EPA-600/4-77-010
February 1977
MESOSCALE AIR POLLUTION TRANSPORT
IN SOUTHEAST WISCONSIN
by
Walter A. Lyons
University of Wisconsin-Milwaukee
Milwaukee, Wisconsin 53201
Grant R-800873
Project Officer
Paul A. Humphrey
Environmental Sciences Research Laboratory
Research Triangle Park, North Carolina 27711
Environmental Sciences Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into five series. These five broad
categories were established to facilitate further development and application of
environmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The five series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
This report has been assigned to the ENVIRONMENTAL MONITORING series.
This series describes research conducted to develop new or improved methods
and instrumentation for the identification and quantification of environmental
pollutants at the lowest conceivably significant concentrations. It also includes
studies to determine the ambient concentrations of pollutants in the environment
and/or the variance of pollutants as a function of time or meteorological factors.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
DISCLAIMER
This report has been reviewed by the Environmental
Sciences Research Laboratory, U.S. Environmental Protection
Agency, and approved for publication. Approval does not
signify that the contents necessarily reflect the views and
policies of the U.S. Environmental Protection Agency, nor
does mention of trade names or commercial products constitute
endorsement or recommendation for use.
-------
PREFACE
During the period 1970-1976, the University of Wisconsin-
Milwaukee's Air Pollution Analysis Laboratory (College of Engi-
neering and Applied Science) engaged in extensive studies on the
mesometeorology of the Great Lakes. In spite of the fact that
these massive bodies of water are convenient natural laborator-
ies for the study of mesoscale air mass transformation, the body
of scientific knowledge relating to such phenomena as lake
breezes, snow squalls, etc., was surprisingly small. The extreme
climatic and weather fluctuations generated by the lakes are of
great importance to the region. The Great Lakes themselves cover
an area of 246,000 km2. Over 15% of the total U.S. population,
and more than 25% of the nation's industrial capacity, is loca-
ted near or on their shorelines.
The U.S. Environmental Protection Agency (1972) funded a
large program to further investigate the effects of lake induced
meteorological systems upon air pollution diffusion and transport
using the western shore of Lake Michigan (near Milwaukee) as the
test site. This program was then integrated into a much larger
effort studying the Great Lakes' mesometeorology as a whole. The
lake effects on the atmosphere in general must be first under-
stood in greater detail before it is possible to better define
the behavior of air pollutants. The combined research programs
11
-------
sponsored by the National Aeronautics -and Space Administration,
the National Science Foundation, U.S. E.P.A., Office of Water
Resources Research, the UWM Center for Great Lakes Studies, and
several utilities, generated over three dozen papers, project
reports, theses and films (totally almost 2000 pages of text).
Obviously summarization, even of only that portion pertinent to
air pollution becomes very difficult if one wishes to keep the
document to readible dimensions. This report acts more as an
"executive summary" of the pollution related aspects of the pro-
gram rather than a complete stand-alone final report.The reader
will be directed to the appropriate documents for details that
may be of interest (see Appendix).
A reasonably complete summary of lake effects upon air pol-
lution was published under the auspices of the American Meteoro-
logical Society (Lyons, 1975) in "Lectures on Air Pollution and
Environmental Impact Analysis." Chapter 5 of this book, "Turbu-
lent Diffusion and Pollutant Transport in Shoreline Environments"
is also a fairly extensive review of what is known about Great
Lakes mesometeorology.
This paper has been written to highlight the important
findings of six years of Great Lakes air pollution related
research. Those more recent developments not covered in the
AMS review paper will be presented in somewhat greater detail.
It is felt that the results reported herein will be used
not only by those responsible for environmental planning on the
Great Lakes but also generalized for use along the entire
26,000 km of the United States coastline.
IV
-------
ABSTRACT
This research program comprised a comprehensive study of
mesoscale meteorological regimes on the western shore of Lake
Michigan and their effect upon air pollution dispersion and
transport. It is felt that the results are applicable in a
generic way to other mid-latitude coastal zones.
Continuous fumigation from elevated sources in shoreline
zones during stable daytime onshore flow was intensively inves-
tigated by a large scale field program. A model was proposed,
constructed, validated and calibrated. It was shown that the
fumigation spot, while causing very high surface S02 concentrations,
was so highly mobile as to generally reduce dosages below the
three-hour standard Cat least for the plants studied).
Instrumented aircraft profiles the structure of the
thermal internal boundary layer (using 1/3 measurements). An
acoustic sounder at 1 km from the Milwaukee shoreline revealed
drastically reduced daytime mixing depths.
An intensive case study of a lake breeze was performed.
Data were used as input to a Kinematic Diagnostic Model (KDM)
which simulated mesoscale trajectories for pollutants released
within the coastal zone.
Both mesoscale and synoptic scale transport of photo-
chemical oxidants were found to be a significant problem in the
Milwaukee area.
-------
CONTENTS
Preface j .,• -j
Abstract v
Figures viii
Acknowledgements xvi
1. Introduction 1
2. Conclusions and Recommendations 8
3. Data Collection and Field Experiments 11
4. Fumigation and the Thermal Internal
Bounday Layer 26
5. The GLUMP Regional Fumigation Model 51
6. Model Calibration and Validation 62
7. Studies of Elevated Point Source Plume
Fumigation 76
8. Urban Scale Pollution Patterns Extension
from Point to Regional Modeling . . 98
9. Acoustic Sounder Derived Mixing Depths 107
10. Lake Breeze Structure 123
11. The Kinematic Diagnostic Model 146
12. Results of the KDM 159
13. Long Range Pollution Transport 175
References 208
Appendices
Publications 215
Theses 218
vii
-------
FIGURES
Number page
1-1 Photograph taken during Gemini manned orbital
mission of a coastal region showing fumi-
gation from numerous points 6
1-2 Structure and details of a typical lake breeze 7
3-1 Milwaukee area field program network 18
3-2 Aerial view of Jack Benny Jr. High School base
station in Waukegan 19
3-3 Waukegan, Illinois field program network ... 20
3-4 View from top of UWM Science Complex building,
showing instrumentation used 21
3-5 Cessna 182 instrumentation and data acquisition
system 22
3-6 Tetroon with pilot balloons over Lake Michigan 23
3-7 Aerial view showing various launch and track-
ing sites for tetroons and pibals around
Milwaukee 24
3-8 Specially modified NWS-VIZ radiosonde, a
RASOT, used with tetroon system 25
4-1 View of Waukegan Power Plant plume from
Waukegan Airport 28 June 1974 38
4-2 Vertical and Horizontal Plume Geometry .... 39
4-3 Wind hodograph at 271 m AGL, 0900-1600 CST,
16 July 1973 along Lake Michigan shoreline
at Milwaukee, Wisconsin 40
4-4 Estimated profiles of the top of the thermal
internal boundary layer (TIBL) used in
calculations 41
4-5 Calculated surface S09 concentrations (in
PPM) at 1100 CST / 42
VI 1 1
-------
4-6 Time history diagram of predicted fumigation
patterns from three power plant plumes
(0900-1600 CST, 16 July 1973) 43
4-7. Predicted time history of S02 concentra-
tions at receptor point receiving
highest instantaneous concentration 44
4-8 Surface temperature time versus distance
from lakeshore (2 °F isotherm), Waukegan,
27 June 1974 45
4-9 Potential temperature cross section fol-
lowing an east-west traverse of Lake
Michigan shoreline at Oak Creek, Wis. 46
4-10 Flight track turbulence reports taken
during the same time period as Figure
4-9 46
4-11 Aircraft measurement of eddy dissipation
rate on an east-west traverse near the
Illinois-Wisconsin border 47
4-12 Ten selected TIBL tops monitored by air-
craft during mid-afternoon during 1974
field program 48
4-13 Linear regression analysis between TIBL dep
depth at 5 km fetch and initial shore-
line potential lapse 49
4-14 Plots of £V3 t S02, and 0.3-1.3^m
aerosol concentrations for top of
mixed layer 50
5-1 Portion of 1970 emission inventory of
S02 sources in southeastern Wisconsin
showing Milwaukee County major point
sources 58
5-2 Horizontal schematic of GLUMP model 59
5-3 Representation of vertical structure of
the GLUMP model 60
5-4 Printer plot of a typical TIBL top profile
with an east-southeasterly flow 61
IX
-------
6-1 East-west aircraft S02 traverse, roughly down
mean plume centerline path, 1547 CDT, 8
August 1974 70
6-2 Helicopter $03 sounding, 0.9 km downwind of
power plant, 1053-1058 CDT, 8 August 1974 ... 71
6-3 Same as 6-2 except at 6.3 km downwind, 1116-
1120 CDT 71
6-4 Same as 6-2 except at 11.4 km, 1106-1110 CDT ... 71
6-5 Composite of the 2.3, 4.5, and 11.8 km S02
traverses on 8 August 1974 72
6-6 S02 record from the ERT fixed site monitor
located 5.3 km from the Oak Creek power
Plant 73
6-7 Ratio of ten minute average to instantan-
eous peak S02 values measured by ERT
mobile van 74
6-8 Mass balance for six selected plumes 75
7-1 Hourly pibal ascents at the Milwaukee shore-
line, 8 August 1974 87
7-2 Hourly hodograph of wind at effective stack
height (H) for OCPP stack 4 88
7-3 Horizontal $62 patterns measured in a saw
tooth trajectory flown west of OCPP at
150 m AGL between 1358 and 1431 CDT 89
7-4a GLUMP model predictions of surface S02 con-
centrations at 1400 CDT, 8 August 1974 90
7-4b GLUMP model prediction of surface S0? con-
centrations at 1600 CDT, 8 August T974 91
7-5 Summation of the predicted maximum 7r for
ththe plume from OCPP Stack 1 92
7-6 Same as Figure 7-5, but for OCPP Stack 4 92
7-7 Time history of predicted surface S0? G
values for the sum of all four OCPP
stacks 93
-------
7-8 The areas of the individual fumigation spots
on 8 August 1974 94
7-9 The speed at which the fumigation 'spots moved
on 8 August 1974 95
7-10 G"z Plotted as a function of
-------
9-4 Mixing depth vs. time (offshore flow, clear
and scattered sky) 118
9-5 Mixing depth vs. time (onshore flow, clear
and scattered sky) ng
9-6 Mixing depth wind rose with adjusted data
(clear and scattered sky) 120
9-7 Acoustic sounder trace with turbulence data
(4 September 1974) 121
9-8 Acoustic sounder trace with turbulence data
(16 July 1974) 122
10-1 A depiction of the Classical Land Breeze at
about 0500 LSI 134
10-2 A depiction of the Classical Lake or Sea
Breeze, fully mature, at about 1500 LSI 134
10-3 Summary of the observed characteristics of
a well developed lake breeze during mid-
afternoon 135
10-4 Summary of the observed characteristics of
a land breeze near dawn 135
10-5 The 1200 CDT SMS-1 satellite photograph on
which is superimposed the 1200 CDT surface
isobars and frontal positions 136
10-6 Lake water isotherms (°C) compiled from th
three sources 137
10-7 Summary diagram of the acoustic sounder
return heights and the wind velocities
at each hour during 4 September 1974 138
10-8 Photograph looking south along Lake Michigan
shore, 0830 CDT, 4 September 1974 138
10-9 Hourly lake breeze wind shift positions on
4 September 1974 139
10-10 The 1000 CDT and 1100 CDT pibal soundings
on 4 September 1974 140
10-11 Composite photograph of 3 frames from Land-
sat-1 of the western shoreline of Lake
Michigan at 1103 CDT, 4 September 1974 141
XI 1
-------
10-12 Plan view of the trajectories of 3 tetroons
launched on 4 September 1974 142
10-13 Cross sectional view of the trajectories of
the three tetroons 143
10-14 Temperature (dry bulb) and humidity profiles
from the RASOT package attached to the 1048
CDT tetroon 143
10-15 Measurements of eddy dissipation rate on a
traverse along Wisconsin Avenue 144
10-16 Pattern of distribution of the 0.3-1.3/tm
range aerosols between 1351 and 1541 CDT . . . 144
10-17 Pattern of distribution of the 7-9/tm
range aerosols between 1351 and 1541 CDT . . . 145
11-1 Flow diagram of the Kinematic Diagnostic
Model 155
11-2 Detailed flow diagram of Kinematic Diag-
nostic Model 156
11-3 Detailed flow diagram of Kinematic Diag-
nostic Model, continued 157
11-4 Detailed flow diagram of Kinematic Diag-
nostic Model, continued 158
12-1 The xz plot of the 1048 CDT tetroon super-
imposed upon the 1130 CDT 2-dimensional
uw wind vectors 157
12-2 Plots of particle trajectories from a six
point "line" source orientated SSW to NNE . . .168
12-3 Plots of particle trajectories from a line
source with a release time of 1200 CDT .... 169
12-4 Plots of particle trajectories from a multi-
stack source on the shore with release heights
of 20, 50, 100, 200, and 300 m 170
12-5 Particle trajectories from a multistack
source with a release time of 0930 CDT .... 171
12-6 The xy projection of the positions of a
series of particles at 1800 CDT 172
.XI 1 1
-------
12-7 A histogram of the relative distribution
of particles according to size groups
across the lake shoreline 173
12-8 A 3-D plot of a simulated aerial 'burst.' 174
13-1 Landsat-1 , Band 5 image of Chicago-Gary
area, 1003 LSI, 14 October 1973 195
13-2 Landsat-1 image, Band 6 over southern
Lake Michigan, 1003 LSI, 24 November
1972. 196
13-3 Computer processed Landsat-1 digital data,
small segment of frame taken along
southern Lake Michigan shoreline 197
13-4 Landsat-1 image, Band 6, of 20 August
1972 showing a portion of Lake Ontario
and the eastern end of Lake Erie 198
13-5 Aircraft measurements of SO? (pphm) pro-
filed in an east-west path along the
Wisconsin-Illinois state border 199
13-6 Same as 13-5, except for concentrations of
aerisols in the 0.3 to 1 . 3 *m size range 199
13-7 Data from six monitoring sites from 15
July through 31 August 1973 200
13-8 Isopleths of percentage of sensitive plants
experiencing photochemical oxidant damage
in Milwaukee during summer of 1972 201
13-9 Schematic showing the mechanism by which
a lake breeze causes elevated ozone
levels in a narrow band parallel to
the shore but several kilometers inland 202
13-10 Hypothetical trajectories of air parcels
(and pollutants) along the western
shoreline of Lake Michigan during a
lake breeze event 203
13-11 Pollution wind rose for Poynette for days
on which ozone exceeded episode alert
levels in summer of 1973 204
xiv
-------
13-12a Landsat-1 image, Band 4, 0945 LSI, 23 March
1973, of the eastern Lake Ontario region
on a cloud-free day with low atmospheric
turbidity 205
13-12b Identical geographic area as Figure 13-12a,
except on 1 September 1973, with highly
polluted atmosphere associated with
synoptic scale air stagnation sulfate
episode 205
13-13 Portion of an SMS-1 visible image taken
1445 GMT, 30 June 1975. 206
13-14 Surface synoptic chart, 1200 GMT, 30 June
1975, with contoured shadings representing
areas of reduced visibility 207
13-15 Location of the 1020 mb isobar on six con-
secutive synoptic charts during the
period 25 June-30 June 1975 207
-------
ACKNOWLEDGEMENTS
Though the bulk of the results reported herein were the
direct result of funding fromthe U.S. Environmental Protection
Agency under Grant No. R-800873, it must be remembered that this
research formed a major subset of an intensive and comprehensive
study of Great Lakes mesometeorological systems. Thus the faci-
lities, data and techniques used and developed were an integrated
whole. For this reason the support of other organizations and
agencies must be gratefully acknowledged, including: National
Aeronautics and Space Administration, the National Science
Foundation, the Graduate School of the University of Wisconsin-
Milwaukee, and the UWM Center for Great Lakes Studies, and the
UWM Social Science Research Facility, the National Center for
Atmospheric Research, Commonwealth Edison Company, and Wisconsin
Electric Power Company. This final report was produced using the
facilities of COMPUMET/Meteorological and Environmental Services,
Minneapolis, Minnesota.
Graduate students at the Air Pollution Analysis Laboratory
at the University of Wisconsin-Milwaukee, deserve special thanks
for making significant contributions to both the data gathering
and analysis phases. In particular, Dr. Cecil S. Keen (now,
University of Cape Town) served as field coordinator and devel-
oper of the Kinematic Diagnostic Program. Jerome Schuh super-
xvi
-------
vised general software development, and in particular, the GLUMP
Diffusion Model. John C. Dooley and Kenneth R. Rizzo labored
long hours in equipment maintenance and data abstraction. Others
including Eugene M. Rubin, Norman Knox, and as many as twenty
additional undergraduate and graduate students played key roles
in this effort.
In particular,the assistance of Professor Henry S. Cole,
University of Wisconsin-Parkside, is hereby acknowledged for his
role as co-principal investigator.
Other agencies, organizations and individuals who were most
helpful in various phases of data collection include: Wisconsin
Department of Natural Resources, Milwaukee County Department of
Air Pollution Control, First Wisconsin Center, Towne Realty,
Miller's Brewery, Johnston Candy Co., Milwaukee Water Works,
City of Waukegan, Illinois Toll Highway Authority, Waukesha
County Airport, and the Milwaukee County Institutions.
Technical manuscript preparation was admirably handled by
Kathleen Walker.
xvi i
-------
SECTION 1
INTRODUCTION
GREAT LAKES MESOMETEOROLOGY
Covering almost a quarter of a million square kilometers,
and with a shoreline extending 5,600 km, the Great Lakes repre-
sent the largest reservoir of fresh water in the world. Their
thermal mass, represented by a volume of 22,910 km3 makes them
a tremendous heat source or sink, depending on the temperature
of the overlying air mass. Since the region possesses a dis-
tinctly continental climate with recorded temperature extremes
ranging from +47C to -53C, the Lakes are the generator of dra-
matic mesoscale air mass transformations. Thus the meteorology
of Great Lakes shoreline regions provides an excellent natural
laboratory for studying pollution diffusion and transport under
conditions of transitional stability and complex mesoscale cir-
culation regimes. But this opportunity is of more than academic
importance inasmuch as fully 25% of the U.S. industrial capacity
lies in areas affected. The Lake Michigan shoreline alone counts
over 25 power plants.
During the winter (cold season), frigid continental air
masses stream southwards over the largely unfrozen waters. Exten-
sive observational (Lenschow, 1973) and theoretical studies
(Lavoie, 1972) of lake snow squalls exist. Lyons and Pease (1972)
1
-------
presented pictures of "steam devils" over Lake Michigan as -25°C
air flowed over + 2°C water, indicative of the extreme instability
that can be found over and near the lakes during such events.
This of course would have significant impact on the diffusion
of effluents released into such a highly modified boundary layer.
The summer (warm season) has not been as extensively re-
searched since the phenomena produced by the lakes are far less
spectacular. In fact often the lakes eliminate rather than cause
phenomena, such as total suppression of convective cloud systems
over the water and downwind shoreline (Lyons, 1966). Bellaire
(1965) was among the first to document the extreme but shallow
(less than 150 m) conduction inversions which can form as warm
tropical air advects over the cold water, which in the center of
Lake Michigan can remain as cold as 4°C well into June. Lyons
(1971) developed a model for such conduction inversions. Since
low-level thermal stabilities are equivalent to those found
within the polar night they would allow effluents to travel long
distances virtually undiluted. As the stabilized air mass ad-
vects inland over the downwind shore on a sunny day, intense
remodification occurs. Herkoff (1969) used surface temperature
patterns obtained via an instrumented automobile to show that
ground heating restores the temperature to typical inland values
often within 20 km. Bierly (1963) studied the vertical aspect
of this thermal remodification on the eastern Lake Michigan
shoreline. The top of this modified layer, the thermal internal
boundary layer (TIBL), marks the upward extent of newly generated
2
-------
penetrative convection. Collins (1971) and Van der Hoven (1967),
along with Hewson, et al. (1963) were among the first to point
out that the generation of a TIBL could lead to the continuous
(dynamic) fumigation of plumes released at an elevated level into
the stable onshore flowing air (see Figure 1-1). Lyons and Cole
(1973) further refined the concept and proposed a Gaussian type
model for the phenomenon. The implications were that since
unusually high concentrations of pollutants would be mixed to the
surface for many consecutive hours, continuous fumigation poten-
tially represented the most serious (and frequent) possible
cause of exceedences of three-hour federal standards for such
pollutants as SO .
The early studies of fumigation presumed a simple gradient
onshore flow. The addition of a shoreline lake/sea breeze circu-
lation compounds the problem many orders of magnitude. This
diurnal mesoscale wind regime has been known since the time of
the Greeks (Baralt and Brown, 1965) and appears in American sci-
entific literature in almost Revolutionary times (Endicott ,1799).
It is only in the last decade or so that its rather profound
impact upon shoreline air quality has begun to be recognized. It
is still possible to find occasional state implementation docu-
ments or air quality advisories in which the presumption is made
that a lake breeze will result in the onshore advection of clean
air and thus the alleviation of a pollution episode. Lyons and
Olsson (1972) used mesoscale data analyses and photography to
show how, that in spite of apparently good ventilation, shoreline
3
-------
lake breezes could cause serious degradation of urban shoreline
air quality. An addition to continuous fumigation and generally
severely limited mixing depths within roughly 20 km of the shore,
it appeared that pollutants could recirculate within the lake
breeze cell. Indeed, in Chicago and Milwaukee it was found that
the highest values of most pollutants tended to occur during
those months when lake breezes were most frequent. Figure 1-2
is a schematic of the lake breeze structure and its possible
effects upon air pollutants. It becomes immediately apparent
that conventional air quality models, which assume a steady-state
homogeneous wind field, would fail in such cases. Furthermore,
aerosols of various sizes could be expected to travel along dif-
ferent paths due to size sorting effects (Lyons and Keen, 1976a).
And while Figure 1-2 is necessarily a two-dimensional represen-
tation, as shown by the Chicago tetroon experiments of 1967
(Lyons and Olsson, 1973), the complex transport phenomena within
lake breezes are eminently three-dimensional.
Also apparent from numerous studies (Lyons, 1975a) was the
fact that conventionally gathered meteorological data were totally
inadequate to describe the atmosphere in coastal environments.
Radiosonde derived mean mixing depth climatologies such as those
generated by Holzworth (1967, 1972) had little validity within
20 km of a coastline, precisely were most of the Great Lakes air
pollution sources (and receptors) were located. The need for
improved mesoscale climatologies of wind, stability and mixing
depths became imperative.
4
-------
The several projects undertaken by the UWM Air Pollution
Analysis Laboratory were aimed at arriving at some first solutions
to the problems discussed above.
With regards to fumigation, it was necessary to actually
test and validate and calibrate the model proposed by Lyons and
Cole (1973). It was necessary to first expand the model to a
multi-source configuration, out of which grew the GLUMP Regional
Fumigation Model. Field studies in 1974 essentially validated
the basic geometry, at least for high stack shoreline plumes.
Urban scale fumigation was found to be indeed important in
causing higher air pollution levels. Studies of the,mixing
depth profiles revealed the TIBL formation to be complex and its
height and shape extremely significant in determining surface
pollution concentrations. It was furthermore found that the
"fumigation" spot exhibited a high degree of mobility, making
the calculation of total dosages received at any one point a
di fficult problem.
The lake breeze structure was further studied and new de-
tails were revealed. Most importantly, it was possible to acquire
•
extensive data showing the concept of recirculation and complex
transport phenomena to be correct. A kinematic diagnostic model
was developed which shed further light on how mesoscale shoreline
flow regimes effect pollutants emitted from a variety of sources.
The following sections will detail the most significant
results from these studies.
-------
Figure 1 -1 . Photograph taken during Gemini manned orbital mis-
sion of a coastal region during midday.A fully developed boun-
dary layer is present inland several tens of kilometers as
marked by cumulus convection. However smoke released from
numerous points undoubtedly experiences fumigation as it passes
inland through a zone of rapidly changing atmospheric stabili-
ties .
-------
STRUCTURE AND DETAILS OF A TYPICAL LAKE BREEZE
STREAMLINES CALCULATED FROM PIBAL
POTENTIAL TEMPERATURE
FYPICAL SMOKE AND CLOUD PATTERNS
\ INLAND
CUMULUS
"•> IADVECT OVER
FRONT &
DISSIPATE
SMOKE LAYERS ALOFT
OVER LAKE
SUSPECTED TRAJECTORIES
GIANT PARTlCULATL-L2^g
Figure 1-2. (a) Typical streamline patter
lake breeze cell during mid-afternoon. T h
the top boundary of the inflow. View is 1
lake on the left. (b) Schematic isopleths
with packing indicating synoptic and mesos
temperature soundings over land and lake a
dry adiabatic lapse. (c) General smoke pa
the fumigating plume from an elevated shor
(d) Hypothesized trajectories of both fine
aerosols emitted in the area.
ns in a well developed
e heavy line represents
coking south, with the
of potential temperature
cale inversions. Typical
re shown compared to the
tterns found, including
eli ne poi nt source.
and large particulate
-------
SECTION 2
CONCLUSIONS AND RECOMMENDATIONS
While considerable information was generated regarding lake-
induced effects upon air pollution transport and diffusion in
coastal zones, if anything, the phenomena were shown to be even
more complex and multi-faceted than originally anticipated.
An extensive field program on the western shore of Lake
Michigan during the summer of 1974 essentially validated the fumi-
gation model proposed by Lyons and Cole (1973). The three regime
plume geometry was indeed found using helicopter and aircraft
measurements. These data also showed the necessity of employing
split-simga modeling inasmuch as plume lateral and vertical spreads
were often as much as three Pasqui11-Gifford stability classes
apart (the effect of wind shear the primary factor in causing
enhanced lateral spread). The fumigation phenomenon is highly
dependent on the shape and depth of the Thermal Internal Boundary
Layer (TIBL). Aircraft profiling of the eddy dissipation rate
(€ ' ) proved most effective in gathering TIBL data, a highly
complex, temporally and spatially, variable phenomenon. Actual
power plant plume S02 measurements showed high surface concentra-
tions. However, the fumigation spots moved rapidly about the
landscape at speeds upward of 10 kmph, reducing dosages at any
given point significantly. Extreme vertical direction wind shears
frequently caused plumes to travel along separate axes. Any fumi-
gation model for multi-stack power plants must take account of
this effect or gross overestimates of concentration/dosages could
8
-------
occur. For ascertaining "worst case" surface concentrations of
fumigating power plant plumes, mobile monitoring (vans, aircraft,
helicopter) is superior to the establishment of a fixed ground
network. The size of the fumigation spot is too small, perhaps
making the required network density prohibitively expensive for
adequate data return in a reasonable amount of time.
The GLUMP Fumigation Model was calibrated for multi plume
power plants with a range dependent calibration factor. The model
predicts within a factor of two in the cases studied, even before
the calibration is applied. The model was extended to cover re-
gional patterns (Milwaukee County) and shows the fumigation regime
significant enhances ground level pollution concentrations. Sur-
face data and emission inventories were unfortunately insufficient
to allow calibration of the model in the regional mode.
Studies of lake meteorology showed that even during supposed-
ly "steady state" onshore gradient flows, there are complicated
wind patterns, including the development of low level jet streams
associated with intense inversion layers. The strong wind direc-
tion shears at plume 1evel,high!ight the essential invalidity of surface
or even tower wind data for modeling individual cases. It is neces-
sary to know exactly how the wind is behaving at the height of each
individual plume through time.
The acoustic sounder is highly useful in showing the struc-
ture of lakeshore environment. For example, five straight days
of severely reduced mixing depths occurred during one period of
onshore flow. Conventional data would have suggested excellent
dispersion conditions, when in fact quite the opposite was true.
9
-------
A climatology of mixing depths at 1 km inland found the mean hourly
daytime mixing depth (clear and scattered skies) to be reduced
2.73 times during onshore as opposed to offshore flow during summer
(656 vs 240 m). Mid-day mixing depths could be suppressed by more
than a factor of five.
An extensive analysis of a lake breeze illustrated via air-
craft and tetroon data the recirculatory nature of lake breeze
circulations. The pibal data obtained was used as input into the
Kinematic Diagnostic Model (KDM), designed to use real rather than
numerically generated wind data to simulate lake breeze transport.
It proved a most useful experimental technique, confirming pol-
lutant recirculation, aerosol size sorting, the role of the con-
vergence zone in redistributing pollutants, and the essential
futility of using surface data in predicting the spread of hazard-
ous materials and pollutants in coastal zones.
Satellite data were shown highly useful in monitoring meso-
scale, regional and synoptic scale transport. Individual plumes
were detected for more than 150 km over Lake Michigan by Landsat,
whereas SMS imaged a major sulfate haze aerosol episode over the
central U.S. Also lake effects complicate even further mesoscale
and synoptic scale transport. A model was proposed to explain
the inland band of elevated ozone levels running parallel to the
shoreline. The Chicago metropolitan area was shown to be a major
contributor to the high oxidant levels recorded in southeastern
Wisconsin. Also, aircraft monitoring of the Chicago urban plume
revealed interstate transport of 25 tons per hour of S0? from
Illinois into Wisconsin on one occasion. With regards to photo-
10
-------
chemical oxidants and sulfates, the concept of Air Quality Control
Regions clearly has to be severely modified or abandoned altogether.
In general, most required meteorological data taking in coastal
environments is highly unrelated to the true needs of describing
the atmosphere into which effluents are released. Needless to say,
the inapplicability of most existing short term prediction models
in coastal zones is painfully evident!
lOa
-------
SECTION 3
DATA COLLECTION AND FIELD EXPERIMENTS
CHRONOLOGY
Each summer from 1970 through 1974 inclusive, some field
data gathering took place in the Milwaukee area in direct or in-
direct support of this project.
