00007
905R87102
'' " "'''
;' /poHege' of;lEnyjronmental !5ciences.;-::;..-', ::-~:--/jp^
S^i&ernQrS State'Ohjyersity'-'.- - , -''-ILll:'-'"' -: '"
;^^JF^st^fh/'lii:t604b6 l^'. --^p^;-;.
V_Principal*; lnVesffgat.br,'.'-H. .Sieve"7ng .'. ' Wv.V'-.V.
:';. :3i2-534-$bo;o;:£xt;;2498;.;;..;.-. ';; .v;:' T^^i
'.S:l*V':;'-^;.'J.. ".";;.': .;.**£> *."*"; .'.-; ' ^j^J^SSI^ia: .. ''. ''-.. '. - "^TJ!? I^TT^^!
;'..''" t''.- .':'..'''' ':~^ff^?^^^r^^r'-''-',''.<'' '"-' ''.- j_^2^
'\ljV]u:»n^,'^-l''" ' ""'-":''.'-'-; -j ,-->*«*f^^?irv'.:-'..;;.'".': :-'::«
t-^t^-^^t^r:.-,.. . -.''^^ " . .. .;..-':....''.'.~«'^TO*s5?r:-:.«l
:/r ..:V:v:-:-.V:-:rj*?SST8h?-.-.. . ' :- .-.-. ;.-. ../......:v''-'-l1iW!«*
K5,.?%^».-. '.-. . i«Sfc
i^ff";! i^V^1'. ".. ;.' ;:-:
'W*^*s>'''- - . , ,""
*"- .- " . ' ' ;"'-1?3.
' .''. ..'-.'' .'*.' "
^» ' '" r ..-^»**M'-t* *r. -' . . "- ' -''.'tHftmff ill i'i» ' ' -'' ".' " ' '*
- ..-.- ."'*':"-.' .- -f-****ft!l%5^*?*&?~*--i.-'-
6*5-?-f.vi-- ...-: : , ' ..>.'-'. .... ' !_
.. - --; - .;.. :...-- -.-.:.:' .--- --''
..'. " - Prepared for: . ..;.:.:;.. .. -.-v1-'/
- ' : U.S:Enyirbnmerital Protection = Agency/: :
' :':'''.-.' -.Chjcago,; III. ;. 60.604 .;/_.; ;/;_-'.; : --_":/,
'''
i '".': ' ^ :.Cphtract No.' ROip
'/ -:' :";:". -jpaie-: a/ 1/78;- :"
'" *'--- '' ''-' " ''"'
".."":"'" Ar^X*'-""
-, ^-'.-<»^->'.-:Fr!vvT??^? v7«i^ff^;. r
' * .- .w'j^BS^^^Sfi*^-;' ' .'."
;; ;;-.-..'. ,-.,;..: ::'-^'^^:s::
-------
SUMMARY REPORT
FOR
TWO YEAR PERIOD
June 1976 - July 1978
OF
AN EXPERIMENTAL STUDY OF LAKE
LOADING BY AEROSOL TRANSPORT
AND DEPOSITION IN THE LAKE MICHIGAN BASIN
Principal Investigator:
Grant:
Project Officer:
II. Sievering, Professor of Engineering Sciences
College of Environmental and Applied Sciences
Governors State University
Park Forest South, IL 60k66
312/53^-5000 x2H98
#R00530101
J. Regan, Chief
Air Surveillance Branch
Region V
U. S. Environmental Protection Agency
Chicago, IL
August 1, 1978
-------
PREFACE
This Gummury of the first two years' data from the Lake Michigan Aerosol
Loading study should be viewed as just that - a summary. We have confirmed
the presented data base to be a quality one unfettered by contamination or
instrument calibration errors. However, the large size of the data base has
allowed us to only recently begin a detailed interpretation. It is for this
reason that the present reporting is but a summary. The Interpretation sec-
tion must be considered preliminary. At this writing only two of some fifteen
or more scientific articles have been submitted for publication. Not only a
thoroughgoing interpretation on our part but also a thorough critique by the
scientific community are, thus, still wanting.
Because much of the data has not yet been analyzed only a small portion
appears in this document. The oft referenced Appendix I computer printout of
the complete data base is available at cost from the Principal Investigator
upon approval from the U. S. EPA. The one other major data base not part of
the computer printout is the NCAR air data. Because this data base is still in
the raw interpretation phase it will be several months before the mesometeoro-
logical analysis is available.
We are indebted to many, many people who supported this effort. Mehul
Dave' and Donald Dolske, full-time research associates, are most prominent in
this regard. GRU students Mike Eason, Jil Forst, Vic Jensen, Pat McCoy, Rich
Rupert, Nell Sutton and Keith Walther all contributed "flesh and bone" as well
as Brainpower. Grant secretaries Rebecca Borter and Elaine Sherman are espec-
iai±y acknowledged for their dedication in spite of the chaos presented to them.
To rj caff of the U.S. EPA, especially Bob Bowden and Jerry Regan, we are most
graceful. To Capt. McLain and his crew of the R/V Simons - thanks for bringing
-------
us through those storms, for without you, we couldn't "be writing this report.
Finally, we thank the National Center for Atmospheric Research staff, especially
Paul Spyers-Duran and Pete Orurn for their assistance.
This work was done under U.S. EPA Grant #R00530101. I gladly accept full
responsibility for this document. Any errors of omission or commission are
my responsibility.
August, 19T8 H. S.
Park Forest South, 111.
-------
INTRODUCTION
As recognized in the original proposal for this grant project the
need for meteorological and ambient air data over and on the Great Lakes
is self evident upon reviewing model predictions of atmospheric loading
to these lakes (Winchester, 1971, Skibin, 1973, Gatz, 1975, Sievering, 1976,
and Eisenreich, 1977)- Specifically, the transport processes as a func-
tion of climatology and, even more importantly, the deposition of low
lying - hereafter, referred to as surface layer - ambient air constituents
to Great Lakes water bodies is essential before any model calculations
can be considered credible. The measurement of wet deposition as a func-
tion of over-lake rain climatology is being considered by other researchers
(Bolsenga, 197^ and Murphy, 1976). The measurement of dry deposition to
large lakes,especially of aerosols, has been largely neglected. A major
purpose of this work is to determine the dry deposition of surface layer
aerosol to Lake Michigan as a function of aerosol size and of surface lay-
er meteorology. This surface layer meteorology or micrometeorology should
be measured under sufficiently varying conditions to identify a microclim-
atology for aerosol dry deposition. This is so because of the much greater
potential for Great Lakes loading in an unstable surface layer - i.e.,
when the air temperature, Ta, is less than the water surface temperature,
Ts. Further, Sehmel and Hodgson (197*0 found that for aerosol deposition
to a water surface_ in_ a wind tunnel the rate of deposition or deposition
velocity, V~d, of 2 pm aerosols was several times that for 0.2 ym aerosols
(see Figure l). A second purpose of this work is to determine trace metal
and nutrient concentrations at representative locations on Lake Michigan
to enable flux estimates. That is,
Flux (F) = Concentration (C^) x Deposition Rate
where F is in ug/sq. meter/second
-------
-2-
Cj_ is the concentration of constituent i in ug/cubic meter and
V, is the deposition velocity in meters/second
Once having calculated these fluxes an overall loading in Tons/Year
can be determined if a micro-and mesoclimatology for over lake waters is
available. Determination of the microclimatology has been identified as
a part of the first purpose of this work. An additional third purpose
requires that a meso-climatology for the Lake Michigan Basin be deter-
mined. In essence this meso-climatology identifies constituent concen-
tration as a function of location and time of year. No fully adequate
way of doing this is available, primarily due to the extremely large
spatial and temporal data base required. The approach used here is the
intense collection of data during preselected periods of time. This
data is then used to generate forward and back-trajectories at the meso-
scale. With these trajectories in hand concentration variations on the
lake-spatially and temporally-can be estimated and then combined with
deposition velocities as a function of climatology to arrive at overall
loadings.
ri'o accomplish the three major purposes identified above a rather exten-
sive sampling program is required. First, meteorological and aerosol
sar~ling instruments were used aboard the R/V Simons at a midlake location-
s':"".")' and U2°OC' - for approximately fifteen 2U hour sampling days during
-.; nl to September 1977. These same instruments have been located at a
^io./ cf Chicago nearshore location since May 1978. Second, aerosol samples
collected on low trace metal and nutrient concentration filters have been
analyzed by inductively coupled argon plasma. (ICAP) emission spectroscopy
-------
-3-
at the U.G.EPA's Central Regional Laboratory. The filters each carry
only a three to six hour aerosol sample so that each filter set is
obtained within a restricted micrometeorological regime. Because of
this sampling requirement some trace metal filter concentrations are
below 1CA? detection limits.
