TECHNICAL REPORT A62-5
symposium
Air over Cities
\
PUBLIC HEALTH SERVICE
ROBERT A. TAFT SANITARY ENGINEERING CENTER
-------
Symposium
Air Over Cities
Sponsored by the
Laboratory of Engineering and
Physical Sciences
of the
Division of Air Pollution
U. S. Department of
Health, Education, and Welfare
Public Health Service
Robert A. Taft
Sanitary Engineering Center
Held November 6-7, 1961
Cincinnati, Ohio
-------
CONTENTS
Page
INTRODUCTION '. v
SESSION 1: CITY AIR - BETTER OR WORSE?
Chairman: H. E. LANDSBERG 1
Air Pollution Studies in the Netherlands. F. H. SCHMIDT .... 23
Recent Developments in the Chemistry of Urban
Atmospheres. J. P. LODGE, JR 31
A Climatological Evaluation of Precipitation Patterns
Over an Urban Area. S. A. CHANGNON, JR 37
Some Effects of Air Pollution on Visibility In and
Near Cities. G. C. HOLZWORTH 69
Smoke Concentrations in Montreal Related to Local
Meteorological Factors. P. W. SUMMERS 89
The Air Over Philadelphia, F. K. DAVIS, JR 115
The Thermal Climate of Cities. J. M. MITCHELL, JR 131
Some Observations of Cloud Initiation In Industrial
Areas. G. E. STOUT 147
SESSION 2: THE DISPERSION AND DEPOSITION
OF AIR POLLUTANTS OVER CITIES
Chairman: M. NEIBERGER 155
Analog Computing Techniques Applied to Atmospheric
Diffusion: Continuous Area Source. F. V. BROCK 173
Dispersion Calculations for Multiple Sources.
F. POOLER, JR 189
Some Effects of City Structure on the Transport of Air-
borne Material in Urban Areas. W. A. PERKINS 197
Source Configurations and Atmospheric Dispersion in
Mathematical Models of Urban Pollution Distributions.
G. R HILST 209
Some Aspects of Atmospheric Diffusion in Urban
Areas. JAMES HALITSKY 217
SESSION 3: PRESENT AND FUTURE NEEDS FOR
METEOROLOGICAL AND AIR QUALITY
OBSERVATIONS
Chairman: J. J. SCHUENEMAN
The Relative Importance of Some Meteorological
Factors in Urban Air Pollution. ELMER ROBINSON 229
Measurement Programs Required for Evaluation of
Man-Made and Natural Contaminants in Urban
Areas. E. W. HEWSON, E. W. BIERLY, AND J. C. GILL 239
The Representativeness of Local Observations in Air
Pollution Surveys. M. E. SMITH 259
Present and Future Needs for Meteorological and Air
Quality Observations in Canada. R. E. MUNN 267
Problems Associated with Forecasting Air Pollution
Over an Urban Area. E. K. KAUPER 269
The Need for More Meaningful Meteorological and Air
Quality Observations for Mortality and Morbidity
Studies. F. FIELD AND J. K. MCGUIRE 277
m
-------
Introduction
ARTHUR C. STERN, Chief, Laboratory of
Engineering and Physical Sciences,
Division of Air Pollution:
My function here is to say welcome, but I have chosen in-
stead to take the 10 minutes allotted to me to forecast what the
role of the meteorologist, particularly the one in charge of a
Weather Bureau station in a large metropolitan area, will be in
the year 1985. This I hope may give you a little pattern for the
talks to follow, and if you succeed in the following 2 days in set-
ting all of the groundwork for this chore in 1985, we will have ful-
filled our reason for getting together here.
Now, firstly, in 1985 the people will be more interested in
whether the air is safe to breathe than whether it will rain. This,
of course, will completely change many of the paramount func-
tions in the normal operation of a Weather Bureau station.
During a typical day in the Weather Bureau station, the
staff will not only study the classical weather maps at ground
level and aloft, but also a recently developed (before 1985) basic
horizontal plane air quality isoconcentration map. In addition to
the latter horizontal plane maps, we will, of course, by then have
a very well-developed set of vertical elevation data for a number
of latitudes and longitudes, from the ground up to 5, 000 feet. At
particular stations, we will get from the facsimile elevation maps
showing temperature, pressure, air movement, humidity, and
air quality in the vertical plane. The station staff will choose the
latitudes and longitudes that intersect the particular urban com-
plex where their major interest lies, and also those proximate
thereto. And although advisories will be sent from Washington --
I presume Washington will still be the capital of the United States
then -- giving 72-hour, 48-hour, and 24-hour air quality fore-
casts, these forecasts will be routinely combined with the data
obtained from the local vertical elevation maps and isoconcentra-
tion maps for air quality. The meteorologist will combine these
with the daily source strength prediction, which he will receive
each day from the local air pollution control agency; these two
sets of data will be fed into the station electronic computer, which
I am sure each of the stations by then will have.
To obtain these vertical elevation maps, we must have by
1985 a very well-developed national network for vertical sounding
from the ground to 5, 000 feet; this network will be as intense
-------
vertically as our horizontal network is now intense horizontal y.
The electronic computer will develop local forecasts of air
quality and the maximum source strength distribution that will
keep pollutant emissions within air quality standard limits.
These latter data, that is the maximum source strength dis-
tributions, will simultaneously be typed out not only at the com-
puter in the Weather Bureau station but also at the air pollution
control headquarters, which will use them to provide quotas to
various agencies.
I don't know whether by 1985 we can give hourly quotas, but
certainly we will be giving quotas to the highway entrance control
points as to the number of vehicles permitted to pass each hour,
and to industrial sources for their hourly emission quotas. Also
we will determine which portion of energy used in household and
commercial enterprises is to be drawn hourly from energy stor-
age reserves maintained for that purpose, and which portion may
come from new energy conversion processes. Thus on bad days
we will balance the ability of the atmosphere to handle the pollu-
tion with the amount that we throw into the atmosphere.
About at this point the computer will be hooked up to the
telemetered local air quality monitoring network to make hourly
comparisons between the predicted and the actual concentrations
and to prepare hourly amendments to the source strength quotas.
VI
-------
Session 1: City Air - Better or Worse
H. E. LANDSBERG,* chairman
Summary
Human activity has caused considerable changes in local climates. These modifi-
cations in turn have affected the temperatures of the lowest layer of the atmosphere, the
diurnal temperature range, the shape of the diurnal temperature curve, the local relative
humidity, the local electric field, the patterns and amounts of precipitation, and the speed
and direction of winds. These effects have probably only a minor influence upon human
well-being. Other changes, however, have potentially harmful effects. The most radical
effect has been on atmospheric suspensions and admixtures. None of these changes have
been beneficial. The growth of nearly all urban areas and industrial complexes has out-
paced the engineering and legal efforts to minimize the nuisance and the possible dangers
of contamination. Already the ill winds from one settlement can influence the next town
downwind. The day of planning in terms of single communities is over, and whole region-
al patterns now must be viewed together. Our knowledge of air quality and its effects on
health is not yet adequate. In the interest of public hygiene, an intense effort in bio-
meteorological and medical research is required.
Census figures show that an ever-increasing percentage of
our citizens have become urbanites or, at least, suburbanites.
Eighty-five percent of the 28-million increase in the U. S. pop-
ulation in the 1951-1960 decade occurred in the standard metro-
politan areas (as designated by the U. S. Bureau of the Census in
1961). These cover only 9 percent of the area of the country.
This concentration of population has given rise to many problems.
As this trend continues, major problems in planning will arise.
Clean water for the population has been of public concern for a
long time, and it is time that we do something about the air in
the metropolitan areas. The reasons for this are clear and need
not be repeated. As in all planning, however, we need a quanti-
tative basis -- and this is only beginning to emerge.
There is no gainsaying that the towns and the cities consti-
tute a radical change from the natural environment. These dense
settlements help in commerce and trade, in culture and services
--and in a bygone age they were even useful in common defense.
But what price do we pay for our gregariousness ? Some might
point to slums, others to the rise in crime, still others to traf-
fic difficulties. As meteorologists we are particularly interest-
ed in the changes of the atmosphere in urban areas. These are
of considerable magnitude and of concern to our health and well-
being.
Just 300 years ago there was problably only one place in
the world which found itself vis-a-vis a notable man-made cli-
matic change: London, England. The chronicler of the day,
*Director, Office of Climatology, U.S. Weather Bureau
-------
2 LANDSBERG
John Evelyn, F. R. S. (1661), gave a classical account of it. He
focused attention on the problem of smoke pollution, which he
attributed to channel coal used in some early manufacturing
processes. Since that day and age one metropolis after another
has had to face the same difficulty from various causes.
In colonial America with its small towns, wide-open spaces,
and good natural ventilation, about a century after Evelyn a trav-
eler could still admire the clean air. Thus, on his trip to the
New World in 1761, the Vicar of Greenwich, the Reverend Andrew
Burnaby (1775), a neighbor of smoggy London, could still rave
about New York: "It lies in a fine climate, and enjoys a very
wholesome air. " But a century after the American Revolution
we find a much less sanguine appraisal. No less an authority
than the medical correspondent of Godey's Lady Book (1873) stat-
ed that the air of cities was thoroughly polluted by organic dust,
thought to be spores or germs. Our great-grandmothers were
enjoined to scrupulous cleanliness and an unceasing war against
dust lest contagious diseases would threaten their health and even
their lives. A Mr. Spence in England and the Reverend Mr.
Gibsone on this side of the Atlantic suggested that smoke be dis-
charged into the sewers instead of being poured into the atmos-
phere (Harpers, 1872).
In the twentieth century the problem has become truly world-
wide. It affects in a melancholic fasion us, our allies, the neu-
trals, and the Soviets alike. Urbanization and industrialization
have been, and will remain, in the ascendance. Thus the prob-
lem has become a matter of concern even to the World Health
Organization (1961). It is the central theme with which we want
to grapple in this symposium.
In the scientific approach of our day we have a few small
advantages over Evelyn. He had only his senses of sight and
smell--admittedly formidable qualitative tools--to diagnose the
difference between the natural state and the man-spoiled environ-
ment. Since Luke Howard published for the first time in 1818
his treatises on the climate of London, we have been trying to
ascertain by quantitative measures the magnitude of the change.
The result is truly startling. Not a single atmospheric element
has remained unchanged in the urban areas. Although we have
only a few sets of "before" and "after" measurements, there are
a large number of running accounts and comparative values be-
tween cities and surroundings.
More comprehensive surveys in the literature (Kratzer,
1956; Landsberg, 1956) present a more-detailed documentation
thanl can give here. Kratzer's monograph cites 533 references
up to 1955. We now accumulate pertinent papers on this subject
SEC TECHNICAL REPORT A62-5
-------
CITY AIR - BETTER OR WORSE 3
at the rate of 100 every 5 years, excluding those which deal only
with techniques and effects in the air pollution area.
Tables 1 and 2 illustrate the city effect in a gross way.
Table 1 deals primarily with the general climatic elements
(Landsberg, 1958), and Table 2 with atmospheric admixtures
(Katz, 1961). Where base figures of undisturbed areas are avail-
able, the differences of the city air environment are stressed.
TABLE 1
CLIMATIC CHANGES PRODUCED BY CITIES
Element Comparison with Rural Environs
Contaminants :
dust particles 10 times more
sulfur dioxide 5 times more
carbon dioxide 10 times more
carbon monoxide 25 times more
Radiation:
total on horizontal surface 15 to 20% less
ultraviolet, winter 30% less
ultraviolet, summer 5% less
Cloudiness:
clouds 5 to 10% more
fog, winter 100% more
fog, summer 30% more
Precipitation:
amounts 5 to 10% more
days with 0. 2 in 10% more
Temperature:
annual mean , 1 to 1. 5 F more
winter minima 2 to 3 F more
Relative Humidity:
annual mean 6% less
winter 2% less
summer 8% less
Wind Speed:
annual mean 20 to 30% less
extreme gusts 10 to 20% less
calms 5 to 20% more
SYMPOSIUM: AIR OVER CITIES
-------
4 LANDSBERG
TABLE 2
CONCENTRATION OF SOME AIR POLLUTANTS
IN THE ATMOSPHERE OF URBAN AREAS
Concentration,
Pollutant ppm
Carbon Dioxide . . 300 to 1000
Carbon Monoxide . . . 1. 0 to 200
Oxides of Nitrogen . ..0.01 to 1.0
Sulfur Dioxide 0. 01 to 3. 0
Aldehydes . . 0.01 to 1.0
Chlorides. ... 0. 00 to 0. 3
Oxidants 0. 00 to 0. 8
Ammonia . . . 0. 00 to 0. 21
Fluorides 0. 00 to 0. 08
Dust . . . . .... 5 to 100 milliona
Also highly variable amounts of gross
dust, micro-organisms and pollen.
Dust concentration in particles per
cubic foot.
This raises the following fundamental question: are there
any undisturbed atmospheric conditions left in the mechanized
areas of the world? The answer is probably "No, " and we have
to be satisfied with using the less-perturbed conditions in the rural
zone near the city for our base values. These data give us an
approximate picture of how much human activity has unwittingly
altered the parameters of the lowest layer of the atmosphere.
Two major changes have been introduced by man. One is
the radical alteration of the surface, the other is the continuous
addition of a wide variety of substances -- gaseous, liquid, and
solid --to the air. It is not entirely possible to separate the
effects of these alterations; however, some of them can be stated
in general terms.
The change of surface involves particularly the radiation
balance. In some locations a new effective surface is created;
run-off and evaporation character of the surface is sweepingly
altered. The air motion is basically affected by changes in the
roughness parameters. On a microscale, wind channeling occurs.
Thus, we see that even a city without heating devices, industry,
or motorized traffic would create a climate different from that of
the surroundings. A mental picture can be made by imagining the
replacement of fields, forests, streams, and ponds by a barren
SEC TECHNICAL REPORT A62-5
-------
CITY AIR BETTER OR WORSE 5
rock formation with deep canyon-like cuts.
The other change is even more complex. The simplest
phase involves the addition of heat from local sources, ranging
from the metabolism of the inhabitants to domestic and industrial
heating processes. Substantial amounts of water vapor also are
added from combustion processes, and by steam releases from
power plants, laundries, and other industrial establishments.
The gamut of chemicals injected into the atmosphere is yet to be
listed anywhere. From a meteorological point of view dust par-
ticles in the sub-micron and micron ranges, carbon dioxide, and
a few hygroscopic substances are probably most important. From
the aspect of health and hygiene many more substances could be
cited. Perhaps carbon monoxide, sulfur dioxide, some alde-
hydes, and fluorine compounds should be singled out because of
their immediate noxious effects.
Here might be a good point to digress. Although a great
deal has been learned about the special climatic characteristics
of cities, progress has been hampered for two reasons:
1. Too little information is available on a strictly com-
parative basis from series of observations in the city and
in rural proximity. Generally, such parallel series are
available by accident rather than by design. This is the
case for the usually observed weather elements at a city
station and an outlying airport. These intercomparisons
are often complicated by the fact that airports are located
in selected microclimates and at considerable distances.
Lacking particularly are simultaneous observations of the
radiation balance and of the vertical wind structure in the
lowest 1000 feet.
2. The second point covers the special orographic and
geographic peculiarities of various locales. With some
luck we might contrive to arrive at an abstraction of a city
climate per se, in contrast to a forest climate or a moun-
tain climate ~o~r an island climate or other types of meso-
climates. For practical purposes, however, this will
accomplish little. In addition each city climate must be
studied for its own peculiarities. This is particularly es-
sential since many settlements are in a specific locality
because of some "favorable" geographical features, such as
a river, a lake, a natural harbor, a plain in front of a
mountain range, or a valley system. All these features
introduce microclimatic variants that couple with the city
influences to create special combinations. Even the macro-
climatic setting is of importance. The problems of South-
ern California are a striking example.
SYMPOSIUM: AIR OVER CITIES
-------
6 LANDSBERG
In addition to the broad general studies of city climates,
therefore, we need detailed analyses of each locality. In a
modern dynamic approach only a bare start has been made. The
classical studies of Besson (1931) placed the problem in the right
framework of scientific inquiry and recent publications such as
those on Vienna (Steinhauser _et _al. 1955, 1957, 1959) have shown
most exhaustively what can be done. In this country we have con-
siderable information on the Los Angeles Basin by many contrib-
utors. (Only a few are cited here: Edinger, 1958; Edinger and
Helvey, 1961; Neiburger, 1955a, 1955b; Haagen-Smit, 1958;
Frenkiel, 1957. ) Even these fine studies remain fragmented re-
ports and papers. And what about the other 74 standard metro-
politan areas? There exist only a few scattered studies on their
atmospheric environment. Private enterprise, the planners of
the future, and even the intelligent citizen-voters need author-
itative and comprehensive reference sources on the mesoclimate
of the metropolitan areas; these, however, remain to be written.
In returning to the main theme, I would like to cover some
of the principal alterations in climate caused by cities. Much of
this will illustrate points already made. In view of the system
employed by past studies, it is easiest to proceed by meteorolog-
ical elements.
There is some information --if rather limited --on radi-
ation conditions. The gist is that the city environment, particu-
larly smoke, reduces the total duration of sunshine and the
intensity of solar radiation. The latter is particularly notable in
the short wave lengths.
This is easily demonstrated for illumination. The city
generally is darker than the environs. A survey in the industrial
section of Zaporozhe in the Ukraine, a center for the production
of steel and other metals in the Dnieper Valley, showed this well
(Fedorov, 1958). There the midday diffuse illumination was re-
duced by about 5 percent in May and June and by 21 percent in
December, with an average annual reduction of 13 percent.
There is need for a great deal more information on com-
parative values of spectral intensity of solar radiation at the
earth's surface inside and outside the city atmosphere. This is
the more desirable because of the bactericidal action of ultra-
violet radiation below 3000 A. Even in polluted city air there is
still some residual killing action of solar radiation on microbes,
as experiments in Leningrad (Krupins et al. , 1954) demonstrated.
We know very little about conditions across a metropolitan area,
based on standardized micro-organisms and simultaneous spectral
pyrheliometer observations. We do not even know with certainty
that some of the pollution products did not kill the microbes used
SEC TECHNICAL REPORT A62-5
-------
CITY AIR BETTER OR WORSE 7
in the experiments carried on so far.
Locally, the absorption of radiation by the polluted city air
might have fairly profound influences on the temperature of the
absorbing layer. In England recent measurements in the infra-
red from aircraft have shown that in extreme cases this effect
may account for temperature rises of as much as 5°C (Roach,
1961). The magnitude of the attenuations is shown in Table 3.
TABLE 3
COMPONENTS OF ATTENUATION IN LOWEST 1, 000 FEET
OF THE ATMOSPHEREa
Visible x , % Infra-red A., % Total xi, %
Absorption 5 35 20
Back- scattering 5 55
Forward-scattering 90 60 75
Total 100 100 100
aAfter W. T. Roach
The conviction of the author is that the absorption 'effects
contribute in reality only in a minor way to the temperature rises
that have been observed in cities. These have been documented
for many locations, ever since Luke Howard's early effort. With
all the orographic varieties that are found in city locations, the
"heat island" effect is well demonstrated. One of the most thor-
ough recent studies in a not-too-complicated setting is that of
Sundberg for Uppsala (1950, 1951). Mitchell (1953) has shown
that population increases account reasonably well for some of the
secular changes observed in metropolitan areas.
As for many other micro- and mesoclimatic effects, city
influence becomes most obvious on clear days and nights. (It is,
of course, always there. ) An interesting effect is notable in the
diurnal variation of temperature. In the hours after sunrise, city
and country show little difference; by and large, the maxima are
not too different. But in the late afternoon the differences begin
to show. In the country cooling with setting sun is rapid. Out-
going radiation is unimpeded. Grass and soil lose their heat by
radiation rapidly; little heat is supplied from deeper layers. In
the city, pavements and house walls absorb a lot of heat during
the hours of high sun. After sunset they begin to radiate toward
each other instead of against the sky. They are better heat con-
ductors than open land with vegetation. Hence they can draw on
conducted heat from lower layers. Direct heat sources are also
involved (furnaces etc. ). Their effect is greatest during the cool
hours of the day. All this combines into a slower decline of tem-
perature in the evening, higher temperatures at night, and
SYMPOSIUM: AIR OVER CITIES
-------
8 LANDSBERG
notably higher minima. This is demonstrated by frequency dis-
tributions of country-city temperature differences for the daily
extremes. In the simplest orographic settings there is little
difference in the daily maxima; the differences are almost nor-
mally distributed. In the minima, however, the distribution is
notably skewed. Both in summer and in winter there is a high
preponderance of excess temperatures in the city, with the
largest departures noted in winter.
The larger the city, the more pronounced the effect between
center and outskirts becomes. The complications that arise are
often the effect of microclimatic settings. These are particularly
notable in the warm season when land, lake, or sea breezes and
mountain and valley breezes attain their greatest extent. The
literature contains some exceptional examples of contrasts in
temperature between a city and its surroundings. In many cases
only a fraction of the observed temperature contrast can be as-
cribed to city influence (Duckworth and Sandberg, 1954).
There are, however, some beneficial effects of the nightly
temperature excess in the city. It reduces the heating degree-
day values and hence the fuel requirements. A comparison of 13
pairs of city-versus-airport station records in the Weather Bur-
eau network shows an average of 10 percent reduction in degree-
days between the two locations. Baltimore is a typical case. The
mean annual value of degree-days at the Friendship International
Airport is 4787 but only 4203 in the downtown district. For a
completely fair picture one should also present comparative data
for cooling degree-days in summer, but, unfortunately, no series
of such observations has as yet been worked up.
A direct effect of the slightly warmer city climate is a re-
duction in seasonal snowfall. This has recently been shown for
Montreal and Toronto (Potter, 1961). In Toronto, which has a
seasonal total of 60 inches, a decrease of about 2 inches can be
ascribed to city effect.
In a number of cities the increased minima lead to an ear-
lier cessation of freezes in spring and later initiation of the
freezing season in autumn. The values are quite variable accord-
ing to the microclimatic settings, but differences of a week on
either end of the freeze-free season apply to many cities.
Meteorologically most intriguing is the behavior of precip-
itation over city areas. It is of interest in the air pollution prob-
lem because of the substantial wash-out effects of precipitation.
These are generally beneficial, except for the presence of fission
products from weapons tests. It should be stated at the outset
that the influences of the city on precipitation are most complex
and the various skeins are not easily unraveled. We can say with
SEC TECHNICAL REPORT A62-5
-------
CITY AIR BETTER OR WORSE 9
reasonable confidence, however, that most of them tend to in-
crease precipitation. These causes are enumerated without
emphasis on the order: water vapor addition from combustion
processes and factories; thermal updrafts from local heating; up-
drafts from increased friction turbulence; added nuclei of conden-
sation leading more readily to cloud formation; and added nuclei
which might possibly act as freezing centers for subcooled cloud
particles.
Only a few pieces of evidence are cited here. The case of
the gradual increase of rainfall over Tulsa, Oklahoma, with its
growth from a village to a city in 5 decades, is presented with
diffidence. The fact that current rainfall comparisons between
city and airport station still show about an 8 percent difference
lends some support to the hypothesis that in this orographically
uninfluenced area the city really has led to the precipitation in-
crease. The winter values are relatively higher; this is also
favorable to the hypothesis of city influence because shower con-
ditions in summer are not likely to be too much affected.
The excess production of condensation nuclei over the city
is a long-established fact (Landsberg, 1938) and has been re-
affirmed with many amplifying circumstances (Georgii, 1959). We
do not yet have complete evidence for the freezing nuclei, although
measurements in the Washington, D. C. , area suggest that for
the low-temperature end there is a considerable surplus in the
metropolitan area (Kline and Brier, 1961). There is even a sug-
ges'tion that giant nuclei, which may initiate the coalescence
process, are more abundant in an industrial area than elsewhere
(Semonin and Stouk, 1961).
We can hardly escape the conclusion that a pollution effect
is at work when a weekly cycle is shown to be operative in indus-
trial areas. This was first suggested by Ashworth (1929) in
England's industrial area. Frederick recently (1961) made a new
analysis of this phenomenon in the United States. He showed a
definite minimum of Sunday rainfall for a 10-year period at Louis-
ville, Pittsburgh, and Buffalo. These cities had fewer total days
with precipitation on Sundays than on weekdays and the least aver-
age amounts for rainy Sundays compared to rainy weekdays.
This study is still incomplete but is certainly suggestive. The
possibility of "seeding1' over metropolitan areas is strongly sus-
pected in a snowfall case reported (v. Kienle, 1952) for Mannheim,
Germany, and might well have been the case for a rather remark-
able snowfall in Leicestershire, England (Murray, 1957). A
strong city influence is also suggested in the snow patterns pub-
lished (Potter, 1961) for Toronto, Ontario. Here, as in other
cities, mesoclimatic influences such as Lake Ontario may play a
role. Such influences may not be entirely lacking in the rainfall
SYMPOSIUM: AIR OVER CITIES
-------
10 LANDSBERG
patterns over Chicago, although an urban effect on rainfall there
seems to be present almost without doubt. The study by Chagnon
(1961) estimates this effect accounts for 2. 3 inches of an annual
total around 35 inches.
In the smaller town of Champaign-Urbana, Illinois, Huff
and Chagnon (1960) found no higher frequency of excessive rain-
falls, of practical importance, than in the surroundings. Their
study is a reminder of an earlier, almost unknown dissertation by
Belger (1940), who investigated some frontal storms over Berlin,
Germany. He studied only four cases in detail but in two found
very marked increases in rain intensity over the metropolis; a
similar case has been discussed by Hull (1957) for Washington,
D. C. More important, perhaps, is Belger's rather remarkable
finding that in all the Berlin cases a notable retardation of the
squall lines or rain fronts over the city took place.
This brings us to the consideration of an important question:
what influence does a city exercise on the circulation? We do
know that the wind speed is considerably reduced in the built-up
area. This is a simple consequence of increased friction. There
have been suggestions from time to time that with weak gradient
winds the city is capable of setting up its own circulation. The
late Professor Berg in Cologna (1959) tried to obtain evidence for
a city and country breeze system. He concluded that a general
wind of 1 to 2 meters per second is sufficient to suppress the
nightly country breeze, which otherwise can be set up by the 3°
to 3. 5°C temperature difference between the city and the country.
I know of only one well-documented case of complete convergence
of wind from all sides into the city. This was based on rime
deposits in the Japanese city of Asahikawa (Okita, 1960).
With the help of data from meteorologically instrumented
towers we are beginning to get a feel for the temperature field in
the vertical over a city area, as shown by the studies of DeMarrais
(1961) for Louisville and earlier by Duckworth and Sandberg (1954)
for San Francisco. These studies indicate that the city undoubt-
edly superimposes a wind field of its own upon the general flow
pattern. Observations of slow pilot balloons and floating balloons,
used over some west coast cities, have given similar indications.
Nevertheless it is almost impossible with the material now avail-
able to disentangle land and sea breeze or mountain and valley
breeze from the city and country breeze. Investigations in much
simpler orographic settings are needed.
From a practical point of view the general flow patterns
and dynamic conditions in the atmosphere are probably more im-
portant. These will be discussed later. Here may be a good
place to insert information on the material human activities add
SEC TECHNICAL REPORT A62-5
-------
CITY AIR - BETTER OR WORSE 11
to the air of our cities. These cover probably the whole range of
chemical substances used in modern civilization and all the by-
products produced incidentally in their manufacture. In addition,
there are all the intermediate and end products of combustion
processes, not the least of which are those of internal combustion
engines of motor vehicles.
Much of the material is in solid form, ranging from sub-
micron particles to flyash in the millimeter sizes. The larger
particles fall out by gravity and generally stay in the urban or
suburban areas. In the modern metropolitan districts this dust
amounts to 100 tons per square mile per month. It is likely to be
more a nuisance than a menace. McDonald (1961) recently es-
timated the mean residence time of this material at a few hours
only. It must be assumed that the smaller particles stay sus-
pended for a much longer time (Hess, 1959). The largest fraction
appears to be in the micron size. This material without doubt
penetrates into the upper respiratory passages and, according to
its composition, might readily serve as an irritant. There is
also persistent evidence that the number of microbial organisms
shows a high correlation to the number of suspended particles.
Although it is not proven, there is a suspicion that some of these
organisms attach themselves to the suspensions. The influence
of the suspensions on the electric properties of the atmosphere
are well known (Beckett, 1958; Miihleisen, 1959). Whether or
not this has health effects remains to be explored (Reiter, 1960).
There is a better estimate on the effects of the gaseous ad-
mixtures. Only 03, CO, CO2, and SO2 are mentioned here. The
oxidants usually represented by ozone have been most widely dis-
cussed (Neiburger, 1959); they are by no means restricted to
Southern California, (see e. g. , Wanta et al. , 1961). These ox-
idants show definite adverse effects on both crop plants and
humans. Much less demonstrated is the influence of CO. This
insidious compound is present in appreciable quantities in the
traffic-clogged canyon streets of our cities. A survey in Ham-
burg, Germany, revealed some remarkable consequences (Effen-
berger, 1957) of carbon monoxide in city air. Suburban residents
showed 1. 56 percent carboxyhemoglobin in blood samples; those
living in city apartments 1. 82 percent and traffic policemen as
much as 9. 5 percent. We may wonder how many headaches can
be attributed to this pollutant. Under inversion conditions with
calm, concentrations of 0. 005 volume percent of CO were found;
this is decidedly high value for a gas that shows such deadly af-
finity for our blood.
The production of CO2 by combustion processes is another
source of urban and industrial air pollution. As far as health
effects go, it probably can be dismissed, even though it may
SYMPOSIUM: AIR OVER CITIES
-------
12 LANDS BERG
increase by 10 percent over the normal in polluted city air. An
example of this is shown in the difference between December and
April mean values by Steinhauser (1959) for Vienna. Much has
been written about the possible influence of CO2 concentration
upon heat absorption in the atmosphere. Some have adduced it as
the cause of rising temperatures of the earth (Callendar, 1961),
and although this is only a hypothesis at present, the matter
deserves close watch in the future.
SO2 is probably another pollution product (Weiss and
Frenzel, 1956; Lenshin, 1958). There is good evidence, for ex-
ample that SO2 increased markedly in the fatal London smog of
December 1952. Russian workers (Bushtuveva, 1954) have shown
that such increases in SO2 are related linearly to simultaneous
presence and increases in H2SO4. These irritants are likely to
be among those causing the well-known symptoms of air pollution
afflictions.
At present we can only wonder how much city air contrib-
utes to general morbidity. It is certainly a problem that requires
more detailed attention than it has been given heretofore (Hei-
mann, 1961). Offhand it appears that the city influence would be
more pronounced on respiratory diseases than on diseases affect-
ing any other system of the organism. Just contemplate this
formidable list: asthma, bronchitis, bronchiectasis, common
cold, diptheria, emphysema, the influenzas, the pneumonias,
laryngitis, phthisis, pleurisy, tuberculosis, and whooping cough.
Although many of these are caused by specific microorganisms,
others are simply symptom-complexes possibly due to chemical
irritations, which in many cases are followed by infections
(Babayants, 1949). Respiratory complications are often the cause
of death in the elderly afflicted with cardiac and circulatory ail-
ments. Although, except for episodes of the Donora type, these
relations cannot readily be discovered (Hechter and Goldsmith,
1961), the subjective complaints should not be overlooked. In a
recent survey, for example, residents in the Los Angeles area,
showed a much higher incidence of malaise attributed to air pol-
lution than other inhabitants of California (Table 4) (Breslow,
1961). Although psychological factors might have inflated these
figures, they cannot be completely dismissed.
In spite of the many contributing local micrometeorological
factors, the general meteorological pattern of air pollution is
simple. There is essentially only one factor that causes accumu-
lation of waste products — lack of ventilation. Two elements con-
tribute to this: (1) vertical stability or low-level inversions that
prevent vertical motions and turbulence and (2) calms and slight
wind motion. From general climatological observations and the
upper air records a picture of the air pollution potential of various
SEC TECHNICAL REPORT A62-5
-------
CITY AIR BETTER OR WORSE 13
TABLE 4
AIR POLLUTION EFFECTS REPORTED BY
CALIFORNIA RESIDENTS21
(Percent of people surveyed who registered complaints in Los
Angeles County compared with those in remainder of State. )
Ailment Excess % Cases
Eye complaints 22
Nose complaints 21
Asthma 17
Throat complaints 16
Hay fever 13
Shortness of breath 11
Coughing 10
Bad headaches 10
&From Breslow (1961?).
localities or broad geographical regions can be gained. The most
recent publications in this field are by Dickson (1961), Boettger
(1961), and Hosier (1961). These studies show that vast areas of
the United States have these two meteorological prerequisites for
temporary air pollution accumulations. Fortunately, the self-
cleansing mechanisms of the atmosphere are also at work. Grav-
ity helps in promoting fallout; basically, however, strong wind
and washout by rain are the effective agents of pollution dispersal
and dilution. Frequent precipitation and winds above 10 miles per
hour can be counted upon in many of our industrialized regions to
hold air pollution to proportions which do not lead to acutely dan-
gerous episodes. In spite of the remarkable natural dissipation
system of pollutants, however, constant vigilance through con-
tinuous surveillance is important.
A major problem of the future is large-area pollution. Area-
wide difficulties have existed for some time in England's Mid-
lands and in the German Ruhr region. In the United States as
population primarily in metropolitan areas increases, towns
grow into cities, and cities into metropolises. With sprawling
suburban areas and industrialization along belt highways, separ-
ate urban entities grow together and form a conurbation which
some have dubbed a megalopolis. Under these circumstances the
pollution spread is no longer a mathematically neat problem of a
point source influenced by diffusion and turbulence. Only under
specialized conditions, such as atomic plant operation, is this
still an acute problem. We do, however, have to face the fact
that there are already innumerable single pollution sources in
various nuclei over large areas -- and in the not-too-distant
future in long chains of cities.
SYMPOSIUM: AIR OVER CITIES
-------
14
LANDSBERG
One of these sources will be the city system from Richmond,
Virginia, to Portland, Maine. This stretched-out chain of settle-
ments, industries, and traffic soon will have to be studied as a
single entity. As an example the author tried a very rudimentary
approach as a crude first approximation. Since January 1961 the
Weather Bureau has been publishing regularly the daily wind re-
sultant for all major stations, based on the 24hourly observations
BRIDGEPORT.
^
NEWARK»»NEW"YORK
PHILADELPHIA*
BALTIMORE
WASHINGTON, 0. C. •
Figure 1. Daily wind motion, from 24-hourly wind resultant, along the densely settled NE
coastal sector of the U. S. for first 8 months of 1961. Black area in frequency
diagram at right shows cases when air motion is slight and pollution products
are apt to stay within settled area.
SEC TECHNICAL REPORT A62-5
-------
CITY AIR BETTER OR WORSE 15
during the day. The resultant gives a quick, if not too accurate,
idea of the air transport during the day. Figure 1 shows a break-
down by resultant distances of wind movement, for the first 8
months of 1961, for the following sequence of 10 cities: Richmond,
Va. , Washington, D. C. , Baltimore, Md. , Philadelphia, Pa. ,
Newark, N. J. , New York (La Guardia Field), Bridgeport, Conn. ,
Providence, R. I. , Boston, Mass. , Portland, Me. Airport winds
can be taken as fairly representative for a city in general, as
shown in a study of Nashville by Frederick (1961).
A considerable number of days showed less than 100 miles
air transport per day. This means that with suitable wind direc-
tion pollution products will stay in the general districts of which
these cities are centers. As might be expected from the general
meteorological pattern of the area, the southern part of the chain
shows the greater number of slow transports. On the other hand
the 8-month sample, which includes data for winter and spring,
shows a fair percentage of movements over 300 miles per day.
These can be considered as a cleansing influence in the area.
Data for autumn will undoubtedly boost the share of the weak
winds, but these were not available when this was written.
Typical situations at the extremes of this distribution are
shown in Figures 2 and 3. Figure 2 shows generally slow winds
along the chain of cities on June 12, 1961. The very weak pres-
sure field of the synoptic situation is a primary indication. Under
such circumstances pollution products can be expected to stay in
the local area or to be wafted slowly toward the next city in the
chain. If visibility is used as a primitive index of pollution, very
low minima are noted. Such weather conditions, at least for
parts of the chain, are not exceptionally rare.
A converse case is shown in Figure 3. This occurred on
February 9, 1961, when an offshore low-pressure system with a
tight pressure gradient caused the resultant winds all along the
chain to be at right angles to the conurbation. Pollution products
cannot accumulate under these circumstances and would be blown
far out to sea. Even the lowest visibilities of the day were good.
Obviously, these cases should be studied with more refined
analyses, but this primitive approach gives an idea of future prob-
lems.
Are there other problems that should receive more atten-
tion? In the author's opinion, in our study of the basic problems
of city climate, we have neglected those man-made factors which
can lead to improvements (Katz, 1956). Most of these are of a
microclimatic nature. In the old days it was generally considered
adequate to place smoke sources in the lee of the city according
to the generally prevailing wind direction; it would be more
SYMPOSIUM: AIR OVER CITIES
-------
16
LANDSBERG
• CITY
• END-POINT OF DAILY RESULTANT WIND VECTOR
\/ FALLOUT AREA OF POLLUTION PRODUCTS
Figure 2. A cose of weak winds. Upper left inset shows pressure distribution at 0100
EST. Figures at right show lowest visibilities during 24 hours of 12 June 1961.
SEC TECHNICAL REPORT A62-5
-------
CITY AIR BETTER OR WORSE
17
• CITY
• END-POINT OF DAILY RESULTANT WIND VECTOR
\l FALLOUT AREA OF POLLUTION PRODUCTS
Figure 3. A case of strong winds. In this type of situation pollution products are apt to
be dissipated to sea.
SYMPOSIUM: AIR OVER CITIES
-------
18 LANDS BERG
appropriate to substitute the direction prevailing for the slight
winds under inversion conditions. Local air drainage problems
also should be given greater consideration.
Among the positive measures are the creation of open
spaces and parklands, avoidance of solid walls of buildings, and
construction of buildings of irregular heights. All of these coun-
teract the blocking of winds. For the gross dust, shrubbery,
trees, and hedges have proven to be good filters and could be used
to much greater extent.
In industrial areas the truly noxious substances must be
suppressed to a much greater extent. This will require further
research to distinguish between these and the effluents which are
harmless to health. Even now the question might be asked; why
permit any SO2 emission at all? There are other culprits equally
well known. Perhaps the idea of a smoke sewer is not as com-
pletely harebrained as it sounds. In fact it has been in use for
years in Europe in connection with some smelters. Certainly
smokestacks on mountain and hill tops will reach over many a
low-level inversion and into the layers of higher wind speeds. The
ratio of cost to benefit is not easily worked out for such an installa
tion on a large scale, but it certainly deserves a most serious
engineering study with a realistic assessment of the public health
benefits.
Cars are likely to remain a major difficulty for many dec-
ades to come. It is too much to hope for complete electrification
of surface transportation, but more of it may come in the metro-
politan areas before they choke completely in traffic tangles and
exhaust fumes. In connection with the latter the automotive in-
dustry certainly can use more inventors.
For decades safe drinking water has been a basic right. The
body needs only a few liters of water each day, whereas, the body
at rest needs 12, 000 liters of air in a day and at work twice as
19
much. At this rate we strain or pass through lO1^ particles per
day. A contaminant that is present at only one part in a million
moves through the system at the rate of 12 to 20 cubic centimeters
in a day. For many of these substances we would shudder to think
of getting that much in a single dose. If we get it diluted we live
in the hope that it is harmless. Is it too much to expect clean air,
as we have come to expect clean water?
REFERENCES
Anonymous (1873); Floating Dust; Godey's Lady Book (March
1873), _86, p. 280.
SEC TECHNICAL REPORT A62-5
-------
CITY AIR BETTER OR WORSE 19
Anonymous (1872); Harper's New Monthly Magazine (April 1872),
^4, p. 786.
Ashworth, J. R. (1929); The influence of smoke and hot gases
from factory chimneys on rainfall; Qu. Journ. R. Meteorol.
Soc. _55, 341-350.
Babayants, R. A. (1949); The dependence of diseases of the res-
piratory tract upon polluted air in the cities (translated
title); Gigiyena i Sanitariya 12, 3-11.
Beckett, J. C. (1958); The influence~of air pollution on the elec-
trical properties of air; Natl. Conf. on Air Pollut. Proceed
1958; 114-115.
Belger, W. (1940), Der Groszstadteinfluss auf nichtstationare
Regenfronten und ein Beitrag zur Bildung lokaler Warme-
gewitter; Berlin Dissertation, Koln, 69 pp.
Berg, H. (1959); Nachtlicher Temperaturanstiej am Rande einer
Groszstadt; Wetter und Leben 11, 9-11.
Besson, L. (1931); L'alteration du climat d'une grande ville;
Annales d'Hygiene _9, (Aout 1931), 34pp.
Boettger, C. M. (1961); Air pollution potential east of the Rocky
Mountains: Fall 1959; Bull. Am. Meteorol. Soc. _42, 615-
620.
Breslow, L. (1961); Air pollution effects reported by California
residents; Cal. Dept. of Publ. Health, Berkeley, 56 pp.
Burnaby, A. (1775); Travels through the middle settlements in
North America in the years 1759 and 1760; Reprinted, Great
Seal Books (Cornall Univ. Press), Ithaca, N. Y. , I960,
154 pp.
Bushtuveva, K. A. (1954); Ratio of sulfur dioxide and sulfuric
acid aerosol in atmospheric air in relation to meteorological
conditions (translated title); Gigiyena i Sanitariya, No. 11,
11-13.
Callendar, G.S. (1961); Temperature fluctuations and trends over
the earth; Qu. Journ. R. Meteorol. Soc., 37, 1-12.
Chagnon, S.A. , Jr. (1961); Precipitation contrasts between the
Chicago urban area and an offshore station in southern Lake
Michigan; Bull. Am. Meteorol. Soc. 42, 1-10.
DeMarrais, G. A. (1961); Vertical temperature difference ob-
served over an urban area; Bull. Am. Meteorol. Soc. 42,
548-554.
Dickson, R. R. (1961); Meteorological factors affecting particulate
air pollution of a city; Bull. Am. Meteorol. Soc. 42, 556-
560.
Duckworth, F. S. , and Sandberg, J. S. (1954); The effect of cities
upon horizontal and vertical temperature gradients; Bull.
Am. Meteorol. Soc. 35, 198-207.
SYMPOSIUM: AIR OVER CITIES
-------
20 LANDSBERG
Edinger, J. G. (1958); Research problems on the meteorology of
Los Angeles air pollution; Univ. of Calif. , Dept. of
Meteorol. ; Report on contract CWB-9309, 48 pp.
Edinger, J. G. , and Helvey, R. A. (1961); The San Fernando con-
vergence zone; Bull. Am. Meteorol. Soc. 42, 626-635.
Effenberger, E. (1957); Das Kohlenoxyd und dessen Bedentung in
der Hygiene; Medizin-Meteorol. Hefte, Nr. 12., 128 pp.
Evelyn, J. (1661); Fumifugium; The inconvenience of the air, and
smoke of London dissipated. Oxford; Reprinted by National
Smoke Abatement Society, Manchester 1933, 18 pp.
Federov, M. M. (1958); The effect of smoke on the illumination
conditions of a city (translated title); Gigiena i Sanitarya
_2J3, 14-18.
Frederick,R. H. (1961a); personal communication.
Frederick, R.H. (1961b); Surface Wind Study -- Nashville,
Tennessee; U.S. Weather Bureau manuscript.
Frenkiel, F. N. (1957); Atmospheric pollution in growing commu-
nities; Smithson. Instit. Rep. for 1956, 269-299.
Georgii, H. W. (1959); Probleme und Stand der Erforschung des
atmospharischen Aerosols; Ber. d. Dt. Wetterdienstes, 1_
(Nr. 51), 44-52. ~~
Haagen-Smith, A.J. (1958); Air conservation; Science 128, 869-
878.
Heimann, H. (1961); Effects of air pollution on human health;
"Air Pollution", WHO Monogr. Ser. 46, 159-220.
Hechter, H.H. , and Goldsmith, J. R. (1961); Air pollution and
daily mortality, Am. J. Med. Sci. 241, 581-588.