The 1970 program was limited to a few aircraft photo-
reconnaissance flights inspecting the plumes of local sources
under differing regimes. These however quickly identified the
continuous fumigation and plume trapping potentials of the Lake
Michigan shoreline environment. The result was the proposed model
of Lyons and Cole (1973).
The summer of 1971 saw a two week field program in the
first half of August. An instrumented Queen Air was supplied by
the National Center for Atmospheric Research. The prime objec-
tive was aerosol monitoring in conjunction with limited pibal
wind measurements of lake breeze structure. Only limited success
resulted due to highly unfavorable weather conditions along with
malfunctions of both sensors and recording devices on the air-
craft. Valuable qualitative data was obtained however with
regards to the structure of the TIBL during stable onshore flow
situations. In fact, one of the best developed TIBLs ever moni-r-
tored (5 August 1971) was mapped (Lyons, 1975a). In addition, a
rough correlation between aircraft sensed turbulence and the shape
and depth of the thermal internal boundary layer was developed.
11
-------
Also noted were some definite relationships between surface ozone
levels and lake breeze frontal passages (Lyons and Cole, 1976).
The 1972 season saw relatively limited field data gather-
Ing other than pibal wind soundings for climatological purposes.
The summer of 1973 was dedicated to the testing of new
techniques for data gathering and the development of instrumen-
tation packages. An airborne meteorological/air quality monitor-
ing and data logging package was designed and test flown. The
Wisconsin Department of Natural Resources (DMR) graciously al-
lowed the use of a twin engine Cessna 337 for this task. There
was limited data acquisition, in part due to relatively unevent-
ful weather, but also due to several electronic component failures
within the data logging system that were inconspicuous only until
post-program data analysis. Several test flights of optically-
tracked tetroons were made in order to establish firm procedures
for weigh-off, balloon release, and tracking via double theodo-
lites. Software was developed and tested for analyses of tetroon
and pibal balloon data. Field sites were selected for pibal
releases. Attempts at offshore releases of pibals and tetroons
were made but found to be operationally difficult to coordinate.
Pollution photography techniques were improved upon. Data
gathered during mid-July proved most useful in the analysis of
an ozone episode along the western shore of the lake (Lyons and
Cole, 1976). Also, a more complete model study of power plant
fumigation potential was carried out using field data gathered
on 16 July 1973 (Lyons and Dooley, 1974). The most important
result of the 1973 effort was to prepare for a highly successful
12
-------
summer of 1974.
THE 1974 SUMMER FIELD PROGRAMS
With additional sponsorship of Wisconsin Electric Power, a
three month long field program was mounted to study both the
mechanism of continuous fumigation and lake breeze phenomena along
the western shore of Lake Michigan. The first phase of the pro-
ject was conducted in and around Waukegan, Illinois from 3 June
through 28 June 1974. The prime target of study was the four
stack shoreline Waukegan Power Plant (WPP) on the shoreline. The
project moved to the Milwaukee area from 16 July through 4 Sep-
tember 1974 where the four stack shoreline Oak Creek Power Plant
plume was intensively monitored, along with the general urban
pollution patterns. A total of 21 days of intensive field data
gathering were undertaken. As many as 37 individuals were em-
ployed in various phases of the data taking and processing. An
extensive summary of the equipment and data processing procedures
used appears in Lyons, et al . (1974). However, a brief summary
is presented here.
The Milwaukee Area
Figure 3-1 is a map of the various sites for the second
end more complete stage of the field program. In addition, within
this area were nine DNR air quality monitoring stations (not
shown) which provided additional data on such pollutants as S02.
CO, HC, NOX and oxidants, as well as partial meteorological data
(wind and temperature). In addition to maintaining a file of all
National Weather Service (NWS) hourly and synoptic teletype,
13
-------
transmissions, NAFAX facsimile charts, and SMS, DMSP, and perti-
nent Landsat satellite images were archived. The base of opera-
tions in Milwaukee was the 12th floor Air Pollution Analysis
Laboratory at UWM. On the roof top directly above was constructed
an observation platform (Figure 3-2) for siting the radiosonde,
FM communications, and acoustic sounder antennae, optical theo-
dolites for tetroon and pibal tracking, and camera systems.
The Waukegan Area
The first phase of the project was directed from a mobile
laboratory established in Waukegan, Illinois, just south of the
Wisconsin-Illinois state line. Figure 3-3 shows the deployment
of instruments. The main attraction of this site was a ten station
telemetered S02/wind monitoring network around the coal burning
Waukegan power plant (WPP). A base station was established 3 km
from the lakeshore on the grounds of the Jack Benny Junior High
School (JBJHS). Figure 3-4 shows the headquarters which housed
one of the two wiresonde tethered balloons, the white sound
baffle for the acoustic sounder, radio, and radiosonde antennae,
etc.
Instrumentation
During 1974 a Cessna 182 was instrumented to monitor air
quality and meteorological variables (temperature, relative humi-
dity, pressure altitude). A Sign-X fast response conductimetric
sensor was used for S02. A two-channel Royco particle counter
sampled the airstreara isokinetically. A Universal indicating
Turbulence System (UITS) was installed which produces a linear
14
-------
output from 0.0 to 10.0 units which is equivalent to £ 1/3,
where £ is the eddy dissipation rate. The output closely cor-
responds to the turbulence sensed by the aircraft passengers and
thus is comparable to the subjective methods used to locate the
TIBL. In addition, DME, air speed and two VOR signals were
acquired. All data were recorded on magnetic tape cartridge for
later ground processing by MODCOMP 11/25 computer. Back-up strip
charts for SO,, and particulate data were also available. The air-
craft instrument package is shown in Figure 3-5.
An instrumentation package was designed for use in either
a Bell 47J or 206 Jet Ranger helicopter. A Sign-X S02 monitor
with a strip chart electrostatic recorder also monitored pressure
height. Dry and wet bulb temperatures were recorded on a second
chart. Tests showed that in normal flying configuration rotor
downwash did not interfere with any of the sensors.
A mobile van was supplied by Environmental Research and
Technology, Inc., which contained a fast response flame photo-
metric S02 sensor. The van was held in place for at least ten
minutes for each reading beneath a fumigating plume. A second
car was instrumented with a temperature probe and routinely
measured the air temperature gradients normal to the shoreline.
All the above were in constant UHF FM communication with
a base station and each other. The aircraft was generally used
to spot the location of the fumigation zone and to vector the
others to it.
A chain of four single theodolite pibal wind stations was
established in a line normal to the shoreline to roughly 30 km
15
-------
inland. Thirty gram helium filled balloons were tracked at 30
second intervals, and data were processed by the Univac 1110
computer yielding mean wind speeds and direction in roughly 100 m
thick 1ayers.
A monostatic acoustic sounder was operated continuously at
the Waukegan base station and then later at the Milwaukee base,
which was atop the 60 m building about 1.5 km inland.
Two wiresondes were constructed for temperature profiles
in the lowest 350 m. Radiation shielded YSI precision thermistors
were attached to 600-ft3 finned balloons. Temperature accuracy
was estimated to better than 0.3°C. Actual heights were deter-
mined with the aid of balloon zenith angle measurements made by a
theodolite. In Waukegan, wiresondes were stationed at the shore-
line (WPP) and at the 3 km base station. In Milwaukee, they were
located at the shoreline and at a site 16 km inland. Additional
data was gathered by an RD-65 portable radiosonde used at both
the Waukegan and Milwaukee base stations. Generally three sound-
ings were made each operational day.
Considerable photographic data were taken. Standard 35 mm
cameras were used °n board the aircraft and helicopter. A time-lapse
all sky fish eye camera was placed beneath the WPP stacks to
record smoke behavior. Also conventional 16 mm time-lapse movies
were made of smoke and cloud patterns from both base stations. In
addition, hourly 360° cloud panoramas were made at each base
station. The use of polarizing filters and/or color infrared
films often produced superior photographs of clouds and smoke
16
-------
patterns that were otherwise obscured in the hazy atmosphere.
Lagrangian air trajectories were determined using 2.5 m3
constant volume super pressured balloons (tetroons) made of 2-mil
thickness mylar, filled with a helium air mixture. By following
certain inflation procedures, it was possible to set the tetroon
to drift with the wind on a predetermined isopycnic surface
(nominally weighted to float at 250 or 300 m AGL), towed to these
altitudes by two or three pilot balloons. Figure 3-6 shows a
balloon in flight.
Tetroons were tracked by two theodolites using 25-power
standard theodolites atop tall buildings (Figure 3^-7). The base-
line between the theodolite pair was 3740 m. Azimuth and eleva-
tion angles were recorded every 30 seconds to 0.1° resolution.
Two-way radio communication assured the synchronous coordination
of readings. The double theodolite computer program gave x, y,
and z positions (in shoreline coordinates and absolute UTM co-
ordinates) and computed the u, v, and w wind components at 30 sec
intervals.
The tetroons have a free lift on the order of 1500 gm,
which is sufficient to fly a standard radiosonde package from
beneath the balloon. A standard NW^VIZ radiosonde was modified
to cycle automatically through temperature, humidity, and refer-
ence contacts (Figure 3-8). Analysis of the data from the RASOT
(Radiosonde-tetroon) technique gave a three-dimensional picture of
the winds, as well as the temperature and humidity variations over
time and space.
17
-------
w
R
A
+
-5-
i fr—
LEGEND
DNR AIR QUALITY MONITOR
ERT SO, MONITOR
HOURLY WEATHER OBSERVATION
PIBAL
TETROON TRACKING SITE
TETROON RELEASE
WIRESONDE
RADIOSONDE
ACOUSTIC SOUNDER
HYGROTHERMO GRAPH
ANEMOMETER
ALL-SKY CAMERA
Radio Co.
Figure 3-1. The Milwaukee area field program network, summer
1974. Not shown are nine air quality monitoring stations
operated by the state of Wisconsin, Department of Natural
Resources.
18
-------
Figure 3-2. Aerial view of Jack Benny Jr. High School (JBJHS)
base station in Waukegan. The large trailer houses both office
space and storage space, especially for the wiresonde balloon.
The acoustic sounder shelter is visible on the scaffold at the
lower left of the enclosure.
19
-------
A
N
A
WAUKEGAN, ILL. FIELD OBSERVATION NETWORK
LEGEND
HYGROTHERMO GRAPH 03 ERT SO2 MONITOR
ACOUSTIC SOUNDER -6- ALL-SKY CAMERA
ZION NUCLEAR R RADIOSONDE; W WIRESONDE
ANEMOMETER
PIBAL
TEMPERATUR
TRAVERSES
WAUKEGAN
MEMORIAL
AIRPORT
Figure 3-3. The Waukegan, Illinois field program network,
summer 1974.
20
-------
Figure 3-4. Widelux wide
as viewed from the north.
The radiosonde antenna is
angle camera view from top of UHM Science Complex building,
Tetroon tracks and the 16 mm and 35 mm cameras visible.
located to the right of the picture.
-------
Figure 3-5. Cessna 182 instrumentation and data acquisition
system. Sign-X at bottom with Royco above. The UITS read-out
dial is on upper left, next to HP multichannel recorder used for
hard copy of S02 traces.
22
-------
Figure 3-6. A tetroon being towed to its neutral density level
by two pilot balloons floating over Lake Michigan just east of
the University of Wisconsin-Milwaukee campus. The RASOT (modi-
fied radiosonde package) can be seen dangling below.
23
-------
Figure 3-7. Aerial view showing various launch and tracking sites
for tetroons and pibals around Milwaukee. The laser range finder
determined base lines for the tetroon trail ings are indicated.
24
-------
Figure 3-8. Specially modified NWS-VIZ radiosonde, a RASOT,
used with tetroon system.
25
-------
SECTION 4
FUMIGATION AND THE THERMAL INTERNAL BOUNDARY LAYER
GENERAL COMMENTS
An airmass that advects over a colder water surface is both
conductively cooled from below in the lowest 100-150 m (Lyons,
1971) and not destabilized by the penetrative convective heat
transport that is present over land during periods of solar inso-
lation (Lyons and Wilson, 1968). When this stabilized airmass
reaches a down wind shoreline on a sunny day, it experiences an
almost step function jump in surface roughness lengths and heat
input. In an analogy to laboratory flow experiments over a heated
plate, a thermal internal boundary layer (TIBL) develops. This
has in fact been extensive physically modeled (Ogawa, 1973). It
has been shown (Raynor, et al., 1974) that even before land fall,
stabilized marine air has diffusion characteristics substantially
different from those deduced from over-land diffusion experiments
under equivalent thermal stabilities. When this airmass reaches
land, the diffusion in a highly transitional stability state be-
comes even more complex and atypical.
The implications of variable diffusion characteristics in
such shoreline "transitional states" was gradually recognized by
a number of researchers, perhaps beginning with Munn (1959).
Turner (1969) adequately described nocturnal radiation inversion
breakup fumigation, a phenomenon which typically produces high
26
-------
surface concentrations from elevated plumes for periods of on the
order of 30 minutes. As early as 1962, Bierly and Hewson discus-
sed "Type III Fumigation" occurring when plume matter is emitted
into an elevated stable layer at the shoreline and then intensely
mixed downward as the convective turbulence generated during over-
land travel reaches plume altitude. They further illustrated and
discussed this as the result of field experiments on the Lake
Michigan shoreline (Hewson, Gill, and Walke, 1963). Limited
studies on the west coast (Robinson, Eberly and Cramer, 1965)
further focused attention on the possible serious implications of
what is now called continuous (dynamic) fumigation of high-stack
plumes in coastal zones. Bierly (1968) studied the characteris-
tics of lakeshore internal boundary layers, though with relatively
little emphasis on pollution aspects. Van der Hoven (1967) and
Collins (1971) made summary studies of what was known about fumi-
gation and presented simple models to account for the formation
of TIBLs. Figure 4-1 is a dramatic photograph of the stable plume
from a lakeshore power plant drifting inland on a typical fumi-
gation day: strong insolation and onshore flow of stable lake-
modified air. The view taken in the opposite direction (downwind)
shows no plume matter - it having been rapidly convected to the
surface in the vicinity of the photographer by the fumigation
mechanism. Accurately modeling this dynamic process was the first
major goal of these projects.
A FUMIGATION MODEL
As a result of a limited field study in which shoreline
27
-------
power plant plumes were photographed, Lyons and Cole (1973)
extended Turner's (1969) fumigation modeling procedure to the
case of continuous (dynamic) fumigation. The model was phenom-
enologically rather simple. Complete details of the model as
applied to a single point source are contained in the original
paper. Figure 4-2 however, highlights its basic characteristics.
Three regimes are postulated. In the first (Figure 4~2a), the
plume from a high stack is emitted into a stable layer and drifts
inland with relatively little diffusion, and plume
-------
of the process), and yields concentrations presumably consistent
with ten minute sample averages. Needed, in addition to the usual
model input parameters are specified effective plume heights and
most importantly, the depth of the top of the TIBL as a function
distance along the plume centerline. The distance between Xb
and Xe is called the "fumigation zone" and the quasi-el 1iptical
area with the 50% of ^^ isopleth is termed the "fumigation
spot."
In that first paper, using estimated input parameters, in-
cluding a crudely determined TIBL profile, it was calculated that
maximum surface S02 concentrations from the OCPP could exceed
1.00 PPM. It should be noted that all four stack effluents were
combined into one plume. This high level, plus the possibility
of the mechanism continuing for over six hours^raised the specu-
lation of widespread exceedences of the federal three-hour SO
standards downwind of shoreline power plants. However it was
also pointed out that the model needed extensive field testing,
validation, and calibration.
A FURTHER APPLICATION
Even before extensive field testing, additional computa-
tions based on the Lyons and Cole model revealed the fumigation
phenomena to be far more complex than perhaps originally envis-
ioned.
While the plume(s) from a shoreline tall stack may indeed
fumigate for many hours on end, fluctuations in wind direction
and speed, plume rise(s), and the height and shape of the thermal
29
-------
internal boundary layer (TIBL) cause the location of the highest
ground level concentrations (the fumigation spot) to move about
the landscape with great speed, therefore reducing significantly
the total dosage received by any given receptor.
To aid in the design of a 10-station S02 network, a model-
ing effort to show fumigation patterns from a coal burning power
plant at Waukegan, Illinois was conducted (Lyons and Dooley,
1974). The simulation covers an 8-hour period and uses both
actual and estimated meteorological data for 16 July 1973. This
day, assumed to approximate a "worst case" example was simulated
on an hour-by-hour basis, the time history of the fumigation
patterns from a multi-plume source then being integrated to show
final concentrations and dosages.
The coal-burning 964 MWe Waukegan power plant (WPP) is
located on the western Lake Michigan shoreline halfway between
Chicago, Illinois and Milwaukee, Wisconsin. It has four stacks
of 102, 121, 137, and 137 m height above ground. The two tallest
stacks had nearly identical emission parameters, and were consid-
ered to form a single combined plume. The annual average S02
emission rate for the three plumes was 485, 448, and 2220 gm/sec,
thus making the combined plume from the tallest stacks predomi-
nant. Calculated plume rise showed considerable variation with
wind speed, from as low as 30 m for the middle stack at 10 m/sec
wind to as great as 780 m on the combined plume at 1 m/sec wind
speed.
From those days during the summer of 1973 when hourly
shoreline pibal wind data were available in Milwaukee, 16 July
30
-------
was chosen as most likely having "worst case" characteristics,
that is, causing the most intense and long lasting fumigation
potential. A high pressure cell centered over 1ower .Michigan
maintained fair skies and light gradient easterly flow over
Wisconsin and Illinois. Landsat-1 high resolution images of the
area during mid-morning revealed cumuli forming 10-16 km inland
from the western shore. Though no lake breeze per se developed,
a probable mesohigh over the lake caused the usual clockwise
rotation of the low level winds. Figure 4-3 is a hodograph of the
pibal measured winds at 271 m over the Milwaukee shoreline between
0900 and 1600 CST. It was assumed that all three plume axes lay
along these azimuths at each hour. Thus neither velocity nor
directional shear is accounted for in these calculations. Due
to the lack (at that time) of adequate aircraft turbulence data,
the profile of the top of the TIBL was estimated on an hourly
basis (Figure 4-4). P-G class F was assumed in the stable air
while P-G class C was used in the unstable air.
Figure 4-5 is a plot of the S02 isopleths of the three
plumes at 1100 CST. More importantly is the combined pattern of
all three plumes at the time of maximum fumigation intensity
(1300 CST) showing the average maximum concentration of 1.30 PPM,
some 4.6 km from the power plant. The fumigation spot is very
small, however, with concentrations greater than 0.5 PPM cover-
ing only 2.7 km2. Figure 4-6 is a time history plot of the S02
concentrations from each of the three plumes showing the large
area affected during this 8 hour period. Federal threes-hour SOo
quality standards are expressed in terms of total dosage received
31
-------
at any one point. Figure 4-9 shows the reconctructed trace of
an SOp sensor that might have been located at the point of maxi-
mum concentration. It was assumed that the surface plume con-
centration pattern remained steady state during a two-hour period
while rotating clockwise with the winds past this receptor point.
The resulting dosage calculated is 1.04 PPM*Hr, considerably less
than the three-hour standard of 1.50 PPM*Hr. The 24-hour average
of 0.14 PPM (or 3.36 PPH*Hr) was not even remotely approached.
Virtually all of the assumptions used in this model have
been shown to be highly conservative. No doubt locally high
values of pollution do occur beneath fumigating plumes. However,
due to the rapid movement of the fumigation spots, the potential
for exceeding three-hour dosage standards at any given point are
smaller than might be expected bas.ed on a mere analysis of in-
stantaneous concentration values. While fumigation effects most
certainly must be carefully considered in any environmental im-
pact assessment for an elevated source in the shoreline environ-
ment, proper modeling may reveal that some such sources can be
legally operated within federal standards. Those occasional
times when limits might be exceeded would be prime candidates
for the application of supplementary emission control strategies
(fuel switching, etc.). This study also revealed that fixed
monitoring sites would only poorly resolve the nature of the
fumigation processes.
It must also be remembered that in all of the above, no
validating field data was available, and no correction factors
have yet been applied to the predicted concentrations. The 1974
32
-------
field program was directed at alleviating these problems.
THERMAL INTERNAL BOUNDARY LAYERS - TIBLs
Clearly the most sensitive input parameter into a fumi-
gation model is the shape of the top of the thermal internal
boundary layer. Laboratory experiments (Ogawa, et al . , 1973)
and some field investigators (Weisraann and Hirt, 1975) have
assumed its shape to be parabolic and assymptotically approaching
a maximum depth equivalent to the inland mixing layer depth.
Eariler investigations however a priori assumed the top of the
layer to be linear, and the models of Van der Hoven (1967) and
Collins (1971) proceeded accordingly. In fact, the Lake Michigan
field studies showed the measured TIBLs to have a variety of
shapes with no easy parameterization scheme apparent.
A fundamental question must be raised as to how the TIBL
is defined within the context of a measurement program. In fact,
separate boundary layers develop in terms of moisture, heat,
moisture flux, etc. as the lake air flows inland and is modified
(Bierly, 1968). For the purposes of defining elevated plume
fumigation, the parameter of importance is vertical mixing gen-
erated by columnar penetrative convective elements raising from
the superadiabatically stratified ground layer (Warner and Tel-
ford, 1967). The top of the TIBL in fact marks the level of
maximum penetration of these convective motions. Inferring this
structure from conventional meteorological data is definitely
not straightforward. Surface temperatures, chilled by passage
over the cold lake generally are restored to inland values after
33
-------
10-30 km of overland fetch, as shown in Figure 4-8. But this type
of information gives little clue as to vertical structure above.
A way is needed to discriminate between "laminar" and turbulent
atmospheric flow regimes. The former refers to stabilized lake
air masses which have minimal turbulent fluctuations due to
either thermal or mechanically generated eddies. The latter is
characterized by both mechanical but primarily thermally driven
eddy motions. In level flight in a small aircraft, this sharp
distinction can be readily noted by an observer. Crossing into
the TIBL results in the flight transforming from generally smooth
to rather bouncy. A field experiment, using an NCAR Queen Air,
refined this simple TIBL detection approach used in the original
Lyons and Cole (1973) paper. Data were collected on 5 August
1971 during a series of east-west traverses normal to the Mil-
waukee shoreline during mid-afternoon with strong sunshine and
stable onshore flowing non-lake breeze winds. The potential
temperature analysis (Figure 3-9) shows the development of the
TIBL. A corresponding plot of the observers subjective charac-
terization of encountered turbulence (Figure 4-10) shows a very
strong correlation indeed. The power plant plume under study
that day was seen to begin fumigating at about the point where
it was engulfed by turbulence of "level II",noticeable bouncing
of the aircraft.
This, and other similar experiences, lead to the decision
to define the TIBL not by radiosondes, etc., but rather by air-
craft turbulence monitoring. The development of the Universal
indicating Turbulence System (UITS) by MacCready (1964) suggested
34
-------
a fairly simple method of profiling the TIBL from a light air-
craft. The UITS, a wing mounted gust probe essentially senses
1/3 ^
£ , where C is the eddy dissipation rate of turbulent energy
within the inertia! subrange. Experience has shown that the
"feel" of turbulence to an aircraft observer is very closely
1/3
correlated to the values. The £ values recorded on tape
(and monitored via dial by the aircraft observer) are independent
1 / O
of aircraft speed. Values of £ below 0.3-0.5 cm2/3sec~^ were
associated with "smooth" air. As C.1/3 approached 1.0 cm2/3sec~'
noticeable bouncing would be felt, which would be described as
the onset of "light turbulence" by an experienced pilot. This
level of turbulence, almost always found to be associated with
penetrative convection (often made visible by rising columns of
smoke) was chosen as the criteria for defining the top of the
TIBL. Values above 3,5 cra2/3sec~1 , occasionally found within
lake breeze convergence zones and beneath growing cumulus, would
be characterized by a pilot as marginally heavy turbulence.
Aircraft flight tracks were chosen to fly east-west tra-
verses normal to the shoreline at various altitudes, in part to
2/3
measure the £ patterns. Experience showed that they were
remarkably revealing of the gross structure of the atmospheric
boundary layer. Figure 4-11 is a good example, Isopleths of
1 ' 3
£, ;' drawn therein must be properly interpreted however. Rather
than representing lines of equal £.1/3 value, they should rather
be considered "envelopes" within which the values indicated fre-
quently and repeatedly reoccurred. Thus outside the 0.5 era2/3
_ •]
sec isopleth one finds no exceedences of the value (smooth air),
35
-------
and within 3.0 cm2/3sec~1 there are wild fluctuations of C1//3>
many exceeding that value. Again, the top of the TIBL was assumed
to be roughly equivalent to the 1.0 cm2/3sec'1 isopleth.
The 27 June 1974 case shown in Figure 4-11 represent early
morning, near noon, and mid-afternoon traverses of the shoreline
in the Waukegan area. On this day, a weak lake breeze did in fact
form. The front was clearly delineated by a vertical wall of
turbulence, indicated by the^1/3 gradient several kilometers
inland. Rapidly increasing southeasterly gradient winds however
quickly destroyed the lake breeze and the last vestiges of its
return flow by 1000 CDT. By mid-day one could almost invision a
"plume" of turbulent heated air starting at the shoreline.
This technique proved quite successful and numerous TIBL
profiles were obtained under varying conditions. As Figure 4-12
reveals, a plot of ten typical mid-afternoon profiles of TIBL tops
shows that they can be highly variable indeed, ranging from linear
to parabolic to hyperbolic, etc. The shape and depth of the TIBL
was found to be very complex and clearly a function of many fac-
tors including initial potential temperature lapse of the onshore
flow, vertical wind speed and direction profiles, insolation,
intensity angle of the mean flow to the shoreline and shoreline
shape itself, etc. An appempt was made to correlate the depth
of the TIBL at 5 km inland fetch with the shoreline potential
temperature lapse (as suggested by Van der Hoven). Linear re-
gression analysis yielded (Figure 4-13) a poor correlation co-
efficient of 0.35 (which was the best of any relationship found).
Despite the wide scatter, a trend could be noted showing that
36
-------
the TIBL would be deeper at 5 km the less the initial stability
of the lake airmass. This is due to the fact that penetrative
convection will grow more rapidly the less of an overlying inver-
sion it needs to destroy. This also means that fumigation poten-
tials are likely to be higher for onshore flows which are just
barely stable, rather than those which are the most stable, a
fact not always appreciated by those dealing with the phenomenon.
An interesting observation was made on 28 June 1974. The
aircraft was flying just at the top of the haze layer through
which columns of smoke could be seen protruding. As each was
traversed the turbulence values rose accordingly (Figure 4-14),
as did the fine aerosol particle count and the SO- concentration.
This graphically illustrates the role that penetrative convection
can play in vertical transport of atmospheric pollutants as well
as the successful use of £ 1/3 as a first approximation to the
upward limit of vertical mixing.
37
-------
OJ
co
Figure 4-1. Print made from a polarized color infrared 35 mm slide, 1330 CDT, 28 June
1974, on Ground at Waukegan Airport. The Waukegan Power Plant Plume (dominated by stack
5) was stratifying west-southwestward before fumigating.
-------
A. VERTICAL PLUME GEOMETRY
STABLE LAYER ALOFT
CLASS XS'
H
EFFECTIVE
STACK HEIGHT
TURBULENT
LAYER
CLASS V
B. HORIZONTAL PLUME GEOMETRY
x'= x-x '
Figure 4-2. (a) Schematic of plurrte geometry in vertical (XZ)
plane used in modeling continuous fumigation, (b) horizontal
(XY) plume geometry used in the Lyons and Cole continuous
fumigation model.
39
-------
WIND HODOGRAPH
AT 271 M
WIND SPEED SCALE (m/iec
Figure 4-3. Wind holograph at 271 ra AGL, 0900-1600 CST, 16 July
1973 along Lake Michigan shoreline at Milwaukee, Wisconsin.
40
-------
1000
800
Q
z
o
Of
o
>600
2
400
200
TOP OF THERMAL INTERNAL
. BOUNDARY LAYER (TIBL)
16 JULY 1973
10
KILOMETERS FROM SHORELINE
15
20
Figure 4-4.
nal boundary
Estimated profiles of the top of the thermal inter-
layer (TIBL) used in calculations.
41
-------
WAUKEGAN, ILLINOIS
COMBINED PLUME
AT 1300 CST
surface S0? concentrations (in PPM) at
Figure 4-5. Calculated
1100 CST (showing three fumigation^spots along same axis) and
combined total for three plumes at 1300 CST, the time of the
maximum predicted surface concentrations.
42
-------
WAUKEGAN, ILLINOIS
Figure 4-6. Time history diagram of predicted fumigation patterns
from three power plant plumes (0900-1600 CST, 16 July 1973) Plume
azimuths shown at hourly intervals. Interpretation: at a point
on a given concentration isopleth, S00 levels were predicted to
have reached the indicated level at Ifeast for one instant during
43
-------
1.0-
Q.
Q.
*>
8
.5
.TOTAL DOSAGE =
AREA UNDER CURVE
1300
140O
1900
Ti
Figure 4-7. Predicted time history of S0? concentrations at re-
ceptor point receiving highest instantaneous concentration (1.30
PPM). Total dosage received during the approximately 2 hours
necessary for fumigation spot passage was 1.04 PPM*Hr.
44
-------
-Pi
en
(/) T]
3~ -J.
O tQ
-s c
CD -5
o OO
3- -h
0) OJ
-s o
3 CD
fD
s: 3
eu -a
<= ro
ro G>
IQ r+
Oi C
ho
o ro
c
3 <
CD CD
-s
— i in
(£> c:
>~-J U)
-P*
• Q.