A third major sampling developed a mesometeorological data base for
trajectory analyses. In addition to obtaining National Weather Service
and Coast Guard meteorological data during midlake and nearshore sampling
an intense mesometeorological and aerosol concentration data base was
generated through aircraft overflights in June and September of 1977 and
May of 1978. This data has been supplied by the National Center for
Atmospheric Research C^ueeri Air Aircraft through an NSF supporting grant
of nearly $100,000.
The next section of this summary report reviews the data base so
far collected and explains the intermediate stage of analysis which com-
puter interpretation has afforded. To better appreciate this interpreta-
tion a brief review of the diabatic drag coefficient approach to the V^
calculation is warranted. A more complete description is found in the
original proposal narrative.
Usinpr the eddy view of atmospheric turbulence and assuming the sur-
face layer over Lake Michigan to be a layer of constant mean aerosol
f^ux (i.e., that the average rate at which'material moves up and down is
a -rr.-tant with height within the surface layer) an expression for that
f'^'.x Is given "by
= o .
-------
-k-
where F. is the mean flux of constituent A
p is the density of air
a" is the mean concentration of A at some reference
ht. within the surface layer (in this case 5 m)
aQ is the mean concentration of A at the air/water
interface
DD is the diabatic drag coefficient
na is the mean wind speed at the reference ht.
TT^, is the flow speed of the air/water interface
The method of obtaining CDD values is quite involved and is detailed
in the Appendix of the original proposal. Suffice it to say that CLp
is determined from micrometeorological measurements of the air tempera-
ture, air/water interface temperature and of the surface layer wind
speed. The mean wind speed, mean air concentration of a constituent and
flow speed of the air/water interface are readily obtained. It is only
regarding the air/water interface concentration of the constituent of
interest that approximating assumptions need to be made. One appromi-
nation not invalid during unstable air sampling is that aQ - 0. This
approximation is suspect for stable air sampling and clearly invalid when
surface rr.icrolayers with high trace metal concentrations are present. The
assumption that ao = 0 will first be made in the computer interpreta-
tion phase and then scrutinized in the section headed Interpretation.
-------
-5-
1 DESCRIPTION OF THE DATA
The first two years of work have shown that our field study of midlake
aerosol characterization and deposition, and the consequent application of
the data to trace metal and nutrient loading estimations has been successful.
The 19TT field program was supported by access to the R/V Simons as a mid-
lake sampling platform, CRL's analysis of filter samples for trace metals
and nutrients, and National Center for Atmospheric Research (NCAR) air-
craft overflights for spatially distributed aerosol and mesometeorological
data. The following section of this report is a description of the 1977
data base. Appendix A contains copies of all major computer printout com-
ponents of the data.
The raw data collected on board the R/V Simons, manually calculated
parameters, and results of chemical analyses from CRL have all been entered
on permanent GSU computer system files. Additionally, copies of the data
files are kept on magnetic tape cassettes at GSU as a safeguard against
possible loss of data due to system failures. All of the computer data
files nave been manually edited and verified to ensure conformity to the
original raw data. A series of twenty-five FORTRAN IV programs has been
written to process the field data into a readable intermediate data base,
presented in Appendix A and split into eight major divisions:
A. I'icrometeorology
3. Trace Metal Concentrations in the Atmosphere
C. 'Irace Metal Mass Flux
D. Total Aerosol Mass Flux
r.. Phosphorus, Nitrate, and Sulfate Atmospheric Concentrations
F. rno3phorus , i.itrate, and Sulfate Mass Flux
C.. Trace Metal Concentrations in Lake Water
-------
-6-
Each of these eight divisions is described below. The description
relates the raw data elements used in processing to the final printed
results shown in the various tables.
A_- __ Micrometeorology
The first section in this division contains micrometeorological data
for each of the filter set sampling periods. The mean, standard deviation,
and number of measurements in each set are shown for water surface tempera-
ture (Ts), air temperature (Ta), and wind speed (U ). Ts is measured by a
downward looking infrared sensor, while Ta and Ua are measured at the five
meter sampling height with standard meteorological instruments. Mean sur-
face water current speed (Us) is also shown for each set.
These data enter into the calculation of the diabatic drag coefficient
(C,^r; which is "at the heart" of this experiment. The critical parameters
for this calculation are shown (mean and standard deviation included) on
che _cwer part of each page. They are (T& - TS ) , the temperature difference
between the air at 5 meters and the surface, and (U& - Us), the difference
in speed letween the air et 5 meters and the surface of the lake. The
stability parameters), a function of wind speed and (Ta - Ts), identifies
the extent to which loading to the lake is affected by micrometeorology.
r,esu_i.ts uf the C^j calculation were obtained manually but are now displayed
in ^eoticr; la.
The ~- -, values are shown as a mean result with a minimum and maximum
val^e whicn represents one standard deviation extremes. To additionally
characterize micrometeorological conditions averaged across a filter set
period, the local Richardson number (Rj) was calculated. R. is a measure
of the tendency for turbulence initiated in the surface layer to be self-
-------
-7-
sustaining. Lastly, the deposition velocity (V.) f°r each filter set
is shown. This value is based on the mean Cn~ for each set.
From this micrometeorological data, it is possible to exclude from
further flux calculations certain filter sets. Using the criteria that
(Ua - Us) should be greater than 2.0 m/s for the CTJTN theory to be reliable,
and that R^ should be less than 0.25 (an empirical critical value, above
which turbulence may be suppressed), sets 10010, 10020, 100^0, 20080, 20120,
30210 must be excluded. For these sets, and during stable air summer season
periods with (Ua - Us) ^2.0 m/s or R.r > 0.25, an assumption that no
deposition takes place is made and is certainly conservative. Climatolo-
gically this situation occurs no more than 2.5% in March, k% in April and
May, 1+. 5% in June, 5-5/<> in July, 6.0% in August, 3.5% in September, and
3.0/0 in October and probably not at all in November through February
(NOAA SSMO, 1975)
Section 2 of the micrometeorology printout contains the means on, and
standard deviation in, wind direction data for each filter set. Also printed
are barometric pressure data for the sampling periods. An assumption under-
lying the application of CDD theory requires the wind direction to be fairly
steady. No strictly quantitative definition of this steadiness is available,
however. Here, a value of 30° to U0° standard deviation about the mean
defines "unsteady" wind field conditions. Thus, two filter sets, 20150 and
^03^0, must be excluded, in addition to the six above.
Although not strictly part of the microscale data base, an additional
consideration for CQQ theory application is that relatively constant meteorol-
ogy prevails for at least 5 km upwind of the sampling location during stable
air conditions. (2 km or even 1 km suffices during unstable periods.) When-
-------
-8-
ever macroscale fronts appeared on southern Lake Michigan, the related
filter set was excluded. This eliminates sets 30220 and U0310. In summary,
then,of the 56 total filter sets collected, ten must be excluded from flux
calculations (and four were blanks), leaving k2 sets which warrant appli-
cation of the CDD theory and trajectory analysis.
A' . Mesometeorology arid Tr_aJectpry__Analysi_s
Though not part of the computerized data base, National Weather Service,
Coast Guard, and City of Chicago meteorological data were plotted on bi-hourly
soutnern Lake Michigan maps along with the shipboard data at 8T°00' and
U2°00'. This data base, along with calculated geostrophic wind vectors at
four points over the southern portion of the Lake, allowed self-consistent
air parcel streamlines to be drawn on each bi-hourly mapexcept for the May
outing, when iriesometeorological conditions were too complex. Forward tra-
jectories, initiated from Midway Airport, temporally linked the individual
map;-,. Air parcel trend linesboth spatial (streamlines) and temporal
(forward trajectories)are thus available to relate aerosol and aerosol
constituent concentrations at the midlake sampling point to source regions.
An example is shown in Figure 2.
An alternative method for source identification is the use of back-
trajectories. In principle a more accurate approach than forward trajec-
tories (simply because the trajectory is initiated at the sampling point),
cr.e .ick of meteorological data near to the sampling point (ship) forced
trie ase of the ship wind vector for most of the mesoscale trajectory (i.e.,
to shore). An example back-trajectory is shown in Figure 3. Macroscale
oacK-trajectories based upon the model of Heffter and Taylor (1975) were
also plotted by EPA Research Triangle Park.
-------
-9-
During the June and September 1977 and May 1978 outings, an NCAR
j4ueen_AjLr aircraft collected mesometeorological and vertical sounding
data to enable mesometeorological analysis and improved back trajectories.
