the Chicago source region will have been transformed to SO,
upon arrival at the crib site. Direct transport from the urban
shoreline to the crib site would produce well less than 1 percent
transformation. To complete the S02 to SO. transformation calculation,
the rate at which growth occurs through the particle size spectrum
is needed. McMurry and Wilson (1982) show that sulfate particle
growth per unit of time Dp/dt) is about 0.02 ym hr~ and that
this growth is essentially independent of the initial particle size.
A nominal growth time for growth of an additional 0.3 pm in diameter
is then about 15 hr. Although this time is about twice as long as the
8 hr mean lake breeze travel time it should be noted that the major, ,
Chicago region S02 sources are not at the lake's shoreline. These
sources, clustered in the vicinity of Joliet, IL, were from 3 to 5 hr
removed from the lakeshore (synoptic flow was from the southwest) on
the days of interest (Table 1). Further, the 8 hr mean lake breeze
travel time would afford an additional sulfate particle growth of 0.1
to 0.2 ym which is exactly the size increment needed to explain the
shift in the peak volume distribution from the non-lake breeze cases
-------
65
of 0.2 ym to those lake breeze cases for which the volume peak occurred
in the d > 0.3 ym range. It would appear that no anomalous bump
appeared during the September flights because of the short travel
time from SOp sources.
6. Summary
Data gathered over Lake Michigan aboard an aircraft during lake
breeze events revealed the occurrence of an anomalous local maximum in
the aerosol size distribution in the 0.3 - 0.4 ym diameter particle
size range. It is suggested that the re-circulation of the lake
breeze provides a material contribution to this anomaly, perhaps
providing sufficient time for gas-to-particle conversion to take
place. Sulfate fractions appear to be enhanced relative to other
chemicals during those lake breeze events sampled.
7. Acknowledgements
Appreciation is extended to the personnel of the Research
Aviation Facility and the Computing Facility of the National Center
for Atmospheric Research. Assistance in the manual plotting of soundings
and trajectories was provided by Mehul Dave, Patric McCoy, and Vic
Jensen. Gratitude is expressed to Dixie Butz for her preparation of
the manuscript and to John Eastman for his review of a draft.
This work was supported by the U.S. Environmental Protection
Agency under grants R00530101 and R00542101.
-------
66
References
Bridgeman, Howard A., 1979: Aerosol
and Rural Air at Milwaukee in
18, 105-115.
Extinction at 500 nm in Urban
April 1976. J. Appl. Meteor.,
de Pena, J.A., J.M. Norman, and D.W. Thompson, 1975:
Sampler for Continuous Airborne Measurements.
Poll. Control Assoc.,, Boston.
Isokinetic
Proc. 68th
Air
Glaser, Walter, 1973: The Airs III Data System.
Technology, March, 61-66.
Atmospheric
Hobbs, Peter V., Dean A. Hegg, Mark W. Eltgroth, and Lawrence F.
Radke, 1979: Evolution of Particles in the Plumes of Coal-
Fired Power Plants--!. Deductions from Field Measurements.
Atmos. Environ., 1_2, 935-951.
Junge, C.E., 1972: Our Knowledge of the Physico-Chemistry of Aerosols
in the Undisturbed Marine Environment. J. Geophys. Res., 77,
5183-5200.
Lyons, Walter A., 1975: Turbulent
in Shoreline Environments.
Environmental Impact Analysis,
Mass. American Meteorological
Lyons, Walter A. and Lars E. Olsson,
Transport in the Chicago Lake
Assoc.. 22, 876-881.
Diffusion and Pollutant Transport
Lectures on Air Pollution and
Duane A.
Society,
Haugen,
296 pp.
ed., Boston,
1972: Mesoscale Air Pollution
Breeze. J. Air Poll. Control
Lyons, Walter A. and Lars E. Olsson, 1973: Detailed Mesometeorologica
Pollution Dispersion in the Chicago Lake Breeze.
101, 387-403.
Studies of Air
Mon. Wea. Rev..
Detailed Mesometeorological
Meszaros, Agnes, 1977: On the Size Distribution of Atmospheric
Aerosol Particles of Different Composition. Atmos. Environ., V\_
1075-1081.
Schuster, B.G. and R. Knollenberg, 1972: Detection and Sizing of Small
Particles in an Open Cavity Laser. Appl. Opt., 11, 1515-1529.
Sievering, H., M. Dave, D.A.
Experimental Study of
Dolske, R.L.
Lake Loading
Hughes, P.
By Aerosol
Deposition in the Southern Lake Michigan Basin.
McCoy, 1979: An_
Transport and Dry
U.S. Environmental
4-79-016. 180 pp.
Protection Agency Pulbication
Chicago, IL.
No. EPA-905/
Sievering, H., M. Dave, D.A. Dolske, and P. McCoy, 1980: Trace Element
Concentrations over Mid-Lake Michigan as a Function of Meteorology
and Source Region. Atmos. Environ., 14, 39-53.
-------
67
Sievering, H. and D.A. Dolske 1982: Chemical Elements in Atmospheric
Aerosols Over Southern Lake Michigan: The Contribution of the
Lake Aerosol Source. (Submitted to Journal of Great Lakes
Research.)
Whitby, K.T., 1978: The Physical Characteristics of Sulfur Aerosols.
Atmos. Environ., 12, 135-159.
Whitby, K.T., 1973: On the Multimodal Nature of Atmospheric Aerosol
Size Distribution, presented at VIII Int. Conf. on Nucleation,
Leningrad, U.S.S.R.
Wilson, William E., Robert J. Charlson, Rudolf B. Husar, Kenneth Whitby,
and Donald Blumenthal, 1977: Sulfates in the Atmosphere, Research
Triangle Park, North Carolina. U.S. Environmental Protection
Agency Publication No. EPA-600/7-77-021.
1
Winchester, T.W. and C.D. Nifong, 1971: Water Pollution in Lake
Michigan by Trace Elements from Pollution Aerosol Fallout.
Water, Air, and Soil Poll., 1, 50-64.
-------
Date
Sept 26
Sept 26
Sept 27
Sept 27
Sept 27
May 17
May 18
May 19
May 22
May 24
May 25
May 25
Height of Thermal
Flight Inversion
Times (CDT) Boundary Layer (m)
0645
1641
0639
1120
1636
1130
1130
1145
1133
1130
0715
1145
- 0915
- 1810X
- 0856
- 1344
- 1826
- 1411
- 1252
- 1313
- 1352
- 1419
- 0858
- 1443
None
None
None
None
None
90 - 260
None
5 - 10
10 - 80
80 - 220
110 - 160
5-190
5-m Air Temperature 10-m Height Direction of Backward-
Minus Surface Water Wind Speed in-Time Trajectory
Temperature (°C) (m/s) (+ 15°) From Crib Site.
8.1
5.7
7.2
6.3
5.9
1.3
6.4
4.7
-0.9
-1.9
-
3.2
8.3
6.0
7.4
6.5
6.1
3.5
3.0
3.5
2.2
1.4
-
5.7
270°
250°
300°
310°
15°
45°
100°
30°
5°
-
90°
CTl
co
Table 1. Meteorological parameters measured during September, 1977 and May5 1978.
-------
69
Lake-Breeze Events Non-Lake-Breeze Events
"S04] / [Pb]
[S04]/[A1]
]S04] / [Zn]
[S04J/[Fe]
[S04J/[Mn]
[so4]/[p-po4]
61
52
94
4.2
67
1605
38
50
64
1.7
20
808
Table 2. Ratio of sulfate mass concentration ((SOJ) to mass
concentration of certain trace metals and phosphate
for two lake breeze (May 18 and 25) and two non-
lake breeze (May 17 and 24) periods.
-------
70
Figure Captions
1. Location of Sampling sites.
2. Sample tubing loss correction factors as a function of particle
diameter.
3. DN/D (log d) as a function of particle diameter. Straight line
is -3 Jungian slope, included for reference. 18 May, 1978,
12:42 - 12:45 CDT altitude 60 m, nearshore shows the "bump"
whereas 18 May, 1978, 12:12 - 12:19 CDT altitude 1900 m,
midlake does not.
4. Same as Figure 3 except the plots are of DV/D (log d) as a function
of particle diameter.
5. Wind directions measured by aircraft over Lake Michigan during a
lake breeze event.
6. Backward-in-time trajectories on May 18, 22, and 25, 1978.
7. Total particle volume measured in the range 0.11 < d < 2.0 ym as
a function of time.
-------
71
43° 00'
o
o
o
00
CO
42° 30'
o
ro
o
1^
CO
O
O
o
h
00
O
rO
o
CD
00
LAKE MICHIGAN
O
O
o
(£>
CO
42°00
MIDLAKE
HALFWAY POINT
CHICAGO •
CRIB SITE
0 10 20 30 40 50
4I°30
Figure 1. Location of sampling sites.
-------
b
CO
b
CORRECTION FACTOR
^ CTI C55
bob
O
00
0
I
I
I
B se SJO^DEJ. UOL^DBJUOO sso|_ 6uiqn^ 9[dujps
ZL
-------
73
5000 •
I
Whitby Grand Average
Continental Dist. FIT
May 18, 1978
1
0.5-^
12:42 - 12:45 CDT
60 m altitude, nearshore
12:12- 12:19 CDT
1900 m altitude, midlake-
0.1
Junge
-3 slope
TT~
0.1
—r~
0.2
0.3 0.4
0.5 0.7
D,yum
1.0
Figure 3. DN/D (log d) as a function of particle diameter.
Straight line is -3 Jungian slope, included for
reference. 18 May, 1978, 12:42 - 12:45 CDT altitude
60 m, nearshore shows the "bump" whereas 18 May 1978,
12:12 - 12:19 CDT altitude 1900 m, midlake does not.
-------
74
ioo H
m
E 50-!
o
I
Q
D) 10-
May 18, 1978
12:42- 12:45 CDT
60 m altitude, nearshore
5-J
0.1-J
T
0.1
Whitby Grand Average
Continental Dist. FIT
T
0.2
0.3
1
0.4
I I I I I
0.5 0.7 1.0
D,yum
Figure 4. Same as Figure 3 except the plots are of DV/D (log d)
as a function of particle diameter.
-------
75
r
800
850
900
950
960
970
980
99C
100C
MB
Aircraft-measured wind directions -
May 25. 1978 1145-1330 CDT ^
I = Northward w
-
-
-
-
nd \
\
\
i
1 ^ /
\ ^
\ ^
7 • t
t f
I t 1 -
N' t
: ^ , ! :
X N \ / -1
_ -«-
CRIE
i *"" HALFWAY | MIDLAKEl
Figure 5. Wind directions measured by aircraft over Lake Michigan
during a lake breeze event.
-------
76
Figure 6a. Backward-in-time trajectories on May 18, 22,
and 25, 1978.
-------
-5
ro
en
cr
May 22
-------
-s
ro
May 25
c»
-------
79
34
32
30
28
26
24
C\j
o
x 22
en
I 20
u
^
18
16
LL
5
O = Sept. 26, 1977
A = Sept. 27,1977
I
6:30 7:00 7:30 8.00
TIME (CDT)
8:30
9:00
Figure 7. Total particle volume measured in the range
0.11 < d < 2.0 ym as a function of time.
-------
80
APPENDIX B
SHORT COMMUNICATION
Chemical Uniformity of Atmospheric Aerosol
Its Violation at a Mid-Lake Erie Site
bv
H. Sievering, D.A. Dolske, and V.E. Jensen
Environmental Sciences
College of Applied Sciences
Governors State University
Park Forest South, IL 60466
USA
ABSTRACT
The hypothesis of elemental mass uniformity of continental aerosol
is not supported by data collected over the Great Lakes. The twofold
higher sulfate concentration in air over Lake Erie compared with that
over Lake Michigan may be the dominant factor in producing the compositional
nonuniformity.
-------
81
Peirson et al . (1974) observed a uniformity in the elemental
composition of aerosol at eight locations throughout the United Kingdom
(UK). Rahn (1976) also noted a striking uniformity in relative elemental
composition in remote areas of the world; a similar uniformity has been
observed in urban (Israel, 1974) and rural (Stolzenburg and Andren, 1981)
areas of the United States. Rahn (1976) has suggested that urban, rural
and remote continental aerosol types should all display the same relative
proportions of elements. Further, it has been proposed (Stolzenburg
and Andren, 1981) that simple dilution can account for most of the observed
mass concentration differences between remote continental and urban
aerosols. Taken together these arguments will here be called the
compositional uniformity hypothesis. This hypothesis is tested below
against two aerosol/elemental data bases gathered over the Great Lakes.