_j.
in
<->•
a>
3
n
ro
-s
o
3
CU
?r
a>
DISTANCE FROM SHORE (km)
~-1>°1 0 w S ~ % #
O ^*.
NX,
s ^
0 r
o
0 f
°l
8
_4
= 8f,l
" ,
S*l/_
- r'
o ~ii —
0 ob
/
S
8
8
S
8 \
- \
8 TP
s • — i —
n ^^
^^*6
^^«0^
^-^:
• —
/- —
?/""
'
/ /-
//
//,
/ /
\ \
- /
s s
— (•— 1
\
, \
\ \
\ N
\
43 —
"54
J~-58 —
•-6(1 —
^__
S
(
\
f
S
\
-\
\
\
NX^
— - — .
1
^_______
\^
1
\
^v.
ouu S 5t g g 88
SUHhACE TFMPFRA-nilRES Location WAUK^AM
SUMMER 1974 FIELD PROJECT Date ft JUHE ,974
-------
HEIGHT ABOVE GROUND, FT.
< 3- S O O
fD fD ->• O O
O Cu rt- 3 3
rt- < 3-
-••<<
O Cu
3 —' 3
n>
Cu
to
_i. CU -S
rt- ••
-h
-s
0
3
rt-
3-
fD
to
c
-s
-h
CU
O
fD
. .
rt-
3"
fD
3
fD
CL
fD
3
O
rt-
fD
to
rt-
3-
fD
e+
O
T3
O
— h
fD
— i
fD
<
Cu
rt-
fD
Q.
to
*<
3
0
T3
f-j.
— i.
O
to
c
cr
to
o
3
— J.
3
<
fD
-s
to
_1,
0
3
— 1>
(/>
O
Cu
T3
T3
fD
Gi-
cn
3>
c:
(O
c
to
rt-
— i
IQ
•^4
— J
u
s:
_j.
rt-
3~
Z
0
s
fD
rt-
rt-
-s
Cu
<
fD
-s
to
fD
O
-h
1 —
Cu
?r
fD
1
^
t
-C
O
rt
fD
3
3
^
rt-
— i
Cu
rt
fD
3
T3
fD
-s
Cu
i
3" 3-
fD fD
-S
3 —'
Cu Cu
fD
-"• ~S
3
rt- 3-
fD rt>
-S Cu
3 rt
Cu fD
—• O.
cr CT
O <<
m5 tj
Q- fD
Qj ^3
-5 fD
-s
—' CU
Cu rt-
<< -••
fD <
-S fD
^-^» O
-i O
•-H 3
DO I
O. cr
fD <<
3
O Cu
fD
-"• rt-
3 CU
^C tlT"
fD —i
"^ fD
to
O CU
3<<
fD
Cu "^
CT
O O
< <
fD fD
O r+
O 3-
O fD
-h — •
rt- Q>
O -5
->• rr ro
3 -••
(/) IQ O
rt- Cu -s
-S 3 O
c to
3 to to
fD 3-
3 O to
rt- -S fD
f> ft) O
D- —' rt-
JO 3 O
C fD 3
fD
fD O) S
3 rt- CU
to
> O
-j. Cu 3
Q-
C~) fD
-S
3» fD -h
fD O
fD fD «• —'
» CU O
HEIGHT ABOVE GROUND, FT.
O O i"i f*i
s
-H
3-
fD
_i. -J.
V) 3
-------
DISTANCE FROM SHORE [KM;
1407-1559 COT
EOOY DISSIPATION BATE [CM^SEC'j
I 2S
w
1
DISTANCE FROM SHORE [KM]
Figure 4-11.. (top) Aircraft .measurement of eddy dissipation rate
(.£;', cit]2/3sec-l ) on an east^west traverse near the Illinois-
Wisconsin border, at 0912-1008 CDT, 27 June 1974. (middle) same,
but for 1127-1119 CDT, (bottom) same, but for 1407-1559 CDT.
47
-------
TEN THERMAL INTERNAL BOUNDARY LAYERS [TIBLS]
SHORE
DISTANCE INLAND (KILOMETERS)
Figure 4-12. Ten selected TIBL tops monitored by aircraft during
mid-afternoon during 1974 field program. Shown also are computed
effective stack heights (H) for highest and lowest OCPP stacks
with wind speed varying from 1 to 10 m/sec.
48
-------
CTI ~n
7TIQ
3 C
-s
-h fD
fD
r+ -P»
n i
-fs»
to
cu ,
3
Q-
-•• fD
e-t- CU
-i. -s
fu
—« -S
0>
W (O
3- -S
O (T>
-S 00
fD V)
-i. O
3 3
tt>
OJ
T3 3
O OJ
rt- —'
0> <<
3- V)
C+ ->•
-i. (fi
fu
—• cr
o>
—' c-t-
0) S.
-a ro
^ fD
fD 3
TIBL HEIGHT (m) AFTER 5km LAND FETCH
Q.
fD
TJ
rh
a>
e+
-------
1142-1146CDT
3000 ft. AGL
(«astbound)
1143
1144
1145
1146
Figure 4-14. Plots of £1/3j 502, and 0.3-1.3 micrometer aerosol
concentrations for a four minute section of flight just at the
top of the mixed layer showing the turbulent and polluted columns
of air just penetrating into the stable layer above at 3000 ft
AGL.
50
-------
SECTION 5
THE GLUMP REGIONAL FUMIGATION MODEL
GENERAL COMMENTS
The 1973 Waukegan analysis showed the clear necessity of
treating separately the several plumes of a multi-stack power
plant. Differences in plume effective stack height and the effect
of the strong directional wind shear could cause the fumigation
spots from plumes of a "point source" to be separated by many
kilometers at any given instant.
The model was tailored to compute the plume characteristics
separately and then sum the results. Conceptually, the step to
a large and distributed number of point and area sources was a
simple one. The GLUMP Regional Diffusion Model was therefore de-
veloped for the western shoreline of Lake Michigan. It can, of
course, be reconfigured for any other shoreline area that is rel-
atively straight and has quasi-uniform coastal terrain.
MODEL CHARACTERISTICS
The point source model of Lyons & Cole (1973) served as the
basis for a regional model. The GLUMP Model (Great Ukes LM4M
Mesometeorology Project) is in many ways similar to steady state
Gaussian approaches that have been employed in the past, but with
one major difference. It was specifically designed to model
shoreline fumigation, predicting resultant concentrations from
51
-------
several hundred point and area sources at a given point in time.
A 1970 emission inventory of S02 and suspended particulates
for southeastern Wisconsin was available for use. Figure 5-1
shows the major point sources of SC>2 in Milwaukee County, circa
1970. Note that they cluster along the shoreline and the east-
west industrial valley. The model also separately simulates the
individual plumes from multi-stack sources, such as a power plant.
The GLUMP model was programmed for the University of Wis-
consin System Univac 1110 dual processor computer. At maximum,
it requires 190 k words of core, and 30+ minutes of run time.
The program is constructed in a modular fashion. Plume
rise can be calculated from any number of equations. Typically,
large sources are estimated by Briggs' formulae (Briggs, 1969).
Other equations can be used depending on atmospheric conditions
and source characteristics (cold plumes, etc.).
The model was designed to be versatile. The number of
receptor grids is variable, with over 100,000 points used if the
computations require. Most of the examples shown herein used a
129(X), 153(Y), 6(Z) grid, or 118,422 receptors. Also,the com-
putational grid size is arbitrarily chosen, perhaps as small as
100 m or larger than 5 km. The source emission and the receptor
grids are independent of one another. Typically all sources
within the seven counties of southeastern Wisconsin and extreme
northeast Illinois are available for use. The receptor grid is
positioned by assigning the location of a movable base point
(the upper left hand grid point). UTM coordinates are used
throughout. Figure 5-2 is a horizontal schematic of the model.
52
-------
The number of vertical layers is also arbitrary, with six
being most frequently used. Each layer has its own wind speed
and direction, P-G stability class for both (Pz and (j^ (a "split
sigma" approach), temperature, pressure, and background level of
the source effective plume height is calculated, it is assigned
to a given layer, and diffuses with the conditions at that level.
Thus for a multi-plume power plant, it is possible to diffuse
each plume in different directions, etc. This partially corrects
for the inability to simulate wind shear effects found in some
Gaussian plume models.
Area sources are given pre-assigned effective plume heights
and are treated as virtual sources.
Figure 5-3 is a schematic showing the vertical structure
of the model .
In order to reduce computation time, several steps are
undertaken. When a TIBL is entered, it extends from the lake
shore to 20 km inland,at 0.1 km increments. TIBL heights beyond
20 km are calculated by extrapolating the TIBL's slope in the
19.3-20.0 km zone. A "TIBL wind" is assigned which allows the
computation of the TIBL height along a mean wind vector from the
shore. Thus, full mixing depths may be attained after 20 km
fetch over the heated land -- but this may be very close to the
lake shore for a north-northeast flow and actually 20 km inland
for a due east flow. The presence of an irregular shoreline
often creates some highly distorted TIBL surfaces as winds depart
from the easterly direction (Figure 5-4).
53
-------
The first step computes all receptor fetches with respect
to the lake shore, yielding the height of the TIBL at that point.
After calculation of H, this allows the plume to be assigned to
the proper computational scheme (conventional dispersion, fumi-
gation, or lid trapping). With this information, the plume's
limits are computed. One solves for a lateral (Y) and a downwind
(X) distance along the plume centerline, at which the concentra-
tion drops off below a minimum desired cutoff. This approach
cuts down the number of computations or throws out the source
entirely if it does not enter the receptor grid. The cutoff con-
centration of SO- used was 0.0002 PPM. Along with these compu-
tations, two other factors are computed, Xg and Xr, the points
at which fumigation begins and ends. These are distances down-
wind from the source, not necessarily (and not usually) equivalent
to the "fetch" of the "TIBL wind" from the shoreline.
MODEL INPUT DATA REQUIREMENTS
The following is a listing of the information needed to
run the GLUMP model for shoreline fumigation.
1. One must specify the receptor grid base point, grid
spacing, x and y dimensions, number and height of the various
layers, and a plume concentration cut-off value.
2. The shore is entered in UTM coordinates with the number
of points being some multiple of 2n, where n is greater than 3 but
less than or equal to 11. This requirement is derived from the
fact that a smooth shoreline profile is determined by a modified
binary-lookup technique employing a maximum of n searches. The
54
-------
shoreline from the northern edge of Milwaukee County to just south
of Waukegan is specified by 2048 points and is assumed straight
north and south of these points.
3. The emission inventory consists of all-point sources
(over 200) with their UTM location, physical stack height, inside
stack diameter, exit velocity, stack temperature, emission rate,
company name, and hours per year of operation. These are updated
on an hour-by-hour basis for the power plant plumes simulated
here. Area sources, if used, must be specified by location, size,
source strength, and effective stack height.
4. The TIBL height is read in at 0.1 km increments to 20
km of inland fetch along the "TIBL wind."
5. The "environment" is specified for the number of layers
chosen with wind speed, wind direction, background level, temp-
erature, mean pressure,
-------
2. Each plume computed is listed along with all relevant
data such as Q,£H, H, lateral spread, plume centerline concen-
tration/Cat 0.5 km increments downwind, and location of xe and
xb-
3. A statistical table is generated giving the maximum
and minimum*^ at each level, along with mean and standard devia-
tions of all non-zero receptor points.
4. A printer plot echoes the TIBL profile as a function
of fetch along the "TIBL wind." (Figure 5-4).
5. More useful is a CALCOMP plot of the TIBL surface with
contours every 50 meters showing the actual pattern of mixing
heights over the region.
6. If desired, a 3-D plot of a TIBL surface can be gen-
erated, with the perspective of the viewing angle being a varia-
ble.
7. Pollution concentrations can be generated in any xy,
yz, or xz plane used in the three-dimensional receptor grid, in-
cludi ng:
i. Standard numerical grid print output, showing
actual values ,
ii. Printer plots which can be very effective for
viewing individual plumes,
iii. CALCOMP plots with any desired isopleth contour
levels, scaled to overlay directly a 1:62500 USGS
topographic map, etc.,
iv. 3-D surfaces of concentrations.
56
-------
Once a series of computations are made, the values obtained
at each receptor are stored on magnetic tape for any later ref-
erence .
57
-------
POINT SOURCES-SO,
TONS/DAY
• > 10
• IO-I.O
• IXMXI
Figure 5-1. Portion of 1970 emission inventory of S02 sources
in southeastern Wisconsin, showing Milwaukee County major point
sources. Particulate emissions have a similar geographic
distribution.
58
-------
ilON INVENTORY PERIMETER UTM COORDINATES
cn
=5 A)
n> -s
•o o cr rs
—• c+ n> o> to o
ro
COMPUTATION PERIMETER
-------
-LEVEL 15-
L AVER 15
G*L*U*M*P MODEL SCHEMATIC
EACH LAYER HAS DISCRETE VALUES OF EACH PARAMETER
PLUME DIFFUSES ACCORDING TO CONDITIONS
IN LAYER IN WHICH H IS COMPUTED.
"ir
BACKGROUND
VALUES
AREA SOURCES
H-5,10,or30m
Figure 5-3. Representation of vertical structure of the GLUMP
model. Each individual effective plume height is computed. The
plume is assigned to a given level having its own wind speed and
direction, background level,
-------
•iiiiiiitfiiililtiiiif
figure 5-4. Printer plot of a typical TIBL top profile with an
east-southeasterly flow. As flow becomes more parallel to the
shore, TIBL top profiles become increasingly complex and distor-
ted .
61
-------
SECTION 6
MODEL CALIBRATION AND VALIDATION
GENERAL COMMENTS
All numerical models of geophysical phenomena are simpli-
fied approximations of reality and are assemblages of assumptions,
compromises, parameterizations, and idealizations, tying together
a conceptualization hopefully based upon the phenomenon as it ac-
tually occurs and not how it is merely perceived. Accuracy,
whether relative or absolute, is never presumed, merely hoped for.
In general a model would be considered successful if the assump-
tions and techniques are chosen in such a way as to consistently
over/under predict in such a manner as to be subjected to a
rational calibration scheme. The 1974 Lake Michigan field pro-
grams attempted to validate the basic assumptions made in the
model and to find proper calibration factors so that the compu-
tations could be used with confidence for planning purposes.
PLUME GEOMETRY
The most extensively studied day of the 1974 Milwaukee
portion of the field programs was 8 August 1974. Nearly ideal
fumigation conditions were combined with unusually high data ac-
quisition rates. A high pressure ridge over the eastern U.S.
resulted in light southeasterly gradient flow over Lake Michigan.
No lake breeze formed, Pibal ascents showed that even at that,
low level wind profiles were complex and exhibited great spatial
62
-------
and temporal changes. Wind speeds near plume altitudes generally
decreased from over 5 m/sec to under 2 m/sec by late afternoon.
Considerable wind direction shear in the vertical caused the
separate plumes of the OCPP occasionally to be distinct, although
they were more cohesive than many other days studied. Winds at
plume levels early in the day were east-northeast, shifted to
southeast, and then returned to northeast by late afternoon. In-
solation was strong. Cumulus clouds formed about 10-15 km inland
during the afternoon, and a maximum mixing depth of 1250 m was
attained by 15 km inland. The onshore flowing airmass was only
slightly stable inasmuch as lake water temperature was consid-
erably warmer than in the early summer and spring. A rather weak but
persistent elevated inversion was found through the day with a
base ranging from 50 to 150 m. Inland wiresonde runs showed a
ground-based 100 m deep superadiabatic layer during the greater
portion of the day.
Though the east-west aircraft traverses were not designed
to monitor S02 plumes per se, chance allowed the plane to fly
nearly along the mean plume axes during mid-afternoon. Figure
6-1 shows an excellent resemblence to the profiles hypothesized
by Lyons and Cole (1973). Values above 3 ppm were found aloft.
It should be remembered that the plumes were actually drifting
through the aircraft track at angles of up to several tens of
degrees. Yet the overall pattern is obvious.
In order to test the correspondence between measured hori-
zontal and vertical plume profiles and the GLUMP model predic-
63
-------
tions, specific aircraft and helicopter data taking traverses
were run. A complete study of model calibration and validation
is made by Dool ey (1976) .
Figures 6-2, 6-3, and 6-4 are plots of the vertical heli-
copter soundings made on 8 August 1974. Sounding number one
(Figure 6-2) was taken 0.9 km northwest of the power plant and
shows the plumes traveling downwind. The lower plume has layered
out between 125 and 275 m while a portion of the upper plume is
centered at 350 meters. This is typical of a pre-fumigation
sounding. By 6.3 km (Figure 6-3), the lower plumes have fumigated
but the upper plume, above the TIBL, is still layered just below
500 meters. At a distance of 11.4 km (Figure 6-4), all the S02
has been thoroughly mixed below the TIBL and is reaching a nearly
vertically uniform concentration profile. A nearly identical
sequence was recorded on the prior day (Dooley, 1976). Thus,
there is an excellent qualitative agreement between observation
and model (see Figure 4-2). In particular measured and computed
xe and xb values were satisfyingly similar.
Horizontal plume traverses likewise conform to the model.
Figure 6-5 is the plot of the observed SO concentrations in a
series of north-south sections during mid-day at 2.3, 4.5, and
11.8 km downwind of the Oak Creek Power Plant. Each individual
cross section has a vertical scale for reference. The two closer
runs were made by the Cessna 182 and have their lowest travers at
152 m AGL (500 ft). The helicopter made the 11.8 km traverse and
it was able to traverse from 30 m to 760 m AGL at 30 m increments.
North on this cross section is to the right.
64
-------
There are some obvious features in the cross sections. The
first is that the plumes are expanding with distance, and the max-
imum concentrations decreaseas the plume diffuses. Second, by
11.8 km, vertical 862 concentrations are quasi-homogeneous. Finally,
the plume and its maximum concentration tilt slightly to the
north. Brown and Michael (1974) noted that if no other outside
forces acted on the plume, the veering action of the wind alone
would cause the distinct tilt observed. A direct consequence of
this shear is to spread the plumes over a greater horizontal width
when projected onto a plan service. Dooley (1976) shows that the
changes in horizontal plume width, as specified by chosing two
0rates (one in the stable air and a second in the mixing zone
related to a virtual source) likewise show a satisfactory degree
of comparison between observed and modeled.
CALIBRATION PROCEDURE
While geometric similarity between modeled and observed plumes
is necessary, is not sufficient to assure that computed concentra-
tions are accurate. The values generated must be scaled (cali-
brated) by comparison with actual measurements.
The concentrations predicted by Gaussian plume models are
typically held to be representative of ten minute averages. It is
unlikely that this is the case for a fumigation regime. The mobile
van, positioned by the aircraft and helicopter in a fumigation
spot, took S02 readings for a full ten minutes. Very rapid fluc-
tuations about a mean occurred associated with both the azimuthal
and range variations of the spot and with thermal convective eddies.
65
-------
Figure 6-6, from a telemetered surface station beneath the OCPP
plumes on 8 August is a good example.
The mobile van data from seven days representing fumigation
conditions were examined. The ratios of the observed ten minute
means to instantaneous peaks were plotted as a function of range
from the power plant. While there is considerable scatter, a
curve visually fitted through the data seems to represent the ob-
served trent (Figure 6-7).
The ratio R-|0(x), approached 0.9 beyond 15 km. For the
few readings within 5 km of the stacks, values fluctuated around
0.5. The greatest scatter is found between 6 and 15 km, the zone
where the most active fumigation would be expected. The plume
matter thus becomes more nearly vertically homogeneous as the
distance from the power plant increases. This confirms the pat-
terns measured by the helicopter and aircraft.
The following method appears a reasonable and generally con-
servative calibration technique.
It is assumed that the GLUMP Model predicts essentially the
instantaneous concentrations that are measured by the fast res-
ponse sensors on the helicopter, aircraft, or the peaks monitored
by the van. The conversion to ten minute average SOo values at
the ground .^(.^ Q(X ), is given by:
(x )
where R-| Q is the distant dependent ratio of the 10-minute mean
to peak concentration,^ is the prediction of the GLUMP Model,
66
-------
and-xL-is a calibration factor.
ThelLterm is essentially the ratio of the maximum ^( g pre-
dicted by the GLUMP Model at a given time to the maximum instan-
taneous concentrations obtained by either the ERT Van or one of
the fixed monitors. There were five hours during which it is
reasonably certain that the measured S02 peaks were the highest
peaks. Of these, four were on 8 August, and the fifth was on 28
June. The average value ofj^was found to be 1.28, with all
values higher than 1.00. Thus, as expected, the GLUMP Model tends
to over-predict surface concentrations, by factors ranging from
over 2.5 within 5 km to less than 1.5 at 15-20 km downwind. Note
that in examples shown below, the computer values are uncorrected
and represent
.
PLUMF MASS BALANCE
It has been assumed above that measured S02 values were in
fact accurate. All ground sensors were of course frequently cali-
grated. In particular the flame photometric unit in the mobile
van was checked at least daily against a calibration source. The
conductometric instruments in the Cessna 182 and the helicopters
were also calibrated before each days flights began. Yet numerous
errors can be expected in airborne sampling programs. One method
to assure that instruments are working satisfactorily, the data
are analysed properly, and that all plume matter was monitored, is
to perform a plume mass balance.
In order to perform the balance we need as input data: the
emissions from the power plant, the reconstructed plume cross
67
-------
section, and the wind field into which the plume was emitted. Six
cross sections were selected for this analysis, five taken on 8
August, the day most intensively investigated. The sixth section
was from 6 August, when the flow was offshore and the plume was
over the lake.
The mass balance consisted of digitizing the cross section
and arriving at a total amount of SCL (gm) contained therein.
Then, using the winds obtained from the pibal sites at the shore-
line, 7 km, and 15 km sites, the travel time fromthe power plant
to the location of the cross section could be estimated, and the
emission for the hour nearest the sample would be used then in
the balance.
In attempting this type of mass balance there are several
factors influencing the final results. The first factor will be
the analysis of the plume S02. The technique of extrapolating
a series of horizontal traverses through the plume into a contin-
uous concentration field lends itself to the possibility of sam-
pling errors. The exact centerline value may be missed using
this method and this would reduce the total SO mass. In lieu
2
of some remote sensing technique, this method, while practical,
is crude, and clearly needs improvement. The response of the
SC>2 monitor is not instantaneous and this too could introduce
some errors, especially in large gradients of SC^ concentrations.
Thirdly, SOo undergoes transformation to aerosol in the plume and
this factor is not incorporated in the mass balance. Finally, it
is to be remembered that power plant emissions themselves are
68
-------
only rough estimates.
With these possible errors in mind, Figure 6-8 shows the
results. Computed SC>2 mass captured from the aircraft/helicopter
traverse data ranged from 74.5 to 101.0% (at distances from 2.3
to 11.8 km). The average was 87.3%. This is felt to be accep-
table in light of the various factors mentioned above.
69
-------
8 AUGUST 1974
1347-1647 CDT
20 15 10
DISTANCE MOM SHORE [KM]
Figure 6-1. East-west aircraft S02 traverse, roughly down mean
plume centerline path, 1547 CDT, 8 August 1974. Peak values in
excess of 3.0 PPM remained well aloft.
70
-------
—i O> IQ
O 0- 3 C
O <-h -s
o s: - CD
a 3
-H S —' 01
• ->• O I
3 en ro
a. oo •
—' O 31
—' en a>
—" oo —•
CT> -J-
i o o
—•00
ro» c-t-
O CD
00 -S
o
O 3» CO
—ICO
• taro
c
C/J (/>
*T1 c~h O
-". C
CQ —' 3
C 10 Q.
~J ^J —'•
fD -P» 3
cn
I
-F^ -n o
CQ 10
c^> c
0> -S 77-
3 n> 3
o>
« CT> Q-
I O
c+» 3
BI 3
3 CL
HEIGHT ABOVE GROUND
"O
TJ
o S
o o
HEIGHT ABOVE GROUND [M]
HEIGHT ABOVE GROUND [FT]
HEIGHT ABOVE GROUND
CO
HEIGHT ABOVE GROUND [M]
--
HEIGHT ABOVE GROUND [M]
— « • a>
O OJ -S
CT>
I
CTl
01
i
co
CT>
I
ro
-------
1117-1159
8/8/74
1157-1212
1214-1234
2.3
Figure 6-5. Composite of the 2.3, 4.5, and 11.8 km SO? Traverses
on 8 August 1974. The right hand scale is the distance from the
power plant in km. Each individual traverse has its own vertical
scale for reference but each scale is the same. Isopleths at .01
.05, .25, .50, 1.0, 2.0, 2.0 ppm.
72
-------
rt- -z. — ' tn ~n
3" O to • -'•
fD rt- — I OJ U3
fD -P> C
•-•J
CO
en 3" to
eu -h 01
TT rt- rt- -S '
3 << o en
-h~0 3 •
r+ c -j.
-S 3 O rt-
B» -•• cu 3-00
< to — i O
CD n» ro
-S c+ O O
(/) _J. -f, Q) -J
(T> O TT- fD
• 3 <-*• n
=r o o
cr ro -s -s
0> fD CL
U3 -5 fD
Cu fD 7T -h
=50 -S
O T3 O
C/l -S O 3
=TQ-S
O fD rt-
-S O -S =r
r+ -h fD
— • -a
<< -t> — ' m
C CU X)
cu 3 3 H
-h -•• rt-
fD Cu -J.
-S rt- X
-•• — I n>
rt- O =T D-
3- 3 -<•
fD VI W
3 -"•
fD fD rt- c+
3 Cu -s n>
Q. (/) Cu
C O 3
— ^T fD O
— ' fD 3
ro Q. o -j-
00 3 rt-
-P» CU O
rt- 00 -S
o
O Cu 3» — i
—I CO
— ^o to n
o c: cu
O -". V) rt-
-h 3 rt- fD
rt- a.
-------
RATIOS OF INSTANTANEOUS PEAKS TO 10 MINUTE MEANS (7 DAYS)
KILOMETERS FROM POWERPC ANT
Figure 6-7. Ratio of ten minute average to instantaneous peak
S02 values measured by ERT mobile van. All fumigation measure-
ments for seven days in Oak Creek and Waukegan were plotted.
The line is a rough fit of the data as a function of distance
from the power plant.
74
-------
en
O* Q. — J r+ "~n
— • ro vo -s -••
c/i -P=. cu c:
T3 — '• • 3 ~S
-* _k «*"i /^
o">» — I ro 01
& 3- m i
o- -". ro oo
—•3 PJ •
«< m o -s
t-t- -•• ro
ro -s -s 2
~i C O f+ fa
-i 3 -^3-in
o ro ro ro t/i
-s 3
tn ,-t- -s -+, cr
ro -•• &>
3- ro -s ro BJ
o> i/> ro 3
ri--a in T3 O
o ro — ' ro
S 3 3 C
ro m H- g -+,
-s ro in ro o
ro »* in •"$
o>
3 D» 3 c-h 10
03 3- -••
<-+• Q- O OJ i<
-h rf
r+ i" -h in
Oi fZ in 3- ro
?r — ' 3- O> — '
ro-hOD.ro
3 n» -s o
r+ ro o c+
— '• ro o ro
3 -h 3 O-
o ro o — '-a
-s s ro — '
wo rt- c:
o to o ro 3
o o o> ro
o — ' in CL m
c ro a> •
3 -h • c+
r+ O CU
• "^ — H
3 -< o 3-
n» -5 3 ro
-'• < 00 W
o ro o
3 -S > — '
trt f~" '
Wl W- ^^ •
-s c
ro w
PERCENTAGE OF TOTAL
s s s s a I g
K»
«•
w
> •.
s
I
? s
>
H
JC
3 M
X
»
•
»
»
>
>
• >
o> oo
> >
C C
25 o
&
£
1
3^
0)
O
m
-------
SECTION 7
STUDIES OF ELEVATED POINT SOURCE PLUME FUMIGATION
OVERALL FINDINGS
The field studies did indeed confirm that continuous fumi-
gation of high-stack shoreline power plant plumes could produce
significant surface S02 concentrations. Both the field data and
the refined modeling efforts made it very clear in addition that
the rapid movement of the fumigation spot resulted in much lower
dosages than originally expected. This is not to say that shore-
line fumigation is not a serious problem, but that a sufficiently
detailed model and the proper input data are required to assess
its true impact. The phemenon is in fact an extremely complex
one.
The use of fixed surface networks in order to study shoreline
fumigation is likely to be frustrating unless one is prepared to
accept an unusually long data gathering period (several years?)
or install a very dense network (prohibitively expensive). The
ten station network around the Waukegan Power Plant, designed in
part to monitor its fumigating plume, only sporadically had any
of its stations intercepted by the "fumigation spot" itself. Simi-
larly for several monitors installed near the Oak Creek Power
Plant. On only several occasions during a three-month summer 1974
period did any site clearly record the passage of the spot. Yet on
many days, high surface concentrations of SOo were present somewhere
within the several hundred square kilometer area of potential
76
-------
impact.
The strategy of using aircraft and helicopters to locate the
fumigation spot, and to deploy a mobile van via FM radio voice
link proved to be correct. It was not necessarily easy however.
The spot moved rapidly and of course not along roads that the van
necessarily had to use. All problems aside, several dozen readings
on excellent fumigation days were obtained, which were adequate
for model calibration.