These data have been received (see example midlake sounding Figure ^) and
will be used over the coming months. Plans for vertical soundings to be
taken on board the Siinons throughout the field program failed because the
Contel Corp. Metrosonde purchased for this purpose never functioned pro-
perly. Emphasis may have to be placed on the June and September 1977 and
May 1978 sampling periods for which the NCAR Queen Air gave a full comple-
ment of mesometeorological data.
B. Trace Metal Concentration and Enrichment Factors
Trace metal concentration data is specified within each set by stage
of each filter set. Figure 5 displays the collection efficiency of each
stage using Misco filter media for ambient lake aerosol. It is clear that
the 1 ym ana larger aerosol usually referred to as primary aerosolis
most efficiently collected by the first stage, with the remainder of the
primary aerosol collected by the second stage. The third stage or backup
filter most efficiently collects the 0.7 ym and smaller aerosolreferred
to as the secondary aerosol. The trace metal concentration data as a
function of stage therefore carry a physical significance. That is, the
first stage data, identify that mass portion of the trace metal under con-
siaeratlon which is related to primary (d >_ 1.0 urn) aerosol while the back-
up stage specifies the mass portion related to secondary (d <^ 0.7 Jim)
aerosol. I\iote that al] these trace metal concentrations have been blank
corrected by subtracting the mean metal concentration of eight blank Misco
-------
-10-
filters taken through all handling procedures except that no ambient air
was drawn through them (Bee Table 1 for blank filter concentrations).
Section three of Appendix A, then, contains the trace metal concentra-
tion data as measured at midlake. The top portion of each page gives the
"run time" for each set. This is the time in minutes over which the
filters were exposed at Ho cfm to the ambient air. Also given are the
results of three supportive measurements: Condensation Nuclei Counter,
Integrating Nephelometer, and Total Mass Monitor data (where available
from the University of Wisconsin).
A brief explanation of the filter set numbering system here will make
the trace metal data in Section 3 . much more readable. All samples taken
during a given set have a five-digit number, which encodes the following
information:
set number: A 3 C D_ E
A first digit corresponds to the outing number, where
1 = April 1977
2 = May 1977
3 = June 1977
1+ = August 1977
5 = September 1977
BCJ three digit sequential sampling period number,
000 - 056 for 56 sets completed in 1977
E last digit corresponds to type of sample. These additional
key values for the last digit will be explained when first
used in each section of Appendix A.
The results of ICAP analysis were blank-corrected, Yttrium-normalized,
ana run time compensated, resulting in the values shown in Section 3 in
uni~3 cf yg of metal per cubic meter of air. There are four rows of data
for ea.cn filter set:
AnCDl = first stage collected aerosol
ABCD2 = second stage collected aerosol
ABCD3 = backup filter collected aerosol
TOTAL = arithmetic sum of all stages
-------
-11-
These four rows of concentration data appear in IT columns, one for each
metal analyzed for by ICAP. These concentration data form a second critical
part of the data base required for performing lake loading calculations by
the CDD method. Too little data were obtained by ICAP analysis for Na, Cd,
Co, Ni, and V to be meaningfully used in further statistical calculations.
This was due to the relatively short average run time (2^0 minutes) for
the filters which caused very low concentration elements to fall below the
ICAP detection limits (L,) . In the case of Na, a very high L^ applied
from September 1977 onward effectively negated earlier reported Na results.
This came about because, in an effort to be self-consistent despite changing
ICAP Lj values, the highest L, used during the April to September 1977
period was applied to the whole period. This resulted in some of the blank
samples being essentially based on high L(j values . Immediately following
the data for set 50560, the original data from CRL's ICAP analyses is in-
cluded in the print.
From the chemical characterization viewpoint the concentration data
can be looked at aggregated as an overall average for each metal. This
compilation is given in Table 2.
A common term used to describe the relative contribution to trace
metal mass in aerosol of a particular metal is the enrichment factor (EF).
THIS factor is defined as a ratio of concentrations:
c _i:: air_of metal x
c in air of standard metal
c in crustal rock of
c in crustal rock of standard metal
Typical standard metals are Al , Fe, Na , and Ti . The most consistent
and reliable results for EF were obtained by using Al as the standard. In
-------
-12-
Table 3, overall average EF numbers show the extent to which certain
aerosol components are increased in relative concentration over that
which could be expected if all cf that aerosol component were crustal-
derived.
It is immediately apparent from these EF data that Fb, Zn, and Cu
are extremely enriched in low lying air over the lake.
Lastly, another type of information that can be gleaned from the con-
centration data is the percentage of each metal's concentration associated
with primary and secondary aerosol. In Table 3, note that the percentage
of each metal's mass collected on the second stage filter is not identi-
fied with either primary or secondary aerosol. This is because the second
stage acts to improve distinctness in primary/secondary aerosol between
the first stage and backup, but does not itself represent either of these
aerosol size classes clearly (see Figure 5)-
~ Trace Metal_Mass _Flux
The data in this section are calculated from the equation:
Flu* = Vd . CiJ
where C'-p is the concentration of metal i on filter state j. Note that
this calculation assumes that V^ is not a function of particle size and
tnat the concentration of aerosol metal at the air/water interface is
very small compared to the air side concentration. More will be said
about these two assumptions at the conclusion of the next section. The
^ata shown in the computer printout is of value principally in comparing
loading rates ir. one set to those in another particular set. Of more
importance and impact are the values shown in Table 5- .Each of the
columns in Table 5 represents the results of a method of estimating a
flux from the ctata obtained in the 1977 field program.
-------
-13-
First, the overall average method takes the mean V^ for all non-
exciuaed filter sets (0.69 + 0.51 cm/s) and products this by the mean
of all data on each metal. This flux is then converted to an annual
loading by an overall time-area factor, Ft&. F^a is the product of the
area (of the surface to which deposition is occurring) and the time
(during which deposition occurs). The area here is 2.9 x 10^ m^ for
the southern basin of Lake Michigan, and time is 0.85 year. Fifteen
percent of the time on average across a year, trace or more precipita-
tion occurs over the lake (f.ievering, 1976). Thus, 0.85 year is the
fraction of time when dry deposition can occur.
Second, it is possible to group filter sets together by simi-
larities in meteorological conditions. Wind speed "bins", defined by
ranges of wind speeds in meters/second as follows: 0 - 2.2, 2.3 - 3-55
3.6 - 5.2, 5.3 - 8.5, and 8.5+» were set up. Filter sets with average
wind speeds falling into these bins were grouped together, and actual V,
values for the bins were then determined. Using climatological data
(i-iOAA SSMO, 19T5)5 ?ta ^or eac^ ^in could be determined. The loadings
for these bins were summed up to give an annual loading, which was further
normalized to the 0.85 year dry deposition period. The results of this
calculation are shown in the second column of Table 5-
Third, exactly as the wind speed bin calculation was done, filter
sets were grouped together by temperature stability, (Ta - Ts) or AT.
riinb for this parameter, in units of °C, were set up as follows:
AT <_ 0.9, -0.9 to +0.8, 1.0 to 2.7, 2.8 to +8.1, and >8.2+.
This calculation results in the lowest annual loading totals of any of the
three estimation methods given in Table 5-
-------
-Ih-
D. Aerosol MassFlux
The data in this section are sampling set totals of particle number
counts, obtained from the Active Scattering Aerosol Spectrometer (ASAS).
The ASAS counts aerosol particles in 15 size channels in each of four
size ranges, i.e., a total of 60 size channels covering the overall range
0.09 - 3-53 pm diameter. Table 5 is a complete listing of the geometric
mean size aerosol counted in each channel. The mean flux printout in
Appendix A. not only gives the total counts in each of the 60 channels
for each sampling set, but also expresses a total flux in ng - m~2 sec~l«
The actual particle counts make up the bulk of the table. However,
it is important to note that these counts must be normalized to unit
time with the accumulation time shown in the upper right part of the table,
This is due to the elimination of certain portions of the data when actual
number counts exceeded the memory capacity of the data accululation
system of the ASAS. Thus, the total counts shown represent different
total times during which the ASAS was "looking1' at a given size range.
A mass flux for the total counts in each size range is given at the
bottom of each page. At the left top, two non-overlap total ASAS fluxes
are given. As can be seen in Table 6, there is some duplication of
counts in several channels toward the ends of adjacent ASAS size ranges.