Great Lakes sampling and analysis methods have been published
previously (Sievering et al., 1980) and will be restated only briefly.
Aerosol particulate matter was collected on three-part cellulose filters
in a three-stage cascade impactor. This configuration provided fine
(diameter, r < 0.5 ym) and coarse (r > 1 ym) particulate fractions for
subsequent elemental mass analysis by inductively coupled argon plasma
emission spectroscopy (ICAP) and for sulfate mass analysis by the BaSO»
turbidimetric method. Recovery studies indicate > 90% for all except
Al, for which 50-70% recoveries were determined. Al concentrations
determined by ICAP were enhanced by a factor of 60°/ to compensate for
poor recovery. Gravimetric analysis for total aerosol mass was also
performed. Shipboard sampling was performed using a boom approximately
5 m ahead of the ship's bow; the individual filter sets consisted
of air volumes from 200 to 700 m . A light-scattering device,
the active scattering aerosol spectrometer (ASAS), was used to
-------
82
monitor the in situ number concentration distributions for the 0.1- to
3.5-ym diameter size range. Meteorological parameters were monitored
aboard ship. These data, in conjunction with land-based meteorological
data, afforded the construction of back-trajectories that will be described
later in this paper. Great Lakes data considered here were obtained on
Lake Erie during the summer of 1979 at 81°30'W, 4?°00'N and on Lake Michigan
during the summer of 1977 at 87°00'W and 42°00'N (see large + symbols in
Figure 1). Both sites are from 40 to 100 km removed from shoreline aerosol
sources. The Lake Michigan site was only 40 to 50 km from, and often
downwind of, the Chicago urban complex whereas the Cleveland area was
about 70 km from, and less often upwind of, the Lake Erie site.
Comparison of the mass percents (i.e., percentage of total aerosol
mass) for each of seven elements observed by Peirson et al. (1974--Britain),
King et al. (1976--Cleveland), the 1964-75 U.S. urban mean (Israel, 1974),
Stolzenburg and Andren (1981—U.S. rural), and the Lakes Michigan and
Erie data sets (after corrections for poor Al recovery) are shown in Tab-le 1 .
With the occasional deviation of Na the decreasing mass order of the
elements is Fe, Al, Pb, Na, Zn, Mn, and Cu for all locations except the
Lake Erie data set. In this last case the order is Fe, Zn, Al , Cu, Pb,
and Mn, the Zn and Cu relative percents being especially high. Scrutiny
of this last column of Table 1 indicates the soil source may have
contributed a full order of magnitude less to Lake Erie aerosol mass
and chemical character compared with the other locations; the mass
percents for Al and Fe (largely soil-derived elements) are tenfold
lower in the Lake Erie case. Wet soil conditions and washout by rain
showers in the vicinity of and over Lake Erie during and before ship
outings is probably the best explanation for the very low soil contribution.
-------
83
-3
However, the relatively high value of 35 yg m for the total aerosol
mass observed over Lake Erie does not readily support the notion that
a major source such as soil was so lacking. This dichotomy may be
_ O
explained by the Lake Erie sulfate (SO.) mass of 9.6 yg m~ versus
only 5.4 yg m over Lake Michigan; the greater sulfate mass over Lake
Erie may have sufficiently offset the low soil source contribution to
_3
cause the only 10 yg m total aerosol mass difference over the two
lakes.
Can the depleted soil source explain the Lake Erie data set's
violation of the compositional uniformity hypothesis? Closer scrutiny
of the Great Lakes data is required. Table 2 presents a comparative
listing of enrichment factors (EF) for the Lakes Erie and Michigan and
UK data sets (including sulfate), and the mass percents and fine/coarse
ratios (i.e., ratio of elemental mass concentration in the aerosol of
r < 0.5 ym to that in the aerosol of r > 1 ym) for Lakes Erie and
Michigan. The enrichment factor is a ratio of ratios; the numerator
is a ratio of trace elemental concentration to Al concentration for air,
and the denominator is the same ratio for soil . Here we use the average
soil ratios given by Bowen (1966), but note that all values are considered
only approximate. This is not only caused by the poor Al recovery but also
by the fact that local Ohio soil dust is probably enriched in Zn, Cu,
Mn, and Pb compared to Bowen's bulk soil analysis (Stol zenburq, 1982).
The EF Comparison does suggest that, ove>" Lake Erie, 7n and Cu were
enhanced, whereas Mn and Pb were depleted. Any contribution by a lake
aerosol source could not have caused this result (Sievering et al., 1983).
This data set's violation of the compositional uniformity hypothesis does
not, then, appear to result wholly from the depleted soil source.
-------
84
Stolzenburg and Andren (1981) have shown that local industrial
sources may cause the contribution of Fe to the aerosol mass to
be greater than that of Al and, less often, those for Zn and f1n
to be greater than that for Pb. This may have been the cause for
[Fe] > [Al ] in some cases of Table 1. However, since no industrial
sources were within 40 km of the Great Lakes sampling sites, a
combination of more distant anthropogenic sources, transport and
transformation must have caused Lake Erie EF for Zn and Cu to be so large
at the same time that EFs for Mn and Pb are so small.
The Lake Erie data set presents a clear violation of the
compositional uniformity hypothesis, despite the fact that the
fine/coarse ratios for the elements (except Pb) and also for sulfate
are essentially the same over both Lakes Michigan and Erie.
Note, however, the more than twofold larger S04 mass percent
contribution over Lake Erie; the SO^ mass concentrations of
9.6 yg m over Lake Erie and 5.4 yg m~ over Lake Michigan
constitute 27% and 12% of their respective total aerosol masses.
Although no additional chemical analyses were available to
characterize the Great Lakes aerosol mass, the substantially
greater S04 contribution to the aerosol mass over Lake Erie may
provide an explanation for the observed violation of the compositional
uniformity hypothesis. Further data obtained over Lake Erie provide
clues in this regard: 1) aerosol size and number distributions
obtained by the ASAS and 2) meteorological back-trajectories.
-------
85
A simple back-trajectory calculation followed the sampled aerosol
from the Lake Erie ship location back toward shore. The back-trajectory
procedure druing traverse over water was calculated by a height-weighted
i
average of the ship's horizontal velocity and a triangulation average
of reported wind data at 1000 and 2000 ft from the National Weather
Service (NWS) upper-level sounding stations at Buffalo, New York,
i
Pittsburgh, Pennsylvania, and Flint, Michigan. When the aerosol reached
shore, weighted averages of NWS station data were used to complete the
back-trajectory. This overall back-trajectory approach has been compared
I
with seven other trajectory
techni
iques for application to the Lake Erie
data base (Jensen, 1981). It was! found to be quite accurate even
!
relative to complex computer backj-trajectory techniques such as that
of Ferber and Heffter (1977).
Figure 1 shows the most likely 24-h, straight-line, back-trajectories,
More highly time-resolved trajectories will, of course, follow a different
path (but reach the same point after 24 h). The figure shows 21
trajectories; in four cases out of the 25 filter set samples obtained
over Lake Erie the meteorological
setting was too complex (e.g., lake
breeze events) to establish a trajectory. Note that numbers from 1 to
42 appear on Figure 1 next to vertical bars of various lengths. These
bars are quantitative indicators of the relative magnitude (and the
!
adjacent numbers qualitative indicators) of the 42 largest S0~ point
sources (USEPA, 1981) in the states of Ohio, Indiana, Illinois, Michigan,
and Wisconsin. The 10 largest SOU sources on a U.S. state-by-state
! I
aggregate basis (Hileman, 1982) a|^e the following (in kilotons yr ):
Ohio (2600), Pennsylvania (2000), Indiana (1950), Illinois (1450),
-------
86
Missouri (1300), Texas (1200), Kentucky (1100), and Florida, Nest
Virginia, and Tennessee (all 1050). Michigan at 900 kilotons yr~ ,
Wisconsin and the lower peninsular portion of Ontario province, Canada
(both at 650 kilotons yr~ ) are each substantially smaller SO- source
areas than Ohio, Pennsylvania, Indiana, or Illinois. It is well known
that S02 gas is a precursor for SO, aerosol. The back-trajectories of
Figure 1 show that in all but a few cases (e.g., filter set 1050)
there is a very large SOo contribution (e.g., filter sets 1090 and 1170)
to air parcels later sampled over Lake Erie.
Figure 2 shows a plot of the aerosol number (AN/Ar) versus size
(radius, r) for set 1050 averaged across more than 6-h of ASAS 5-min
sampling periods. The sloping solid line is the least-squares best
fit to the AN/Ar data for 0.13 < r < 1.3 ym. Figures 3 and 4 are
similar plots but for sets 1090 and 1170. The slope of the solid lines
are the following: for set 1050, -4.2; for 1090, -5.3; and for 1170,
-5.35. Junge (1956) calculated that for an aerosol population to be
classified as of natural continental origin the AN/Ar slope in this
radius range should equal -4.0. Other investigators have since shown
that limits of 3.5 to 4.5 are reasonable for classification as a natural
continental aerosol population (Israel, 1974). On this basis the aerosol
distribution for set 1050 may be claimed to be continental, whereas the
distributions for sets 1090 and 1170 are clearly not. Of the 21 filter
sets for which ASAS data were obtained, only five had slopes fitting
the above continental criterion; they are the five that traversed the
state of Michigan (see Figure 1). The mean AN/Ar slopes for these five
-------
87
sets (including 1050) is -4.2 ± 0.2; four of these five sets have
elemental mass percents that fit the compositional uniformity hypothesis.
The mean AN/Ar slopes for the remaining 16 filter sets in the ASAS
group (including 1090 and 1170) is -5.2 ± 0.4; this is significantly
different from that which characterizes a natural continental aerosol
population. In all but six of these 16 cases the compositional uniformity
hypothesis is clearly violated.
The mean SO. concentration for the ten filter sets that do not
_3
violate the hypothesis is 3.8 yg m ; for the remaining 15 sets the
mean SO. concentration is 11.9 yg m . This is especially noteworthy
in light of the fact that the S04 contribution is largely by small
particles as is shown by the fine/coarse ratios in Table 2. Further,
the example comparison of Figures 3 and 4 with Figure 2 shows that
below 0.2 ym radius the aerosol population is a full order of magnitude
greater for those sets that violate the hypothesis (sets 1090 and 1170)
than for the set that does not (set 1050). These data suggest that the
S04 conversion process may be an important contributor to the compositional
nonuniformity observed in the case of the Lake Erie data set. Certainly
simple dilution can not account for the observed elemental concentration
differences, since the continental slope criterion v;ould then be
observed. In any event, there is a clear enhancement of < 0.2 ym radius
aerosol for those cases when the chemical uniformity hypothesis is violated.
This in turn suggests that more distant anthropogenic source aerosol
may cause nonuniformity in chemical composition just as local sources
are known to cause this nonuniformity. Thus, it does not appear that
chemical uniformity of continental atmospheric aerosol removed ^100 km
from anthropogenic sources is necessarily to be expected. Until further
evidence is found identifying and explaining exceptions to the "rule" as well as
-------
supporting generality of the chemical uniformity hypothesis, it
should remain a hypothesis only.
-------
89
REFERENCES
Bowen, H.J.M. Trace Elements in Biochemistry (Academic Press, New York,
1966).
Ferber, G.J. and J.L. Heffter. Development and Verification of the ARL
Regional-Continental Transport and Dispersion Model (NOAA Air Resources
Lab., Washington, D.C., 1977).~~~
Hileman, B. Environ. Sci . & Tech. 16, 323A-327A (1982).
Israel, H. Trace Elements in the Atmosphere (Ann Arbor Science, Ann
Arbor, MI, 1974).~
Jensen, V.E. An Evaluation of Several Methods of Atmospheric Trajectories
(Dept. of Geog., Northern Illinois Univ., DeKalb, IL, 1981).
Junge, C.E. Tellus 8, 127-139 (1956).