Certainly obvious was that conventionally acquired meteoro-
logical data (on 200 ft towers, etc.) is totally inadequate to de-
fine the atmospheric structure resulting in fumigation. The exact
orientation of the several plumes from a multi-stack power plant
must be known. Even the serial pibal chains with hourly ascents
insufficiently defined the details of the wind which had a major
impact on surface concentrations during supposedly "steady state"
gradient onshore flows. In any case, a 60 m tower will not yield
wind directions relevant to a plume with an effective stack height
of 300 m. So marked were the temperature inversions and vector
wind shears with height that little could be infered from near-
surface observations. Plume diffusive characteristics likewise
were totally unrelated to the P-G stability classes that might
be estimated by using surface data alone and Turner's (1969)
simplified classification scheme. In fact, even temperature
lapse rates determined by radiosonde/wiresonde proved useful only
in the grossest sense of the word. In particular horizontal
plume spread was poorly correlated to temperature lapse due to
the marked effect of vertical wind direction shear.
77
-------
8 AUGUST 1974
As mentioned in Section 6, the most intensely studied day
was 8 August 1974, which was nearly a "classic" in terms of fumi-
gation potential. A weak high pressure ridge was stalled over
the Greal Lakes; there was no lake breeze, but the apparent
steadiness of the onshore flow from the synoptic charts was how-
ever partly illusory. Mean wind directions varied from 070°
during the morning to 135° by raid-afternoon, and back to 085°
by late afternoon. These fluctuations are brought about in part
by the development of a highly ageostrophic wind component associ-
ated with a lake mesohigh (Lyons, 1970). Vertical wind direction
shears, which on this day were comparatively small, acting on
the highest and lowest plumes can have a significant impact on
the location of fumigation spots. Velocity shears, associated
with wind speed maxima similar to "low level jets" also result
in plumes emitted at different levels having markedly differing
plume rise and dilution factors (Figure 7-1). Very telling is a
wind hodograph (Figure 7-2) of the azimuth of the highest plume
from OCPP. This was constructed by calculating the effective
plume height on an hourly basis and then interpolating the wind
direction and speed from the shoreline pibal-derived winds. Each
of the four plumes had similar but still distinctly different
hodographs.
By early August the lake had become considerably warmer,
with surface temperatures approaching 20 °C. The onshore flow
was thus not as severely stabilized as in early summer and spring,
The shoreline radiosonde measured slightly stable layers aloft,
78
-------
and generally P-G stability Class E prevailed at the shoreline.
The maximum surface temperature differential between the shore
and the inland sites was only about 4.5 °C. The wiresonde located
15 km inland generally measured D stabilities in the lower 300 m
during the bulk of the day. Had not a rather thick sulfate haze
covered the area, the surface temperature remodification would
undoubtedly have been stronger, and the TIBL even deeper. The
acoustic sounder showed little indication of significant low-
level stability during the day. It is important to note that the
minimal initial stability played a key role in the relatively rapid
deepening of the TIBL, even under reduced insolation. Had an
intense low level inversion been present, it would have taken con-
siderably greater fetch before the weak insolation could have
caused penetrative convection to erode the overlvinq stable
air while flowing inland.
Various plume measurements on this day are seen in Section
6 above. Figure 7-3 is a horizontal traverse of the S02 plume
made by the aircraft during mid-afternoon. Of some interest is
the plume's rather sharp bend to the left after reaching approxi-
mately 10 km inland. This could very well be a manifestation of
two phenomena. The first is that as the plume experiences vertical
mixing due to convection, the vertical direction wind shear also
enhances lateral spread. The second point is that there appears
to be a "wobble" on the entire boundary layer that occurs during
the stable onshore flow situation. It appears to be something
of a higher speed version of a typical stable flow meander, and
the entire boundary layer appears to undergo alterations of wind
79
-------
direction of several tens of degrees at a frequency of several
times per hour. It should be noted on many days with stable
onshore flow, the plume cannot be tracked as easily as might be
suggested by Figure 7-3. On some days the intense vertical wind
direction shear had the four plumes fanning off in entirely sepa-
rate directions, in one case with axes separated by up to 120°.
Even a fast moving aircraft is incapable of adequately mapping
S02 fields under these circumstances.
MODEL RESULTS
The GLUMP model was employed to produce hour-by-hour simu-
lations of the four Oak Creek Power Plant plumes, each treated
separately. Brigg's stable plume rise equation estimated ef-
fective stack height for each plume, adjusted by observations when
available. The effective stack heights of the four plumes ranged
from less than 200 meters to above 525 meters in the time period
under discussion, in almost all cases but one (discussed below)
there were at least several degrees of wind direction shear be-
tween the four plumes. A typical plot of surface S02 concentra-
tions (uncalibrated, instantaneous values) is shown in Figure
7-4a for 1400 CDT. In this case there appear to be two distinct
fumigation spots on the surface. A ten minute computed value of
0.47 ppm (calibrated) is found, while the mobile van reported
a 10-minute average of .42 ppm. In this case the two lower plumes
had azimuths of 124° with upper plume azimuths being 137°. By
1600 CDT (Figure 7-4b) the winds became more easterly, but more
importantly, for a short time, there was relatively little wind
80
-------
direction shear, less than 1° between the upper and lower plumes.
Thus all four plumes coincidentally lay along the same azimuth
resulting in almost the maximum additive effect. The uncali-
brated SO,, surface concentrations were approximately 1.06 ppm
(similar to those reported by Lyons and Cole (1973) in their
initial case studies). It was during this hour that the highest
surface values were monitored by the mobile van, 0.79 PPM (ins-
tantaneous ) .
While both observations and computations suggest very high
instantaneous S02 values at the ground, the maximum one hour
average recorded by the fixed monitor 5.3 km west of the power
plant was only .16 PPM. This is a manifestation of the fact that
the fumigation spots are moving about the surface with great
speed lowering dosage at any given point. Figure 7-5 represents
a summation of the computations for OCPP stack 1.
The diagram was produced by overlaying all the plumes from
this stack for all hours and drawing isopleths around concentra-
tions of equal value. In other words, any point within the 0.50
PPM contour had values in excess of that figure for at least one
instant during the period 1000 to 1730 CDT. The heavy line con-
nects the center of the fumigation spot for the plume. Figure 7-6
is the same presentation except for the highest stack (number 4)
which also has had the highest S02 emissions. (The extremely
rapid motion of the four separate fumigation spots around the area
is evident.) The patterns from stack 1 and stack 4 were similar
but by no means identical. It is imperative that for multi-plume
sources that each plume be treated separately with the final S02
81
-------
concentrations achieved by summation. Figure 7-7 is another time/
history diagram of the integrated behavior of all four plumes
during the time period 1300-1730 CDT. The tracks of the primary
and, when present, secondary fumigation spots, are noted. The
pattern is clearly dominated by the peak calculated at 1600 CDT,
which may be a partial artifact of the pibal wind data. It seems
likely that a 15-20 km2 area experienced maximum instantaneous
S02 values in excess of 0.70 PPM. Upon application of the 1.28
Q_calibration factor, and a mean distance of RIQ(X) °f 0.75,
the 'Xio(x) va1ue resulting is about 0.45 PPM. Especially in
view of the speed at which fumigation spots travel (see below)
it seems certain that the values exceeding the 3-hour 0.50 PPM
standard did not occur on this day.
FUMIGATION SPOT CHARACTERISTICS
Using the model computations for 8 August 1974 plume simu-
lation, the fumigation spots were investigated in greater detail.
For purposes of definition, a fumigation spot is the area enclosed
within the isopleth representing half of the peak surface SOo
concentration.
Figure 7-8 shows the fumigation spots and areas for each
plume on 8 August 1974. Each spot generally covered an area less
than 20 km2, and occasionally less than 5 km2. Thus the area
affected at any given instance is quite small. The area of a
fumigation spot is largely determined by the slope of the TIBL
at the point where it intersects a given plume. If the TIBL is
rising rapidly, the area will be quite small, although peak surface
82
-------
Thus it can be concluded that, unless actual observations of
plume (j~~ ' s are at hand, modeling efforts in the shoreline boun-
dary layer should include corrective terras for C7"~v to account
»/
for shear diffusion phenomena.
Figure 7-11 shows all observed CP 's as a function of dis-
J
tance downwind from the power plant. What becomes clear is that
the main line through all of these data does not appear to follow
the general curve slope, but rather shows a distance-dependent
decrease in the rate of vertical spread. Thus the data appear
to be illustrating the significant effect of active diffusion
due to the plume's inherent initial turbulence at emission into
the atmosphere for the first several kilometers. Aircraft $} /3
measurements within plumes the first several kilometers downwind
found values equivalent to that in a fully developed boundary
layer. The fact that the plume was actively entraining environ-
mental air could be seen by the turbulent nature of the plume.
Only after several kilometers did the plume become smoothly
stratified (as long as it remained above the TIBL). Thus while
it has not been incorporated into the GLUMP model, it would be
wise to investigate the use of a
-------
concentrations will be higher.
Figure 7-9 charts the speed of the fumigation spots across
the landscape through the day showing a range from less than two
to almost 10 kilometers per hour. The higher stacks (numbers 3
and 4) of course moved faster, since they intersect the TIBL at
greater heights than the lower, thus at greater range, and a
given angular variation results in a higher surface translational
speed. On this day, which was probably rather typical of many
summer days, these spots traversed the landscape approximately
four kilometers per hour. The difficulty in obtaining an adequate
10-rainute sample for a spot perhaps less than a kilometer in width
is obvious.
It should be reeraphasized that if a lake breeze were in
progress all of the above described fumigation spot behaviors
would be even more erratic and complex. Wind shears within the
lake breeze circulation cell are considerably greater, and winds
are more temporally and spatially inhomogeneous. Fumigation
spots would then be found over a much wider area and travel at
even greater speeds, making tracking all the more difficult, and
interpretation of fiaed station monitoring data almost impossible.
SPLIT SIGMAS
Dooley (1976) studied the 1974 field data to determine the
validity and/or necessity of employing the split sigraa approach
in the GLUMP Fumigation Model. Based on initial visual obser-
vations that plume lateral spread seemed to be disproportionately
large in comparison to the vertical spread within the lake en-
84
-------
vironment, Lyons and Cole (1973) allowed the option of using dif-
ferent P-G stability classes in their single point model. This
capability was continued in the GLUMP Regional Fumigation Model.
In general it was found that the measured (J~2 values were
often close to, but sometimes a class or more greater than the
expected based on vertical temperature lapse rate. But the plume
lateral spreads <7~ were often much larger than the temperature
*/
might lead one to expect. This effect is undoubtedly due largely
to the vertical directional wind shear. Tyldesley and Wallington
(1975) could account almost entirely for the horizontal diffusion
of plumes. Brown and Michael (1974) noted wind shear effects
from power plant plumes to be significant within Long Island sea
breezes .
Figure 7-10 shows the observed plume (J~z ' s plotted as a
function of the observed plume (77, ' s . Note that in no case did
•J
they fall within the same P-G stability class. Of those studied,
29% were one P-G class apart, 33% were two P-G classes apart, and
38% were three or more P-G classes apart. The use of a G~y belong-
ing to the same stability class as the
-------
projected Pleasant Prairie power plant (Wisconsin Electric Power,
southwest of Kenosha, Wisconsin) in Lyons et al . (1974). In this
particular case, a very slowly developing TIBL remained below the
top of the proposed physical stack height for much of 27 June
1974. This resulted in a classic case of fumigation at a point
five kilometers further inland. Again, quite high instantaneous
surface S02 concentrations were predicted, but the plume's extreme
meandering effects discussed above prevented any predicted exceed-
ences of 1, 3, or 24-hour standards.
86
-------
Figure 7-1. Hourly pibal ascents at the Milwaukee shoreline, 8
August 1974. Top shows plotted winds (1 barb = 2.5 m/sec), Wind
speed (u) in 1 m/sec isopleths shown at the bottom.
87
-------
STACK HEIGHT UNIT 4
1-1-M
Figure 7-2. Hourly hodograph of wind at effective stack height
(H) for OCPP stack 4. The holographs for each of the four plumes
were similar but by no means identical.
88
-------
8 AUGUST 1974
13S8 - 1431 COT
SO, MEASUREMENTS
5OOH. ACL
Figure 7-3. Shows the horizontal S02 patterns measured in a saw
tooth trajectory flown west of OCPP at 150 m AGL between 1358 CDT
and 1431 CDT. Once again, as at Waukegan, a sharp bend in the
plume axis appears. This is probably the result of a shift in
the boundary layer wind direction associated with the meander
phenomenon
fumi oat ion
discussed above. The peak values
zone were between 0.5 and 0.6 ppm
of SOn noted in the
89
-------
LAKE
MICHIGAN
MILES
Figure 7-4a. GLUMP model predictions of surface S02 concentrations
(uncalibrated, instantaneous readings) at 1400 CDT, 8 August 1974
in Oak Creek area showing double fumigation spots due to large wind
shear.
90
-------
LAKE
MICHIGAN
Figure 7-4b. GLUMP model prediction of surface S09 concentration,
1600 CDT, 8 August 1974, in Oak Creek. (The figure shows a single
fumigation spot with 1.06 ppm maximum during a period of negligible
directional wind shear in vertical.) The values shown here are un-
calibrated instantaneous predictions.
91
-------
ro
Figure 7-5.
predicted max
Stack 1. The
laying all of
for all hours
concentration
words, any po
had values in
one time duri
A heavy line
gation spots
rapid motion
spots around
Represents a summation of the
imum^G for the plume from OCPP
diagrams were produced by over-
the plumes* from a given stack
and drawing isopleths around
s of equal value. In other
int within the 0.50 ppm contour
excess of 0.50 ppm for at least
ng the period 1000 to 1730 CDT.
connects the centers of the furai-
for each plume. The extremely
of the four separate fumigation
the area is evident.
Figure 7-6. Same as Figure 7-5, but
for OCPP Stack 4.
-------
Figure 7-7. Time history of predicted surface SO? "XG values for
the sum of all four OCPP stacks, 1300-1730 CDT, 8 August 1974.
Plume fumigation track is heavy line, and a secondary is shown
as dashed.
93
-------
Ul
oc
25
20
15
10
AREA OF FUMIGATION SPOT 8 AUG74
\
10 11 12 13 14 15 16
TIME CDT
Figure 7-8. The areas of the individual fumigation spots on 8
August 1974. The areas of fumigation at 1500 CDT were not
available because portions of the plumes were outside of the
computational grid.
94
-------
SPEED OF FUMIGATION SPOT 8 AUG 74
75
2-5
•o
o>
0)
a
0)
mean all
4 stacks
time cdt
Figure 7-9. The speed at which the fumigation spots moved on
8 August 1974. The upper graph is the mean speed for all four
plumes and the lower is for each individual plume.
95
-------
N
<
Q
111
>
cc
LU
(/> C
m
o
B
BCD
OBSERVED SIGMA Y
Figure 7-10. c?"2 plotted as a function of #
model did not use split sigma then most of
would have to fall along the diagonal line.
y. Note that if a
the observed values
96
-------
1000
to
Of.
fc>N
10
1 10
DISTANCE DOWNWIND , km
100
Figure 7-11.
-------
SECTION 8
URBAN SCALE POLLUTION PATTERNS
EXTENSION FROM POINT TO REGIONAL MODELING
This report so far has discussed fumigation from single or
multi-plume "point" sources. What happens when an entire urban-
ized area undergoes continuous fumigation? The GLUMP model was
designed to simulate this phenomena. The mesometeorological data
gathered on 28 June and 8 August 1974 were used in predicting S02
and suspended particulate patterns at the surface over Milwaukee
County and environs for several selected hours. Only point source
calculations are shown here (since the accuracy of the area source
emission inventory available was questionable). In face, it is
fairly certain that the 1970 emission inventory for point sources
had become invalid by 1974 due to rather major changes in fuel
combustion patterns in Milwaukee County in that time period. Thus
no calibration of the model was attempted. The calculations are
still useful of course for the manner in which they illustrate
the deleterious effects of onshore flow in a heavily industrial-
ized urban area. The complete model and results are described in
Schuh (1975).
RESULTS
The urban model is essentially that described in Section 5
and used in Section 7, except that over 75 major point sources of
98
-------
SO,, and participates were introduced. A 0.33 km horizontal grid
spacing was used, resulting in a receptor grid 43 km in the x dir-
ection, 51 km in the y direction, and with 6 levels (118,422
receptor points). Except where noted, plume rise is computed
using Briggs1 equation based on hourly inputs of emission factors
(when available), wind speed and direction from pibals, aircraft-
measured TIBL profiles (Figure 8-1), wiresonde temperature profiles
(for plume rise), etc. Pasqui11-Gifford stability classes for
each level above and below the TIBL, were those used in discussing
the power plant plume fumigations.
Figure 8-2 shows the predicted (and uncalibrated) ground
level concentrations of S02 for 1100 CDT, 28 June 1974. The top
portion of the illustration are the predicted concentrations for
essentially a plume (lid) trapping regime with the lid constant
everywhere at 700 m. The bottom portion shows the fumigation
regime values for identical conditions except the TIBL surface
rising along curve b (Figure 8-1) to a final height of 700 m. In
essence, this compares the difference between lid trapping and
fumigation on a regional basis. Careful examination shows that
the fumigation regime is indeed a more restrictive one in terms
of surface air quality. Peak values for fumigation are noted in
excess of 0.50 PPM. The typical CALCOMP concentration isopleth
plots are somewhat difficult to use when attempting to visualize
the overall pattern. Thus the GLUMP model output data are pref-
erably presented as three-dimensional surfaces covering the same
area as Figure 8-2 but viewed from above and to the southwest.
99
-------
Figures 8-3 through 8-8 have all been smoothed to eliminate, among
other things, "orifice errors", that is, extreme values resulting
from a receptor grid being unusually close to a source.
Figure 8-3 is the exact same information as shown in Figure
8-2. It is clear that substantially higher surface S02 values
should be expected during the fumigation episode as compared to
lid trapping. Similar area averaged S02 levels were 0.018 vs.
0.009 PPM. That is to say, average S02 readings over Milwaukee
County can be 100% higher during fumigation episodes (at least
under this set of meteorological conditions). The suspended par-
ticulate computations (Figure 8-4) show a similar pattern. Fumi-
gation peak values of 4552,ugm/]T]3 are compared to 1362^gm/m3 for
lid trapping. Area! averages show fumigation levels from point
source emissions 30% above those for lid trapping.
Figures 8-5 and 8-6 show much the same for 1600 CDT, 8
August 1974, when the final depth of the TIBL was deeper (almost
1000 m). For SO,,, peak surface values for fumigation were 31%
higher than for lid trapping, and area! average concentrations
were 42% higher. It should be noted that during "fumigation
episodes" not all sources, especially low-level ones, actually
experience fumigation. They do, however, diffuse within very
limited and slowly increasing mixing depths for the first 10 to
20 km of inland drift.
Figure 8-7 and 8-8 are for 8 August 1974, but for 1730 CDT,
when increasing high cloudiness causes a rapid reduction in TIBL
depth to a maximum of less than 400 m (there was also a sharp
100
-------
drop-off in v/ind speed and a shift to more northeasterly direction)
For SO^, peak values for fumigation were actually less than for
lid trapping. This is partly an artifact of the plume rise com-
putational scheme which caused several large power plant plumes
to remain entirely aloft. But the area! average for fumigation
was a full 144% above that for lid trapping. Areal averages of
particulates (Figure 8-8) were 41% greater for the fumigation
case.
Thus it appears that., for short-term or "worst case" epi-
sodes, continuous fumigation from both isolated sources and for
entire urban areas must be considered a major concern in any po-
tential violation of standards. The regional computations also
show that the lake effects of fumigation are insignificant beyond
about 20 km under most circumstances (note, major differences in
concentrations are most pronounced near the shoreline).
101
-------
THERMAL INTERNAL BOUNDARY LAYERS
6 8 10 12 I4~
DISTANCE FROM LAKESHORE (Km.)
Figure 8-1 .
(a) 1600 CDT
(c) 1730 CDT,
Three TIBL tops
8 August 1974,
8 August 1974.
used in regional
(b) 1100 CDT, 28
computations.
June 1974, and
102
-------
Figure 8-2. CALCOMP contour plots of uncalibrated GLUMP model
surface SO? concentrations for 1100 CDT, 28 June 1974 for a
uniform 700 m deep mixing layer (top) or for a variable depth
TIBL (bottom) yielding fumigation.
103
-------
Figure 8-3. Three-dimensional surface of SO2 concentrations pre-
dicted by GLUMP model for Milwaukee County with (top) a uniform
700 m lid, and (bottom) fumigation according to conditions of
1100 CDT, 28 June 1974.
Figure 8-4. Same as Figure 8-3, except for suspended participates
104
-------
Figure 8-5. Three-dimensional surface of S02 concentrations pre-
dicted by GLUMP model for Milwaukee County with (top) a uniform
lid of 1000 m, and (bottom) fumigation according to conditions at
1600 CDT, 8 August 1974.
Figure 8-6. Same as Figure 0-5,except for suspended participates
1 05
-------
(D
-D 3
c
£
O
o ro
o r-~
Q- fO
O
4- T3
•4-J O
3 O
O
** [ •>
uo
i en
CO E
•r—
CD -a
S- S-
=3 O
CO U
•r- O
1_1_ fO
(^ E
(13 O
QJ 4->
E rti
fT3 O^
oo •«-
\
o
S-
ra
Q.
-a
O)
-a
E
0)
Q.
CO
3
to
-M
CL
(11
O
X
a>
i
CO
GJ
i-
CD
CO
03
a>
i
CO'—
E
O) O
S- 4-> •
3 -4-> -vl-
ai o r^.
•i- JD C1
co
i
CO
O)
cn
-------
SECTION 9
ACOUSTIC SOUNDER DERIVED MIXING DEPTHS
COASTAL CLIMATOLOGY
The vast amount of meteorological data collected in this
country serves the needs of aviation and National Weather Service
numerical forecast models. Air pollution meteorologists, when in
need of climatological data have had two choices: make do with
existing networks, or install their own sensors -and wait a con-
siderable amount of time. When attempting to define aeral dis-
tributions of mean mixing depths for instance, the synoptic radio-
sonde network is all that is available to use (Holzworth, 1972).
It is known that the numbers derived are highly unrepresentative
of mountainous regions, cities, and of course coastal zones. Not
only are temperature regimes very spatially variable on the meso-
scale, but frequent mesoscale wind flows (urban heat islands,
mountain/valley winds, land/sea breezes) require vertical wind
profile climatologies. EMSU special radiosonde stations have only
slightly filled the void. Since radiosonde technology is highly
expensive in terms of both equipment and manpower, remote sensing
technology seems more likely to yield mesoscale time-continuous
data in a cost effective manner. Vertical wind measurements via
laser or acoustic doppler anemometry will be feasible in the near
future. At this time, continuous estimates of mixing depths using
107
-------
lidars and acoustic sounders are now quite routine, especially
the latter. The installation of an acoustic sounder along the
western shore of Lake Michigan by UWM in 1974 gave rise to the
acquisition of data for a "cliraatological" study of mixing depths
near a large lake during various synoptic regimes. A complete
survey of the field of acoustic sounding technology can be found
in Hall (1972), and Thompson (1975), among others.
An example of the complex thermal regimes found in the Lake
Michigan coastal zone was acquired at Waukegan, Illinois during a
five-day period of sustained onshore stable flow. Figure 9-1 shows
analyses of wiresonde vertical temperature profiles at the shore-
line and at 3 km inland on one of the days. An intense elevated
inversion was present at the shoreline power plant. The shallow
surface based neutral layer was due to the air's passage over a
narrow rim of warm water just offshore. By the time the air tra-
veled 3 km inland, the inversion had partially burned off, but as
afternoon onshore flow increased, the capping inversion could still
be found. Figure 9-2, usi'ng data taken on the prior day when con-
ditions were very similar, shows how these complex lake-induced
thermal patterns can lead to potential errors in diffusion esti-
mates. If one were to use the table presented in Turner (1969)
to convert surface data into P-G stability classes at Milwaukee
airports some 6 and 16 km inland, one sees highly unstable A, B,
and C conditions suggested. Yet a quick analysis of wiresonde
vertical temperature lapse rates converted into P-6 equivalent
classes (Pendergast and Crawford, 1974) at the shoreline and 3 km
108
-------
inland reveals that while indeed the immediate surface layer may
be remodified, intensely stable layers remain aloft. Thus the
use of surface data to infer dispersion conditions at the alti-
tudes of elevated plumes under the influence of lake effects is
highly questionable at best. An acoustic sounder, while not pro-
viding quantitative data (in the configuration used) allows one
to monitor continuously the approximate depth of the mixed layer
at any given point at relatively little cost.
THE ACOUSTIC SOUNDER
The AeroVironment Model 300 acoustic sounder was operated at
ground level some 3 km inland at Waukegan and atop al5 m high
building in Milwaukee about 1 km inland. The sounder, which had
an input power of 35 watts, transmitted typically at 100 micro-
second pulses at 1600 Hz, and was enclosed within an 8 ft high
plywood/lead sound baffle lined with convoluted urethane. The
data were recorded on facsimile type paper with 15 cm vertical
scale (either 500 or 1000 m mode) at the rate of 32 mm per hour,
and were cut into 24-hour (Midnight to Midnight local time) strips.
Since the sounder operated in a monostatic mode, the signal return
was generated only by temperature fluctuations on the order of
half the wavelength, typically 20 cm.
With some experience, and especially with the-vast amount of
ancillary data available, it became quite straight forward to in-
terpret the traces obtained. Figure 9-3a is a trace taken at
Milwaukee, 13 July 1974, on a day with brisk offshore southwest
winds and strong sunshine. The nocturnal radiation inversion,
109
-------
approximately 200 m deep, was plainly visible and began burning
off by 0700 CDT, and the top of the mixed layer went above the
500 m full-scale by 1100 CDT. The vertical streak lines visible
(though poorly in reproduction) mark the characteristic signature
of thermal convective plumes rising from the surface superadiabatic
layer. Vector-vane G@ val ues suggested surface P-G classes B and
C during the bulk of the daylight hours.
Figure 9-3b was taken under very different conditions, namely
stable onshore flow of fog on 6 May 1975. The echo returns appear
to represent the top of the fog layer where turbulent interactions
are taking place. Heating during the day caused the layer to
expand from 300 m to 400 m around noon only to return to its ori-
ginal depth by evening. Figure 9-3c, taken 24 June 1974 at 3 km,
occurred during stable onshore flow, but clear skies. This was
the first of five consecutive days of continuous stable northeast
flow. Though the sunshine was strong, remodification of the lake
air was sufficiently slow so that a mixing depth of more than 500 m
was never attained at this site throughout the entire period. Vector
vane 07^ readings, even during mid-day suggested no better than
P-G class D at its most unstable. Apparently the extremely limited
vertical depth of the mixed layer prevented the development of
vertical convective plumes, thus limiting the energy of turbulent
wind fluctuations in the surface layer. This was noted many times
throughout the summer. Thus a sounder, especially within 10 km
of the shoreline, can be used to monitor the upper limit to which
near surface effluents can mix at a given point, which is also the
time profile of the top of the TIBL. Rapid changes in mixing depth
110
-------
with the onset of lake breezes will of course also be noted (Fig-
ure 9-3d). On 17 June 1974, weak southwest flow prevailed during
the day and the mixing depth rose above 1000 in by early afternoon.
But a 1ate-onsetting lake breeze front passed the sounder at 1715
CDT. The frontal slope as well as mixing depths reduced to about
400 m are evident.
A CLIMATOLOGY OF MIXING DEPTHS
As discussed by Rizzo (1975), it is possible to use a sounder
to compile a climatology of shoreline mixing depths. During the
period 2 July - 15 September 1974, 1618 hours of mixing depth
heights were extracted from the sounder traces (at the 1 km inland
Milwaukee site). Wind directions and speeds were available from
the co-located vector vane and cloud cover conditions were noted
from the nearby Milwaukee NWS station, as well as logs and photo-
graphs .
Thus for each hour, Rizzo cataloged apparent sounder-derived
mixing depths (when an echo present), wind direction, cloud cover
(clear, scattered, broken, or overcast) and time after surface
or sunset. On those afternoons when the mixing depth exceeded
sounder range (500 m) or the returns were too weak or diffuse to
allow assigning a specific value, a mixing depth was assumed if
the flow was offshore. Figure 9-4 includes a curve of mixing
depths rising to Holzworth's published mean summer afternoon
maximum depth for clear or scattered clouds (1200 m). For days
with broken or overcast conditions, 1000 m was assumed to be the
maximum depth.
Ill
-------
Figure 9-4 then is a graph of the measured (and partially
assumed) hourly mixing depth heights for offshore flow during
clear or scattered cloud conditions plotted in terms of hours
after sunrise and hours after sunset. It is rather as one expects
to find. The contrast with Figure 9-5, the measured mixing depths
during stable onshore flow (clear and scattered) is notable indeed.
In addition to the mixing depths or base of the elevated inversion
being generally in the 175-250m range,there is virtually no diurnal
fluctuation. The same pattern holds forth during broken and cloudy
onshore flow. Thus it can be said that during stable onshore flow,
at any given point, it remains "nighttime" for as long as the con-
dition persists, with mixing depths severely limited and changing
only slowly over time.
A mixing depth wind rose (Figure 9-6) was prepared which shows,
as expected, that night values are lower than day values for via-
tually all directions during clear and scattered cloud conditions.
More importantly is the directional dependence of daytime mixing
depths between onshore and offshore flow (with a lesser but detec-
table nocturnal analog). Mean daytime (clear and scattered) mixing
depths (for all hours, not just maximum) for southwest flow is
656 m as opposed to 240 m for northeast flow (a factor- of 2.73
times greater). During the peak hours of afternoon heating, it
is likely the difference becomes a factor of 5 or more. It should
be noted that the UWM campus is on a small peninsula and both due
north and south winds have considerable over water fetch.