Hess, P. (1959); Untersuchungen iiber den Ausfall von Aerosol-
partikeln durch Niederschlage und Wolkenbildung; Ber. d.
Dt. Wetterdienstes _7 (Nr. 51); 67-71.
Hosier, C. R. (1961); Low-level inversion frequency in the con-
tiguous United States; Mo. Wea. Rev. 89, 319-339.
Howard, L. (1818); The Climate of London, Tst. ed. (2nd. ed.
1833, 3 vols. )
Huff, F. A. , and Chagnon, S. A. (1960); Distribution of excessive
rainfall amounts over an urban area. Journ. Geophys. Res.
_65, 3759-3765.
Hull, B. B. (1957); Once-in-a-hundred-year rainstorm, Washing-
ton, D. C. 4 September 1939; Weatherwise 10, 128-131, 139.
Katz, M. (1956); City planning, industrial plant location, and air
pollution; Sect. 2, Magill, P. L. , etc, al. , Air Pollution
Handbook, McGraw Hill Book Co.
Katz, M. (1961); Some aspects of the physical and chemical nature
of air pollution; "Air Pollution" WHO Monogr. Ser. 46,
pp. 97-158.
Kienle, J. von (1952); Ein stadtgebundener Schneefall in Mann-
heim; Meteorol. Rundschau 5, 132-133.
SEC TECHNICAL REPORT A62-5
-------
CITY AIR BETTER OR WORSE 21
Kline, D. B. , and Brier, G. W. (1961); Some experiments on the
measurement of natural ice nuclei; Me. Wea. Rev. 89,
263-272.
Kratzer, A. (1956); Das Stadtklims, 2nd ed. , Die Wissenschaft
Bd. 90, Braunschweig, 184 pp.
Krupins, A. P. , Tyukov, D. M. , and Ponomareva, A. M. (1954).
Effect of atmospheric pollution on the bactericidal effective-
ness of solar radiation, (translated title); Gigiena i
Santiariya, No. 8, 15-18.
Landsberg, H. (1938); Atmospheric condensation nuclei, Erg.
Kosm. Physik, _3, 155-252.
Landsberg, H. E. (1956); The climate of towns; in: Man's Role
in Changing the Face of the Earth, Univ. of Chicago Press,
584-606.
Landsberg, H.E. (1958); Physical Climatology; 2nd ed. Gray
Printing Co. 446 pp.
Lenshin, V. T. (1958); Sulfurous anhydride in the atmospheric
air of Leningrad; Gigiena i Sanitariya 23, No. 8, 57-59.
McDonald, J. F. (1961); Mean atmospheric residence times for
particulate air pollutants; Bull. Am. Meteorol. Soc. 42,
664-665.
Mitchell, J. M. , Jr. (1953); On the causes of instrumentally ob-
served secular temperature trends; J. Meteorol. 10, 244-
261.
Muhleisen, R. (1959); Elektrische Ladungen an Aerosol-. Dunst-
und Nebelteilchen; Ber. d. DT. Wetterdianstes 7 (Nr. 51),
62-66. ~
Murray, M. (1957); Snowfall in Leicestershire 18-19 February
1957; Weather 1J2, 329-330.
Neiburger, M. (1955a); Tracer test of the accuracy of trajector-
ies computed from observed winds in the Los Angeles area;
Air Pollut. Foundat. Rep. No. 7, 55 pp.
Neiburger, M. (1955b); Visibility trends in Los Angeles; Air
Pollut. Foundat. Rep. No. 11, 45 pp.
Neiburger, M. (1959); Meteorological aspects of oxidation type
air pollution; in: The Atmosphere and the Sea in Motion,
New York, Rockefeller Inst. Press and Oxford University
Press, 158-169.
Okita, T. (1960); Estimation of air flow from observation of rime
ice; J. Meteorol. Soc. of Japan, Ser II, 38, 207-209.
Potter, J. G. (1961); Changes in seasonal snowfall in cities;
Canad. Geographer _5, 37-42.
Potter, J. G. (1961); Areal variation of snowfall in metropolitan
Toronto; Meteorol. Branch-Sept, of Transp. -Canada, CIR-
3431, TEC-342, 7 pp.
Reiter, R. (1960); Meteorobiologie und Elektrizitat der Atmos-
phare. Probl. d. Biokl. Ed. 6) Akad. Verlagsges Leipzig
1960, 424 pp.
SYMPOSIUM: AIR OVER CITIES
-------
22 LANDSBERG
Roach, W. T. (1961); Some aircraft observations of fluxes of
solar radiation in the atmosphere; Qu. Journ. R. Meteorol.
Sec. _8_7, 346-363
Semonin, R. C. , and Stout, G. E. (1961); Atmospheric particu-
lates in precipitation physics; 111. State Water Survey,
Meteorol. Lab., Final Rep. Nat'l.'Sci. Found G-6352,
18 pp.
Steinhauser, F. (1959); Der Kohlenoxyd-Gehalt der Lufts in Wien
and seine Abbangigkeit von verschiedenen Faktoren; Ber.
d. Dt. Wetterdienstes _7 (Nr. 51); 53-61.
Steinhauser, F. , Eckel, O. ,~and Sauberer, F. (1955, 1957,
1959); Klima und Bioklima von Wien; Wetter und Leben,
Pt. I Sonderheft 3 Vol. T_> 12° PP-
Pt. II Sonderheft 5 Vol. 1), 134 pp.
Pt. Ill Sonderheft 7 Vol. Tl, 136 pp.
Sundberg, A. (1950); Local climatological studies of the temper-
ature conditions in an urban area; Tellus 2, 222-232.
Sundberg, A. (1951); Climatological studies in Uppsala, with
special regard to the temperature conditions in the urban
area. Uppsala Univ. Geog. Inst. , Geographies No. 22.
U. S. Bureau of the Census (1961); United States Census of Popu-
lation, United States Summary, Number of Inhabitants,
Final Report PC(1), Washington, 140 pp.
Wanta, R. C. , Moreland, W. B. , and Heggestad, H.E. (1961);
Tropospheric ozone: an air pollution problem arising in
the Washington, D. C. , metropolitan area; Mo. Wea. Rev.
89_, 289-296.
Weiss, E., and Frenzel, J. W. (1956); Untersuchungen von
Luftverunreinigungen durch Rauch-und Industriegase im
Raum von Linz; Wetter und Leben 8, 131-147.
World Health Organization (1961); Air Pollution, Monogr. Series
46, Columbia Univ. Press, New York, 442 pp.
SEC TECHNICAL REPORT A62-5
-------
Air Pollution Studies in the Netherlands
F. H. SCHMIDT, Royal Netherlands
Meteorological Institute
Summary
A study is made of the increase in concentration of atmospheric pollution at the
earth's surface in large cities and industrial areas when stability increases. By means
of some simplifying assumptions, it is shown theoretically that when stability increases
discontinuously the increase in concentration of pollution initially is proportional to time
and for extended periods of stability is proportional to the square root of the time. A
linear increase of concentration with time is also found with a slow increase of stability.
The theory is illustrated numerically by data from measurements obtained in three case
studies. The empirical data are reasonably well represented by theoretical formulas.
Air pollution began to be felt as a problem in the Nether-
lands almost immediately after the second World War, when the
industrialization of the country progressed rapidly. Its harmful
effects, even in a flat and windy country like the Netherlands,
were experienced for the first time in an area northeast of big
steelworks to the west of Amsterdam when severe damage was
done to flower estates. Figure 1 shows the total damage observed
at the end of summer, expressed in a relative measure; the
Sutton pattern can be recognized clearly. The curves are less
flat than they should be according to theory, because the wind
direction was, of course, not constant during the whole season.
The problem has been solved, more or less, since the steel-
works built a 150-meter chimney.
At the same time the ever-increasing industrialization in
and around the second largest harbour of the world, (New York
being the largest) Rotterdam, just to the south of one of the most
important horticultural districts of the country, became more
and more a nuisance. In this case we could not attribute any
damage to one special industry but had to consider the integrated
effect of a large number of sources at various heights; this led
us automatically to a study of the problems that are specific for
the topic of this symposium.
We considered several aspects of the behaviour of pollu-
tion in such an area; special attention has been given to the role
of atmospheric stability in some cases. Three examples are
given in this paper and are illustrated with some figures.
First we tried to get an idea of what happens to concentra-
23
-------
24
SCHMIDT
Figure 1. Patterns of pollutant damage to vegetation near Amsterdam.
tions in an urban area when stability in that area increases sud-
denly. In such cases, at least in the dangerous ones, there is
generally little or no wind and the pollution balance is almost ex-
clusively governed by the production of pollution and its upward
transportation by turbulence.
Assuming equilibrium during the period preceding the
stability increase, including the effect of advection into the ver-
tical transport, which is of course physically not correct, and
taking an area so large and homogeneously contaminated that the
variations of concentration in horizontal directions may be ig-
nored, another relatively rough approximation, we get equations
that can be solved by Laplace transform (see Bouman and
Schmidt, 1961). We are interested in the variation of ground
concentration with time after the moment stability increases sud-
SEC TECHNICAL REPORT A62-5
-------
AIR POLLUTION STUDIES IN THE NETHERLANDS
25
denly. The result is that after a short period, during which only
the sources in ground level automobiles for example add to
the ground concentration, all sources start to increase, resulting
in an increase of that concentration proportional with the square
root of time.
The result has been tested successfully in a number of
cases. Figure 2 shows as an example the SO2 measurements
made in the centre of London during the great smog of December
1952. The theory provides a possibility within certain limits to
forecast to what level concentrations may rise under such cir-
cumstances.
5 05
8 04
o*
Q3
0.2
O.t
December
Figure 2. Measured concentrations of S02 in London during smog episode, December 1952.
Secondly the hourly observations of the dust content of the
air, made in the centre of Rotterdam and measured by the well-
known filter method, have been analysed harmonically. General-
ly, the dust content over the day, averaged for instance for every
month of the year, is such that we find the curve reaches a pro-
nounced maximum in the morning and more or less flattens for
the rest of the day. It may be assumed, therefore, that it is
especially the morning part of the curve that determines the har-
SYMPOSIUM: AIR OVER CITIES
-------
26
SCHMIDT
monic components. It is found that the first harmonic generally
has the largest amplitude of all.
If stability plays a role in the daily variation of the dust
content of the air, its influence may be assumed to appear espe-
cially in the first harmonic. If the morning observations are
assumed the most important ones, therefore, through the year
we may expect a shift of the time at which the first harmonic
shows its maximum.
Figure 3 shows the time of sunrise and the times at which
the first harmonic and the combination of the second and the third
harmonics obtain maximum values. It seems that the part of the
morning maximum that is due to stability occurs about 1 hour
after sunrise.
JAN FEB MAR APR MAY JUNE JULY AUG SEPT OCT NO
Figure 3. Harmonic component patterns based on hourly dustfall observations in Rotter-
dam.
Finally, if the first harmonic depends mainly on stability,
its relative amplitude, i.e., its contribution to the total morning
maximum, may be assumed to depend on cloudiness. High values
of cloudiness result in small variations in stability, and vice
versa. This is even more true with respect to precipitation.
Apart from its direct connection with cloudiness, precipitation
wets the earth's surface and thereby reduces the daily variation
SEC TECHNICAL REPORT A62-5
-------
AIR POLLUTION STUDIES IN THE NETHERLANDS
27
in stability. In Figure 4 the total monthly amount of precipitation
is plotted against the amplitude of the first harmonic over the
total morning amplitude; the effect of precipitation can be observ-
ed clearly.
JAN FEB MAR APR MAY JUNE JULY AUG SEPT OCT HOY DEC
Figure 4. Relationship of harmonic pattern to precipitation data.
Cities influence the circulation, as do large industries that
have a considerable heat output. It was observed that in Vlaar-
dingen, a small town west of Rotterdam, the volatile products of
the Shell refineries south of it could be smelled during situations
with westerly winds. This could only be explained by assuming
a vertical circulation due to the heating of the atmosphere over
the Refineries, air rising over them, and eventually descending
over the town.
In order to detect this circulation, wind observations were
made around the Shell complex during situations with westerly
winds, and the divergence was determined.
Figure 5 shows the divergence pattern that has been com-
puted from these observations, translated into vertical velocities
in centimeters per second, based on the assumption that the di-
vergence pattern is constant between the ground and 100 meters,
the maximum height of the chimney stacks. The figure confirms
that air rises over the region with largest heat output and de-
scends to the north of the refineries. This result was based on a
relatively small number of observations (17); however, direct
measurements of the vertical motions with pilot balloons have
given similar results.
SYMPOSIUM: AIR OVER CITIES
-------
28
SCHMIDT
-10
0 Vz in cm/sec.
Figure 5. Computed wind divergence over industrial complex near Vlaardingen during
westerly winds.
BIBLIOGRAPHY
F. H. Schmidt, 1957 : On the diffusion of stack gases in the at-
mosphere; Koninklijk Nederlands Meteor-
ological Institute, Mededelingen en
Verhandelingen 68.
F. H. Schmidt, 1960 : On the dependence on stability of the para-
meters in Button's diffusion formula;
Beitrage zur Physik der Atmosphare 33,
p. 112.
SEC TECHNICAL REPORT A62-5
-------
AIR POLLUTION STUDIES IN THE NETHERLANDS 29
D. J. Bouman and F, H. Schmidt, 1961 : On the growth of ground
concentrations of atmospheric pollution in
cities during stable atmospheric condi-
tions; Beitrage zur Physik der
Atmosphare 33, p. 215.
DISCUSSION
FROM THE FLOOR: Plant damage around the steel mill,
was it sulfur dioxide damage or chloride damage?
DR. SCHMIDT: Chloride.
MR. ROBINSON: In the diurnal stability pattern did you
include a diurnal wind factor, also?
DR. SCHMIDT: No, it is very small. There may be an
influence of land or sea breeze, even at Rotterdam, although it is
30 kilometers from the city.
We didn't have a measure for the daily variation of
stability. We just looked at pollution and said, "Well, pollution
shows a daily variation. And, can we understand that by ascrib-
ing it to the daily variation of stability. " At the present time we
do not have the means in the Netherlands to measure stability
correctly. We have radiosonde observations, but it is hardly
possible to say something about stability in the lower 100 or 200
meters with the observations as they are confined to two per day.
What we did was just look at the daily variation of pollution,
taking that daily variation as a measure of the daily variation of
stability, then compare that daily variation of pollution with the
amount of rain, for instance, the amount of rain giving an
indication of the cloudiness, not a very exact indication, but
giving an indication. And when we did that we found a connection
between the two that we think we can understand.
SYMPOSIUM: AIR OVER CITIES
-------
Recent Developments in the Chemistry
of Urban Atmospheres
JAMES P. LODGE, Jr., National Center
for Atmospheric Research
Summary
The role of meterolog ica I factors in the transport of air pollutants is widely known.
The effect of air pollutants upon weather and climate, however, has received far less at-
tention. Probably the substance that has been studied most is carbon dioxide. It is
generally agreed that the concentration of this compound in the earth's atmosphere has
increased since the turn of the century, though the exact amount of increase is still un-
settled. There is still less agreement on the resulting effect on the global heat balance.
Urban air contains many more condensation nuclei than country air, and numerous un-
supported statements have related this fact to smog formation. A better case can be
made for an effect of sulfur dioxide, which is absorbed by fog droplets and is partially
oxidized to sulfuric acid. This process should have a profound effect on fog persistence.
Numerous other substances can be cited that must have some sort of modifying effect
upon the urban microclimate.
What I would like to talk about has been in some measure
preempted by Dr. Landsberg, but I am going to go over it again
because perhaps stated from the viewpoint of a chemist it will be
received with a little different emphasis. I want to talk about the
atmospheric effects of chemical pollutants.
Considerable work has been done on the effect of the atmos-
phere on its load of chemical species. Some of this work has
been of good quality, and some very poor, but nonetheless there
has been substantial continuing research effort on the atmospheric
transport of pollutants and on the changes undergone by pollutants
as they are being carried. On the other hand, with one or two
rather small exceptions astonishingly little active work has been
done on the effect of chemical species, as such, on the weather,
microclimate and climate of the urban area.
Certainly the best-known and most-studied effect is the so-
called greenhouse effect of carbon dioxide; however, the exact
magnitude of this effect still appears, so far as I can tell from
current literature, to be under debate.
The amount of increase of carbon dioxide in the atmosphere
over the past 60 years or so is not known. In 1961 we still do not
have sufficiently accurate methods of measuring carbon dioxide to
permit determination of this increase over a short period. The
severity of the need for such methods was brought to my attention
just a few days ago by a colleague who is trying to measure carbon
31
-------
32 LODGE
dioxide at Point Barrow. He needs a method or device that will
measure the very small gradients we would expect to find in the
proximity of sinks, i. e. , a device that might measure hundredths
of a part per million carbon dioxide. I did not unduly encourage
him, since this calls for an accuracy of roughly one in 30, 000,
which is very good accuracy for the best-established techniques
of quantitative analysis of substances much easier to determine
than carbon dioxide.
It has frequently been stated and was reiterated by Dr.
Landsberg that cities are a source of very large numbers of con-
densation nuclei. The exact nature of their effect remains un-
demonstrated. Some very careful experiments concerning the
nature of condensation nuclei formation in flames have been
made; however, to my knowledge, only the rate of production of
condensation nuclei by individual fuels has been demonstrated.
Many of the condensation nuclei produced by human activity are
so small they simply are not activated in the natural atmosphere.
Certain other individual species present in the particulate
state in urban air have a substantial effect on fog persistence.
For example, the output of sulfur dioxide reflected in an increase
in the amount of sulfur trioxide or sulfuric acid has a substantial
effect on fog persistence, as is now clearly demonstrated by work
in London and elsewhere. Be-e-ause of the extremely hygroscopic
nature of sulfuric acid, there can be no question whatever that the
presence of sulfur dioxide does result in an increase in the sul-
furic acid content of the air, and thereby frequently stabilizes and
perpetuates fogs at conditions well below water saturation.
There is still another effect of city gases such as sulfur
dioxide, and this is the conversion of certain other species to
sulfates and nitrates, and the like. For example, it has been
shown by Jung that the one place large amounts of particulate
nitrate are found in the air seems to be over coastal cities. In
these cities salt fresh from the sea is brought into contact with
oxides of nitrogen. Chemically this is very fascinating. The
sulfates present in urban air seem to be ammonium sulfate and
calcium sulfate. Sulfur dioxide is converted to sulfuric acid and
thence to sulfate and seems in that state to tie up primarily with
ammonia. Since lime and ammonia are the most active materials,
they present the finest chance for reaction; nitrogen dioxide then
comes along but must be contented to react with the only thing
left, sodium chloride, which is less favorable for reaction. The
sulfur dioxide has grabbed off everything else.
In a very clean atmosphere, if the sulfur dioxide were
forced to react with the sodium chloride because of the absence of
other good surfaces, we would have an interesting situation.
SEC TECHNICAL REPORT A62-5
-------
CHEMISTRY OF URBAN ATMOSPHERES 33
According to the data that I can find, the transition humidity for
sodium chloride, i. e. , the relative humidity at about room tem-
perature at which it undergoes transition from a solid particle to
a liquid drop, is about 74 percent. On the other hand, the cor-
responding transition humidity for sodium sulfate decahydrate,
which is certainly the species that would be present under these
conditions, is 93 percent. This would have a profound effect on
fog persistence under these conditions.
I am not able to find the transition humidity for sodium
nitrate in the literature; however, this would be very interesting
since these chemical reactions do result in a shift in the relative
humidity at which haze formation is initiated. This is an effect
that to my knowledge has not really been investigated in relation
to our present knowledge.
Another species put into the air of cities in quantity is soot.
I find it very fascinating that soot is put out and we have talked
about sootfall, and so forth, and yet nobody has tied this in with
the series of experiments done by, among others, Dr. Florence
Van Straten. These experiments indicate at least a good possi-
bility of the initiation of convective activity by heat absorbed by
clouds of soot.
Without passing for the moment on the validity of Dr. Van
Straten's experiments, which I am not competent to assess and
which I think no one else has fully assessed as yet, this certainly
presents another source of possible meteorological change from
pollutants.
Another interesting bit of recent knowledge is the fact that
a number of species, most particularly metallurgical fumes,
seem quite active as freezing nuclei at reasonably elevated tem-
peratures; and still more recent knowledge, completely unassim-
ilated as yet, that certain organic compounds are extremely ac-
tive in promoting freezing of supercooled water.
Head in Australia demonstrated freezing on the surfaces
of some fairly esoteric organic compounds at temperatures as
warm as -2. 5°. I will say, lest I frighten anyone, that this was
under highly unrealistic conditions. Attempts to get classical-
type cold-box nucleation by the same species have not been suc-
cessful as yet. Head's work certainly indicates the possibility
that cities can act as rather potent sources of freezing nuclei,
and that this may, in fact, affect precipitation patterns over
cities.
Another substance briefly mentioned by Dr. Landsberg, and
certainly on a large scale neglected as a pollutant, is water vapor.
Not too long ago, while I was still working here at Taft Center, I
SYMPOSIUM: AIR OVER CITIES
-------
34 LODGE
had a visitor who wanted to know how to remove his chief prob-
lem pollutant, which was water vapor. His stacks discharge
quantities of water vapor just upwind of a major highway, and
under these conditions you can have a severe pollution problem
from water. There is, furthermore, the possibility already
raised by Dr. Landsberg that this added water has an impact upon
the total water balance of the city, in terms of increasing precip-
itation.
Finally, we have a rather more subtle effect of chemical
species on the weather, a secondary effect. I can point out areas
such as that region between the City of Houston, Texas, and the
Gulf, for example, which have been thoroughly denuded of vege-
tation by uncontrolled discharge of pollutants trees are dead,
grass in many cases is gone. I think this is an effect that has
been very little thought of, but we have denuded these acres of
ground. We have reduced them to the bare earth. Certainly the
effect of this on the heat balance must be great. No one would
claim that the albedo of bare earth was identical with that of a
grass- and tree-covered expanse.
In fact, as is well known, Dessens, working in Africa, has
attempted to show, I think with relatively little success,, actual
cloud formation over a simply large black area of ground. If he
puts in a little additional heat from a few burners, which we fre-
quently do ourselves, by building a plant in the middle of such a
desert, then he can get very fine formation of convective clouds.
On one noteworthy occasion Dessens produced a tornado.
I am not trying to use scare tactics here. I am merely
pointing out that we can hardly do anything without upsetting some-
thing in our balance of nature. We must weigh very carefully
whether the results are worth the upset.
DISCUSSION
DR. MITCHELL: I was, of course, interested in the allu-
sion you made to the fact that certain organic compounds might
act as freezing nuclei at as high a temperature as -2. 5°, and I
just wanted to have you clarify, if you could, whether the evidence
for this was in context of the small droplets as we find them in
the atmosphere or at larger bulk water masses. This would be
highly interesting.
DR. LODGE: Probably the first publication on this work
was a brief note in Science, I think this past September 9 by R
B. Head of Melbourne. The chemical compound was melted on a
microscope slide and allowed to solidify; very small drops were
SEC TECHNICAL REPORT A62-5
-------
CHEMISTRY OF URBAN ATMOSPHERES 35
condensed on the solidified compound by chilling the microscope
slide, and then observed for freezing.
Now the droplets themselves are small. The number of
active sites is rather slight, and appears to coincide with some
sort of unidentified lattice falls. The organic compounds in ques-
tion are pure only to the extent of giving sharp melting points, and
consequently I can't say beyond a shadow of a doubt that it is not
due to some impurity. This I am planning myself to do a little
work on. But there was one compound that did nucleate under
these conditions at some times at temperatures as warm as -2. 5°,
and several above -10°. Primarily steroids, but also cholester-
ol, which is quite active in this as well as in some other things,
and fluorene derivatives.
DR. HEWSON: I would like to ask Dr. Lodge what he es-
timates the improvements in the techniques for measuring carbon
dioxide have been since the beginning of the century.
DR. LODGE: That's sort of a tricky question. I would say
this: at best our methods now are very much better, a fine ex-
ample being, on the days when they are working, the nondisper-
sive infrared instruments, such as those used by Keeling and
Kanwisher, and others.
As I know you realize, there are many days on which these
instruments are not working and our big difficulty is the total lack
of a true referee method which permits calculation from first
principles of what the concentration is that the instruments are
measuring.
Their reproducibility can be made very good. Their ab-
solute accuracy I think is largely unknown. It is my opinion that
within a few years there should be methods that would be good to
at least a tenth of a part per million. I think we are at last be-
ginning to learn a few things about this, and I believe that the
accuracy, precision, or discrimination, of the order of a part
per million, necessary to sense gradients over water and so on,
can be achieved by simply dodging the issue and measuring the
gradient itself, directly, say, with a concentration cell, rather
than undertaking to make two separate determinations and sub-
tract.
DR. HILST: Dr. Lodge, what scale of length or area do
you have in mind with regard to the changing of surface conditions
and how this might affect the area. Is it an area of a square mile
or are you thinking of counties, states, continents? Or all of it?
DR. LODGE: The specific case to which I alluded would be
the area that we all know about, of the order of less than a square
mile, an area which has been denuded simply by chemical action.
SYMPOSIUM: AIR OVER CITIES
-------
36 LODGE
I am not referring specifically to paving. I am referring simply
to the destruction of vegetation by uncontrolled discharge of pol-
lutants.
DR. HILST: The effect of it, of course, would depend on
how dynamically or hydrostatically stable or unstable the atmos-
phere reacted.
DR. LODGE: Absolutely.
FROM THE FLOOR: In view of the available methods for
measuring pollution, data which are available from history, and
what is liable to happen in the future, what is your view on the
best indicator of trends of pollution, manmade pollution?
DR. LODGE: As things stand, and assuming that we don't
learn a tremendous amount more about air cleaning than seems
likely, I would stick by my statement of some years ago that car-
bon dioxide was probably the best index. Because it is produced
in all human activity. ,
REFERENCES
Lodge, J. P. , Ed. , Atmospheric Chemistry of Chlorine and Sul-
fur Compounds, NAS-NRC, Publ. No. 652, 1959, page 47.
Van Straten, Dr. Florence, U.S. Naval Res. Lab. Rep. 5235,
Washington, D. C. , 25 pp. 1958.
Head, R. B. , (1961), Nature, 191, 1058.
Dessens, H. , (1960), in H. Weichmann, Physics of Precipitation,
NAS-NRC Publ. No. 746, pp 396-401.
SEC TECHNICAL REPORT A62-5
-------
A ClimatologicaI Evaluation of
Precipitation Patterns Over An Urban Area
STANLEY A. CHANGNON, Jr.,
Climatologisl, Illinois State Water Survey*
Summary
A 13-year measurement program of urban rainfall distribution in a moderate-sized
Illinois municipality is described. A dense raingage network with one gage per square
mile has been in operation since 1948 in a community of 70,000 population. The month-
ly, seasonal, and annual precipitation patterns derived from the network data are dis-
cussed, and suggestions are offered concerning the possible urban influences on the
patterns. The annual average precipitation pattern is compared with that from a nearby
rural network to help evaluate the apparent urban-affected precipitation pattern obtained
over the urban area. Recommendations are presented for the type of measurement pro-
gram necessary to determine whether urban influences are producing measureable effects
on the amount of precipitation over and downwind from the urban areas.
INTRODUCTION
Precipitation data from a dense network of raingages lo-
cated in a moderate-sized urban area in central Illinois have pro-
duced a 10-year precipitation pattern which suggests the presence
of urban influences. Statistical tests and theoretical considera-
tions have been applied to these findings in hopes of determining
whether the urban influences were responsible for additional pre-
cipitation on the downwind side of the urban area. From these
investigations, conclusions as to other possible causes or explan-
ations of the additional downwind precipitation are drawn, and
suggestions are offered concerning urban precipitation measure-
ment programs.
PREVIOUS FINDINGS
Before the Illinois study is discussed, it is appropriate to
report briefly on some of the earlier findings. Undoubtedly, the
most recent comprehensive resume of findings relating to urban
effects on precipitation is the one by Dr. Landsberg. * Dr. Lands-
berg presented information on the various urban conditions affect-
ing precipitation conditions as well as information on the observed
precipitation increases in urban areas. He refers to other pub-
lications which show that urban areas, as opposed to rural areas,
*Read by Glenn E. Stout for Stanley A. Changnon, Jr.
37
-------
38 CHANGNON
have been noted to have (1) more rainy days, (2) more thunder-
storm days, and (3) more total precipitation. Various studies ot
European and North American conditions have shown that in urban
areas precipitation can be expected to range from 5 to 10 percent
above that in nearby rural areas. The number of days of light
rain and of thunderstorms were reported to average from 11 to 18
percent more in urban areas than in nearby rural areas.
Three factors are listed as principal causes of urban-
induced increases in precipitation. These urban factors are addi-
tional condensation nuclei, additional water vapor, and turbulence
resulting from thermal convection and increased surface rough-
ness. Additional nuclei are designated as playing a more im-
portant role in the colder half-year when frontal precipitation in
the central United States is most frequent. Because of this, Dr.
Landsberg stated that the urban effect on precipitation is most
pronounced in the colder half-year of humid continental climates.
DESCRIPTION OF ILLINOIS PROJECT
As so often occurs in science, a project started for one
purpose eventually finds another and often an equally useful scien-
tific objective. Such is the case with our urban precipitation
project in Illinois.
In late 1948, the Illinois State Water Survey, in collabora-
tion with-the University of Illinois College of Engineering, in-
stalled nine recording raingages in Champaign-Urbana, Illinois.
The locations of these gages, numbers 1 through 9, are shown in
Figure I. These were installed to furnish data for a hydrologic
study of rainfall-runoff relationships on an urban watershed lo-
cated in these adjoining communities. At that time, two other
recording raingages, numbers 11 and 12, were already in opera-
tion in the urban area, and by 1950 a twelfth gage, number 10 on
the map (Figure 1), was installed. Data collection and routine
analysis have been continuous since these 12 gages were installed.
The data collected in the 1950-1959 period have been carefully
re-analyzed and the results are used as a basis for the urban
precipitation study. Extreme care was exercised in the installa-
tion of these gages in an effort to obtain comparable exposures,
and since installation, the gages have had regular maintenance
and servicing by one technician. Thus within normal performance
limitations, the data from this network are considered to be of
exceptional quality.
In many respects Champaign-Urbana, which is about 130
miles south of Chicago in east central Illinois, is in a uniauelv
suitable location for evaluating urban influences on precipitation
SEC TECHNICAL REPORT A62.5
-------
URBAN PRECIPITATION PATTERNS 39
These adjoining cities, which are actually a single urban area,
lie in an exceptionally flat featureless glacial plain with no sig-
nificant changes in relief within 100 miles. The entire surround-
ing area also is devoid of any large-sized water bodies. The
rural land is devoted to extensive cash-grain farming, chiefly
corn and soybeans. The nearest urban communities with appre-
ciable industrial development are more than 40 miles away. Dur-
ing the 1950-59 period, a complex of petro-chemical industries
was built in a rural area 23 miles south of Champaign-Urbana.
In general, it is believed that no natural or cultural factors that
might significantly affect the distribution of precipitation exist in
the region outside the urban area.
The twin cities function basically as a university-residential
area. There are a few local industrial and heating plants as
shown on Figure 1, but these are of minor importance and are
not heavy industries. During the 1950-59 decade, the urban area
has undergone growth. The corporate area was 9. 8 square miles
in 1950 and was 11. 2 square miles by 1959; therefore, the rain-
gage density has been about 1 gage per square mile. The urban
population increased from 62, 000 in 1950 to nearly 77, 000 in
1959. As shown in Figure 1, the urban area is rectangular with
the longer dimension of 5 miles lying east-west. The mean north-
south width is 2. 5 miles. Impervious surfaces have been com-
puted to be 38 percent of the total corporate area. ^ Until recent
years, most of the urban area including the University of Illinois
campus had an extensive tree growth but this has been greatly re-
duced by tree diseases. The only other local change of any pos-
sible significance was the addition in 1957 of a soybean processing
plant in the vicinity of gage 4 (Figure 1).
In order to have a measure of precipitation for the surround-
ing rural area, records from three raingages, which were in op-
eration during most of the 10-year period, were employed. These
gages, as shown on Figure 1, are at Rising, a village 3 miles
west-northwest of the urban area; at the University of Illinois
Airport, 3. 5 miles south of the urban area; and at St. Joseph,
about 7 miles east of the urban area. These are all unshielded
recording raingages operated by the Illinois State Water Survey.
Four weather stations in the area (Figure 1) were operated simul-
taneously during the July 1959 to June 1960 period to obtain de-
tailed surface temperature measurements.
INTRODUCTION TO PROBLEM
The possibility of the existence of urban effects on the pre-
cipitation distribution in Champaign-Urbana is based on the pre-
mise that these effects will produce more precipitation in the
SYMPOSIUM: AIR OVER CITIES
-------
40
CHANGNON
• RISING
URBAN
BOUNDARY-^
X
I
I
r
DMW (
«4
•5
CHAMPAIGN
x«7
I 1
—i i r
1_J L_
.9 L
URBANA i
JSCENTRAL
ST JOSEPH
AIRPORT
-KEY-
X Principal Sources of Smoke
and Water Vapor
Q Weather Station
• Recording Raingage
0123
SCALE, Miles
Figure 1. Champaign - Urbana and Vicinity
eastern portions of the urban area than elsewhere in the urban
area. This is considered valid since most precipitation systems
and individual cells generally move from a westerly direction in
this area. ^' With such an average movement it is believed that
urban effects on precipitation would not be realized until the pre-
cipitation system had passed over the western portion of the urban
area. Fluctuations in the speed of cell movement, the stage of
storm development, and the speed and direction of the surface
winds would cause urban-affected precipitation conditions to max-
imize at different locations in the eastern half of the urban area,
or possibly in the rural area east of Urbana.
Radar studies of precipitation cells and organized lines of
precipitation in this area have shown that precipitation cells and
lines generally move at speeds of 10 to 30 mph. ^> 5, 6 ^ an
average speed of 20 mph and an average cell diameter of 5
miles, 6 a cell moving from the west over the urban area would
require 30 minutes to pass entirely across the urban area. This
is from the time the cell's leading edge reaches the western
boundary until the trailing edge leaves the eastern edge of Urbana.
Such a situation would require urban effects from Champaign
(west) to act on the cell during the 15 to 20 minutes the cell is
passing over Champaign. Since the average life cycle of a
SEC TECHNICAL REPORT A62-5
-------
URBAN PRECIPITATION PATTERNS 41
thunderstorm is on the order of 60 minutes, this means that
most thundershowers and rainshowers passing over Champaign-
Urbana from the west spend nearly one-half their life cycle under
the possible influence of the city. These statistics give some
order of magnitude for the time-space relationship between pre-
cipitation conditions and this specific urban area.
If a 10-year ma'ximization of precipitation occurred in the
eastern urban area, this could be explained by any one of three
possible factors: first, urban conditions may have produced addi-
tional precipitation (urban effects); second, more rain cells may
have formed or maximized naturally over Urbana (natural varia-
tion of precipitation); and third, more sheltered exposures for
the raingages may have existed in the eastern area (variability in
gage exposures).
Description of the 1950-59 annual and seasonal precipita-
tion patterns and the precipitation patterns displayed by individual
precipitation periods in the 1958-1959 period will be presented,
and from these and other supporting data, explanations will be
offered relating to the three possible causes of the precipitation
patterns observed.
AVERAGE ANNUAL PRECIPITATION
Figure 2 shows the average annual precipitation pattern
31
30
Figure 2. Average annual precipitation for 1950-1959, in inches.
SYMPOSIUM: AIR OVER CITIES
-------
42 CHANGNON
based on the data for 1950-1959. Note that the highest values oc-
curred over the ESE portions of the urban area. The entire
urban area is the center of the region of greatest precipitation
which diminishes in all directions away from the urban area. In
general the rural to urban relationship is 30 to 34 inches, re-
spectively and the difference amounts to 12 percent. This com-
pares favorably with annual percentage differences of 5 to 10 per-
cent reported by Dr. Landsberg.
Although not discernible in Figure 2, the average annual
precipitation pattern for 1950-1959 across Missouri, Illinois,
and Indiana had a west-to-east increase, as based on First-Order
Station data. In the area of Champaign-Urbana this west-to-east
increase was at the average rate of one inch per 20 miles; this
indicates that with no urban influence a west-to-east increase of
0. 25 inch should have occurred from the west side of the urban
area to the east side.
A 10-year precipitation record in a humid continental cli-
mate can be considered of questionable significance, since it may
represent an abnormal period because of temporal and spatial
precipitation variations. One estimate of the possible validity of
the maximum in the east urban area was obtained by comparing
long-term records of two urban raingages. Gage 11 in the high
area (Figure 2) has been in operation since 1903, and gage 12,
with a 10-year average nearly 2 inches lower than gage 11, has
been in operation since 1930. The 1930-1959 average annual pre-
cipitation at gage 11 is 37. 3 inches compared to 34. 5 inches at
gage 12. The 30-year average annual difference of 2. 8 inches is
in the same order of magnitude as that shown in Figure 2, and
thus lends credulence to the 10-year pattern. The 45-year nor-
mal annual value for gage 11 is 36. 6 inches; therefore, the 34. 4
inches recorded as an average for the 1950-1959 period repre-
sents a below-normal departure, which is largely a result of
drought conditions in 1952-1955. 7
The validity of the pattern shown in Figure 2 also was in-
vestigated by inspecting the frequency distribution of the annual
ranks based on the annual totals of each of the 12 urban gages.
The 12 values in each year were ranked from high (first) to low
(twelfth), and the rank frequencies for the 10 years were then in-
spected. This was used to determine whether the areas of high
and low average precipitation were a result of extremes in just a
few of the 10 years. The rank frequencies for the 12 gages and
an average rank determined for each gage are given in Table 1.
The 6 gages located in the east half of the urban area had 25 of
the first, second, or third ranks, and this total is 10 more than
would be expected if the ranks were normally distributed among
all 12 gages. There was a distinct absence of low ranks (tenth,
SEC TECHNICAL REPORT A62-5
-------
URBAN PRECIPITATION PATTERNS
43
TABLE 1
NUMBER OF TIMES EACH GAGE ACHIEVED VARIOUS RANKS
BASED ON ANNUAL TOTALS AT EACH URBAN GAGE
1950-1959 PERIOD
Distribution of rank by gage number
Ranks
1-2-3
4-5-6
7-8-9
10-11-12
2
1
4
3
2
3
1
2
6
1
4
1
2
0
7
5
1
2
3
4
6
1
4
3
2
7
0
0
5
5
8a
2
1
4
3
_ŁŁ
2
5
3
0
10a
6
3
0
1
ii!
8
2
0
0
12a
5
2
1
2
ll
2
3
2
3
Average Rank 6911 10 5 12 8 3 2
1 4 7
Indicates gages located in the east half of the urban area.
eleventh, or twelfth) among the six easternmost gages. Only nine
of these low ranks were recorded, whereas 15 would normally be
distributed in the east. Also four of the six easternmost gages had
the highest average ranks. These data support the climatological
validity of the annual pattern shown in Figure 2 and indicate that
this annual maximization in the east was a repetitive process in
60
55
70
Figure 3. Total number of 30-minute to 24-hour amounts Ł 2-year recurrence value,
1950-1959
SYMPOSIUM: AIR OVER CITIES
-------
44 CHANGNON
most years. Either gage 10 or 11 ranked first or second in 7 of
the 10 years. In two of the remaining three years, gages 8 and 9
achieved rank one. In only one year, 1959, did the maximum pre-
cipitation and highest ranks occur on the west side of the cities.
The possibility that the 10-year annual pattern might be a
derivative of the frequent occurrence of excessive rainstorms in
Urbana was also investigated. In Figure 3 the urban distribution
of storms in 1950-1959 with recurrence intervals equal to or
greater than 2 years and with durations ranging from 30 minutes
to 24 hours is shown. 8 The greatest frequency of such storms
was in the southwestern urban area and the lowest in the eastern
area, indicating that the average annual precipitation maximum in
the east urban area was not derived from repeated occurrences
of excessive rainstorms in that area.
Several studies in Europe have shown that an urban area,
through urban influences, experiences more thunderstorms than
nearby rural areas. *- An examination of the records from the
Airport weather station and the Central weather station (Figure 1)
revealed that during a 5-year period of comparable records for
the hours between 0700 and 1800 CST the urban station recorded
on the average three more thunderstorm days per year than the
Airport station. The urban average annual 5-year value was 45
thunderstorm days, which is 2 days less than the normal, based
on 54 years of record. This finding tends to corroborate the
belief that urban effects produced by Champaign-Urbana influence
precipitation systems passing and developing over the cities.
Other investigators have analyzed data on the frequency of
days with precipitation and concluded that an urban effect exists.
A European study revealed Munich had 11 percent more days with
precipitation in the range of 0. 004-0. 2 inch than the surrounding
country. 1 An urban increase of 12 to 18 percent in these light
rains was found in the Ruhr region of Germany. *• In Illinois a
comparison of 8 years of records showed that the mean annual
number of days with light precipitation (0. 004-0. 24 inch) was 69
at the Airport gage (Figure 1), compared with 78 at urban gage
11. This is an increase of 13 percent, which compares favorably
with the percentages noted in Europe. For days with precipita-
tion of 0. 25 inch or more, there was no significant difference be-
tween the average for the Airport (41 days) and that for gage 11
(42 days).
AVERAGE SEASONAL PRECIPITATION
In Figure 4 the four seasonal average isohyetal patterns
are portrayed. Maximization of precipitation over the eastern
half of the urban area is evident in all four seasons. In Table 2
SEC TECHNICAL REPORT A62-5
-------
URBAN PRECIPITATION PATTERNS
45
the values considered the best approximations of the rural or un-
modified urban precipitation and those best representing the mod-
ified urban precipitation are shown. These data reveal that the
greatest percentage excess in the urban area occurs during the
colder seasons; this agrees with Landsberg's findings for Tulsa.
The least excess occurs during the summer.
The 30-year average seasonal values for gages 11 and 12
1
Figure 4. Average seasonal precipitation patterns for 1950-1959 in Champaign-
Urbana and Vicinity.
SYMPOSIUM: AIR OVER CITIES
-------
46
CHANGNON
TABLE 2
COMPARISON OF AVERAGE SEASONAL PRECIPITATION FROM
RURAL AND UNMODIFIED URBAN AREAS
WITH THAT FROM MODIFIED URBAN AREA
1950-1959
Season
Rural and unmodified,
urban precipitation, in.
Modified urban
precipitation, in.
Difference between rural
and urban-modified, in.
Percentage excess in
modified urban area
Winter
5. 25
6. 25
1. 00
16
Spring
8. 25
9. 50
1. 25
13
Summer
10. 25
11. 00
0. 75
7
Fall
6.25
7. 25
1. 00
14
TABLE 3
COMPARISON OF AVERAGE SEASONAL PRECIPITATION
AMOUNTS AT GAGES 11 AND 12 BASED ON
30-YEAR AND 10-YEAR PERIODS OF RECORD
Seasonal precipitation, in.
1930-1959 Period
Gage 11 average
Gage 12 average
Difference, 11-12
1950-1959 Period
Gage 11 average
Gage 12 average
Difference, 11-12
Winter
6. 55
5. 43
1. 12
6. 47
5. 63
0. 84
Spring
11. 05
10. 11
0. 94
9. 50
8. 64
0. 86
Summer
11.05
10. 65
0. 40
11.08
11. 16
-0. 08
Fall
8. 67
8. 17
0.50
7.31
7.21
0. 10
are shown in Table 3. Gage 12 represents a more rural location
than gage 11 in the urban modified area. The differences between
these two gages, based on data for the 1950-1959 period are
shown also in Table 3. In general, the average seasonal values
SEC TECHNICAL REPORT A62-5
-------
URBAN PRECIPITATION PATTERNS 47
for both gages, based on the more reliable long-term record,
compare favorably with those based on data for 1950-1959. Only
the 1950-1959 spring and fall averages are lower than the 30-year
values. A comparison of the differences between the values from
the two gages reveals that seasonal differences based on the two
periods of record are remarkably similar. These findings tend
to substantiate the fact that although the 1950-1959 precipitation
was somewhat below normal at a point, the variations shown be-
tween points may be representative of those expected over a
longer and, climatologically, a more reliable period of time.