These overlaps can be excluded in two ways: by excluding the higher size
ran^e overlap portion (labelled >), or by excluding the lower size range
portions (labelled <). The total number counts generate mass flux re-
sults for total aerosol in the 0.09 - 3.53 vm range by the following equa-
tion, which is an application of the C theory:
-------
-15-
where p^ = mean aerosol density,
assumed here =1.5 S/cm
= diabatic drag coefficient
= total number count
= wind speed at 5 rc, minus
___ lake surface current
F Y" = volume of particle at
mean radius of size range
| / for which En was calculated
v/5 = sampling volume per second of
ASAS, a constant of 0.28 cm3 s"1
: total accumulation time
in seconds for the size range
being considered
An important consideration in characterizing midlake aerosols is a
description of the particle size distribution. For the 0.09 to 3.53jum
diameter particles "seen" by the ASAS, this distribution is best visualized
by a plot of An/log Ar versus r (see Figure 6 for plots of monthly averages).
i, Phosphorus , Nitrate, and Sulfate Concentration Data
Concentration data for total inorganic Phosphorus, Nitrate/Nitrite,
and Sulfate (hereafter P-N-S) is reported as a function of filter set and
stage of each filter set. As was done with trace metal data in section B,
it is possible to relate the portion of P-N-S filter concentrations to be
blank corrected by subtracting the concentrations reported for filters
taken through each step of the procedure except ambient air sampling.
Table 7 gives the mean concentration data of P-N-S observed during
the April through September 1977 field program. It is significant that
all of these potentially nutrient-available species are strongly associated
with secondary aerosol (d <_0.7 vim). The secondary aerosol, having a higher
surface-t-o-vulume ratio, may tend to make the aerosol forms of P-N-S more
soluble and therefore more available to the lake biota. Also, in the form
of secondary aerosol, atmospheric transformation reactions are more likely
to affect the concentration and forms in which Phosphorus, Nitrogen, and
-------
-16-
Sulfur are to be found.
Note that in Appendix A., the sample identification key for the
P-N-S samples is as follows:
ABCD6 - 1st stage collected (primary aerosol)
ABCDT - 2nd stage collected aerosol
ABCD8 - backup collected (secondary aerosol)
ABCD0 - Total c of all stages
F. Phosphorus_,^j'iitrateJ, and Sulfate Mass Flux
Using the same method by which trace metal fluxes and annual loadings
were calculated in section C. , similar results are shown for the P-N-S
data. In Appendix A., the sample set derived results are shown. Table 8
is a listing of annual P-N-S loadings by overall average and micrometeor-
ological data groupings, similar to Table 5 for trace metals. The rela-
tive importance of atmospheric route loadings of P-N-S are shown in Table 9
by comparison to water shed input and wet deposition. However, the ultimate
importance for these nutrient species may be of a more subtle and insidious
nature. Due to the critical roles of these nutrients in the lake ecosystem,
any source of input of P-N-S will have direct effects.
G. Trace Metal Concentration in Lake Water
Two tables of trace metal concentration data are presented in Appendix A.
Tne first table contains the results of ICAP analysis of bulk water, i.e.
composite samples of water taken a 3, 5, and 7 meter depths. The second table,
where the sample identification number is ABODUO, contains results for sur-
war,er samples, taken with a surface microlayer screen sampler (Elzerman, 1976).
All of the water data is intended in this study to enter into the interface
aerosol metal concentration term of the CDI) - derived flux equations . How-
ever, inadequacy of current theoretical models for transfer mechanisms of
-------
-17-
aerosol across the air/water interface has forced this term to be assumed as
small compared to the air side concentration. This assumption vill be dis-
cussed in detail in the Interpretation section. Table 10 is a comparison of
the mean concentrations of several metals in surface versus bulk lake water
for the 5BCDH series of samples. These data clearly suggest a trend towards
surface microlayer enrichment for many metals.
-------
-18-
II INTERPRETATION
A. Concentrations
Since the flux is a product of specie concentration and deposition velo-
city, the interpretation section is divided into Concentration, Deposition
Velocity, and Consequences.
The average midlake aerosol concentration was found to be ^5 i 10 ug/m-^.
About 10 to 20% of this midlake aerosol is trace metal contributed with sulfate
and nitrate/nitrite each contributing another 20% or more. A comparison of aero-
sol mass concentration (M) with integrating nephelometer optical scattering
(bscat) signals suggests that a linear regression with strong correlation exists
over Lake Michigan. The equation is
b = 0.298 + 0.0138M (l)
S C 9. U
with an r = 0.97. A strong linear correlation has been found by other researchers
in urban environments (CharIson, et.aL (1969), Kretzschmar (1975)), but the slope
of equation (l) is much less than urban related correlations. This linear regres-
sion equation and the high correlation between bscat and M suggest that fewer
large aerosol may contribute to midlake mass concentration than in urban environ-
ments. If one assumes the aerosol density to be constant with aerosol size,
then n.ass is proportional to aerosol size. One can then expect less variability
in aerosol mass as measured by the integrating nephelometer for midlake aerosol.
This Is an important reason for the correlation coefficient of the above
equation to be 0.97.
7r.e log-log plot of Af.AS aerosol counts (see Figure 6) confirms the IN
;ata In suggesting that nidlake aerosol concentration may be dominated by second-
ary aerosol. A best fit straight line to such log-log plots results in a slope
of -2 to -3 for urban polluted and rural continental air. Marine air and upper
-------
-19-
level air often exhibit slopes of -3 to -k. The fact that the average slope for
the 1977 field program is -3.25 is strong confirmation that the midlake aerosol
is depleted of primary aerosol. Further resolution of the data base gives
slopes of -2.3, -3.1*, -3.3 and -3.8 for May, June, August and September data
respectively. The stable air of the May Outing forces many of the secondary
aerosol to remain above the thermal internal boundary layer (described later)
while increasing the number of primary aerosol by suppressing dispersion within
the surface layer. Except in this strongly stable air the slope of the log-log
plots support the conclusion that secondary aerosol predominate the midlake
aerosol.
This conclusion explains the predominance of trace metal on the backup
filter even for metals such as Fe and Zn (see Table ^). If the concentration
of metals is dominated by that portion associated with the secondary aerosol
one might conjecture that much of these metal concentrations in the atmospheric
surface layer at midlake is man derived. Lake derived contributions would tend
to be associated with large aerosols (Duce, 1976 and Maclntyre, 197M- By using
average crustal rock trace metal concentrations for rock most prevalent in the
Great Lakes basin (Wedepohl, 1971) and the average bulk water trace metal con-
centrations measured during the 1977 field program the percent of each trace metal
that is derived can be estimated. Using Al again as the reference metal in air
and Mg as the reference metal in Lake Michigan waters, the resulting estimates
are shown in Table 11. That 90% and more (99.9% of Pb) of several metal midlake
concentrations are neither crustal nor lake derived is quite dramatic. One is
tempted to relabel "unexplained" as "anthropogenic". However, one should not
rule out the possibility that large lake-derived aerosols may fractionate in air
to create snail aerosol with metal concentrations no longer representative of
their relative concentrations in lake water (Duce, 1976).
-------
-20-
Correlation coefficients using filter set average concentrations of
metals, P, NO^ and micrometeorological data are shown in Table 12. Any cor-
relation coefficient 0.7 or greater is circled. Almost no correlation of
micromet data with metal P or NCU data is found, whereas inter metal and NCU
correlation is quite prevalent. The correlation of P with other data is
present but usually not with high correlation. The generally negative correla-
tion between wind speed and other variables indicates lower midlake concentra-
tions with higher wind speeds. However, the small magnitude suggests very
little dependence on this factor. The high correlation of NOo with AT sup-
ports the notion that in situ chemical generation of NOo is significant.
The correlation coefficient matrix of Table 12 does not well identify
groups of factors that may be co-related. Factor analysis is a statistical
technique which can give additional information by using linear combinations of
the factors in the correlation coefficient matrix. A preliminary use of fac-
tor analysis has identified the following elemental groups:
Group 1 - Cu, Fe, Mn, Mo, Pb, Zn and Mass
Group 2 - Ca, Mg, Al, Ba, Fe, Mn and Ti
Group 3 - Ti, P and S0|
Group k - NO-j, S0£ and AT
The first group is clearly the secondary aerosol metals. Note that mass
is correlated with these metals. The second group is the primary "soil
derived" aerosol. Pollutant and lake-derived source contributions may, however,
contribute to this group. Group 3 may be a lake-derived source and group U is
clearly the chemically (in air) generated aerosol. Of the 9Q% or more "unex-
plained" elements in Table 11 only Ti is not part of Group 1. This grouping
by factor analysis is strong evidence that, except for Ti, the lake-derived
large aerosol fractionation is not a contributor to the otherwise anthropogenic
source of tr.e metals Cu, Pb , Zn, Mn, Mo and Fe.