During this period the mean mixing depth over Milwaukee at
112
-------
1 km inland for all hours and all wind and sky conditions was 300
m. For daytime offshore flow it averaged 480 m compared to 240 m
for onshore. Thus any regional dispersion model based on radiosonde
derived mixing depths (which is devoid of any lake suppressive ef-
fects) would likely underestimate mixing depths at this point by
a factor of two, and all other factors aside, underestimate pol-
lution concentrations from local sources by the same factor. It
must be pointed out that these data are only valid for this exact
point due to the extreme variability of lake effects upon TIBL
height. However, the sounder has proven itself highly useful and
it would appear that a network of several sounders in a line nor-
mal to the shoreline in perhaps two seasons would provide an
adequate estimate of lake suppression of mixing depths on both
a day-by-day as well as seasonal basis.
There is of course an overiding question as to whether or
not the structures seen on the sounder trace do in fact properly
define the top of the mixed layer. Schubert (1975) suggests
that sounders tend to yield estimates less than those of conven-
tional temperature sounding techniques (although that is not to
imply the data are not more accurate). On several occasions
when the UITS-equipped Cessna 182 flew nearby the sounder, a
comparison was run between the £ 1/3 and the sounder patterns
(Figure 9-7). On a section of the 4 September 1974 trace, a
shallow but intense nocturnal inversion layer was present below
100 m. Turbulent fluctuations caused by mechanical eddies were
sufficient to cause a strong sounder return. Sections of the
113
-------
1/3
£ trace as the airplane flew at different levels from the
shoreline inland are superimposed. The 0800 CDT data shows that
above the inversion the air was virtually smooth (g,1/3 less than
0.3 cm2/3sec-1) due to the lack of both convective and mechanical
turbulence. By 0915 CDT, just before the onset of a lake breeze,
heating had allowed the mixed layer to grow over 500 m. Values
i/o 2/^1
of C were almost everywhere above 1.0 cm sec . Other
studies show that the intensity of the sounder return is fairly
well correlated, thus linking the aircraft derived TIBL profiles
to the sounder estimated mixing depths. Figure 9-8, from mid-
afternoon, 16 July 1974, clearly shows the top of the sounder
return separates the surface layers with C,1/3 values above 1.0
O / O T
cm sec" from the layer aloft with £1/3 less than 0.3 cm2/3sec"1
114
-------
WIRESONDE TIME/HEIGHT SECTIONS
nATF 6-28-74
08 O9
Figure 9-1. Temperature cross sections for 28 June 1974 at Waukegan
Power Plant (top) and 3 km inland (bottom) by smoothing plots of
hourly wiresonde runs. (Isotherms every 0.5 °C. Heavy line indicates
base of elevated inversions.)
115
-------
27 JUN 74 P-G CLASSES AIRPORT DATA
- Timmerman Field
-Mitchell Field .
27 JUN 74 P-G CLASSES (significant levels) WIRESONDE SHORELINE
27 JUN 74 P-G CLASSES (significant levels) RADIOSONDE INLAND
5
TIME ICDT) .
Figure 9-2. Measurements of shoreline meteorological conditions
made 27 June 1974.
116
-------
July 3
COT
•*•-
Figure 9-3a. Acoustic sounder trace, Milwaukee, 13 July 1974
6 0
! * ,!
' " '
i: »
U 10
i i t
Figure 9-3b. Acoustic sounder trace, 6 May 1975.
Figure 9-3c. Acoustic sounder trace, Waukegan, 24 June 1974
n
COT
'i"'l' i,
v *
,"< .<*. •
^^ M^ i'.^.
Figure 9-3d. Acoustic sounder trace, Waukegan, 17 June 1974
117
-------
If)
DC
UJ
l
X
5
HOLZWORTH'S MEAN MAXIMUM
MIXING DEPTH
JULY 1- SEPTEMBER 15, 1974
- I VALUE OF ONE
[STANDARD DEVIATION
o
SUNRISE
- i-
\ OFFSHORE FLOW
CLEAR AND SCATTERED SKY
SUNSET
TIME
NO DATA PTS 3SS
Figure 9-4. Mixing depth vs. time (offshore flow, clear and
scattered sky).
118
-------
E
55
JUIT l-SIPTIM
VM.UE OF ONE STANDHflO OCVIATIOM
ONSHORE FLOW
CLEAR AND SCATTERED SKY
4 - i
TIME
NO DATA PTS 176
Figure 9-5. Mixing depth vs. time (onshore flow, clear and
scattered sky).
119
-------
MIXING DEPTH WIND ROSE
CLEAR AND SCATTERED SKY
JULY 2 - SEPTEMBER 15, W4
NW
NE
MIXING DEPTH
DIRECTION
N
NE
E
SE
S
SW
W
NW
DAY
188
240
245
252
170
379
307
380
NIGHT
287
276
75
ff w
170
142
252
249
325
DAY VALUES
NIGHT VALUES
100
(METERS)
200
Figure 9-6. Mixing depth wind rose with adjusted data (clear
and scattered sky).
120
-------
4l9iiQ7A?~7' Acoustic sounder f"ace with turbulence data (September
"5 I .7 / T1 y #
121
-------
Figure 9-8. Acoustic sounder trace with turbulence data (16
July 1974).
122
-------
SECTION 10
LAKE BREEZE STRUCTURE
GENERAL COMMENTS
Since, by meteorological standards, the lake/sea breeze is
comparatively easy to observe and has aspects appealing to numer-
ical modelers, there has developed an extensive literature on the
subject. Baralt and Brown (1965) and Jehn (1973) have indexed
literally hundreds of articles. Upon considering the basic nature
of these dirunal mesoscale flows, it is obvious they would have a
profound effect upon both air pollution dispersion and transport.
Figure 10-1 shows a schematic of an idealized land breeze cell,
collecting urban pollutants in its offshore flow, some of which
rises in the offshore convergence zone into the return layer aloft,
then partially returning into the urban area. The sea or lake
breeze (which are dynamically equivalent and often of comparable
strength on the Great Lakes) would act in reverse, only this cir-
culation is characteristically more vigorous (Figure 10-2).
Urbanized shoreline areas during daytime may appear to experience
adequate ventilation, yet the combination of partial recirculation
of pollutants as well as restricted mixing depths could cause
locally serious air pollution potentials.
It was not however until the mid-1960's that the air quality
aspects of shoreline mesoscale regimes began to be appreciated.
Lyons and Olsson (1972) were among the first to illustrate how
the complex wind and thermal patterns dramatically and consistently
effect shoreline air quality.
123
-------
Numerical models of the sea breeze began very modestly indeed
with studies such as that of Haurwitz (1947). As computational
power increased they advanced in complexity (Pearce., 1 955), and
finally with Estoque (1961, 1962) began to achieve a certain re-
semblance to reality, including showing the effects of environmen-
tal winds upon the locally developed circulation. Moroz (1967)
extended Estoque's model to the Great Lakes. McPherson (1970)
was able to make many improvements, including the inclusion of
shoreline irregularities in his model of the Texas coast sea breeze.
The model of Neumarmand Mahrer (1971, 1975) allowed the prediction
of surface pressure fields after the abandonment of the hydrostatic
approximation. Pielke (1974a, 1974b) simulated in three dimensions
the sea breeze of the Florida peninsula, and its convection patterns,
with an eight-level primitive equation model, and a detailed boun-
dary layer parameterization scheme. The 11 km grid used, while
less than in many preceeding models, and adequate for his purposes,
also resulted in much of the detail of the circulation being masked.
Typically, most models underestimate vertical motions in the con-
vergence zone by a factor of two to an order of magnitude or more.
Some of the strongest lake breeze frontal motions computed were
in the model of Sheih and Moroz (1975) which found upward of 95
cm/sec. Numerical models to date have not been configured to
provide three-dimensional trajectories of air motion, which of
course is the relevant parameter for air pollution studies. The
model now under development by Dieterle and Tingle (1976) at
Brookhaven appears to be the first to have this capability.
124
-------
Observational studies of sea/lake breezes, while numerous,
were rarely of sufficeint scope to provide adequate data to allow
verification of models being developed. The study of Fischer (1960)
on the New England coast was perhaps the earliest major program.
The Texas coast (Hsu, 1969), the eastern shore of Lake Michigan
(Olsson, et al., 1968) and the Chicago area (Lyons and Olsson,
1973) said major field programs, however, even these collected
only fragmentary information relevant for modeling purposes. Fig-
ure 10-3 is a summary of literature reports regarding the structure
of well developed lake (sea) breezes. Keen (1976) discusses these
in great detail.
The land breeze is even more limited in terms of numerical
modeling and relevant complimentary observational programs. What
is known about the nocturnal analog of the lake/sea breeze is sum-
marized in Figure 10-4 (also taken from Keen, 1976). Reports by
Olsson et al. (1969) on the eastern Lake Michigan shore, Feit
(1969) on the Texas coast, and radar observations by Meyer (1971)
along the Atlantic coast) are among the more complete studies. In
general numerical models have been plagued by difficulties in
parameterizing the nocturnal boundary layer, which has led to
generally unsatisfactory results.
The bottom line is that models, while clearly becoming in-
creasingly better able to simulate the real atmosphere, are not
yet of sufficient precision and versatility to allow direct ap-
plications to mesoscale air pollution transport problems. Obser-
vational programs, costly and difficult to manage, have generally
125
-------
failed to provide sufficient verification of models, although
improvements continue here also.
The lake breeze observations made on the western shore of
Lake Michigan in 1974 were useful inasmuch as they confirmed many
of the findings of Lyons and Olsson (1973) in Chicago, but more
importantly provided raw data for an attempt at a purely kine-
matical approach to estimating air pollution transport within com-
plex three-dimensional time dependent coastal flows. The basic
observational results are summarized below, after Keen (1976).
THE LAKE BREEZE OF 4 SEPTEMBER 1974
By far the best lake breeze studied during the rather poor
summer season of 1974 occurred on 4 September. In many ways it
resembled that discussed by Lyons and Olsson (1973) and provides
for some interesting comparisons.
A large unseasonably cold cP high pressure system drifted
south of Lake Michigan on 4 September 1974 (Figure 10-5 shows the
noon synoptic chart and SMS-1 visible satellite). The pressure
gradient over the southern basin of the lake that day never ex-
ceeded 1.2 mb/200 km. Clear skies and light winds produced an
intense nocturnal radiation inversion. Surface low temperatures
at dawn set'numerous records (2 °C to 7 °C). The development of
intense lake breezes after unusually cool nites, contrary to what
might be expected, is often quite common. It appears these air
masses are thermally unstable in the lowest layers which aids the
development of intense upward convective heat transport over land
during the day. Figure 10-6 shows lake surface water temperatures
126
-------
increasing from 10 °C on the western shore to 21 °C on the eastern
shore. Also shown are the 0600 CDT surface observations. A
shallow but well defined land breeze was present along the entire
shoreline, persisting for two to three hours after sunrise. Visual
observations of the Milwaukee urban smoke dome combined with the
acoustic sounder trace (Figure 10-7) suggested the top of this
inversion (and perhaps the land breeze inflow) to be about 100 m.
Peak offshore wind speeds were about 2.3 m/sec. Aircraft obser-
vations above this layer found the air to be non-turbul ent (G1/3
2/3 -1
less than 0.3 cm sec~ ), with extremely low aerosol concentrations
Figure 10-8 shows a plume from a fertilizer plant in a rural
area south of Milwaukee draining into the lake in the waning
moments of the land breeze (about 0830 CDT). Vector vaneCT^ob-
servations suggested P-G stability classes D and E existed even in
the urban area within the land breeze outflow before dawn (0620
CDT). The plume from the OCPP, visible to the south in Figure
10-8, was nearly vertically into a small cumulus of its own making
base about 800 m, with the effluent then drifting generally north-
ward.
The sounder trace, as well as the smoke photographs, re-
vealed that convective plumes began to develop over land about
0830 CDT. The aircraft first detected a developing lake breeze
wind shift zone by the appearance of white caps about 3 km off-
shore at 0850 CDT. By 0945 CDT, the front began moving rapidly
inland, undercutting the now vertically mixing urban smoke pall
and injecting it into the return flow layer aloft. The windshift
127
-------
line (Figure 10-9) at first moved at 7-8 km/hr inland, but grad-
ually slowed to about 3 km/hr by mid-afternoon, finally becoming
indistinguishable approximately 30 km inland at sunset. At 1100
CDT pibals revealed the inflow layer to be about 500-600 m deep
with a 1200 m thick return flow layer above (Figure 10-10). Maxi-
mum inflow velocities of 4.0 m/sec were achieved at 150 m above
the lakeshore around 1400 CDT. The surface inflow layer winds
showed the apparent effect of the Coriolis acceleration by shifting
from northeast to southeast as the day wore on.
Cumulus clouds associated with the frontal convergence
zone gradually moved further inland during the day (Figure 10-11).
The Landsat image also shows cumulus and altocumulus over the
lake, the remnants of the vigorous convection present over the
warm lake water on the edge of the land breeze outflow. This
remained vigorous for more than an hour after the lake breeze
onset. This represents a vivid illustration of the fact that
shoreline circulations are driven by differences in upward heat
transport between land and water - even if they are positive over
both. Due to the weak synoptic pressure gradient, the lake breeze
convergence zone was detectable along the entire shoreline of Lake
Michigan.
TETROON OBSERVATIONS
Three tetroons were released on 4 September 1974. Fig-
ure 10-12 is the plan view of their trajectories. Figure 10-13
is a cross-sectional view of the same trajectories.
The first tetroon, launched at 0850 CDT, initially
128
-------
moved toward the southeast in the land breeze outflow, but at 10
minutes into the flight it met the lake breeze convergence zone
and was lifted to about 600 m. For the following 30 minutes it
oscillated between 600 and 700 m, drifting northeastward, after
which it slowly began to sink and move back toward land. The
maximum lakeward fetch of the tetroon was 10 km. After 74 minutes
it was lost from the view of one of the tracking theodolites, to
be recovered later than afternoon at the position marked in Fig-
ure 10-12. This helical trajectory is similar to that found by
Lyons and Olsson (1973).
Tetroon two was launched at 1048 CDT shortly after the
onset of the lake breeze. It drifted slightly south of west, with
the lake breeze, performing a cycloidal loop at about 25 minutes
into the launch. The floating height, between 400-500 m, was in
the upper part of the lake breeze inflow layer. Oscillations of
of this nature have been observed by Angel! (1975) and Pack et al.
(1972) in a Los Angeles sea breeze study. By 50 minutes into the
launch the tetroon had reached the lake breeze convergence zone
(20 km inland) and rose rapidly in the updraft, disappearing
from sight in haze. This tetroon was found on the ground the
following morning at the position marked in the figure.
The third tetroon was weighed off to float at a lower
altitude than the previous two, and 10 minutes after launch (1310
CDT) it leveled off at 400 m, drifting slowly toward the west.
Sudden updrafts took the tetroon to 500 m and 700 m but after 70
minutes into the launch, it began to settle back to its floating
129
-------
altitude. By 90 minutes the tetroon was more than 25 km from the
tracking theodolites and was occasionally lost from the view of
theodolite 1. (The trajectory path from 95 minutes to 110 minutes
was extrapolated between those missing observations.) After 110
minutes, both theodolites lost sight of the balloon, it was re-
covered in a field just north of 194 later that afternoon.
Cross sections of dry bulb temperature and humidity :
profiles from the RASOT package attached to the 1048 CDT tetroon
are given in Figure 10-14. The temperature profile went from a
relatively smooth to a highly fluctuating trace as the thermal
internal boundary layer (TIBL) grew to balloon altitude. The
humidity trace roughly paralleled the dry-bulb curve, indicating
that the periodic fluctuations were associated with penetrative
convective thermals, estimated to be between 200 and 300 m in
width.
AIRCRAFT OBSERVATIONS
Three flights were made in the instrumented Cessna 182
aircraft on 4 September. The first flight was between 0835 and
0948 CDT, the second between 1118 and 1215 CDT and the third
flight was from 1351 to 1542 CDT. East-west transects were flown
from approximately 8 km offshore to about 30 km inland. The
flight of 1351-1542 CDT was meteorologically the most interesting
and will therefore be the focus of discussion. By 1400 CDT the
lake breeze had penetrated 20-25 km inland and distinct areas of
turbulence and particulate concentrations were observed along the
flight tracks.
130
-------
1/3
The £ profile in Figure 10-15 shows a pattern with
smooth air over the lake (values 0.3 cm2/3sec~1) overland within
8 km of the lakeshore. The TIBL upper limit (1.0 cm2/3sec~1)
is not confined to the inflow layer. The effect of forced convection
in the lake breeze convergence is marked by a "column" of air
several kilometers wide with C1/3 above 2.5 cm2>/3sec~1 . Temp-
eratures frequently fall within this updraft zone - and indication
of strong vertical motion. The bending of the isopleths first
inland in the lake breeze inflow layer and then lakeward at the
base of the return flow suggests that turbulence was also being
advected eastward in the return flow aloft while dissipating.
The various features of the lake breeze are strikingly il-
lustrated by the aerosol cross sections (Figure 10-16 and Figure
10-17). For particles in the size range 0.3-1.3/im, nominal ter-
minal velocities from 0.01 cm/sec. In effect they drift with the
air motions. Particles in the size range 7.0-9.0yUm have nominal
terminal velocities between 0.5 and 1.1 cm/sec in calm air and
would tend to settle much more rapidly than the smaller particles.
Figures 10-16 and 10-10 show the distribution of the two size
ranges of aerosol particles obtained from flight transects between
1351 and 1541 CDT. Analyses were made from the Rustrack charts
of the Royco particle counter with a full-scale deflection being
plotted at 100 in the case of the 0.3-1.3 yUm range, and as 10 in
the case of the 7-9 urn range. These illustrations are not indi-
cations, therefore, of absolute particle counts but rather of
patterns of the fine and giant aerosol distributions in the lake
131
-------
breeze circulation.
Figure 10-16 is a plot of aerosols in the 0.3-1.3yftm range
and clearly evident is the effect of the TIBL which starts at the
shoreline and quickly begins urban scale fumigation of the pollutants
stored in the upper portion of the inflow within 3 km. That the
upper half of the inflow over the lake (as determined by shoreline
pibals) is as heavily polluted is strong additional evidence that
much of the return flow mass gradually settles back into the
onshore flow layer causing at least partial recircul ation of pol-
lutants. In addition, pollutants from sources in the Milwaukee
industrial valley (80 isopleth) were being advected inland and
mixed upward in the turbulent lower portions of the inflow layer.
The bending of the isopleths eastward in the region of the frontal
zone together with low particle values (relatively clean air) on
the landward side of the front, suggests that strong subsidence
was present just ahead of the front. With the pollution advected
lakeward in the return flow experiencing a steady decay in tur-
bulence (Figure 10-15) the dipping of the isopleths into the in-
flow layer suggests the presence of subsidence and implies re-
circulation. This is supported by the pattern found in Figure
10-17 where in this region the air is clean in terms of giant
aerosols.
The pattern of the 7-9yu,m range particles (Figure 10-17)
does not show the patterns found in the fine aerosol analysis.
This range of aerosols is produced largely by mechanical abrasion
and geographically random sources accounting for their non-uniform
distribution. Notable .though ,is the high concentration of large
132
-------
aerosols at very low levels over the lake. In all probability
this represents the fall-out of these aerosols into the lake and
is further evidence of the size sorting hypothesis by Lyons and
Olsson (1973). The importance of this to dry deposition lake
pollution has been noted by Lyons and Keen (1976).
Thus 4 September represents a well documented lake breeze.
The observed wind, temperature and turbulence fields can be used
to test existing or developmental numerical models, and plans are
underway to do just this. An immediate use was as input for the
UWM Kinematic Diagnostic Model (Keen, 1976).
133
-------
(b) 1500 L3T
SOLAR RADIATION
i M LUIJJ i 5
nJnn ,%T }'• fp1S*l!n °f the Classica1 Land Breeze at about
0500 LSI, driven by differential .relational' cooling and showing
the urban pollution plume advected over the lake. The dashed line
shows the top of the land breeze outflow layer.
THE CLASSICAL LAKE BREEZE
(•) 1000 1ST
SOLAR RADIATION
i I M i i M M i
LAND HEATED
Figure 10-2. A depiction of the Classical Lake or Sea Breeze,
fully mature, at about 1500 LSI, driven by differential radia-
tional heating, with elemental motions of air pollution trans-
port noted.
134
-------
CHARACTERISTICS OF A LAKE BREEZE
RETURN
FLOW
400-2OOOm
FRONTAL MOVEMENT L5m Me"
CONVERGENCE
FRONTAL
ZONE
1 -2 Km
INCREASING TO 3.5-5 m MC"'
Figure 10-3. Summary of the observed characteristics of a well
developed lake breeze during mid-afternoon.
CHARACTERISTICS OF A LAND BREEZE
I INVERSION ' LAYER
•*» LAND BREEZE OUTFLOW
100-4OOm
SOhra
CONVERGENCE ZONE
OS-15 km
UkND BREEZE FRONT
20 30
fUDUTIOWAL COOLINO
Figure 10-4. Summary of the observed characteristics of a land
breeze near dawn.
135
-------
Figure 10-5. The 1200 CDT SMS-1 satellite photograph on which
is superimposed the 1200 CDT surface isobars and frontal positions
136
-------
LAKE WATER TEMPERATURES
Figure 10-6. Lake water isotherms (°C), compiled from (i) 4 ship
transects on 2, 3, 4 and 7 September, (ii) mean daily water intake
temperatures at Sheboygan, Milwaukee, Grand Haven, and Muskegon,
(iii) averaged from an infrared airborne radiation thermometer
from a flight on 6 September 1974. The dotted lines indicate the
routes of the ships. Surface winds and temperatures (°C) are for
0600 CDT for 4 September 1974. Each wind barb equals 1 m/sec.
137
-------
m
MO
ACOUSTIC 730
SOUNDER 200
RETURN )50
HEIGHTS ,00
50
UNIVERSITY
1 km
intend
VAME
MITCHELL
FIELD
6 km
Inland
COUNTY
NSTITUTIONS
U km
Inland
; y
/
"
0 o 0 \ \ \ \ \ \ \ A/^x^ — N-> \ \ \ \
r ?> & 9 -=•
<< , < < (
-------
Figure 10-9. Hourly lake breeze wind shift positions on 4 Sept-
ember 1974. Initially the lake breeze traveled inland at between
8 and 9 km/hour but in the late afternoon slowed to under 3 km/hour
FP= filtration plant, VH = Veterans Hospital, CI = County Insti-
tutions. Mitchell Field, Timmerman Field and Waukesha are marked
by MKE, MWC, and WAU respectively.
139
-------
1000
1000
Figure 10-10. The 1000 CDT and 1100 CDT pibal soundings on 4
September 1974; 1 barb equals 1.0 m/sec. Pibal sites located
on Landsat image.
140
-------
Figure 10-11. Composite photograph of 3 frames from Landsat-1
of the western shoreline of Lake Michigan at 1103 CDT, 4 Sept-
ember 1974. Band 5 is shown. Note the inland penetration of
the lake breeze by the line of cumulus clouds paralleling the
coast and the clear zone within the lake breeze.
141
-------
r\3
QL £ CT 3 D" O T|
-i. Qi -S -". -s 3 -••
O1 01 (D 3 fD IQ
c-t- (T) C ft) -P» C
EU CT N c+ N -S
3 fD fD fD fD CO fD
O rt- 00 (D
fD £ cr • -5 -a —j
fD fD fD <-t- o
3 fD -h r+ fD I
(D 3 O —I C 3 —i
E" -s 3- -s o- ro
00 r+ fD 3 (T> •
C 3- -S
-S (T) D- —' -h
n> -j. o —•* —' -a
Q. c: oo -p^ o vo —-"
SO) CO S -J QJ
<
_l.
I fD
SOUTH
UTM COORDINATES
NORTH
"< T3 O CT
O fD D fD
—i O» O) —I -h
B> 3 ~S O
C 3 3 fD fD
-S 01 Id a. O
D. O -h
G) n» -h —' -s co
rt- 3 -S CO O en rt-
O- o —'-a o 3-
CO 3 OT3 fD
*>! rt" —'• O
•P* 3- < O 3 O rt-
O fD -•• Old —I -5
fD —I Q>
ro -n S ->• c-t- t-i.
-*• ' f+ 3 fD fD
3 -S fD ct- c+ O
• 01 <-h O -S rl-
c+ —I ~S O O
3- O rt- O -S
S fD O 3-3 -"•
->• 3 (0 fD
Oi rt- (/) Q. Oi
O 3- -•. -S
O fD O. 3 -J- O
3 O -S -t> -h -h
Oi Q. -i. —i rt-
-"• O -h O fD CO
3 —• rf- S O.
—'• fD rt-
O rt- D. —' —'• fD
fD fD Ol 3 rt-
3 S. <<. -S
rt- CT —'• fD rt- O
fD O> rl- ~5 3" O
-S 01 3- fD 3
fD B) 01
I rt- -h —•
—' 3- rt- Q» —'
&> —'• fD fD 7\~ d>
3 -S fD C
fD —' 3
O> CT> O
TV- O 3-
fD fD
Q.
-------
TETROON TRAJECTORIES
4 SEPTEMBER 1974
ru\j
800
700
600
500
400
300
200
100
LOST :
IN HAZE * —
V **-^ \«0 13I° C
"~^ioo "/Xsl^0 •^ir'^C^V*
"•/ \i / J^ IAI^.,./* x M,"^
K s' ^ "~V\
_ OUT Of VIEW N \f
r\ yc^Vr
h /-^^ y
»/ /x
; OUT Of VIEW
/ Of THEODOLITE 1
/
Of THEODOLITE 1 \ \ f~*^
\ \
104« CDT V\
Launch \
\
> OS9O CDT
/ Launch
i
j
_
_
-
-
-
km 30 2O 10 0 10 km
Figure 10-13. Cross sectional view of the trajectories of the three
tetroons. The weigh-off height of the 0850 CDT and 1048 CDT tetroons
was 500 m while the 1310 CDT tetroon was 400 m. Note the cycloidal
oscillation in the 1048 CDT tetroon at about 26 minutes into the
1 aunch.
RASOTTEMPERATURE & HUMIDITY PLOTS 1048CDT LAUNCH
UJ
tr
I
DC
UJ
0.
5
UJ
20
15
10
5
0
- 8
fN
-
^v
- A
--'"' \
1 2 *
F *O
^TempeR
- " /'
> o <
4 N>
) IT) i
ture
^. H
•^^
Kelat
Humii
— ^_
ve
dity
T 1
> o <
t •* •>
^^S"
1
? § '
vV\/
i 1 <
\^
^
>
<
V.
-
0 5 10 15 20 25 30 35 40 45 SO 55 6O 65 7€
Minutes into Launch
80 ^
70 >
H
50 X
UJ
40 £
Ul
3O a:
20
>
Figure 10-14. Temperature (dry bulb) and humidity profiles from the
RASOT package attached to the 1048 CDT tetroon. Times shown are
elapsed time into the launch. Diagonal numbers at the top of the
diagram give the heights (m) of the tetroons at 5-minute intervals.
143
-------
4 SEPT. 1974
mi -1541 en
EDDY DISSIPATION RATE [«*>.
Figure 10-1 5. Measurements of Eddy Dissipation Rate (£1/3)
cm2/3 sec-') on a traverse along Wisconsin Avenue. The inflow
layer of the lake breeze is shown as a dotted line.
4 SEPT. 1974
1351 - 1541 en
ROYCO-CHANNEL
Figure 10-16. The pattern of distribution of the 0.3-1.3^m
range aerosols between 1351 and 1541 CDT. The numbers represent
the percent scale deflection on the Rustrack recorder with a
full deflection being recorded as 100.
144
-------
-pi
en
3" fD —'•
CO O £
CO -S
O O fD
CO to —i
fD O
CT I
S fD —'
OJ c"f" ***J
co s; .
fD
-S fD
fD 3 —I
o 3-
O —' fD
-s co
O- OTO
fD —' a>
Q- rl-
tu c+
Q) 3 fD
to D. -S
O Ol O
• *> -h
D.
r> -••
o co
—I rt
• -S
CT
— • o
rs
CO
n o
01 -h
fD <-+
3T
Q. fD
fD
-h ^1
— > i
fD IQ
-••3
O
=5 -S
DJ
->• 3
3 CQ
fD
HEIGHT (nwtara « 100)
8 S 8
-------
SECTION 11
THE KINEMATIC DIAGNOSTIC MODEL
RATIONALE AND SCOPE
The kinematic diagnostic model (KDM) of a lake breeze-land
breeze system uses observed data as input rather than a computed
flow field. This largely avoids the scale-ratio problems faced
in numerical models, and provides a straightforward method of
analyzing the trajectories of particles within the circulation
system. By assuming that the breezes are laminar flows and ig-
noring the effects of microscale turbulence, a model can simulate
the mean transport wind, which determines a plume centerl.ine.
While the simulation may represent a rather unsophisticated
approach to the problem, it is felt it also reveals with great
clarity the great complexity of shoreline circulations.
The lake breeze-land breeze sequence on 4 September 1974
was chosen as a case study. Computational restrictions had to
be balanced against spatial and temporal resolution requirements,
and the following compromises resulted:
(i) The horizontal computational area extends 35 km inland
and 15 km into the lake, making a total length of 51
km across the coast.
(ii) The vertical computational area extends to 2200 m, with
calculations nominally starting at 1 m above ground.
146
-------
(iii) Horizontal resolution would be refined to 250 m.
(iv) Vertical resolution would be 100 m.
(v) The coastline would be considered straight.
(vi) The individual soundings would be plotted on a cross-
sectional plane, normal to the shore.