Winter The winter precipitation pattern (Figure 4) re-
veals a condition not found in the other three seasons. The east
side urban area high extends to the east into the rural area, in-
dicating that if urban influences do produce additional winter pre-
cipitation the resultant greater precipitation is prolongated in
time and continues over a greater distance than in the other sea-
sons. This could be explained partially by the greater winter
frequency of more extensive frontal precipitation systems, and
by the fact that, in general, urban nuclei effects on precipitation
systems are most pronounced in the winter season. The per-
centage excess listed in Table 2 is highest for winter also.
Spring The area of maximum precipitation in spring ap-
parently extends 2 or 3 miles east of Urbana, but at the St. Joseph
rural location the average value is much lower than the urban
maximum values. A secondary high in the isohyetal pattern ap-
pears at the Airport gage. The increase in precipitation across
the urban area is much more rapid over a short distance, from
8. 75 to 9. 5 inches in 1 mile, than any comparable increase over
a short distance in winter, although this could be largely a sam-
pling variation.
Summer Summer is the only season in which the east-
side urban high apparently does not extend into the rural area to
the east. The presence of a rural average, at the Airport,
greater than the values in the urban high (Figure 4) strongly sug-
gests that the summer pattern may largely be a result of natural
rainfall variations. At least, natural variability, which maxi-
mizes in the summer, helps to confuse the assignment of any in-
creases to urban influences. The differences between the aver-
ages of the three rural gages also substantiate this conclusion.
The Airport value is 11. 49 inches, the Rising value is 10. 41
inches, and the St. Joseph value is 8. 33 inches. The maximum
difference between these three values exceeds any variation ob-
served between any of the 12 urban gages. Landsberg has stated
that urban influences on precipitation have least effect in the sum-
mer, or warm season, and this is supported in Table 2,which
reveals that the lowest seasonal percentages of urban excess
SYMPOSIUM: AIR OVER CITIES
-------
48
occurred in the summer.
Fall
CHANGNON
The fall isohyetal pattern closely resembles the
pattern; however,, changes in precipitation across the
The ur-
sprin
urban area are more gradual in fall than in the spring.
ban excess in fall produces the second greatest seasonal per-
centage, as shown in Table 2.
EVALUATION OF INDIVIDUAL PRECIPITATION
PERIODS IN 1958-1959
All precipitation periods in 1958 and 1959 on the urban
area network were analyzed to investigate more closely the con-
ditions with which precipitation increases occurred in the east-
ern portion of the urban area. As stated previously, precipita-
tion maximization in the east was considered indicative of pos-
sible urban influence. A precipitation period was defined as a
distinct period of precipitation, with no precipitation 6 hours
before the period began and none for 6 hours after the period
ended. This delineation of the data yielded 262 precipitation per-
iods in the 2 years. Each of these precipitation periods was
mapped and isohyetal patterns were drawn.
The urban area was divided by two lines, as shown in
Figure 5, so that a western, eastern, northern, or southern half
1 t
I '
3 I1
^- I1-
~"*5 -~~. !•'
.7^
'<-. -- 1 •"--
-.- _,._!".;-">
u
WEST 1 EAST
_ _ —1
! ,-'
i i__
•9 L
1
I
1
1
— .. _4_NORTH
10?-' SOUTH
i
i
.1
Figure 5. Division of urban area for study of precipitation pattern
SEC TECHNICAL REPORT A62-5
-------
URBAN PRECIPITATION PATTERNS 49
of the urban area could be delineated. Each directional half had
six raingages. Based on this form of separation each isohyetal'
pattern for a precipitation period was classified according to the
half containing the maximum precipitation. This half was defined
as the area containing the four highest precipitation values; if no
half had the highest four values, or if the 12 values showed very
little percentage difference, the pattern was considered "flat" or
without a definite maximum. Consequently, there were five po-
tential pattern classifications.
A pattern with a maximum in the eastern half was consid-
ered to be a pattern representative of urban influences on precip-
itation. The other four patterns were considered to represent
natural variations in precipitation. Therefore, the frequency of
periods with east-half maxima was examined statistically in rela-
tion to the frequency of other patterns. The temporal distribution
of precipitation occurrence, the surface winds, and the types of
synoptic situations which occurred with these 262 rain periods
also were studied, compared, and evaluated.
Frequency Distribution. The annual and seasonal frequen-
cies of the periods associated with the five different patterns are
shown in Table 4. If all patterns were normally distributed, each
of the percentage values in Table 4 would be 20; however, exam-
ination of the percentages for the cases with east-half maximum
values indicated that in each season and for the year this pattern
easily exceeded the normal expected frequency of occurrence. In
the winter and fall seasons the percentages are almost twice those
expected, and the spring and annual percentages are more than
50 percent greater than expected. Only in summer is the per-
centage near the normal expected value. Without elaborate sta-
tistical testing, which is not in order with this sample size, it
appears that the high percentage of cases with east-area max-
ima in the fall and winter achieve particular significance because
no other comparable high percentages occurred in these two sea-
sons. The high east area percentage in spring is exceeded by the
flat pattern percentages, and this factor tends to lessen the sig-
nificance of the high spring value in the east.
From these data it could be concluded that urban influences
affect one precipitation period in 10, causing a maximization in
the eastern half of the urban area. Normally, 2 out of 10 precip-
itation periods are expected to maximize there, but on an annual
basis the 1958-1959 data indicate that 3 of 10 periods maximized
in that urban locality. In all four seasons more precipitation per-
iods with an east-half maximum occurred than expected. In both
the fall and winter season, nearly 4 out of every 10 precipitation
periods maximized in the eastern half of the urban area.
SYMPOSIUM: AIR OVER CITIES
-------
50
CHANGNON
TABLE 4
„„ FREQUENCIES Ol
,__ WITH DIFFERENT TYPES
OF"ISOHYETAL PATTERNS
Area of Maximum Precipitation Values
' " " None Total
West North South (flat) periods
Winter
Number of periods
Percent of total
Spring
Number of periods
Percent of total
Summer
Number of periods
Percent of total
Fall
Number of periods
Percent of total
Annual
Number of periods
Percent of total
East
18
39
28
34
18
24
22
37
86
33
5
11
14
17
18
24
12
20
49
19
1
2
9
12
4
7
19
7
17
15
20
7
12
36
14
14
31
29
36
15
20
14
24
72
27
46
82
75
59
262
Hourly Occurrence of Precipitation. The time of occur-
rence of precipitation with these various precipitation patterns
also was investigated. A count of the number of times each hour
had precipitation was made for (1) the precipitation periods with
maxima in the east half and (2) all the remaining precipitation
periods. The results of this analysis are shown in Table 5. If it
is assumed the data derived from the other four patterns repre-
sent a near normal hourly distribution of precipitation, any sig-
nificant deviations from these distributions exhibited by data from
periods maximizing in the east would be of interest.
In the winter the east-half maximization periods produced
precipitation most frequently in the hours from 1400 to 2200 CST,
whereas for all the other periods precipitation was most frequent
SEC TECHNICAL REPORT A62-5
-------
URBAN PRECIPITATION PATTERNS 51
from 0400 to 1200. In the spring the periods maximizing in the
east area were most frequent from 1200 to 2000 CST, whereas
the other patterns maximized from 1800 to 0200 CST. In the
summer, the precipitation periods with east-half maximization
were most frequent at hours nearly the same as those with the
TABLE 5
DISTRIBUTION OF HOURS WITH PRECIPITATION
ASSOCIATED WITH PRECIPITATION PERIODS
MAXIMIZING IN THE EAST
AND WITH ALL OTHER PRECIPITATION PERIODS
Percent of total hours
Winter Spring Summer Fall Annual
2-hr period East Other East Other East Other East Other East Other
(CST) max. patterns max. patterns max. patterns max. patterns max. patterns
2400-0200
0200-0400
0400-0600
0600-0800
0800-1000
1000-1200
1200-1400
1400-1600
1600-1800
1800-2000
2000-2200
2200-2400
4 6
5 8
5 9
7 11
8 11
9 9
10 9
11 8
10 8
12 7
12 7
7 7
3
4
5
6
8
11
12
14
13
12
a
4
10
7
7
8
6
7
6
8
9
10
10
12
5
7
5
10
11
13
10
11
8
7
6
6
4
6
6
7
8
9
14
12
8
12
8
6
6
5
7
8
9
10
10
9
9
10
9
8
8
7
7
11
12
8
8
8
8
7
8
8
5
5
5
7
9
10
11
11
10
11
9
7
7
77
8
9
9
8
9
9
8
9
8
9
other periods of precipitation, although the other patterns indicated
a secondary maximization of hourly frequencies from 1800 to
2000 CST. In the fall the east-half maximum periods produced
precipitation most frequently between 1000 and 1400 CST, com-
pared with 0600 to 1000 CST for the other patterns. This ten-
dency for a greater frequency later in the day with the east-half
maximization periods also was characteristic in winter. On an
annual basis, the precipitation periods in which a maximization
occurred on the east half of the urban area produced precipitation
more frequently during the afternoon and early evening hours,
whereas the other types of patterns produced precipitation more
uniformly distributed in time. The possible causes for these
peculiar time distributions associated with precipitation periods
SYMPOSIUM: AIR OVER CITIES
-------
52 CHANGNON
maximizing in the east will be discussed later.
The average precipitation periods with an east-half max-
imization were 12. 1 hours in winter, 6. 5 hours in spring, 3. 4
hours in summer, and 10. 1 hours in fall. The durations for fall
and spring were slightly greater than those based on all the other
patterns.
Surface Wind Conditions. It is assumed that two of the
urban influences on precipitation are additional condensation
nuclei and water vapor. For these factors to influence the mod-
erately small Champaign-Urbana area most actively, they would
have to be brought into the precipitation systems west of the
urban area, since most precipitation systems move from the
west. This would give these two factors more time to act on the
precipitation systems. To produce an east-area maximization,
the low-level winds prior to the start of precipitation should be
moving across the urban area in a direction which would allow
these influences to be swept downwind into the clouds to be
affected. ^ Because of the relatively small size of the urban
area under consideration, however, a northeast, east, or south-
east low-level wind direction would appear to be most favorable
to obtain an upwind movement of atmospheric nuclei and water
vapor for earlier entrainment into the precipitation-producing
system entering the urban area from the west. Winds at and
prior to precipitation are considered in this evaluation rather
than wind conditions during precipitation, because precipitation
tends to remove nuclei from the air thereby lessening this urban
influence. 1 •*• Previous radar studies have shown that most pre-
cipitation cells in this area move from the west-southwest, west,
and west-northwest. To support this supposition on air move-
ment, the precipitation periods with maximization in the east
should be preceded by a greater frequency of winds from the
northeast, east, and southeast than are all the other four types
of precipitation patterns. Data presented in Table 6 do not sub-
stantiate this assumption,although surface winds are not always
TABLE 6
FREQUENCY OF SURFACE WIND DIRECTIONS PRIOR TO
AND AT THE BEGINNING OF PRECIPITATION PERIODS
Direction frequency, %
Total
Precipitation pattern periods N NE E SE S SW W NW
Maximum in east 86 41
All other patterns 176 14
.19 5 12 37 15 7
8 17 12 16 19 95
SEC TECHNICAL REPORT A62-5
-------
URBAN PRECIPITATION PATTERNS 53
indicative of the low-level wind field. 25 percent of the east-half
maximum periods had surface winds moving from the three
easterly azimuths across the city, whereas of all other periods
37 percent had surface winds from these three directions.
These findings derived from the surface wind data could be
interpreted as refuting the theory that urban effects helped to pro-
duce more frequent maximization of precipitation in the east,
because favorable surface wind directions were not frequent. At
least, the surface wind conditions with east-side precipitation
maxima do not substantiate the theoretical considerations.
Synoptic Conditions with Precipitation Periods. The fre-
quency of various synoptic situations associated with the 262
periods is shown in Table 7. Cold and warm front classifications
include pre-frontal activity related to their presence. Relating
the cause of precipitation to various synoptic types is fraught with
problems of interpretation, but it is believed that the frequencies
shown in Table 7 are in the right order of magnitude and that it is
safe to use such data to draw limited conclusions. As expected,
TABLE 7
ANNUAL AND SEASONAL FREQUENCIES
OF SYNOPTIC TYPES ASSOCIATED WITH
THE PRECIPITATION PERIODS IN 1958-1959
Surface
Season
Winter
Spring
Summer
Fall
Annual
Warm
Front
2
11
3
7
23
Cold
Front
11
20
29
21
81
Stationary
Front
5
15
23
11
54
Occluded
Front
10
10
0
4
24
Low or
Trough
8
13
9
8
38
Upper
Trough
4
2
3
5
14
Air
Mass
6
11
8
3
28
Total
Periods
46
82
75
59
262
cold fronts were the most frequent synoptic condition that pro-
duced precipitation, and stationary fronts were the second most
frequent type. Surface lows, air mass instability, occluded
fronts, and warm fronts had almost equal frequencies.
In Table 8 the numbers of different synoptic types associated
with the five different precipitation patterns are expressed as per-
centages of the pattern total. Only with the air mass and station-
ary front types is there a semblance of comparable percentages
among the five patterns. If the difference between the number of N
west and east maximizations (Table 4) is assumed to represent
SYMPOSIUM: AIR OVER CITIES
-------
East
maximum
7
35
20
12
13
1
12
West
maximum
10
33
30
0
15
2
10
South
maximum
0
36
22
0
11
17
14
North
maximum
16
48
21
0
5
5
5
flat
13
17
14
19
21
6
10
54 CHANGNON
TABLE 8
SYNOPTIC TYPES ASSOCIATED WITH
FIVE URBAN PRECIPITATION PATTERNS
Percent of each pattern total
East
Synoptic type
Warm front
Cold front
Stationary front
Occluded front
Surface low-trough
Upper trough
Air mass
urban influences, then the significant differences between east
and west synoptic percentages shown in Table 8 could indicate the
synoptic condition most favorable for maximizations in the east.
The only difference favoring the east is associated with occluded
fronts. Comparison of all the percentages in Table 8 reveals
that the flat pattern percentage for occluded fronts exceeded that
same percentage with east maximizations; however, it could be
concluded that the only significant synoptic condition relating to
maximization in the eastern urban area is the occluded frontal
type.
For a closer inspection of the synoptic conditions associated
with eastern urban area maximizations, the seasonal frequencies
are portrayed in Table 9, and these frequencies are also ex-
pressed as percentages of the seasonal totals in Table 7. It has
been stated previously in this section that apparently: (1) the
seasons during which the east-half storms occurred most fre-
quently were winter, fall, and spring; and (2) the synoptic type
exhibiting the most significant difference in percentages, associ-
ated with east-half patterns, was the occluded front. These
statements are augmented by the data in Table 9. In winter and
spring 40 percent of the occluded front periods maximized in the
east, and in the fall, 50 percent.
Summary. The theory that urban effects increase precip-
itation appears to be substantiated by many of the results derived
from the investigation of 262 precipitation periods. Almost twice
as many periods as expected had a maximum of precipitation in
the eastern half of the area. Furthermore, the seasons when
these events occurred most frequently were winter and fall,
closely followed by spring. Therefore," the cases that supported
SEC TECHNICAL REPORT A62-5
-------
Season
Winter
Number
Percent of total
seasonal cases
Spring
Number
Percent of total
seasonal cases
Summer
Number
Percent of total
seasonal cases
Fall
Number
Percent of total
seasonal cases
Warm
front
1
50
3
27
0
0
2
29
Cold
front
5
46
8
40
8
27
9
43
Stationary
front
3
60
3
25
7
30
4
36
Occluded
front
4
40
4
40
0
0
2
50
URBAN PRECIPITATION PATTERNS 55
TABLE 9
SEASONAL NUMBER AND PERCENT OF SYNOPTIC TYPES
ASSOCIATED WITH PRECIPITATION MAXIMUMS
IN THE EASTERN HALF OF THE URBAN AREA
Surface Upper Air
low-trough trough mass
401
50 0 16
316
23 50 56
102
11 0 25
401
50 0 33
the hypothesis of urban-induced precipitation occurred largely in
the colder half-year, and this agrees with previous findings. 1
The marked tendency for precipitation with an east-half
maximization to occur later in the day than the other precipitation
patterns during the winter and fall, and more frequently during
the period from midday to late afternoon than the other precipita-
tion patterns suggests the possibility that a diurnal heating influ-
ence might be acting to produce maximization of precipitation in
the east. Surface winds prior to the beginning of precipitation
periods maximizing in the eastern area were not frequently from
easterly azimuths, however, a condition believed favorable for
bringing the urban nuclei and water vapor influences into the cloud
and precipitation systems. The only synoptic type associated
with precipitation periods maximizing in the east more frequently
than elsewhere was the occluded front.
EVALUATION OF CAUSES FOR URBAN PRECIPITATION
MAXIMIZATION.
Considerable data have been presented concerning the daily,
seasonal, and annual distribution of precipitation during all or
parts of the 1950-1959 decade. It has been repeatedly suggested
that urban influences could be largely responsible for the max-
imization of precipitation in east side urban areas. Champaign-
Urbana urban-rural differences in total precipitation, days of
rain, and days with thunderstorms were statistically similar to
SYMPOSIUM: AIR OVER CITIES
-------
56 CHANGNON
results found elsewhere, and attributed to urban influences on at-
mospheric conditions. Based on this comparison it could be
superficially concluded that the findings of the 10 years of precip-
itation data collected in Champaign-Urbana were a result of urban
meteorological influences. Knowledge of the spatial precipitation
variability encountered in this climate through natural influences
and from varying raingage exposures, plus the lack of definite
measured proof of the action of urban influences, however, leaves
the discerning observer unsure of the causes for the observed
variations in precipitation. Therefore, a discussion of each of
three possible causes for the variation follows.
Raingage Exposures. Evaluating precipitation catch has
long been a problem in meteorology and climatologyl2 because of
differing raingage exposures. Also the effects of various methods
of shielding raingages have been measured in many different ex-
periments with varying results. Weiss and Wilson 13 reported
that amounts of precipitation measured when various shielding
techniques are used vary from 87 percent to as high as 340 per-
cent of unshielded catches. From the values of the 17 experiments
they listed, the median value selected indicated that shielded rain-
gages catch 9 percent more precipitation than unshielded gages.
This percentage difference is comparable with the annual differ-
ence of 12 percent found for the 10-year period in Champaign-
Urbana. That is, if the gages in the eastern urban area gener-
ally were more shielded than those in the west, a west-to-east
increase comparable to that experienced in 1950-1959 might re-
sult. To derive such a conclusion it must be assumed that ex-
posure differences caused by obstacles in an urban area might be
comparable with those differences between shielded and unshield-
ed gages. None of the 12 gages were shielded. All the raingages
were assumed to have similar gage wind eddies, but it is possible
that site and environment wind eddies, because of differences in
gage exposures, varied sufficiently in magnitude to create wind
eddy differences which approximated the differences in gage
eddies obtained between shielded and unshielded raingages.
In further support of the possibility that gage exposures are
responsible for the Champaign-Urbana differences, it should be
noted that shielding effects are much more pronounced for snow-
fall than for rainfall. i3 Since the average urban snowfall in the
10-year period was 11. 8 inches, or about 1. 2 inches of water,
extreme exposure differences between the rural gages and the
central urban gages might produce a water measurement differ-
ence of about 0. 5 inch. This difference is based on a median per-
centage determined from 18 field experiments reported by Weiss
and Wilson; these experiments showed that the snow catch of a
shielded gage is 160 percent of the snow catch of an unshielded
SEC TECHNICAL REPORT A62-5
-------
URBAN PRECIPITATION PATTERNS 57
gage. As noted in Table 2, the average urban-rural precipitation
difference in winter was 1. 0 inch; shielding variations therefore
could account for 50 percent of the observed difference.
Since exposure variability can be shown to be a possible
cause for observed precipitation difference, a method was devel-
oped to obtain a quantitative measure of the exposures of the 12
urban gages for comparison purposes. No wind instruments were
available to perform this.
The method initially consisted of measuring with a transit
the amount of horizon blocking afforded by surrounding obstacles
which extended above the horizon line of the raingage tops. Since
the amount of wind passing a gage orifice is inversely related to
the gage catch, the amount of blocking or shielding by surround-
ing objects should be directly related to the catch. The exposure
in blocking was computed as the number of degrees of vertical
blocking in each of the eight 45-degree sectors in azimuth, with
each sector centered on the eight basic compass directions.
Since no object at any gage had a vertical angle greater than 30
degrees, the total possible blocking in any sector was 45 degrees
times 30 degrees, or 1350 degrees, and the amount of blocking in
each sector at each gage was expressed as a percent of this total.
Secondly, these sector blocking percentages were normal-
ized to wind directions by multiplying each by the corresponding
annual wind direction frequency during the occurrence of precip-
itation. Therefore, excessive gage blocking to the north from
where precipitation winds are infrequent would not achieve as
much statistical significance as equal sector blocking to the south
from where precipitation winds are frequent. Unfortunately this
calculation does not include wind speed, but it does provide a
relative measure of the effect of site and environment shielding
around the gages.
The results of this method for evaluating exposures of the
12 gages are shown in Table 10. A comparison of these results
with those in Table 1 indicates that, in general, there is very
little agreement for any gage between the rank of the precipitation
catch and the shielding rank. A high shielding factor should re-
late directly with a greater precipitation catch. If the rank val-
ues based on precipitation amounts in Table 1 are added, the
easternmost six urban gages have a score of 25, compared with
a score of 53 for the westernmost six gages. A similar summa-
tion of ranks based on the shielding factors in Table 10 provides
a total score of 41 for the easternmost gages and 37 for the west-
ernmost gages. This indicates no material difference between the
overall exposures of the gages in the western and eastern halves
of the urban area.
SYMPOSIUM: AIR OVER CITIES
-------
58 CHANGNON
TABLE 10
ENVIRONMENTAL SHIELDING FACTORS
FOR 12 URBAN RAINGAGES
Total
shielding
Gage
1
2
3
1
5
6
7
8
9
10
11
N
8. 4
3.0
1. 7
1. 3
1. 3
0
1.0
1. 3
0. 3
1.0
3. 3
NE
2. 6
1. 1
4. 6
1.8
1. 3
0. 7
1. 3
0.2
3. 7
3. 1
1.8
E
3.2
0. 3
3. 7
1. 1
1. 3
1. 8
1. 3
0
2. 4
0. 8
2. 6
SE
2. 9
0.2
1. 5
0. 9
0. 9
3.2
0. 3
0
0. 7
1. 7
1. 7
S
13.8
2. 9
7. 8
2.8
2. 5
9. 1
0. 6
0
2. 3
3. 9
4. 7
SW
4. 4
0. 7
0.8
1. 3
2. 1
0.2
0. 4
3.0
0.8
0.8
1.6
1
1
2
3
0
1
0
1
0
2
W
. 5
. 0
. 4
0
. 5
.2
.9
. 6
. 0
. 4
. 1
NW
4.
2.
1.
1.
1.
1.
0.
1.
3.
0
8
1
0
3
0
5
2
7
9
.9
value
40.
12.
23.
9.
14.
15.
8.
6.
11.
13.
21.
8
0
6
2
2
2
3
3
9
6
7
Rank based
on
shielding total
1
7
2
9
5
4
10
11
8
6
3
0 0. 4 1. 4 0 2.0 12
A further corroboration was obtained by comparing summa-
tions of the shielding factors for the four highest-ranked gages and
the four lowest. The four highest-ranked raingages, numbers 9,
10, 11, and 12, had a combined shielding factor of 49. 2, whereas
the four lowest-ranked raingages, numbers 3, 4, 5, and 7, had a
combined factor of 55. 3.
The results of the investigation of gage exposure variability
as a possible cause or explanation for the distribution of precip-
itation in 1950-1959 have indicated that, except for a few un-
measurable variations, differences in exposures in the urban
area probably were not responsible for the observed variations in
precipitation. Any shielding differences that did affect the dis-
tribution of precipitation were probably most active during snow-
fall in the winter season. Quantitatively, this effect on snowfall
catch could have been 0. 5 inch of water, which accounts for only
12 percent of the 4. 0-inch annual difference in precipitation for
the eastern urban area and that for the western urban and rural
areas.
A raingage located in an urban area, even with a very un-
sheltered exposure site, must be considered to have more en-
vironmental shielding than most rural raingages, and therefore,
the urban raingage will generally catch more precipitation. In
larger urban areas it is conceivable that the increased turbulence
and friction effects might cause an overall reduction in the speed
of storm movement and increase the time required for passage
over the city, thereby allowing more precipitation to fall over
the urban area.
SEC TECHNICAL REPORT A62-5
-------
URBAN PRECIPITATION PATTERNS 59
Urban Effects. The second possible cause for the max-
imization of precipitation in the eastern half of the urban area is
the influence on precipitation systems from urban effects. These
effects are considered to be additional condensation nuclei, addi-
tional water vapor, and increased turbulence from surface rough-
ness and thermal heating. In previous descriptions of the data
associated with precipitation quantity, seasonal distribution, num-
ber of rainy days, number of thunderstorm days, and synoptic
conditions, it was indicated that much of these data could be in-
terpreted as indicative of influences from urban effects.
Additional condensation nuclei are presumed to be largely
derived from combustion processes. Champaign-Urbana has
very few industries and no heavy industries. The five principal
point sources of smoke are shown on Figure 1, and none, except
for the heating plant near gage 9 and the University heating plant
near gage 1, produces any large volume of smoke. Furthermore,
the heating plant operations are restricted largely to the colder
seasons. The only other potential sources are home heating
plants, but only during the cold season. Telford 1" recently
stated that industries, other than heavy industries such as steel
plants, probably have little influence on the freezing-nuclei count
in the atmosphere. Therefore, except for some possible minor
influence in the winter season, it is believed that the Champaign-
Urbana urban area does not produce sufficient condensation nuclei
to have an effect on precipitation systems.
Much the same conclusion can be drawn concerning water
vapor production by the Champaign-Urbana urban area. Without
heavy industries no great volume of water vapor is released to
the atmosphere. The only other possibility for additional water
vapor might be evapotranspiration, from urban trees and lawns,
which could be greater than that from corn and soybean crops in
rural areas; however, data published by Blaney, '•^ concerning
consumptive water use of plants, which is indicative of the rate
of evapotranspiration, indicate that corn and soybeans have a
greater combined rate of evapotranspiration than grass and decid-
uous trees. This indicates that the foliage in the urban area would
release less water vapor through evapotranspiration than foliage
in rural areas.
It thus appears that air turbulence produced by the Cham-
paign-Urbana urban area could be the only major effect which
could influence precipitation. Unfortunately, no measurements
of the roughness effect on turbulence are available, but this effect
may be considerable, especially when the flat, featureless sur-
rounding rural area is taken into consideration. Thermal heating
over the cities also could produce turbulence by lifting the air.
SYMPOSIUM: AIR OVER CITIES
-------
60 CHANGNON
In a recent article DeMarrais 15 has shown that air over an urban
area because of local temperature conditions, is relatively less
stable throughout the day than air over rural areas.
A comparison of the mean monthly minimum and maximum
temperatures, based on 6 years of comparative data, has shown
that the maximum values at the rural Airport station (Figure 1)
range from 0. 5°, in the summer-fall period, to 1. 3° in the
winter-spring period, lower than those at the urban Central sta-
tion. 16 The rural mean monthly minimum temperatures vary
from 0. 8°, in summer, to 2. 0°, in winter, lower than the urban
values. This proves that on the average the urban area is warmer
than the rural area.
To further examine the urban-rural temperature difference
in time and space, two additional weather stations were installed
on the fringes of the urban area (Figure 1) for a 12-month period
beginning in July 1959. The 2-hourly mean seasonal values from
each of the stations were compared with those from the Central
urban station, and the results are given in Table 11. Space does
not permit a thorough discussion of all the many interesting fea-
tures displayed in Table 11; however, a few pertinent observa-
tions require mention. In all four seasons there are times of the
day when the rural (Airport) and outlying urban stations (NW and
SE) have temperatures exceeding those at the urban Central sta-
tion. The time of day when these excesses occur varies from
late in the afternoon in winter and fall to mid-morning in spring
and summer. The more rapid afternoon cooling at the Central
urban station in winter and fall suggests the occurrence of greatei
radiational heat loss over the urban area.
For each season data in Table 5 were used to select the 8-
hour periods of maximum precipitation associated with the pre-
cipitation periods which maximized in the eastern half of the
cities, and these periods have been indicated in Table 11. Com-
parison of these 8-hour periods with the mean temperatures of
the Central urban station in each season reveals that the period,
except in summer, occurred during or just after the hours of
maximum heating. Further, comparison of these 8-hour periods
with the temperature departures of the rural and outlying urban
stations reveals that the 8-hour period in each season always
began at hours when the temperature departures began to exceed
the Central station mean values. From these relationships it
would appear reasonable to conclude that the urban area, because
of higher temperatures and greater radiation at the ground sur-
face than in the rural areas, may have been producing more
thermal lifting of air and a generally more unstable lower atmos-
phere, and consequently was affecting precipitation systems.
DeMarrais 15 has shownj howeverj that urb&n &nd ^^ ^
SEC TECHNICAL REPORT A62-5
-------
URBAN PRECIPITATION PATTERNS 61
COMPARISON OF TEMPERATURES BETWEEN
CENTRAL AND FRINGE STATIONS
July 1959-June 1960
00 02
Winter
Central
mean 28.8 28.2
Station departures
SE -0.3 -0.1
NW -0.5 -0.3
AP -1.- -1.0
8-hour period of most
frequent precipitation
Spring
Central
mean 42.8 41.4
Station departures
SE
NW -1.1 -1.0
AP -2.6 -2.5
8-hour period of most
frequent precipitation
04
27. 8
-0. 3
-0.4
-1.6
40. 1
-0. 7
-2. 5
06 08
27.6 28.
-0.4 -1.
-0.4 -1.
-1.1 -2.
39. 3 43.
-0.9 -0.
-1.3 -0.
7
2
3
0
2
1
2
Time,
10
31. 5
-1.2
-1. 1
-0.6
47. 9
+0. 4
+0. 5
X
CST
12 14
33.8 34.3
-0.6 -0.1
-1.0 +0.3
-0.4 +0.2
X
50.8 52.6
+0. 4 +0. 2
+0.2 -0.3
1000-1800
16 18
33.1 31.2
+0.4 +0.1
+0. 7 +0
+0.4 -0. 6
1400-2200
52.8 51.0
+0 -0. 7
-0.8 -1.7
X
20
30.2
-0. 2
-0. 4
-1.0
47. 8
-1. 6
-2. 7
22 Average
28.9 30.3
-0.2 -0.3
-0.1 -0.4
-1.4 -0.7
X
45.1 46.2
-1.2 -0.3
-2.4 -1.3
Summer
Central
mean 68.2 66.5 65.3 65.9 70.8 76.4 79.5 81.1 80.9 78.4 74.1 70.3 73.1
Station departures
SE -0.7 -0.7 -1.2 -0.8 +1.8 +1.3 +1.3 +2.2 +1.4 +0.9 -1.1 -0.8 +0.3
NW -0.2 -0.5 -0.5 -1.5 -0.6 -0.3 +0.6 +0.7 +0.8 +1.0 -0.1 -0.2 -0.1
AP -1.7 -1.5 -1.9 -0.4 +2.1 +1.3 +1.0 +0.1 -0.4 -0.3 -2.4 -1.9 -0.3
8-hour period of most X 0800-1600 X
frequent precipitation
Fall
Central
mean 47.1 46.3 45.6 46.8 50.6 55.3 58.3 58.7 56.4 52.4 49.7 48.0 51.3
Station departures
SE +0.4 +0.1 -0.2 -1.9 -1.2 -0.9 +0 +1.0 +2.1 +1.3 +0.3 +0.1 +0.3
NW +0.7 +0.2 +0 -2.1 -2.4 -2.0 -0.8 +1.1 +2.4 +2.4 +1.0 +0.8 +0.3
AP -0.5 -0.7 -1.1 -2.9 -1.8 -0.8 +0 +1.0 +2.1 +0.7 -0.2 -0.5 -0.4
8-hour period of most X " 1000-1800 X
frequent precipitation
temperatures several hundred feet above the surface are not too
dissimilar in daylight hours, the period when precipitation max-
imization in the east was most frequent. DeMarrais also indicated
that the urban influence on air temperature above a city was
greatest at night, tending to break down the inversion which re-
mains over the rural areas. Thus, his findings tend to refute the
claim that thermal effects were a major urban influence on
precipitation.
Natural Variability of Precipitation The third possible
cause for the precipitation distribution in the Champaign-Urbana
area in the 1950-1959 period could be natural spatial variations
in precipitation. In this area nearly 40 percent of the average
annual precipitation is derived from thunderstorms, and the ex-
extreme variability of storm and monthly precipitation in the area
is well known. ^ Unfortunately, similar statistical measures of
SYMPOSIUM: AIR OVER CITIES
-------
62
CHANGNON
the spatial variability of seasonal and annual precipitation, es-
pecially over short distances, are not yet available. Therefore,
it is difficult to evaluate statistically the 1950-1959 Champaign-
Urbana seasonal and annual patterns in regard to significant dif-
ferences which may have arisen from natural rainfall variability.
During this same 10-year period the Illinois State Water
Survey operated another raingage network in a completely rural
area 55 miles northwest of Champaign-Urbana. ^ This network,
the Panther Creek Network, was comprised of nine recording
raingages located in a 100-square-mile area, as shown in
Figure 6.
31.5,
\
\
N
A '
AREAS EQUAL TO SIZE ^ )
OF CHAMPAIGN-URBANA K^
URBAN AREA
\ •
31.5 32
35
SCALE
MILES
NETWORK
BOUNDARY
• RAINGAGES
Figure 6. Annual average precipitation on Panther Creek raingage network, 1950-59
SEC TECHNICAL REPORT A62-5
-------
URBAN PRECIPITATION PATTERNS 63
In the results here reported for this network, data from two
Weather Bureau raingages located on the periphery of this net-
work were incorporated. The precipitation data from this net-
work are considered to be representative of natural variations in
precipitation, and the network is near enough to Champaign-Urbana
to experience comparable climatological conditions. Therefore,
data from this rural network were compared with that from the
urban area, even though the gage densities differed, because the
rural network data were the only available expression of natural
variability of precipitation for this 10-year period and this area.
The average annual pattern of precipitation in the rural net-
work is shown in Figure 6. Superimposed on this pattern are two
rectangles, each the size of the Champaign-Urbana urban area.
These rectangles have been placed in two portions of the rural
pattern where variations were comparable with those exhibited in
the urban network. This indicated that the annual urban pattern
(Figure 2) could be entirely a result of natural causes.
The rural network data for each year and each gage were
ranked from high to low, as were the urban annual data shown in
Table 1. In the rural network, 20 of the 30 possible highest three
ranks were achieved by the five easternmost gages. This is com-
parable with the value of 25 of 30 three highest ranks obtained in
the urban area by the six easternmost gages. If the six north-
easternmost gages in the rural network are chosen as a group for
a similar comparison, we find that these gages had 24 of the 30
highest three ranks. These findings also suggest that the urban
precipitation distribution was of a magnitude exhibited by natural
variation. In 8 of the 10 years the annual totals of the three east-
ernmost gages in the rural network ranked first. In the urban
area the highest annual rank occurred at one of the six eastern-
most gages in 9 of the 10 years.
The seasonal patterns in the rural network did not show the
consistent maximization in the eastern area in each season that
was seen in the urban network. In the rural network definite sea-
sonal maximums occurred in the eastern area only in the fall and
spring. In the winter a north-south latitudinal increase was ap-
parent; in the summer the precipitation pattern was more ambig-
uous, and no definite regional maximization occurred. These
findings, especially for the winter season, reflect the fact that
the rural network was experiencing true natural variations in pre-
cipitation. In this climatic region, the distribution of precipita-
tion in winter is much less variable than in the other seasons and
also is related closely to change in latitude.
Another measure of rural or natural precipitation variability
was made using the data from the individual precipitation periods
SYMPOSIUM: AIR OVER CITIES
-------
64 CHANGNON
in
... the 1958-1959 period. As shown in Table 4, 33 percent of all
1958-1959 precipitation periods in the urban network maximized
in the eastern half of the urban area. To obtain comparable val-
ues from the larger rural network, it was decided to determine
how many of the 245 precipitation periods on the rural network
maximized at any one of the three easternmost gages. In the 2-
year period, 122 precipitation periods maximized at these three
gages. This frequency, expressed as a percent of the total per-
iods, was almost 50 percent, compared with the 33 percent fre-
quency in the urban area. This comparison may be questionable
because of differences in area, but it does indicate that a frequent
concentration of storm precipitation can occur naturally in a very
small portion of a given area. Time did not permit further de-
tailed investigation of the precipitation data from the rural net-
work, but the results presented here appear to prove that most of
the urban measurements of precipitation could have been entirely
a result of natural variations rather than of any urban influence.
SUMMARY AND SUGGESTIONS
In reporting on a study of precipitation in a moderate-sized
urban area with very little industry and no pollution problem,
attempts have been made to illustrate some of the important prob-
lems associated with the overall question of whether urban effects
influence the spatial distribution of precipitation. It might be
argued that some of the problems raised by the Champaign-Urbana
site would not be applicable in larger cities. Similar data from a
larger urban area probably would not provide more enlightening
answers, however, because urban-rural percentage differences in
precipitation conditions measured in Champaign-Urbana were
comparable to those measured at other much larger urban areas.
As stated previously, the present general lack of knowledge
of the microscale atmospheric processes does not permit the
evaluation and measurement of urban effects on precipitation
processes with the degree of accuracy desired. In general, these
effects have been claimed to produce increases in precipitation on
the order of 5 to 15 percent, but meteorologists cannot presently
measure or describe the atmospheric processes well enough to
achieve a high degree of accuracy in theoretical calculations.
The only other hope for defining the relation between urban
conditions and precipitation is through an intensive measurement
program to be used with generalized inferences. The results of
the Champaign-Urbana study have shown that the precipitation
maximization in Urbana could easily be claimed to result either
from urban effects or from natural variations of precipitation.
Raingage exposures presented a problem in the evaluation of
effects; and admittedly, not all of the instruments and data needed
SEC TECHNICAL REPORT A62-5
-------
URBAN PRECIPITATION PATTERNS 65
to make such decisions were available. To instrument an urban
area for the purpose of evaluating urban influences on precipita-
tion would be a difficult and tremendously expensive project, and
investigators would want to be confident that results from such an
undertaking would provide reliable answers. Another and probably
more serious problem to overcome, even with a thoroughly in-
strumented project, is that of natural variation of precipitation,
especially in climatic regions where variability is known to be
excessive because of the predominance of shower-type rainfall.
Much more statistical information is needed concerning the vari-
ability of seasonal and annual precipitation in small densely-gaged
networks. Until these data are available one cannot provide a
statistical description of the natural spatial variability of precip-
itation; and until this description is available, it will be difficult
to evaluate the natural variability of precipitation in a study of
the relationship between urban influences and precipitation. Many
large urban areas in the United States are near to physical fea-
tures, such as large water bodies, hills, and mountains, which
make urban-precipitation studies even more difficult.
Whether a given city produces, or can produce, a 5, 10, or
15 percent increase in rainfall and number of rainy days in por-
tions of its urban area is a question that neither meteorology nor
climatology can at present answer accurately. Gross or general-
ized urban effects may be claimed, but to obtain reliable evidence
of their existence, we must increase our knowledge of atmospher-
ic processes and undertake extensive measurement programs.
REFERENCES
1. Landsberg, H.E., "The Climate of Towns", Man's Role in
Changing the Face of the Earth. University of Chicago
Press, Chicago, Illinois, 1956, pp. 584-603.
2. Chow, V.T., Hydrologic Studies of Urban Watersheds. Hy-
draulic Engineering Series No. 2, University of Illinois,
Urbana, 1955.
3. Huff, F. A. , and S. A. Changnon, Jr., Severe Rainstorms in
Illinois 1958-1959. Report of Investigation 42, Illinois
State Water Survey, Urbana, 1961.
4. Changnon, S. A. , andF.A. Huff, Studies of Radar-Depected
Precipitation Lines.. Scientific Report No. 2, Contract No.
AF 19(604)-4940, Urbana, Illinois, 1961.
5. Hiser, H. W. , and S. G. Bigler, Wind Data from Radar Echoes.
Tech. Report No. 1, Navy BuAer Contract No. N189s-88164,
Urbana, Illinois, 1953.
SYMPOSIUM: AIR OVER CITIES
-------
66 CHANGNON
6. Byers, H. R. , and others, The Thunderstorm. U.S. Govern-
ment Printing Office, Washington, D. C. , 1949.
7. Hudson, H. E. , and W. J. Roberts, 1952-55 Illinois Drought
with Special Reference to Impounding Reservoir Design.
Bulletin 43, Illinois State Water Survey, Urbana, 1955.
8. Huff, F.A. , andS.A. Changnon, Jr., "Distribution of Exces-
sive Rainfall Amounts Over an Urban Area", Journal of
Geophysical Research, 65(11): 3759-3765, November, 1960.
9. Changnon, S. A. , Jr. , Thunderstorm-Precipitation Relations
in Illinois. Report of Investigation 34, Illinois State Water
Survey, Urbana, Illinois, 1957.
10. Telford, J.W. , "Freezing Nuclei from Industrial Processes",
Journal of Meteorology, 17(6):676-679, December, 1960.
pp. 676-679.
11. Dickson, R. R. , "Meteorological Factors Affecting Particulate
Air Pollution of A City", Bulletin of AMS, 42(8):556-560,
August, 1961.
12. Kurtyka, J. C. , Precipitation Measurements Study. Report of
Investigation 20, Illinois State Water Survey, Urbana, 1953.
13. Weiss, L. L. , and W. T. Wilson, "Precipitation Gage Shields",
Comptes Rendus et Rapports. Toronto, 1958, pp. 462-484.
14. Blaney, H. F. , "Irrigation Requirements of Crops", Agricul-
tural Engineering, 32( 12):665-668, December, 195T!
15. DeMarrais, G. A. , "Vertical Temperature Difference Ob-
served Over An Urban Area", Bulletin of AMS, 42(8):548-
554, August, 1961.
16. Changnon, S. A. , Jr. , Summary of Weather Conditions at
Champaign-Urbana, Illinois. Bulletin 47, Illinois State
Water Survey, Urbana, 1959.
17. Huff, F.A. , and J. C. Neill, Rainfall Relations on Small Areas
in Illinois. Bulletin 44, Illinois State Water Survey, Urbana,
1957.
DISCUSSION
CHAIRMAN LANDSBERG: This was certainly very gracious
of Mr. Stout to present Mr. Changnon's paper. If Mr. Stout is
willing to stand in for his colleague, we have a minute or two for
questions.
DR. SCHMIDT: One thing occurs to me in regards to this.
Due consideration has been shown to the individual exposure of
the stations, but I wonder if there isn't some possibility that the
urban area, the built-up area itself, creates to a general de-
crease in the low-level wind flow in that region. Would it be pos-
sible that the lighter winds in the eastern region might result in
a larger apparent catch of precipitation? The rain gages are so
SEC TECHNICAL REPORT A62-5
-------
URBAN PRECIPITATION PATTERNS 67
sensitive to wind speeds that if the mean winds in that region were
generally lower, it is a possible thing that I think you could look
into fairly easily. Because if it is so, then the catch would be
most different during higher winds and less different during low
winds.
MR. STOUT: This is true. We are all aware of this, but
as yet we have no data with which to correct for this apparent
defect. But I think you are right that the general wind movement
is much less in the built-up areas than in the rural areas, or at
the outer fringes of town.
FROM THE FLOOR: Is your Panther Creek network totally
free of urban influence?
MR. STOUT: Yes.
FROM THE FLOOR: So the variations that you found here
which are equal in magnitude to the urban area could not in any
way be charged to wind speed?
MR. STOUT: The topmost gage was in a town of about 750
or a thousand people, and there were no industries there. But
over to the right there that was all rural areas, all open, country
and no problem as far as exposures.
DR. HEWSON: Are there any other orographic effects other
than the Panther Creek?
MR. STOUT: No, the change in elevation at Panther Creek
is a maximum of about 100 feet in that area, so it is a relatively
flat land.