-------
-21-
Another fruitful way to view the metal concentrations is through the
use of scattei-nliagrams (Rshn, 1976). Quite characteristic patterns appear
when the enrichment factor for a metal of interest is plotted versus the
concentration of the reference metal - in this case Al. An example is shown
in Figure 7, the scatter-diagram of Pb versus Al. Note that data collected in
April 1977 is also plotted here even though it was not used in loading cal-
culations. This scatter^diagram is not significantly different from that found
in other literature. The area within the dashed line is where the bulk of
other reported Pb enrichment factor data resides. Data points "below the dashed
line are not uncommon and indicate cases in which the Pb concentration is
about equally contributed to by primary and secondary aerosols. Data points
to the upper left of the dashed line are rare. A goodly number of June,
August and September data points appear in this area. This is strong evi-
dence that the primary aerosol containing ^0% or more of the atmospheric Al
concentration is depleted before the midlake sampling point. In addition,
relatively little Pb has been depleted. Thus, a very large Pb enrichment
factor with a low Al concentration results.
A second scattepdiagram is presented in Figure 8 - that for Ti. Most
researchers have reported a trend toward constant enrichment factor. The
data of Figure 8 suggests instead a trend toward constant concentration.
This would indicate that a relatively constant source of Ti is present at
midlake. Either the lake is indeed a source for Ti or despite all precau-
tions the Simons effluent may have been a contaminant source. That the
lake is a source for Ti is also supported by the factor analysis results and
the ciata of Table ^. Fully hO% of the Ti concentration at midlake is asso-
ciated with the primary aerosol. This portion of the Ti concentration is
-------
likely to be lake-derived as against contaminant derived. Other scattepdia-
grams riot shown give additionally useful interpretation. One particular
example is the scatteinliagram for Mg which substantiates the very strong
source the lake is for mid-Lake Michigan Mg concentrations. The use of Mg
as our lake source reference metal is, therefore, substantiated.
The preceding discussion of trace metal and P, N, S specie as well as
mass and number concentration is incomplete without an understanding of the
mesometeorology surrounding the concentration data base. As pointed out in
the Introduction the concentration data must be viewed as a function of loca-
tion and time of year. Though some of this mesoclimatological view can be
obtained from the previous discussion it is the trajectory analyses at the
mesoscale that can best generate mesoclimatology data needed for a good under-
standing of spatia] and temporal variability in concentration. Some prelim-
inary analysis will be reviewed here.
A first useful parameter for mesoclimatology is the so called dispersion
factor, the product of the mixing height and mean wind speed. Edwards and
Wheat (1973) have shown this dispersion factor to be well correlated with
atmospheric lead concentrations in Denver. A correlation between monthly
mean dispersion factors across a H2-month period concludes that variability
in the dispersion factor alone accounts for 69% of the variability in the lead
concentrations. Monthly mean mixed layer wind speeds would have given a
better correlation. Still better would be to use emission factors producted
by the inverse of the dispersion factor.
, ;;ta for the southern basin of Lake Michigan may not suffice to determine
emission factors (although this will be scrutinized). However a four year
record of j_ow icvel soundings at Midway Airport does allow the determination
-------
-23-
of mean mixed layer winds and of the mean mixed layer height. Use of the
mean mixed layer winds makes their extrapolation to the southern "basin much
more acceptable. By the two major periods over Lake Michigan (i.e., warm
season - May through October and cold season - November through April) the
mixed layer heights are 1^00 m in the warm season and 850 m in the cold
season. The mean mixed layer winds are 6.8 m/s in the warm season and 7.8 m/s
in the cold season. Thus, the dispersion factor is 9520 in the warm season and
6630 in the cold season. Concentrations of metals such as Pb are then propor-
tional to the inverse of the dispersion factor. This level of disaggregation is,
of course, not sufficient to specify a concentration mesoclimatology for the
southern basin of Lake Michigan. One could disaggregate by month although the
concentration climatology is more likely to be dependent upon air mass type. An
analysis of the Midway soundings by air mass type tentatively has concluded:
Warm Season - 30% continental polar, 1500 m MLH
35% maritime tropical, 1275 m MLH
25% uncertain, 1150 m MLH
Cold Season - 15% continental arctic, 1000 m MLH
25% maritime polar, 900 m MLH
10% maritime tropical, 875 m MLH
25% uncertain, 775 m MLH
10% of the warm season and 25% of the cold season air masses can be
designated washout air masses due to precipitation of 0.01" or more.
Clearly, further consideration will have to be made of the meteorological
data oase in the nouihern basin of Lake Michigan. In particular, soundings at
Midway rr.ust be compared to those taken over Lake Michigan by the NCAR aircraft
to outain a mesociimatology of the dispersion factor. Analysis such as
the foregoing should eventually lead to a statement on the concentration
climatology over Lake Michigan.
-------
B. Deposition Velocity
Use of the diabatic drag coefficient theory has given a grand average V-,
encountered during the 1977 field program of 0.8 +_ 0.5 cm/sec. Extreme
values were 0.09 an(i 1.66 cm/sec. Means for the May, June, August and September
Outings were 0.22 +_ O.I1!, 0.28 +_ 0.17, 0.89 +_ 0.31 and 1.12 +_ O.UO respectively.
This V^ is obtained by first determining a neutral drag coefficient for the
wind speed regime in which one sampled (See Figure 9). Then a ratio of dia-
batic to neutral drag coefficient is obtained for the temperature stability
regime present during sampling (See Figure 10). As already pointed out, the dia-
batic drag coefficient approach should give good results except when (ua - us)
$2 m/s or the local Richardson number, R^, is greater than 0.25. At these
values surface drag and surface layer turbulence have been suppressed to the
point that a laminar sublayer may develop at the air/water interface. Once
this phenomenon occurs there is likely to be little or no deposition occurring.
Only a few such cases appear in the 1977 data base.
The mean V^ by Outing (0.22, 0.28, 0.89 and 1.12) compare favorably with
other data found. Sehmel and Sutter (197^) found values of 0.01 cm/s and larger
for deposition of particles of density 1.5 gm/cm depositing on a water surface
in a wind tunnel. However, a strong dependence on surface roughness was found.
Cawse (197^) found V of 0.1 cm/s and larger for a filter surface exposed
to ambient air. The real environment surface of Lake Michigan not only affords
substantial roughness compared to a wind tunnel having a water surface but
also may afford other and more efficient mechanisms for deposition than a filter
surface exposed to ambient air. In particular, a recent study by Stulov, et.
al. (,.978) considers the efficiency of collision of aerosols with water surfaces.
It vis- shown that :it a clean water surface all collisions are effective
-------
-25-
and lead to aerosol transfer into the water medium. However, when aerosols
collide with other aerosols previously deposited on the water surface they
usually rebound. Thu;-> a V of 0.2 up to 1.0 cm/s and larger is certainly
feasible.
The work by Stulov also addresses the assumption that a =0. It was
noted in the Introduction that this assumption was made for all calculations
in the Description of Data section. Mass flux calculations using the ASAS
number concentration data can safely assume a =0 for the surface of Lake
o
Michigan rarely affords a significant probability for aerosol-aerosol colli-
sions. Possibly when the laminar sublayer is present aerosol may collide
with aerosol instead of the water surface itself. However, in these instances
V has already been assumed to be zero. Further support for the appropriate-
ness of a =0 is given by the work of Meszaros (1977). She has found that
90% of summer continental aerosol not unlike that seen over Lake Michigan and
in the size range 0.02 ^ r < 100 jam is water-soluble and 50% or more is water-
soluble in winter months.
regarding specie specific fluxes the a =0 assumption is more difficult
to accept. However, the combination of arguments applied to metal concentra-
tions earlier also applies here. The Table 11. result suggesting 1.0% or less
of Pb, Zn, Ou, Mn, Fe and Ti and only 2% of Mo midlake concentrations are lake-
derivea is a strong indicator that the air/water interface concentrations
of these metals are generally far less than that measured at the 5m height.
The tentative grouping of metals by factor analysis supports the notion that
these metals, except for the anomalous case of Ti, are not lake-derived des-
pite the possible fractionation of large lake-derived aerosol into less than
1 jim aiameter aerosol. More importantly, the work of Stulov (1978) and of
-------
-26-
Meszaros (1977) suggest a near 100% efficiency of collision and better
than 50% water solubility of aerosols throughout the year. Thus, the number
concentration of aerosol and mass concentration of non-volatile, non-reactive
elemental species should change across the air/water interface. Of course,
the exception to this rule is when the laminar sublayer prevails and greatly
reduces the probability of collision between aerosol and the lake surface.