(vii) A 24-hour lake breeze-land breeze cycle would be con-
sidered, and by repetition, trajectories of up to 48
hours could be computed.
DATA RECTIFICATION AND PREPARATION
The individual pibal soundings at the four observing sta-
tions yielded 'u' and 'v1 component winds at 100 m height incre-
ments above each station. These component winds were separately
plotted to scale on cross-sectional maps of 1 km horizontal and
100 m vertical scales. Isotachs were then drawn over an area
of 51 km in the horizontal and 2200 m in the vertical. As no
soundings were made further than 24 km inland nor over the lake,
isotach values in these areas were estimated. Before 0800 CDT -
and after 1600 CDT, no hourly soundings were made and the cross
sections were estimated from surface recording anemometers and detailed
knowledge of the behavior of the night time wind flow was not
as important as its gross features.
With hourly 'u1 and 'v1 component isotach fields compiled
for the 24-hour period beginning 0500 CDT 4 September, further
interpolations were made at sections between hours, as often as
every 15 minutes, primarily to describe the inland movement of
the lake breeze frontal zone in more detail. This was necessary
147
-------
since u component velocities in the frontal zone varied from
strong negative values on one side (easterly flow) to weaker
positive values on the other (westerly flow) over an area of
less than 2 km, and further, since the horizontal advance of
the front through time was not uniform. In total 65 u-compo-
nent sections and 24 v-component sections were prepared.
Digitization of the individual sections was accomplished
by using a Bendix Datagrid precision digitizer connected to a
card punch. Measurements were made to a thousandth of an inch
and point values were taken about 0.5 cm apart (approximately
800 m) along single isotachs. Depending on the complexity of
the section, the u-component points varied in number between 860
and 1500 points per section. Since the v-component sections were
not as complex, these varied in number between 650 and 800 points
per section. Approximately 24,000 cards were used for data
storage in the 89 sections digitized.
The Computer Sciences-Statistics Center of the University
of Wisconsin-Madison provided graphics routines for two-dimensional
interpolation, smoothing, and refining, including four interpo-
lation routines called RANGRD, INTERP, INTRP and INTRPB. These
routines do not interpolate in the classical sense, but rather
use algorithms to find the values at every point on a rectangular
grid given a set of N arbitrary values with corresponding x, z
locations in a two-dimensional domain. Keen (1976) gives a com-
plete discussion and should be consulted for details.
The interpolation routine INTRP took the arbitrarily spaced
data points on each time section and produced point values at
148
-------
each of 1173 points in a 51 X 23 grid matrix, and for this a
smoothing routine called SMOOTH was used. The horizontal scale
is in kilometers and the vertical scale in hundreds of meters.
In order to achieve the desired horizontal resolution of 250 m,
each sectional matrix had to be subjected to a refining technique
to produce a final matrix of 201 X 23. For this the routine
REFN1 was used, which finds the values at new grid points bv
bilinear interpolation in a two-dimensional domain from the values
supplied in an existing grid of either equally or unequally spaced
rows and columns.
The output from REFN1 produced a rectangular grid matrix
with 201 horizontal spacings (of 250 m in actual resolution) and
23 vertical spacings (of 100 m in actual resolution). In total
4,623 node point values per section were obtained.
One more data manipulation is necessary before any trajec-
tories could be produced. The period of analysis for this study
was 24 hours, in which time sections were made at 15 minute inter-
vals during the daytime hours and SO^minutes or 60 minute inter-
vals during the nighttime hours. Since a final resolution of 15
minutes was desired, a further series of interpolations had to be
made. In this case a simple linear interpolation between the two
time periods was used.
After these matrices at intermediate times were calculated,
96 u-component matrices and 96 v-component matrices were available
at 15 minute intervals for a 24-hour period. By now 887,616 dis-
crete values have been calculated to produce a time series of
matrices, 201 (0.25 km) by 23 (100 m) at 15 minute intervals.
149
-------
The data were stored on tape in files and it was now possible to
proceed with particle trajectory analysis.
TRAJECTORY CALCULATIONS
The Mainline Program was written to be independent of the
data series, so that the technique can be used with a data series
in which the spatial scales and/or temporal scales are different
from those in the above example. For ease of description, the
program will be explained with the aid of aschematic flow chart,
Figure 11-1.
The first requirement is to initialize the program with the
space and time scales of the study and to decide upon the number
and starting time of the particles.
Input
Parameters
Initialize 1
xUTM, yUTM
SCALEX, SCALEY
NHOURS
NREAD
NP
The UTM coordinates of the lower
left hand corner of the matrix
(0.0)2
The distance between the matrix
columns (in km) and between the
rows (in m) (0.25, 100)
The total number of time periods
to be analyzed. If a 24-hour
period is desired and each time
was 15 minutes apart, then (24 x 4)
+ 1 = 97 (193)
The number of intervals in a time
period
The number of'time period per hour
(4)
"-The figure in brackets fol7owing the descriptions are the
input parameters used in this study.
150
-------
NPTS
START
CORR.
PARTICLE
POSITIONS
AND
DIMENSIONS
INITIAL
U-MATRIX
V-MATRIX
Total number of particles
considered (up to 1000).
Time at which trajectory
tions are to start, i.e.
time in 24 hour time
to be
(273)
computa-
the first
(0500)
Correction factor to increase or
decrease the u and v vector velo-
cities. Given as a decimal.(0)
The spatial starting positions (x,
y, z) of each particle are read in,
followed by its fall speed (in cm
sec-1) and its actual start time.
The u-matrix and the v-matrix of
the first time period are read in.
COMPUTE
'W
MATRIX
From the u-matrix, the w-components
are calculated from the formula
w(z) = w,
dz
z = o
where w(z) is
at height z-|,
the vertical motion
w0 the vertical motion
at the lowest level (assumed zero for
the ground), and x the distance nor-
mal to the shore. These are then
stored into a w-matrix array.
MAIN LOOP
1=1, NHOURS
FLIP-FLOP
COUNTER
INTERCHANGE
MATRIX
This is the beginning of the main
loop of the program. Since each
matrix uses a considerable amount
of core storage, a flip-flop tech-
nique was devised whereby only two
time sections are in core at any
one time. This is accomplished by
means of a four dimensional array
(VCTR(FLIP, IZ, IX, IDIM)) where
the first subscript (FLIP) is used
as a pointer, indicating the loca-
tion of the old and the new time
section.
151
-------
NEW
U-MATRIX
V-MATRIX
COMPUTE
'W
MATRIX
The next time period is read in
and the w-components of that time
are calculated as above.
SUB-LOOP
FOR
POSITIONS
J = l,NREAD
NEW
STARTING
TIME
TEST
PARTICLE
IMPACT
A secondary loop is now used for
computing the actual positions of
the particles between the main loop
time period (15 minutes). In the
secondary loop, positions are cal-
culated every minutes. Since new
particles may start at any time
during the total computational
period (24 hours), a test is inc-
luded to determine whether any new
particles are to start in the time
interval being computed. Another
test is included to stop computa-
tion of a particle if it has
descended to ground level.
COMPUTE
VELOCITY &
DISTANCE
FOR NEW
POSITIONS
NEW
x, y & z
POSITIONS
OF PARTIC.
A subroutine is now called to com-
pute the u, v and w velocities
effecting each of the particles.
The particles are translated at
constant vector component velocities
during each time period and their
positions are then determined from
the following formulae:
NEW(km)=OLDX(km)+3600(sec)xFACTOR(l/sec)xANS(u)(m/sec)x0.001(to km)
NEWY(km) = OLDY(km)+3600(sec)xFACTOR(l/sec)xANS(v)(m/sec)x0.001 (to km)
NEWZ(m)=OLDZ(m)+3600(sec)xFACTOR(l/sec)x(ANS(w)(cm/sec)-FALL(cm/sec))
xO.Ol (to m)
where FACTOR is the time interval over which the velocity
is held constant, i.e. 15 minutes.
152
-------
Winds outside the initial 50 km by
2200 m domain were set equal to
those at the boundaries.
The new x, y and z positions for
each of the particles at every min-
ute can be written out to a file
from which various types of computer
plots may be generated.
By means of a CalComp plotter and standard plot routines,
XY, XZ, and YZ plots were made of the trajectories of the selected
particles. A three-dimensional plot routine was adapted for use
with this program to illustrate graphically the pattern of dis-
persion in a lake breeze-land breeze circulation system.
All computations were performed on the UNIVAC 1106 computer
operated by the Computer Services Division of the University of
Wisconsin -Milwaukee. On this system the program required 27K
words of memory; for each additional particle only 21 extra words
of memory were required.
The efficiency of the program is shown by considering that
during a 24-hour sequence, 1,331,424 individual values are handled,
of which 443,808 (the w-components) have to be calculated. These
computations took between 8 and 10 minutes of computation time.
Some problems were met, however, in handling the plotting
file. As an example, with 273 particles, a file was produced which
contained 1,180,171 point values which on the UNIVAC 1100 series
took 659 tracks of disc or approximately 800 feet of high density
(1600 BPI) tape. For convenience, the file was broken into sub-
files with particles being handled and plotted in groups. Although
153
-------
the x, y and z positions of each particle were given every minute,
plotting tests showed that a point of economic efficiency was
reached when points were plotted at 6-minute intervals. Plotting
them every minute more than doubled the plotting time (and cost)
while longer time intervals became noticeable in departures in
the trajectory paths.
A complete listing of the program and routines used is given
in Appendix II of Keen (1976).
SUMMARY
Annotated flow-diagrams of the procedures followed in the
development of the KDM are presented as a summary of this section
using data at 1100 CDT, 4 September 1974 (Figures 11-2, 11-3, and
11-4).
154
-------
C
START
1
INPUT
PARAMETERS
#1
f PARTICLE
POSITIONS &
DIMENSIONS #2
f INITIAL
V MATRIX
V MATRIX #3
COMPUTE
'w1
MATRIX
#4
FLOW DIAGRAM
OF
KINEMATIC DIAGNOSTIC MODEL
MAIN
LOOP
I=1,NHOURS
#5
FLIP-FLOP
COUNTER
INTERCHANGE
MATRICES #6
bUB-LUOP
FOR
POSITIONS
J=1.NREAD
NEW
MATRIX
'V MATRIX #7
COMPUTE
VELOCITY
'OIST. FOR
POSITIONSi
mt—
x. y. z
POSITIONS
OF PARTICLES
C
E"
Figure 11-1. Flow diagram of the Kinematic Diagnostic Model
155
-------
110O WT
k:
p
U Component
1100 CDT 4 S«pt. 19W
V Component
1100 CDT 4 S«pt. IBM
FROM
FIELD STUDY
c
t
SERIAL PIBAL ASCENTS
ARE USED
TO
PROVIDE
AND
WIND
COMPONENT
MAPS
OF THE
STUDY
AREA
Figure 11-2. Detailed flow diagram of Kinematic Diagnostic Model.
156
-------
DIGITIZED MATRIX
IS
10
15
1O
THESE MAPS
ARE THEN
DIGITIZED
[AND!
RUN THROUGH AN
INTERPOLATING
ROUTINE WHICH
CONVERTS THEM
INTO A
51 x 23
GRID MATRIX
WHICH IS
THEN
SMOOTHED
AND
REFINED TO A
201 x 23
GRID MATRIX
AFTER
INTERPOLATING
FOR ALL MISSING
TIME PERIODS
T
Figure 11-3.
conti nued.
i
Detailed flow diagram of Kinematic Diagnostic Model,
157
-------
11 :00 R VECTORS
\ \ \
\ \ \
\ \ \
•till
'fill
• rill
•lit
AND
CALCULATING
W
COMPONENT
FROM THE
U FIELD
FROM WHICH
'uw'
WIND VECTOR
PLOTS MAY BE
PRODUCED
THE
MAINLINE PROGRAM
IS RUN WHICH
COMPUTES THE
TRAJECTORIES OF
EACH PARTICLE
STANDARD
PLOTTING
ROUTINES
MAY BE USED
TO PRODUCE
XY
YZ
XZ
{PROJECTIONS
^-DIMENSIONAL
DISPLAYS
Figure '11-4.
continued.
Detailed flow diagram of Kinematic Diagnostic Model,
158
-------
SECTION 12
RESULTS OF THE KDM
TRAJECTORY COMPUTATIONS
The land/lake breeze sequence of 4 September 1974 was used
as a case study in the development of the Kinematic Diagnostic
Model (KDM). Time integrations allowed the construction of par-
ticle trajectories, released at any point in space or time within
the computation grid. Both gases and aerosols can be submitted.
Aerosol deposition rates are crudely approximated by using esti-
mates of terminal velocity (a constant vertical speed imposed
upon the advected particle). Particles impacting upon the surface
were "deposited" and calculations were terminated. Those passing
beyond the computation perimeter could be brought back if wind
conditions so dictated.
The KDM allows the simulation of transport from line sources,
multi-stack sources, area and volume sources, and aerial bursts.
The use of observed data hopefully provides a more realistic
wind field than that computed via complex (and expensive) numerical
models, although notable data voids, such as over water, do indeed
exist. The use of tetroons was in part motivated in an attempt to
validate the calculated trajectories. Figure 12-1 shows the xz
view of the 1048 CDT tetroon superimposed upon the 1130 CDT uw-
vector cross-section plotted on the same scale. The authenticity
of such cycloidal movement within the general lake breeze circu-
lation seems probable, and while the coincidence between observed
159
-------
and predicted may be rather fortuitous in this case, an inspection
of the overall tetroon data (Section 10) suggests a gratifying
degree of similarity in most respects.
It is felt that the KDM works best by far in the frontal
convergence zone and is weakest in the overwater areas where flows
were in fact merely "guesstimated." In particular it is felt that
the data used for this trial underestimated overwater subsidence.
On the whole the KDM is believed to be a useful tool and a starting
point at refining modeling philosophies.
SIMULATION EXPERIMENTS
A gradation of particle sizes was used in the* KDM trajec-
tory simulations, spanning the gaseous state and the fine and the
coarse size ranges. Fall speeds of 0.0, 0.1, 0.4, 0.7, and 1.0
cm sec'1 were used in the various simulations.
LINE SOIIRCF
Particles (0.1 and 0.4 cm sec'1 fall speeds) were simul-
taneously released at six points along a line running southwest-
northeast from an inland point 13 km from the lakeshore to the
most easterly release 2 km inland from the shore. The release
altitude was 10 m AGL. Figure 12-2 (a) of various particulates
released at 0700 CDT.
Initially the trajectories of the 0.1 and 0.4 cm/sec par-
ticles were similar. The particles responded to the land breeze
and stayed within 20 m of the surface (heavier particles descended
to ground level shortly after release) for several hours until
160
-------
being rapidly lifted in the lake breeze convergence zone and ad-
vected first inland then lakeward with the return flow. The
lighter particles (0.1 cm sec"1) showed a greater lateral spread
in their downwind travel than did the heavier particles. The
effect of the two different fall speeds is clearly seen in the xz
projections.
The release of an identical series of particles at 1200 CDT
is shown in Figure 12-3. With the lake breeze convergence zone
by this time nearly 16 km inland, the particles were injected
into the inflow layer and were carried westward for 10-15 km
before being lifted to 1200-1600 m in the frontal convergence
zone. Thereafter they traveled in the lake breeze return flow
layer, at least for some distance before some were affected by
vortex circulations, whereupon opposite paths were taken.
If these trajectories are somewhat analagous to the behavior
of automotive pollutants released that day, it is interesting to
note elevated layers of NOX were seen that afternoon over the
lake. This suggests that the convergence zone acts like a chim-
ney injecting many pollutants into the upper portion of the boun-
dary layer. Only dynamical recirculation and fumigation in the
lake breeze, or washout/fallout mechanism should involve them
again with near surface processes in the short run.
.TI-STACK LAKESHORE SOURCE
In this series of simulations five sizes of particles were
considered. The particles of fall speeds 0.0 and the 0.1 cm sec"1
were released from 5 levels (20, 50, 100, 200, and 300 m) and the
161
-------
0.4, 0.7 and 1.0 cm sec'1 particles at 3 levels (100, 200, and
300 m). The trajectories of the particles were calculated over
a maximum period of 24 hours. Computations terminated if the
particle went beyond the plotting grid, or else at 0800 CDT on
the following morning (5 September 1974).
Figure 12-4 are the xy, xz, and yz projections of the tra-
jectories of the particles simultaneously released at 0700 CDT.
During the first three hours after release, all particles res-
ponded to the land breeze circulation, veering to a northerly
component under the apparent effect of the Coriolis force. The
most westerly pair of trajectories in Figure 12-4 are those of
particles emitted at the lower two levels (20 and 50 m). The
effect of greater height is to cause the trajectories (plumes) to
diverge more rapidly.
During the first three hours after release the trajectories
of the particles from the lower three levels (20, 50 and 100 m)
closely approximates the trajectory of the low-level plume shown
in the photograph in Figure 10-8, showing not only a similar dir-
ectional path but also very little upward movement over the lake
for the first 2-1/2 hours. By 0930, 2-1/2 hours after emission
when the lake breeze was approaching shore, the particles became
progressively influenced by the vertical component in the advecting
convergence zone. After rising to over 700 m, they were carried
in the weaker return flow, the lighter particles with a more
easterly component than the heavier ones. The trajectories of the
heavier particles (Figure 12-4) not only show a greater divergence
162
-------
with downwind advection, but also a significant fall-out during
the night hours.
Heavier particles, released at the same elevations as the
latter three, were similarly advected to the north, also in a
complex series of helical circulations. The greater particle
sizes (greater fall speeds) produced a significant fall-out during
the night, suggesting that an apparent size sorting occurred across
theshoreline.
Figure 12-5 illustrates the phenomenon of split-flow depic-
ted by Cole and Lyons (1972) in which power plant plumes with
differing effective stack heights were found with radically dif-
fering orientations.
Numerous three-dimensional views such as Figure 12-5 con-
firmed and emphasized many features of lake breeze flows including
(i) the helical circulations that occur within the whole
ci rculation eel 1 ;
(ii) the large lateral spread of "plume" produced by the
highly complex trajectories caused partly by the dif-
ferent fall-out velocities of the particles, but to a
greater extent by the vertical wind shear;
(iii) the partial recirculation of gases and aerosols
within the circulation cell.
(iv) the size-sorting phenomena allowing particles from
common sources with differing fall speeds to take often
radically different trajectories.
163
-------
SIZE SORTING OF PARTICLES
In an attempt to investigate size sorting of the particles
within a lake-breeze system, a composite diagram was constructed
from the trajectories of 180 particles. Release heights of 20,
50, 100, 150, 200, and 300 m were chosen at a lake-shore source.
Particles were divided into three groups; group A comprised par-
ticles with fall speeds of 1.0 and 0.7 cm sec'1, group B particles
of 0.5 and 0.4 cm sec'1 and group C, 0.1 and 0.0 cm sec'1. Par-
ticles of each size were released from each of the above noted
heights at the following hours: 0700, 0900, 1100, 1300, and 1500
CDT. The trajectories were traced, and the position of every
particle at 1800 CDT was marked. Figure 12-6 is the xy projection
of the 1800 CDT positions. No easily recognizable grouping could
be distinguished, so the x-distance (i.e. across shore) was divided
into four equal zones, each 20 km wide. The over-lake zone covered
a distance from 30 km to 10 km out from the lakeshore, the shore
zone spanned the area 10 km on either side of the lakeshore, and
the two inland zones covered distances from 10 to 30 and 30 to 50
km, respectively. The distribution of the particle groups within
each zone was totaled and graphed in Figure 12-7. The most obvious
variation occurs with Group A, the largest particles considered,
which display a distince maximum over the lake, diminishing steadily
with inland distance. Group B, particles with intermediate fall
speeds, shows only a slight maximum in the shoreline zone, and no
definite conclusions can be made about their distribution. Group
C, the gases range, shows a maximum concentration in the inland
164
-------
zone 10-30 km from the shore. For the large, particles (fall speeds
greater than 0.7 cm sec-1 ) , a sorting across the shore does seem
apparent, suggesting that with an obvious concentration of par-
ticles over the lake, enhanced deposition in the lake should
occur.
AERIAL BURST
The simulation of an aerial burst was accomplished by con-
sidering an instantaneous release of 52 particles from a position
200 m above the lakeshore. The release time chosen was 0700 CDT
and trajectories were computed for a 12-hour period thereafter.
A release height of 200 m was chosen because this would place the
burst within the zone of maximum inflow velocities and thereby
cause the plume to experience a maximum downwind divergence.
A release time of 0700 CDT was chosen to emphasize the
complexity of flow within a lake breeze system as well as to il-
lustrate the inaccuracies in the present procedure to handle an
emergency situation such as an accidental release of a toxic sub-
stance. In the event of such an explosion, authorities might
consider the prevailing wind direction and speed and calculate an
arc 20° on either side of the mean direction, and then assumed
that 95 per cent of the contaminants would be contained within
this cone advecting at the speed of the mean wind. At 0700 CDT
the mean wind direction was northwesterly and the mean speed es-
timated at 1.2 m sec" . This would imply that the contamination
would travel and slowly diffuse over the lake, averting the
populated shoreline areas especially the downtown area. However,
165
-------
one glance at Figure 12-8 shows that just the opposite would
occur. The trajectories are initially toward the southeast, but
on meeting the lake breeze convergence zone, are convected into
the return flow layer. There they slowly sink into the inflow
layer with a northerly he!ical circulation advecting with the
lake-breeze front and concentrating in a 10 km zone along the
shoreline at approximately the same latitude as the burst origin.
With the possible occurrence of fumigation within the lake breeze
inflow layer it is quite likely that the contaminants could reach
the ground within 6 or 7 hours after release.
The use of a model such as described above to simulate
problems such as explosions, dry deposition trajectories of single
or multiple particle releases, etc., is clear. It remains to
improve the model technique further and to make other simulations
within lake breeze systems.
In summary the KDM shows the lake breeze circulation pro-
duces transport patterns of a highly complex nature. Keen (1976)
presents many more examples showing the futility of attempting
to apply "conventional" methods of air quality analysis to meso-
scale flow systems.
166
-------
SI
«•
0 1»
" t»
•
• '«•
•
S '*"
1»
„ ••
m
- ••
z
4<
s-
t
i
k
1 1 1 1 1
1 1 1 1 1
1 1 1 I 1
1 / / 1 1
1 1 1 1 1
lilt
1 1 1 1
1 1 1 1
' f t i
'III
1 1 1 1
1 1 I I
t t 1 I
11 I I
* > V I
\ » \ \
V \ V \
;--
>::::
>•
\
\
•x
i
\
\
s
•x
I
\
\
•X
\
\
\
^
• 1
V
\
\
\
«
12:00 R VECTORS
\ \ \ \ \ \ \ \ \ \ \ \ \ \ v v v
M/\\\\\\\\\\\\\\Vllv''''''''''''//''''~
i i \\\\\\\\\\\ \\\\\* • ' ' ' ' ' ' i i ' ' ' I I • ' ' ' '
/ \ \ \ \ 1 1 t t 1 1 | t i i i , ' - > ' ' ' i i t i • ' i i • • i •
' \ \ t i i i i i i i i i i t ; • i i t • • i i • • • •
'\\iiiitiitiiii,' » * > > % ' * > « • v * •
' t 1 ///////////// , '. X X V X x x ^, x . x x -
'. 1 I 1 1 / t t / / t / / / / , , , XXXN-XXX^-XX-
;///////////////': xxxv-xxx--- —
/'/'///I////////, x'
/•////// r ///////,, ~x^ x^x~xxx--x^-
, \ ////// fl t 1 t t ,,,,, xsx-~xv-
*\ ////// tf t t 1 1 t , , , , XXX~~XX-
V X ^ ////// f 9 | U.^-f-'.'-i. -.^.W.-.ii;,^.^ x — s v -
x \ - ////// 1 >TT\ \vvvxv f\J.-< ' • ' rr%-— — %-.
\\-'/////' \W\\ A \ \ \ v ^ -^-^ \ '-..,,,.,,-
\ ^ ~'/ /j t '/\ \ vv
-------
Hi
Figure 12-2. Plots of particle trajectories from a six point "line"
source orientated SSW to NNE. Release time was 0700 CDT. xy (50
x 50 km) and xz (50 x 2.5 km) projections are shown and the '0'
mark along the x-distance indicates the position of the shoreline.
(a) shows the trajectories of the particles with a 0.-4 cm sec"'
fall speed, (b) shows the trajectories of the particles with a
0.1 cm sec-1 fall speed.
168
-------
Figure 12-3. Plots of particle trajectories from a line source
with a release time of 1200 CDT. (a) shows the trajectories of
the particles with a 0.4 cm sec-1 fall speed, (b) shows the tra-
jectories of the particles with a 0.1 cm sec"1 fall speed.
169
-------
IRMEIHOK DIM 0. MM) .1 ITMII
/>Y /
/ / ! I /
// \ I/,-"
LflKESHORE 0700 0. RNO .1 STBKES
Figure 12-4. Plots of particle trajectories from a multistack
source on the shore with release heights of 20, 50, 100, 200, and
300 m. Release time was 0700 CDT and trajectories were computed
for an 18 hour period, (a) the xy, yz, and xz projections of the
particles with fall speeds of 0.4, 0.7, and 1.0 cm sec-1 . (b) the
xy, yz, and xz projections of the particles with fall speeds of
0.1 and 0.0 cm sec" ' .
170
-------
«v--,
Figure 12-5. Particle trajectories from a multistack source with
a release time of 0930 CDT. (a) shows the trajectories of the par-
speeds of 0.4, 0.7, and 1.0 cm sec-"1, (b) shows
the particles with fall speeds of 0.1 and 0.0
a 3-dimensional view of all the trajectories.
tides with fall
the trajectories of
cm sec-', (c) shows
171
-------
•V
-*»•" . -•••• -"-,' -"•,' . -"-f
mm • • •
." , , .If;' .
*••
Figure 12-6. The xy projection of the positions of a series of
particles at 1800 CDT. Particles were released from a multistack
source on the lakeshore (marked by a large dot). The particles
were grouped into 3 size ranges with the largest squares showing
the positions of the particles with fall speeds of 0.7 and 1.0
cm sec-', the middle size squares mark the positions of the par-
ticles with fall speeds of 0.4 and 0.5 cm sec-1 and 0.0 cm sec'1.
172
-------
DISTRIBUTION OF
PARTICLES BY SIZE
ACROSS LAKE SHORELINE
km SO
Figure 12-7. A histogram of the relative distribution of particles
according to size groups across the lake shoreline. Each zone was
20 km wide and any particles having a greater than 30 km over-lake
fetch or 50 km over-land fetch were disregarded.
173
-------
Figure 12-8. A 3-dimensional plot of a simulated aerial 'burst
Particles were simultaneously released from a cube 200 m above
the lakeshore at 0700 CDT and trajectories shown represent the
travel over a period of 12 hours.
174
-------
SECTION 13
LONG RANGE POLLUTION TRANSPORT
GENERAL COMMENTS
An integral part of the rationale behind the Clean Air Act
of 1967-1970 was the assumption that air pollution measured at a
given point emanated from "local" sources. Air Quality Control
Regions (AQCRs) were established on the principle that local
political jurisdictions could best cope with emissions from local
sources which cause their own peculiar pollution problems. Since
1970 however, evidence has been steadily mounting that air pol-
lution can be anything but a "local" problem. It now seems to be
a certainty that certain pollutants, in particular photochemical
oxidants and sulfate aerosols can and do travel considerable dis-
tances, in many cases in excess of 1000 kilometers, while under-
going complex chemical and physical transformations en route.
Thus a monitor situated at any given point registers concentra-
tions reflecting the result of local emissions, plus mesoscale
and even synoptic scale transport and transformation, all com-
mingled in a manner making it virtually impossible to separate
the factions resulting from each mechanism. At least in the case
of these two major pollutants, the term "natural background" has
become a harder and harder one to define. Widespread occurrences
of elevated rural ozone levels (Coffey and Stasiuk, 1975) attest
to the viability of this chemical while traveling long distances.
Although ground level S02 levels in cities are declining slowly
175
-------
(A1 tshul 1 er, 1976) increased sulfate monitoring in rural areas
shows no decline, and if anything perhaps a slight rise over the
last few years. Thus the reduction in near-ground level releases
of S02, often by injection of S02 into the upper portion of the
boundary layer via high stacks, is causing a new problem to re-
place an older one. Rural sulfate values several times the pro-
posed standard are now being routinely monitored over large areas
of the nation simultaneously (EPA, 1975). The Sulfate Regional
Experiment (SURE) as described by Hidy, et al. (1976), is just
one of several efforts now focusing on the problem of meso- and
synoptic-scale air pollution. The Midwest Interstate Sulfur
Transport and Transformation program (MISTT) in the St. Louis
area is yet another. Already individual power plant plumes have
been found traveling in a cohesive manner for distances over
several hundred kilometers. The reality of meso- and synoptic-
scale transport of pollutants can no longer be denied. Moreover
these events do not occur on an unfrequent, but rather on a
routine basi s.
Various aspects of the research performed at the UWM Air
Pollution Analysis Laboratory, as described below, adds several
more facets to this complex phenomena.