SYMPOSIUM: AIR OVER CITIES
-------
Some Effects of Air Pollution on
Visibility In and Near Cities*
GEORGE C. HOLZWORTH, V. $. Weather
Bureau, Los Angeles, California
Summary
Regularly obtained visibility observations are examined with regard to effects due
to nearby air pollution sources. Some inherent difficulties in dealing with visibility data
are pointed out. The effects on visibility of wind speed and of wind direction relative to
pollution sources are described. A method is presented for analyzing possible trends in
visibility.
One of the more profound effects of air pollution is on visi-
bility. To some extent such effects are experienced by most mem-
bers of a community and therefore are worthy of consideration in
air pollution appraisals. To obtain a record of visibility observa-
tions suitable for study, one turns almost invariably to the nearest
observing office of the Weather Bureau. Typically, such offices
are at the airport, often near the outskirts of the city. Observa-
tions at the edge of a city offer an opportunity to assess the effects
of the urban area as a general pollution source. As shown in Fig-
ure 1, the urban area of Sacramento, California, acts as a source
of pollutants that reduce the visibility at the municipal airport,
located 4. 5 miles south of downtown Sacramento. Relative to the
airport, most of the Sacramento urban area lies in the sector
northwest clockwise through east-southeast. As the drawing in-
dicates, the higher percentages of poor visibilities, 0-10 miles,
are associated with winds from the urban directions.
The data in Figure 1 are based on hourly observations dur-
ing May, July, September, and November for 18 years. Only day-
time observations have been considered, since visibility during
darkness depends on different factors and is considered less re-
liable. In an attempt to eliminate visibility reduction due to natur-
al causes, the observations considered were restricted to periods
in which wind speeds were 1 to 10 miles per hour (mph), no pre-
cipitation was occurring, and relative humidity was less than 90
percent. Even with these restrictions, there remain 13, 726 ob-
servations. For small samples such visibility wind roses are
often less regular than that shown in Figure 1.
In this analysis the implied assumption is that the wind
direction at the time of the visibility observation is roughly
*This work is supported by the U. S. Public Health Service.
69
-------
70
HOLZWORTH
n7lP '*l > *"^»T 'V^O^—^J-
-;^.. iSlfl^SlS^
Figure 1. Sacramento metropolitan area with wind rose of the percent frequency for each
wind direction for visibilities of 0-10 miles. Wind rose centered on observation
site, municipal airport. Data based on daytime hourly observations during May,
July, September, and November for 18 years. Observations with precipitation
and/or relative humidity greater than 90 percent, or with wind ipvedi other than
1-10 mph omitted.
SEC TECHNICAL REPORT A62-5
-------
EFFECTS OF POLLUTION ON VISIBILITY 71
representative of the trajectory of the air since it passed over
significant pollutant sources. This assumption appears valid at
Sacramento but may not apply in areas of small-scale eddies or
where there are marked diurnal variations in air flow.
The diurnal variation in visibility at Chicago's Midway
Airport from January to March 1930 was discussed by Week.
He observed that the main contribution to reduced visibility was
smoke. An average daily minimum visibility of 3. 0 miles at
8:00 a. m. was attributed to the smoke released after the man of
the house arose and stoked the furnace. As the day advanced,
the smoke became lighter (probably in part because of increased
atmospheric mixing) and the visibility improved to 4. 5 miles
near 3:00 p. m. In the late afternoon firing at residences and
banking of factory furnaces reduced the visibility to a secondary
minimum of 3. 5 miles near 5:00 to 6:00 p. m. , after which it
slowly rose again to an average daily maximum of 7. 0 miles at
4:00 a. m.
At Columbus, Ohio, daily visibility observations made at
the center of the city at 8:00 a, m. for 11 years have been com-
pared to corresponding Weather Bureau observations at the air-
port, about 6 miles to the east-northeast. The airport is well
beyond the city limits and is considered an out-of-town location.
Figure 2 shows that the average annual visibility at the city lo-
cation is consistently about 1 mile lower than that at the airport.
No observations were included if precipitation was occurring. It
is interesting to note in these data a trend of improving visibility
through 1958 at both sites. This trend may be related to the av-
erage annual dustfall rates in Columbus, shown in Figure 2. The
steady decline in these rates is consistent with the trend of in-
creasing visibility, and both trends reflect the effective program
of the Division of Smoke Regulation and Inspection of the Columbus
Department of Public Safety.
The term "visibility" is used in this paper in its meteoro-
logical sense, which is more properly "visual range, " or the
greatest distance at which prominent objects can just be seen.
This use should be distinguished from visibility in terms of the
clarity with which objects can be seen, although the two uses
are not unrelated.
An important use of visibility data is as an indicator of air
pollution trends, as described for Columbus. To determine the
visibility trend for the United States in general the percent fre-
quencies of all hourly visibilities less than 7 miles (due to all
causes) have been compared at a number of locations for two per-
iods separated by about 15 to 20 years. The data are presented
in Table 1.
SYMPOSIUM: AIR OVER CITIES
-------
72 HOLZWORTH
TABLE 1
PER CENT FREQUENCIES OF ALL VISIBILITIES
LESS THAN 7 MILES AT
WEATHER BUREAU AIRPORT STATIONS &
JAN FEE MAR APR MAY JUN JUL AUG SEPT OCT NOV DEC
BAKERSFIELD 1/29-12/38
CALIFORNIA 7/55-6/59
BURBANK 10/31-12/38
CALIFORNIA 5/50-4/55
CARIBOU 1/38-12/41
MAINE 7/59-3/61
CHICAGO 1/30-12/38
ILLINOIS 1/59-12/60
COLUMBIA 6/30-12/38
MISSOURI 7/59-3/61
COLUMBUS 1/30-12/38
OHIO 7/58-6/61
DBS MOINES 4/33-12/38
IOWA 7/58-6/61
EL PASO 7/30-12/38
TEXAS 1/49-12/53
GRAND ISLAND 9/31-12/38
NEBRASKA 7/58-6/61
GREENSBORO 1/30-12/38
N. CAROLINA 7/58-6/61
INDIANAPOLIS 7/29-12/38
INDIANA 7/58-6/61
LAKE CHARLES 3/39-4/42
LOUISIANA 7/59-3/61
MEDFORD 7/29-12/38
OREGON 8/50-7/55
MILWAUKEE 1/30-12/38
WISCONSIN 7/58-6/61
MOLINE 1/30-12/38
ILLINOIS 7/58-6/61
NASHVILLE 7/37-12/41
TENNESSEE 7/58-6/61
OAKLAND 7/29-12/38
CALIFORNIA 5/50-4/55
23 9
47 29
[Taj 24
J25J 14
36 30
21 30
72 70
57 37
31 34
21 28
63 57
46 37
54 46
23 30
[I] 5
111 4
25 23
15 23
31 24
23 24
61 56
35 40
11 12
20 24
35 12
40 23
50 45
29 30
63 55
31 34
45 41
26 27
26 19
28 26
PEORIA 3/36-12/38 68 56
ILLINOIS 7/58-6/61 36 36
RICHMOND 6/29-12/38 39 33
VIRGINIA 7/58-6/61 22 24
3
6
24
21
24
28
65
48
29
24
45
29
41
28
9
4
20
20
23
19
50
33
17
16
5
4
40
0
5
1
2
29
32
19
21
62
32
22
4
28
19
29
11
10
4
23
10
18
11
38
14
11
9
PI
14
28 17
32
37
15
13
54
34
12
2
19
16
21
9
4
2
13
10
10
12
26
14
6
3
2
2
32
18
51 40 25
34 14
32 11
17 9
9
8
4
9
40 29
33 12
27 22
20 13
11
10
7
2
8
20
11
19
18
1
3
48
42
16
14
48-
30
7
6
16
20
14
9
2
1
m
lil
6
17
18 1
18 )
3
4
1
0
27
2
2
3 7
5 9
I
11 29
15 40 58
43 42 43
45 51 51
18 16 19
10 13 10
47 52 52
32 38 33
m
lil
8 13
4 2
16 23 23
26 32 24
~il 20 20
io|
1
1
3
3
8 9
0 2
0 1
8 10
2 5
10 11 16
22 28 22
Til 31 32
22| 30 19
356
8 10 9
! [
2] 9
3 5
22 30 32
11 11 22 12
21 [Til 37 35
14 |_20J 19 14
8 13 16 12
10 15 17 17
3
5
7 11 25
1 10 18
16 17 28 30
11 16 20 12
18 20 22 26
22 26 41 27
38 22 21
47 34 26
20 31 39
14 23 18
58
40
18
14
35
25
26
12
1
1
10
3
15
24
38
21
12
15
56 73
29 39
23 36
5 15
53 68
23 41
32 52
11 17
1 3
1 3
15 20
6 9
26 34
14 13
46 65
18 31
11 19
15 20
13
29
38
19
45
21
23
19
28
27
37
20
26
23
23 37
49 46
36 43
18 21
51 66
12 26
43 44
16 25
41 30
37 30
41 62
17 26
34 38
18 18
SEC TECHNICAL REPORT A62-5
-------
EFFECTS OF POLLUTION ON VISIBILITY
TABLE 1 (Continued)
73
SACRAMENTO
CALIFORNIA
SALEM
OREGON
SAN DIEGO
CALIFORNIA
SEATTLE
WASHINGTON
SIOUX CITY
IOWA
SOUTH BEND
INDIANA
ST. LOUIS
MISSOURI
TULSA
OKLAHOMA
WINSLOW
ARIZONA
1/30-12/38
3/50-2/55
1/34-12/38
7/55-6/59
1/30-12/38
8/49-7/54
1/30-12/38
7/55-6/59
1/47-12/51
7/58-6/61
6/30-12/38
7/58-6/61
1/35-12/41
1/58-12/60
3/30-12/36
7/58-6/61
2/31-12/38
7/55-6/59
JAN FEE MAR APR MAY JUN JUL AUG SEPT OCT NOV DEC
42 [22J
39 30
4
9
2 1 1
324
2
4
13
8
16
19
37
44
55
46
35 29
25 19
90 85 86
79 78 88
50 45 39
34 18 28
26 24 17
14 19 12
0 0
0 0
51
24
87 81 77
75 67 71
30 44 53
13 28 34
37
24
33
20
46
27
86
73
46
38
45
27
40
13
37
35
88
82
36
28
11 17
9 11
67
41
52
22
14 22
4 12
0 0 0 Fol 4
0 00 2 [lj 0
Observations at hourly intervals.
Data Sources: U. S. Weather Bureau Local Climatological Data, Climatography of the U. S.
No. 30-(for selected stations), and Normal Plying Weather for the U. S. (New Orleans, 1945).
The stations listed in the table were selected because the obser-
vation sites did not change significantly during the time between
the two periods compared. Where the frequency of visibilities
less than 7 miles is greater in the later period (lower row) than
in the earlier period (upper row), the figures are boxed in. Low
visibilities are more frequent in the later period in less than 26
percent of the comparisons. Further, these increases in poor
visibility are for the most part rather small: of the 336 compar-
isons only 13 are greater than 10 percent, 4 are greater than 20
percent, and none are greater than 30 percent. On the other
hand, many of the decreases in frequency of low visibilities are
large: 108 are greater than 10 percent, 35 are greater than 20
percent, and 10 are greater than 30 percent. These changes are
all the more significant in view of a revised method of reporting
visibility. As pointed out by Robinson 2* visibility was reported
before January 1, 1939, as the greatest visibility over the hori-
zon. Since then it has been reported as the lowest visibility over
the half of the horizon (not necessarily continuous) with the
*This reference includes an excellent discussion of the the-
ory of visibility and many of the practical considerations to be
made in utilizing visibility data for air pollution purposes.
SYMPOSIUM: AIR OVER CITIES
-------
K
tq
O
H
M
O
ffi
2
1 — I
O
o
z>
H
AVERAGE 8:00 A.M. VISIBILITY
MILES
123456
COLUMBUS, OHIO
30
AVERAGE DUSTFALL
TONS PER SQUARE MILE PER MONTH
25 20 15 10 5
1 [_
1
•• DOWNTOWN ISsai^ AIRPORT IN CITY
DATA SOURCE: COLUMBUS DIVISION OF SMOKE REGULATION AND INSPECTION 14
Figure 2. Average 8:00 A.M. visibility at downtown Columbus, Ohio, and at the municipal airport,
6 miles east-northeast; and average dustfall rates in Columbus.
ffi
o
r
O
SJ
H
ffi
-------
EFFECTS OF POLLUTION ON VISIBILITY 75
greatest visibility.
Although the data of Table 1 are meager, the differences
seem much too large to attribute to chance alone. It appears
that visibility in many urban areas of the United States is im-
proving. Such a trend is perhaps surprising, but there is some
evidence to indicate that it is real. Prior to control of the qual-
ity of solid fuel in St. Louis (beginning about 1941) and in Pitts-
burgh (beginning about 1945) pollution in winter, due largely to
smoke from coal combustion, was often dense enough to require
the use of street lamps and automobile headlights at midday. It
has been estimated that in St. Louis smoke was reduced 75 per-
cent and in Pittsburgh, 70 percent. 3 Visibility data for St. Louis,
included in Table 1, show a marked reduction in poor visibilities,
especially in the colder months.
Some idea of the contaminants produced by the combustion
of coal, oil, and natural gas may be obtained from Table 2. The
weight of particulate material produced per BTU decreases by
about an order of magnitude for each conversion, from coal to
oil and from oil to gas. Coal and oil combustion produce about
the same quantity of gaseous material, but natural gas produces
only one-fifth that amount.
TABLE 2
COMPARISON OF CONTAMINANTS FROM
COMBUSTION OF COAL, OIL AND
AND NATURAL GAS (INDUSTRIAL BOILERS) a
(lb/107 Btu)
Coal Oil Gas
Particulates 14.1 1.93 0.14
Gases 26.8 26.62 5.24
Total 40.9 28.55 5.38
a After Neiburger (Ref. 15)
A specific example of the effect of fuel conversion on vis-
ibility improvement has been described by Bloodworth. At the
municipal airport at Atlanta, Georgia, a railroad switching yard
2 miles to the northwest was listed as a principal smoke source
(for reduced visibilities). The total number of regular hourly
observations when smoke alone reduced the visibility to less than
SYMPOSIUM: AIR OVER CITIES
-------
76
HOLZWORTH
1 mile was tabulated for the period from 1944 to 1951 Table 3
h™wthat as the locomotives in the yard were converted from
steam (coal) to diesel, the annual number of poor visibilities due
to smoke alone was greatly reduced. This effect was most pro-
nounced during the period from 1947 to 1949. Thereafter the
continued conversion of locomotives to diesel fuel had little ap-
parent effect. Bloodworth noted in addition that smoke reduction
at the airport was also augmented by a general fuel change in the
area from coal to natural gas.
TABLE 3
TOTAL NUMBER OF VISIBILITY OBSERVATIONS
IN GIVEN VISIBILITY CATEGORIES DUE TO SMOKE ALONE
AT ATLANTA MUNICIPAL AIRPORT, GEORGIA a
Percent Conversion of Steam
Visibility Less Than (Coal) to Diesel Locomotives
1 mile 3 miles
17
20
17
15
5
1
0
2
191
182
135
139
113
17
6
22
1944
1945
1946
1947
1948
1949
1950
1951
aAfter Bloodworth (Ref. 4)
in Nearby Yards
Started
20
60
90
5
Nearby railroad yards have been listed as a common pri-
mary source of smoke and reduced visibility at many airports
The American Petroleum Institute 6 reports that before World
War II less than 2 percent of the locomotives in use on the major
railroads burned diesel fuel. In 1958 more than 90 percent of
such locomotives were diesel fueled.
Besides the changing of fuels there have been improve-
ments in combustion techniques and equipment and in the methods
of capturing pollutants before they are emitted to the atmosphere,
The general public is certainly becoming more aware of air pol-
lution, with the result that increasing efforts are being made to
reduce emissions to the atmosphere.
SEC TECHNICAL REPORT A62-5
-------
EFFECTS OF POLLUTION ON VISIBILITY 77
It is not intended here to settle the matter of whether vis-
ibility in the United States has been generally improving or de-
teriorating. Several significant matters have not even been men-
tioned. But the information presented here suggests that the
emission of visibility-reducing materials to the atmosphere well
may have declined.
For specific locations the evaluation of visibility data with
reference to air pollution is not often simple. In such studies it
is desirable to eliminate those cases of visibility reduction due
to natural causes. For precipitation this is straightforward.
For fog it is much less so because many man-made pollutants
are hygroscopic and may act as condensation nuclei in the for-
mation of fog. Further the meteorological conditions favorable
to fog formation are also conducive to poor dispersion of air
pollutants. The usual procedure for eliminating cases of poor
visibility due to natural causes is to select some relative humid-
ity above which the visibility observations are neglected. In trend
studies it would be helpful if the dispersive characteristics of the
atmosphere and their effects on production of pollutants would not
vary from year to year. But they do vary, and there are seldom
sufficient data to allow for such effects.
In addition to the troublesome influences of the weather
there are other difficulties in dealing with visibility data. Since
visibility is defined as the greatest distance at which objects can
just be seen, there must be objects present. In some localities
the number and distribution of objects, or markers, may be un-
satisfactory. Also, the erection of new markers may change the
spectrum of reported visibilities. Visibility observations are
subjective estimates made by trained and experienced observers.
These estimates are based on the ability to "see" and it is known
that such capacities are far from constant, either among different
people or in the same person at different times.
Although not all of the factors influencing visibility have
been touched on, it should be apparent that there are good reasons
for visibilities to fluctuate markedly from year to year (and over
even shorter spans). In trend studies such variations are often
observed, with the result that the trend is difficult to evaluate,
especially over short periods. It is usually reasonable to as-
sume that such fluctuations are random, however.
One of the more extensive studies of the visibility trend
relative to air pollution was carried out by Neiburger 7 for down-
town Los Angeles for the years 1933 through 1954. From obser-
vations at 5:00 a. m., noon, and 5:00 p. m. a great many tabula-
tions of visibility by wind direction, relative humidity, time of
SYMPOSIUM: AIR OVER CITIES
-------
78 HOLZWORTH
day season, and year were prepared. It was shown that the
frequency of very good visibilities decreased markedly as the
population and industry of the area grew. After about 1947 to
1948, when the control of pollutants was initiated, visibility
seems to have improved in spite of a continued increase of pop-
ulation and industry. Neiburger concluded that these trends were
not due to differences in the weather.
Using a similar type of analysis, Robinson, Curie, and
James ^ considered noon visibilities at downtown San Francisco
for the period 1933 to 1956. They noted that for various visibi-
lity ranges the annual frequencies varied from year to year.
After restricting the observations to cases of light wind and low
relative humidity, they concluded that it was difficult to observe
any long-term visibility trend related to air pollution.
Holzworth and Maga ^ have suggested a method for analyz-
ing the trend of visibility. This method, which will be illustra-
ted here, has been applied to data for Sacramento and Bakers-
field, California, with the conclusion that the visibility has
deteriorated at both cities, supporting the belief that air pollution
in California's Central Valley is increasing. The method of
Holzworth and Maga was also applied to Neiburger's Los Angeles
data, which were extended 5 years through 1959. Before controls
were initiated, a trend of rapidly declining visibility was indi-
cated for Los Angeles. After controls there was little trend in the
visibility. From a control standpoint this latter is significant in
view of a rapid rise in population and the attendant increase in
automobile emissions.
The method of trend analysis suggested by Holzworth and
Maga will now be illustrated by application to data for the muni-
cipal airport at Salt Lake City, Utah. Only hourly daytime ob-
servations (about 8:00 a.m. to 5:00 p.m.) during periods with no
precipitation and with relative humidities less than 90 percent
are considered. The total range of visibility is broken down into
a number of sub-ranges, and the percent frequencies of occur-
rence in each of these ranges is tabulated for each year. The
advantage of considering the visibility by ranges rather than using
median or mean values is that the changes that occur can be de-
tected more precisely. In the left portion of Figure 3, the per-
cent frequencies of visibilities in the indicated ranges are plotted
for October of each year from 1946 to 1960. There are some
rather large variations from year to year, but as a whole the
linear regression lines, fitted by the method of least squares,
depict the general trend in each range fairly well. In this par-
ticular case there is a clear trend of improving visibility. The
SEC TECHNICAL REPORT A62-5
-------
EFFECTS OF POLLUTION ON VISIBILITY
79
frequencies in the two higher visibility ranges are both increas-
ing, while the frequencies in the three lower ranges are all
decreasing.
PER CENT FREQUENCIES OF
VISIBILITIES IN GIVEN RANGES
BY YEARS WITH LINEAR TREND LINE.
64 j
56--
FLUX OF VISIBILITY
FREQUENCY CHANGES
FREQUENCY
-- 1946
1960
-- 34.6
44.2
NET
CHANGE
46 60
+9.6
FLUX OF RESULT-
ANT CHANGES
-9.6
46 48 SO 52 54 56 58 60
-- 28.9 55.8 +26.9 + -36.5
-9.6
10.6 0.8 -9.8 + -26.7
-36.5
-- 14.4 0.7 -13.7 + -13.0
-26.7
-- 11.5
-1.5
-13.0 +
-13.0
46
Figure 3. Percent frequencies of visibilities in given ranges by years (left), and schemat-
ic shift of visibility frequency changes (right) at Salt Lake Municipal Airport
in October. Linear regression lines fitted by method of least squares. Data
based on daytime hourly observations; observations with precipitation and/or
relative humidity greater than 90 per cent omitted.
SYMPOSIUM: AIR OVER CITIES
-------
80 HOLZWORTH
But the matter may be pursued further in order to include
some numbers in the trend, and this will also be helpful in.the
evaluation of more complicated trends. For any one year the
total frequency of observed visibilities in all ranges is 100 per-
cent; this is true also for the frequencies determined from the
linear regression lines. Therefore, the initial and terminal
points of the regression lines may be used to obtain the net per-
centage frequency changes over the span of years considered, as
shown in the right portion of Figure 3. For instance, from the
regression line the frequency of visibilities in the range greater
than 30 miles is 34. 6 percent in 1946 and in 1960 it is 44. 2 per-
cent. The net change is plus 9. 6 percent, and so on for the other
ranges. The algebraic sum of the changes in all ranges is zero,
and the increases are compensated for by decreases elsewhere.
But now it is important to consider how such compensations
are likely to occur. A negative net change represents a surplus
of visibilities in that particular range, and a positive change, a
deficit. The problem is to determine how the surpluses and de-
ficits in the various visibility ranges interact.
The matter of why the net frequency changes in visibility
ranges should interact may be considered from another point of
view. If other things are the same, when the concentration of
visibility reducing materials changes, a similar effect may be
expected on all visibilities that would have occurred otherwise.
Let us take the very simplified case of three visibility ranges,
low, intermediate, and high, each with frequency of occurrence
of 33. 3 percent at some initial time. At some later time if the
general concentration of visibility reducing materials is less we
may expect, for instance, that 10. 0 percent of the frequencies
that would otherwise have been poor are now intermediate, 10
percent of the intermediate are high, and 10 percent of the high
are just higher. The result is that at the later time the resulting
frequencies are low: 23. 3 percent, intermediate: 33. 3 percent,
and high 43. 3 percent. It would appear that the only visibilities
affected were those in the upper and lower ranges. But actually
there has been a flux of visibilities upward to higher ranges.
Consider a case of the visibility improving, say, from 3
to 25 miles. Such changes are not discontinuous, but the visi-
bility passes continuously through intermediate valves. Cor-
respondingly, with a trend of improving visibility, as in Figure
3, it is expected that the surpluses (net frequency decreases) in
the lower visibility ranges are shifting upward to each next higher
range. In each case the amount shifted to the adjacent range is
algebraically added to the net change there and the resultant
SEC TECHNICAL REPORT A62-5
-------
EFFECTS OF POLLUTION ON VISIBILITY 81
value is shifted to the next range. These resultant shifts are in-
dicated by the arrows in the right portion of Figure 3. For in-
stance, in the lowest range, 0 to 6 miles, the net change from
1946 to 1960 is minus 13. 0 percent. This surplus is shifted up
to the next higher range, 7 to 12 miles, where it is added to the
net change there, minus 13. 7 percent. The resultant is minus
26. 7 percent which is shifted up to the next range, 13 to 19 miles,
where the net change is minus 9. 8 percent. The resultant change
in the 13-to-19 mile range is minus 36. 5 percent and this is shift-
ed to the 20~to-30 mile range, where the net change is plus 26. 9
percent. The resultant change in the 20-to-30 mile range is minus
9. 6 percent, which, when shifted to the highest range, greater
than 30 miles, just balances the net change there of plus 9. 6
percent. In Figure 3 it is shown that in all cases the resultant
shifts (of surplus frequencies) are all upward to higher visibility
ranges. The total resultant shift upward is 85. 8 percent, and it
is clear that there is a marked trend of improving daytime visi-
bility at Salt Lake City in October.
The visibility trend at Salt Lake City was also studied for
December; the analysis is shown in Figure 4. Here the major
visibility changes are in the two upper ranges; the best visibili-
ties are declining in frequency and the second-best are increasing.
As shown on the right side of Figure 4, the total resultant shift
downward to lower visibility ranges is 18. 0 percent, the total up-
ward to higher ranges is 2. 5 percent, and the sum is 15. 5 percent
downward. Therefore, there is a slight trend of deteriorating
visibility in December.
In view of the improving trend in October this is surprising.
But a deeper look into the matter will indicate that such opposing
trends are not without some basis. In 1941 Salt Lake City passed
an air pollution control ordinance, aimed primarily at reducing
smoke emissions. It has been stated that emission of particulate
matter within Salt Lake City limits has been reduced substantial-
ly as a result of enforecement activities. ® Natural gas was
brought to the area about 1941, and steady conversion from coal
to gas was responsible for much of the reduction in particulate
emissions. It is significant that Salt Lake City is the only muni-
cipality in the state having an air pollution control ordinance. As
shown in Figure 5, the city is located near the center of a large
intermountain valley. This valley is developing rapidly. It is
estimated that from 1940 to 1958 the population grew from
347, 000 to 632,000. 10
In October, when there is a trend of improving visibility,
space heating is necessary, especially at night. The mean daily
maximum temperature for October is 66. 5 F and the minimum
SYMPOSIUM: AIR OVER CITIES
-------
82
HOLZWORTH
PER CENT FREQUENCIES OF
VISIBILITIES IN GIVEN RANGES
BY YEARS WITH LINEAR TREND LINE
FLUX OF VISIBILITY
FREQUENCY CHANGES
NET
FREQUENCY CHANGE FLUX OF RESULT-
1960 46 60
6.9 -16.5 +
!_ 22.9 37.9 +15.0 + -16.5
".I
84
ni
A
13-19 MILES
%f 6-8
i L
8.6
+1.8 +
-1.5
-0.3
it —
'14-
1(H
8-
n -
• 7-12 MILES A
A A '
: X/^ "--*'"* V
I...J_I — 1 — 1 — 1 — 1 — 1 — 1 — 1 — 1 — 1 — 1 — 1_
- 13.9 15.8
+ 1.9 + -2.2
60-
52-
44--
36-
28-
20-'
12-
4--
l ! I I i i i | i I I I I I
46 48 50 52 54 56 58 60
YEARS
33.0
30.8
-2.2 +
Figure 4. Percent frequencies of visibilities in given ranges by years (left), and schemat-
ic shift of visibility frequency changes (right) at Salt Lake Municipal Airport
in December. Linear regression lines fitted by method of least squares. Data
based on daytime hourly observations; observations with precipitation and/or
relative humidity greater than 90 per cent omitted.
SEC TECHNICAL REPORT A62-5
-------
EFFECTS OF POLLUTION ON VISIBILITY
83
Figure 5. Salt Lake Valley, Utah. (Elevation contours in thousands of feet).
is 39. 2 F. The dispersive capacity of the atmosphere is much
less at night than during the day; low-level temperature inver-
sions are frequent at night, and it is estimated that the mean
daily maximum mixing depth is 1000 meters. In the forenoon
wind directions are primarily from the city area, south-south-
east, and in the afternoon from the lake, northwest. With such
a set of conditions a pattern of high pollution concentration in the
morning with much lower values during the afternoon is reason-
able.
SYMPOSIUM: AIR OVER CITIES
-------
84
HOLZWORTH
Therefore, pollutants that reach the airport in the morning, be-
fore the atmosphere is well-stirred, will be most effective in
reducing the visibility. Figure 6 shows the percent frequencies
of visibilities of 0 to 12 miles for each wind direction (for wind
speed of 1 to 10 mph in October. Poor visibilities are by far
most frequent with winds from the direction of Salt Lake City,
from the sector east through south. It is clear that Salt Lake
City is a direct source of poor visibilities at the airport during
October. Recall that in Figure 3 the trend in frequency of visi-
bilities in the ranges 0 to 6 and 7 to 12 miles was downward. The
fact that the visibility trend in October is upward reflects the
effectiveness of the control program in Salt Lake City.
30
Figure 6. Wind rose of percent frequency for each wind direction for visibilities of 0-12
miles at Salt Lake Municipal Airport in October. Data are the same as in
Figure 3 except that only wind speeds of 1-10 mph are considered.
The visibility wind rose for December, shown in Figure 7,
is based on the same type of analysis as that in Figure 6. Al-
though visibilities of 0 to 12 miles are most frequent with winds
from the northwest, there is no particular sector for which the
frequencies are especially greater than for others. Therefore,
low visibilities are about as likely from one sector as another.
It is concluded, then, that concentrated pollutants are trapped
in the valley for several days at a time and simply recirculated,
Schmalz 3 nas explained such occurrences in association with
stagnating anticyclones over Utah, especially in winter. He noted
that after about a week of stagnation, smoke (reported by pilots)
SEC TECHNICAL REPORT A62-5
-------
EFFECTS OF POLLUTION ON VISIBILITY
85
extended from Salt Lake City to the Nevada State line, 110 miles
away. Schmalz also noted the relationship between low visibili-
ties and wind direction, shown in Figure 6.
Figure 7. Wind rose of percent frequency for each wind direction for visibilities of 0-12
miles at Salt Lake Municipal Airport in December. Data are the same as in
Figure 3 except that only wind speeds of 1-10 mph are considered.
In December the mean maximum mixing depth is estimated
to be 500 meters, just half of that in October. Furthermore, the
lowest average wind speed of the year, 7. 4 mph, 3-1 occurs in
December. With such a low dispersive capacity atmospheric
pollutants may be transported throughout the Salt Lake Valley
and remain in a relatively concentrated state throughout the day.
Therefore, pollution from various sources throughout the valley,
where there are no control ordinances, is effective in reducing
the daytime visibility at Salt Lake Airport in December. At the
airport the effect of the control measures is expected to be most
SYMPOSIUM: AIR OVER CITIES
-------
86 HOLZWORTH
apparent in the forenoon, whereas the effect of uncontrolled
emissions is expected to be most apparent m the atternoon.
Consequently, the visibility improvement effected by control
measures within Salt Lake City is overwhelmed by the uncon-
trolled emissions from developing areas throughout the valley.
And in December there is a trend of declining visibility in the
daytime.
Visibility data can be an important source of information
in the evaluation of air pollution problems. In some cases the
conclusions reached are straightforward, in more complex
situations, however, additional information and more detailed
analyses are required.
REFERENCES
1. Week, F. H. , 1930: Average Visibility at Chicago airport.
Monthly Weather Review, vol. 58, p. 204.
2. Robinson, E., 1962: Effects of air pollution on visibility,
Air Pollution, A Comprehensive Treatise, Edited by
A. C. Stern, Academic Press, New York.
3. Faith, W. L. , 1959: Air Pollution Control, John Wiley,
New York, p. 4.
4. Bloodworth, S. H., 1953: The decreasing importance of
smoke in reducing visibilities at Atlanta, Georgia,
Bulletin of the American Meteorological Society, vol 34.,
p. 78.
5. U. S. Weather Bureau: Terminal reference forecasting man-
ual, available for various stations in the U. S., Govern-
ment Printing Office, Washington, D. C.
6. American Petroleum Institute, 1959: Petroleum Facts and
Figures, American Petroleum Institute, p. 369.
7. Neiburger, M. , 1955: Visibility Trend in Los Angeles, Air
Pollution Foundation, Rpt. No. 11, Los Angeles, 45 pp,
8. Robinson, E. , H. Currie, and H. A. James, 1960: Aspects
of San Francisco visibility climatology, presented at 187th
national meeting Amer. meteor. Soc. , Eugene, Oregon,
June 14-16.
9. Holzworth, G. C. and J. A. Maga, 1960: A method for
analyzing the trend in visibility, Air Pollution Control
Assn. Journ. , vol. 10, pp. 430-436.
10. U.S. Public Health Service, 1959: A Review of the Air
Pollution Situation in The Salt Lake Valley, Utah, R. A.
Taft Sanitary Engineering Center, Cincinnati, Ohio, 32 pp,
SEC TECHNICAL REPORT A62-5
-------
EFFECTS OF POLLUTION ON VISIBILITY 87
11. U. S. Weather Bureau, 1959: Local Climatological Data
with Comparative Data, Salt Lake City, Utah, issued an-
nually, Government Printing Office, Washington, D. C.
12. Hosier, C. R. , 1961: Low-level inversion frequency in
the contiguous United States, Monthly Weather Review,
vol. 89, pp. 319-339.
13. Schmalz, W. M., 1947: Some notes on visibilities at Salt
Lake City Airport, Bulletin American Meteorological
Society, vol. 28, ppT 179-186.
14. Division of Smoke Regulation and Inspection, Dept. of
Public Safety, City of Columbus, Ohio: 1960 Annual Report.
15. Neiburger, M. , 1959: Meteorological aspects of oxidation
type air pollution, The Rossby Memorial Volume, Rocke-
feller Institute Press in association with Oxford University
Press, New York, pp. 158-169.
DISCUSSION
PROFESSOR J. O. LEDBETTER: How were the regres-
sion lines fitted?
MR. HOLZWORTH: By the method of least squares.
If you have reason to suspect that there is some other trend in-
volved, you can compute other lines more detailed. We have
done this in a few cases but it doesn't lead to any improvement.
MR. LICHTBLAU: Did you draw the conclusions on the
deterioration of visibility at Lake Charles, as indicated from
your diagram? Lake Charles was outstanding.
MR. HOLZWORTH: Yes, the frequencies of poor
visibilities have increased in recent years at Lake Charles. It
was brought to my attention only recently, in a compilation made
here at the Taft Center. It is interesting to consider which of
the cities that were listed have any type of control, and most of
them do. Those that don't have any control are Caribou, Maine;
Lake Charles, Louisiana; Pocatello, Idaho; and until very re-
cently Oakland, California. So Lake Charles is one of the cities
where there are no controls. What conclusions you draw from
this, I am not sure. It would probably depend on having more
data.
DR. SCHMIDT: It may be of interest to you that we studied
the visibility from our light vessels that are situated 10 to 20 or
perhaps 30 miles out from the coast. Comparing visibility be-
fore and after the war, we found a deteoriation of the visibility.
SYMPOSIUM: AIR OVER CITIES
-------
cn HOLZWORTH
DO
DR HILST- What is the-implication of before and after
the war, 'increased industrialization after the war?
DR. SCHMIDT: Yes, sir.
MR WOHLERS: I have a question. I am not a meteorolo-
gist. You have given quite a bit of data ranging from 1930 up to
1950 some odd, at various stations. How many men may have
been on duty at those stations?
MR. HOLZWORTH: This is one of the matters that I hoped
to bring out in listing the difficulties of using visibility data --
some of the factors that go into this year-to-year variation that
you see. The only way I think that the data can be used is to
assume that such variations are random. Obviously a great
number of different people have taken part in those observations,
SEC TECHNICAL REPORT A62-5
-------
Smoke Concentrations in Montreal Related
to Local Meteorological Factors
PETER W. SUMMERS, Weather Engineering
Corporation of Canada Ltd. and McGill
University, Montreal
Summary
Smoke concentrations in central Montreal have been measured by Weather Engineer-
ing Corporation since January 1960. The seasonal variations are related to the changes
in atmospheric stability. Early in 1961 measurements were started by McGill University
on the campus and on top of Mount Royal. Data from the three locations are compared
and discussed in terms of the vertical distribution of smoke and local topographical ef-
fects. The advantage of a "smokeless zone" is clearly demonstrated. Some synoptic
situations producing high smoke concentrations are illustrated. Application of these re-
sults to similar locations in eastern Canada and the northern United States with a long
winter snow cover is discussed.
The first analysis of air pollution in Montreal in relation
to local meteorological variables was presented one year ago at
the Kingston meeting of the Royal Meteorological Society,
Canadian Branch. Although based on a 4-month sample of smoke
concentration measurements from only one location, in the down-
town area, this study presented several interesting results and
some tentative conclusions.
Since that time, measurements have continued and the
seasonal and diurnal variations can now be analyzed in more de-
tail. Measurements were started at the MacDonald Physics
Building on the McGill University Campus at the end of January
1961 and on top of Mount Royal at the Canadian Broadcasting
Corporation transmitter building in the middle of March 1961.
The measurements from these two sampling locations can now be
compared to determine the effects of the Mount Royal "smokeless
zone". Also the measurements made at the top of Mount Royal
at a height of 600 feet above the city can be used to evaluate the
effect of inversions on the vertical transport of pollution. Some
preliminary results of these studies will be presented.
SAMPLING EQUIPMENT
The equipment used for these studies is the Model E auto-
matic spot sampler developed for the American Iron and Steel
Institute by Hemeon ^ and manufactured by the Research Appli-
ance Company of Pittsburgh.
89
-------
90 SUMMERS
In sampling with this equipment, a known volume of outside
air is drawn through a given area of filter paper and the amount
of dust and smoke particles deposited in a given time is evaluated
by measuring the optical density of the soiled spot with a photo-
meter. The method is described in full detail elsewhere. 2
UNITS
The optical density is calculated as follows:
Optical density (O. D. ) log IP
where:
I0 the intensity of transmitted light through the clean
paper
I the intensity of transmitted light through the soiled
paper
For comparison with other samples the optical density is
converted to a unit scale called the COH unit, defined as follows:
1 COH unit 100 x O. D.
This value is still dependent on the volume of air drawn
through the filter paper. If L is the quantity of air sampled ex-
pressed in thousands of linear feet then:
L flow (cu. ft. per min. ) x sampling time (min.)
1000 x area of spot (sq. ft. )
The standard unit used to express smoke concentration is:
COH units per 1000 linear feet O. D. x 100
L
INTERPRETATION OF COH UNITS
The New Jersey State Department of Health3 carried out an
extensive statewide survey of air pollution by means of smoke
samplers in 1956. On the basis of informed opinion they assigned
the following adjectival ratings to various levels of smoke con-
centration:
Smoke Concentration Adjectival Rating
(COH units /1000 linear feet)
0-0.9 light
1.0 1.9 moderate
2.0 - 2.9 heavy
3.0 3.9 very heavy
4. 0 plus extremely heavy
SEC TECHNICAL REPORT A62-5
GPO 8251II-*
-------
SMOKE CONCENTRATIONS IN MONTREAL 91
Many cities have recorded concentrations in the range of
8.0tol2.0 COH units for sampling periods of 2hours. In Montreal
on April 14th, 1960, a maximum of 12. 8 was registered between
1900 and 2100 EST. During one half-hour period from 1930 to
2000 EST a peak of 15.4 COH units was reached, under a very
strong low-level warm frontal inversion.
During a period of forest fires in the fall of 1952 concentra-
tions produced by a mixture of forest smoke and urban smoke
reached 18. 0 COH units in Pittsburgh. Areadingof 23. Ounits has
been reported from Sydney, Australia, presumably from bush
fires inland.
The adjectival ratings have been used by several other
agencies and are used throughout this paper.
SAMPLING TIME
The relation between the quantity of smoke in the atmos-
phere and the otpical density was found to be linear for values of
I less than 50 percent. With the flow rate and spot size used in
our samplers this value often was exceeded for 2-hour samples in
the winter months. Therefore, a standard 1-hour sampling period
was used at all three locations from October to May. One-hour
samples have the added advantage of convenient correlation with
meteorological variables such as wind and temperature, which
are normally tabulated on an hourly basis in published summaries.
During most of the first year the Weather Engineering
sampler was operated on a 2-hour sampling period and was
changed to a 1-hour period only when necessary. Thus some of
the data represent averages over a 2-hour sampling period.
LOCATION OF SAMPLERS
The location of the three smoke samplers with respect to
Mount Royal and the downtown Montreal area is shown in
Figure 1, which also indicates the height of the air intake
above sea-level. The McGill and Weather Engineering intakes
are situated about 35 feet above the ground; the CBC intake is 12
feet above ground.
The following abbreviations are used throughout:
WEC -- Weather Engineering Corporation
sampler located on Crescent Street
McGill -- McGill University sampler located on the
Campus
CBC -- McGill University sampler located in the
SYMPOSIUM: AIR OVER CITIES
-------
92
SUMMERS
Canadian Broadcasting Corporation building
on top of Mount Royal
?<,VWEATHER ENGINEERING (I851)
SCALE IN MILES
Figure 1. Location of samplers with respect of Mount Royal
SEASONAL VARIATION IN SMOKE CONCENTRATION
Figure 2 shows a plot of the monthly averages at WEC
from January 1960* to September 1961. This figure indicates a
very marked seasonal variation. Readings remained in the top
half of the heavy range from January to April 1960 and then fell
sharply during May to relatively low readings from June to
August. Only in July, however, did the monthly average fall into
the light range. Beginning in August the concentrations increased
steadily, reaching the heavy range again by November. During
January 1961 concentrations reached a peak in the very heavy
range. From January on they declined steadily until May. During
the summer of 1961 values remained almost constant near 1.0
COH unit. September 1961 was the warmest on record in Montreal,
with a mean temperature nearer to that expected in July or
August; the smoke concentration during this month was lower by
27 percent than the corresponding value for September 1960.
There are two reasons for the marked seasonal trend:
*Note the January 1960 average is based on 16 days only.
SEC TECHNICAL REPORT A62-5
-------
SMOKE CONCENTRATIONS IN MONTREAL
93
3-0
'2-0
o
§
o
F
0Ł
I-
Ł
K
8
VERY HEAVY
t
•—• W.E.C.
o—o MCQILL
X-—X C.B.C.
i i i i i i i
J FMAMJJASONDJ FMAMJJAS
I960 1961
MONTH
Figure 2. Average monthly smoke concentrations in Central Montreal, January 1960 to
September 1961
1. The emission of smoke into the atmosphere is much
higher during the long winter heating season.
2. The ventilation of the city's air is severely restricted
in the winter months because of the almost continuous
stability of the atmosphere and frequent temperature
inversions caused by snow-covered or frozen ground.
These two factors combine to give smoke concentrations
•SYMPOSIUM: AIR OVER CITIES
-------
94 SUMMERS
in midwinter nearly three times those of midsummer.
The effect of snow cover on smoke concentrations in the
spring is shown in Figure 3. This diagram shows a plot of the
3-0
x2-0
l
"" 1-0
I960
1961-
-i 1 1 1 1 1 i i i i
3 9 15 21 27 5 II 17 23 29 4 10 16 22 28 4 10 16
FEBRUARY MARCH APRIL MAY
MID-DATE OF 21-DAY RUNNING MEAN
Figure 3. Twenty-one-day running mean at WEC late winter and spring 1960, 1961
21-day running means from February to mid-May for 1960 and
1961. The readings are very close until about March 17th, when
the 1961 readings began to fall below the corresponding 1960
values; by April 10th they were 0. 8 COH units lower. During
April 1961 the smoke concentrations averaged 24% lower than in
1960:
There are two possible reasons for this: either less pollu-
tion was emitted into the atmosphere during April 1961 or the
dispersion of the pollution by the atmosphere was much greater.
Since the mean temperature was 1. 0°F lower in 1961 the
first reason can be discounted. Heating plants should have pro-
duced about the same amount of smoke, and there is no evidence
of any reduction in emissions from other sources such as trans-
portation and industry.
One must look, then, for a meteorological reason to ex-
plain the increase in ventilation. The percentage variation in
mean wind speed and precipitation between the two Aprils was
less than that for some of the other months, February, March
and May; therefore winds or precipitation cannot be the cause.
Synoptically the weather patterns were rather similar, with fre-
quent storms in both Aprils. In the absence of any low-level
temperature soundings, the frequency of temperature inversions
cannot be determined accurately.