King, R.B., J.S.Fordyce, A.C. Antoine, H.F. Leibecki, H.E. Neustadter,
and S.M. Sidik. J. Air Poll. Cont. Assoc. 26, 1073-1078 (1976).
Peirson, D.H., P.A. Cawse, and R.S. Cambray. Nature 251 , 675-679 (1974).
Rahn, K.A. The Chemical Composition of Atmospheric Aerosol (Qrad. School
of Oceanog., Univ. of Rhode Island, Kingston, RI, 1976).
Sievering, H., M. Dave, D.A. Dolske, P.L. Huges, and P. McCoy, 1979. An^
experimental study of lake loading by aerosol transport and dry
deposition in the southern Lake Michigan basin. EPA-905/4-79-016.
Sievering, H., M. Dave, D.A. Dolske, and P.A. McCoy. Atmos. Environ. 14,
39-53 (1980).
Stolzenburg, T.R.(1982). Trace element composition of local Ohio soil
dust. Private communique.
Stolzenburg, T.R. and A.W. Andren. Water, Air, and Soil Poll. 15, 263-270
(1981 ).
-------
TABLE 1. Percentage of Total Aerosol Mass for Each of Seven Elements
Element
Fe
Al
Pb
Na
Zn
Mn
Cu
Aerosol
Mass.,
Great Britain
0.9
0.6
0.35
2.5
0.35
0.05
0.04
-45
Clevel
Urban
3.8
2.5
0.65
0.7
0.35
0.1
0.1
115
and
Suburban
3.8
4.0
1.1
0.8
0.55
0.15
0.15
45
U.S. Urban U.S. Rural
1.5 1.8
* 2.0
0.75 0.5
* 0.4
0.65 0.1
0.10 0.06
0.09 0.05
105 40
Lake
Michigan
1.1
0,1 .0
0.5
*
0.2
0.05
0.04
45
Lake
Erie
0.13
-0.1
0.01
*
0.11
0.002 °
0.03
35
* no data available
-------
TABLE 2. Aerosol statistics for Lakes Erie and Michigan, and comparisons of their
enrichment factors with those observed by Peirson et al. over Great Britian
Element
Al
Fe
Pb
Zn
Mn
Cu
SO,
Dry Mass Percent
Erie Michigan
0,0
0
0
0
0
0
27
.1
.13
.01
.11
.002
.03
0
1
0
0
0
0
12
.6
.1
.5
.2
.05
.04
Fine/ Coarse Patio
Erie Michigan
1
1
5
4
2
3
23
.8
.9
.9
.8
.1
.3
1.2
1.6
13
6.1
2.0
3.9
27
Enrichment Factors
Erie Michigan UK
1 1
o,2 . 4 0,2 . 1
o,720 o,3200
0/1500 o,320
o,2 . 0 o,5 . 9
o,l 1 00 o,l 20
n .a . n .a .
1
o,2.
3700
925
9.
165
n .a
9
9
m
-------
Figure 1. Great Lakes study region with the most likely 24-h, straight-line
back-trajectories for each of 21 filter set samples. The 42
largest SO,, point sources in this reqion are also shown.
-------
10'
10'
E
h_
O
0.01
93
f.\
i
. Set # 1050
On 7 Sept. 1319
Off 7 Sept. 1935
0.1
1.0
10.0
Figure 2. Aerosol number versus aerosol radius for filter set #1050,
-------
94
108
105
10"
I
10=
i02
10
0.01
JL.
.Set # 1090
On 17 Oct. 0500
Off 17 Oct. 1157
0.1
1.0
10.0
Figure 3. Aeorsol number versus aerosol radius for filter set #1090.
-------
95
10'
10s
104
u
10=
102
0.01
Set # 1170
. On 19 Oct. 0748 .
Off 19 Oct. 1050
0.1
1.0
10.0
Figure 4. Aerosol number versus aerosol radius for filter set #1170,
-------
96
APPENDIX C
TECHNICAL NOTE
TRACE ELEMENT PASS-THROUGH FOP CELLULOSE IMRACTOR .SUBSTRATES AMD
FILTERS WHEN USED FOR AEROSOL COLLECTION
Donald A. Dolske* and H. Sievering
Environmental Science Program
College of Anplied Science
Governors State University
Park Forest South, Illinois 60466
ABSTRACT
Filter papers and impaction substrates made of cellulose fibers,
such as Whatman 41 and Misco P810/252, are of considerable utility in
the collection of aerosol for subsequent trace elenental analysis.
This experiment evaluated the performance of Misco P810/252 in collecting
trace elements, relative to a co-located standard glass fiber filter
hi-vol collection. Sampling was conducted in varying meteorological
conditions, so that results might be expressed in terms of environmental
variables such as temperature and relative humidity. The pass-throuqh
factors presented here were derived from a series of environmental samnles
collected over land and over water. Overall mean cellulose Mi'sco filter
collection of Pb and Zn was found to be 38% and 32?^ less
than that collected on the alass fiber filter.
Present address: Atmospheric Chemistry Section, Illinois Department
of Energy and Natural Resources, State Mater Survey Division, P.O. Box
5050, Station A, Champaign, Illinois 61820.
-------
97
EXPERIMENTAL DESIGN AND EXECUTION
The purpose of the experiment was to compare the collection efficiency
of the Misco cellulose impactor substrates and filter in a modified Sierra
1-, 3-, 6-slot impactor with the assumed near-100-percent collection
efficiency standard type A glass fiber filter. The problem, restated,
is to determine if the total concentration computed from the sum of
impactor stages 1 and 3, and backup filter, stage 6, can be related
to the total concentration of aerosol gravimetrically measured with
standard type A glass fiber filters. This experiment, then, has particular
significance for data of Sievering e_t aj_. (1980) and Puce e_t aj_. (1976),
since all trace element and nutrient data for aerosol over Lakes Michigan
and Erie were collected using the ,f/isco cellulose media.
All aerosol collection was done with paired flow-controlled hi-vol
3 -1
(1.1 m rm'n ') samplers. Flow rates of the samplers were recalibrated
after each group of five sampling runs, approximately one- to two-week
intervals, to assure comparability of sample volumes. The samplers were
set up not loss than three meters apart, and were fitted with plastic
exhaust hoses to minimize reentrainment of cooper aerosol shed by pumn
commutators. One pair of samplers was operated at the Governors State
University air quality monitoring site from March to August, 1980. A
second sampler pair collected aerosol at the city of Chicago 68th Street
crib from June to August, 1980. A filter holder cassette containing,
in order, Misco slotted substrates on stages 1 and 3 and. a 20 by 25
centimeter Misco filter on one hi-vol sampler; a standard type A glass
fiber 20 by 25 centimeter mat was used on the other sampler. Behind
each of these filter cassettes, separated by 0.5 centimeters and supported
by fine stainless steel screen, was mounted a "second backup," a type A
glass fiber mat, of Gelmari Spectrograde material. The Spectrograde filters
-------
98
collect aerosol as the standard type A, but have a lower trace element
blank. In all, then, there were six parts to each concurrently collected
set of filters: the three-stage Misco impactor/backup with its spectroqrade
second backup, and the type A filter with its Spectrograde backup. These
parts were designated Ml, M3, M6, S^M, and A, SRA, resnectively.
Sets of filters were exposed for periods ranging from 24 to 96 hours,
depending on variability of meteorological conditions over the sampling
period and degree of pollution loading prevailing. In general, an effort
was made to conduct sampling on a given set during similar conditions, so
that parameters derived for that sample could be related to meteorological
variables. Each fifth sample was a field blank, in which the filters
were carried through the entire procedure, except that the hi-vol motors
remained off during the exposure period, usually 24 hours for the field
blanks. Several unexnosed blanks were also used, where the filters were
carried through all of the procedure except mounting on the hi-vols.
In all, 37 samples and 16 blanks were collected during the course of the
experiment.
Continuous monitoring of ambient temperature, wind speed, wind
direction, barometric pressure, and relative humidity was done at Governors
State throughout the experiment. In addition, wind speed, wind direction,
temperature, and relative humidity were recorded hourly at the 68th Street
Crib while sampling was being conducted there. These data were averaged
for each period during which a set of filters was being exposed. ~!"he
mean values were then used to characterize each period. At the Governors
State site, two optical particle-counting devices continuously monitored
particulate levels. Results fron an automatic condensation nuclei counter
(Environment One Corn. Rich 100) and an integrating nephelometer (Meteorology
Research Inc. Model 1550) were similarly averaged over each sampling period.
-------
99
Analysis of the various filter stages followed several different
procedures. The Ml, M3, and M6 stages were first trimmed with Teflon-
coated stainless steel scissors to renove the unexposed edges. The
exposed portions were then placed in fused quartz conbustion boats
which had been acid washed with Ultrex HC1. The boats were then run in
a low temperature oxygen plasma asher (Plasmod Inc.) at 75 watts PF
for about 24 hours, i.e., until all the cellulose filter material was
oxidized. The residue was then brought into solution with hot 12N Ultrex
HC1 , and volune-normalized with distilled deionized water. The SPA and
SGM stages were gravimetrically analyzed for total aerosol mass, and
then were trimmed of unexposed parts, cut into strips, and leached in
a boiling HC1 + HNCL bath for 60 minutes. These filters were then
rinsed with distilled deionized water and the leachate brought to a
normalizing volume. The type A filters were -gravimetrically analyzed
for total aerosol mass only. The liquid samples resulting from the
preparation of the Ml, M3, M6, SGA, and SGM filters were then analyzed
by atomic absorption spectrophotometry for Pb, Zn, and Fe.
RESULTS
As can be seen in Table 1, the overall nean concentrations of Pb,
Zn, and Fe fall within the range of concentrations for these elements
in aerosol in the Great Lakes region and elsev;here. Samples in
this study were collected on a routine 24- to 96-h run-time basis,
while the overlake samples' run-times were much shorter, usually 3- to 6-h,
The longer run-times for the pass-through experiment were necessary in
order to collect sufficient material on the second stage backup filters,
-------
100
SGA and SGM. Even with the extended run-times, Fe on these stages was
quite often below analytical detection limits. (NOTE: detection limits,
laboratory and field blanks for the pass-through samples were comparable
to the values reported for Lake Michigan samoling [Sievering e_t a]_. , 1980]
Still, the pass-through factors (RTF) reported here,
'
PTF =
,
Ml ,3, 6 SGM
should be representative of the percentage anounts of material missed by
the three-stage modified Sierra 1, 3, 6 slot impactor arid Misco cellulose
media, as used in several studies. The PTF values given here would
most likely be slight underestimates, as loading of the filter below a
critical stage should only slightly increase its collection efficiency.
The value of PTF for Pb and Zn not only can be applied to those
elements, but also to other fine particle (n < 1.0 ym) associated elements.
Similarly, the PTF for Fe reported here could also apply to similarly
particle-size associated elements (D > 1.0 un). Table 1 gives the overall
mean PTF for each of the three elements measured. Table 2 shows some of
the stronger associations that could be found between the measured
parameters. Although the correlation coefficients are very weak, expected
general trends appear. PTF and percent concentration fine aerosol (°tf)
seem to be directly related. RH is weakly and inversely related with PTF,
as would be expected due to the hygroscopic nature of the cellulosic
media. Percent fine aerosol association is defined here as:
100%
Separation of the five highest and lowest RH cases and computation
of PTF values for those regimes, shown in Table 1, again reflects the
-------
101
inverse relationship, but without strong statistical significance.
(NOTE: Fe was undetectable in most of these ten cases.)
The main conclusion of this study, then, is that the overall mean
RTF values in Table 1 are relevant correction factors for the size-
fractionate trace element in aerosol data of Sievering et_ al_. (1980),
and Duce ejt al_. (1976) and others. Based on %F as a convenient, although
low-resolution, size-association indicator, the RTF results for Pb, Zn,
and Fe can probably be reasonably extended to other elements in the
overlake detabase. Applicability of these RTF values to fine aerosol
data collected via cellulosic media by other researchers would require
a careful evaluation of the exact sampling technique employed, but will
hopefully be useful.
REFERENCES
Duce, R.A., Ray, B.J., Hoffman, R.L., and Walsh, P.R. (1976).