SATELLITE DETECTION OF AIR POLLUTANTS
With the launching of Tiros I in 1960, a succession of over
40 meteorological satellites have relayed a vast amount of infor-
mation gathered by an impressive array of sensors. Sensor reso-
lution, using visible light, infrared and other wavelengths, has
176
-------
consistently improved over time. The first satellite imagery,
with resolutions on the order of 10-20 nautical miles, were
designed to see large-scale cloud patterns, with image dynamic
range so poor as to totally preclude the observation of such
subtle features as smoke and haze. The ability of an orbital
sensor to detect visible air pollutants however was quickly
noted by 70 mm color hand-held photographs taken by astronauts
as early as 1960 (see Figure 1-1). By the late 1960's, sensor
resolution and spectral sensitivity had improved to the point
where plumes from volcanic eruptions, Saharan dust storms, and
the snoke from forest and agricultural slash burning fires could
be seen. Even these, had they been interpreted in the proper
light, would have been convincing proof of the frequent occurrence
of long-range pollutant transport. It was not until the advent
of NASA's Landsat-1 satellite however that observations of longer
range pollutant transport became more routinely available. The
spatial resolution of the satellite was better than 100 meters,
and in addition, worked in four spectral bands: Band 4 - 0.5-0.6
m; Band 5 - 0.6-0.7/(m; Band 6 - 0.7-0.8^m; and Band 7 - 0.8-1.1
. Lyons and Pease (1973) published Landsat images of the
southern basin of Lake Michigan which very clearly showed plumes
of smoke from Chicago-Northern Indiana sources advecting with
relatively little dilution for more than 60 km northeastward
over the relatively cold lake. It was found that for various
reasons smoke plumes could be best detected in Landsat MSS Band
5 (the "red" band). Also of interest was that the same smoke
177
-------
plumes were not as readily visible over the mottled land surface
inasmuch as inherent image contrast between ground smoke was
exceedingly poor (Lyons, 1975).
It had long been suspected by Wisconsin DNR officials that
the Chicago metropolitan area was a major source of pollutants
measured in the Southeastern Wisconsin AQCR. The original NAPCA
implementation documents for this region suggested that up to 10%
of the suspended participates had Chicago as their original source
There was no easy way to prove this correlation however, in spite
of the occasional observations of dense clouds of iron-oxide-red
smoke drifting into the Milwaukee area from the southeast during
stable cross-lake flow conditions. Figure 13-1 is an excellent
example of a Landsat image of interregional pollutant transport.
In this particular case the plumes drifted on southwest winds
over a colder lake for distances of over 160 km with relatively
little lateral diffusion. A plume CT"y of 4.5 km measured at 60
km downwind suggesting a Pasquil1-Gifford stability Class E. It
also appears that as distance downwind increases, the rate of
lateral increase in plume width decreases. In this particular
example, plumes crossed the southwestern Lake Michigan shoreline
in the vicinity of Benton Harbor. Given the meteorological condi-
tions of this date, it seems almost certain that the plume matter
fumigated to the surface after a passage of only five to ten
kilometers inland. Had local control officials been monitoring
in this vicinity, they might have been quite perplexed at the
high levels of suspended particulates, and other pollutants con-
tained within these industrial plumes. Quite clearly there is
178
-------
no way for the receptor AQCR to defend itself against the "sins
of emissions" of the source AQCR, particularly if they are unaware
of the phenomenon just described.
The same plumes have been implicated in regional-seale inad-
vertent weather modification. Specifically, Chicago industrial
complex pollutants have been postulated as a cause of the LaPorte,
Indiana precipitation anomoly (Changnon, 1968). As described by
Lyons (1974), these plumes were "caught in the act" of apparently
modifying cloud streets formed during a period of cold air advec-
tion over the southern end of Lake Michigan on 24 November 1972(^9-13-2)
While the satellite image does not indicate the active agents in
the cloud modification (heat, moisture, cloud nuclei, ice nuclei,
etc.) there seems to be little doubt that cumulus clouds began
forming out of the smoke plumes considerably closer to the shore-
line and that those cloud lines were also better developed. Great
Lakes water management via cloud seeding has been discussed for
more than five years. If large-scale inadvertent seeding is oc-
curring from pollution sources, its effect must be clearly under-
stood in order to assess the overall impact of advertent weather
mod ificat ion.
While subjective interpretation of Landsat and other satel-
lite imagery is of tremendous value, it is generally preferable
to use objective machine-oriented data analysis. As described
by Lyons, Keen, and Northouse (1974), automatic image interpre-
tation techniques were applied with success to Landsat digital
data tapes. The MODCOMP 11/25 computer system developed by the
179
-------
UWM Robotics and Artificial Intelligence Laboratory interpreted
Landsat tapes using pattern recognition techniques, specifically
the cluster analysis algorithm of Eigen and Northouse (1973).
Figure 13-3 shows the results. In this example water is shown
by a dash (-), all land spectral signatures are suppressed and
therefore blank, and those 60 by 80 meter data pixels having a
spectral signature thought to be smoke are shown as filled in
circles (®). More sophisticated techniques would be needed in
order to detect smoke over land surfaces having a wide spectrum
of spectral characteristics.
On 20 August 1972 (Figure 13-4) an exceptionally well -
developed lake breeze cloud system was noted over the eastern
end of Lake Erie and the southern shore of Lake Ontario, along
with a "smudge" over the eastern end of Lake Erie west of Buffalo,
New York. This was interpreted to be the signature of smoke layers
in the developing lake breeze return flow, a graphic example of
the phenomenon described in the Chicago area by Lyons and Olsson
(1973). As pointed out by Raynor, Hays, and Ogden (1974) any gas,
aerosol , or pollen would be transported in a similarly complex
manner in a lake or sea breeze circulation cell.
THE CHICAGO URBAN PLUME
During the 1974 field program conducted at Waukegan, Illinois,
the major effort was concentrated on the local power plant plumes
as described in the above sections. On 28 June 1974 a serendipi-
tous measurement of the Chicago urban plume was obtained. This
180
-------
case study is described in considerable detail by Lyons and Rubin
(1976).
On this day, there was a brisk southeast gradient flow.
Winds gradually changed from south through southeast to east,
from daybreak through late afternoon. Extensive fumigation of the
Waukegan power plant plume was measured. Intense elevated inver-
sions were present over the lake from 50 to above 300 meters
throughout the entire day. SMS-1 satellite imagery showed a
cumulus-free zone extending several tens of kilometers inland
during late afternoon. Aircraft traverses were run in east-west
traverses along the Illinois-Wisconsin border throughout much of
the day. The £ ''3 field showed that the TIBL reached a semi-
steady state after 1100 CDT, with the top of the TIBL extending
to approximately 1000 meters after 20 km of inland fetch.
During the flight (1006-1154 CDT) the S02 plumes from the
Waukegan power plant and smaller local ground sources were noted.
As shown in Figure 13-5 a very concentrated plume of SOg with peak
values above 20 pphm was present offshore at approximately 400
meters. The reasonable presumption is that this plume had ema-
nated from the general vicinity of Gary, Indiana. More strikingly
however, embedded within the overall haze was found an urban scale
plume from the Chicago metropolitan area. It began approximately
15 km inland and extended to a sharp western edge approximately
28 km inland. The Chicago plume consisted of elevated values of
SOp and fine particulates(and a noxious odor).
The Chicago urban plume h.ad clearly traveled as a cohesive
mass during the early morning hours. It was initially very
181
-------
shallow, probably trapped beneath the shallow nocturnal inversion
of the prior night, growing directly proportional to the pene-
trative convective mixing after sunrise. Aircraft measurements
showed it rose almost in unison with the height of the l.oe1/3
isopleth. The peak concentrations of S02 within the plume were
145 pphm, or about 120 pphm above apparent regional "background."
The fine particle concentration was a factor of five above values
present within the boundary layer but outside the plume during
mid-day. Visibilities outside the Chicago plume were typically
15 to 25 km, but only 5-10 km within its core. During the early
afternoon the plume became more and more vertically and laterally
diffuse and was found further and further westward. By the time
the afternoon traverses were run, the Chicago plume was not noted
within the first 30 km of the shoreline.
Using mean pibal winds within the mixed layer, a back tra-
jectory was estimated. Air crossing the Wisconsin border at 1200
CDT was calculated to have been over industrialized extreme north-
western Indiana at 0600 CDT (100 km, 152° from Waukegan). The
plume matter continued into Wisconsin where maximum hourly ozone
levels (probably also associated with this plume) of almost 10
pphm were recorded between 1500-1700 CDT at Poynette, Wisconsin,
some 35 km north of Madison. The amount of S0? entering the state
of Wisconsin was considerable. An integration of plume matter
contained within the 4 pphm isopleth yielded transport rate of
25 tons/hour at 1100 CDT. This would compute to an annual rate
of 22,000 tons/year.
182
-------
From this observation alone it is impossible to assess the
annual average impact upon air quality of Wisconsin AQCRs by the
Chicago metropolitan area. But given the relatively high fre-
quency of southeasterly gradient flow in this area, and for those
pollutants where federal standards are just marginally being
exceeded, at least legally, the impact might be considerable.
MESOSCALE TRANSPORT OF PHOTOCHEMICAL OXIDANTS WITHIN LAKE BREEZES
The traditional "recipe" for photochemical "smog" (which
is primarily ozone) that is given in textbooks includes, 1) large
quantities of precursors, reactive hydrocarbons (RHC), and the
oxides of nitrogen (NO ), 2) strong solar radiation, 3) rela-
s\
tively high ambient temperatures, 4) relatively light wind speeds,
and 5) limited mixing depths. It is assumed the RHC and NO pre-
x
cursors react in the presence of sunlight within some unspecified
distance, perhaps several kilometers downwind of the sources,
producing a peak of photochemical oxidants shortly after noon
with a gradual fall-off to near zero by and during the following
nighttime hours. As pointed out by Cole and Lyons (1972) the
high emissions of primary pollutants in the Gary-Chicago-Milwaukee
corridor combined with lake effects make this region a prime
suspect for photochemical oxidant episodes. This was confirmed
during 1971, when EPA monitored for photochemical oxidants 99
days at a suburban site 2 km inland from Lake Michigan, and 10 km
north-northwest of the central business district. Maximum hourly
averages of 19 pphm were reached - the highest for any city over
183
-------
200,000 population monitored. On 18 days readings in excess of
the 8 pphm hourly average federal standard were observed. Most
interesting though was the fact that fully 94$ of the exceedences
occurred with winds from west-southwest through southeast, and 67%
from the south through southeast alone. Inspection of the data
showed frequent sharp rises in ozone after the passage of lake breeze
fronts. Furthermore peak values frequently occurred during late
afternoon and sometimes even early evening after sunset. These
data were among the first to suggest that the source of southeastern
Wisconsin't ozone might be at least in part, the Chicago metro-
politan area.
During 1973 the Wisconsin DNR installed a network of air
quality stations around Milwaukee and also Racine, Wisconsin, near
the lakeshore, 30 km south of the Milwaukee CBD. Also an additional
"rural background" installation was made at Poynette, Wisconsin.
In a 47-day period during late summer 1973, at least two
or more stations in the network (excluding Poynette) reported
readings greater than 8 pphm on a total of 27 days. On nine of
these days episode alert dosage thresholds (40 pphm h) were ob-
tained at one or more sites. Hourly averages as high as 27 pphm
were measured at Racine. A pollution wind rose, based on the MKE
mean resultant daily wind, showed that 92% of the days with stan-
dard exceedences had daytime winds from the southwest through east-
southeast. Racine frequently reported the highest ozone readings,
which is somewhat odd if one considers that the area only has
approximately 12% of the RHC and NO emissions of Milwaukee
j\
184
-------
County. During 1974 and 1975 even higher values have been recor-
ded in the Racine area. Thus with the noted strong correlation
between high ozone levels and southerly winds the Chicago and
northern Indiana industrial complex looms as the immediate can-
didate for the source of a large fration of the ozone measured
in southeastern Wisconsin.
Some local patterns in ozone distribution however also
emerged (Lyons and Cole, 1976). Figure 13-7 shows data taken
from six monitoring sites during the 1973 summer. It appears
that ozone levels near the immediate shoreline (less than 1 km)
are relatively low, and then there is a sharp rise in about 1-4
km inland, with a gradual fall-off further to the west. More
complete 1974 data confirm this pattern. Esser (1973) used
specifically chosen bioindicator plants (Bel W-3 tobacco) to
monitor photochemical oxidant patterns in the greater Milwaukee
area (Figure 13-8). Analysis of his results shows that damage
is relatively infrequent in the first kilometer form the shore-
line but then rises sharply as one proceeds inland. A band of
25% or more damage runs north-south parallel to the shore from
approximately 1.5 to 8 km inland, then there is a gradual fall-
off further inland. Since higher ozone values are usually found
with an onshore component of the wind, why are the values near
the immediate shoreline lower?
A mechanism is proposed to explain this pattern. A typical
lake breeze day dawns with a weak land breeze in progress (Figure
13-9). Pollutants produced by morning rush-hour traffic drift
185
-------
offshore over the lake. Since the air becomes warmer than the
lake water after sunrise, a strong low-level conduction inversion
is present over the lake, perhaps 100-200 m in depth. Ship wire-
sonde observations (Bellaire, 1965) indicate that the offshore
flowing air and its pollutants should tend to flow up and over
the shallow surface inversion leaving the air near the lake re-
latively "clean." Lack of penetrative convection over the cold
lake allows the air above the inversion to become stratified, and
sunlight acts on the primary pollutants, forming ozone. The
newly produced ozone cannot penetrate the surface inversion, and
little scavenging of the gas occurs and concentrations increase.
After the onset of the lake breeze inflow, the onshore flowing
air has little ozone near the surface, with the bulk of it being
stratified from perhaps 150-500 meters aloft. In a manner ana-
logous to continuous fumigation of elevated shoreline plumes, the
ozone aloft is intercepted by the deepening TIBL as the airstream
moves inland. The ozone and other pollutants are fumigated to
the surface in relatively high concentrations beginning 1 or 2 km
inland. After perhaps 5 km inland fetch, the bulk of the ozone
has been mixed within the deepening TIBL. Rapid destruction now
begins by contact with the surface plus the addition of NO. Simi-
lar patterns have been discovered at the inland boundary of the
west coast marine inversion by Miller and Ahrens (1969), Edinger,
et al . (1972), and Blumenthal et al. (1974). While Figure 13-9
assumes that the air flow is purely two-dimensional (in the plane
of the paper) it provides a reasonable explanation for the obser-
ved patterns.
186
-------
The experience in southern California proves photochemical
oxidants can indeed be transported for long distances. Edinger,
et al . (1972) show tree damage in mountainous areas 120 km distant
from the source of the primary pollutants. Blumenthal et al .
(.1974) computes over 100 tons per hour being advected from the Los
Angeles basin to areas further east. That the Chicago metropolitan
area could be acting as giant source adding to locally produced
photochemical oxidants in southeastern Wisconsin is therefore not
surprising. But since lake breezes have a highly complex trans-
port wind as described above, the exact mechanism by which the
transport occurs is hard to specify.
Figure 13-10 represents hypothetical air trajectories used
to explain an ozone episode in southeast Wisconsin on 17 July
1973. Trajectories A and B represent air (and pollutant) motions
near the surface which moved roughly northward during the day,
totally unaffected by the lake. Trajectory C marks an air parcel
released near the ground around 0800 LST, which drifted north-
ward only to be undercut by the inland rushing lake breeze. Tra-
jectory D represents a low-level release around sunrise. With
the formation of the lake breeze, this air, which by now has
developed a considerable photochemical oxidant burden, crosses
the shoreline at Racine about 1-2 km inland, and fumigation of
ozone to the surface begins (heavy line). Trajectory F is some-
what similar in behavior except it leaves the shore perhaps
around 0900 LST. This parcel may fumigate further south around
Kenosha. The air then rises into the return flow layer near the
187
-------
lake breeze front and after traveling northeastward, it may sink
back down into the upper portion of the inflow to once again
fumigate while passing inland. In the case of trajectory E, this
parcel is released near the shoreline within the lake breeze in-
flow during late morning. It spirals northward and finally fumi-
gates over Racine late in the day.
In reality, there are an infinite variety of possible tra-
jectories and many of them may be variations of those above. The
results of the KDM would suggest the air motions, if anything,
are even more complex than described in Figure 13-10. This mech-
anism is consistent with the simpler two-dimensional view of the
lake breeze in Figure 13-9 inasmuch as it still allows for the
fumigation mechanism to produce the inland ozone maximum. It
also explains the late afternoon and even nighttime elevated
ozone readings in Milwaukee and Racine. Parcels of air arriving
in Wisconsin come from a spectrum of distances and release times.
Air streaming onshore after dark, which left the Chicago area
during morning rush-hour, can cause high surface ozone readings
simply due to mechanical mixing of the air from above the 100 m
level to the ground. This is a "mechanical fumigation" which may
well be aided and abetted by "urban heat island fumigation." The
lack of appreciation by some officials of the complex meteorology
involved in ozone transport is emphasized by Altshuller (1975)
who discovered some monitoring reports where high nocturnal ozone
values were sumarily dismissed as inaccurate per se.
When the southeasterly gradient flow is strong enough so
188
-------
that no lake breeze circulation is present, relatively high ozone
levels still occur in the nearshore areas of southeastern Wiscon-
sin. The primary pollutants released at the southern end of the
lake still travel with relatively little vertical dilution across
a cold lake, reaching the downwind shoreline with a considerable
ozone burden. In this case the air trajectories are far less
convoluted.
LONGER RANGE PHOTOCHEMICAL OXIBANT AND SULFATE TRANSPORT
The distinction between roesoscale and synoptic scale trans-
port of pollutants may be almost semantical. However in the case
of photochemical oxidants, it may be appropriate for photochemical
reasons to classify as synoptic transport those events taking
more than one day's time.
The literature is now replete with reports studying wide-
spread rural elevated ozone levels. Rural ozone monitors within
the state of Wisconsin have not been exempted from this phenomenon
During 1973 the Wisconsin DNR ozone monitor at Poynette, which
is 140 km west of Lake Michigan, began collecting rural "natural
background" data. During the 1973 summer at least 189 hours of
episode alert dosages were recorded on a total of 24 different
days, the first as early as 24 May. While the values were not
as high as in the lakeshore region (maximum only 16 pphm), 8
pphm was frequently exceeded. The pollution wind rose (Figure
13-11) based on the Madison, Wisconsin resultant daytime surface
winds, shows once again a strong directional preference, with
84% of the alert level hours occurring with winds from the
189
-------
southwest through the east-southeast. The 41% of the events oc-
curring with winds from the southwest quadrant cannot easily be
attributed to the Chicago, Milwaukee, or Madison area urban plumes.
On 27 August 1973, typical of days with elevated rural ozone
levels, there was a general southwest surface flow through the
central United States around a high pressure system moving slowly
through the Appalachinas. The Poynette site recorded 12 consecu-
tive hours of alert levels beginning at 1000 1ST. A backtrack of
air trajectories showed these air parcels had traversed such major
urban areas as Kansas City, Dallas-Fort Worth, and Houston in the
previous three days. Remembering that the low-level jet stream
mechanism (Bonner, 1968) has high velocity wind cores exceeding
25 m/sec within the nocturnal boundary layer, it is quite possible
for materials to be advected more than 1000 km within a 12-hour
period. In effect, large metropolitan areas within the eastern
2/3 of the United States can be considered as a patchwork of
large volume sources of RHC and NO , which is vertically distrib-
J\
uted within the daytime mixed layer. Downwind, in tie absence of
additional NO emissions, ozone forms, or if the sun has set, the
primary pollutants stay in relatively high concentrations within
the urban plumes. The development of the nocturnal radiation in-
version allows frictional decoupling from the surface producing
the low-level wind maximum aloft, and also encapsulating the
pollutants aloft in what was the prior day's mixed layer. This
also prevents further input of fresh NO. By sunrise the next
morning either the primary pollutants begin forming oxidants or
an existing pool of ozone is present aloft, typically stored in
190
-------
a layer below the synoptic scale subsidence inversion and above
the top of the prior evening's radiation inversion break-up fumi-
gation. It has been noted on numerous occasions in Milwaukee
during southwesterly flow that ozone suddenly increases at the
surface during mid-morning at the time an acoustic sounder shows
the rapid burn-off of the inversion with the development of con-
vective plumes.
In effect, entire air masses can become polluted. Ripper-
ton et al. (1974) made airborne ozone measurements during anti-
cyclonic conditions over North Carolina, Ohio, West Virginia,
and Pennsylvania, finding widespread ozone levels near or above
8 pphm rather uniformly vertically mixed within the daytime mixed
layer. There was no apparent specific source of the pollutants.
Air masses characterized by elevated ozone levels also have an
appreciable degree of haziness. The Landsat-1 satellite has
observed these widespread turbidity episodes. Figure 13-12a is
a picture of the southeastern shoreline of Lake Ontario on a late
winter day when the region was covered by a very clean cP air
mass. The exact same region was monitored on 1 September 1973
(Figure 13-12b) during a widespread air stagnation episode. Very
light southwesterly winds covered the region. Sparse precipitation
and relatively shallow mixing depths allowed effluents from much
of the Ohio Valley and Appalachian region to commingle for several
days producing a widespread air mass turbidity which made the
ground virtually invisible to the Landsat sensors. Ernst (1975)
has noted in SMS imagery large hazy masses of air drifting off
191
-------
the U.S. east coast. The occurrence of widespread air mass pol-
lution was noted by Hall et al. (1973) in their study of an
August 1970 air stagnation. The movement of a large volume of
air characterized by visibilities severely reduced due to smoke
and haze for several thousand kilometers was studied. As shown
by Altshuller (1976) and Wilson et al. (1975), the haziness is
largely associated with sulfate aerosols formed by complex gas
phase reactions in the atmosphere. These sulfate episodes, which
could also be as easily termed "smog blobs" due to their frequent
coincidence with elevated ozone levels, are associated with mT
anticyclones, and their attendent synoptic scale subsidence, lack
of precipitation and therefore wash-out of pollutants, low visi-
bilities, very warm temperatures, high dewpoints, low mixing depths,
and light transport winds. It is noted by Husar et al . (1976)
that these episodes can be present over the eastern 2/3 of the
United States for up to two weeks at a time. The latter part of
June and early July 1975 was characterized by a significant sul-
fate episode in the central and eastern U.S. Most amazingly is
the fact that standard SMS satellite imagery was able to detect
this event (Figure 13-13). Extremely large areas of turbid air
covering many states could be tracked on a day-to-day basis by
inspection of the SMS imagery.
Fortuitously and coiricidentally the MISTT program was con-
ducting field studies in St. Louis during that period. Mid-day
visibilities in eastern Missouri deteriorated to about 4 km
during the peak of the episode. The low visibilities were as-
sociated with relative humidities as low as 40% elimanating
192
-------
consideration of fog or "water haze" situations. Aircraft and
ground instruments showed degraded visibility to be widespread
and not a local phenomenon. At the peak of the episode, total
suspended aerosol mass was not unusually high, approximately
lOO^gm/m3. Chemical analyses however showed the total sulfate
aerosol content to be as high as 35 >ugm/ni3, nearly three times
the proposed national standard for sulfates. Furthermore, of
that, nearly 40% appeared to be comprised of sulfuric acid aerosol
A conversion of Service A teletype visibility reports to equiva-
lent bscat values (Husar, 1975) showed a very high degree of
correlation between the satellite photograph turbidity producing
reduced ground contrast and the geographical extent of the turbid
air mass (Figure 13-14).
The rather amorphous and slowly changing shape of the
"smog blob" makes the origins of the aerosols hard to determine.
An attempt at plotting the path of the center of the stagnant
high pressure cell was inconclusive due to minute pressure
changes resulting in an apparent erratic translation from chart
to chart. More illustrative however is a plot of the 1020 mb
isobar position for each 1200 GMT surface map, 25 June through
30 June 1975 (Figure 13-15). Since the situation was so nearly
steady-state, this ensemble can be thought of as a crude approxi-
mation of the trajectories of boundary layer air parcels. Note
the general appearance of flow from the high power plant density
region of the Ohio Valley into the Midwest where the "smog blob"
is most obvious on 30 June. As the haze area gradually pene-
trated into the Minneapolis/St. Paul area, surface visibilities
193
-------
degraded from 40+ km on 27 June to 5 km on 29 June. Elevated
oxidant values throughout this entire region were also simultan-
eously recorded. An important finding is that initial studies
suggest that the degree of haziness seen on the photograph is
surprisingly wel1-correlated to mid-day surface visibility reports,
which in turn has a somewhat consistent relationship to sulfate
aerosol concentrations (Husar, 1975).
194
-------
Figure 13-1. Landsat-1, Band 5 (0.6-0.7/um) image of Chicago-Gary
area, 1003 LSI, 14 October 1973. Smoke plumes advect across the
relatively cold lake showing little diffusion, only to fumigate
on downwind shoreline near Benton Harbor, Michigan.
195
-------
CHICAGQ
31
10 n
27WMDW
4
Figure 13-2. Landsat-1 image, Band 6 (0.7-0.8 ^um), over southern
Lake Michigan, 1003 LSI, 24 November 1972. Four major particulate
plumes emanate from shoreline sources. Cumulus clouds form over
relatively warm lake, and those arising out of smoke plumes form
closer to shore and become larger and brighter.
196
-------
Figure 13-3 Computer processed Landsat-1 digital data small
oct'o eTl97a2keVJ°ng l°Ut^n Uke M1chl>n shoreHne 1003
October 1972. Lake water indicated by (-), all land sia-
natures are suppressed (and therefore blank)! and smoke is denoted
?Li ^rJ' T e man:niade Peninsula and breakwater protrudinq into
IhP An Pjume^on the left is from a steel mill complex, while
the one on the right is from a cement plant.
197
-------
Landsat-1 image, band 6, of 20 August 1972 showing
Lake Ontario and the eastern end of Lake Erie. A
lake breeze rings both lakes (with the front indicated)
an enlargement of the Buffalo, N.Y., area over which
return flow layer can be seen extending over the lake.
Water temperatures (°C) are shown for Lake Ontario and the 1000
CST hourly aviation weather data is plotted. (Each barb equals
2 .5 m/sec) .
Figure 13-4.
a portion of
well defined
The inset is
smoke in the
198
-------
to
ID
CU T|
(T> ->•
-S tQ
O C
to -S
o ro
fD
o
oo to
i cu
(D
to oo
N cn
fD *•
-s n>
su x
3 O
tQ ft)
ro T3
• t-t-
o
•s
o
o
3
o
fD
3
c-h
-S
ai
c-t-
—i.
o
3
to
HEIGHT ABOVE SURFACE JM)
HEIGHT ABOVE SURFACE (FT)
CU
-S — ' -S
O CU fD
3 3
ct- rt-
£D>» 3"
OJ fD
-S — '
^< O O
O 3"
<-~.fu -i.
S — 'O
— ' fD ~n
— • CU -J.
on to to
rl-
3- CO O
n> -s
oa o"
O S fD
O fD
— I to -u
- t-h OJ
I
rv>T3 cn
O> cu .
c+
C-. 3-
£= 3=
3 Cu _i.
fD — ' -S
O O
— ' 3 -S
IQIQ Co
o
O l/> "O
30—1
o c c
fD -S 3
3 O fD
rh fD v
~S to
rt- 3-
O 3
3 O-TD
to —'
T3 C
Co -S 3
cr o fD
O cr
< cu -h
fD cr -5
—' O
CU 3C
T3 Eu
T3 "O C
3- — ' 7C
3 C fD
^3 CO
• fD CU
3
fD -
3 co
CU
3T3
CU O
n-s
-j. fD
3 -S
IQ
fD 3
to fD
c s: cu
-S -..(/>
-h fD
<-t- ->' 3
O 3 c+
I to
—' i—i
cn —" O
O —• -h
O ->•
3 OO
3 O O
fD -"fM
r+ to
fD *-»
-5 to -a
to r+T3
• CO 3r
rt- 3
fD *—
O
—' D"O
fD O -S
Cu -s O
-S Q. -h
—' fD -•.
<< -S —'
<• fD
< CL
to O ->•
-1-03
U~ CTi
—' i cu
fD 3
HEIGHT ABOVE SURFACE (M)
HEIGHT ABOVE SURFACE (FT)
-------
1
f-
li
i -r
^ Al
&o"
It
if
P
6°
28
24
20
16
12
8
4
O
SUMMER
1973
A
• •
•i
'Highest / A , ^ . llTL-m i
^ ' / r z.**-ii.*5Km — i
. peak/ data gap (no station)
/
• /
• 0 ^
•
/Ve. inst ^ — . .
L (U ^B
J^m > •
«o Si = ^ '
" 8 .£5^ 1 f '
c y ^*'> o^ «
0 1.6 3.2 11.3 12.9
Kilometers from lake
July through
3.
Figure 13-7. Data from six monitoring sites from 15
31 August 1973 (except Jones Island which began 4 August). Data
plotted as a function of distance from the lake. Dots indicate
average instantaneous peaks measured on all days. Triangles are
the highest instantaneous peaks recorded. Squares denote number
of days with readings >8 pphm.
200
-------
BIO-INDICATOR SITE
AIR QUALITY
MONITORING SITE
Lake
Michigan
1234
kilometers
(After Esser, 1973)
Figure 13-8. Isopleths of percentage of sensitive plants experi-
encing photochemical o^idant damage in Milwaukee during the summer
of 1972. Wisconsin DNR air quality monitoring sensor indicated
by numbers in boxes. The "X" locates the EPA monitoring site
active in the summer of 1971. MKE is the National Weather Service
station at Mitchell Field.
201
-------
MORNING RUSH HOUR -8 am
MID AFTERNOON ~ 2pm
FUMIGATION ZONE
Figure 13-9. Schematic showing the mechanism by which a lake
breeze causes elevated ozone levels in a narrow band parallel
to the shore but several kilometers inland. This diagram refers
to idealized conditions when the lake breeze flow is purely two-
dimensional (in the plane of the paper).
202
-------
OSHKOSN
MILWAUKEE
___ Michigan
oooooo ABOVE CONDUCTION LA'
• TETROON RELEASE
O TETROON RECOVERY
I
0 kilometers 50
Figure 13-10. Hypothetical trajectories of air parcels (and
pollutants) along the western shoreline of Lake Michigan during
a lake breeze event.