SEC TECHNICAL REPORT A62-5
-------
5MOKE CONCENTRATIONS IN MONTREAL 95
The only significant difference between the two winters
vas the snow cover. The total snowfall in Montreal in the 1959-
1960 winter season was 120 inches; in the 1960-1961 season it
Has only 86 inches.
By March 8th, 1961, the snow cover at Dorval Airport was
down to 2 inches and had disappeared from many parts of the
:ity. Two heavy snowstorms in mid-March increased the depth
again, but the last snow cover disappeared one week earlier than
in 1960. Throughout April there were frequent rains, which
speeded the thawing-out of the ground so that the chilling effect
of the ground on the air ended much earlier in 1961 than in 1960.
Thus the lowest layers of the atmosphere became unstable earlier
in 1961 and allowed increased ventilation, which caused the lower
smoke concentrations in April 1961.
The readings at McGill show exactly the same seasonal
trend, but the magnitude of the readings is lower by a factor of
one-third to one-half. This difference between the two locations
will be discussed later.
The few months of data available from the CBC location
show little evidence of any marked seasonal variation. The
average smoke concentrations from April through September
have remained almost steady, whereas at WEC and McGill the
values have halved. This suggests that although much less
smoke is produced in the summer months, the increased vertical
mixing due to convection carries the smoke aloft more readily.
In an earlier report * it was shown that during the period
January 15th to April 30th, 1960, the highest smoke concentra-
tions all occurred under inversion conditions caused either by
the advectionof warmer air from the south over the snow-
:overed ground or by low-level warm frontal surfaces over the
=ity.
The stalling of warm fronts just south of the St. Lawrence
ind Ottawa valleys in the winter months is a feature well-known
;o meteorologists in the area. At times this stagnation can lead
;o prolonged spells of high smoke concentrations (well above 25
Dercent more than the 21-day running mean) as happened for
;5-l/2 days in February 1960.
During the 1960-61 winter season high readings again oc-
::urred in these circumstances, but there was also a much greater
'requency of anticyclonic weather, which produced equally high
smoke concentrations. It therefore appears that two frequently
occurring types of weather situations can lead to extremely high
>ollution levels in Montreal during the winter months. In most
rther large cities only one situation, the anticyclone, regularly
iYMPOSIUM: AIR OVER CITIES
-------
96
SUMMERS
causes high pollution levels. These meteorological influences
will be illustrated by examples.
DIURNAL VARIATION OF SMOKE CONCENTRATION
Seasonal Change in Week-day Diurnal Variations at WEC.
A study of the week-day diurnal curves for the months
January 1960 to September 1961 indicates the same basic daily
variation in every month. Each day has two maxima (shortly after
sunrise and in the evening) and two minima (early morning and
early afternoon).
The shape of the curves shows a continual transition from
month to month. The curves can be classified into two basic
types, examples of which are illustrated in Figure 4. Type A
3-0
2-0
•0
TYPE A
DEC I960
I 3579 II I 3579
MORNING AFTERNOON
MID-TIME OF SAMPLE, E.S.T.
Figure 4. Examples of weekday diurnal curves at WEC
SEC TECHNICAL REPORT A62-5
-------
5MOKE CONCENTRATIONS IN MONTREAL
97
ms a small diurnal amplitude of less than 35 percent. The
naximum occurs in the morning; the early evening peak is near
md at times higher than the morning peak. Type A occurs from
November through April. Type B has a large diurnal amplitude
}f more than 50 percent with a strong peak in the morning. It oc-
curs from May through October.
In Figure 5 the top curve shows the diurnal amplitude ex-
pressed as a percentage of the monthly average and indicates a
100
80
60
40
20
0
-10
• • DIURNAL AMPLITUDE
X X MORNING MAXIMUM-EVENING MAXIMUM
o o AFTERNOON MINIMUM-OVERNIGHT MIN.
F M A M
A S 0 N D
J J
MONTH
Figure 5. Diurnal amplitude, morning maximum minus evening maximum, afternoon min-
imum minus overnight minimum as a percentage of average weekday values
pronounced seasonal variation. In the winter months the atmos-
phere remains stable for most of the day and overnight; hence
the diurnal variation is relatively small. In the late spring and
particularly in the late summer and fall., great stability develops
SYMPOSIUM: AIR OVER CITIES
-------
98 SUMMERS
overnight and is followed by daytime instability leading to large
diurnal variations. In mid-summer (July) the atmosphere tends
to remain unstable at times through the night hours or at least
develops no great stability. This leads to somewhat smaller di-
urnal variations as indicated in Figures 4 and 5.
The middle curve on Figure 5 shows the difference between
the morning maximum and the evening maximum again expressed
as a percentage of the average monthly smoke concentration.
This curve indicates that the morning maximum is most pro-
nounced in the fall, whilst in the winter there is little difference
between the two maxima.
The bottom curve on Figure 5 shows that the two minima
are almost of equal magnitude, although in the fall the lowest
readings tend to occur in the afternoon rather than overnight.
A full explanation of the shape of these diurnal curves can-
not be given at this time. The actual smoke concentration at any
given time is a function of both the rate of smoke production and
the rate of ventilation during the preceding few hours. Both of
these vary throughout the day and although the diurnal changes
can be assessed subjectively, the absolute magnitudes cannot be
obtained at present.
The morning peak shortly after sunrise is generally ascribed
to a Hewson^ fumigation and although this undoubtedly occurs on
occasions in Montreal, the curves in Figure 4 show little indica-
tion of this on the average. Rather there is an almost steady in-
crease in concentration from 0200 EST until shortly after sunrise,
The gradual buildup probably occurs because during these hours
more pollution is produced than can be ventilated. The rate of
ventilation increases sharply after sunrise and more than offsets
the big increase in production due to daytime traffic and industry
until late afternoon. By late evening the rate of ventilation has
fallen off again but is still greater than the rate of production
after man's daytime activities have ceased. Pollution therefore
falls to a minimum again shortly after midnight.
Comparison of Week-day with Week-end Diurnal Variations at
WEC.
From November through March the day-to-day variations
in pollution are so large that even with a sample of 30 observations
at any given time for each day of the week no significant variations
were found.
Average values of smoke concentration were found to be lOfi
lower on week-ends, but the individual diurnal curves were very
erratic. Once a larger sample is obtained allowing the effects of
random day-to-day meteorological changes to be supressed then
SEC TECHNICAL REPORT A62-5
-------
SMOKE CONCENTRATIONS IN MONTREAL
99
a comparison can be made between rates of production of pollu-
tion on week-days and week-ends.
During the summer months June to August 1960 and 1961
the day-to-day variation in the meteorological variables was not
so great. Although the sample consisted of only 26 observations
for each day of the week, significant variations were found be-
tween week-ends and week-days. The daily average was higher
for every week-day than for the week-end. On Saturdays the
average smoke concentration was 20 percent less than on week-
days; on Sundays it was 25 percent less. The curves are shown
for comparison in Figure 6.
1-6
0-8
0-4
week days
Saturdays
x x Sundays
5 7
MORNING
579
AFTERNOON
MID-TIME OF SAMPLE, E.S.T.
Figure 6. Comparison of weekday with weekend diurnal variations at WEC for the summer
months June to August 1960, 1961
SYMPOSIUM: AIR OVER CITIES
-------
100 SUMMERS
If we assume that the factors affecting ventilation are the
same on week-days as on week-ends., then the differences between
the curves in Figure 6 must be explained in terms of differing
rates of production of pollution. We know that industrial activity
is much less over the week-end and that the flow of traffic is very
different.
Perhaps the most significant feature of Figure 6 is the
complete breakdown of the normal diurnal variation on Sunday.
There is no morning peak because most of man's activities are
at a standstill in the early hours of Sunday.
Comparison of Diurnal Variations at Top and Bottom of Mount
Royal
By April 1961 the sampler on top of Mount Royal was op-
erating on a regular basis. The 1-hourly readings are compared
with those at Me Gill for the month of April. This small sample,
based on 20 days, could not be combined with observations for
May since the change to Daylight Saving Time shifts the daily
cycle of production one hour with respect to the daily cycle of
ventilation.
Figure 7 is an example of the type of comparison that may
be made when more data become available.
The Me Gill curve shows evidence of a Hewson fumigation
immediately after sunrise between 0500 and 0600 EST.
The maximum at the top of Mount Royal at 0800 EST indi-
cates a lag of 2 hours after the overnight stability has broken
down sufficiently to allow maximum transfer of pollution up to
that level.
EFFECT OF MOUNT ROYAL "SMOKELESS
ZONE" ON SMOKE CONCENTRATIONS
Nine hundred and fifty-seven simultaneous hourly obser-
vations of smoke concentration for week-days only at WEC and
Me Gill during February and March 1961 were tabulated accord-
ing to the wind directions recorded at Dorval Airport. The
mean concentrations for each of the eight cardinal compass
points are shown in Table 1. The following conclusions can be
made from this table:
1. Averaged over the quadrant north through east to
south, where the Mount Royal smokeless zone can have
no effect, the smoke concentrations at Me Gill are 33
percent lower than at WEC. The reason is that the im-
mediate surroundings pf Me Gill have fewer sources of
pollution.
SEC TECHNICAL REPORT A62-5
-------
SMOKE CONCENTRATIONS IN MONTREAL
101
1-2
z
LU
(J
0-8
0-4
0
CBC
_L
SUNRISE
J_
SUNSET
J
LL
_L
I
5 7
MORNING
I
I
J 5 7 9 I
AFTERNOON
END-TIME OF SAMPLE, E.S.T.
Figure 7. Comparison of diurnal variations at top and bottom of Mount Royal for April
1961
2. When winds are tabulated on a 16-point compass the
greatest frequency is from the west-southwest rather
than southwest (see Figure 8); the majority of the
southwest winds shown in Table I are really west-
southwest winds. With winds from this direction the
air at the McGill Sampler has passed over Westmount
Summit Park and skirted Mount Royal Park and has
SYMPOSIUM: AIR OVER CITIES
-------
102 SUMMERS
picked up very little pollution during the last 2 miles of
its trajectory. Thus the Me Gill readings drop to 0.8
COH units. With southwest winds the buffering effect of
the smokeless zone is not felt at WEC, where the average
concentration is 2. 7 COH units.
TABLE I
EFFECT OF MOUNT ROYAL "CLEAN AREA" ON
SMOKE CONCENTRATIONS AT WEC AND McGILL
DURING FEBRUARY AND MARCH 1961
_N NE _E_ SE _S_ SW W NW Calm
Percentage
Frequency 15 20 55 6 6 31 14 3 1/2
Mean Wind
Speed, mph 9.8 14.2 10.3 11.2 7.7 13.4 15.4 9.8
Mean Smoke
Concentration
at WEC,
COH units 3.4 2.9 3.3 2.9 3.9 2.7 1.6 1.8 4.1
Mean Smoke
Concentration
at Me Gill,
COH units 2.2 1.9 2.4 2.0 2.6 0.8 0.6 1.4 3.5
3. The Me Gill site has even better protection with winds
from the west, and concentrations drop to 0. 6 COH
units. At WEC the west winds have passed over the
edge of the Park, and concentrations drop to 1.6 COH
units.
4. With northwest winds both McGill and WEC are pro-
tected by the Park, and smoke concentrations are
similar.
5. In calm conditions the Park has no effect and readings
at both locations are high.
Even without the Park, winds from southwest through
north could be expected to give lower smoke concentra-
tions because stability is lower and ventilation is there-
fore increased, (see Figure 8).
A similar analysis has been done for June to August 1961.
This analysis shows the same general trend in variations of
smoke concentration but with the following important differences:
SEC TECHNICAL REPORT A62-5
-------
SMOKE CONCENTRATIONS IN MONTREAL
103
24% SLIGHTLY UNSTABLE
ATMOSPHERE
No temperature liwerslor
Clean Arctic air,
Goad Vortical Mixing.
Low Pollution Potential
777A- HEAVY INDUSTRY
40% STABLE ATMOSPHERE
Frequent strong tempera-
ture Inversions.
No Vertical Mixing
High Pollution Potential
32% NEUTRAL TO SLIGHTLY
STABLE ATMOSPHERE
No temperature Inversions.
But only slight Vertical
mixing.
Light to Moderate
Pollution Potential.
4% CALM WINDS
No Horizontal or Vertical
Mixing
High Pollution Potential
Note: These composite wind data are for the months November to April
for the 10-year period 1947 to 1956 at Dorval Airport and were
supplied by the Meteorological Branch, Department of Transport.
Figure 8. Location of heavy industry on Island of Montreal together with the wind fre-
quencies for the months November to April
1.
2.
Averaged over the quadrant north through east to
south the smoke concentrations at McGill are only 16
percent lower than those at WEC.
With southwest and west winds the reduction at McGill
is only 45 percent, compared to a reduction of 60 to 70
percent in the winter.
These values indicate that although Mount Royal has an
appreciable effect in the summer, it is not nearly so marked as
in the winter. This is to be expected, since a much larger, pro-
portion of the total smoke in summer is due to the more distant
industrial sources rather than to local household and commercial
heating plants.
SYMPOSIUM: AIR OVER CITIES
-------
SUMMERS
RELATION OF SMOKE CONCENTRATIONS TO
SYNOPTIC FEATURES ON THE WEATHER MAP
Two cases will be discussed to illustrate how changes in
the synotpic features of the weather map affect the smoke con-
centrations in Montreal.
Prolonged Spell of Anticyclonic Weather, March 20th to 24th,
1961
The surface weather maps during this period are shown in
Figures 9 to 11. An Arctic cold front passed through Montreal
Figure 9. Surface Weather Map
0100 EST 20 March 1961
late on March 19th, and the high-pressure area drifted slowly
southeast from James Bay to a position just east of Montreal by
the morning of the 22nd. The vertical temperature structure at
Maniwaki, Quebec, (about 125 miles northwest of Montreal)
during this period is shown in Figure 12. At 1200 G. M.T. (0700
EST) on the 20th the profile shows a shallow layer of cold air up
to 940 mb. By the 21st subsidence associated with the high and
advection of warmer air aloft, combined with strong nocturnal
cooling at the ground, had created an extremely strong tempera-
ture inversion of 16°C in the lowest 600 feet of the atmosphere.
This inversion persisted for three days, but by the 23rd its
SEC TECHNICAL REPORT A62-5
-------
SMOKE CONCENTRATION IN MONTREAL
105
-
•
Figur. 11. Surtoc. W.otKct Map
0100 EST 24 Morch 1961
SYMPOSIUM: AIR OVER CITIES
-------
106 SUMMERS
magnitude had decreased to 9° C in the lowest 400 feet.
On the 24th the high had moved off to the east as a small
low drifted across New England from the west, and the anti-
cyclonic inversion had completely disappeared. The cold front
moved down rapidly from the northwest and passed through
Montreal early on the 25th.
700
800
900
1000
I200Z
22 MARCH
J_
-20
-10
TEMPERATURE, "C
Figure 12. Vertical temperature profiles at Maniwaki -- March 20-24, 1961
SEC TECHNICAL REPORT A62-5
-------
SMOKE CONCENTRATION IN MONTREAL 107
The pollution record during this period for WEC and CBC
is shown in Figure 13. The readings at McGill were in very close
agreement with WEC and so are omitted from the diagram. Un-
fortunately, the CBC sampler was not operating at the beginning
8-0
I e-o
2-0
•«•• ' ' M A/V \ A^
] \ . / V/^T
\A
12 18 00 06 12 18 00 06 12 18 00 06 12 18 OO 06 12 18 00
MARCH 20 MARCH 21 MARCH 22 MARCH 23 MARCH 24
END-TIME OF SAMPLE, E.S.T.
Figure 13. Smoke concentrations at WEC and CBC during prolonged spell of anticyclonic
weather, March 20th - 24th, 1961
of the period. The morning peak occurred at WEC at varying
times between 0400 and 1000 EST on each of the four mornings.
At CBC the concentration dropped to near zero at about
0600 EST on the 22nd and 23rd, an indication that the inversion
was below the top of Mount Royal.
Between 1200 and 1300 EST on the 23rd the temperatures
at Dorval Airport increased by 9° F indicating the final break-
down of the inversion. With only very light winds and a slightly
stable atmosphere (see Figure 12), the concentration was again
high at WEC on the morning of the 24th. At the top of Mount
Royal, however, readings remained relatively high through the
night of the 23rd/24th in the absence of the inversion. Smoke
concentrations finally fell off on the night of 24th/25th because of
freshening westerly winds and the passage of the cold front, which
brought in clean Arctic air.
It is interesting to note the differences between the minimum
temperatures recorded at the McGill Observatory and Dorval
Airport during this period:
SYMPOSIUM: AIR OVER CITIES
-------
108
Date
March 21
22
23
24
SUMMERS
Minimum
Temper ature^F
Dorval Me Gill
2
10
10
3 1
11
14
26
34
Temperature
Difference, °F
9
4
16
3
There is no doubt that smog is one of the factors that con- |
tributes to large differences in temperatures between cities and »
the surrounding countryside on a night of strong nocturnal radia-
tion cooling. In Figure 13 the average smoke concentration is
lower during the night of the 21st/22nd than on the preceding and
following nights and could account for the much lower tempera-
ture difference. Cloud cover on the night of 23rd/24th produced
a general restriction in nocturnal cooling.
Variations in Smoke Concentrations during the Passage of a
Storm
Figure 14 shows the variations at WEC and McGill during
the period February 12th to 15th, 1961. At this time the sampler
6-0 -
ANTICYCLONIC
_ INVERSION
MARITIME
WEAK ARCTIC / TROWAL
WARM FRONT
" \\ ^
12 18 00 06 12 18 00 06 12 18 00 06 12
FEE 12 FEB 13 FEB 14 FEB 15
END-TIME OF SAMPLE, E.S.T.
Figure 14. Variations of smoke concentration at WEC and McGill during passage of ridge
of high pressure and storm, February 12th - 15th, 1961
SEC TECHNICAL REPORT A62-5
-------
SMOKE CONCENTRATIONS IN MONTREAL
109
had not been installed on Mount Royal. The surface weather maps
are shown in Figures 15 through 17.
-rx^"~ / V'
t, ,,"*>^v-c V
*•-v B iis» T
SjK- /-Kt- l>
15*,* *Ł&*
*
**\'^^~
, *&jgsi!^^jaup|Ł
:"'/ /A^ ttr^Ł^^-
Figure 15. Surfaco Weather Map
0100 EST 13 Feb. 1961
Smoke concentrations were extremely high on the morning
of the 13th under a strong anticyclonic inversion. On this occasion
the minimum temperature at Dorval Airport was -6°F compared
to 8°F at the Me Gill Observatory.
At noon on the 13th the winds at Dorval increased to 15 mph
from the east or northeast, but smoke concentrations remained
high under a warm frontal inversion. At 2100 EST a weak Arctic
warm front (not shown on Figures 12 and 13) passed through
Montreal. It was followed by a strong southerly gradient with
southeast winds between 15 and 20 mph, which produced low
smoke concentrations. With the approach of a trowal (shown
as an occlusion on Figure 16) the surface winds decreased and
the increased stability due to over-running warm air at low
levels produced another sharp peak in pollution at 1000 EST. For
the remainder of the day winds were steady from the southwest at
20 mph, but the concentrations at VVEC did not decrease signifi-
cantly until after the passage of the Arctic cold front at midnight,
which brought westerly winds at 20 to 30 mph.
SYMPOSIUM: AIR OVER CITIES
-------
110
SUMMERS
The lower concentrations at McGill during the after-
noon and evening of the 14th are due to the effects of Mount
Royal.
SEC TECHNICAL REPORT A62-5
-------
SMOKE CONCENTRATIONS IN MONTREAL
111
*"?rv -^^^^-^
.** "* ***P_T..> - 1 7 . ^^^-^f ""_*• -<•?
Figur* 17. Surface Weather Map
0100 EST 15 F«b. 1961
Summary and Conclusions
The preceding analysis clearly shows that the seasonal
and daily variations of smoke concentration in central Montreal
are intimately related to changes in the meteorological variables.
In particular the effect of winter snow cover is very important in
that it produces almost continuous stability and frequent tempera-
ture inversions in the lowest levels of the atmosphere at a time of
the year when the emission of pollution is at a maximum.
The problem is further aggravated by the relative location
of the heavy industrial areas in Montreal. Pollution from these
industrial sources is blown directly across some or all of the
business and residential sections of the Island during the 40 per-
cent of the time in the winter months, when the atmospheric con-
ditions are least able to disperse it. This is illustrated diagram-
aticallv in Figure 8.*
The advantage of a "smokeless zone" such as Mount Royal
Park is very evident in that pollution concentrations in the sur-
rounding areas are reduced considerably.
*Wind frequencies were compiled from data supplied by the
Meteorological Branch, Department of Transport.
SYMPOSIUM: AIR OVER CITIES
-------
112 SUMMERS
Although strong inversions form in Montreal under anti-
cyclonic weather conditions, they occur just as frequently due to
the advection of warm air at low levels in the atmosphere over
the cold ground.
The second effect is most pronounced at a location which:
1. has a long and continuous period of snow cover;
2. lies in a valley orienated in a general east to west
direction so that cold air is either trapped or funnelled
in at ground level;
3. is near to or north of the main winter storm tracks to
provide frequent periods of low-level warm air ad-
vectation just above the cold air.
Montreal, with a population of 2 million, is the largest
urban area in North America to satisfy all the requirements for
a pronounced inversion of the second type, a factor that hitherto
has not been considered in air pollution studies. Since these
conditions probably occur in other rapidly growing urban areas
in central Canada and the northern United States, they should be
taken into account for urban planning and the location of industrial
plants.
REFERENCES
Denison, P. J. , B. A. Power and P-W. Summers, "Analyses
of Air Pollution Levels in Montreal Related to Meteorological
Variables, " presented at Kingston Meeting, Royal Meteoro-
logical Society, June 1960 (unpublished).
Hemeon, W. C. L. , George F. Haines, Jr. and Harold H. Ide,
"Determination of Haze and Smoke Concentrations by Filter
Paper Samplers, " Journal Air Pollution Control Assn.,
Vol. 3, No. 1, pp.22. August 1953.
New Jersey State Department of Health, "State-wide Air Pollu-
tion Survey -- Smoke Index, " 1958.
Hewson, E. Wendell, "Atmospheric Pollution, " Compendium
of Meteorology, pp. 1147, 1951.
DISCUSSION
DR. HEWSON: Do you have other examples of trapping
below warm fronts, especially slowly moving warm-front
SEC TECHNICAL REPORT A62-5
-------
SMOKE CONCENTRATIONS IN MONTREAL 113
systems? It seems to me that this is a situation in which one
might expect quite high concentrations. You had one brief period
of it during your study.
MR. SUMMERS: Yes, we did a study in the first winter,
and we found that 20 percent of the time the air was being
trapped under a warm front. This trapping effect is quite a well-
known feature of the St. Lawrence and Ottawa valley region. It
can persist for several hours, and on one occasion did persist for
2-1/2 days, a very high concentration.
DR. HILST: In your comparison of the McGill station and
the effect of the Mount Royal Park, you pointed out that the McGill
station was affected first, then it improved; the concentrations
there went quite low, while yours stayed up. But then when you
were both affected, their concentrations came up, rather than
yours coming down. Do you have any explanation for this?
MR. SUMMERS: Not really, except that with the north-
west wind the air is not blowing directly across the mountain but
is going around the mountain, which would possibly affect it. This
occurred in a very small percentage of the time, 4 percent. It
was a very small sample.
DR. SCHMIDT: Mr. Summers, you talked about that
morning maximum, and you talked a bit of the cause of that
maximum. Did you also see something of a shift throughout the
year ?
MR. SUMMERS: Yes. The shift occurred within 2 hours
after sunrise. And about 4 hours in the midwinter.
MR. MUNN: I have some comments about the evening peak
that might be helpful. Recently I have been looking at data from
a 200-foot tower in Ottawa, and also from a tower at Douglas
Point in Ontario. We have found that in the early evenings, under
light-wind, high-pressure situations, the inversion forms later on
in the evening. The inversion doesn't collapse completely, but it
diminishes somewhat, and comparable with this the wind at 200
feet is a maximum in the afternoon. It decreases in the early
evening when the inversion reduces in intensity, and in the late
evening the 200-foot wind increases. It is at that point that the
pollution level goes up. I have found that the 200-foot wind in-
crease is a very good predictor of an evening peak.
SYMPOSIUM: AIR OVER CITIES
-------
The Air Over Philadelphia
FRANCIS K. DAVIS, Jr., Professor of
Physics, Drexel Institute of Technology and
Staff Meteorologist, WFIL-TV
Summary
Simple air pollution theory indicates that relatively few meteorological and air
quality observations are necessary in general air pollution studies. Even these relative-
ly few observations are usually lacking, however, in studies of the behavior and quality
of the air over a particular city. The observations of meteorological quantities related to
air pollution should be standardized and the number of observations increased. Theory
should be extended to include the complications brought about by characteristics of urban
areas. At the same time, some useful rules for forecasting and regulating air pollution
can be formulated for a city such as Philadelphia on the basis of currently available ob-
servational data.
THE PROBLEM
A steadily increasing population and a continuing trend to-
ward industrialization within that population have resulted in an
ever-increasing volume of the various waste products being poured
into the atmosphere. Wanta and Stern (1957) point out that 21 per-
cent of the world's people now live in cities with populations of
20, 000 or greater and that by the middle of the next century the
number will have increased to 90 percent. The population of the
United States is increasing at a rate of 15 percent per decade,
and the amount of energy expended per person is increasing
yearly.
In spite of engineering efforts to reduce air pollutants to
their least offensive form, they could become, if they are not
already, a very serious health and economic problem in the Phil-
adelphia area and in many other areas. Maneri and Megonnell
(1960) describe the result of three statewide surveys of the extent
and seriousness of air pollution in New York. Although these sur-
veys were made independently by different agencies and with dif-
ferent techniques, a composite conclusion was reached: "Air
pollution is a widespread complaint in New York State communi-
ties of all sizes, and its abatement and control receive too little
attention. "
When waste products have been discharged into the atmos-
phere, their subsequent life history is a meteorological problem.
Even though pollutants are put into the atmosphere at a constant
rate, the condition of the atmosphere will determine whether
these contaminants accumulate to cause discomfort and damage
115
-------
116 DAVIS
or disperse enough that no problem results. Atmospheric condi-
tions also determine whether such a problem will be localized in
one small section of the community or will be widespread. Cer-
tainly any wise city planning activity should consider local weather
conditions and their relation to air pollution when zones are set
for industrial and residential development.
DIFFUSION PARAMETERS
When contaminants are released at a particular point, the
atmosphere acts as its own cleansing agent by transporting and
diffusing such contaminants, thus reducing their concentrations.
A complete mathematical analysis of diffusion processes which
would allow quantitative calculations is difficult. Diffusion equa-
tions are available from several sources, some of the most widely
used being those developed by Sutton (1932, 1947). Such equations
indicate that for a given contaminant emission the properties of
the atmosphere of primary concern in determining concentrations
at any point are wind speed, wind direction, and air stability as
reflected by the temperature profile in the vertical direction.
Very recent observations reported by Maneri and Megonnell (1960),
Boettger (1961), Dickson (1961), Markee (1961), and Turner(1961)
tend to confirm this indication. For example, Dickson reports a
significant correlation of atmospheric pollution with all three
properties; Markee reports high correlations between certain
pollution levels and vertical temperature differences in Louis-
ville; and Turner reports that the meteorological variables of
wind speed, temperature, and stability account for about half of
the variance in daily city-wide sulfur dioxide concentrations and
soiling indices in Nashville, Tennessee.
DATA REQUIREMENTS
Relating meteorological parameters to air pollution concen-
trations in a given city is not a simple job. Although most large
cities across the nation make air pollution measurements of some
kind, there are no fixed standards for the type of measuring in-
struments to be used, for the general location and height at which
such measuring instruments are exposed, or for the atmospheric
conditions under which such measurements should be made. As
a result, collection and comparison of pollution data in an effort
to consider relative pollution levels in various cities across the
nation, and even within a single city, are almost meaningless.
It is highly desirable to accumulate data that would lead to stand-
ardization of pollution-measuring instruments and techniques so
that a picture of pollution distribution on a nationwide basis and
within a city can be established. Furthermore, Hemeon (1958)
points out that we have yet to determine which of the many air
SEC TECHNICAL REPORT A62-5
-------
THE AIR OVER PHILADELPHIA 117
pollution effects are objectionable, and that from an engineering
viewpoint we have no objective basis for evaluating these air pol-
lution nuisances.
In the meteorological department, measurements of wind
speed and direction are available at many U.S. Weather Bureau
Stations. These measurements, however, are frequently made
at relatively high levels, often at airports located at some dis-
tance from urban areas and in topographical surroundings quite
different from the urban area they represent. Low-level wind
flow patterns throughout urban areas ought to be much more def-
initely understood under conditions of high pollution potential.
The vertical temperature structure of the air over a city is
probably as important as the wind data in analysis of pollution
problems. But in most cities no vertical temperature measure-
ments are available. The vertical temperature distribution must
be inferred from other measurements. For example, Turner
(1961), in his Nashville study, proceeded on the premise that sta-
bility depends mainly on net radiation and wind speed and that net
radiation is determined by the elevation of the sun and by the
amount and height of cloud cover. He used these parameters to
set up a stability classification.
Observations of the vertical temperature profiles have been
made recently over Louisville, Kentucky. DeMarrais (1961) re-
ports that these data show that stability conditions over an urban
complex are quite different from those over open areas, where
most other observations have been made.
Thus, it would also be highly desirable to standardize and
increase the meteorological observations related to air pollution
and to extend the observations and the theory to cover the com-
plications brought about by the physical characteristics of urban
areas.
METEOROLOGY OF THE PHILADELPHIA AREA
Although there is great need for additional meteorological
and air pollution measurements to define more clearly the rela-
tionships between the two, existing observations and knowledge
can often be combined and used to gain a greater measure of
understanding of the air pollution problem in a particular city. In
Philadelphia we recently undertook an investigation of the weather
influences on urban air pollution with the aim of determining
whether periods of potentially high air contamination might be
anticipated on the basis of changing weather patterns.
To establish some of the detailed weather patterns for
Philadelphia, a collection of hourly surface weather observations
SYMPOSIUM: AIR OVER CITIES
-------
118 DAVIS
taken at the Philadelphia Weather Bureau Station at International
Airport was analyzed. Through the cooperation of the National
Weather Records Center in Asheville, North Carolina, these rec-
ords were acquired in the form of WBAN- 1 punch cards suitable
for machine computations. They cover the period July 1949
through June 1959.
Studies such as those by Kleinsasser and Wanta (1956),
Machta, Scott, and Korshover (1958), Keagy and Schueneman
(1958), Neiburger (1959), and Niemeyer (1960) indicate that light
winds are very highly correlated with high degrees of atmospher-
ic pollution. Analysis of the Philadelphia data shows that the
period of lightest winds is summer and fall, particularly late
summer and early fall. This coincides with the period of max-
imum frequency of clear skies at night, which contribute to low-
level cooling at night and to high stability. The large-scale
weather observations indicate that poorest diffusion conditions in
the Philadelphia area occur in the late summer and early fall and
generally are associated with winds from the southwest. The late
summer and fall are potentially the seasons of greatest air pollu-
tion risk in Philadelphia on the basis of low wind speeds, or low
ventilation rate, alone. A detailed analysis of the general mete-
orological patterns for Philadelphia is presented in a report pub-
lished by the Air Pollution Control Section, Department of Public
Health, City of Philadelphia [Davis (1960)J.
RELATION OF HIGH POLLUTION LEVELS TO
METEOROLOGICAL CONDITIONS
In an effort to relate air pollution conditions in Philadelphia
to weather conditions and weather patterns, dates of high particu-
late concentration in Philadelphia were obtained for the years
1957 through 1959. The particulate concentrations were recorded
at two sites by high-volume and AISI samplers. The basis for
classifying a daily reading as high was arbitrarily set at 1. 5 times
the monthly average. Since the list of dates selected on this basis
was quite long, an additional criterion was imposed: the high-
pollution dates must cover a period of 2 to 5 days in which read-
ings for at least one of the days were high at both sampling sites
and on both sensing devices.
This list totals 24 "episodes" covering 68 days, a little over
6 percent of the days in the 3-year period. The original list, how-
ever, included 125 days, or more than 11 percent of the total days.
Seasonally, the distribution shows eight "episodes" in the winter
months, four in the spring, two in the summer, and eight in the
fall.
The general weather patterns associated with each of these
SEC TECHNICAL REPORT A62-5
-------
THE AIR OVER PHILADELPHIA 119
pollution periods were investigated. Although these patterns
were varied, some characteristics were persistent. On the basis
of these a forecasting system could be set up to provide a warning
when weather conditions likely to aggravate air pollution problems
are developing. These conditions have a typical general history:
a continental polar high-pressure center moves out of Canada
across the Great Lakes, sometimes going slowly across New York
State and off the Northeast coast, sometimes down into the Ohio
Valley and eastward across the Middle Atlantic area to the Mid-
east or Northeast coast. These highs become stationary or very
slow-moving at the coast, and the Philadelphia area is influenced
by very light winds for several days. During most of that period
the high-pressure center is generally to the east, so the air flow
is predominantly from a direction between southeast and south-
west, with southwest heavily predominant.
Wind speeds during the problem periods were predominantly
less than 8 mph; during 15 of those periods such speeds were re-
corded continuously for 14 hours or longer. High pollution levels
were recorded with somewhat higher wind speeds in the spring
season.
To associate wind directions with high pollution "episodes'1
more specifically, the hourly wind directions were tabulated for
these periods. Figure 1 shows a composite wind rose for these
high pollution periods. The high percentage of winds from west-
southwest through south is apparent.
Because of the higher wind speeds associated with the spring
episodes (March, April, and May), the high pollution periods for
those months were analyzed separately. The over-all pattern for
wind direction is not greatly changed, but the predominance of
southwesterly winds is even stronger. Figure 2 shows the wind
rose associated with high pollution episodes in the spring.
RELATION OF LOW POLLUTION LEVELS TO
METEOROLOGICAL CONDITIONS
The weather conditions related to low levels of air pollution
in Philadelphia were studied for the same 3-year period. Low
levels were determined by records from the same samplers and
were arbitrarily set at 0. 6 of the monthly average or lower. The
list included only those dates for which low readings were record-
ed at both locations. This list was heavily weighted with dates
that fell on Saturdays and Sundays; these dates were eliminated
because studies by the Air Pollution Control Section have demon-
strated that the dramatic lowering of air pollution levels on week-
ends results from reduced community activity rather than from
meteorological factors.
SYMPOSIUM: AIR OVER CITIES
-------
120
DAVIS
NW
W —
SW
NE
— E
SE
Figure 1. Percentage of wind directions during periods of high pollution, 1957-1959-
These periods of low pollutant levels showed a very con-
sistent weather pattern dominated by high, gusty winds and fresh
continental polar air. Most often the weather map showed a low
pressure center near the mideast or northeast coast and a high
center west of the Philadelphia area. Winds were thus from a
direction between northeast and northwest, mostly northwest,
with speeds often around 20 mph and gusts over 30 mph.
VENTILATION RATES
The horizontal motion of the air transports pollutants across
the city. Contaminants put into the air may be spread out verti-
cally so that surface concentrations will be low if the air is un-
stable. If wind speeds are high, concentrations will be relatively
low because the volume of air into which the pollutants are dis-
charged is large.
SEC TECHNICAL REPORT A62-5
-------
THE AIR OVER PHILADELPHIA
121
NW
HE
SW
SE
Figure 2. Percentage frequency of wind directions during high pollution periods in the
spring season 1957-1959-
If we assume that contaminants put into the air in and
around Philadelphia are all discharged at levels below 500 feet
and that under stable conditions they remain within that layer, the
maximum allowable discharge rates can be estimated crudely for
certain wind speeds.
The average wind speed in Philadelphia is between 9 and 10
mph. Wind speeds of less than 8 mph are recorded 37 percent of
the time, and speeds of 12 mph or less are recorded 73 percent
of the time. On this basis it seems reasonable to use a speed of
10 mph as a first approximation in computing an average ventila-
tion rate. Wind speeds of this magnitude and less are often asso-
ciated with stable air, but this is not necessarily true during the
daylight hours. Thus any assumption that the contaminants are
mainly confined in the lower 500 feet is open to question and
SYMPOSIUM: AIR OVER CITIES
-------
122 DAVIS
should be evaluated on the basis of additional data.
In spite of the limiting assumptions it seems worthwhile to
consider quantitatively the way in which concentrations of pollu-
tants may be expected to build up in the air over Philadelphia
under certain circumstances.
The physical layout of the city is such that an area 10 miles
wide and 15 miles long along a southwest to northeast axis should
be representative of the surface area to be ventilated. If the con-
taminants are considered to be concentrated in a 500-foot depth,
the volume of air to be ventilated is 14. 2 cubic miles. The aver-
age rate of generation of sulfur dioxide in Philadelphia is fairly
accurately known to be about 830 tons per day. Since most of
this is generated in the southern or southwestern to central sec-
tions of the city, this seems to be a suitable contaminant to work
with in quantitative calculations.
Let us assume that completely fresh air is flowing through
our space at a certain rate (the ventilation rate), that sulfur diox^
ide is being generated in that space at a fixed rate, and that the
contaminant is instantaneously distributed throughout the air in
the space. Then, the rate at which the total volume of sulfur di-
oxide in the space builds up with time can be expressed as
dV Pdt -lAj.-^- dt
a
where V volume of sulfur dioxide in the air at any time
P rate of production of sulfur dioxide
830 tons per day - 2. 64 x 10"6 cubic miles per hour
^r ventilation rate in cubic miles per hour
V total volume of air 14. 2 cubic miles.
a
V
Note that— - concentration, C.
a
For a 10-mph wind speed, the ventilation rate is 9. 47 cubic
miles per hour, and
dV 2.64xlO-6dt 9.41--dt
2. 64 x 10"6- V
1. 5
dt
SEC TECHNICAL REPORT A62-5
GPO 8251II-!
-------
AIR OVER PHILADELPHIA 123
1.5 In (2. 64 x 1(T6- ~) t + constant
_
Now, at t 0, V 0. . ' . A = 2. 64 x 10~6, so
V fi -t/1'5
2.64 x 10-6 _J_ = 2.64 x 10~be
1. 5
and V 1.5 (2. 64 x 10~6) (l-e~t/1-5)
To obtain the maximum V, note that as t —• °°, e ' -* 0,
io, V approaches 1. 5 (2. 64 x 1CT6) as a limit. Therefore, after
i long period of time,
: ~ L 5 ^2' 64 x 10 * ^ 2. 8 x 10" 7 parts of SO2 per part air
L ^t, &
or, C 0.3 ppm.
It is hardly likely that pollutants would be restricted to the
ower 500 feet of the atmosphere when the wind speeds average
[0 mph. The expected concentrations would be one-half of the
:alculated value or less, if the vertical mixing extended through
t layer 1000 feet deep or more.
In many of the periods considered earlier the wind speed
vas recorded as less than 4 mph on many observations. If we
:alculate for a wind speed of 3 mph, the ventilation rate becomes
i. 84 cubic miles per hour and the maximum concentration is
L 0 ppm.
The concentration buildup, of course, is quite rapid in the
;arly stages and then tends to level off, approaching the maximum
'or any given set of conditions. For example, if we use about
lalf of the maximum concentration for a wind speed of 3 mph; that
s, 0. 5 ppm, we find that it takes about 4 hours to attain that
ralue. A graph showing the rate at which concentrations build up
'or a low ventilation rate is plotted in Figure 3. Note that this is
jased on the assumptions that the air coming into the city is en-
irely clean and that the sulfur dioxide is restricted to the lower
iOO feet of air.
These derived concentrations are in good agreement with
SYMPOSIUM: AIR OVER CITIES
-------
124
DAVIS
observed concentrations of oxides of sulfur in the city. The ob-
served concentrations averaged about 0. 15 ppm over the past 3
years, with an estimated rise to 0. 5 to 1. 5 ppm if these oxides
varied with the soiling index during the periods of minimum venti-
lation.
o
p
8
1.0
0.8
0.6
0.4
0.2
2 4 6 8 10 12
TIME, hours
Figure 3. S02 concentrations versus time for a wind speed of 3 mph.
EXPERIMENTAL FORECAST PERIOD
At the suggestion of Mr. Raymond Smith, Chief of the Air
Pollution Control Section, Department of Public Health, of Phila-
delphia, a short experimental program was set up to predict high
pollution episodes for the city. This program embraced the per-
iod January 1 through April 30, 1961, but excluded the first 2
weeks in March. During this program we were to alert the City
to the probability of high pollution levels at 9 AM at least 24 hours
before the dates on which such high pollution levels were expected
to occur. We were to issue a definite high pollution level fore-
cast by 9 AM on the date of expected occurrence and to give 24
hours notice of the expected time of relief.
SEC TECHNICAL REPORT A62-5
-------
THE AIR OVER PHILADELPHIA 125
Five air pollution "episodes'1 were recorded during the ex-
perimental program. Four of these were successfully forecast.
For two episodes we were able to give 48 hours advance warning,
and for the other two, 24 hours. On three occasions we alerted
the city 24 hours in advance but cancelled the alert by 9 AM on
the alert date. On these dates no high pollution levels were re-
corded. Thus, periods of high pollution level in Philadelphia
were successfully forecast at least 24 hours in advance with 80
percent accuracy.
SUMMARY
Simple air pollution theory indicates that relatively few me-
teorological and air quality observations are necessary in general
air pollution studies. Even these relatively few observations are
usually lacking, however, in studies of the behavior and quality
of the air over a particular city. The observations of meteoro-
logical quantities related to air pollution should be standardized
and the number of observations increased. Theory should be ex-
tended to include the complications brought about by characteris-
tics of urban areas. At the same time, some useful rules for
forecasting and regulating air pollution can be formulated for a
city such as Philadelphia on the basis of currently available ob-
servational data.
BIBLIOGRAPHY
Boettger, C. M. , 1961: Air pollution potential east of the
Rocky Mountains: Fall, 1959. Bull. Amer. Meteor. Soc. ,
vol. 42, no. 9, pp 615-620. "
Davis, F. K. , I960: The Atmosphere Over Philadelphia. Air
Poll. Control Sec. , Dept. of Public Health, City of Phila-
delphia, 56 pp.
DeMarrais, G. A. , 1961: Vertical temperature difference ob-
served over an urban area. Bull. Amer. Meteor. Soc. ,
Vol. 42, No. 8, pp 548-554.
Dickson, R. R. , 1961: Meteorological factors affecting par-
ticulate air pollution of a city. Bull. Amer. Meteor. Soc.
Vol. 42, No. 8, pp 556-560.
Hemeon, W. C. L. , 1958: What needs to be done in the future.
Proc. Nat'l Conf. on Air Pollution, U.S. Dept. of Health,
Educ. and Welfare, pp 326-330.
Keagy, D. M. and Schueneman, J.J., 1958: Air pollution in
the Birmingham, Alabama area. Tech. Rep. A 58-8, U.S.
Dept. of Health, Educ. and Welfare, 17 pp.
SYMPOSIUM: AIR OVER CITIES
-------
126 DAVIS
7. KLeinsasser, T.W. and Wanta, R. C. , 1956: The development
of a forecasting service for use in air pollution control.
Paper presented APCA meeting, Buffalo, N. Y. , May, I95ji
8. Machta, L. , Scott, G. , and Korshover, J. , 1958: Weather
Bureau Research on air pollution potential. APCA Proc,,
58-25, 19 pp. ~'
9. Maneri, C. S. and Megonnell, W. H. , I960: Comprehensive
area surveys inN.Y. State. Journ. APCA, Vol. 10, No. 5
pp 374-377.
10. Markee, E. H. Jr., 1961: Effects of vertical temperature
difference on soiling index. Journ. APCA, Vol. 11, No. 3,
pp 118-119.
11. Neiburger, M. , 1959: Meteorological aspects of oxidation
type air pollution. The Atm. and the Sea in Motion, Rocke-
feller Inst. Press, N. Y.
12. Niemeyer, L. E., I960: Forecasting air pollution potential.
Mon. Wea. Rev. , Vol. 88, No. 3, pp 88-96.