Trace metal concentration as a function of particle size in
marine aerosols from Bermuda. Geophys. Res. Lett, .3,
339-342.
Sievering, H., Dave, f*., n0lske, n., and McCoy, P. (1980).
Trace element concentrations over midlake Michigan as a
function of Meteorology and Source Region, Atmospheric
Environment, 14, 39-53.
-------
102
Table 1. Mean Pass-Through Factors (RTF) and Concentrations (f).
C, ng m
__ _____ __
Pb Zn Fe Pb " Zn "" ' ce
Overall 109 61 165 37.6 ± 6.4 31.9 ± 6.1 11.9 ± 3.3
RH > 88 202 84 nd 36.1 ± 12.1 30.7 ± 10.3 nd
RH < 65 80 47 nd 42.8 ± 10.8 38.2 ± 12.6 nd
Table 2. Some Correlations Between Variables.
RTF (Zn) 2F (Pb) *F (Zn) RH
RTF (Pb) +.52 '-.25 +.44 -.24
RTF (Zn) +.23 +.84 -.19
°/F (Pb) -.19 +.15
XF (Zn) + .11
F = Fine aerosol (D < 1 vim)
-------
103
APPENDIX D
CHEMICAL ELEMENTS IN ATMOSPHERIC AEROSOL
OVER SOUTHERN LAKE MICHIGAN:
THE CONTRIBUTION OF THE LAKE SOURCE
Herman Sievering and Donald A. Dolske
Envirormienta1 Science Program
College of Applied Sciences
Governors State University
Park Forest South, IL 60466
-------
104
ABSTRACT
Chemical element balance analysis was applied to atmospheric aerosol
data obtained over Lake Michigan for 12 elements, in order to ascertain
the contribution of each of seven source types, especially Lake Michigan
itself. It was found that the lake contributed about twice as much to
total aerosol mass in both diameter d <1ym (fine) and d > lym (coarse)
aerosols as did soil, although the soil source aerosol contribution may
have been anomalously low. The fine aerosol lake source appears to be
dependent upon mean windspeed; the coarse aerosol lake source appears
to function most of the time.
The samples were collected from May 1978 through January 1979, on
cellulose media with modified Sierra cascade impactors. Summer and fall
sampling predominated. Sampling was done at a City of Chicago water
intake crib, 3.2 km from shore on southern Lake Michigan, 87°32'W, 41°47'N.
Air concentrations were determined by inductively coupled Ar plasma
atomic emission spectroscopy. Mean concentrations were generally 2 to
4 times higher than means observed one year earlier at midlake, 87 OO'W,
42°00'N, 50 km from the nearshore crib site. Size distributions of the.
elements were similar at both sites, except for those of Ca and Mg which
appear to be due to the lake source.
-------
105
INTRODUCTION
The urban-industrial complex surrounding the southwestern and
southern shore of Lake Michigan contributes significant amounts of
anthropogenic aerosol to the atmosphere. Prevailing winds in this
region give rise to particle trajectories that pass over the lake more
than 70 percent of the year (NOAA, 1975). Mesoscale recirculation by
lake breeze events (Lyons and Olsson, 1973), unstable spring and autumnal
air overlake, and enhanced turbulence during non-steady-state conditions
(Hughes and Sievering, 1982) increase the probability of atmospheric
pollutant aerosol contributing to the chemical loading of Lake Michigan.
The extent and intensity of the Chicago regional air pollution source
(Winchester and Nifong, 1971; Gatz, 1975) is such that moderately
efficient (^10 percent) transfer between air and water would be detrimental
to water quality in the lake. Trace metals (Eis^hreich, Emmling, and
Beeton, 1977; Dolske and Sievering, 1979) and nutrients (Murphy and Doskey,
1977; Dolske and Sievering, 1980) have been increasingly implicated as
significant atmospheric contributions to the total pollutant burden of
Lake Michigan.
Processes affecting the overlake transport and deposition of
pollutant aerosol are complex and not well understood. In order to
expand the chemical and physical characterization of aerosols over
southern Lake Michigan, much research has been conducted in recent years.
Sievering et al. (1979) collected aerosol and meteorological data in
1977 at a midlake site (87°00'W, 42°00'N). Data, reported here, were
collected at the City of Chicago 68th Street water intake crib (87°32'W,
-------
106
41°47'N) from May 1978 through January 1979. Several recent aerosol
sampling programs (Williams and Muhlbaier, 1978; Fingleton and Robbins,
1980} also collected data at the crib site. Comparison of these crib
data with the 1977 midlake data is here undertaken to lend insight
regarding the overlake transport and chemical character of particulate
matter. Further, chemical element balance (Gordon, 1980; Cooper and
Watson, 1980) analysis was undertaken to quantify the contribution of
several aerosol source types to the particulate matter observed at the
crib site.
-------
107
METHODS
In May and June 1978, flow-controlled high-volume samplers (General
Metal works, GMW 2000) were operated at the upper box-beam bridge which
runs between the two main buildings of the crib (see figure 1). The
sampler intakes were 13 m above the lake surface, midway between the
two structures. Standard meteorological sensors were also located at
the position. A digital system recorded and printed 15-min means for
the continuously-monitored meteorological parameters. From July 1978
to January 1979, the hi-vol samplers were operated on the lower bridge
of the crib 6 m above the surface. During these months, the meteorological
sensors were placed on a movable, 3-m mast which was erected on the upwind
side of the hexagonal sea wall for each day's sampling.
In order to eliminate contamination from daily operation at the
crib, permissible wind direction (WD) sectors were defined. Samples
were considered valid only if WD + la for the entire sampling period
was within the sectors 40° < WD < 130° or 210° < WD < 330° (see figure 1).
The samples were collected on cellulose impactor substrates (MISCO P252A)
and 20 x 25 cm backup filters (MISCO P810A). Sierra #235 five-stage
impactors—modified to use only stages #1 and #3 along with the backup--
were exposed at 1.13 m min (40 SCFP1) for periods of 3 to 7 hours. This
length of exposure was sufficient to collect enough aerosol for the
chemical analyses, yet short enough for meteorological parameters to
remain fairly constant during each sampling period.
The exposed cellulose filters were analyzed by two separate methods.
The impactor strips and about one-half of the backup filter were oxygen-
-------
108
plasma ashed at 75 watts RF. The ash residue was dissolved in hot
6 N HNO~ and brought to a standard volume with distilled deionized
water. This procedure gave three subsamples for each filter set:
two impactor stages (#1 and #3) collect coarse particles of diameter
(d) > 1.0 ym while the backup collects fine particles, d <1ym. All
three subsamples were analyzed by inductively coupled Ar plasma atomic
emission spectroscopy (ICAP). Elements identified were Al, Ca, Cd, Cu,
Fe, Mg, Mn, Mo, Na, Ni, Pb, Ti, Sn, V, and Zn. Samples taken concurrently
and prepared by an alternate method (Dolske and Sievering, 1980) were
analyzed at USEPA-Central Regional Laboratory for total phosphorus,
nitrate, and sulfate by automated colorimetric methods. A summary of
trace metal analytic detection limits, procedural blanks, and typical
sample values is given in Table I. Note that Al values may be somewhat
low since HF dissolution was not used. It should also be noted that ICAP
Cu data are considered valid not only because the procedural blank value
is small when compared with typical samples, but also because the hi-vol-^
motor exhaust was separated from the filter sampling.
Duplicate analyses of replicate samples and samples into which knewn
amounts of certain elements were spiked (Cu, Pb, Zn) indicate that ICAP
results were reproducible to within + 20 percent. A second portion of
the backup filter from 12 filter sets was sent to Argonne National Laboratory
for x-ray fluorescence (XRF) analysis of Br, Fe, Mn, Ni, Pb, Se, and Zn
content. The results of this limited analytical intercomparison are given
in Table II. For Pb, Zn, and Fe the concentrations were in fairly good
agreement, but the XRF results for Mn and Ni gave consistently higher
concentrations, resulting in large percent differences from ICAP concen-
trations. The concentration values in the second column of Table II are
-------
109
means of all analytical results from both ICAP and XRF. Br and Se were
not included in the ICAP analyses.
Meteorological data collected at the crib were complemented by
hourly reports from National Weather Service and U.S. Coast Guard stations
around southern Lake Michigan. Also, during the May 1978 sampling period,
a National Center for Atmospheric Research aircraft collected meteorological
and aerosol number concentration data during flights that included the
crib site as a sampling point (Hughes and Sievering, 1982). From these
data, a simplified back-trajectory calculation determined the horizontal
and vertical spread of the aerosol collected at the crib. The horizontal
spread in the trajectory was assumed to lie within two standard deviations
about the mean wind direction. The vertical spread was determined through
the application of stability category dispersion coefficients determined
as a function of the bulk Richardson number (Nagib, 1978).
-------
no
RESULTS
The back-trajectory data sets are divided into two distinct groups.
The first consists of data representing a direct Chicago-area-to-crib
traverse (figure 2a, for example). The second group consists of data
representing aerosol trajectories strongly influenced by a lake breeze
recirculation (figure 2b). The source area for all the data sets appears
to have been the general Chicago region (including Gary, Indiana),
regardless of the wind direction prevailing at the crib during aerosol
collection. This was the result of restrictions placed on mean WD for
the entire set for valid, uncontaminated samples, and the small vertical
rise in trajectories upon reaching shore.
Rahn (1976) has concluded that geometric means may be the most
appropriate representation of environmental concentrations of trace
elements for they tend to have lognormal distributions. A direct comparison
of nearshore and midlake geometric mean concentrations of 16 elements and
compounds is shown in Table III. The concentrations are geometric means
for all data sets collected where the wind direction was within permissible
sectors; each data set's concentration is the arithmetic sum of analytical
results for the impactor stages and the backup filter. For aerosol
collected at the midlake site (87°00'W, 42°00'N), sampling, analytical,'
and trajectory-plotting methods used in 1977 (Sievering et al., 1979)
were the same as those used at the crib in 1978, so the two data sets
should be comparable.
In general, mean concentrations at the crib site were two to four
times greater than the midlake means (Table III). A few elements (Cd,
Mo, Ni, V) that were below ICAP detection limit at midlake appeared in
-------
in
the nearshore samples. This was not an unexpected result because of the
proximity of the crib site to the intense Chicago regional aerosol source.
The mean concentrations at the midlake point were, in turn, five to more
than thirty times higher than the continental background concentrations
given by Rahn (1976).
It is interesting to note, however, that while the concentration of
nearly all elements were much higher at the nearshore site, aerosol
elemental composition was essentially unchanged. Enrichment factors for
each element of interest, x, were calculated using the average composition
of soils given by Bowen (1966):
EF =
(aerosol) /-\ $d1cator) (soil)
C (indicator)
Al was used as the soil-derived or natural-background aerosol indicator
element, with EF = 1. Values of EF > 1 indicate that an element is present
in amounts greater than might be expected from the natural, soil-derived
source. From Table IV, it can be seen that the most highly enriched
elements are Cd, Pb, Zn, Mo, Ni, and Cu.
The impactor arrangement used allows the first impactor stage to be
taken as collecting only coarse (d > 1 ym) particles. The second impactor
stage improves resolution between the coarse mode (1st stage) and fine
mode (backup). The distribution of an element with respect to particle
size can be crudely shown by a ratio between the fine and coarse particle
concentrations of that element:
Fine/Coarse Fraction -
-------
112
In Table IV, the fine/coarse fractions for the crib samples and
midlake samples are compared. For most elements there are no significant
differences between the two data sets. A fraction greater than one
indicates that the element is concentrated in the d < 1 ym or fine particle
range. Note that this is the case for all of the high-EF, anthropogenic
elements.
Another descriptor of the elemental content of the aerosol is percent
mass composition:
C (x)
% mass composition = —L-L
C (total aerosol)
where C (total aerosol) has been gravimetrically determined from standard
type-A glass fiber filters (20 x 25 cm). The values calculated here
(Table IV) conform closely to the data of Gatz (1975) for Chicago aerosol
and to the midlake values that had trajectories indicating the presence
of the Chicago source aerosol.
-------
113
DISCUSSION
Comparing the 1978 crib-site concentrations with the 1977 midlake-
site concentrations leads to two observations regarding qualitative
effects of overlake transport on the Chicago region source aerosol.