Under the assumption of a smooth continuum of concentration the profile
method of measuring V (call it V ) may be applied. The principle behind
this method requires that measurements of the concentration of interest be
made at two heights. If the instrumentation used does, in fact, discrimi-
nate between the concentrations at the two heights, call them H for high
position and L for low, then a mean profile method deposition velocity, V
is determined by: _ _
CH ~ CT
where u is the wind speed at the standard 5 m height.
During tne 1977 field program measurement of Vn was attempted for aerosols,
dp
and effectively for at least the metals Pb, Zn, Cu, Mn, Fe and Mo, by use
of ASAS number concentration data at H = 6 . 5 m and L = 3.7 m. Preliminary
analysis of the profile data showed over half of the data had to be discarded
iioe the ship was not sufficiently close to being a fixed platform.
Th Is , Despite the fact that V means for 30 minutes and more were sought
dp
after. '.h*1 ship's rise and fall with wave action caused the CTT and c
n L
values to fall within one standard deviation of each other. Only nine pro-
fi±e measurements gave statistically significant c - c . However, the
n L
nine remaining cases do give substantial insight into the question of a £f
-------
-27-
One case haa uc = h.l m/s and (T - T ) = 2.0° C. Using Figures 9
_p £1 S
and 10, C = 1.16 X 10 . This then gives a Vd = 0.58 cm/s. By using the
number concentration data of the ASAS in the above equation the following
V, were calculated:
dp
V = 1.09 cm/s 0.75 urn < D < 3.5 Jim
dp "^
V 0.36 crn/s 0.25 < D < 0.75 jam
V. - 0.62 cm/s 0.10 urn < D < 0.25 jam
dp ^
Consideration of the nine profile case studies suggest:
V V (0.75< D< 3.5
-L V -, v", (0.25 < D < 0.75 jam') < 7,
10 d dp d
A V, V, (0.10 < D < 0.25 Jjm) < V,
10 a dp ^ d
The deposition velocity seems, indeed, to vary with aerosol size but not
as strongly as Sehrael and Hodgson (197M suggest. However, the conclusions
from these profile measurements must be considered quite tentative.
Until, further profile measurements are made a tentative conclusion
regarding a. ±
-------
-28-
Filter sets were grouped together according to averaged micrometeoro-
logical condition:; for the time the sets were taken. Two parameters were used,
each in a separate aggregation scheme: temperature stability (T - T ),
s. s
and mean wind speed (u). An at least ten-year-long record of synoptic condi-
tions over the lake (NOAA SSMO, 19T5) was used to determine the April to
September frequency of occurrence (F ) for given values of each parameter.
"Bins" were set up to correspond with reported ranges of values for (T - T )
a s
and u? and the filter sets were then sorted into these "bins".
The (T - T ) bins are based directly on values and ranges for air/sea
a s
temperature difference reported in the climatological history. Thus, calcu-
lating an annual F was a matter of arithmetically summing the monthly-mean
values. It should be noted here that the historical data reports air/water
temperature difference whereas the micrometeorological data of this
work is for air/surface temperature difference. T , in this work, was
o
measured by a downward looking infrared sensor and represents a true sur-
face temperature. During periods of intense insolation, and particularly
in the presence of organic surface films, T may be somewhat higher than
o
the water temperature. The result is a possible bias in the F for the
(T - T ) bins towards more stable conditions. This bias would lead to
a s
lower estimates for total annual loadings, because F will be larger for
the stable bins, which have lower average deposition velocities. Table 13A
shows the li-ita for the ('1\ - T^) bins. Each bin is identified by a range
of v,i'ties J'or (T_ -'!'_), C. The next row is the F values corresponding
to Jiimatological data (NOAA SSMO, 1975). The number of actual sets in each
bin is given, and an average V , cm/s for those sets is shown in the last
row. For each metal, mean concentrations, c, for each bin were found, and
-------
annual loadings were calculated from this equation:
Bin Loading = V , I1' c A k
d o
2
where: A = area of southern basin = 2.9X10 m
k = constant to correct units
Finally, the five bin loadings in metric tons/yr. were summed to give
an annual loading for the 0.8l year the F values for April to September
represent. The result was normalized to 0.85 year loadings to compare with
other dry deposition calculations of this work. The normalization of the
(T - T ) bin loadings tends to further bias the results toward a least-likely
as
total annual loading; as can be seen from Table ISA, our data is unstable-
case deficient. There is a six-fold increase in Vn from the most to least
d
stable bin, and more unstable (T - T <0) data would likely result in a bin
cL S
with a still higher V . Bin loading from that unstable data would substan-
tially increase the annual loading total.
Table 13B shows thr u bins, similar to the (T - T ) bins discussed above.
a s
Bin .1. jadings and annual lake loadings were calculated in the same manner
as for the temperature stability result. In the case of climatological data
for u, the Historical record was not reported in units or ranges conveniently
compatible with our own data. By calculating annual F values for the
o
reported ranges, and plotting cumulative F versus u, it was possible to
interpolate F values for the "bins" used in this work. The u bins are
o
basea very closely on the Beaufort wind scale increments. Note that again,
I1^) result, our data is severe (high u) condition deficient,
and leaas to a least-likely total annual loading. If more high wind data
were available, a. higher V bin would result and increase the final sum
loading substantially.
-------
C. Consequences
Given the data generated and interpretation to date some discussion
of the consequences of the present results is certainly warranted. Though
no person on the grant project team is in a position to assess these impacts
we hope others may do so. There are a few literature sources which may help
the reader in this assessment.
At first thought a diffuse area source of toxic metals depositing on
the lake's surface might be thought of less damaging to the lake ecosystem.
However, several Great Lakes biological researchers have suggested (in private
communiques)that lower organisms in the near surface waters or in the sedi-
ment may ingest significant amounts of trace metals such as Pb due to the
atmospheric source. Bioaccumulation may then result in toxic levels of these
metals in edible fish.
Pb is a most important candidate for investigation regarding atmospheric
inputs. The enrichment factor of over 3300 and essentially zero concentration
contribution from other than anthropogenic sources make it the most lethal -
potentially - of the elements analyzed in this data base. Patterson, et. al.
(1976) finu a small Pb flux to the oceans. Yet ecosystem impacts are noted.
They also point out the selective food chain transport of Pb between albacore
and riCtruen. Maclntyre (197^) identifies a major ecosystem impact when the
laminar sublayer is present. Under this condition of greatly reduced transport,
airnorne metals rna.y collect in a surface microlayer with enrichment factors
greater i.hau 10000.
T'u accumulation in freshwater fishes is a strong function of lake chemistry
(Torrey, 1976). For example, Merlini and Pozzi (1977) studied the accumulation
-------
-31-
of Pb by an edible freshwater fish at pH 7.5 and pH 6.0. Although the sites
of lead concentration were not altered, the fish concentrated almost three
times more Pb at the lower pH than at the higher pH. Whether the Pb remains
in particulate form or dissolves is certainly another significant factor.
Each of the metals have unique pathways into edible freshwater fish. With
the sharp increase in coal utilization for energy production the next decades ,
the impact of atmospheric trace metals on Great Lakes waters will surely
increase.
The impact of nutrients input to the lake via the atmospheric route may
be more direct than the impact of metals. In a phosphorus-limited ecosystem,
an addition of that nutrient in small particle or soluble form will cause an
immediate response by the primary producers. Moreover, in areas of the lake
far removed from nutrient sources other than the atmosphere, the availability
and relative concentrations of nutrient materials may differ from that of near
shore regions. The implications for the diversity in and abundance of aquatic
biota are not well known. It is hoped that atmospheric loading impacts will be
increasingly addressed in the literature.
-------
-32-
III FUTURE WORK
Review of the present data base suggests at least three areas for
improvement:
1. more unstable surface layer data and nearshore concentration
measurements
2. more and better measurements of V.
d
3. the development of trajectory analysis programs and a meso-
climatology for aerosol transport
The comparison of the 1977 field program micrometeorology with Lake
Michigan's southern basin climatology has shown the need to obtain data in
high wind speeds (u > 10 m/s) and late fall or winter months (T - T < -2 C).
3, S
Such data will be very difficult to obtain at midlake but may be reasonable
at a nearshore fixed platform. For this reason and to characterize the near-
shore aerosol concentrations the 1978 field program is taking place at a
City of Chicago water intake crib 2 miles offshore. This sampling should
create a data base to address the first area for improvement.