13. Sutton, O. G. , 1932: A theory of eddy diffusion in the atmos-
phere. Proc. Roy. Soc. , London, A, Vol. 135, pp 143-165,
14. Sutton, O. G. , 1947: The problem of diffusion in the lower
atmosphere. Quart. Journ. Roy. Meteor. Soc. , Vol. 73,
pp 426-436.
15. Turner, D. B. , 1961: Relationships between 24-hr, mean air
quality measurements and meteorological factors in Nash-
ville, Tennessee. Journ. APCA, Vol. 11, No. 10, pp 483-
489.
16. Wanta, R. C. and Stern, A. C. , 1957: Classification of air
pollution exposures. Amer. Ind. Hyg. Assoc. Quarterly,
Vol. 18, No. 2, pp 156-160.
DISCUSSION
MR. SCHUENEMAN: I wonder if you have an estimate of
the additional time it might take a regular weather bureau station
operator to make pollution forecasts once he has run through the
exercises of developing and selecting parameters. Assuming
that he is making forecasts for other purposes, how much addi-
tional work would it be?
PROFESSOR DAVIS; As a rough estimate, I think it wouldn't
take him any more than an extra 45 or 60 minutes to make a spec-
ial air pollution forecast for a city like Philadelphia.
DR. SCHMIDT: I was very much interested in the last
..graph you showed giving the increase of air pollution as a func-
tion of time. I just tried to put it in another way by taking the
square root of time and I think that you found also that the increase
is proportional.
SEC TECHNICAL REPORT A62-5
-------
THE AIR OVER PHILADELPHIA 127
PROFESSOR DAVIS: I made a note of that in the paper you
gave, and I am sure that is a pretty close approximation.
DR. SCHMIDT: I wonder if you could briefly tell us how
many stations you used in gaining your pollutant levels for Phila-
delphia. Did you find any great variation statistically between
these stations? And were these only particulates, or did you also
include the gaseous material?
PROFESSOR DAVIS: We used only two stations in the
City of Philadelphia for the reason that over that 3-year period
only two stations were making continuous observations. The
measurements at other stations were sporadic, and we wanted
observations on a continuing basis. These were particulate
measurements made with the AISI and the high volume samplers.
DR. SCHMIDT: Was there a great difference between these
two stations? Did you reach a mean or an average, or just how
did you come to a particular level, a danger level of contaminants
in your city?
PROFESSOR DAVIS: We didn't establish a danger level. As
a basis for classifying the pollution level as high, we set a purely
arbitrary figure: when a reading on an instrument was 1-1/2
times the monthly average. We called this a critical value, but
one should not read into this any implication that this level is
dangerous to health. I have no way of knowing whether this is a
critical value in regard to health.
The samplers at both stations were very consistent, I think,
in the over-all readings.
MR. FULKS: I would like to hear something more specific
about the type of wind exposure you would like.
PROFESSOR DAVIS: Well, as a physicist, about the best
thing you can say is that you would like to get all the data you can.
If we could stick an anemometer at every corner, and 50 feet
above the surface, and all the way up to 500 feet we would like to
do this, but it is not feasible. At this point, I am not sure what
is required to really describe the wind flow in a city like Phila-
delphia.
FROM THE FLOOR: Two related values are the SO2 level
and the soiling index. When one exceeded the limit, did the other
exceed the limit? And you talk about 1-1/2 times the monthly
average - which month is that? If that occurs in the beginning
of the month, how do you determine what the monthly average is?
PROFESSOR DAVIS: With regard to the second question,
the data were furnished me by the City. The Air Pollution Con-
trol Section of the City simply averaged all the values from the
SYMPOSIUM: AIR OVER CITIES
-------
128 DAVIS
beginning to the end of the month, and any individual reading
that turned out to be 1-1/2 times that average was considered to
be a high pollution level..
Now, in regard to your question about the SO2 values and
the soiling index, these values also were given me by the Air
Pollution Control Section of the City of Philadelphia. They are
confident of their SC>2 levels. They didn't have measurements on
a parallel basis, but they assumed that the soiling index followed
the same pattern as the SO2-
MR. ROBINSON: I am concerned about the first statement
in your answer. In air pollution studies, I think that one thing we
come across is a great deal of variability, both in meteorological
and in air pollution measurements. This makes your use of at
most two stations questionable. I would like to see you expand a
bit on this idea of simple relationships and few measurements.
PROFESSOR DAVIS: Well, I think during the early part of
the paper I cited an equation such as those derived by Sutton as
the basis for the theory behind diffusion. Now, obviously simpli-
fying assumptions are made with regard to the microscopic con-
dition of the atmosphere and so on, but if you work with equations
such as Button's and others of similar nature it appears that wind
speed, wind direction, and air stability are all you need.
MR. ROBINSON: This is fine in theory, but when you bring
these together with your measurements, you don't come up with
a situation that makes you very satisfied with just a few observa-
tions.
PROFESSOR DAVIS: I thought that was the whole point of
my discussion: the theory indicates that only a relatively few ob-
servations are needed, but even these are lacking.
DR. NEIBERGER: I only want to comment that the theory
is based on very simplified assumptions, which are not ordinar-
ily met in nature, and perhaps this is the difference that you are
talking about.
PROFESSOR DAVIS: That's right. That's the only theory
we have, I think.
DR. GIFFORD: It is true that Button's method involves a
number of simplifying assumptions; but I think Sutton would be the
first person to point out, were he here, that he never intended
that the method that he presented should be applied to the distances
comparable with the size of a city, nor to deviations from the
smooth, gently rolling land, for which his method was first pre-
sented. The fact that it has been applied in these other ways in
no way reflects upon the quality of his contribution, which was
great.
SEC TECHNICAL REPORT A62-5
-------
THE AIR OVER PHILADELPHIA 129
Very recently there has been a method used in connection
with the Atomic Energy applications. That sort of air pollution
is restricted at present to isolated or groups of isolated sources.
This method has been put forward by many of us, particularly by
Mr. Pasquill in England. It avoids the simplifying assumptions
by essentially relating the theoretical problem directly to the
high-grade measurements of atmospheric dispersion that have
been made by some of the people in this room and others in the
last few years. This method is compatible with the sort of treat-
ment that Button presented. And I don't want to present an unin-
vited paper at this point, but I would like to say that I will have
available for distribution tomorrow, to any of those of you who are
interested, a little pamphlet published by the Atomic Energy
Commission and my office, the Oak Ridge Weather Bureau Office,
which lists the essential technical information that is needed in
order to use this method and gives something of the rationale.
This pamphlet tells you what observations the method is based on.
-------
The Thermal Climate of Cities*
J. MURRAY MITCHELL, Jr., Office of
Climatology, U. S. Weather Bureau,
Washington, D.C.
Summary
The urban "heat island" and other aspects of the anomalous temperature distribu-
tion in cities are discussed. Evidence is presented of the long-term intensification of
the heat islands of several U. S. cities, associated with their physical growth. Con-
tributory causes of the heat island are outlined, including local heating plants (winter),
insolational superheating (summer), and the influence of aerosol and gaseous pollutants
on the urban radiation budget. The influence of the heat island on other aspects of
urban climate, such as lapse rate, cloudiness, and turbulent mixing, is briefly con-
sidered.
THE URBAN "HEAT ISLAND"
Many familiar urban-rural contrasts of weather and cli-
mate are due more or less directly to the fact that cities are
warmer than their environs. This characteristic warmth of a
city, appropriately dubbed the urban "heat island, " is particu-
larly noticeable at night when skies are clear and winds are
light. Indeed the "heat island" is so characteristic of urban cli-
mate that local Weather Bureau forecasts must usually account
for it with such familiar wording as, "clear tonight, low temper-
ature 40° in the city and about 30° in the suburbs. . . "
That certain cities are warmer than their environs has
been known for a very long time. London's heat island was doc-
umented by Luke Howard as long ago, apparently, as 1818.
More than a century later, Vienna's was described in great de-
tail, among others by Wilhelm Schmidt, who in 1927 was the
first to use an automobile to obtain thermal cross-sections of a
city. Since then, the temperature distributions of many cities
have been sampled. The example of Washington, D. C. , has been
discussed by Landsberg. Undoubtedly the most exhaustive study
of urban temperatures yet published is that of Sundborg4' for
Uppsala, Sweden.
r>
Duckworth and Sandberg have recently surveyed the tem-
perature distribution of three Californian communities by auto-
mobile. Their study, the most comprehensive of its kind in the
United States to date, demonstrated that relatively small towns
have heat islands too. One survey, made on a clear spring night, is
*A revised version of this paper appeared in Weatherwise, vol. 14, no. 6, December 1961
131
-------
M
n
H
M
n
a
x
i— <
n
33
W
~
O
S3
HI I-1
C c/j
-a to
•a
o
a>
a
o
13
~
s
O
(W
Figure 1. Isotherm pattern for 2320 PST on 4 April 1952 superimposed on an aerial photograph of San Francisco.
C/3
p
cn
o
O
cn
~
O
Tl
n
cc
w
r
r
-------
THE THERMAL CLIMATE OF CITIES
133
Another survey by the same investigators is shown in Figure 2.
On these occasions, which were not particularly unusual, the
temperature of the densely built-up business district was some
20°F higher than the lowest observed suburban temperature, and
7 or 8 F higher than the average temperature of the urban per-
iphery.
.e
ttothtrm pattern for
PST, 26 March 1952, in
San Francisco
Figure 2.
IS THE HEAT ISLAND REALLY MAN-MADE?
Most cities were originally settled where they were be-
cause of desirable topographical features, such as a river ad-
vantageous to commerce or hills tactically valuable in their
defense. Such features are certain to produce microclimatic
anomalies. Someone, for example, who knows both San Fran-
cisco and a smattering of microclimatology, might well look at
Figure 1 and reasonably conclude that the same temperature pat-
tern would have existed on the eve of the California gold rush in
1848, when San Francisco was a mere village of 800 inhabitants.
After all, the highest temperatures are to be found on Nob Hill,
lying above the nocturnal inversion likely to spread over the rest
of the city in clear calm weather, and the lowest temperatures
are in Golden Gate Park, which at its minimal elevation would
be chilled the most under this same inversion.
SYMPOSIUM: AIR OVER CITIES
-------
134
MITCHELL
How, then, can we know that urban heat islands are not
topoclimatic anomalies that would exist, city or no city?
First, we can point to the fact that cities in entirely dif-
ferent topographic settings typically possess heat islands. Ex-
amples of heat islands in smaller cities, again measured by
Duckworth and Sandberg, are shown in Figures 3 and 4.
Isotherm pattern for 2200
PDT, 22 May 1952, in
San Jose, California
Figure 3.
Isotherm pattern for 2055
PST, 25 March 1952, in
Palo Alto, California
Figure 4.
SEC TECHNICAL REPORT A62-5
-------
THE THERMAL CLIMATE OF CITIES 135
Second, we can sometimes demonstrate that the heat island
of a city is weaker on Sundays, when the city is comparatively
dormant, than on other days of the week, when it is teeming with
people and bustling with activity. For example, a comparison of
daily maximum and minimum temperatures in the city of New
Haven and at its neighboring airport, averaged for four winter
seasons, is shown in Table 1. The relative warmth is seen to
have averaged only about half as much on Sunday as on any other
day. (This result is statistically significant at the 99 percent
level.)
TABLE 1
EXCESS WARMTH OF CITY COMPARED WITH AIRPORT (F°)
Data for New Haven, Conn. , in Four
Winter Seasons 1939 1943
Maximum Minimum Mean daily
Day of week temperature temperature temperature
Sunday 0.1 1.2 0.6
Monday 0.0 2.0 1.0
Tuesday 0.2 2.4 1.3
Wednesday 0.0 2.1 1.1
Thursday 0.0 2.3 1.2
Friday 0.0 2.4 1.1
Saturday 0.0 2.1 1.0
Third, it is frequently possible to show a gradual increase
of the intensity of an urban heat island as the city grows in size
and population over the years. ° Examples are Washington, D. C.
and Baltimore, Md., whose seasonal mean temperatures as
measured at their downtown Weather Bureau stations, are shown
in Figure 5 in the form of 10-year moving averages from 1894 to
1954, For comparison, the combined temperature series for a
large number of rural climatological stations, all within about 50
miles of these cities, is also shown in this figure. One can see
that, especially in the summer season, the cities have been grad-
ually warming up in relation to their rural surroundings. Cur-
iously, this relative warming has been almost entirely absent in
winter (see bottom of Figure 5).
The rates at which ten cities in the eastern half of the
United States have warmed relative to their environs since the
19th Century are compared for each season in Table 2. Here,
the cities have been ranked in order of growth rate, measured
by each city and its suburbs. Such a measure of city growth has
been used because it approximates the average length of travel
SYMPOSIUM: AIR OVER CITIES
-------
136
MITCHELL
1995-1904 1905-14 1915-24 1925-34
AVERAGING PERIOD
Figure 5. Ten-year moving averages of seasonal mean temperature in eastern Maryland,
shown as departures from 1945-54 average. Solid curve is average of data (or
17 rural cooperative stations; dashed curve is Baltimore city data; dotted curve
is Washington, D. C., city data.
over the city of air arriving at the Weather Station thermometer
and the total urban heating is assumed to be proportional to this
travel distance. Table 2 shows that the highest rates of urban
warming generally are associated with the most rapidly growing
cities. The relationship is close in summer, as shown in Figure6,
but not nearly so close in winter or for the year as a whole. A
different measure of city growth would apparently be more appro-
priate in the colder half of the year, when the urban heat island is
maintained by different kinds of heat sources than those prevail-
ing in summer. We will have a little more to say about these
presently.
SEC TECHNICAL REPORT A62-5
-------
THE THERMAL CLIMATE OF CITIES 137
TABLE 2
RATES OF URBAN WARMING OF SELECTED CITIES,
BY SEASONS
(Excess over Rural Environs)
(All values shown are significantly different from zero at 99 percent confidence level)
City
Cleveland, Ohio
Boston, Mass.
Washington, D. C.
Tampa, Fla,
Baltimore, Md.
Charlotte, N. C.
Rochester, N. Y.
Nashville, Tenn.
Lincoln, Nebr.
Marquette, Mich,
Period of
record a
1895-1941
1895-1933
1893-1954
1895-1931
1894-1954
1897-1951
1914-1940
1897-1948
1895-1954
1899-1954
Growth
rate
index
12. 9
10. 5
9. 9
6. 6
6.0
5. 1
4. 5
3. 0
2.4
0. 6
Urban warming rate (degrees F
per 100 yrs)
Winter
1. 8
2. 9
c
4. 8
c
c
c
c
c
c
Spring
3. 8
3. 2
1. 5
2. 6
2. 0
c
c
c
c
c
Summer
5. 1
4. 7
4. 3
2. 9
3.0
1. 7
c
c
2. 2
c
Autumn
2. 7
3. 7
2. 7
3.4
2. 7
c
c
c
c
c
a Limited to period of available city-office data only.
b Rate of change of root population (units per year).
c Not significantly different from zero.
6 r
A. YEAR
o 4
o
X
O
5
I
r = .59
B. SUMMER
A
A
A
A
r = .86
10 15 0 5
CITY GROWTH RATE
10
15
rigurc 6. Change of urban temperature with city growth: annual mean temperature on left,
ana summer-season mean on right. Cities are those listed in Table 2. "r" is
correlation coefficient.
SYMPOSIUM: AIR OVER CITIES
-------
138
MITCHELL
In an earlier study along similar lines, I compared the
net temperature changes between the 20-year periods 1900
through 1919 and 1920 through 1939, at a total of 77 cities through-
out the United States, with the corresponding changes of square
root of urban population. Regression lines were fitted to these
data in each of six geographical sections of the country. For the
annual mean temperature, the results are reproduced in Figure 7.
In all six zones the regression lines have a similar slope, an in-
dication that the more rapidly growing cities in each zone tended
to experience faster rates of warming than the slower-growing
cities. This result is quantitatively as well as qualitatively con-
sistent with the more direct measure of the effect of urban growth
in 10 cities shown in Figure 6.
MEAN TEMPERATURE CHANGE (DEC. F)
Ld
2 or
I
+3
20120
<« 12-MONTH ANNUAL AVERAGE
Figure 7. 20-yr changes of annual mean temperature at 77 selected cities in the United
States, as (unction of 20-yr changes in square root of their urban population, by
geographical zones. Computed regression lines are included. Data cover
period 1900 - 1939.
CITY-RURAL TEMPERATURE DIFFERENCES IN RELATION
TO TIME OF DAY
It has already been mentioned that the urban heat island is
most evident at night. Figure 8 shows comparative monthly av-
erage temperature in the center and at the edge of a city for each
hour of the day. The figure refers to Vienna, Austria, for which
SEC TECHNICAL REPORT A62-5
-------
THE THERMAL CLIMATE OF CITIES
139
hourly mean temperatures are readily available, but there is
no reason to doubt that data for American cities would show sim-
ilar comparisons.
JULY, 1956
— SCHOTTENSTIFT
(URBAN STATIONI
--- HOME WARTE
{SUBURBAN STATION)
FEBRUARY, 1956
12 2 4 6 8 10 12 2 4 6 8 10 12
A.M. NOON RM.
Figure 8. Diurnal variation of temperature in Vienna, Austria.
In July the relative warmth of the city is greatest at night,
and virtually absent during most of the daytime. In fact, during
mid-day in summer, it is not unusual for the city to be a fraction
of a degree cooler than its environs, owing presumably to the at-
tentuation of sunlight over the city by dust and smoke. The city
continues to hold its heat in the late afternoon, after tempera-
tures have begun to slide in the suburbs. Maximum daily temper-
atures in the city may thus be reached an hour or two after they
are reached in the suburbs. Minimum temperatures are also
reached somewhat later in the city, and the daily temperature
range is smaller.
In February (the coolest month of the year in Vienna) the
city tends to remain warmer than its suburbs both day and night.
Again, the daytime differences are smaller than the nighttime
ones. Daily maxima and minima of temperature are reached
slightly later in the city than in the suburbs, and the daily range
is less in the city.
SYMPOSIUM: AIR OVER CITIES
-------
140
MITCHELL
CITY-RURAL TEMPERATURE DIFFERENCES IN RELATION
TO TIME OF YEAR
Figure 9 shows two examples of the annual march of tem-
perature in and outside cities. Comparative monthly normals of
daily maximum, mean, and minimum temperature are included
in this figure. These examples are typical in showing the urban
heat island to be reflected most in minimum temperatures, and
least in maximum temperatures, throughout the year.
90
so
70
60
50
40
30
u 90
K
60
70
60
50
40
30
20
10
DENVER, COLO.
CITY STATION
AIRPORT STATION
BALTIMORE, MD.
A S 0 N
Figure 9. Annual variation of average monthly maximum, mean, and minimum temperatures
in and outside selected cities.
CONTRIBUTORY CAUSES OF THE HEAT ISLAND
Various physical factors are probably responsible in part
for the formation of urban heat islands. Some are more influen-
tial in summer, and others more so in winter.
On a summer day solar radiation is more or less readily
absorbed in the city by building and paving materials possessing
large heat-storage capacities. There is virtually no vegetation
or moist soil that can soak up some of this solar heat by evapor-
ation, as it does in the country. The temperature of streets and
SEC TECHNICAL REPORT A62-5.
-------
THE THERMAL CLIMATE OF CITIES 141
south-exposed building surfaces may soar well above 100°. Even
though air temperatures may be reasonably low, these hot sur-
faces irradiate us poor pedestrians, seemingly like so many
turkeys in a slow oven! To make matters worse, the buildings
obstruct ambient wind that might otherwise offer relief.
As night falls, the streets and buildings begin to cool down,
but very slowly. Their radiation continues all night, and their el-
evated temperatures prevent the air itself from cooling as much
as it otherwise would. By morning this heat has still not entirely
expended itself, and the next sizzling day begins with this therm-
al handicap.
On a winter day matters are a bit different - at least in our
relatively high latitudes. The solar rays may have appreciable
heat in them, but the streets and buildings are warmed perhaps
as much by the furnace heat of our office buildings and homes.
This heat seeps through their imperfectly insulated walls and
windows and disgorges through countless rooftop chimneys.
During the day the atmosphere is comparatively turbulent,
and the urban warmth is dissipated before street-level temper-
atures can be much affected. By nightfall, however, the-furnaces
begin to work harder against the cold, and the increasing slug-
gishness and stability of the atmosphere allow their heat to ac-
cumulate more readily in the city. In addition, the smoke, water
vapor, and carbon dioxide escaping from the chimneys spread
out in a "blanket" above roof-level that supplements the building
walls themselves in intercepting some of the heat radiation from
the streets and re-radiating part of their own heat downward.
Under these circumstances temperatures remain relatively high
in the city throughout the night, all the way from street level to
heights often considerably above the roof tops.
Surmounting this warm layer, a nocturnal inversion is com-
monly encountered. This inversion inhibits the upward dispersion
of smoke and other pollutants in the city. In fact, the inversion
may become intensified by the pollution layer whose upper surface
cools by radiation to the sky, just as in the country the ground it-
self cools by radiation. This, in turn, promotes the further in-
crease of pollution concentration in the city, until arrival of
morning, when the sun's heat can finally destroy the inversion.
EFFECT OF WEATHER ON THE HEAT ISLAND
In the preceding discussion, we have been presupposing
average weather conditions, characterized by relatively clear
skies and light winds. Since cloudiness and strong winds would
inhibit the formation of ground inversions, and since wind speed
SYMPOSIUM: AIR OVER CITIES
-------
142 MITCHELL
would also affect the rate of "ventilation" of the city, it is to be
expected that the intensity of the urban heat island would depend
substantially on these weather variables. Sundborg,4 who studied
the matter in some detail for Uppsala, found that the nighttime
urban-rural temperature difference, D, could be roughly approx-
imated by an equation of the form
D=
V
where N is percent cloud cover, V is wind speed, and a and-b are
constants. Undoubtedly such a formula could be used for cities
other than Uppsala; appropriate values of the constants would de-
pend on the city and on the locations of the wind speed and com-
parative temperature measurements. *
In this discussion it has been possible merely to sketch in
some of the salient characteristics of the urban heat island. Much
remains to be done in establishing the relative importance of the
various physical factors involved. Of particular interest to us
here, of course, is the role of atmospheric pollution — water
vapor and carbon dioxide as well as smoke— in altering the ra-
diative heat balance of cities. Unfortunately very few theoretical
studies of this subject have been made. We hope that this matter
will receive much more attention in the near future.
* Note that calm winds would have to be treated as a special case
case (in Uppsala, calms were rare). Sundborg also found that
temperature and humidity influence D somewhat. At night, how-
ever, these factors are negligible in comparison with wind and
cloudiness.
REFERENCES
Howard, L. : Climate of London Deduced from Meteorologic-
al Observations. Harvey and Darton, London (3rd Ed), 1833.
Schmidt, W. : Distribution of Minimum Temperature during
the Frost Night of May 12, 1927 within Vienna. Fortschritte
der Landwirtschaft, 2-21, 1929,681-686. (In German. )
Landsberg, H. E. : Comfortable Living Depends on Micro-
climate. Weatherwise, 3-1, 1950, 7-10.
Sundborg, A. : Climatological studies of the Special Regard
to the Temperature Conditions in the Urban Area. Geo-
graphica (Geographical Institute, University of Uppsala,)
22, 1951. (Ill pages.)
Sundborg, A. : Local Climatological Studies of Area. Tellus,
2-3, 1959, 222-232.
SEC TECHNICAL REPORT A62-5
-------
THE THERMAL CLIMATE OF CITIES 143
6. Duckworth, F. S., and J. S. Sandberg. The Effect of Cities
upon Horizontal and Vertical Temperature Gradients.
Bulletin American Meteorological Society, 35-5, 1954,
198-207.
7. Mitchell, J. M. , Jr.: On the Causes of Instrumentally Ob-
served Secular Temperature Trends. Journal of Meteor-
ology, 10-4, 1953, 244-261.
8. Mitchell, J. M., Jr. : The Measurement of Secular Temper-
ature Change in the Eastern United States. Research Paper
No. 43, U. S. Weather Bureau, Washington, 1961.
(80 pages.)
9. Steinhauser, F. , O. Eckel, and F. Sauberer. The Climate
and Bioclimate of Vienna (Part II). Osterreich Gesellschaft
fur Meteorologie, Vienna, 1957. (136 pages, in German.)
DISCUSSION
MR. MILLER: I have two points. One is the possibility
that Washington, D. C. is different from other cities as to the
limitation on vertical construction. I am not talking about the
growth of the government, but the buildings.
Second, I would like to have Dr. Mitchell comment on
what geographers are calling "strip cities, " such as the urban
complex between Cleveland and Pittsburgh, or between Washing-
ton and New York.
DR. MITCHELL: Well, the government in Washington is
certainly growing, but mostly it is growing in the suburbs. As
a matter of fact, looking at the horizon from 24th and M Streets,
where the long series of observations cited here were made, I
am impressed by the fact that it doesn't differ very much from
what it looked like in photographs 20, 30, or 40 years ago. I
think that in winter, at least, the intensity of the heat island,
from the point of view of a single spot in the city, is importantly
affected by what is happening right in the local area, not so much
by what is happening at the distance of the suburbs.
Another way of saying the same thing is that in winter, at
least, the heat island will grow in area with the growth of a city,
but will not intensify much in terms of its central superheating
value. But in summer apparently other things are involved --
what is going on in the suburbs does affect temperatures down-
wind in other parts of the city.
Now, concerning these "strip cities, " I don't really know
what can be said about them. Apparently, even a moderately
small amount of human development is enough to affect
SYMPOSIUM: AIR OVER CITIES
-------
144 MITCHELL
temperatures, and especially in winter it doesn't take very much
development to affect temperatures quite noticeably. From that
point on, further physical development will have a relatively
small effect on temperature. So I think that if I were to predict
how temperature conditions in strip cities and other newer rural
developments would change, I would expect to find initially a
rather rapid increase of nocturnal temperatures in these areas,
Then, as these communities continue to grow, in a peripheral
sense, the effect will probably become less pronounced.
MR. ROBINSON: I think somebody should speak from
Palo Alto.
I noticed something on the slides that I hadn't noticed be-
fore, and perhaps it is significant. The temperature differences
from San Francisco to Palo Alto were not great in the sense of
the built-up part of the city, each case being in order of 10 de-
grees. But the area covered by this high-temperature isotherm
was primarily, if my memory serves me rightly, the area in
which we have reinforced concrete buildings. In Palo Alto, this
is an area of a few square blocks; in San Francisco it is possibly
about a square mile, or maybe 2 square miles. The area for
San Jose would be in between. But if this is a factor, perhaps,
in the growth of cities, the development of the downtown area
and the growth of reinforced concrete construction is a signifi-
cant factor, and then, perhaps, the central portions of the city
have seen the maximum as far as this is concerned. I noticed,
that for Palo Alto there was a slight indication of an isotherm
even around the building complex of Stanford University. This,
again, is a stone and concrete group of buildings in an isolated,
really a rural setting. This might also be a factor of difference
to look for if careful observation should come about between an
eastern city with brick construction in the homes and residential
areas, compared to a western city with wood and stucco, that
retains much less heat.
OR. GIFFORD: One of the graphs you presented showed
the graduaTlnc-p-ease of the winter temperature in Washington.
I recall that in the climatology that I took some time ago, we
considered an over-all temperature increase for the whole world.
I wonder if you could comment oh the effect of this city heat is-
land phenomenon on the possibilities of detecting an over-all
worldwide temperature or of defining an overall worldwide
temperature? Since this increase is apparently an urbanization
effect, could we now know actually what the global temperature
is doing?
DR. MITCHELL: I think so. You may remember that I
SEC TECHNICAL REPORT A62-5
-------
THE THERMAL CLIMATE OF CITIES 145
showed a series of six regressions of the temperature trend
versus the population growth of the cities. (Figure 7) The
point at which these regression lines intersect the zero line
of city growth was not at zero temperature change. This is
one indication that there is a residual trend that is certainly
due to large-scale climatic variations. Taking records from
various nonurbanized locations we come to the same conclusion,
that definite temperature trends exist over and above what the
cities are contributing. The two effects are very real, and for
many purposes it is essential to separate the two; that's what I
tried to do here.
DR. GIFFORD: Are these trends of about the same mag-
nitude?
DR. MITCHELL: That depends on what time interval you
consider. They have been about equal since the turn of the cen-
tury, yes. This was only a coincidence.
Mr. STOUT: Mr. Changnon has some data on thermograph
records from the four stations around Champaign-Urbana, a
small city with a population of 7, 000. The interesting thing is
that we do have the cooling off at night, which you would expect
at the rural stations. But in the afternoon these- stations are
warmer than the downtown stations. So there is sort of a ring,
or a donut effect taking place, in the afternoon, anyway, that I
have no explanation for.
DR. HILST: Has any attempt been made to use these data
to determine the thermal energy excess in the heat island and
compare this with the energy consumption in the city?
DR. .MITCHELL: You mean to make direct calculations of
the amount of warming to be expected from local fuel consump-
tion? Did I understand you correctly?
DR. HILST: Just what do these numbers represent in the
way of excess thermal energy in the atmosphere?
DR. MITCHELL: They are quite consistent with estimates
of the heat produced by fuel combustion in the winter seasons. I
can cite a few studies, but they put a highlight only on this matter
of fuel consumption and not other heat-budget factors involved.
SYMPOSIUM: AIR OVER CITIES
-------
Some Observations of Cloud Initiation in
Industrial Areas
GLENN E. STOUT, Head, Meteorology
Section, Illinois State Water Survey,
Urbana, Illinois
Summary
A petro-chemical industry, located in the rural areas of central Illinois, has been
observed to initiate clouds. At least five cases are partially documented by surface,
airborne, or radar observations. On one occasion a cumulus congestus cloud formed and
produced tornado funnels. No other clouds were present in the area. Other cases are
discussed and the available data presented. Preliminary occasional observations suggest
that this industrial source could be used for an interesting research study of the possible
influence of industrial pollution on cloud formations.
Various investigators have suggested that pollution from in-
dustrial sources enhances precipitation. To the best of my knowl-
edge, no one has proposed that pollution decreases rainfall down-
wind from the source. For example, Landsberg and Changnon,
in earlier papers of this symposium, have suggested that precip-
itation over an urban area is greater than over the rural areas.
To test a hypothesis that nuclei from specific industrial sources
might increase precipitation in the atmosphere, the staff at Gen-
eral Electric Company, Falconer and Schaefer, •"• studied rainfall
patterns downwind from lead-smelting plants in the West. Re-
sults were inconclusive, since the lack of sufficient rain gages in
the area prevented a comprehensive study.
More recently, F. W. Van Straten,2 of the U.S. Navy,
proposed that carbon black would produce clouds under certain
conditions and would enhance the development of clouds. Tests
were made by spreading carbon black in warm maritime air. Al-
though there was some suggestion that clouds could be produced,
the program has not been fully tested.
Telford, ^ of the Commonwealth Scientific and Industrial
Research Organization in Australia, recently concluded that a
portion of the freezing nuclei in the atmosphere comes from cer-
tain types of industrial activity. The smoke from a steel factory
was identified as a prolific .source. He also concluded that other
industries probably have little influence on the freezing nuclei
count. It is quite possible that some may produce inhibiting agents.
Studies of rainfall in Illinois by Huff4 have suggested that
the greatest frequency of heavy rainfalls occurred east of St.
Louis, Missouri. The annual mean rainfall at Edwardsville,
147
-------
148
STOUT
Illinois, has increased by 10 percent in the last two decades. A
detailed climatological study has shown a higher incidence of
fronts in this region. Therefore, the increase may be due to pol-
lution or frontal stagnation.
The rainfall patterns east of Chicago have been of consider-
able interest to us. For example, LaPorte, Indiana, shows an
average annual rainfall of 50 inches, while nearby stations re-
port 36 inches. Correspondence with the State Climatologist of
Indiana has not produced a satisfactory explanation of this great
rainfall anomaly.
The large concentration of steel mills around Gary could
very well support the high incidence of rainfall at the LaPorte
station. Figure 1 shows the annual precipitation for LaPorte,
LA PORTE
VALPARAISO
SOUTH BEND
WHITING
1900 1905 1910 1915 1920 1925 1930 1935 1940 1945 1950 1955 I960
—YEARS —
Figure 1. Annual precipitation values determined by averaging 5-year moving totals for
selected Indiana stations
Valparaiso, South Bend, and Whiting, Indiana. The increase in
precipitation beginning about 1922 is quite evident at the three
stations in operation at that time. The continuing increase in the
SEC TECHNICAL REPORT A62-5
-------
CLOUD INITIATION IN INDUSTRIAL AREAS
149
annual precipitation at LaPorte is quite striking. Either the ex-
posure of the gage or the pollution from the steel mills has con-
tributed to the anomaly. The LaPorte observer convinced the
author that there is an unusual wintertime snowfall belt in a small
area around LaPorte. The heavy snowfall arrives with north-
westerly flow, which may be due to lake effect. The pronounced
increase in rainfall in summer is difficult to explain. Figure 2
shows the annual steel production in the Chicago and Gary indus-
trial complex. The great increase in annual rainfall after 1933
100 xlO7
I00xl0s
100x10*
I
in
Ł100x10*
H
Ł 100X10'
I
lOOxlO2
1910 1915 1920 1925 1930 1935 1940 1945 1950 1955 I960
- YEARS-
Figure 2. Annual steel production in tons, 1910-1960
coincides with the greater increase in steel production. Exact
figures for the Gary steel manufacturing complex were not avail-
able. We have also looked at the rainfall patterns downwind from
other industrial areas such as Peoria, Illinois, and find no anom-
alies of any significance. There are no steel furnaces in the
Peoria area.
Johnson reported that a cumulus cloud developed over a
prairie grass fire. The radiosonde data did not indicate any con-
vective activity possible. He thought that the water vapor for the
cloud over the fire came from evaporation of the moisture from
the green grass, from the'top layers of the ground, and also from
the combustion of the grass. Hoddinott ^ reported observing a
funnel cloud that was due to an industrial activity. He reported
that the area of Chester, England, has been a beehive of indus-
trial activity since 1955 and that new cumulonimbus developments
were often observed downwind from the steel works of Chester.
SYMPOSIUM: AIR OVER CITIES
-------
150 STOUT
The flat prairie land of east central Illinois was trans-
formed from an agricultural area by the construction of a large
industrial activity just west of Tuscola, Illinois, in about 1951.
The industrial complex is now valued at $130 million and covers
approximately a square mile. Five hundred thousand cubic feet
of natural gas is used daily in the extraction of various elements
including ethane, which is the main raw material used in making
petrochemical products. The plant also houses the largest unit
in the world for making industrial alcohol.
On December 24, 1954, the Midwest area was under a large,
high-pressure system of continental origin. At around 1600, I
observed from our laboratory a low stratus-type cloud, the only
cloud in the sky, directly south of the airport. There was also
one small cumulus turret evident. I drove 20 miles south to the
source of the cloud and observed that the stratus was actually a
dark smoke cloud below an inversion of 1700 feet. The white
cumulus turret was located near the power plant stacks and ex-
tended several hundred feet above the inversion. Fallout was oc-
curring about three miles from the stack. Upon investigation of
possible contaminations in the air, I learned unofficially that there
had been a burnoff of alcohol prior to my first cloud observations.
On March 1, 1960, clouds of unusual origin were again ob-
served near Tuscola and at Decatur, approximately 30 miles due
west. A pilot reported that these clouds were the only ones pres-
ent. Clouds formed near the top of the inversion layer and were
approximately 200 feet in depth. Below the base of the clouds,
industrial smoke was clearly visible and seemed to be feeding into
the cloud formation.
In May 1960, Dr. B. Vonnegut from Arthur D. Little, Inc.,
and I observed a low-hanging white cloud in the area of the indus-
trial activity in the late evening. A squall line had passed the
area in the late "afternoon. The industrial cloud extended 15 to 20
miles from the source and was intense enough for radar detection.
The top of the cloud as determined by radar was about 6000 feet.
No rainfall measurements are available to verify a precipitation
mechanism within the storm.
Our most striking observation occurred on May 27, 1960.
As we drove to work at 0700 CST, we observed one large cumulus
cloud about 20 miles south of the airport. There were no other
clouds present within visible range. We immediately dispatched
an aircraft, which at that time was outfitted to collect samples of
the particulates in the atmosphere, to investigate. As the ob-
server approached the area, he called back to report a funnel
hanging from the base of the cloud. In the meantime, radar ob-
servations indicated the cloud top was 20, 000 feet. A few minutes
SEC TECHNICAL REPORT A62-5
-------
CLOUD INITIATION IN INDUSTRIAL AREAS 151
later, the state police called to report funnels from this isolated
cloud. They observed five funnels during the 75-minute period
of observation. The particle concentration within the cloud was
of sufficient quantity and size that again the radar was able to
detect it. To the best of our knowledge there was no precipita-
tion from the cloud. This observation appears to be similar to
Hoddinott's in Chester, England.
On October 12, 1961, clear skies prevailed over East
Central Illinois except for intense smoke layers over the cities
and the Petro Chemical Company area, where inversion around
2500 feet produced a dense smoke layer extending westward from
Tuscola. At the same time, a lower inversion with a top at 100
feet trapped a white hygroscopic smoke, which appeared to be a
fog layer. This layer extended northeastward from the Tuscola
area for approximately 20 miles.
Two types of data should be explored for further study of
this phenomenon. First, a radar climatological study of the ex-
istence of echoes in and downwind of industrial areas should be
undertaken. Secondly, the use of TIROS cloud data would help to
support some of these findings.
Wilk 7 of the Illinois State Water Survey has been using well-
calibrated quantitized radar information to determine the areas of
greatest radar reflectivity during hailstorms. During several of
his case studies, he has noticed that the strongest echoes from
precipitation masses occur over industrial areas. On other occa-
sions he has observed new echo developments ahead of squall
lines downwind of industrial areas such as St. Louis, Missouri.
No quantitative comprehensive study has been undertaken to date,
but the significance of the data suggests that such work should be
performed.
Satellite information also provides a new source of data.
Inspection of TIROS photos has on several occasions shown in-
creased cloud activity just east of Gary, Indiana. Maps of rain-
fall showed localized rainfall in the area during the approximate
time ot the satellite passage. Therefore, a climatological study
of aerial cloud photos in and around major industrial areas with
and without other influences, such as large bodies of water, would
prove most interesting.
I have tried to illustrate that industrial complexes produce
clouds and enhance precipitation. Conversely, there is always a
chance that pollution will have no effect on the precipitation proc-
ess. Several techniques have been suggested for future data col-
lection and research on the influence of industrial activities on
precipitation.
SYMPOSIUM: AIR OVER CITIES
-------
152 STOUT
REFERENCES
1. Falconer, R. E. andV.J. Schaefer, Final Report No. RL-
1007, ONR Project, Contract No. NONR-925(00), Decem-
ber, 1953.
2. Van Straten, F. W. , "Preliminary Experiments using Carbon
Black for Cloud Modification and Formation, " NRL Report
5235, October 28, 1958.
3. Telford, J. W. , "Freezing Nuclei from Industrial Processes,"
Journal of Meteorology, Vol. 17, No. 6, December 1960,
676-679.
4. Stout, G.E. and F. A. Huff, "Studies of Hydrometeorological
Factors Influencing Severe Rainstorms on Small Water-
sheds. " Presented at Tenth Hydraulics Division Conference
of ASCE in Champaign-Urbana, Illinois, August 16-18, 1961.
5. Johnson, Oliver, "The Development of a Cumulus Cloud over
a Prairie Grass Fire. " Weather, London, 14(6), June
1959, 212-215.
6. Hoddinott, M.H.O. , "Funnel Cloud at Chester, July 28, 1959."
Meteorological Magazine, London, 89(1053), April 1960,
124-125.
7. Wilk, Kenneth E. , "Radar Investigations of Illinois Hail-
storms. " Scientific Report No. 1, Contract No. AF 19(604)-
4940, January 15, 1961.
DISCUSSION
DR. NEIBERGER: Just one or two comments about some
experiments that Dr. Dessens in France has been conducting on
the French side of the Pyrenees. He has been conducting what he
calls a meteorotone, which consists of a square of oil burners
about 100 meters square. I have forgotten the amount of oil that
is consumed and the corresponding amount of energy produced,
but in his experiments under appropriate conditions the smoke
column, or the column of heat, develops convective activity to the
extent that cumulus clouds are formed. In two or three cases he
has had funneled clouds developed which had quite intense rota-
tions, corresponding to small tornados.
MR. KALSTROM: May I ask in what direction these funnel
clouds were in relation to the plant and the heat sources -- how
far downwind? Or were they quite close to the energy?
MR. STOUT: They were downwind and about 5 miles away.
But the cloud kept moving away, so that at the end of an hour the
cloud was about 25 miles from the source. It was a slow-moving
cloud, but it did migrate away from the source.
SEC TECHNICAL REPORT A62-5
-------
CLOUD INITIATION IN INDUSTRIAL AREAS 153
MR. KALSTROM: The funnels occurred some distance
away from the heat ?
MR. STOUT: Yes, that's right.
FROM THE FLOOR: I might mention a rather well-docu-
mented case, I believe in the 1920's. An oil fire was reported,
where tornados actually occurred.
SYMPOSIUM: AIR OVER CITIES
GPO 825111—6
-------
Session 2: The Dispersion and Deposition of Air Potiutants Over Cities
M. NEIBERGER,* Chairman
Summary
In setting limits for the control of pollution sources in industrial and urban com-
plexes, limits must not be established solely on the basis of individual stacks and
plants; the basic concept must be the area-source strength, in terms of total emissions
per square mile, since the emissions from separate stacks and plants are additive as the
air moves across them toward residential and commercial communities. The dispersion of
pollutants from vehicle exhausts along congested streets and roads deserves attention.
Such questions as the influence of heat from motors and the motion of the vehicles need
examination. Studies of dispersion have generally assumed flat uniform terrain and wind
conditions in which the direction is steady and the average speed is high compared to
turbulent fluctuations. Studies must be performed on the dispersion of pollutants over an
irregular complex of buildings and under the influence of wind conditions in which the
average speed is low and the magnitude of windspeeds in fluctuations is as great or
greater.
In introducing this session on the dispersion and deposition
of atmospheric pollution over cities, I want to point out the ques-
tions I regard as most important in the consideration of the dis-
persion of air pollution over cities. For the most part I shall
leave the answers to the other members of the panel insofar as
answers are known, and to further research insofar as they are
not. It is my hope that in the course of this morning's session
the foundation for this afternoon's program on present and future
needs for observational data will be laid. At the same time, the
needs for future theoretical studies will be developed.
In the past almost all studies of the dispersion of pollution
have dealt with point sources, and line sources have been con-
sidered occasionally. The problem of area sources, and partic-
ularly the effective evaluation of how the concentrations due to
self- contamination within large source areas depend on the var-
ious contributing factors, has begun to be investigated only
recently.
We can get a simple indication of this dependence by mak-
ing the assumption, as is frequently done, that the flux, F, of a
contaminant of concentration C in any direction is proportional
to its gradient in that direction.
Here Fz is the flux in the vertical of a contaminant of con-
centration C, and Az is the coefficient of proportionality, ordi-
narily called the coefficient of turbulent exchange, or sometimes
the eddy diffusivity.
'Chairman, Dept. of Meteorohogy, University of California at Los Angeles
155
-------
156 NEIBERGER
Fx and Fy are similar expressions for the flux and in x and
y directions.
As is familiar to all of you, consideration of the conver-
gence of this flux leads to the expression for the rate of change of
the concentration with time given in equation 2.
— — ( Ax ) + f Ay — ) + -— I Az T— ) (2)
at a x V a * / dy \ dy J dz \ dz /
For an individual point source, integration of this equation
leads under simplifying assumptions to the formulas of Roberts,
Bosanquet and Pearson, and Button.
In a general way their formulas show the concentration at
the ground downwind of a large elevated source in the form shown
in Figure 1. The position of the large elevated source is taken at
the origin of the coordinate axis; the maximum ground concentra-
tion is some distance downwind because of the time that is re-
quired for the diffusion to carry the pollutant down to the ground,
and the further diffusion spreads it out so that beyond this point
along the plume axis you have a more or less exponential decrease.
1000 1500 2000 2500 3000 3500
DISTANCE DOWNWIND ALONG PLUME AXIS, meters
Figure 1. Variation with distance of surface concentrations from elevated source
(schematic).