(Caution is expressed regarding quantitative comparison of the two
data sets since a number of factors may contribute to their differences.)
First, the fine/coarse fractions and % mass compositions (Table IV)
reveal three groups of elements. For Pb, Zn, and S the fine/coarse
fractions are slightly greater at midlake than at the crib while their
% mass compositions are less. (Of the elements identified by ICAP, these
are more strongly associated with fine particles.) Sedimentation loss of
coarse particles, gas-to-particle reactions, and coagulative growth
could all contribute to this outcome. For Al, Cu, Fe, and Mn (and
possibly N and P) the fine/coarse fraction was slightly larger nearshore.
In contrast to these two groups, Ca and Mg show large, significant
differences between fine/coarse fractions at midlake and nearshore (a
factor of three or more). The decrease in Ca and Mg fine/coarse fractions
during overlake transport might be due to the generation of aerosol by
the lake itself.
A second qualitative observation concerns the effect of lake-breeze-
induced mesoscale recirculation on the nature of crib-site aerosol. Of
the 23 crib filter sets, 18 could be separated into two groups; those
with which no lake breeze event was associated, and those for which a
well-developed mesoscale recirculation was indicated by surface and aircraft
data. For the lake breeze case data, overall concentrations at the crib
site were slightly lower than at midlake. This point is consistent with
large-particle sedimentation loss and dilution, as well as aerosol aging
within the recirculating plume (Lyons and Olsson, 1973). EFs for the two
-------
114
groupings of 8 lake-breeze and 10 no-lake-breeze filter sets show no
substantial differences except for Ca and Mg. For these two elements,
EFs are two and one-half times larger for the no-lake-breeze group than
for the lake-breeze group. This appears, at first, peculiar since
aerosol in a lake-breeze trajectory are present over the lake far longer
than are those aerosol in a nearly direct trajectory from shore to the
crib site. However, the mean windspeed at the crib was 3.4 + 1.5 ms for
the 8 lake-breeze filter sets and 6.0 + 1.0 ms for the 10 no-lake-
breeze filter sets. The combination of light winds (< 5 ms ) and
thermally stable air present over the lake during lake-breeze events
could "shut off" the lake aerosol source whereas winds greater than
5 ms"^ might introduce a lake source component to the total aerosol
observed at the crib site.
-------
115
Chemical Element Balance (CEB)
Since both qualitative observations above suggest a lake source
contribution to aerosol chemical character over Lake Michigan, a source
reconciliation was attempted using the chemical element balance (CEB)
approach (Gordon, 1980; Cooper and Watson, 1980). The computer analysis
developed by Watson (1979) was used, affording both fine and coarse fraction
aerosol source reconciliation. This particular computer analysis is
especially useful when considering the crib site data since the uncertainty
in each calculated source contribution is also determined. The reason
this latter point is useful is that the identification of a lake aerosol
source may be confused by the cement limestone manufacturing (or, less
likely, a soil-like anthropogenic) aerosol source; or highway or road
dust which are all similar to the lake source and also prevalent in
the Chicago region (Gatz, 1975; Winchester and Nifong, 1971). This
problem may be ameliorated somewhat by adding Na filter sets' data
since Lake Michigan dissolved and suspended solids are about 2%
Na by weight (Torrey, 1976) whereas the cement manufacturing source
aerosols are known to be less than 0.001% Na by weight (Kowalcyzk
and Gordon, 1979). Because of high blank levels, Na concentrations
could not be determined for 5 of 23 filter sets' data aggregated to give
Tables III and IV results. The remaining 18 may, however, be considered
in discriminating between the lake and cement manufacturing sources.
Seventeen of these 18 sets were obtained before December so that the
road-salting Na source may be discounted. Of course, Na in soil, fuel
oil, and coal must also be considered. Then there is the ubiquitous
sea salt source. The purpose of the CEB approach is to account for
these multiple source contributions for each of the elements monitored.
-------
116
Seven source types were here considered: the lake source (hereafter,
LAKE); a soil source (SOIL); a cement manufacturing source (CEMENT); an
iron and steel manufacturing source (STEEL); a coal burning source
(COAL); a fuel oil burning source (OIL); and an automobile source
(AUTO). Table V lists the elemental composition--including uncertainties--
for each of these seven source types. Notice that Ti was included.
(For all 18 filter sets being considered, Ti concentrations were
significantly above their procedural blank values.) Since Ti may help
in distinguishing between SOIL and COAL it was included in the CEB
analysis. Ni, Mo, and Sn were not included since there were too many
instances for which their fine or coarse mode concentrations were not
sufficiently above procedural blanks. (These three elements contribute
< 0.01% [or an unknown amount] to all seven of the aerosol sources in
Table V.)
The seven source types were chosen on the basis that they probably
contribute the large majority of mass for the 12 elements in Table V.
except Na and possibly Zn (Gatz, 1975). The elemental composition of
each source (its "fingerprint") was determined by combining Chicago-
region data with published statements of relative mass contribution--
primarily those of Gatz, 1975; Kowalcyzk and Gordon, 1979; Ondov et al, ,
1979.
Detailed information on dolomite limestone used to make cement
(Lamar and Thomas, 1956) shows that Chicago-region limestone quarries
contain, on average, only about two and one-half times more Ca than Mg,
whereas, finished cement contains 35 to 40 times more Ca than Mg. The
emission source for cement manufacturing is almost entirely caused by
mining in the quarries; thus, the use of Lamar and Thomas' quarry analyses
instead of finished cement analyses to specify the CEMENT elemental
-------
117
source composition. Notice the large uncertainty in weight percent for
both CEMENT Ca and Mg. This is due to the wide variation in their
content from quarry to quarry. LAKE was determined using Torrey's
(1976) review of Lake Michigan chemical data. Total dissolved solid
and total suspended particulate contributions were summed to arrive at
each element's LAKE contribution. LAKE assumes no enrichment of elements
during aerosol emission at air/water interface.
A comparison of LAKE and CEMENT shows that the ratio of Ca to Mg as
well as the weight percent for Fe and Na may afford the separation of
LAKE contributions from those of CEMENT. No previous literature source
composition for LAKE is available for this comparison, but most literature
cement manufacturing compositions (Gatz, 1975; Winchester and Nifong, 1971)
assumed emissions "to have the same composition as the finished product"
and, thus, an inappropriate low weight percent of Mg. There is also about
65;J and about 75" more Fe and Al , respectively, in CEMENT than was indicated
by Gatz (1975). The Mn content is about the same as was assumed by
Kowalczk. and Gordon (1979) although, as in the case of Mg, the Mn content
in quarries is quite variable or unknown. CEMENT, as specified in
Table V, actually causes more difficulty in distinguishing it from LAKE
than if literature cement source compositions had been used.
Distinguishing SOIL from COAL has also been shown to be quite
difficult. In fact, Cooper and Watson (1980) point out that soil,
road dust, rock crusher, asphalt production and coal source contributions
"cannot readily be distinguished on the basis of their elemental finger-
prints and are usually grouped into a common source category." This is
especially true in the present context since the elemental concentrations
of Br, K and S were not determined. In setting out to distinguish SOIL
-------
118
and COAL it was felt that COAL would be a major contributor to the fine
fraction aerosol but not to the coarse fraction (Ondov et al., 1979)
whereas road dust, rock crusher and asphalt production sources would
fractionate about the same as SOIL by aerosol size. SOIL may therefore
be seen as representing some anthropogenic soil-like aerosol sources
as well as the natural soil source itself.
The midwestern soil composition of Bowen (1966) provided the elemental
fingerprint for SOIL in Table V. This is essentially the same soil
"fingerprint" used by Gatz (1975) in previous Chicago-region CEB
interpretive analysis. The coal source composition was specified using
Ondov et al. (1979) as the primary data source. Commonwealth Edison
(1981) was contacted to confirm that cold-side electrostatic precipitators
are used (almost exclusively) as pollution control devices on their coal-
fired power plants and that western coal is by far the primary coal type
burned. Thus, the data of Ondov et al. (1979) is especially appropriate.
However, it was feund (Commonwealth Edison, 1981) that the concentration
of Cd and Zn are approximately 30 times and 3 times higher, respectively,
and V approximately 3 times lower in the strains of locally burned
coal than in that considered by Ondov et al. (1979). COAL in Table V
directly reflects this in its distinction from the Ondov et al. (1979j<
coal "fingerprint." A comparison of COAL (Table V) with a previously
identified Chicago-region coal source (Gatz, 1975) shows some differences.
The ratio of Gatz's values to COAL are (in parens): Al (1.0), Ca (1.3),
Cd (2.5), Fe (2.1), Mg (0.7), Mn (0.6), Pb (12), V (2.7), Zn (2.2),
Na (0.2) and Ti (1.1). These differences are primarily the result of
a more careful consideration in COAL of the coal type and pollution
control devices used by coal-fired power plants in the Chicago-region.
-------
119
Notice again, though, the large uncertainties specified,especially for
Cd, Pb and V, due to il1-knowledge of certain elemental weight
contri butions.
STEEL in Table V has nearly the same source composition as was
earlier specified by Gatz (1975) and Winchester and Nifong (1971)
except that recent emission measurements from a number of steel
manufacturing plants (Lake-Porter Air Pollution Task Force, 1981)
showed one-third lower concentrations of Ca, and very slightly lower Cu,
Fe and Mg concentrations.
OIL in Table V is almost identical to the oil source composition
of Kowalcyzk and Gordon (1979) since actual oil-burning emissions
were analyzed for Al, Ca, Cu, Fe, Pb and Zn as well as V and Na.
Finally, AUTO is also nearly the same as the auto source composition
of Kowalcyzk and Gordon (1979) except that the weight percent of Pb
was reduced from 40% to 25% given the changing mix of leaded/unleaded
automobile fuel use (Illinois Chamber of Commerce, 1981).
The Table V source types and the 18 crib filter sets for which
all 12 elemental mass concentrations are available were used to perform
CEB analysis on a set-by-set basis for both the fine fraction and
coarse fraction elemental masses. It was first confirmed that each
element's mass concentration could be fully explained by a six-source
linear combination (excluding either CEMENT or LAKE), except Na and Zn
concentrations. The Table VI results show the mean ratio of CEB
calculated to observed concentrations for Ca, Fe, Mg, Mn, as well as
for Na and Zn. The mean percent Na explained (across the 18 sets
considered) is 40% for the fine fraction mass and 88% in the coarse
fraction mass when LAKE is considered in linear combination with the
last five sources of Table V. Invoking a sea salt mass concentration,
-------
120
_ 3
in no case greater than 0.25 yg m , completely explains the residual
Na mass with the concomitant introduction, by the sea source, of no
more than a 3% enhancement of Ca and Mg. The explained Zn may be low
due to neglect of an incineration source because it could not be well
specified. There is also some possibility that one of the anthropogenic
soil-like aerosol sources (such as road dust) which is enriched in Zn
may explain some of the residual Zn. However, it will be seen below
that SOIL, representing these anthropogenic soil-like aerosol sources (partly)
as well as the natural soil source, was a relatively small contributor
to the total aerosol mass. The otherwise exceptional fit of the six-
source linear combination (using either LAKE or CEMENT) to the elemental
mass data base for both fine and coarse fractions warrants the presentation
of the overall CEB results.
CEB Source Reconciliation
The mean ratio of calculated CEB to actual observed elemental concen-
trations was very nearly 1.0 when taking a six-source combination, using
either LAKE or CEMENT separately, as Table VI shows. However, uncer-
tainty in these Table VI ratios were substantially greater when CEMENT
was used as opposed to LAKE. In the case of Mn, an overall uncertainty
of 20% results from use of LAKE versus 37% from use of CEMENT; for Fe_ ^
these uncertainties are 14% and 29%, respectively. Uncertainty for Mg
was 18% and 110% respectively for LAKE and CEMENT. Clearly, LAKE better
explains the set-by-set crib-site data than does CEMENT, when one or the
other is used in combination with the remaining five sources of Table V.
We are also interested in the relative contributions of the seven
sources to the overall elemental concentrations at the crib site. A
linear combination of all seven sources was used to obtain the CEB results
in Table VII for both FINE (fine fraction) and COARSE (coarse fraction)
-------
121
mass loadings--relati ve to a soil source of unit (1.0) na^nitude.