"he second area may be more difficult to address. However, the profile
method has already given better data at the crib's permanent platform sampling
site. Several studies of the mass transfer coefficient from liquid to air
(Cohen, et. al., 1978, Duce, et. al., 19779 Brtko and Kabel, 1976) may result
in an accurate statement of the a correction to earlier flux calculations.
o
Lastly and most difficult will be the determination of concentration var-
iability both spatially and temporally. Some consideration was given to this
q^ues^ion at the conclusion of section II A. Section I A1 described the forward
and back trajectory analyses being plotted on southern Lake Michigan maps.
The manual plotting of back trajectories (example, Figure 3) will allow the
designation of 95% confidence, three-dimensional source regions from which the
-------
-33-
aerosol sampled at midlake originated. If emissions data is available
for the designated source region one can then calculate the product of
emission factor arid the inverse of the dispersion factor for each filter set.
A determination can then be made as to whether the variability in midlake
concentration is highly correlated with this product. If so, computerized
back trajectories using the NCAR aircraft and NOAA SSMO data bases will be
used to attempt developing a mesoclimatology for the southern basin of Lake
Michigan. This mesoclimatology will be relatively easier to define for cold
season conditions. During the warm season the development of thermal internal
boundary layers and lake breezes makes the situation much more complicated.
However, with the goal being a mesoclimatology for concentration variability
on a temporal and spatial basis the task may be feasible if not reasonable.
-------
REFERENCES
Bolsenga, S. N.: 19TH, private communication, Great Lakes Environmental
Research Laboratory, Ann Arbor, Michigan.
Brtko, W. J. and Kabel, R. L.: 191 6, Water , Air and Soil Pollution, 6_, 71.
Cawse, P. A., I9lh: A Survey of Trace Elements in the U.K., 1972-73.
U. K. At. Energy Auth. Rep. AERE-R7669 . Hanvall, 95 pp.
Charlson, R. J., Ahlquist, N. C. and Selvidge, H. : 1969, Jour . of Air Pollut,
Control Assoc. , 10, 937.
Cohen, Y., Cocchio, W. and MacKay, D.: 1978, Environ. Sci. and Tech., 12,
553.
Duce, R. A.; Sea-Air Exchange Study, SEAREX, Univ. of Rhode Island,
Kingston, R. I.
Duce, R. A., et . al.: 1976, Trace Metals in the Marine Atmosphere, in
Marine Pollutant Transfer, Toronto; Heath, 77.
Edwards, II. W. and Wheat, II . G. : 1978, Environ. Sci. and Tech. , IP, 687.
Eisenreich, S. J., Enrolling, P. J. and Beeton, A. M. : 1977» 931.
Maclntyre, F. : 197^, Chemical Fractionation and Sea-Surface Microlayer
Processes in The Sea, _5_, Marine Chemistry, New York: Wiley,
Mer.ini, K. and Pozzi, G.: 1977, Environ. Pollut . , 12, 167.
Mesaaros, A.: 1977, Atmos. Environ. , 11, 1075
Murphy, T. J. and Doskey, P. V.: 1976, «J. Great Lakes Res , , 2, 60.
-------
-35-
NOAA: 1975, uummary of Synoptic Meteorological Obervations for Great Lakes
Areas, Vol. _3, Lake Michigan. National Oceanic and Atmospheric Adminis-
tration, National Climatic Center, Asheville, N. C.
Patterson, C., et. al.: 1976, Transport of Pollutant Lead to the Oceans
and Within Ocean Ecosystems, in Marine Pollutant Transfer, Toronto;
Heath, 23.
Rahn, K.: 1976, The Chemical Composition of the Atmospheric Aerosol, Grad.
School of Oceanography, Univ. of Rhode Island, Kingston, R. I., 265 pp.
Sehmel, G. A. and W. II. Hodgson: 1971!, "Predicted Dry Deposition Veloci-
ties," in Atmosphere-Surface Exchange of Particulate and Gaseous Pollu-
tants - 197^ Symposium. Battelle Pacific Northwest Laboratory, Richland,
Wash., pp. 399-^22.
Sehmel, G. A. and Sutter, S. I,.: 197*+, J_. Rech. Atmo s., III, 911.
Shepherd, J. G. : 19'fH, Atmospheric Environment, _8_, 69.
Sievering, H.: 1976, Water, Air and Soil Pollution, _5_> 309.
Skibin, D.: 1973, Water, Air and Soil Pollution, _2,
Steen, B.: 1977, Atmospheric Environment, 11, 623.
Stuiov, L. D., Murashkevich, F. I. and Fuchs, N. : 1978, J_. Aerosol Sci.,
Q 1
s_1 -L
Torrey, M.: 1976, Chemistry of Lake Michigan, Vol. 3 of Environmental
Status of the Lake Michigan Region, ANL/ES-140, Argonne Natl. Labs.,
Argonne, 111.
Wedepohl: 1971, Geochem. Acta, 10, 999.
Winchester, J. W. and Nifong, G. D.: 1971, Water, Air and Soil Pollution
I, 50. '
-------
-36-
Metal Blank Typical Sample
Backup Cascade Backup Cascade
Ca
Me
Al
ii
3a
Cu
ve
In
io
Fb
Ti
Zn
Table 1. ri'race Metal Concentrations for Typical Sample Filter and Blank Misco
Cellulose Filters, yg/1 in 25 ml ICAP Sample
6150
565
285
5605
12
25
850
30
28
130
^5
190
1970
220
125
3725
5
10
370
10
10
55
20
100
6875
1315
1110
6990
25
100
2810
160
55
1605
75
725
11250
1320
875
3070
35
70
1820
80
10
200
45
245
-------
-37-
H umber of Gets
C > L
Metal
30
1+7
51
33
33
1+3
1+5
46
20
50
Ca
Mg
Al
B
3a
Cu
Fe
Mn
Mo
Pb
Ti
Zn
1st Stage c Backup c
(Primary Aerosol) (Secondary Aerosol)
Total c
(1st, 2nd and
Backup stages )
.67
.125
.07
.121+
.003
.001
.130
.007
.000
.006
.001+
.008
.06
.065
.072
.120
.002
.006
.170
.011
.002
.127
.003
.01+6
1.03
.253
.201
.352
.007
.016
.1+06
.022
.003
.153
.009
.072
Table 2. 1977 Overall Average Trace Metal Concentrations,
yg/cu. m
-------
-38-
Trace Metal EF "Background"
(Al =1.0) Marine Aerosol EF (Duce, 1976)
Fb 3380 180
Zn 520 22
Cu lU5 11
Kn 9
MS 11 2
Ca 8
Fe U.5 1
Sa 2.5
B 7
Ti 1.5
Mo 1.3
Table 3. Overall Average Enrichment Factors for 1977 Aerosol Metal Samples
-------
-39-
Element Percent Primary Percent Second Stage Percent Secondary
(d > 1.0 ym) (d < o.7
Ca
Me
Al
B
Ba
Cu
Fe
Mn
Mo
Pb
Ti
Zn
65
50
35
35
1*3
8
32
29
7
1+
1+0
11
29
2U
29
31
28
56
26
22
13
13
31
26
6
26
36
31+
29
36
1+2
1*9
80
83
29
63
Table i.. Elemental Association with Primary/Secondary Aerosol at Midlake.
-------
-1*0-
Overall Average Wind Speed AT Bin
Element Loading, T/yr. Loading, T/yr. Loading, T/yr.
Pb
Zn
Cu
Mn
Mg
Ca
Fe
Ba
B
Ti
Mo
825
390
86
120
1365
5550
2190
38
1895
^9
16
710
3^0
39
89
1030
33^0
1650
27
1955
57
12
U65
230
35
59
815
2050
1090
16
1386
H5
8
Table 5- Estimated Annual Dry Deposition Loadings of Southern Lake Michigan
Basin
-------
-Ul-
Size Range Number
Channel
1
2
3
U
5
6
7
8
9
10
11
12
13
1U
15
0
.763
.960
1.157
1.353
1.550
1.7k6
l.9l*3
2.139
2.335
2.531
2.727
2.923
3.112
3.286
3.1*50
1
.335
.365
.395
.1*25
.1*55
.1*85
.515
.51*5
575
.605
.635
.663
.692
.721
.750
2
.182
.198
.211*
.230
.21*6
.262
.278
.291*
.310
.326
.31*2
.358
.37^
.390
.1*06
3
.098
.113
.121*
.131*
.11*2
.11*9
.156
.163
.170
.176
.183
.190
.198
.206
.216
Table 6. Geometric Mean Aerosol Diameter (urn) of ASAS Counting Channels
-------
-H2-
Specie
total P
NO'
sok~2
T'o'KT ^ -7
Primary
c
5
726
208
i r>77 na+ o /
Aerosol
(?)