203
-------
10
POYNETTE, WIS.
Summer
1973
Figure 13-11. Pollution wind rose for Poynette for days on which
ozone exceeded episode alert levels in summer of 1973. Winds
measured at Madison, Wisconsin NWS station. Based on DNR report.
204
-------
Figure 13-12a. Landsat-1 image, Band 4 (0 . 5-0 .6/401) , 0945 LSI,
23 March 1973, of the eastern Lake Ontario region, on a cloud-
free day with low atmospheric turbidity. Snow cover is spotty
through the.region. Rochester, New York indicated by ROC.
Figure 13-12b. Identical geographic area as Figure 13-12a, Land-
sat-1 Band 4, 0945 LSI, 1 September 1973 on a cloud-free day but
with highly polluted atmosphere associated with synoptic scale
air stagnation sulfate episode.
205
-------
Figure 13-13. Portion of an SMS-1 visible image (1 NM resolution)
taken 1445 GMT, 30 June 1975, Atmospheric turbidity over the central
U.S. enhanced by re-photography of the original. Datalog print
courtesy National Weather Service Forecast Office, Twin Cities, MM.
206
-------
Figure 13-14. Surface synoptic chart, 1200 GMT, 30 June 1975, with
contoured shadings representing areas of reduced visibility (and
corresponding back-scattering coefficients) acquired from Service
A network data at 1800 GMT.
Figure 13-15. Location of the 1020 mb isobar on six consecutive
1200 GMT synoptic charts during the period 25 June - 30 June 1975
207
-------
REFERENCES
Altshuller, A.P., 1976: Regional transport and transformation of
sulfur dioxide to sulfates in the U.S., J. Air Poll. Control
Assoc., 26, 318-324. ~
Altshuller, A.P., 1975: Evaluation of oxidant results at CAMP
o!te?JU»the United States. J. Air Poll . Control Assoc.,
25, 19-24. ~ ~
Angell, J.K., 1975: The use of tetroons for probing the atmospheric
boundary layer, Atmos. Tech., 7. 38-43.
Baralt, G., and R.A. Brown, 1965: The land and sea breeze: an
annotated bibliography, The University of Chicago, Satel1ite
and Mesometeorology Research Project, 61 pp.
Bellaire, F.R., 1965: The modification of warm air moving over
cold water, Proc. 8th Conf. Great Lakes Res.. Intern. Assoc
Great Lakes Res. , 249-256.
Bierly, E.W., 1968: An investigation of* atmospheric discontinui-
ties inducted by a lake breeze, Ph.D. dissertation, Univ.
Michigan, Dept. Meteor, and Oceanogr., 150 pp.
., and E.W. Hewson, 1962: Some restrictive meteorological
conditions to be considered in the design of stacks, J.Appl
Meteor. , 1, 383-390. " —
Blumenthal, D.L., W.H. White, and R.L. Peace, and T.B. Smith,1974:
Determination of the feasibility of the long-range transport
of ozone or ozone precursors. Meteor. Res. Inc. Contract
No. 68-02-1462. EPA, 92 pp. NTIS No. EPA-450/3-74-061 .
Bonner, W.D., 1968: Climatology of the low-level jet. Mon Wea
Rev.. 96, 833-850. *
Briggs, G.A., 1969: Plume rise, U.S. Atomic Energy Commission,
TID-25075, 81 pp.
Brown, R.M. and P. Michael ,1974: Measured effect of shear on plume
dispersion, Preprint, Symposium on Atmospheric Diffusion
and Air Pollution, Amer. Meteor. Soc., Santa Barbara, 246-
250.
Changnon, Stanley, The LaPorte weather anomaly - fact or fiction? • 1968
Bull. Amer. Meteor. Soc. 49, 4-11.
208
-------
Coffey, P.E., and W.N. Stasiuk, 1975: Evidence of atmospheric
transport of ozone into urban areas, Environ. Sci. Tech., 9,
59-62.
Cole, H.S., and W.A. Lyons, 1972: The impact of the Great Lakes
on the air quality of urban shoreline areas: some practical
application with regard to air pollution control policy and
environmental decision making. Proc. 15th Conf. Great Lakes
Res., Intern. Assoc. Great Lakes Res., 436-463.
Collins, G.F., 1971: Predicting 'Sea-breeze fumigation' from tall
stacks at coastal locations, Nuclear Safety. 12, 110-114.
Dieterle, D.A. and A.G. Tingle, 1976: A numerical study of meso-
scale transport of air pollutants in sea-breeze circulations.
Preprints, 3rd Symposium on Atmospheric Trubulence, Diffusion
and Air Quality, A.M.S., Raleigh, 436-441.
Dooley, J.C., Jr., 1976: Fumigation from power plant plumes in
the lakeshore environment, UWM Air Pollution Analysis Lab-
oratory Report No. 18, May, 119 pp.
Edinger, J.G., M.H. McCutchan, P.R. Miller, B.C. Ryan, M.J. Schroe-
der, and J.V. Behar, 1972: Penetration and duration of
oxidant air pollution in the south caost air basin of Cali-
fornia, J^. Air Poll. Control Assoc., 22, 882-886.
EPA, 1975: Position paper on regulation of atmospheric sulfates,
U.S. Environmental Protection Agency, EPA-450/2-75-007,
85 pp.
Ergen, D.J. and R.A. Northouse, 1973: Unsupervised discrete cluster
analysis TR-A1-73-2, Robotics and Artificial Intelligence
Lab Report, UW-Mi1waukee.
Ernst, J.A., 1975: A different perspective reveals air pollution,
Weatherwise. 28, 215-126.
Esser, J.T., 1973: Bio-assay of ambient pollution in Milwaukee and
environs: effects of photochemical air pollution on vegeta-
tion, M.S. thesis, Botany Dept., UW-Mi1waukee.
Estoque, M.A., 1962: The sea breeze as a function of the prevail-
ing synoptic situation, J. Atm. Sci . , 19, 244-250.
> 1961: A theoretical investigation of the sea breeze, Quart
J.R. Meteor. Soc. , 87, 136-146. '
Fert, D.M., 1969: Analysis of the Texas coast land breeze, Tech.
Report No. 18, Atm. Sci. Group, Univ. of Texas, Austin, 52 pp.
209
-------
Fisher, E.L., 1960: An observational study of the sea breeze, J
Meteor. , 17, 645-660. —
Hall, F.F., Jr., 1972: Temperature and wind structure studies by
acoustic echo sounding,in Remote Sensing of the Troposphere,
V.E. Derr, Ed., GPO, Washington, D.C. —
« C.E. Duchon, L.G. Lee, and R.R. Hagan, 1973: Long-range
transport of air pollution: a case study, August 1970,
Mon. Wea. Rev. , 101 , 404-414.
Haurwitz, B., 1947: Comments on the sea breeze circulation, J.
Meteor., 4, 1-8. —
Herkoff, D., 1969: Observed temperature profiles near the Lake
Michigan shoreline. Tech. Report, Dept. of Meteorology
and Oceanography, Univ. of Michigan, 37 pp.
Hewson, E.W., G.C. Gill, and G.J. Walke, 1963: Smoke plume photo-
graphy study, Big Rock Point nuclear plant, Charlevoix,
Michigan. Pub. No. 04015-3-P, Dept. of Meteorology and
Oceanography, Univ. of Michigan, Ann Arbor, unpublished.
Lyons, W.A., 1971: Low-level divergence and subsidence over the
Great Lakes in summer. Proc. 14th Conf. Great Lakes Res.,
Intern. Assoc. Great Lakes Res. , 467-487.
_, 1974: Inadvertent weather modification by Chicago-North-
ern Indiana pollution sources observed by ERTS-1. Mon.Wea
Rev., 102. 503-508.
_, 1970: Numerical simulation of Great Lakes summertime
conduction inversions, Proc. 13th Conf. Great Lakes Res.,
Intern. Assoc. Great Lakes Res. , 369-387.
1975: Satellite detection of air pollutants. Proc . ,
osium on Remote Sensing Applied to Energy Related
v__ _.-• ; ------ ^ —, ^. >, ,. v , .,« j i i ^ ^ i i <_^\j \*\j L- 11 ^ i yjf i\ c i a u c \4
Problems, sponsored bv University of Miami 3.inhn Wiley
& Sons, 263-290.
_, 1966: Some effects of Lake Michigan upon squall lines
and summertime convection. Prog. 9th Conf. Great Lakes Res
Intern. Assoc. Great Lakes Res. , 259-273.
_, 1975a: Turbulent diffusion and pollutant transport in
shoreline environments, in Lectures on Air Pollution and
Environmental Impact Analysis, A.M.S.. Boston. i36-?na.
_, and H.S. Cole, 1973: Fumigation and plume trapping on
the shores of Lake Michigan during stable onshore flow
J. of Appl Meteor. . 12, 494-510.
210
-------
_, and H.S. Cole, 1976: Photochemical oxidant transport:
mesoscale lake breeze and synoptic-scale aspects, J. Appl.
Meteor., 15, 733-743.
_, and J.C. Dooley, Jr., 1974: Study of fumigation of sulfur
oxides from the Waukegan, Illinois power plant. Report
submitted to Commonewalth Edison Company, Chicago.
_, J.C. Dooley, C.S. Keen, J.A. Schuh, and K.R. Rizzo, 1974:
Detailed field measurements and numerical models of S02
from power plants in the Lake Michigan shoreline environment,
Contract Report to Wisconsin Electric Power Co., by Air
Pollution Analysis Laboratory, UW-Milwaukee, Milwaukee,
Wisconsin, 218 pp.
_, and C.S. Keen, 1976a: Particulate transport in a Great
Lakes coastal environment. In Proceedings. Workshop on
Atmospheric Transport and Removal Processes. 2nd ICMSE
Conf. on the Great Lakes, Argonne National Laboratory.
_, C.S. Keen, and R.A. Northouse, 1974: ERTS-1 satellite
observations of mesoscale air pollution dispersion around
the Great Lakes, preprints, Symposium on Atmospheric Dif-
fusion and Air Pollution, A.M.S., Santa Barbara, 273-280.
_, and L.E. Olsson, 1973: Detailed mesometeorological studies
of air pollution dispersion in the Chicago lake breeze, Mon.
Wea. Rev. . 101. no. 5, 387-403.
_, and L.E. Olsson, 1972: Mesoscale air pollution transport
in the Chicago lake breeze, J. Air Poll. Control Assoc., 22,
876-881.
_, and S.R. Pease, 1973: Detection of particulate air pol-
lution plumes from major-point sources using ERTS-1 imagery,
Bull. Amer. Meteor. Soc., 54. 1163-1170.
_, and S.R. Pease, 1972: 'Steam Devils' over Lake Michigan
during a January arctic outbreak, Mon. Wea. Rev., 100,
235-237.
_, and E.M. Rubin, 1976: Aircraft measurements of the Chicago
urban plume at 100 km downwind, Preprints, 3rd Symposium on
Atmospheric Trubulence, Diffusion, and Air Quality, A.M.S.,
Raleigh, 358-365.
_, and J.W. Wilson, 1968: The control of summertime cumuli
and thunderstorms by Lake Michigan during non-lake breeze
conditions, Res. Paper No. 74, Satellite and Mesometeoro-
logy Res. Proj., Univ. of Chicago, 38 pp.
211
-------
Hidy, G.M., P.K. Mueller, E.Y. Tony, J.R. Mahoney, and N.E. Gaut,
1976: Design of the sulfate regional experiment (SURE),
Final Report of Research Project 485, Electric Power Research
Insti tute.
Hirt, M.S., 1. Shenfeld, G. Lee, H. Whaley, and S.D. Jurtors, 1971:
A study of the meteorological conditions which developed a
classic "fumigation" inland from a large lake shore source,
Paper 71-132, 64th Annual Meeting, Air Poll. Control Assoc.
Holzworth, G.C., 1967: Mixing depths, wind speeds, and air pol-
lution potential for selected locations in the United States,
Journal of Applied Meteorology, 6.
, 1972: Mixing heights, wind speeds, and potential for ur-
ban air pollution throughout the contiguous United States,
Envir. Protection Agency, Office of Air Programs, Pub. No.
AP-101 , xi, 118 pp.
Hsu, S.A., 1969: Mesoscale structure of the Texas coast sea
breeze, Report No. 16, NSF Grant GA-3674, Univ. of Texas
at Austin, Atmospheric Sciences Group, 237 pp.
Husar, R.B., 1975: On the origin of large scale haziness over
the Midwestern United States: anatomy of an air pollution
episode in St. Louis, unpub. report, Air Poll. Res. Lab.,
Washington Univ., St. Louis, 10 pp.
, N.V. Gillani, J.D. Husar, C.C. Paley, and P.N. Turcu, 1976:
Long-range transport of pollutants observed through visibi-
lity contour maps, weather maps, and trajectory analysis,
Preprints, 3rd Symposium on Atmospheric Turbulence, Diffusion
and Air Quality, A.M.S., Raleigh, 344-347.
Jehn, K.H., 1973: A sea breeze bibliography, 1664-1972, Report
No. 37, University of Texas, Atm. Science Group, 51 pp.
Keen, C.S., 1976: Trajectory analyses of mesoscale air pollution
transport in the Lake Michigan shoreline environment, Ph.D.
dissertation, Univ. Wis.-Mi 1waukee , also Special Report No.
29, Center for Great Lakes Studies, 235 pp.
Lavoie, R.L., A mesoscale numerical model of lake-effect storms,
J. Atm. Sci . , 29, 1025-1040.
Lenschow, D.H., 1973: Two examples of planetary boundary layer
modification over the Great Lakes, J. Atmos. Sci., 30 ,
568-581.
MacCready, P.B., Jr., 1964: Standardization of gustiness values from
aircraft. J. Appl. Meteor., 3, 439-449.
212
-------
McPherson, R.D., 1970: A numerical study of the effect of a coastal
irregularity on the sea breeze, J. Appl. Meteor. , 9. 767-777.
Meyer, J.H., 1971: Radar observations of land breeze fronts, J_._
Appl . Meteor. , 10, 1224-T232.
Miller, A., and C.D. Ahrens, 1969: Ozone within and below the
west coast temperature inversion. Report No. 6, San Jose
State College, Dept. of Meteor., 74 pp.
Moroz, W.J., 1967: A lake breeze on the eastern shore of Lake
Michigan: Observations and model, J. Atmos. Sci. , 24,
337-355.
Munn, R.E., 1959: The application of an air pollution climatology
to town planning, Inter. J. Air Pol 1. , 1 , 276-278.
Neumann, J., and Y. Mahrer, 1971: A theoretical study of the land
and sea breeze circulations, J. Atmos. Sci., 28, 532-542.
, and , 1975: A theoretical study of the lake and
land breezes of circular lakes, Mon. Wea. Rev. , 103, 474-
485.
Ogawa, Y., W.H. Hoydysh, and R. Griffith, 1973: A laboratory
simulation of sea breeze effects, TR 115, New York Univ.,
Environ. Res. Labs, 175 pp.
Olsson, L.E., 1969: Lake effects on air pollution dispersion,
Ph.D. dissertation, Dept. of Meteorology and Oceanography,
Univ. of Michigan, 216 pp.
•
Pearce, R.P., 1955: The calculation of a sea breeze circulation
in terms of the differential heating across the coast line,
Quart. J.R. Meteor. Soc.. 81, 351-381.
Pendergast, M.M., and T.V. Crawford, 1974: Actual standard devia-
tions of vertical and horizontal wind direction compared to
estimates from other measurements. Preprint, Symposium on
Atmospheric Diffusion and Air Pollution, A.M.S., Santa
Barbara , 1-7.
Pielke, R.A. , 1974a: A comparison of three-dimensional and two-
dimensional numerical predictions of sea breezes, J. Atm.
Sci., 31, 1577-1585.
, 1974b: A three-dimensional numerical model of the sea
breeze over south Florida, Mon. Wea. Rev. , 102, 115-139.
Raynor, G.S., T.V. Hayes, and E.G. Ogden, 1974: Mesoscale trans-
port and dispersion of airborne pollens, J. Appl . Meteor.,
13, 87-95.
213
-------
> p- Michael, R.M. Brown, and S. Sethuraman, 1974: A research
program on atmospheric diffusion from an oceanic site. Pre-
print, Symposium on Atmospheric Diffusion and Air Pollution,
A.M.S., Santa Barbara, 289-295.
Ripperton, L.A., J.B. Tommerdahl, and J.J.B. Worth, 1974: Airborne
ozone measurements study, Paper 74-42, presented at 67th
Annual Meeting, Air Poll. Control Assoc, Denver, 19 pp.
Rizzo, K.R., 1975: Determining the mixing depth climatology using
an acoustic sounder in a lakeshore environment, M.S. thesis,
UW-Milwaukee, also, Special Report No. 28, Center for Great
Lakes Studies, 78 pp.
Schubert, J.F., 1975: Climatology of the mixed layer using acous-
tic methods, 3rd Symposium on Meteorological Observations
and Instrumentation, Preprint, A.M.S., 10-13 February,
Washington, D.C.
Schuh, J.A., 1975: A mesoscale model of continuous shoreline fumi-
gation and lid trapping in a Wisconsin shoreline environment,
M.S. thesis, UW-Milwaukee, also Special Report No. 27, Center
for Great Lakes Studies,107 pp.
Sheih, C.M., and W.J. Moroz, 1975: Mathematical modeling of lake
breeze, Atmos. Environ. , 9, 1-12.
Thomson, D.W., 1975: ACDAR Meteorology: the application and in-
terpretation of atmospheric acoustic sounding measurements,
Preprint, 3rd Symposium on Meteorological Observations and
Instruments, A.M.S., Washington, D.C.
Turner, D.B., 1969: Workbook of atmospheric dispersion estimates,
Rev. ed., U.S. Dept. Health, Education, and Welfare, 84 pp.
Van der Hoven, I., 1967: Atmospheric transport and diffusion at
coastal sites, Nuclear Safety, 8, No. 5, K.E. Cowser, Ed.,
490-493.
Warner, J., and J.W. Telford, 1967: Convection below cloud base,
J. Atm. Sci. , 24. 374-382.
Weisman, B., and M.S. Hirt, 1975: Dispersion governed by the thermal
internal boundary layer, Paper 75-26.3, presented at 68th
Annual Meeting, Air Pollution Control Assoc., Boston, 13 pp.
Wilson, W.E., R.J. Charlson, R.B. Husar, K.T. Whitby, and B. Blu-
menthal , 1976: Sulfates in the atmosphere, Paper 76-30-06,
69th Annual Meeting, Air Poll. Control Assoc., Portland,
20 pp.
214
-------
APPENDIX
Publications generated via support from the U.S. Environmental
Protection Agency, Grant No. R-800873.
1972
"Climatology and Prediction of the Chicago Lake Breeze," 20-
minute color film, presented , American Meteorological Society
Conference on Weather Analysis and Forecasting, Portland,
Oregon, also Journal of Applied Meteorology, IJ, 1255-1270.
"Mesoscale Air Pollution Transport in the Chicago Lake Breeze"
with L.E. Olsson, Journal of the Air Pollution Control Associa
tlon, 22, 876-881. Also APAL Report No. 3.
"Detailed Mesometeorological
in the Chicago Lake Breeze,"
Review, 101, 387-403. Also
Studies of Air Pollution Dispersion
with L.E. Olsson, Monthly Weather
APAL Report No. 8.
"The Impact of the Great Lakes on the Air Quality of Urban Shore-
line Areas: Some Practical Applications with regard to Air Pol-
lution Control Policy and Environmental Decision Making," with
H.S. Cole, Proc. 15th Conf. Great Lakes Res.. IAGLR, 436-463.
Also APAL Report No. 4.
1973
"Fumigation and Plume Trapping on the Shores of Lake Michiga
During Stable Onshore Flow," Journal of Applied Meteorology,
494-510, Also APAL Report No. 2.
12
"ERTS-1 Views the Great Lakes," with S.R. Pease, presented to 16th
Conf. Great Lakes Research, IAGLR, also NASA, Goddard Space Flight
Center, Greenbelt, MD., Proceedings, "Symposium on Significant
Results Obtained from ERTTH Satellite/1 Vol. 1, Sec. A, 847-854,
Also GLUMP Report No. 15.
"Detection of Particulate Air
Sources Using ERTS-1 Imagery,
American Meteorological Society, 54,
»l ~ J TJ' S\~ ---------- -•- - -— II .- - - --.. L- r ±,.-~.- , :W ;. .. -ii —
No. 10.
Pollution Plumes from Major Point
S.R. Pease, Bulletin of the
1163-1170. Also APAL Report
with
215
-------
The Use of ERTS-1 Imagery in Air Pollution and Mesometeorlogical
Studies Around the Great Lakes," with R.A. Northouse, NASA, God-
dard Space Flight Center, Greenbelt, MD, Proceedings, Third ERTS
Symposium," also, APAL Report No. 11. " ——
"Mesoscale Despersion Regimes on the Shores of the Great Lakes:
Observations and Models," with H.S. Cole, J.A. Schuh, and C.S.
Keen, 20 minute color film, presented Annual Meeting, American
Geophysical Union, San Francisco.
1974
"Inadvertent Weather Modification by Chicago-Northern Indiana
Pollution Sources Observed by ERTS-1, Monthly Weather Review,
102, 1163-1170, Also APAL Report No. 97~~
ii
Great Lakes Springtime Conduction Fogs; ERTS-1 Observations and
Models," with S.R. Pease, 20-minute color film, presented Annual
Meeting, American Meteorological Society, Honolulu.
"Numerical Modeling of Mesoscale Suspended Particulates and Sul-
fur Oxide Patterns in an Urban Great Lakes Shoreline Environment,"
with J.A. Schuh, preprints, AMS 5th Conference, Weather Analysis
and Forecasting, St. Louis, also APAL Report No. 7.
"The Use of Monitoring Network and ERTS-1 Data to Study Inter-
regional Pollution Transport in the Chicago-Gary-Milwaukee Cor-
ridor," with H.S. Cole, presented Annual Meeting, Air Pollution
Control Association, Denver, Paper 74-241, also APAL Report No.
12, 25 pp.
"ERTS-1 Satellite Observations of Mesoscale Air Pollution Disper-
sion Around the Great Lakes," with C.S. Keen and R.A. Northouse,
preprints AMS/WMO Symposium on Atmospheric Diffusion, Turbulence
and Air Pollution, Santa Barbara, 273-280, also APAL Report No.
1 O •
"Study of Fumigation of Sulfur Oxides from the Waukegan, Illinois
Power Plant," with J.C. Dooley, Final Report to Commonwealth
Edison Co., 75 pp.
1975
"Detailed Field Measurements and Numerical Models of SO? from
Power Plants in the Lake Michigan Shoreline Environment," with
J.C. Dooley, C.S. Keen, J.A. Schuh, and K.R. Rizzo, Final Report
to Wisconsin Electric Power Company, 218 pp.
"Satellite Detection of Air Pollutants," Remote Sensing Energy-
Related Studies,T.N. Veziroglu, Ed.,John Wiley"& Sons, New" York,
£ O O ~ £ _? U •
216
-------
"Turbulent Diffusion and Pollutant Transport in Shoreline Environ-
ments," invited contribution, Workshop on Meteorology and Envrroji-
mental Assessment, American Meteorological Society, Boston, 105-
177". (Invited lecture/paper).
1976
"Single Theodolite Pibal Observations on the Western Shoreline of
Lake Michigan: Summers 1973 and 1974," with S. Hentz, O.C. Dooley,
C.S. Keen, GLUMP Project Report No. 16, 190 pp.
"Particulate Transport in a Great Lakes Coastal Environment," with
C.S. Keen, invited paper, Proceedings, 2nd ICMSE Conference on the
Great Lakes, workshop on atmospheric transport and removal proces-
ses, Argonne National Laboratory, 23 pp.
"Computed 24-Hour Trajectories for Aerosols and Gases in a Lake/
Land Breeze Circulation Cell on the Western Shore of Lake Michi-
gan," with C.S. Keen, preprints, Sixth Conference on Weather
Analysis and Forecasting, American Meteorological Society, Albany,
NY, 6 pp.
"Photochemical Oxidant Transport: Mesoscale Lake Breeze and Syn-
optic Scale Aspects," with H.S. Cole, Journal of Applied Meteor-
ology, 15 (in press), pp. 733-743. ""
"Aircraft Measurements of the Chicago Urban Plume at 100 km Down-
wind," with E.M. Rubin, Preprints, Third Symposium on Turbulence,
Diffusion, and Air Pollution, American Meteorological Society,
Raleigh, N. Carolina, pp. 358-365.
"Meteorological Aspects of Air Quality," Bulletin of the Ameri-
can Meteorological Society ? 57, 205^206. ' .-
"SMS/GOES Visible Images Detect Synoptic-Scale Air Pollution
Episode," with R.B. Husar, submitted to Monthly Weather Review.
217
-------
Graduate Theses produced via support from the U.S. Environmental
Protection Agency, Grant No. R-800873. environmental
Ph.D
Keen, Cecil S., 1976: Trajectory Analyses of Mesoscale Air Pol-
lution Transport in the Lake Michigan Shoreline Environment
University of Wisconsin-Milwaukee, Department of Geography ,' pub-
No- 29> UWM center for
M.S.
Dooley, John C., Jr., 1976: Fumigation from Power Plant Plumes
in the Lakeshore Environment. University of Wisconsin-Milwaukee,
Co ege of Engineering and Applied Science, Published as Air
Pollution Analysis Laboratory Report No. 19, July, 119 pp.
Rizzo, Kenneth R., 1975: Determining the Mixing Depth Climatology
Using an Acoustic Sounder in a Lakeshore Environment. University
of Wisconsin-Milwaukee, College of Engineering and Applied Science
Published as Special Report No. 28, UWM Center for Great Lakes
btudies, December, 78 pp.
FMmin;-^1?75: -A Mesoscale Model of Continuous Shoreline
l-umigation and Lid Trapping in a Wisconsin Shoreline Environment
University of Wisconsin-Milwaukee, College of Engineer-inland
Applied Science Published as Special Report No. 27, UWM Center
for Great Lakes Studies, December, 107 pp.
218
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. R. PORT NO.
EPA-600/4-77-010
2.
3. RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE
MESOSCALE AIR POLLUTION TRANSPORT IN SOUTHEAST
WISCONSIN
5. REPORT DATE
February 1977
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
Walter A. Lyons
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
University of Wisconsin-Milwaukee
Air Pollution Analysis Laboratory
College of Engineering and Applied Science
Milwaukee, Wisconsin 53201
10. PROGRAM ELEMENT NO.
1AA603
11. CONTRACT/GRANT NO.
R-800873
12, SPONSORING AGENCY NAME AND ADDRESS
Environmental Sciences Research Laboratory - RTP, N.C.
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, N.C. 27711
13. TYPE OF REPORT AND PERIOD COVERED
Final 1972-1976
14. SPONSORING AGENCY CODE
EPA/600/09
15. SUPPLEMENTARY NOTES
16. ABSTRACT
During the period 1970-1976, the University of Wisconsin-Milwaukee's Air
Pollution Analysis Laboratory (College of Engineering and Applied Science) engaged in
extensive studies on the mesometeorology of the Great Lakes. This report highlights
the important findings of this research that are air pollution-related. An extensive
field study on the western shore of Lake Michigan during the summer of 1974 essential -
l.v validated the GLUMP Fumigation Model, which was calibrated for multi-plume power
plants. Studies of lake meteorology showed that, even during supposedly "steady-
state" onshore gradient flows, complicated wind patterns occurred including the de-
velopment of low-level jet streams associated with intense inversion layers. The
acoustic sounder was found to be highly useful in showing the structure of the lake-
shore environment. Also, satellite data were highly useful in monitoring mesoscale,
regional and synoptic scale transport. Individual plumes were detected for more than
150 km over Lake Michigan, and a major sulfate haze aerosol episode was imaged over
the central U.S. A model was proposed to explain the inland band of elevated ozone
levels running parallel to the shoreline, and the Chicago metropolitan area was shown
to be a major contributor to the high oxidant levels recorded in southeastern Wiscon-
sin. On one occasion, aircraft monitoring of the Chicago urban nlume revealed inter-
state transport of 25 tons per hour of S0? from Illinois into Wisconsin. The finding;
.buyyebt uidL trie concept or Mir quanty tontroi Kegions has to be severely modified
n v ^ hirinHnnrrl ?\~\ +-r\*-ir\ -\-v\r\\* T\V\A <- u^i.,rt ^ -f-hn •J«-» — «T -— -iK-il-i-i-\j- n-F mr\ - -1- - -Y -5 r -i- -i i- -t • - • •
ur auttriuuiiuu a I tuyu Lriui ariu brio WG a uric 1 Hupp 1 1 CaDI 1 1 "V OT mOSt 6X1 Stl 11 CJ SHOT I iGMll
17-prediction models in coastal KEY WORDS AND DOCUMENT ANALYSIS 7nnp<;
a. DESCRIPTORS
*Air pollution Electric power plants
Ozone Meteorological satel-
Sulfur dioxide lites
*Meteorology Plumes
*AtmosDheric circulation
Sea breezes
*Mathematical models
13. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
EPA Form 2220-1 (9-73) 2
b. IDENTIFIERS/OPEN ENDED TERMS
Great Lakes
Southeast Wisconsin
19. SECURITY CLASS (This Report)
UNCLASSIFIED
20. SECURITY CLASS (This page)
UNCLASSIFIED
c. COSATI Field/Group
13B
07B
04B
12A
10B
22B
21B
21. NO. OF PAGES
238
22. PRICE
9
------- |