Since too few filter sets had an associated total aerosol mass, the results
must be presented as relative magnitudes; that is, CEB analysis was
performed instead of chemical mass balance (CMB) analysis. The range
in relative source contributions is quite large (except for COAL).
Therefore, only approximate mean ratios are stated in Table VII, again
excepting COAL. The Table V source contributions of COAL and SOIL are
quite similar, so the mean ratio of FINE COAL to SOIL is a rather constant
0.55 + 0.06. As an example of confidence in Table V source contri-
butions the mean uncertainties for FINE SOIL and FINE COAL are 32% and
34%, respectively. These small uncertainties justify strong confidence
in the value 0.55, considering that uncertainty in the ICAP elemental
analyses is 20% to 30% and that source elemental compositions (Table V)
are even more uncertain.
Of greatest interest is the relative contribution of LAKE. As shown
in Table VII, LAKE has a mean ratio of ~2, relative to CEMENT as well as
to SOIL. This suggests that CEMENT and SOIL together just equal, on a
total aerosol mass basis, LAKE. It is, however, important to note that
the percent mass composition of Al , and even of Fe, is lower at the crib
site than it was the previous year at midlake. The 0.4% mass for Al
during the 1978 crib sampling is, in fact, one-fifth that in an elemental
composition model for Chicago aerosol suggested by Gatz (1975). Thus, the
SOIL contribution was anomalously low during crib sampling. The range
in the ratio of LAKE to SOIL observed across the 18 filter sets is very
wide, being anywhere from 0.5 to 4.8 for FINE and 0.3 to 4.6 for COARSE.
Most of this variability is probably due to that in LAKE contribution
to crib site aerosol mass and must be considered real since the total
mean percent uncertainty for FINE LAKE is 55% and for COARSE LAKE 61%.
-------
122
(These total uncertainties are the sum of the SOIL uncertainty and LAKE
uncertainty since we are here considering a ratio of the two. The
total mean percent uncertainty for CEMENT is 99% for FINE and 101%
for COARSE.) The equally strong contribution of LAKE to fine and coarse
aerosols over Lake Michigan (double that of- SOIL) is surprising, espe-
cially considering the qualitative observation early in this discussion
that fine/coarse fractions for the largely lake-derived Ca and Mg were
reduced at the midlake site compared with the crib site. However, the
fact that LAKE appears to be such a variable source suggests that the
lake may be a strong local aerosol source. This latter point is in
concert with the second qualitative observation that filter sets for
no-lake-breeze conditions had significantly larger Ca and Mg EF's than
the lake-breeze group of filter sets. The short overlake fetch of 3 to
at most 10 km for the no-lake-breeze sets was apparently still sufficient
to allow the lake to contribute Ca and Mg to the aerosol observed at the
crib site so long as at least moderate windspeeds prevailed.
Separation into 8 lake-breeze and 10 no-lake-breeze filter sets results
in the LAKE/SOIL ratios shown in Table VIII. The FINE LAKE mass contribu-
tion during lake-breeze occurrences is significantly below the overall
ratio of two, whereas the ratio for FINE LAKE with no-lake-breeze (and.
short over-water fetch) is higher than this for most of the 10 sets in
this group. There is no distinguishable difference for COARSE LAKE. The
fact that windspeeds were 3.4 + 1.5 m s~ during lake-breeze sets and
6.0 + 1.0 m s"1 during the short-fetch no-lake-breeze sets indicates that
a local lake source can definitely contribute to crib site aerosol mass;
Table VIII indicates a greater lake contribution in the fine fraction. The
Table VIII data suggest that lake-derived fine (d < 1 ym) aerosol may be much
more dependent upon mean windspeed, and that lake-derived coarse (d > 1 ym)
aerosol may be omnipresent.
-------
123
CONCLUSION
CEB analysis was successfully applied to the 68th Street crib
site data base. Quantitative identification of a strong lake source
(possibly local) aerosol supports qualitative indicators, based on Ca
and Mg enrichment factors and fine/coarse aerosol fractions, that a
lake source is significantly contributing to aerosol mass over Lake
Michigan. This is true not only for coarse (diameter, d > 1 pm) aerosol
but also for fine (d < 1 ym) aerosol. The lake source contribution
to the fine aerosol mass is about equal to the sum of anthropogenic
contributions (iron/steel manufacturing, oil burning and automobile
exhaust). The strength of the fine aerosol lake source appears to be
dependent upon mean windspeed whereas the coarse aerosol lake source
appears to be much less so. The coarse aerosol lake source is very
probably due to the jet drop generated during wave breaking whereas we
may speculate that film drops are responsible for the fine aerosol lake
source.
The lake source was found to contribute twice as much mass as did
soil. This was so for both the fine and coarse fractions. However,
the soil source appeared to contribute one-half to one-fifth less mass
during the 1978 crib site field sampling effort than may be considered
typical. If a four-times larger soil source contribution had been present,
the lake source would have been found to contribute about half as much
as the more typical soil source. Nonetheless, the lake as a source of
aerosol mass must be carefully considered whenever aerosol data obtained
over Lake Michigan, or any other of the Great Lakes, are being interpreted.
-------
124
ACKNOWLEDGMENT
The authors wish to thank J. Forst, K. Walther, and W. Ward
for tireless help with the field program; V. Jensen for his analytical
assistance; Dixie Butz and Vera Rhimes for preparing the manuscript.
We also wish to extend our appreciation to Tom Tisue of Argonne
National Lab for the XRF analysis of our filter portions, and J. Schmidt
for review of our manuscript draft. This work was supported by USEPA,
Grant #R005301012.
-------
125
REFERENCES
Bowen, H.J.M., 1966. Trace Elements in Biochemistry, Academic Press,
New York.
Commonwealth Edison, 1981. Private communication.
Cooper. A. and Watson, J.G., 1980. Receptor oriented methods of particle
source apportionment. J. Air Poll. Cont. Assoc., 30, 1116-1122.
Dolske, D.A. and Sievering, H., 1979. Trace element loading of Southern
Lake Michigan by dry deposition of atmospheric aerosol. Water, Air,
Soil and Poll., 12, 485-502.
Dolske, D.A. and Sievering, H., 1980. Nutrient loading of Southern Lake
Michigan by dry deposition of atmospheric aerosol. J. Great Lakes Res.,
6, 184-194.
Eisenreich, S.J., Emmling, P.J. and Beeton, A.M., 1977. Atmospheric
loading of phosphorus and other chemicals to Lake Michigan. CL_
Great Lakes Res., 3, 291-304.
Fingleton, D.J. and Robbins, J.A., 1980. Trace elements in air over Lake
Michigan near Chicago during September, 1973. J. Great Lakes Res., 6,
22-37.
Gatz, D.F., 1975. Relative contributions of different souces of urban
aerosols: application of a new estimation method to multiple sites
in Chicago. Atmos. Env., 9, 1-18.
Gordon, G.E., 1980. Receptor models. Env. Sci. & Tech., 14, 792-799.
Hughes, R.L. and Sievering, H., 1982. Lake breezes on particle size
spectra and sulfate over Lake Michigan, submitted to Jour, of Appl.
Meteor.
Illinois Chamber of Commerce, 1981. Private communication.
Lamar, J.E. and Thomson, K.B., 1956. A description of dolomite limestone
by chemical analysis in the County of Cook, 111. Geo. Survey Circ.
#221.
Kowalcyzk, G.S. and Gordon, G.E., 1979. Source identification of trace
elements and total suspended particulate material in Washington, DC,
presented at the San Francisco A.I.Ch.E. meeting, Nov., 1979 (unpublished),
Lake-Porter Air Pollution Task Force, 1981. Private communication.
Lyons, W.A. and Olsson, L.E., 1973. Detailed mesometeorological studies
of air pollution dispersion in the Chicago Lake breeze. Month.
Weather Rev., 101, 387-403.
-------
126
Mezaros, A., 1977. On the size distribution of atmospheric aerosol
particles of different composition. Atmos. Env., 11_, 1075-1081.
Murphy, T.J. and Doskey, P.V., 1977. Inputs of phosphorus from
precipitation to Lake Michigan. J. Great Lakes Res., 3, 305-312.
Nagib, H., 1978. Private communication.
NOAA, 1975. Summary of Synoptic Meteorological Observations for Great
Lakes area., vol. 3, Lake Michigan. Nat'l Climatic Center,
Asheville, N.C.
Ondov, J.M., Ragaini, R.C. and Biermann, A.M., 1979. Elemental emission
from a coal-fired power plant: Comparison of a venturi wet scrubber
system with a cold-side electrostatic precipitator. Env. Sci. &
Tech., 13. 362-371.
Rahn, K., 1976. The Chemical Composition of Atmospheric Aerosol. Grad.
School or Oceanogr., U. Rhode Island, Kingston, R.I.., 265 pp.
Sievering, H., Dave, M., Dolske, D.A., Hughes, R.L., and McCoy, P., 1979.
An Experimental Study of Lake Loading by Aerosol Transport and
Deposition in the Southern Lake Michigan Basin. EPA-905/4-79-016.
Torrey, M.S., 1976. Environmental Status of the Lake Michigan Region,
vol 3, Chemistry of Lake Michigan. ANL/ES-040, Argonne National
Lab, Argonne, 111.
Watson, J.G., 1979. Chemical elemental balance receptor model methodology
for assessing the sources of fine and total suspended particulate
matter in Portland, Oregon. Ph.D. Dissertation, Oregon Graduate
Center, Beaverton, Oregon.
Williams, R.M. and Muhlbaier, J., 1978. Preliminary findings of wind-
direction controlled aerosol sampling over Lake Michigan. In Argenne
National Lab, RER Division Annual Report: Ecology, Jan-Dec 1978.
ANL-78-65-I11.
Winchester, J.W. and Nifong, G.D., 1971. Water pollution in Lake Michigan
from pollution aerosol fallout. Water, Air and Soil Poll., 1, 50-64.
-------
TABLE I
Element
Al
Ca
Cd
Cu
Fe
Mg
Mn
Mo
Ni
Pb
Sn
V
Zn
Detection
1 inri t
90
5000
2
9
120
100
6
6
5
45
8
2
50
Procedural
blank
290
5100
6
40
240
210
20
30
40
80
10
4
100
Typical
sample
2600
16400
120
230
6300
4900
500
75
75
4100
260
90
1800
Detection limits, procedural blanks, and typical sample concentrations
for the ICAP method (yg £-1)
-------
128
TABLE II
Element
Fe
Mn
Ni
Pb
Zn
Mean Concentration
of ICAP and XRF
(ng m~3)
785
32
27
500
490
Mean Difference
ICAP vs. XRF
(%)
33
66
70
19
18
Results of ICAP and XRF analyses of 12 filter sets
-------
Element
A]
Ca
Cd
Cu
Fe
Mg
Mn
Mo
Ni
Pb
Sn
V
Zn
P
N (N03)
s (so4)
Mass
129
TABLE III
C nearshore,
1978
(ng m~^)
215
1340
10
20
520
405
40
6
6
340
20
7
150
20
3670
5660
57,600
C midlake,
1977
(ng m~3)
180
770
n.d.
6
320
200
20
n.d.
n.d.
140
10
n.d.
55
35
3100
5100
31,800
Geometric mean concentrations of elements in aerosol at the 68th Street
crib and midlake.
-------
Element
Al
Ca
Cd
Cu
Fe
Mg
Mn
Mo
Ni
Pb
Sn
V
Zn
P
N (NO,)
O
s (so4)
EF
1.0
22
4300
190
4.6
26
11
580
230
5200
690
100
580
130
TABLE IV
Nearshore
Fine/Coarse % Mass
Fraction Composition
1.7 0.4
1.2 2.3
15.6 <0.1
5.5 <0.1
3.3 0.9
2.2 0.7
3.1 <0.1
5.8 <0.1
7.2 <0.1
12.7 0.6
4.7 <0.1
7.4 <0.1
4.4 0.3
7.6 <0.1
9.1 6.5
30.8, 9.8
Midlake
Fine/Coarse
EF Fraction
1.0 1.2
22 0.4
170 2.9
3.5 1.6
23 0.6
10 2.0
5200 17.3
530 6.1
4.2
8.0
37.3
% Mass
Composition
0.6
2.4
<0.1
1.0
0.6
<0.1
0.4
0.2
0.1
3.1
5.6
Comparison of parameterized characterizations of nearshore and midlake aerosols.