(16)
(12)
( k)
2nd
c
8
1061+
336
P AT R p^r
Stage
(%)
(21+)
(18)
( 6)
Secondary
c
20
1+158
1+872
_3
Aerosol
(?)
(60)
(70)
(90)
Total c
32.1+
59^8
51+16
Association, per cent.
-------
Specie
total P
NO'
BO,,-"
Overall Average
Dry Loading
170
32000
29000
Wind Speed
Bin Calculation
160
22500
20500
AT Bin
Calculation
100
20500
18500
Table 8. P-N-S Annual Loadings by Atmospheric Dry Deposition.
-------
Specie
Ca
Mg
Al
Cu
Fe
Mn
Pb
Ti
Zn
total P
NCrr
Loadings , 10"kg/yr
Dry Deposition
,
Precipitation
(2)
Run-Off
(3)
2050
815
575
35
1090
59
H65
^5
230
100
20500
18500
560
20
950
53
88
59
52
1°00(1+)
U90000
13^000
13000
lltO/cV
U50(5)
250
100
100
18°(5)
U500(u)
Table 9- Annual Loading of Lake Michigan Comparison of Several Loading Routes,
1. This Work - Southern Basin only, 1977-
2. Gatz, 2nd ICMSE Conf. on G.L., 1975-
3. Wincnester & Nifong, W,A,SP, 1, 50+, 1975-
U. Murphy & Doskey, EPA - 600/3-75-005, 1975-
5. Bobbins, 2nd ICMSE Conf. on G. L., 1975.
-------
Metal Surface Sample Bulk Sample Ratio
Concentration, yg/1 Concentration, yg/1 Surface/Bulk
Ca
Mg
B
Ba
Co
Cu
Fe
Mn
Mo
Pb
Ti
Zn
30928
8872
55
10
1
1*
37
1
3
8
3
20
29806
8589
68
10
-
3
2k
3
6
3
16
l.OU
1.04
.81
1.0
1.33
1.5l»
1.0
1.33
1.0
1.25
Table 10. Comparison of Trace Metal Mean Concentrations in Surface and Bulk
Water Samples (September, 1977 samples).
-------
-U6-
Element
Al
Mg
Ca
Fe
Mn
Mo
TL
Cu
Zn
Pb
Average
Crustal
Conc(ppm)*
T8300
13900
28700
35000
690
1.0
I °
30
60
15
Average
Bulk Water
Cone(ppb)**
37.5
8728
29900
H6.3
2.6
3.3
2.4
it.8
17.7
7.5
Percent
Crustal
100
1U
7
8.1
0.1
0.3
0.2
Percent
Lake Derived
0
86
73
O.U
0.3
2.0
0.7
O.U
0.7
0.1
Percent
"Unexplained"
0
0
20
75-6
91.6
97.9
99
99.7
99.1
99.9
3
* Wedepohl (1971)
** 19 H' grand average of more than
bulk water samples
able II. Percentage of Trace Metals Identified as Crustal, Lake Derived
and "Unexplained"
-------
1 £73 O
^ ^ a?
I
OG
I
Fo
-t
On
ui
cn
y s
8
m
*
CD (go) cn
1
cn
i CD
CD
i
Fo
Correlation Coefficients among metal, P, NO , WS and Temp using
h2 Filter Sets' Data.
-------
-US-
Range of (T^ - T ), °C < -0.9 -0.8 to +0.9 1.0 to 2.1 2.8 to 8.1 > 8.2
ct S
F , fraction of year* 0.26 0.09 0.05 0.22 0.19
Number of sets, this bin 8 IT 9 35
V cm/s I.Oh .95 .^8 .21 .15
Table ISA. Temperature Stability Bins of Filter Set Data
Range of u, m/s < 2.2 2.3 to 3.5 3.6 to 5.2 5.3 to 8.5 > 8.6
F fraction of year* O.OU O.l6 0.15 0.29 0.2U
dumber of sets, this bin h 7 9 19 ^
V"._, cm/s .03 .20 .UU .98 I.^
i^l.
.u.-iie ^3B. Wind Speed Bins of Filter Set Data
-------
Deposition Velocity, Vp,cm/sec
p
b
p
b
0
a.
O
3
5 o
o '-,
o
5'
3
b
0
Ma
U!<
i
13
V\l
Cl
X
X
* <
/
/
<
\
O
-\
X
/
s
.''' X
/
X
X
X
X
X
K
X
X
gss
m
r
»
0
II »
-i aaw
ft N O
o
(D
ffi
(O°
3"
0) <
Q.»
|8
w ?:
O o'
3 3
Figure J. Tlieoretical Aerosol Deposition to a Water Surface as a
Function of Aerosol Size. Note the experimental data
points (x) measured by Sehmel and Sutter (l^jh) for
particles of density 1.5 gm cm
surface in a wind tunnel.
-3
depositing on a water
-------
VO
H
CO
in
\
2:
\
-
\
\
03
JH
^ c^ 05
M -H
Id I/I
U f-i >s
3 S 03
to C c
I O<
N
CO
O
\
Z
<
O
-------
-------
-------
Figure U. Example Midlake Sounding
9/56/30
12. « 12.6 12.8 13.0 13.2 13.d 13.6 13.8 W.O W.2
W.6 1«I.S
-------
.Percentage of Ambient Aerosol Collected (%)
Figure 5. Aerosol Collection Efficiency of Misco Filter Sierra
//I, //3 and Backup Cascade Combination.
-------
10
10
10
IO
E
o
o>
o
10
10
10
0.01
Figure 6a.
MAY 17-20,1977
AVERAGE
SLOPE-
O.I
1.0
10.0
r,jim
Plot of ASAS counts for May Outing.
-------
ro
'£
o
o>
o
0.01
Figure 6b.
Plot of ASAS counts for June Outing.
-------
ro
o
o»
O
AUGUST 14-19, 1977
AVERAGE SLOPE*
WITH r*6.97
0.01
10.0
Figure 6c,
Plot of ASAS counts for August Outing.
-------
10
o
o>
o
AVERAG
WITH r*
Figure 6d.
Plot of ASAS counts for September Outing.
-------
10000
J>C
o
o:
o
i
LU
^
X
O
QC
LJ
3000
1000
300
IOO
30
10
Mid Li
Mass
Heavil;
by Se
Aeros<
GRAN
COEF
CONC
ike Aerosol
Cone.
f Dominate<
juiidury
D AVERAGE
"ICIENT OF
ENTRATIOIV
v48 50f
**\ L_- w<
35 -J
18
/
Z_
! CORREL
Pb and A
= 0.67
V41 Q29
v54
51 /
3 . Vv47 /
1Q S (
'
-
D27
24
n
Mid Lake
About Eqi
by Primary
&TION
/2'«.13
^ 'Jp
326 D*15 (
1 J
D25
8
»6
5
)3
10
20 «14
Aerosol Mqss Cone.
ally Contributed to
and Secondary
I.O
3.0
IO.O
30.0
IOO.
300.
IOOO.
ALUMINUM, ng/m4
O APRIL I977 D AUGUST I977
MAYI977 v SEPTEMBER I977
JUNE I977
Figure 7-
Scatter-dlagram for Pb.
-------
100
o
o
10
o
UJ
o
Ql
UJ
F
1.0
O.I
«.
GRAND
COEFF1
CORRE
Scatter S
Does Not
Source C
AVERAGE
CIENT OF
_ATION=0
uggests £
oil Derived
PredomlnaTe. A sec
f Ti Musi Be Presei
37
36
L?
3
29
40
4*7
31
51
24
) »53
54
41
TION
Aerosol
ond Varia
it!
34
25 *28
55
26 "°
22 .13
21
II
/
4
1
bie
56.20 .5
8 «7
*S .6
14
LO
3.0
10.0 30.0
100. 300.
1000.
ALUMINUM, ng/nrr
'igure
Scatter-diagram for Ti.
-------
Neutral Drag Coefficient CD(5m)
o
o
o
o
NJ
ft
3
3
3
10
o
Figure 9. Neutral Drag Coefficient at 5 m Height
-------
o
«
O
O
w
O
a.
JO
O
O
O.
O
E
Q
rH ^
oj 0)
rH -P
-P
o3
rl -P
Q C
(U
O -H
H O
P -H
cS CM
ft CM
n3 0)
H O
Q O
O
rH
00
'juapi ^303
------- |