Now, if you had initially a background of pollution on which '
the contamination due to this source is superimposed, you would
SEC TECHNICAL REPORT A62-5
-------
DISPERSION AND DEPOSITION OF POLLUTANTS
157
have the situation shown in Figure 2. Here the background con-
centration is shown by the dashed line and the total concentration
is shown by the same curve as in Figure 1 displaced upward. Far
enough downwind of the point source, the effect is simply to in-
crease the background concentration slightly.
TOTAL CONCENTRATION
BACKGROUND CONCENTRATION
1 23456
DISTANCE DOWNWIND, kilometer,
Figure 2. Variation with distance of total concentrations at surface downwind from an
elevated source (schematic).
DISTANCE DOWNWIND, kllomitni
Figure 3. Variation with distance of total concentrations at surface downwind from a
number of elevated sources located along a line parallel to the m»on wind
(schematic).
SYMPOSIUM: AIR OVER CITIES
-------
158 NEIBERGER
Now, if you visualize a series of such point sources all in
a line along the wind you would get the conditions shown in
Figure 3. The first source adds to the background concentration
its typical hump, which falls off and leaves the background slight-
ly higher, and then the next one does the same thing. Thus there
is a gradual increase in the background.
In Figure 3 I have assumed schematically that all of the
sources are very large, each producing a big hump and then de-
clining in the course of over a kilometer, and that all of them are
located exactly along the wind direction. This, of course, in gen-
eral would not be true. If you had random sized and randomly
spaced sources spread over an area, the effects along the axis
would look like that shown in Figure 4. In this case one after
another of the sources would contribute at random intervals and
with random intensities. The background concentration goes up
as a result of the successive decay of each of these sources. The
plume axis of the first one source is assumed to be the line along
which the concentration is computed. Occasionally, this line will
strike a source which is approximately on the same plume axis,
but in general the sources will be some distance from it.
DISTANCE DOWNWIND, kllom.f.r.
Figure 4. Variation with distance of surface concentrations from randomly sized and
spaced elevated sources (schematic).
To get a quantitative estimate of the increase in background
concentration due to an area source of this sort, we must express
the intensity of the source in terms of a smooth functionE(x, y, t),
The boundary conditions for the solution of the equation then would
SEC TECHNICAL REPORT A62-5
-------
DISPERSION AND DEPOSITION OF POLLUTANTS 159
be that at the ground, i. e. , at z = 0, Fz , the upward flux, is just
equal to the source function. (For the moment we assume that
the sources are all at the ground. ) We further assume that at
sufficiently great heights the concentration goes to zero, and at
the point where the air first begins to cross the source area the
concentration is zero. These boundary conditions,
Fz = E ( x, y, t) at z 0
C = 0 at z = -° (3)
C = 0 at x = 0 .
are adequate to determine a solution if AX, Ay, and Az are known.
In general, Ax, Ay, AZ are functions of z, because of the
variation of the lapse rate with height and the variation of the size
of turbulent eddies with height. They are in general also func-
tions of x and of y, because of variations of the lapse rate in the
horizontal and because of the variation in the roughness due to
buildings and other irregularities of terrain.
To solve the diffusion equation under these conditions is a
very difficult problem when you have to allow for the variations
of all these quantities, and even then, as I mentioned, this is a
simple and not a rigorous way of approaching atmospheric
diffusion.
For a first crude approximation, we shall assume that AZ
is constant and equal to A, and that the variation of C in the x
direction, taken to be the wind direction, and the variation in the
y direction are much smaller than the variation in the z direction.
In that case the equation reduces to equation 4.
dC
Fz E, a constant, for x > 0 . Fz = 0 for x < 0 .
With these boundary conditions, assuming also that E is
constant and not dependant on x, y, or t, we can get an analytic
expression for the solution, which is given by equation 5.
2Vt c
WA
At the ground, z equals zero, the concentration reduces to the
very simple expression given in equation 6. At Z - 0,
C =_Ł_ŁlVJi (6)
SYMPOSIUM: AIR OVER CITIES
-------
160 NEIBKRGER
This equation is similar to the one that Dr. Schmidt showed
yesterday.
If, furthermore, we concern ourselves with an air column
which is moving across the area with a constant velocity., U, in
the x direction, then t = x/U, and we have the expression shown
in equations 5a and 6a.
Z2U
c_ E ) 2Vx 4Ax _Z_ rf /_&J3/U\f(5a)
\^, ——— t ^ t- j. j. v_, 1 - - t /
VAT/)
(6a)
Equation 6 and 6a give some indication of the way that you
would expect the background concentration due to an area source
to change as the air moves across it. It varies with the square
root of the time that the air moves along its trajectory over the
source, or, if the wind is constant over the area source, then it
varies as the square root of the distance along the path of the
column as it moves across its source.
For a numerical example, let us assume that the average
emission of hydrocarbons from automobile traffic in Los Angeles
is 2 tons per square mile per day and is evenly distributed. This
corresponds to a value of E of approximately 10"9 gram per
square centimeter per second.
If the wind speed is 1 meter per second and you express the
concentration in parts per million by mass, which is commonly
called in meteorology the mixing ratio, the proportion by mass at
various distances is shown in Table 1. The concentration is
TABLE 1
HYDROCARBON CONCENTRATIONS (ppm by mass)
AT WINDSPEED OF 1 METER PER SECOND
y Distance Downwind, km
Diffusivity
(A)
5- 102
103
5- 103
104
5-104
SEC TECHNICAL REPORT A62-5
0. 1
0. 41
0.29
0. 13
0. 09
0. 04
0. 9
1.2
0. 86
0. 38
0. 27
0. 12
2. 5
2.0
1. 4
0. 64
0. 45
0.20
6.4
3.2
2. 3
1.0
0. 72
0. 32
10.0
4. 1
2. 9
1. 3
0. 91
0. 41
20.0
4.9
3.4
1. 5
1. 1
0. 49
40.0
8.0
5.7
2.5
1.8
0.80
-------
DISPERSION AND DEPOSITION OF POLLUTANTS 161
expressed in parts per million, for various values of A given in
the first column. The distances downwind from the start of the
source area are given in kilometers.
Values of A are really not well known, but in a general way
the values near the top of the table correspond to fairly strong
inversions, or moderate inversions, and values near the bottom
correspond to lapse conditions.
Taking any one value of A, it is seen that the concentration
increases downwind as the air passes the sources, as would be
expected. On the assumptions made, this variation is exactly as
the square root of the distance, so that the rate of increase of
concentration tapers off as the air passes over more and more of
the source area.
Looking now at the effect of the various coefficients of dif-
fusion, we see that the concentration is greatest for the slowest
diffusion, corresponding to the inversion, and is least for the
greatest diffusivity, corresponding to a relatively large lapse
condition.
The computation thus gives a clear indication of the factors
controlling the concentrations of pollutants, namely source in-
tensity, windspeed, and the intensity of turbulence, represented
by the A, which in turn is controlled by roughness of terrain, at-
mosheric stability, windspeed, and windshear.
I have presented this discussion not as an answer to the dif-
fusion problems but rather to focus attention on the elements that
need investigation.
As long ago as 15 years, when I was first consulted on the
air pollution problem, I pointed out the urgent need for a census
of pollution sources to give their locations and intensity. The
Air Pollution Control District of Los Angeles County has carried
out such censuses to some extent.
Figure 5 shows a map made some years ago of pollution
sources over the Los Angeles basin, as an indication of the type
of data that are needed. The chemical and process industries
are very dense in the section called Vernon, southeast of central
Los Angeles, and are sparse in residential areas.
Some of the municipal incinerators shown on the map have
actually been closed down because they didn't meet the rules of
the Air Pollution Control District. This map is based on data
from 1952 for the most part.
The isopleths of traffic density also will have changed con-
siderably from the 1951-52 data represented here. The lines are
SYMPOSIUM: AIR OVER CITIES
-------
NEIBERGER
Figure 5. Map of traffic density and industrial sources of pollution in Los Angeles County.
labelled in terms of the percentage of the county total per square
mile. The one labelled 0. 1, for instance, has one-tenth of a per-
cent of the county total per square mile. In central Los Angeles
it reaches as much as 45/100ths of a percent per square mile.
If one takes this fraction per square mile, multiplies it by the
total amount of gasoline consumed in Los Angeles County in a
day, and then by the fraction that comes out in the exhaust un-
burned or partially burned, one can get the amount of hydro-
carbons lost from automobile exhausts per square mile. So we
have a pattern of isopleths which can be interpreted as the emis-
sion function, E, as a function of x and y, expressed in terms of
tons per square mile, or whatever unit you choose.
If, corresponding to each of these circles and triangles, we
had similar information about the source strength for each par-
ticular pollutant, we could draw isopleths for them also and have
a function E for other pollutants, for example SO^, or whatever
other pollutants one is interested in. This is the kind of map that
I feel is required to come from censuses of the amount of pollu-
tion put into the atmosphere. This map was prepared on a whole-
day basis, but. since traffic varies with time of day hourly maps
of this sort should be plotted also.
SEC TECHNICAL REPORT A62-5
-------
DISPERSION AND DEPOSITION OF POLLUTANTS
163
To evaluate the trajectory of the air as it crosses the
source area, it is necessary to know not only how the emission
intensity varies over an urban area but also how the wind varies
over time and space. Figures 6-10 represent the flow pattern
over Los Angeles during a particular 24-hour period in 1954,
starting at the time of the maximum sea breeze, 2:30 p. m. ,
September 21, and ending at the same time September 22.
In Figure 6, you see that the air is streaming across the
basin from the coast to the mountains at the time of maximum sea
breeze. The solid lines with arrows are streamlines and the
dashed lines are isotachs, or velocity lines in miles per hour. At
the coast the speed is 16 miles per hour, but farther inland it
drops off to 8 miles per hour.
Figure 6. Streamlines (solid lines) and isotachs (lines of equal speed, dashed lines) for
1430 PST September 21, 1954. Speeds are given in miles per hour beside sta-
tion arrows and at ends of isotachs.
Figure 7 shows the pattern 8 hours later. By 10:30 p.m.
the land breeze has already sprung up, the air is flowing out from
the mountains and across the coast, with speeds of less than 4
miles per hour over most of the basin, and there is even a line,
enclosing an area for which the winds are less than 2 miles per
hour. Thus the winds have become very light and from the land
on the whole.
SYMPOSIUM: AIR OVER CITIES
-------
lb-1
NEIBERGER
Figure 7. Streamlines (solid lines) and isotachs (lines of equal speed, dashed lines) for
2230 PST September 21, 1954. Speeds are given in miles per hour beside sta-
tion arrows and at ends of isotachs.
Figure 8, for 6:30 a. m. shows the land breeze at roughly
its maximum, with a 6-mile-an-hour wind right at the coast, but
over most of the area the winds are 2 miles an hour or less, very
light winds carrying the air back from the inland area toward the
coast and then off-shore.
At 10;30 a. m. , Figure 9, the Seabreeze has already started.
One interesting thing, at 2:30 p. m. September 21 the Seabreeze
was all coining from the west coast of the basin, but during the
early part of the Seabreeze there is a line of convergence, with
part of the Seabreeze coming from the south coast. At 10:30 the
speeds range from 4 to 10 miles per hour, rather less than at
2:30 p.m. Figure 10 shows the situation at 2:30 p.m. September
22, when the Seabreeze is again near its maximum.
This series of charts shows how in a complicated topogra-
phy, with light winds such as we have in Los Angeles in the warm
half of year, there is quite a variation of winds with distance in-
land. In computing diffusion in such terrain one cannot assume a
wind in a uniform direction with constant speed.
Figure 11 shows a trajectory computed from hourly maps
such as the ones in the preceding figures. This one is the trajec-
tory reaching Pasadena at noon on the 23rd of September, 1954.
SEC TECHNICAL REPORT A62-5
-------
DISPERSION AND DEPOSITION OF POLLUTANTS
165
" ,.,... '
I' I'' ! 1 "•—)->
K»1I II H>LI>
Figure 8. Streamlines (solid lines) and isotachs (lines of equal speed, dashed lines) for
0630 PST September 22, 1954. Speeds are given in miles per hour beside sta-
tion arrows and at ends of isotachs.
Figure 9. Streamlines (solid lines) and isotachs (lines of equal speed, dashed lines) for
1030 PST September 22, 1954. Speeds are given in miles per hour beside sta-
tion arrows and at ends oi isotachs.
SYMPOSIUM: AIR OVER CITIES
-------
NEIBERGER
Figure 10. Streamlines (solid lines) and isotachs (lines of equal speed) dashed lines) for
1430 PST September 22, 1954. Speeds are given in miles per hour beside sta-
tion arrows and at ends of isotachs.
The large circle is Pasadena, and the smaller circles show the
positions of the air at the hours given by the numbers beside
them. The air went in across the coast with the beginning of the
Seabreeze on September 22, streamed inland, reached a point not
far from Pasadena that evening, meandered southwestward slowly
during the night, started northeastward again with the beginning
of the seabreeze the next day and reached Pasadena 23 hours after
it left the coast. A little later the results of a computation of the
variation of hydrocarbons and carbon monoxides along this tra-
jectory will be presented. To evaluate the concentrations along
the trajectory we should know, in addition to these windflow pat-
terns, the variation of coefficient diffusivity, A, in the horizontal
and in the vertical.
Figure 12 shows a temperature sounding typical of the warm
half of the year near the coast in Los Angeles. This is on a ro-
tated tephigram, so that the constant temperature lines (labelled
in °C) slope upward to the right and the dry adiabatic or constant
potential temperature lines (labelled in °K) slope upward to the
left. The numbers on the left of the sounding curve are heights
in meters, on the right water vapor mixing ratio in parts per
thousand. The temperature decreases with height for the first
SEC TECHNICAL REPORT A62-5
-------
DISPERSION AND DEPOSITION OF POLLUTANTS
167
18
/I6
15
2I/
/ ,/l3 Pasadena
23.. .22,' r< .. 19
23-9-54
TRAJECTORY OF AIR REACHING PASADENA
1200 PST 23 SEPTEMBER 1954
MEASURED VALUES AT PASADENA - Eye irritation 28 (on scale 0-55)
Oxidant; 30 pphm
Hydrocarbons; 31 pphm
Carbon Monoxide. 620 pphm
Ald«hydet: 29 pphm
Figure 11. Trajectory of air arriving at Pasadena at noon September 23, 1954 with high
values of oxidant and eye irritation.
SYMPOSIUM: AIR OVER CITIES
-------
168
NEIBERGER
few hundred meters are practically at the adiabatic rate, and then
there is an inversion of temperature of a considerable thickness,
above which the temperature decreases again.
Thus, instead of a constant value of A, corresponding to
some particular lapse rate, it varies with height. For the com-
putation that I will show, which is an unrealistic one also, I have
assumed that A is practically infinite for the layer of adiabatic
lapse rate, and extremely small beginning where the inversion
starts. The whole computation is intended to be schematic.
Figure 12 shows a typical sounding near the coast, but
actually even there it varies with time of day, and as the air
moves inland during the day there is variation with position. For
an accurate computation, then, one should account for the varia-
tion of A not only with height, but also with time and with distance
600
700
800
900
IOOO —
Figure 12. Typical sounding at southern California coast in summer.
SEC TECHNICAL REPORT A62-5
-------
DISPERSION AND DEPOSITION OF POLLUTANTS
169
in the horizontal. Rigorous account of all these factors would be
too difficult to carry out analytically, but might be attempted by
high-speed electronic digital computers, or perhaps as the next
paper will discuss, by analog computers.
The results of the computation of the hydrocarbon and car-
bon monoxide concentrations for the trajectory that we saw in
Figure 11, using the distribution of E corresponding to the traffic
density pattern shown in 1952, Figure 5, are shown in Figure 13.
We assume a parcel of air starting at the coast containing no hy-
drocarbons and no carbon monoxide, and compute the amount that
is added due to the traffic density as it follows the trajectory
shown in Figure 11. For each segment of the trajectory the
source strength is assumed proportional'to the corresponding
traffic density line, and the diffusion into the air parcel below the
inversion is computed. The computed hydrocarbon value for the
point where the trajectory reaches Pasadena at noon on the 23rd
Sept. 22
Figure 13. Estimated accumulation of hydrocarbons and carbon monoxide from motor traffic
along trajectory of Figure 11.
is about 20 pphm, the observed value measured was 31 parts per
hundred million, so that the computation gives a value of the same
general magnitude, although not very close.
For the carbon monoxide, the computed value is about 700,
and the measured value was 620. Thus even with the very crude
method of evaluation the computation comes not very far from the
SYMPOSIUM: AIR OVER CITIES
-------
170
NEIBERGER
measured values, on the assumption that the measured values are
due to the addition of the hydrocarbons and carbon monoxide from
automobile traffic without any other sources and without any de-
struction of these contaminants. Actually, we know that there
are other sources, and there may also have been a background
value at the start, since the air may have moved out from the
land to sea the previous day and then moved in again. Thus there
are ways of explaining the fact that the hydrocarbons are too low
on this basis. Similarly, the fact that the computed value of the
carbon monoxide is too high can be explained, since presumably
some of the carbon monoxide was oxidized during the passage
across the basin. But also, of course, there are many other
deficiencies in the computations and one should not take the cor-
respondence too seriously.
Even this crude theory gives some promise that if the wind
and temperature conditions are forecast accurately and if we have
some way of deriving from them the values of A, we could make
fair forecasts of the concentrations of pollutants for which the
area emission strength is known. As theory improves and data
for verification become available, the accuracy of forecasts of
concentration will improve correspondingly.
The principle of area emission intensities as opposed to in-
dividual source strengths applies not only to the forecasting prob-
lem, but also to the problem of zoning urban areas with respect
to industrial developments and traffic patterns. The criterion
should be to limit the increases in sources to the maximum area
emission intensity that will not produce undesirable concentra-
tions. This intensity for an area may be produced by individual
large sources or by many small ones.
The zoning restrictions should recognize this fact. To ar-
rive at a criterion for the acceptable area emission intensity, as
for accurate forecasting, it is necessary to know much more
about the process of diffusion over urban areas. Recognizing
that the approach through statistical studies of turbulent diffusion
may turn out to be the best way to get the answer, we may still
couch the question in the terms of the exchange coefficient A. The
question then would be how is A affected by the terrain of complex
urban areas, which consist of tall and low buildings, trees, hills,
heat sources from residences and industrial processes, and mov-
ing vehicles. How does it vary with the wind, especially for low
average wind speeds, in which condition the fluctuations have as
large magnitude as the average? How does it vary with lapse
rate and how does it vary with height under all these circumstances?
When the variation of A is known or the equivalent clarification of
the turbulent diffusion process is obtained by other means, it will
be necessary to apply it to the distribution of E for each urban
SEC TECHNICAL REPORT A62-5
-------
DISPERSION AND DEPOSITION OF POLLUTANTS 171
community under models typical of adverse weather conditions to
estimate the resulting concentrations of pollutants. The integra-
tion of the diffusion equation or its equivalent will probably be too
difficult to carry out analytically, as I have already mentioned.
High-speed computers will have to be used. But when we have
adequate theory and computing techniques and the knowledge of
the area emission functions it will be possible to compute the
effects of prolonged adverse weather conditions not yet experi-
enced or the effect of changing the emission distribution by the
addition of new sources or the development of new commercial,
industrial, and residential areas.
SYMPOSIUM: AIR OVER CITIES
-------
Analog Computing Techniques
Applied to Atmospheric Diffusion:
Continuous Area Source*
FRED V. BROCK, Meteorological
Laboratories, The University of Michigan,
Ann Arbor
Summary
An electronic analog computer has been used to obtain solutions to the diffusion
equation. The model is that of a continuous area source located on the ground in steady-
state conditions. The crosswind-integrated concentration is obtained as a continuous
function of distance downwind for discrete height intervals. The versatility of analog
simulation is demonstrated by introducing a variety of boundary conditions and other para-
meters into the basic model. With this approach the effects of an inversion, radioactive
decay, gravitational settling, ground reflection, ground absorption, etc. are conveniently
included. Windspeed and eddy diffusivity can be arbitrarily varied with height. Analog
simulation may be extended to treat the problem of photochemical reactions that occur
during the diffusion process.
The purpose of this paper is to show how the electronic an-
alog computer may be used to simulate atmospheric diffusion.
The example chosen to illustrate this technique is that of diffusion
of particulate matter from an area source-on the ground in steady-
state conditions. No specific physical problem is envisioned here
but, as the solutions are obtained in nondimensional parameters,
they could apply to a wide variety of situations. For example they
could be applied to the diffusion of industrial effluents from a
group of stacks, of automobile exhaust fumes from a city, or, at
the other end of the scale, of pollen from a cluster of plants. The
only restriction on scale is one implied by the choice of a wind
profile, a power law in this case.
In this problem the coordinate axes are oriented with the
x-axis downwind, the y-axis crosswind, and the z-axis vertical.
The source, as shown in Figure 1, is circular and centered at the
point (a/2 .0.0). The mean wind vector is assumed to have no
components in the y or z directions. The source strength is as-
sumed to be constant over the source area.
* Paper No. 58 from the Meteorological Laboratories, Department of Engineering
Mechanics, The University of Michigan. Publication No. 30 on Atmospheric Pollution
by Aeroallergens. Grant No. Al C6 from the National Institute of Allergy and Infec-
tious Diseases, Public Health Servi ce.
173
-------
174
BROCK
Figure 1. Orientation of the coordinate axis relative to the source and to the mean wind
direction. The center of the circular source is located at the point (a/2, 0, 0)
in the x, y, z space.
THE MATHEMATICAL-PHYSICAL MODEL
The diffusion equation as used here may be stated as
f 1* XX (1)
u(z) AA JL[K(Z) -1*1+ _L
3x az |_ dzj dy
<9y
with the following boundary conditions:
X — 0 as x —0 for all y, z
and x, |y| , z -» ?°
lim
z ^C
J^ + fxl
a z J
where
X concentration
u mean wind speed in the x- direction
K eddy diffusivity
f = fall speed of the particulate
X = decay rate, e. g. radioactive decay
in consistent units.
The source strength is a constant Q over the region defined by the
relation
y.2 + (x a/2)2 a2/4; z 0.
SEC TECHNICAL REPORT A62-5
-------
ANALOG COMPUTING TECHNIQUES 175
The rate of loss of material to the ground is assumed to be pro-
portional to the "local" concentration near the ground with the
proportionality constant p. The statement of the boundary condi-
tion at the ground was adopted from the work of Calder.
If the mean wind speed and eddy diffusivity profiles are
given by
rz i 1/7 -• 6/7
u(z) U, - , K(z) K,
then equation (1) may be written in nondimensional form as
follows:
ld^ -_ z- >/7 d \z 6" as] + F z '/7 _^ _ D z- '" S (2)
c ax az \_ szj az
where S is the nondimensional crosswind-integrated concentra-
tion defined as
1 ' Xdy.
Xo ^
The constant Xo will be defined later. The other nondimensional
parameters are
X Cu,- 2 l~l
Z — z
zo
- 6/7
F *o TO | f
„ .„., 6/7
lT-| P
D -
The boundary conditions for equation (2) are
S — 0 as X — 0 for all Z
and as X, Z — °°
lim
z-0
L 6/7 iŁ + (F P) sl = -B T!AX x1! [i - h(x A)]
SYMPOSIUM: AIR OVER CITIES
-------
176 BROCK
where
2 C O u •
Xo
B K,
All of the constants involved can be grouped into two types.
The first type, scale parameters, includes ui, KI, z\, and z0.
These determine the conversion from nondimensional units to
physical units and, therefore, the scale of the problem. The sec-
ond type may be called convenience parameters and have no phys-
ical significance. These parameters, B and C, are used solely to
scale the problem on the analog computer.
PREPARATION FOR ANALOG COMPUTATION
Equation (2) is stated in terms of two independent variables,
X and Z. Since the analog computer can integrate with respect to
only one independent variable, one of them, Z, must be replaced
with finite differences. Computer time is used to represent X.
Implementation of finite differences requires a new model, which
is shown in Figure 2. The atmosphere is divided into 10 layers
in the vertical, and the thickness of the layer increases exponen-
tially with height. Each layer incorporates the properties of the
atmosphere that it replaces. For convenience, the layers are
labeled 1 through 10. Thus the notation Sn(X) represents the
crosswind-integrated concentration in the nth layer as a function
of distance downwind.
The general finite difference expression for stations 1
through 9 is
1 ds
C dX
Z(n)' Z(n + l/2)6/7 , „
(n)'1/7 ("
Z(n) [_A
A Z(n) [_A Z(n + 1/2) n+i n
Z(n l/2)6/7
AZ(n I72T V n n- i
F Z(n)-1/7
2 A Z(n)
DZ(n)-'"Sn. (3)
The expression for station 1, which is next to the ground, is dif-
ferent because of the boundary conditions that must be inserted.
The boundary condition in finite difference form is
SEC TECHNICAL REPORT A62-5
-------
ANALOG COMPUTING TECHNIQUES
177
, 6/7
(Si S0) + (F P) Sc
(4)
10
n
Q S
9
8
7.5
7
6
K R
5
AG.
4
3
i
2.996
Z
1.897
1.386
1.204
1.050
0.7984
0.5977
0.4305
0.2876
8.1524
0.05163
-
-
-
-
ds UQ\XO x •
Figure 2. Finite difference model showing the vertical coordinate axis relations.
From this, an expression can be found for So, which is a virtual
station below ground. The meaning of this is that in the finite
difference model, a virtual station below ground is given a con-
centration such that the boundary condition for flux at the ground
is satisfied. The boundary condition at Z — °° is satisfied by
setting SIQ 0 which is at infinity due to the exponential cell
spacing.
SYMPOSIUM: AIR OVER CITIES
-------
178
BROCK
ANALOG MODEL
The physical model now consists of a stack of planes spaced
exponentially in the z direction. It is supposed that each plane
has a capacity for containing some of the diffusing substance just
equal to that of the layer of atmosphere that it replaces. At a
given point in x, a traverse in the z direction shows the properties
of the atmosphere lumped at discrete intervals. This model can
be simulated with an electric analog, where each of the circuit
components is analogous to some property of the atmosphere. A
segment of such a circuit is shown in Figure 3. The analogies
shown were developed from the similarity of the node differential
equations for a simple case,
C dV
CndT
J - 2 V + V )
n+1 n n-i
(Az) u
n dx
_
(Az)
n+l
2X + X )
n n-1
To perform the simulation, set the resistor Rn proportional to
Az/k, the capacitor Cn proportional to Az Q, and voltage V pro-
portional to concentration. The problem time t will represent
distance downstream. All these relations are stated as propor-
tionalities so that scale factors can be used to provide reasonable
^WV
V,
-A/W-^A/W
-VVVjAA/V
•AA/V
(AZ)n/K
\n
TIME=x
dx "
Figure 3. Typical station in the passive network analog showing the analogy between
circuit components and the physical properties of the atmosphere pertinent to
the diffusion problem.
SEC TECHNICAL REPORT A62-5
-------
ANALOG COMPUTING TECHNIQUES
179
values of resistance and capacitance. In general, one may relate
electrical resistance to atmospheric resistance to diffusion, and
electrical capacitance to the capacity of a layer of the atmosphere
to hold the diffusing substance. The complete network analog is
shown in Figure 4, along with the electronic analog and a repre-
sentation of the physical model.
-s0
Electronic Analog
Circuit
Finite Difference
Physical Model
Figure 4. Electronic analog, passive network analog, and physical, finite difference
model of the atmosphere. Note that each group of components represents a
layer of the atmosphere.
SYMPOSIUM: AIR OVER CITIES
-------
180
BROCK
The passive-circuit analog was not used in this study, but it
is helpful to construct it on paper because of the insight it provides.
This is particularly useful in determining the method of applying
the boundary conditions.
The electronic analog circuit could be designed directly from
the passive network or from the set of simultaneous differential
equations. The latter method is much the easier course after the
equations are reduced to the form
dS
dX
+ cn
The principal element in the electronic analog is the integrating
amplifier, which can be used to represent one station as shown in
Figure 5. In the passive-network analog, there was a direct
rGH
-s,
O.I or I M
HIGH-GAIN, d-c
AMPLIFIER
—Q— = COEFFICIENT POTENTIOMETER
Figure 5. Typical station in the active or electronic analog circuit. The concentrator! is
represented as a voltage, and equation coefficients as potentiometer settings.
SEC TECHNICAL REPORT A62-5
-------
ANALOG COMPUTING TECHNIQUES 181
correspondence between the circuit elements and the physical
problem, while in the electronic analog, there exists a one-to-one
correspondence between the circuit elements and the defining
equation. The coefficients a, b, and c are set on the correspond-
ingly designated coefficient potentiometers. The integrating am-
plifier sums the three inputs, integrates the sum, and inverts the
sign. Ordinarily, the gain of the integrating amplifier is unity,
set by the 1-megohm input resistor and 1-microfarad feedback
capacitor. If one of the coefficients is greater than unity, the
corresponding input resistor used is 0. 1 megohm, which provides
a gain of 10. This is necessary because the maximum setting of
the coefficient potentiometers is 1.
The electronic circuit is much more flexible than the pas-
sive network, permitting almost any conceivable form of boundary
conditions to be set up readily. In this case, a voltage is gener-
ated and fed into the network at So such that equation (4) is satis-
fied.
The accuracy of the solutions obtained with this method can
be demonstrated by analogy. When the problem is set up on a
computer, it is a physical system that is designed to behave like
the atmosphere with respect to diffusion. This same model has
been tested previously. ^ With a different input arrangement it
simulated diffusion from an infinite line source. The solutions
obtained were compared with available analytical solutions "
where the errors were found to be less than 5 percent.
RESULTS OF COMPUTATION
In Figures 6 through 10 the solutions have been plotted as
lines of constant, crosswind-integrated concentration for cases I
through V, respectively. The following table shows what effects
were used in each case.
TABLE 1
LIST OF SOLUTIONS OBTAINED AND THE VALUES OF
THE PARAMETERS FOR TIME DECAY D,
GRAVITATIONAL SETTLING F, GROUND SINK EFFECT P,
AND THE PRESENCE OF AN INVERSION I
Solution
Parameter
D
F
P
I
SYMPOSIUM: AIR OVER CITIES
I
0
0
0
NO
II
0
0
0
YES
III
0
0
2
NO
IV
0
2
2
NO
V
2
0
0
NO
-------
182
BROCK
A typical set of values will give an idea of the magnitude of
these parameters. If we choose zo = zi 10 cm, KI 10 cm /sec,
the other parameters become
f 20 cm/sec
X 0.002 sec-'
To express X in terms of half-life, this becomes about 6 hours.
0.2 -
0
12
24
28
32
16 20
Distance Downwind, X
Figure 6. Lines of constant, crosswind-integrated concentration in the X, Z plane.
D = F = P = 0. No inversion.
EXTENSION OF THE METHOD
Analog computation can be used profitably in solution of
partial differential equations with up to three independent vari-
ables. In diffusion problems these will be time and two space
variables or three space variables. In the case, say, of three
space variables, x, y, and z, one of them, usually x, will be rep-
resented continuously by computer time. The other two will be
represented by finite differences.
Arbitrary variation of any parameter as a function of any of
the independent variables is readily accomplished. The function
SEC TECHNICAL REPORT A62-5
-------
ANALOG COMPUTING TECHNIQUES
1.4
183
12 16 20 24 28 32 36
Distance Downwind, X
Figure 7. Lines of constant, crosswind-integrated concentration in the X, Z plane.
D = F = P = 0. Inversion at Z = 1.897.
0.2 -
0
8
28
32
12 16 20 24
Distance Downwind. X
Figure 8. Lines of constant, crosswind-integrated concentration in the X, Z plane.
D = F = 0. P = 2. No inversion.
SYMPOSIUM: AIR OVER CITIES
-------
BROCK
12 16 20 24
Distance Downwind, X
28
32
Figure 9. Lines of constant, crosswind-integrated concentration in the X, Z plane.
D = 0. F = P = 2. No inversion
0.2
0
28
32
12 16 20 24
Distance Downwind, X
Figure 10. Lines of constant, crosswind-integrated concentration in the X, Z plane.
D = 2. F = P = 0. No inversion.
SEC TECHNICAL REPORT A62-5
-------
ANALOG COMPUTING TECHNIQUES 185
may be in analytic, graphical, or tabular form. Boundary condi-
tions of almost any desired form can be used, including multiple
sources, ground reflotation of particulates, etc.
Not so readily accomplished, but still possible, is computa-
tion of the diffusion of the products of chemical reactions or radio-
active decay in the atmosphere. It is possible to compute simul-
taneously the diffusion of the original substance and of the products
of reactions.
The most severe restriction on this method is that the
amount of equipment required increases with the complexity of
the problem, so that each analog computer installation will have
its own limit.
REFERENCES
1. Brock, Fred V. , 1961. Analog computing techniques applied
to atmospheric diffusion: continuous line source. Tech-
nical Report No. 2 on National Science Foundation Grant
G-11404. The University of Michigan, Ann Arbor, Michigan.
2. Calder, Kenneth L. , 1961. Atmospheric diffusion of partic-
ulate material, considered as a boundary value problem.
Journal of Meteorology. 18(3):413-416.
3. Button, O. G. , 1953. Micrometeorology. McGraw-Hill Book
Company, New York
The author wishes to acknowledge his grateful appreciation
to Professor E. Wendell Hewson for his consistent interest and
constructive advice. The wholehearted cooperation and assistance
of Mrs. Anne C. Rivette is also appreciated.
DISCUSSION
MR. GIFFORD: Just to throw your last remarks into a
little bolder relief, would you tell us, considering that you showed
really a tremendous amount of computations, five or six complete
fields for essentially different diffusion problems, about how long
it took to program the machinery and to actually run off these
solutions ?
MR. BROCK: The question is a little difficult to answer
because it happens that the problem is similar to a previous one
that I did, but I think that programing the machine itself takes
very little time. Designing the particular circuit would take
SYMPOSIUM: AIR OVER CITIES
-------
186
BROCK
perhaps an hour. It would take several hours to compute the co-
efficients for the various parts in the problem. It took me longer
actually to decide how the boundary condition ought to be set up
than to do all the rest of the work combined.
Running the problem doesn't take very long, because we
set it up so that the particular solution is run out in about 15
minutes, and it is plotted out in graphical form for us. I then
replotted in the form that you saw to make it a little more com-
pact. So this work took me about two weeks, most of which was
spent worrying about how the boundary condition ought to be set
up.
MR. GIFFORD: It seems to me that in connection with
urban pollution problem and area sources, there are essentially
two separate problems, of which you have treated one. In the one
problem you have an area source, and this would correspond
somewhat to the megalopolis problem, where you are interested
maybe in the next state or the next town, that is, what is coming
from a large area upwind.
For the other problem you really don't care very much
about what happens outside of town but you are extremely con-
cerned about what goes on within the area source. And I wondered
if you had given any thought to the simulation of this second prob-
lem.
MR. BROCK: Essentially there is a difference in scale,
although when you deal with what is happening immediately within
the source, so to speak, you have to be more careful about the
simulation of the source. I think that could probably cause more
trouble than anything else. It is reasonable to assume that the
source is constant over a circular area, when you get far enough
away from it. Up close you would probably have a series of point
sources, and the problem then it to simulate these. If you have a
field of point sources you must produce this electrically, and in-
evitably it would come down to a question of resolution. There is
a limit to how fine a resolution one can get in analog computation.
It is usually determined by the amount of equipment. However,
it is like treating the problem of multiple sources, where you
have a scattering of sources. I think that we could reasonably do
something with this.
MR. GIFFORD: I might remark that we have just begun to
think about this as a result of the very interesting visit that we
had from Mr. Brock at Oak Ridge. We can't get it on the com-
puter until December. But as a first approach, what we did was
to apply a mathematical coordinate. This is really what I wanted
to ask you about.
SEC TECHNICAL REPORT A62-5
GPO 825111-7
-------
ANALOG COMPUTING TECHNIQUES 187
To get the detail in the city, we are going to change coor-
dinates from X and Y to R and S, where R 1 - E~x, and
S 1 E~y. That means that when x 0, say, than R 0, and
when x °° , R 1, and when x - 1 it turns out that R = ~ 0. 8.
That means that if you apply this transformation then you can put
most of your equipment inside the city and not worry too much
about what goes on in the suburbs or in the next town. We don't
have any results, but if you have any words of wisdom for us, I
would be glad to hear them. Do you think we will get anywhere
with that kind of approach?
MR. BROCK: I am sure that you will. The computers are
of immense flexibility. You really can do almost anything that
you want to do, if you persevere long enough.
This is an excellent transformation. Of course, the one
that I used to reduce the z axis from infinity down to something
finite that I could handle, and turning it around and applying it to
X or Y is very much the same thing. You simply explode your
scale then, which can be done very easily.
I would be very much interested in seeing the results on
this, because I think you do have a good approach.
MR. LARSEN: Do you believe that a time-sharing of analog
computer components would be a feasible method of cutting down
on the amount of equipment needed for a three-dimensional simu-
lation of air pollution over a city?
MR. BROCK: We have been considering this problem re-
cently. There have been two innovations in the analog art re-
cently. One of them is time sharing -- well, both of them are
time sharing in a sense, and they have very earnest proponents.
Now, so far as I know, and I have been in contact with some of
the leaders in this field, this has not yet resulted in anything that
can be brought to bear on our problem. It has been used in other
problems, which can be described as something essentially like
the diffusion equation. But in other things, such as fluid flowing
through a pipe, for example, this time-sharing business has
worked. The leaders in the field as of August of this year, and
by leaders I mean specifically Professor Howe of the University
of Michigan and some people from the West Coast, have agreed
that at the present time this time-sharing could not be applied to
the problems we have in meteorology. The problem seems to be
that when you time share you introduce more errors, and as you
do more of this to save on equipment, errors multiply and grow
to the point that the computation becomes useless. I don't want
to go on record as saying that this won't work in the future, but
as of the present it does not seem to.
SYMPOSIUM: AIR OVER CITIES
-------
188 BROCK
MR. LARSEN: So if you wish to simulate pollution., say,
over Cincinnati and from all points and from multiple sources
you would still require hundreds of analog components to simulate
a whole town at one time.
MR. BROCK: It depends on the detail you want. If you want
to go into all the gory detail at once, then yes --it would take an
awful lot of equipment. But by no means would it completely
solve the problem of handling partial diffusion by analog
computers.
SEC TECHNICAL REPORT A62-5
-------
Dispersion Calculations for
Multiple Sources
FRANCIS POOLER, Jr., Weather Bureau
Research Station, Laboratory of Engineering
and Physical Sciences, R. A. Taft Sanitary
Engineering Center, Cincinnati, Ohio
Summa ry
Several ways in which multiple sources may be collectively described are ex-
amined, together with appropriate diffusion formulation. Since the main objective in
developing models to calculate concentrations is to eliminate the necessity of direct
air quality measurements, u pragmatic approach is recommended, both to simplify the
required calculations and to incorporate diverse observations of pollutant behavior into
the models used. Some possible uses of the calculations are discussed, and necessary
further observations suggested.
Calculation of pollutant concentrations from the numerous
sources in urban areas is a difficult task. We probably will never
be able, or even be asked, to derive models capable of explaining
all possible variations of concentrations over an area, because no
one has any use for such detailed information. What are desired,
and likely will be for some time to come, are values of calculated
concentration that show details comparable to those that can be
observed. The apparent objective in making calculations is thus,
essentially, to be able to calculate values of concentration that
match measured values. The ultimate objective is to develop
means of calculating concentrations to supplant direct measure-
ments.
This general objective can be divided into at least five cat-
egories, corresponding to the uses that might be made of calcula-
tions. The categories that occur to me are:
1. To estimate concentrations for particular times and places,
for correlation with some known effect.
2. To estimate the contributions to the total pollutant loading
from particular classes of sources: from different source areas,
or from different kinds of sources within the same area.
3. To delineate the spatial distribution of concentrations.
4. To estimate the variability of concentrations for averaging
times less than the sampling interval of the instrument used.
5. To estimate extreme concentrations that could occur at a
given location.
As a general statement of the calculation problem, we are
trying to determine the concentration at any point in space and
189
-------
190 POOLER
time. The concentration is made up of the contributions from a
field of sources, with this source field a function of the space
and time coordinates. The contribution of concentration from any
point in this source field is the product of the field strength and a
spatial probability distribution function, which itself varies with
time. Symbolically,
X(x,y, z, t) = / / / / Q(x,y, z, t)D (t)
7xyyyzvt xyz •
where O is the source strength of a point in the 4-dimensional
field of space (x, y, z) and time (t) and D is a spatial distribution
function, also dependent on time. The total concentration at a
point is obtained by integrating over the 4-dimensional field. It
may be noted that the source field is to be integrated, and not the
concentration field.
In any real case we might be dealing with, we will never be
able to specify either Q or D precisely and totally. We are forced,
by practical necessity, to approximate the sources by a collective
source description, and to approximate the probability density
fields by a collective probability density. In other words, we are
attempting to describe a highly complicated matter by making
some simplifying assumptions, in order to derive mathematical
analogs that can be used to estimate concentrations. In any analogy,
comparison is made only between limited parts of the two subjects
compared; in an analogy of multiple-source dispersion, one must
choose the portion of the problem that is to be most closely de-
scribed by means of the analog. The portion to be described is
determined primarily by the purpose in making the calculations.
In other words, there is no all-purpose model that can be used to
satisfy all the uses to which multiple-source dispersion calcula-
tions might be put. The particular model used will depend on three
factors: (1) the use to be made of the results; (2) the source data
available; and (3) the meteorological data available.
The models that can be employed fall into three general
classes, which can be named according to the source description
employed. These classes I shall call the cloud-source model, the
grid-point-source model, and the area-source model. In the
cloud-source model, the emissions from groups of sources are
followed forward in time, and the concentration from any source
grouping can be calculated for as many points as desired. In
equation form,
xp (t)=ZJ/ Q- (*)D._,_ (t) .
The source grouping can be considered as an emitted cloud per
unit time, the cloud having some definable initial probability
SEC TECHNICAL REPORT A62-5
-------
DISPERSION CALCULATIONS: MULTIPLE SOURCES 191
density. The subsequent travel and spread of the instantaneous
cloud can be approximately described in terms of any available
meteorological measurements, so that it is not necessary to as-
sume spatial or temporal uniformity in either the mean flow or
the turbulence processes. The amount of calculation necessary
with this kind of model increases in direct proportion to the num-
ber of source clouds used to describe the sources, since generally
both the initial cloud distributions and the subsequent mean mo-
tions and diffusion rates will be different for each cloud. When it
is obviously necessary to consider a varying field of motion, which
usually will be the case when pollution must be followed for at
least an appreciable fraction of a day, this kind of model is prob-
ably required. However, if such long travel times must be con-
sidered, the exact details of the real source geometry become
rather unimportant, so that it will generally be possible to keep
the source clouds limited to a manageable number. Dr. Frenkiel
used this kind of approach with, I believe, a digital computer to
estimate concentrations in the Los Angeles basin using clouds for
his model. This same general approach can, of course, be used
with a limited number of individual sources, in which case the
contribution from each source to a point can be calculated individ-
ually by a conventional formula.
In the other two kinds of models, both temporal and hori-
zontal spatial uniformity of motions are assumed, so that the rel-
ative dispersion from source to receptor doesn't depend on the
actual location of either, but only on the relative locations. In
the grid-point-source model, a series of sources of identical
geometry are used to represent the real source, so that the spa-
tial variations of source strength can be simulated to an accuracy
determined by the spacing of the grid-points. The concentration
at a point is then determined as the sum of the contributions from
each source point, or
X --
p —u yz
The most practical use of this kind of model is to calculate
concentration patterns for an area, and usually it will be desirable
to obtain as much detail in the spatial variation of concentration
as is available from the source description. Thus it is easiest to
use this model to calculate values for a grid of points with the
same spacing as the source grid. A grid of D values can be cal-
culated, the products of Q and D at each grid point obtained, and
these products added to calculate the concentration for one point.
The grid of D values must then be shifted relative to the source
grid, and the arithmetic repeated, to obtain concentrations for
each point of the grid. Thus, the amount of simple arithmetic
SYMPOSIUM: AIR OVER CITIES
-------
192 POOLER
necessary is proportional to the square of the number of grid
points. The easiest, cheapest, and most accurate way to do such
arithmetic is with the aid of a digital computer.