-------
131
TABLE V
Percent by Weight
Element
Al
Ca
Cd
Cu
Fe
Mg
Mn
Pb
V
Zn
Na
Ti
LAKE
0.03+0.02
22+5
<0.01
0.02+0.02
0.03+0.03
7.5+1.5
0.01+0.01
<0.01
<0.01
0.02+0.02
2.1+0.4
<0.01
CEMENT
3.2±2.6
32+12
<0.01
<0.01
3.2+2.3
12+11
0.1+0.1
<0.01
<0.01
0.03+0.02
<0.01
<0.01
SOIL
6.5±1.5
1.2+.0.3
<0.01
<0.01
3.7+0.4
0.5+0.2
0.08+0.04
<0.01
0.01+0.01
<0.01
0.6+0.3
0.4+0.1
COAL
14±3
3.1+1.1
0. 01+.0. 01
<0.01
3.4+0.6
1.1+0.4
0.04+0.02
0.01+0.01
0.03+0.02
0.04+0.01
1.8+0.5
0.8+0.2
STEEL
2.3+0.7
3.6+2.0
<0.01
1.4+0.6
36+6
1.4+0.6
2.5+0.5
<0.01
<0.01
1.8+0.6
<0.01
<0.01
OIL
0.210.1
2.9+0.3
<0.01
0.09+0.06
0.9+0.7
0.15+0.05
0.02+0.01
<0.01
3.0+1.0
0.2+0.1
1.6+0.8
<0.01
1
AUTO
0.15+0.09
0.05+0.03
<0.01
<0.01
1.8+1.2
<0.01
<0.01
25+5
<0.01
0.07+0.05
<0.01
<0.01
Aerosol composition for various Chicago-region source types.
-------
132
TABLE VI
Mean Ratio (Calculated to Observed)
Using LAKE | | Using CEMENT
Element
Ca
Fe
Mg
Mn
Na
Zn
Fine Fraction
1 .24
1 .0
Q.85
0.93
0.40
0.24
Coarse Fraction
0.86
1 .01
0.78
0.95
0.88
0.37
Fine Fraction
O.P8
1 .04
0.68
0.98
0.26
0.25
Coarse Fraction
0.87
1 .04
0.82
0.94
0.39
0.38
Ratio of CEB calculated (six-source linear combination) to observed elemental
concentrations.
-------
133
TABLE VII
Relative Weight (SOIL = 1.0)
Aerosol
Fraction LAKE CEMENT
FINE
(d < 1 pm)
Mean Ratio ^2 M
Range 0.5-4.8 0.2-2.8
COARSE
(d > 1 ym)
Mean Ratio ^2 >1
Range 0.3-4.6 0.2-3.3
STEEL OIL AUTO COAL
3/4 1/4 ^1 0.55
0.3-1.5 0.1-0.5 0.2-2.5 0.4-0.7
1/3 0 1/10 0
0.1-0.6 0-0.2
Relative weighting of six source types for fine and coarse aerosol fractions
at the 68th Street crib site.
-------
134
TABLE VIII
Relative weight (SOIL = 1 .0;
Aerosol
Fraction
FINE
(d < 1 ym)
COARSE
(d > 1 ym)
Mean windspeed 3
X. and fetch LAKE
\. (long
.4 ± 1.5 ms'1
with lake breeze
over-water fetch)
1.0 + 0.5
2.4 + 1.0
6.0 i 1.0 ms-"1
LAKE with no lake breeze
(short over-water fetch)
3.7 + 2.2
2.2 + 2.1
Relative weighting of the lake source for fine and coarse aerosol fractions
and as a function of fetch over the lake.
-------
Fi
gure 1. Plan of City of Chicago 68th Street crib,
showing permissible wind direction sector
limits
CO
en
-------
LAKE MICHIGAN
68th Street Crib
Figure 2. Plot of back-trajectory calculation for two sample periods.
a) Direct shore-to-crib trajectory.
-------
137
LAKE MICHIGAN
68th Street Crib
Chicago
o
o
I
oo
00
Kilometers
10 20 25
O
O
o
N-
co
41°30'
o
CO
o
CD
co
052678 1600Z
Figure 2 b) Trajectory influenced by lake breeze event.
-------
138
APPENDIX E
SOME EFFECTS OF WIND-SHIFT ON
OVER-LAKE TURBULENCE
AND AEROSOL DEPOSITION
Richard L. Hughes and Herman Sieverlng
Environmental Science Program
College of Applied Sciences
Governors State University
Park Forest South, IL 60466
-------
139
Abstract
Aircraft measurements of turbulence spectra and surface measurements
of total aerosol mass derived from size distributions were compared for
periods with and without windshift over the southern basin of Lake Michigan
in September of 1977 and May of 1978. It was found that wind shifts are
associated with depletion of near-surface aerosol mass and with departures
from isotropy in the inertia! subrange as indicated by difference of the
ratio of vertical velocity spectra to longitudinal velocity spectra from the
isotropic value of 4/3. Velocity spectra were measured by fixed and rotating
vanes mounted on the aircraft nose boom. Aerosol size distributions were
measured by a laser scattering particle counter.
-------
140
I. Introduction
Dry deposition of atmospheric aerosols to water bodies has, within the
last decade, come to be recognized as an important and, for some substances,
predominant route for the influx of matter. Winchester and Nifong (1971)
calculated atmospheric loading rates for the region of the southern basin of
Lake Michigan. This model has been variously modified by Skibin (1973),
Sievering (1976) and others. These studies and those by Gatz (1975a and
1975b) not only found increased estimated loadings to the lake by the
atmospheric route, but pointed out large uncertainties in these estimates,
stemming from a lack of understanding of the boundary layer dynamics of
aerosols and lack of in situ measurements of dry deposition over the lake.
The most direct method of measuring aerosol flux, the eddy correlation
method, has proved to be unwieldy in the field and especially on a lake
surface (Williams et al., 1980 ). There has been enough evidence, however,
to note a large discrepancy between measurements done in a wind tunnel
(Sehmel and Sutter, 1974) and those taker under field conditions. Construction
of a model capable of accurately predicting deposition of aerosols to a
water surface requires consideration of all phenomena contributing to this
discrepancy. Several explanations have been proposed in the literature
(see review article by SI inn et al., 1978). One possible phenomenon discussed
by Slinn (1979) is the rapid uptake of water vapor by aerosols in the high
humidity environment near the surface and subsequent gravitational fallout.
Another mechanism suggested by Sievering (1980) involves direct
impaction of particles on those elements of the surface which protrude through
the laminar sub-layer. When a continuous laminar sub-layer is present, one
may expect more than an order of magnitude variation in aerosol deposition for
-------
141 2
the 0.1 < d < 2 pm size range. It was found experimentally (Sievering, 1980)
that a two-to three-fold variation prevailed in this size range under conditions
of moderate wind speed (2.4 to 8.2 m sec"1). If these experimental data are
truly representative of air/water aerosol exchange for moderate wind-speed
regimes, then non-steady state conditions must predominate above the interface.
That is, turbulence-induced eddies may advect aerosols to surface roughness
elements protruding outside a possibly discontinuous sublayer resulting in
direct impaction.
One of the ways in which deposition to a water surface differs from
deposition to land or vegetation is that a water body adapts its surface
characteristics to the wind. There is, particularly over large bodies of
water, a lag between a change in the wind regime and the adaptation of the
wave train (SethuRaman, 1978). During this period of maladjustment, elements
of the water surface are most prone to protrusion .through the laminar sub-
layer. A change in wind speed or direction perturbs the equilibrium
between a water surface and a steady-state wind, and would be expected to
also perturb the state of isotropy which is presumed to occur within the
inertia! subrange of the momentum power spectrum in the absence of unusual
sources or sinks of kinetic energy. One necessary, but not sufficient,
indicator of the presence of an isotropic state is the ratio of the vertical
velocity (or lateral velocity) spectrum to the longitudinal velocity spectrum.
In the inertial subrange, this ratio is thought to approach the value 4/3 under
isotropic conditions. Departures from this value in over-water environments
relative to over-land environments have been previously noted using a variety
of measurement techniques (Schmitt, Friehe and Gibson, 1978),
-------
142
3.
2. Instrumentation
Turbulence measurements were taken aboard a Beechcraft £"?eenaire
aircraft provided by the Research Aviation Facility of the National Center
for Atmospheric Research. The aircraft departed from Midway airport and upon
reaching the lake proceeded at a low altitude (usually 30 m) to the midlake
sampling site 87°00'W, 42°00'N (see fig. I). Vertical soundings were taken
in a square spiral "box" pattern over this site, over a point halfway to
shore 87°15'W, 41°55'N (referred to as the "halfway point") and over a nearshore
sampling site, a City of Chicago water intake crib 87°32 'W, 41°37'N. Soundings
extended from about 15 m through about 1800m. Flight times were approximately
0700-0900, 1100-1400 and occasionally 1600-1800 Central Daylight Time (CDT)
during September 26-30 of 1977 and May 17-19, 22, and 24-26 of 1978.
Constrained and rotating vanes mounted on a nose boom of the aircraft
(Lenschow, 1971) were used to measure velocity. Also en the aircraft were
fast-response pressure, temperature and humidity sensors. These data were
recorded at a rate of 10 hz by an on-board data system (Glaser, 1973).
Concurrently with the aircraft data, meteorological and aerosol data
were gathered at the surface. During September of 1977, the data were
collected aboard the R/V -Simons at tne midlake sampling site
During May 1978, the surface sampling site was the water intake crib
Among other analyses of the surface data, trajectories
of the air parcels sampled were calculated backwards in time. The sampling site •
wind vector was used for this trajectory as long as the parcel had not
penetrated the surface layer which was assumed to be at a height of 50 m.
The vertical velocity of the parcel was estimated by a momentum-analogue
deposition velocity calculated from the windspeed at 5m and the temperature
stability. After penetration of the surface layer, the wind vector was
-------
143
4.
calculated from a weighted average of the surface data and a triangulation
average of National Weater Service upper level soundings. The process is more
fully described in Sievering et al. (1979). The trajectories, though they had
large uncertainties, were sufficiently accurate to provide an indication of
the source regions of the air sampled and a" means of calculating the degree
of windshift.
An Active Scattering Aerosol Spectrometer (ASAS), which counted particles
in each of 60 size channels in the range 0.23 < d < 30.0 ym for September
data collection and 0.11 < d < 3.5 ym in May by scattering of a He-Ne laser
source (see Schuster and Knollenberg, 1972), was mounted on board the aircraft.
During May 1978 a second ASAS was operated at the lake surface. ASAS data
obtained during 1977 was collected at the midlake site. Several instrument-
related problems make the validity of absolute numbers suspect, but comparisons
between periods of ASAS operation should be realistic. Integration of the
o
aerosol size spectra multiplied by 4/3 )f r and an assumed particle density of
2 g cm"3 yields a measurement of total aerosol mass per volume of air sampled
assuming sphericity of particles and uniform density. An integrating
nephelomter (IN) allowed independent verification of mass measurements (see
Dave, Dolske, and Sievering, 1979).
3. Analysis
The 10 hz values of the three velocity components were analyzed by a
fast-Fourier transform method and the resultant spectra were plotted. In
addition, the ratio of vertical to longitudinal velocity (SW/SU) was cal-
culated in the wavelength range 10 < X < 300m which was presumed to lie
within the inertial subrange.
-------
144
5.
In an attempt to find a relationship between the degree of variability
in wind direction and the departure from isotropy, a "windshiftiness"
parameter was developed. This parameter estimates the persistence of any
one wind direction as the ratio of path length to displacement or s/£, where
s is the trajectory path length and i is the straight-line displacement of
a parcel from start to end of trajectory. This parameter is unity for perfectly
straight flow and unbounded above for convoluted trajectories'. A total of
eleven flights occurred concurrently with surface sampling during September
of 1977 and May of 1978. Of these, two could not be used because of faulty
data (see Table 1). When a rank correlation was done between the "windshiftiness'
parameter and 4/3 - SWSU I , the rank correlation coefficient was 0.925
(see Table 2) allowing one to reject the hypothesis of zero correlation at
better than the one percent level.
Comparison of aerosol mass measurements with wind direction shifts is
another method of investigating the relationship between windshift and near-
surface effects. It elucidates turbulence behavior less directly, but still
involves a direct measurement of atmospheric aerosols. ~ A case study of
midlake sampling site data for the period from 18 May 1977 to 19 May 1977 is
particularly relevant. Table 3 lists six periods over which data were collected,
and certain associated meteorological and aerosol measurements. In this
table, stability (AT) is the air temperature measured at a height of 5 m minus
the surface temperature (measured by infrared thermometry). The wind speed
and wind direction were also measured at 5m. All values are averages of
measurements taken every fifteen mini/tes across the period indicated. The
mass, calculated by integration of the aerosol size spectra, allows one to
discriminate between mass due to fine particles(0.11 < d < 1 urn) and an
estimated total measured was by the IN (approximately 0.1 < d < 1-0)
-------
145
6. -
or the ASAS (0.11 < d < 3.5 ym).
An abrupt windshift from 240° to 170° took place between 1230 and 1300
CDT on 18 May 1977, i.e., near the start of data set 20080. Before this period the
southern basin of Lake Michigan had been under west-south-westerly winds for
over 10 hours. As a result the midlake area and, presumably^ points south
and southwest were under the influence of the Chicago plume, as evidenced by
the high mass measurements for data sets 20060 and 20070. After the windshift,
it should have taken about 4^ hours for the slightly less polluted south-shore
plume, traveling at the mean wind velocity of 1.9 m sec~l, to reach the sampling
site. Thus, only the last 20% of data set 20080 should have been sampling
south shore air. Yet, in spite of continued strong stability and light wind
speeds, there is a sixfold drop in aerosol mass. The large standard deviation
in the wind direction for data sets 20080 and 20090 resulted from the fact
that during this period, the wind direction swung from 180° to 100° and back
again.
There is another abrupt windshift and, perhaps more important, a
severe decrease in the standard deviation on wind direction, between 2100 and
2200 on 18 May 1977. A parcel which left shore at this time would reach the
sampling site just about the time the heavily-loaded data set 20110 was begun.
In summary, the indication is that the period of continuous windshift is
associated with a severe decrease in total aerosol mass near the surface.
4. Summary and Conclusions
The ratio of vertical to longitudinal velocity in the inertia! subrange
was measured by aircraft over Lake Michigan. When near surface values of this
ratio were compared to an estimator of the degree of windshift, a correlation
was found suggesting that "windshiftiness" over Lake Michigan is associated
with departures from isotropy, as would be expected if increased wind shifts
-------
146
7.
induced protrusions of roughness elements through the laminar sub-layer.
A case study was presented in which wind shifts are followed by a
decrease in aerosol mass concentration, consistent with the hypothesis that
increases in the protrusion of roughness elements will promote direct impaction
of aerosols on the water surface.
-------
147
' 8.
6. Acknowledgements
Appreciation is extended to the personnel of the Research Aviation
Facility (especially P. Spyers-Duran) and the Computing Facility of the National
Center for Atmospheric Research. The plotting of trajectories was performed by
Vic Jensen and Patric McCoy. Gratitude is expressed to Brenda Chapman for
preparation of the manuscript.
This work was supported by the U.S. Environmental Protection Agency
under contracts R00530101 and R00542101.
-------
148
References
Dave, Mehul, Donald Dolske, and Herman Sievering, Short Communication.
"Atmospheric Aerosol Mass Concentration and Scattering Coefficient
in a Midlake Region." Atmos. Environ., J.3, 1979, 1597-1600.
Gatz, D.F. 1975a: Relative Contributions of Different Sources of Urban
Aerosols: "Application of a New Estimation Method to Multiple Sites in
Chicago." Atmos. Environ., 9_, 1-18.
Gatz, D.F., 1975b: "Pollutant Aerosol Deposition into Southern Lake Michigan."
Water Air and Soil Poll., 5, 239-251.
Glaser, Walter, 1973: "The Aris III Data System, Atmospheric Technology,
March, 61-66.
Lenschow, D.H., 1971: "Vanes for Sensing Incidence Angles of the Air from an
Aircraft. J. Appl. Meteor.. 1Q, 1339-1343.
Schmitt, K.F., C.A. Friehe, ane C.H. Gibson, 1978: Sea Surface Stress Measure-
ments. Boundry-Layer Meteor., 15, 215-228.
Schuster, B.G. and R. Knollenberg, 1972: Detection and Sizing of Small Particles
in an Open Cavity Laser. Appl. Opt., H., 1515-1520.
Sehmel, G.A. and S.L. Sutter, 1974: Pacific N.W. Lab Annual Rep, for 1971,
Vol III, pt 1, Richland, WA. A Batelle Northwest Laboratories.
SethuRaman, S., 1978: Influence of Mean Wind Direction on Sea Surface Wave.
Development. J. Phys. Ocean, _8, 926-929.
Sievering, Herman, 1976: Dry Deposition Loading of Lake Michigan by Airborne
Particulate Matter, Water Air and Soil Pollution, _5, 309-318.
Sievering, Herman, Mehul Dave, Donald A. Dolske, Richard L. Hughes and
Patric McCoy, 1979: An Experimental Study of Lake Loading by Aerosol
Transport and Dry Deposition in the Southern Lake Michigan Basin,
Chicago, II, U.S. Environmental Protection AGency publication No.
EPA-905/4-79-016.
Sievering, Herman, 1980: Profile Measurements of Particle Mass Transfer at
the Air/Water Interface [Accepted for publication in Atmospheric
Environment].
Skibin, D. 1973: Comment on Water Pollution in Lake Michigan from Pollution
Aerosol Fallout, Water Air and Soil Poll., 2, 405-407.
Slinn, W.G.N., L. Hosse, B.B. Hicks, A.W., Hogan, P. Lae, P.S. Liss, K.O.J.
Munnich, G.A. Sehmel and 0. Vittori, 1978: Some Aspects of the
Transfer at ATmospheric Trace Constituents Past the Air-Sea Interface,
Atmos. Env., 12, 2055-2087.
-------
149
10.
SI inn, W.G.N., 1979: Predictions for Particle Deposition on Natural Waters,
private communication.
Williams, R.N., M.L. Wesly and B.B. Hicks, 1980: Preliminary Eddy Correlation
Measurements of Momentum, Heat, and Particle Fluxes to Lake Michigan,
Radiological and Environmental Research Division Report, Argonne
National Laboratory report No. ANL-78-65 part III.
-------
150
11,
Table Captions
Various meterological parameters measured over Lake Michigan during
September 1977 and May 1978.
Rank of "windshiftiness" parameter and
Lake Michigan during September 1977 and
4/3 - Sw/Su measured over
Nay 1978.
3. Sample time meteorological parameters and aerosol data measured over
Lake Michigan during May of 1977.
-------
151
12.
TABLE 1
Date
26 Sept.
17 Sept.
27 Sept.
27 Sept.
17 May
18 May
19 May
24 May
25 May
Time
1641
0639
1109
1636
0650
1130
1145
1130
1145
(CDT)
- 1810
- 0919
- 1344
- 1823
- 1810
- 1256
- 1330
- 1420
- 1447
Average
Wind
Direction
265
319
300
295
4
69
91
25
90
Average
Wind
Speed (m sec"1)
6.0
7.4
6.5
6.1
3.5
3.0
3.5
1.4
5,7
Wind-
shiftiness
Parameter
1.0
1.03
1.01
1.01
1.07
1.11
1.004
1.08
1.42
-------
152
13.
TABLE 2
Date
27 Sept
27 Sept
27 Sept
27 Sept
17 May
18 May
19 May
24 May
25 May
Time
1641
0639
1109
1636
0650
1130
1145
1130
1145
(CDT)
- 1810
- 0919
- 1344
- 1823
- 0810
- 1256
- 1330
- 1420
- 1447
Rank of
Wind- Rank of
Shiftiness 4/3 - SW/SU
9 8
5 5
6.5 7
6.5 6
4 2.5
2 1
8 9
3 4
1 2.5
-------
eteorology
Aerosol Mass
Data Set
20060
20070
20080
20090
20100
20110
Time (CDT)
5/18 03:00-5/18 06:00
5/18 08:20-5/18 11:20
5/18 12:00-5/18 17:25
5/18 17:45-5/18 21:00
5/18 22:00-5/19 01:30
5/19 01:45-5/19 06:15
AT
5
6
9
11
13
7
.9
.7
.5
:4
.3
.2
± 1.4
± 3.6
± 3.5
± 3.7
± 2.1
± 2.9
T-
TT(5)
(m/s)
4.5 ±
3.6 ±
1.9 ±
3.8 ±
3.7 ±
2.7 ±
0.2
0.3
0.6
0.3
0.9
0.4
lu
240
239
174
140
212
212
IN-derived
Mass
± 10
± 8
± 46
± 40
± 8
+ 26
125
266
141
110
47
141
ASAS dervied Mass
< 1.0 ym
160
174
29
29
19
133
Total
169
193
31
34
23
333
Table 3. Sample Time, Meteorological Parameters, and
Aerosol Data Measured over Lake Michigan
during May of 1977
-------
154
43°00'
o
o
o
00
O
ro
o
CD
00
LAKE MICHIGAN
O
O
o
CD
00
MIDLAKE
HALFWAY POINT
0 10 20 30 40 50
Figure 1. Location of Sampling Sites.
-------
TECHNICAL REPORT DATA
(I'L ?.!, rcaJlHUnjcrit'ns o>, thrrc>ir\t bclort complctiifi
1 REPORT NO I? 3 REUP
EPA-9Q5/2-84-002 |
4 TITLE ANDSUBTiTLE 5 BEPO
An Experimental Study of Lake Loading By Aerosol J
Transport and Dry Deposition in the Lake Erie Basin S.PERF
7 AUTHORiS) 8 PERF
H. Sievering, D. A. Dolske, V. Jensen and
R. L. Huges
9 PERFORMING ORGANIZATION NAME AND ADDRESS 10 PRO
Governors State University
Univeibily Park, Illinois G04GC 11. CON
ROD
ROO.
12 SPONSORING AGENCY NAME AND ADDRESS 13. TYP
Res
U. 5. Environmental Protection Agency 14. SPO
Great Lakes National Program Office gre
536 South Clark Street, Room 958 Off
Chicaao. : 1! inois 60605
lENT'S ACCESSION NO.
RT DATE
une 1984
DRMING ORGANIZATION CODE
DRMING ORGANIZATION REPORT NO.
GRAM ELEMENT NO.
TRACT/GRANT NO.
33010!
542101
E OF REPORT AND PERIOD COVERED
aarch 979
MSORING AGENCY CODE
at Lakes National Program
ice, U.S. EPA, Region V
15 SUPPLEMENTARY NOTES
Ed Klappenbach
Project Officer
16 ABSTRACT
The purpose of this study of Lake Erie was to obtain an ove" lake data base
and to relate aerosol chemical constituent loadings. Information collected
at each scale is necessary to expand our und9rstand i nq of the complex
aerosol transport and deposition process that occur over the lake,
ft is believed these results obtained from the Great Lakes Atmospheric
Depos ition(GLAD) network, should lead to a more generalized understand ng
of aerosol transport and deposition in the Great Lakes region, as well as
a better estimation of atmospheric-route Great Lakes Pol ution loading.
T7 KE V WORDS AND DOCUMENT ANALYSIS
o DESCRIPTORS b. IDENTIFIERS'OPEN ENDE
Dry loading
Wet loading
Aerosol transport
Chemica I ana I ys is
Monitor! ng
Air po! ! utants
S ite samp I i ng
IS DiSTP.BJTiON STATEMENT IS SE C U R IT Y C L ASS (This t
Document is available to the public through
the National Technical Information Service 20 SECURITY CLASS cr/,,*,
(NT1S), Springfield, VA 22161 Unclassified
D TERMS v COS AT! Held Group
Report) 21 NO OF PAGES
168
lapij 22. PRICE
EPA Forrn 2220-1 (9-73)
------- |