A somewhat simpler source grid can be devised if concen-
trations at only a small number of points are desired, by spacing
the grid points in such a way that the relative contribution from
each source point to the receptor point is about the same; this
can be done by spacing the points so that their density is an in-
verse function of distance from the receptor point.
The third model is the area source: the concentration at a
point is determined as the integrated contribution from the up-
wind source area, or
X - / — D (x)
P ~~ J u z
x
The assumption is made in such a formulation that the
source area is sufficiently extensive in the crosswind dimension
that it can be treated the same as an infinite crosswind source,
As Mr. Brock showed, you could also make the distribution func-
tion two-dimensional rather than one-dimensional as shown. In
view of the fact that there has been quite a bit of coverage of area
sources by Mr. Brock and Dr. Neiberger, and by Dr. Schmidt
yesterday, there is no need to go into any further detail to show
the use that might be made of such calculations.
When it comes to setting down some exact expression for a
model, three steps are required. The first is a visualization of
the entire dispersion process that is to be described. This re-
quires a good imagination, for we will rarely be able to actually
see more than small glimpses of the real situation, and we must
fill in the greater part of the total picture with imaginary plumes,
telescoping time scales, and other devices.
Some attention must also be given to the source inventory
that will be used, for the way in which the sources are to be col-
lectively described is partially dependent on the way the disper-
sion is to be treated, and the dispersion will very obviously de-
pend on the geometry of the sources. It will be necessary to con-
sider what purpose the calculations are to serve, what kind of
detail will be available from the source data, and what sorts of
meteorological data will be available. From these considerations,
we must settle on the type and general form of the model to be
used. The source data can then be adapted to the model.
The second step is to break down the total dispersion proc-
ess into smaller components that can each be described by means
SEC TECHNICAL REPORT A62-5
-------
DISPERSION CALCULATIONS: MULTIPLE SOURCES 193
of some mathematical expression. For this step, I think the best
thing to do is look out the window, or sit on top of Mount Royal,
perhaps to watch the auto-correlation of velocity, or the growth
and decay of eddies, or whatever it is that one chooses for de-
scribing diffusion. This step requires the ability to be a good ob-
server. I think both the first and second steps, the visualization
and the observation, require the capability of a meteorologist, and
therefore I wouldn't trust a model not made up by a meteorologist.
The third step is to use the meteorological measurements
that may be available to indicate the dispersion that has been ob-
served and described. I think a great deal of confusion can be avoided
if we recognize that there is no such thing as a definitive set of
meteorological measurements that will describe dispersion for
even the simplest of circumstances. We must therefore use avail-
able meteorological data as indices of dispersion conditions, and
not regard them as diffusion measurements. If we choose to use a
meteorological parameter as a continuous index, rather than as one
element of a discrete classification system, it will be possible to
end up with some extremely odd dimensional units for numerical
coefficients, even though the original form of an equation is dimen-
sionally logical. This step requires a very careful judgement, and
probably few observers would choose identical values to describe
the same process.
Even with rather scant meteorological data to work with, it
will be possible to devise some elaborate formulations that pro-
vide a quantity of doubtful detail. Since these details cannot be
checked, and may not be valid to start with, it will often be pos-
sible to simplify a formulation without any real loss of accuracy.
It will often be instructive to plot curves of calculated concentra-
tion versus distance from a source, using linear scales, to see
what the influence of varying parameters in a formulation may be.
In this way, it may turn out that a large error factor really doesn't
matter, because the erroneous contribution is only a small
fraction of the total from all sources. The degree to which a form-
ulation is simplified will thus depend in part on the formulation
itself and on the use to be made of it.
Since the form of an equation as used will depend on the
means available for making the calculations, I don't believe there
is much point to setting down specific formulae. With an area-
source model, it is easiest to set up some expression that can be
integrated analytically or electronically; with a grid-point source
model, quite unwieldy functions can be used, since the expression
need be numerically evaluated only for discrete points;and with the
volume-source model, dispersion by the mean flow may mask some
of the smaller-scale diffusion processes, so that mass continuity of
the mean flow will be a primary requirement of a formulation.
SYMPOSIUM: AIR OVER CITIES
-------
194
POOLER
Multiple-source dispersion calculations at present involve
too much guesswork and arbitrary extrapolation to be regarded
as having a scientific merit, and therefore I don't think they
should be judged by scientific standards. Improvements will come
about by narrowing the margin of uncertainty in our estimates of
dispersion parameters and by increasing our knowledge of the
relationships between dispersion and available measurements.
The greatest area of uncertainty is in the area of generally great-
est significance in determining pollutant concentrations in urban
areas: we have very little data concerning the vertical distribution
of mean flow and turbulence, or of pollutant stratification. We
also lack a physical basis for tying such data together, i. e., our
theories are incomplete. We know little concerning the ultimate
disposition of pollution; we can't very well say where something
is going unless we first know what it is, and this requires atten-
tion to the physical and chemical forms in which the pollutants
exist in the atmosphere.
In spite of these inadequacies, I think we can, at present,
make calculations for limited situations that are sufficiently ac-
curate to serve a useful purpose. If such results justify our in-
exact methods we can always incorporate new findings and more
complete theories into the methods, and thereby extend the utility
of such calculations.
DISCUSSION
CHAIRMAN NEIBERGER: I have a feeling it might help
some of the people here if you will describe how one would go
about getting the D function.
MR. POOLER: Well, this is the thing I had in mind when I
said, first look out the window. I don't believe there is any general
way to go about it, primarily because it depends on the model you
are using. For instance, if you are interested in dispersion over
scales of several miles, which would be an urban area diffusion
problem, you would have to come out with some distribution func-
tion that fit what the atmosphere was actually doing. This is the
main requirement, and how you actually do it, I think requires
imagination more than anything. There are certainly numerical
measurements or calculations from some of the work that Dr.
Hilst has done, as an example, or some of the work from large
sources, our TVA studies, where we can see the dispersion
measurement over fairly large scales, and we can use these
numerical data as a beginning point. But I think the problem is
that there is so little that we know about large-scale diffusion
that at this point we simply have to make up a great deal.
SEC TECHNICAL REPORT A62-5
-------
DISPERSION CALCULATIONS: MULTIPLE SOURCES 195
DR. GIFFORD: There are really three parts to this prob-
lem. First, you have got to get the material into the air, and
then there is the part you have been talking about, which is carry-
ing it through the air, and then the other part, which concerns
the effect on people. Also there is the problem of measuring con-
centrations after some residence time of materials in the air.
Since you highlighted the uncertainties and difficulties and general
unsatisfactory nature of the second of these, I wondered if you
would be willing to hazard a guess as to whether the atmospheric
contribution is more or less uncertain, for example, than our
present knowledge of sources and of concentration measurements?
In other words, does it make much difference whether we are a
little uncertain about it?
MR. POOLER: This is a very good point to bring out, be-
cause I think at the present time the estimates we can get of the
sources and also the accuracy of many of the concentration mea-
surements are probably about the same order of accuracy as our
dispersion calculations --of equal uncertainty in all three areas.
MR. HILST: You are not suggesting that on this account it
is worthwhile to take less than the best we can get?
MR. POOLER: We should all aim for perfection, but we
will never know if we have reached it.
SYMPOSIUM: AIR OVER CITIES
-------
Some Effects of City Structure on the
Transport of Airborne Material in
Urban Areas
W. A. PERKINS, Dept. of Chemistry and
Chemical Engineering, Stanford University,
Palo Alto, California
Summary
Transport of airborne material in urban areas is compared to the transport
mechanisms observed and calculated for open areas. The effects of urban areas in modi-
fying general weather patterns are examined, with emphasis on the effects of specific
location, size of the urban area, wind speed and cloud cover, and vertical temperature
gradient. Meteorological and air pollution surveys conducted in several citiss are re-
viewed, and conclusions based on these surveys are presented. The degree of loss of
particulate matter by impaction on exposed surfaces in urban areas is considered;
theoretical and experimental impaction data are reviewed.
There is reason to believe that the transport of airborne
material in urban areas, particularly under certain meteorolog-
ical conditions, might differ substantially from the transport that
has been observed in open areas. Buildings and other structures
increase the surface roughness, thereby enhancing atmospheric
diffusion associated with mechanically induced turbulence. In
addition, the large number of surfaces within an urban area af-
fords a potential loss of particulate airborne material by impac-
tion processes.
Unfortunately the type of experimental data I would like to
present to illustrate these and other effects are not now available.
What are needed, of course, are quantitative observations on the
diffusion of material within cities for direct comparison with
measured diffusion in open areas. To be meaningful these dif-
fusion studies require an air sampling network on a suitable
scale, together with a controlled source of tracer material that
can be emitted in known amounts and, if particulate, in a pre-
scribed particle-size range. Collateral meteorological observa-
tions on a scale commensurate with the travel distances are es-
sential. Results from this type of investigation are a necessary
starting point for the computer techniques suggested by Hilst at
this meeting, if the results of these computations are to be rep-
resentative of actual transport within cities.
In the absence of suitable field data, one must fall on in-
direct evidence that suggests possible differences in the transport
197
-------
198 PERKINS
of airborne materials within cities as compared with open areas.
Perhaps the most pertinent line of evidence is the well-established
fact that air temperatures within cities are significantly different
from those in surrounding open areas. The differences are evi-
dent both in the horizontal temperature distribution near the
ground and in the vertical temperature gradient.
The fact that cities are noticeably warmer than their sur-
roundings has been known for over a century based on records
taken from fixed stations inside and outside of certain cities. '
2, 3 The first measurements of temperature distribution at
ground level within a city were obtained by Schmidt in Vienna^
using equipment carried in a car. A similar technique was used
by Sundborg to obtain comprehensive measurements of air
temperature throughout the city of Uppsala. The first indication
that the vertical temperature gradient was modified in an urban
area was obtained by Blachin and Pye^ during their investigation
of the temperature patterns in Bath, England, which lies in a
steep valley. In a number of cases nighttime temperatures taken
from the hillside showed little or no inversion over the city, al-
though inversions were established within the valley on either
side of the city.
With the above surveys as background, a series of detailed
measurements on urban modification of air temperature was
undertaken by the Aerosol Laboratory, Stanford University. The
results have been reported by Duckworth and Sandberg' and have
been fully described by Mitchell and others at this meeting.
A few comments regarding the objectives of these surveys
may be of interest. Since modification of air temperature by a
city is less pronounced during the day than at night, primary
emphasis was given to nighttime observations. In designing the
program four points were given consideration. First, the attempt
was made to determine by refined techniques whether temperature
effects observed thus far were characteristic of cities in general
or whether these phenomena were peculiar to specific locations.
The fact that urban temperature modification had been noted in
several communities in earlier years certainly suggested that
this was a general characteristic, but additional data obtained
with improved equipment seemed desirable.
Second, the effect of city size on the extent of urban modi-
fication had not been investigated systematically. Accordingly,
measurements were made in San Francisco, San Jose, and Palo
Alto to represent large, intermediate, and small cities located
close enough within the Bay Area to have similar over-all weather
conditions. Conditions are particularly similar during the summer
SEC TECHNICAL REPORT A62-5
-------
SOME EFFECTS OF CITY STRUCTURE 199
months, when the study was undertaken.
Third, additional information was needed to establish the
effect of wind speed and cloud cover on urban temperature
modification. On the basis of available data these two factors
appeared to be the most important over-all weather factors con-
trolling the urban effect.
Finally, and perhaps most important, measurements of
vertical temperature gradients taken concurrently with the
horizontal temperature traverses were needed. It is doubtful
that the hillside observations taken in Bath were truly repre-
sentative of the free air temperature gradient. Accordingly,
wiresounde temperature measurements were made within and
outside each of the three cities in conjunction with the ground-
level measurements taken with sensors mounted on automobiles.
Since the investigation in the Bay Area was completed,
DeMarrais has reported on vertical temperature gradient ob-
servations in Louisville. These urban-area gradients were
compared with typical open-area vertical gradients obtained in
Idaho and New England. The findings of DeMarrais are in accord
with those described by Duckworth and Sandberg.
Typical examples of nighttime temperature surveys in
San Francisco and San Jose have been shown previously. (See
Figures 1, 2, and 3 Mitchell) Air temperature is highest in the
most built-up area and lowest in the open areas. A similar pat-
tern was found for Palo Alto. In all three cities maximum temp-
erature differences were associated with light wind and clear sky,
and under comparable conditions the differences were greater
the larger the city. Thus in order of decreasing city size, the
maximum observed temperature differences between urban and
surrounding areas were 20°F, 14. 2°F, and 12. 6°F. Representa-
tive temperature differences in the three cities were approximate-
ly one half the maximum values.
Figure 1 shows one example of temperature gradients ob-
served in built-up and open areas in San Francisco under condi-
tions favorable for the development of a moderate temperature
differential across the city. Typically, the strong inversion
associated with the open areas is absent within the city at low
level. In the example shown, the gradient in the built-up area is
essentially neutral below 100 feet and isothermal aloft, with a
slight inversion at 125 feet. In many cases the gradient taken
within the city showed a marked inversion at about two building
heights above the surface, with neutral conditions below.
From the several trials conducted in the three Bay Area
cities coupled with earlier and more recent observations obtained
SYMPOSIUM: AIR OVER CITIES
-------
200
PERKINS
TEMPERATURE
-------
SOME EFFECTS OF CITY STRUCTURE 201
From the standpoint of atmospheric diffusion within cities,
the urban temperature modification tends to reduce the extreme
range of conditions expected. Lapse conditions will be found
during the day if such conditions exist in the open. Inversion
conditions in the lower layers within a city are not likely at
night, however, and in any case will be less well-developed than
in open areas. Thus a strong inversion in open areas associated
with light winds and clear sky are accompanied by neutral condi-
tions within built-up areas. If the sky is overcast or the wind
speed is high, the urban temperature differential will be elimin-
ated and neutral atmospheric stability can be expected both inside
and outside the urban area. As a consequence of the urban temp-
erature modification therefore, atmospheric temperature stabil-
ity within a city is generally limited within the range of neutral
to lapse conditions rather than inversion to lapse, as found in
the open. Accordingly the transport of airborne material within
cities may be expected to show less variability than that in the
open.
If the airborne material is composed of particulates, some
of the material may be lost by impaction on the large amount of
surface to which airborne material is exposed in an urban area.
The magnitude of this loss can be estimated by use of experi-
mental data obtained in connection with laboratory studies on
particulate impaction. An airborne particle approaching an ob-
struction may contact the obstructing surface and presumably be
retained, if the particle has sufficient momentum to cross the
streamlines formed around the obstruction. If its momentum is
not sufficient, the particle will be carried by the air stream
around the obstruction and there will be no impaction. As shown
by Johhstone and Roberts, ^ momentum considerations can be used
to define a dimensionless parameter, K, which in turn may be
directly related to particulate impaction efficiency. Specifically,
K 1 d2up (1)
18 DM
where d particle diameter
P particle density
u air speed relative to obstruction
M - viscosity of air
D = minimum crosswind dimension of obstruction
Theoretical and experimental particulate impaction data
reported by Ranz and Wong-'-'-' indicate that no impaction is ex-
pected for K values less than 0. 05. For a nominal wind speed of
10 mph (500 cm/sec), a particle diameter of 2 microns, and
density 1 gm/cc equation (1) becomes
SYMPOSIUM: AIR OVER CITIES
-------
202 PERKINS
v 6.16 x 10"3
K. - —
D
Therefore, in order that K exceed the limiting value of
0. 05, the value of D must be less than 0. 12 cm. Obstructing
surfaces having a minimum dimension greater than this value
will not be effective collectors by the impaction process in
moderate winds unless the particles are substantially larger than
2 microns. Thus buildings and other structures will not be ex-
pected to collect a significant amount of particulate airborne
material.
Although losses might be expected on small surfaces such
as wires and foliage, two factors tend to minimize the over-all
loss on such surfaces in city-wide transport. First, if the foliage
is dense enough to cause significant removal by impaction, the
wind speed and likewise the impaction efficiency will be reduced.
Second, if the wind speed is high, mechanical turbulence will dis-
tribute the airborne material aloft and a smaller percentage of
the total will be in lower layers and in contact with obstructing
surfaces.
At the outset of this discussion I indicated that quantitative
data on the transport of material over cities are not now avail-
able but I must make one exception. Neiberger conducted five
daytime experiments in Los Angeles using a particulate fluor-
escent tracer •"• 1 to investigate the reliability of trajectory analy-
sis from available meteorological data.
In trial 5 all of the tracer material was carried across the
single sampling arc, and further, the arc was wide enough to
bound the airborne tracer on each side. Therefore the results
from this trial can be used for a quantitative comparison between
the amount of material released and the amount found at the
sampling position.
Briefly, the trial was conducted as follows. A total of
2400 grams of fluorescent tracer was released between 5 and 6
a.m. from a point source at Compton Airport, as shown in
Figure 2. Prom appropriate calibration data, it was known that
each gram of tracer contained 5. 5 x Id*" fluorescent particles,
hence the total number of particles released was 1.3 x 10^. A
total of 23 samplers were located along an arc approximately 13
miles from the source. Air samples were collected on membrane
filters. Any fluorescent particles present in the air sample were
retained on the filter surfaces and were readily identified and
counted under ultraviolet light. Filters were changed at hourly
intervals to determine the time of tracer arrival and the time re-
quired for the tracer to pass the sampling arc.
SEC TECHNICAL REPORT A62-5
-------
SOME EFFECTS OF CITY STRUCTURE
203
ARRIVAL TIME
From surface trajectory
From 500ft trajectory
Observed
_I300-I400 hours
_0900-IOOO hours
_ 1000 hours
10
21 Sept. 1954
Figure 2. Results of fluorescent particle tracer test No. 5 conducted by Neiberger in the
Los Angeles Basin. Tracer was released at Compton Airport; samplers were
located on an arc approximately 13 miles distant from the source. Positive
counts were obtained at stations 16 through 23; each circle at these stations
represents one power of ten in particle count. Numbers to right of stations
show hours when tracer particles were collected. Surface wind trajectories by
several analysts are included together with one 500-foot level trajectory. (Air
Pollution Foundation Report No. 7)
During the first 3 hours after the release, winds were
light and variable. The tracer moved an estimated three miles
in a southeasterly direction then later in the morning reversed
its direction. Subsequently, the wind speed increased and the
tracer moved steadily across the center of the sampling arc at
6 mph. The total number of particles collected at each station
SYMPOSIUM: AIR OVER
-------
204 PERKINS
is given below; in Figure 2 each ring around the station repre-
sents particle numbers collected in powers of 10. Sampling rate
was 8. 75 liters/minute at station 23 and 7. 5 liters/minute at all
other stations.
Sampling Total Sampling Total
Station No. Particle Count Station No. Particle Count
15 0 20 864
16 60 21 346
17 1963 22 98
18 3352 23 42
19 1541 24 0
Before the tracer reached the sampling arc, surface heat-
ing created unstable conditions from the ground to 1000 feet. A
strong inversion at the 1000-foot level effectively stopped further
mixing above this height. Under these conditions it is reasonable
to assume that mixing was sufficient below the inversion to pro-
duce a uniform vertical distribution of the tracer between'the
surface and 1000 feet. Thus the amount of tracer recovered at
ground level can be used to compute the total amount of tracer
passing through a vertical plane extending from the sampling arc
to a height of 1000 feet. On the basis of this assumption the total
number of particles crossing the sampling arc from the ground
to the inversion layer is 1. 1 x 10^4. The expected value based
on the total number of particles released is 1. 3 x 10 . In view
of uncertainties in determining source strength, inversion height,
and wind speed through the unstable layer, these two values are in
excellent agreement. These results suggest that the proposed
transport mechanism is reasonably correct and that there was no
significant loss of particles in the course of 5 hours of travel over
the city of Los Angeles.
Obviously, more information is needed before results such
as these, obtained in one trial, can be generalized. If observa-
tions indicate that marked atmospheric stability in the lower lay-
er over a city is normally absent, however, then vertical mixing
of airborne material can be expected both day and night. In this
sense the presence of a city tends to simplify rather than com-
plicate the problem of estimating behavior of airborne material
in its movement over urban areas.
REFERENCES
1. Howard, Luke: Climate of London deduced from meteor-
ological observations. 3rd Ed. London, Harvey and Barton,
1833.
SEC TECHNICAL REPORT A62-5
-------
SOME EFFECTS OF CITY STRUCTURE 205
2. Renou, E. : "Differences de temperature entre Paris et
Choisyle-Roi. ' Societe Meteorologique de France,
Annuaire, 10: 105-109, 1862.
3. Hammon, W. H., and Duenchel, F. W. : "Abstract of a
comparison of the minimum temperatures recorded at the
U. S. Weather Bureau and the Forest Park Meteorological
Observatories, St. Louis, Missouri, for the year 1891. "
Monthly Weather Review, 30(1): 11-12, Jan. 1902.
4. Schmidt, Wilhelm: "Die Verteilung der Minimumtempera-
turen in der Frostnacht des 12 Mai 1927 im Gemeindegebiet
von Wien. " (Distribution of minimum temperatures during
the frost night of May 12, 1927 within the communal limits
of Vienna. ) Fortschritte der Landwirtschaft, 2(21): 681-
686, 1929.
5. Sundborg, A. : "Local climatological studies of the tempera-
ture conditions in an urban area, " Tellus, 2(3): 221-231,
1950.
6. Balchin, W. G. V. , and Pye, N. : "A micro-climatological
investigation of Bath and the surrounding district. "
Quarterly Journal Royal Meteorological Society, 73: 297-
323, 1947.
7. Duckworth, F. S., and Sandberg, J. S. : "The Effect of
Cities upon Horizontal and Vertical Temperature Gradients,
Bulletin American Meteorological Society, 35: 198-207,
1954.
8. DeMarrais, G. A. : "Vertical Temperature Difference Ob-
served over an Urban Area, " Bulletin American Meteor-
ological Society, 42: 548-556, 1961.
9. Johnstone, H. F., and Roberts, M. H. : "Deposition of
Aerosol Particles from Moving Gas Streams, "^Industrial
and Engineering Chemistry, _4Jj 2417-2420, 1949.
10. Ranz, W. E., and Wong, J. B. : "impaction of Dust and
Smoke Particles on Surface and Body Collectors,
Industrial and Engineering Chemistry, 44: 1371-1381,
1952.
11. Neiburger, M. : "Tracer Tests of Trajectories Computed
from Observed Winds", Report No. 7, Air Pollution
Foundation, Los Angeles, California, 1955.
DISCUSSION
MR. YOSHIDA: You mentioned that there were no losses
by impaction on buildings. Have you made any estimate of the
amount of tracers you may lose as a result of a diffusion pro-
cess ?
SYMPOSIUM: AIR OVER CITIES
-------
206 PERKINS
PROFESSOR PERKINS: Yes, we have made a fairly crude
attempt at this and this loss appears to be extremely small.
Much less than that by impaction.
DR. LANDSBERG: Professor Perkins, I have a question
with respect to your d over D factor. That can be valid only at
a certain limiting windspeed.
PROFESSOR PERKINS: You are exactly right. My point
in presenting this was to depict a certain condition at 10 miles
an hour. Of course, as windspeed goes up, the D value would
also go up in proportion. At 100 miles an hour the correspond-
ing value of D would be at 12 centimeters, but at 100 miles an
hour one is not much interested in impaction.
MR. HOLZWORTH: We have seen the slides that were
produced by Duckworth several times here, and you have men-
tioned that there were certain conditions under which one would
get these temperature patterns, the heat islands. With respect
to San Francisco, that area being rather windy and cloudy, we
should point out that these pictures probably are not too repre-
sentative of conditions that occur there often. Is that correct?
PROFESSOR PERKINS: Yes, in the sense that the
measurements were made not necessarily to determine whether
these were characteristic of San Francisco over, say, an appre-
ciable period of time. They are characteristic of a large city,
providing you have the clear conditions and low windspeed.
MR. HOLZWORTH: These were, indeed, rather special-
ized occasions in San Francisco, wouldn't you say?
PROFESSOR PERKINS: I don't represent the San Fran-
cisco Chamber of Commerce but we don't have clouds all the
time. Of course, the stratus is a very common phenomenon
during summer, but much less so in spring and fall.
DR. HEWSON: I would like to ask what might be the time
variation of this rather uniform distribution of turbulence for a
city. For example, don't we begin to get some stabilization,
say, towards dawn?
PROFESSOR PERKINS: The temperature differences
maintain their identity throughout the night; in San Francisco there
is a difference all night long. In a small community, Palo Alto,
the difference is less. So again it is a function of city size.
FROM THE AUDIENCE: In view of the comments about
impaction and d over D, and comments about the effectivity and
desirability of parks in metropolitan areas, would you care to
comment on the effectiveness of parks in helping pollution
problems ?
SEC TECHNICAL REPORT A62-5
-------
SOME EFFECTS OF CITY STRUCTURE 207
PROFESSOR PERKINS: Well, my comments will be rather
qualitative. If one intends to remove particles by impaction, it
will have to be by foliage, since it takes small surfaces to do it.
If the foliage is dense enough to make appreciable reduction in the
concentration of material in an area of 1 or 2 kilometers, then
the windspeed simply couldn't be maintained. "In other words, if
you have a high density of foliage, you lose your windspeed, and
therefore the impaction efficiency drops because it varies in pro-
portion to the windspeed.
We have some reservations as to the possibility of remov-
ing particles effectively by this mechanism. One other consider-
ation is that you remove only the material that passes through
this lower layer. If the material is also distributed substantially
above the lower layer then, of course, the percentage lost by im-
paction is going to be reduced.
MR. KALSTROM: In the studies in San Francisco and San
Jose was there any attempt to differentiate natural differences
from the effects of the city? In other words, were all of the ob-
served effects due to the' city or were part of them due to condi-
tions that were there previously? I don't know how much building
has gone on in San Francisco since 1952, but in both San Jose and
Palo Alto there has been considerable building and enlargement
of the city since that time. Would a new study show the same
pattern or would it show the effects of the more recent building?
PROFESSOR PERKINS: San Jose, I am sure, would show
a new pattern. There isn't the slightest doubt about it. And this
is by virtue of the fact that not only has it expanded, but there
has been an appreciable density increase in certain areas. A re-
study of this area, I think, would show the effects that you men-
tioned.
Now, I think your first point was that you are trying to
separate the effect of the city from the local climatology --
MR. KALSTROM: Pre-existing.
PROFESSOR PERKINS: Pre-existing climatology. I think
San Jose might be a good example to illustrate this point. Were
San Jose completely removed, and the flat area undeveloped or
agricultural as it was in the past, I am sure you wouldn't find the
kind of temperature discontinuity now observed in that area.
SYMPOSIUM: AIR OVER CITIES
-------
Source Configurations and Atmospheric
Dispersion in Mathematical Models of Urban
Pollution Distributions
GLENN R. HILST, The Travelers Research
Center, Inc. Hartford, Connecticut
Summary
The bivariant normal distribution of airborne materials emitted continuously from
individual point (small-area) sources is utilized as the basic mathematical model for at-
mospheric dispersion. The mathematical form of the variance terms .in this model, as a
function of travel distance or time, is reviewed. The integral form of these basic equa-
tions for multiple sources is then derived and examined for tractability under various
source-distribution and dispersion conditions. Simplified conditions are utilized to ob-
tain distributions of air pollutants for steady-source and meteorological conditions and
these are presented to show the effects of concentrated versus widely dispersed sources.
The more general utility of the model in studying various source configurations under a
variety of meteorological conditions and the restrictions that must be imposed for real
sources and atmospheres are reviewed.
The problem of urban air pollution and its impact on the
comfort, safety, health, and economic status of communities of
people are, paradoxically enough, becoming better defined and
more complex at the same time. Better defined as we discover
the underlying biological, physical, and chemical processes at
work; more complex as we recognize the multivariate nature of
the problem and the sometimes subtle, sometimes bold interac-
tions of the atmosphere, sunlight, multiple pollutants, and the dis-
tribution of sources of pollutants in establishing the patterns of
exposure in the urban environment. A pollution problem exists
only if there is a material or combination of materials that ad-
versely affect at least one segment of a population; the effects are
in themselves frequently difficult to define.
The work of many people in the many facets of air pollution
has shown conclusively that the knowledge and understanding we
require to cope with this problem, whether in urban planning or
pollution control, must be derived as a consistent synthesis of the
processes of pollution generation and emission, atmospheric
transport and diffusion, and interactions between the pollutants
and the receptors. It is the purpose of this paper to present a
method for synthesizing the first two of these, source variables
and atmospheric variables, by way of a mathematical model and
in the context of the urban pollution problem. This model is re-
stricted to steady-state meteorological and source conditions and
predicts average concentrations.
209
-------
210 HILST
THE MODEL
Consider a Cartesian coordinate system with an arbitrary
origin but fixed in the earth, with the x axis oriented along the
direction of the mean wind, the y axis oriented horizontally cross-
wind, and the z axis oriented along the vertical. At a point
(Ł, -1}, 3 ) we place a continuous, steady point source of strength
Q( Ł , 1 .5 )- The average concentration of the material emanat-
ing from this source when it reaches the point (x, y, z) is
X(x-Ł, y-ij, z-4). We wish to specify X (x, y, z|Ł , r\, \ ) in
terms of Q( i-, 1, .J), (x-?, y-1, z-.$ ) and the parameters that
characterize the dispersive capacity of the atmosphere. For this
purpose we shall assume that the distribution of X in the (y, z)
plane is bi-variant normal. Then
X(x, y, zlf, , , j ) ' "' "* } exp
1 z
-------
MODELS OF POLLUTION DISTRIBUTIONS 211
SOLUTION OF THE MODEL
Solution of equation (1) and integration of equation (2) re-
quires the specification of o-y2 (x Ł ) and
-------
212
HILST
\
\
\ Line Source
\
\
\
\
\
V
X\
\
P| =
Fi n ite Area
Source
Continuous Area
Source
0
Figure 1. Maximum concentrations as a function of downwind distance from a line source,
a banded source of width w, and a uniform source distribution, all of infinite
crosswind extent.
Another criterion that may be established is, "What source
distribution gives the minimum total dosage over some area of
interest?" In the present steady-state model comparative dosages
can be obtained by integrating equation (2) over the area of interest.
With the infinite crosswind source distributions assumed for our
exemplary models, the integration is over a downwind distance
X from the line source or the upwind edge of the area sources.
Then
X
D:
X (x)
(4)
where Xr (x) is the distribution of concentration in the plane of the
maximum (the plane of the sources), D is the dosage per unit time,
and the subscript i specifies the source configuration (i - J for
line source, i w for banded source, i c for uniformly dis-
tributed source).
SEC TECHNICAL REPORT A62-5
-------
MODELS OF POLLUTION DISTRIBUTIONS 213
Solution of equation (4) for these source configurations
yields
2-P
X 2
F u AzX'(2-p)
W-A w V2Ą u Az#(2-p)(4-p)
2-P
4Q/i X 2
Dc,X
X
4-p"|
2J
2IT u Az/>(2-p)(4-p)
Forming ratios of these dosages, we have
D/,X 4-p
(a)
DC,X 2
(b) (6)
DW,X 2 DW)x , >
~T^ ~A T^ \("'
D/ , X 4-p DC , X
Equations (6) are of particular interest since we wish to know the
comparative dosages. Equations (6a) and (6b) are shown graph-
ically for three values of p in Figure 2, where the total dosages
from a line source and banded sources of varying downwind widths
w are compared with the total dosage from a uniformly distributed
source covering the entire distance of interest, X.
Inspection of Figure 2 shows that the maximum total dosage
is associated with the line source and this dosage is 10 to 15 per-
cent higher than the uniform source case. The total dosage from
the banded source area is intermediate between the line and uni-
form sources and tends to the latter as the fraction of the distance
of interest occupied by the banded source increases (w/X •— 1. 0. )
If one can ignore for the moment all other requirements that
must be met in planning the distribution of sources of atmospheric
Pollutants in urban areas, the ramifications of Figures 1 and 2 are
SYMPOSIUM: AIR OVER CITIES
-------
214
HILST
1.20
pe o
Q IP
1.02
1.00
0.9 1.0
w
X
Figure 2. The effect of downwind source width, w, on total dosage in the plane of max-
imum concentration and over a distance of interest X, for three values of the
diffusion coefficient, p. (w = 0 represents the case of a line source.) (See text
for explanation of ratio Dw \/Dc )(.)
almost self-evident. One can minimize, or at least largely con-
strain, maximum concentrations and can minimize absolutely
dosages over an area of interest by distributing the sources of air
pollutants as uniformly as possible over that area. The reduction
in total dosage by this device is not large, according to this model,
but again the value of a 10 to 15 percent reduction in the general
exposure of the area or population to air pollutants must be weighed
separately.
Some generality can be introduced in the source distributions
assumed above. The more practical problems of urban air pollu-
tion do require the recognition of constraints set by aesthetics,
SEC TECHNICAL REPORT A62-5
-------
MODELS OF POLLUTION DISTRIBUTIONS
215
land utilization, transportation, power and water supplies, top-
ography, and the like. These source distributions are unique to
the locality for which they are derived and must be considered
individually, a procedure for which the models developed here
are equally useful.
We can consider one further problem that is of real interest
in city planning. The question may be phrased, "Of what value
are 'green-belts' (non-source areas) in the context of air pollu-
tion?" An initial solution to this problem may be obtained by solv-
ing the model for an arbitrary arrangement of banded source
strips. An example of such a solution is shown in Figure 3, where
three source bands are arranged, as shown by the shaded strips,
Distance Downwind (km)
Figure 3. Comparison of maximum concentrations as a (unction of distance downwind from
three arbitrarily arranged banded source areas and the same total source strength
distributed uniformly over a distance of 20 km. All source configurations are of
infinite crosswind extent.
at 0 « x < 3 km, 7 ^ x < 9 km and 11 $ x 3 16 km. The resultant
concentration distribution is shown by the solid line in the graph;
for comparison, the concentration distribution that obtains for a
uniformly distributed source configuration is shown as the dashed
line (total source strength is conserved in this comparison).
The comparative excess of concentration (and dosage) in the
compressed source strips and the deficit of these quantities in the
"green-belts" are obvious from the diagram. Concentrations for
SYMPOSIUM: AIR OVER CITIES
-------
216 HILST
these two cases are converging with increasing distance, i. e. , at
large distances the concentrations and dosages become insensi-
tive to the banded source arrangement.
CONCLUSIONS
The power of mathematical models to cope with the joint
effects of source distributions and atmospheric dispersion prop-
erties in determining air pollution distributions has been illustrated.
The ability of these models to handle much more complex systems
of sources and meteorological variability is clearly recognized and
requires only the judicious use of available high-speed electronic
computer equipment.
A limitation must be recognized, however. These models
are no more accurate and complete than the mathematical repre-
sentations of the physical processes that they incorporate. In the
present instance the lack of a thorough knowledge of the vertical
exchange capacity of the atmosphere over a rough urban surface
and the failure to incorporate loss of pollutants by deposition are
most serious areas of ignorance. These must be overcome by
continued research; in the meantime we have the mechanism, in
these models, to synthesize what we do know in a useful way.
Perhaps more importantly, we can use these models as guides to
the design of the highly complex experimental programs required
to further our knowledge in this important problem area.
SEC TECHNICAL REPORT A62-5
-------
Some Aspects of Atmospheric Diffusion
in Urban Areas
JAMES HALITSKY, Research Scientist,
New York University
Summary
The waffle-type topography of a large city creates two general airflow regimes: a
channel (low through streets and among buildings, and an upper airflow. Three types of
discharge of pollutants over cities (ground-level, roof-level, and tall-stack discharge) are
considered in terms of diffusion by the two airflow regimes. Ground-level discharge is in
channel flow, which is principally an aerodynamic phenomenon. Wind-tunnel studies of
concentration patterns would be very useful for this source. Roof-level discharge is in a
transition regime. A full-scale study would require saturation of an area with sampling
equipment for the short duration of a constant wind. A program to develop simulation of
turbulence in wind tunnels would prove helpful for studies of roof-level discharge. Tall-
stack discharge is in free atmospheric flow, with turbulence generated by ground rough-
ness and by heat from building surfaces. A full-scale study would be required to deter-
mine diffusion coefficients over a city.
Gas concentrations in an urban atmosphere may vary over
several orders of magnitude. For example, the concentration of
SO within the chimney of a coal- or oil-fired heating plant may
be of the order of 1CF ppm, while day-to-day samples taken at
street level may be of the order of 10~1 ppm or less. In a given
atmospheric condition, an observer moving among such sources
would not normally experience the highest concentrations. Human
beings, animals, plants, and mechanical equipment sensitive to
corrosion would be speedily removed from an environment con-
taining 10 ppm. SC>2 concentrations of the order of ICr ppm are
found quite frequently on roof tops, however, and concentrations
of 10 ppm may be found in upper-story apartments when the wind
is in the right (or wrong) direction. At a distance from a given
chimney great enough that the contribution of that chimney to the
concentration level is negligible, the observer experiences low
concentrations of the order of 10 ppm. These concentrations
do not vary rapidly with location. They are caused by the in-
tegrated effect of all sources in the area.
Between the recognizable regions of high local concentra-
tion and low diffuse concentration, there exists a poorly defined
region of intermediate concentrations. This is the ordinary en-
vironment of the urban dweller. It is a three-dimensional space
in Which the concentrations fluctuate in time, and through which
the observer moves during his daily routine. Our understanding
of diffusion and our knowledge of pollutant distribution within this
217
-------
218 HALITSKY
intermediate region is practically nil.
A research program designed to evaluate air pollution over
cities should take into account the characteristics of these three
regions, and methods of observation and analysis should be suit-
able to each region.
ANALYTICAL METHODS
In the mathematical treatment of atmospheric diffusion a
real situation is customarily idealized into a simple mathematical
model. A tall chimney may be represented by a continuous point
source. A multitude of small chimneys dispersed over a large
area may be represented by a continuous area source. When
these idealizations are made, it is implied, but not always
stated, that the derived formulas for ground concentration are
valid only beyond a loosely defined distance at which emission ir-
regularities cease to influence the concentration distribution in
the plume.
For the isolated chimney the irregularity may appear as an
initial plume expansion, which becomes insignificant in comparison
with subsequent diffusion after a short travel distance. Since the
plume rarely reaches a receptor on the ground in this distance,
the foregoing limitation is not a handicap. In a multiple-chimney
problem the area-source solution will be valid beyond some dis-
tance from the edge of the area. When the principal receptors are
among the chimneys within the area, the area-source solution is
not adequate.
In a region of multiple sources the technique of distributing
the total effluent uniformly over the total area must lead to an
underestimate of the maximum concentration experienced by a re-
ceptor moving within the area. Diffusion theory generally predicts
that all sources upwind of a point will contribute to the concentra-
tion at that point. Maximum contributions will come from sources
directly upwind. The contributions from sources displaced later-
ally from the upwind axis decrease exponentially with lateral
distance. Practically, the entire contribution will be made by
sources lying within a wedge having its apex at the receptor and
its body symetrically disposed about the upwind axis. In the
mathematical process of transforming the multiple sources into an
area source, some of the contaminant is removed from this wedge
and distributed over the area outside the wedge. The removed
portion makes no contribution to the concentration at the receptor;
this results in a lower estimate of the integrated effect than would
be obtained if the sources within the wedge were not distributed.
Moreover, a receptor immediately downwind of a stack would
SEC TECHNICAL REPORT A62-51
-------
ASPECTS OF ATMOSPHERIC DIFFUSION 219
receive very high peak concentrations.
To overcome this defect one might replace the continuous
area source by small area sources of high concentration at the
actual chimney locations. Concentrations in the plumes for each
chimney might be calculated, and the total concentration at any
point would be derived by a numerical summation of concentrations
from all sources. This approach would be quite feasible if the
physical reality consisted of isolated chimneys rising from level
ground of low uniform roughness. In such a case, the mean wind
speed and direction and diffusion coefficients could be postulated.
In a city, however, most chimneys do not extend more than
a few feet above the roofs of the buildings they serve, and other
sources such as automobile exhausts move about on the ground.
The mean wind high above the city need bear no simple relation-
ship in speed or direction to the wind at roof or street level.
Diffusion coefficients at roof and street level are simply not
known. Thus, it is unlikely that a purely analytical approach will
prove effective within a city. Experimental data are needed to
give clues to diffusion rates in circumstances where the principal
factor is the influence of local topography.
EXPERIMENTAL METHODS
How shall such experiments be conducted? Evidently we
must consider source location, receptor location, building
shape, wind speed, wind direction, and thermal gradients. It is
a huge task to instrument one full-size building to cover all pos-
sible receptor locations and to conduct such a test during the
short time period during which atmospheric conditions remain
constant. Moreover, we must consider whether the information
obtained in this one test is valuable enough to justify the expense.
Generally, such considerations militate against the experiment.
The alternatives to a full-scale field test are spot checks
at a given installation or comprehensive tests on a scale model.
Recent model tests at New York University found that spot checks
are often misleading because neither the instantaneous nor the
mean plumes from a source close to a building can be predicted
with certainty. One cannot tell whether the spot measurement is
a maximum or not. Under controlled wind-tunnel conditions, ex-
haustive studies of concentration distributions can be made at
reasonable cost. Such intensive studies have been performed re-
cently with a model of the Clinical Center of the National Institutes
of Health at Bethesda. 1» 2
Figures 1 and 2 are representative of the smoke pattern
photographs made during tests of smoke patterns around single
SYMPOSIUM: AIR OVER CITIES
-------
CO
W
O
H
M
n
a
^
i—i
n
>
r
SJ
w
^
o
«ind
centerplane section elevation
roof plan
rear face elevation
Figure 1. Smoke Concentrations Around a Cube 0° Orientation
r
H
•s.
7,
-------
r.
o
H
-f.
K
•X
r;
H
wind
c
C/;
C
x
roof plan
centerplane section elevation
elevation
Figure 2. Smoke Concentrations Around a Cube 45° Orientation
-------
222 HALITSKY
block shapes. 3 They indicate that the maximum concentrations
may occur in unexpected locations. For example Figure 1 shows
that when a building is oriented with one face normal to the wind,
the emission from a flush roof exhaust will produce maximum
roof surface concentrations upwind of the exhaust opening. This
condition does not obtain in a corner orientation, as shown in
Figure 2.
AN URBAN AREA DIFFUSION RESEARCH PROGRAM
In evaluation of pollution over and within an urban area,
some generalizations of source distribution are needed to es-
tablish typical pollution patterns. Three major source groupings
can be established. The first contains the very tall chimneys
found on power plants, refineries, and municipal incinerators.
The second contains the numerous heating plant and incinerator
exhausts at roof level of the older five- and six-story buildings
that fill our city blocks. The third group contains the automo-
tive exhausts at street level.
None of these source groupings are amenable to strict
analytical treatment. The tall stacks may be considered as
elevated continuous point sources, but there is a dearth of in-
formation regarding diffusion coefficients over urban areas.
The roof-top emissions may be treated as rectangular area
sources separated by clear bands formed by streets and avenues,
but it is difficult to predict how rapidly the gases will penetrate
downward between buildings. Diffusion from ground-level sources
is greatly affected by the channeling of air currents between
buildings and the back flow in eddies created by building corners;
analytical methods are completely inadequate for this type of
source.
Since it is generally desirable to use analytical methods
where possible, experiments should be performed to determine
diffusion coefficients and mean wind profiles at various eleva-
tions above roof level. These quantities should be measured at
full-scale in the atmosphere. Although the controlled environ-
ment of the wind tunnel makes testing easier, the modeling of
atmospheric turbulence in the wind tunnel has not been developed
sufficiently to give reliable results, especially where thermal
gradients are involved. There is a possibility that wind tunnel
models may be used to determine diffusion coefficients near roof
level under neutral conditions and high wind speeds, since the
mechanical turbulence in this region should be much greater than
thermal turbulence. Until the scaling of turbulence spectra has
been more fully explored, however, this approach can not be
recommended.
SEC TECHNICAL REPORT A62-5,
-------
ASPECTS OF ATMOSPHERIC DIFFUSION 223
Once the diffusion coefficients above the city have been
determined, the analysis of tall stack sources may proceed. If
the coefficients are found to vary considerably with height, it may
be necessary to refine the diffusion formulas to permit arbitrary
variation of wind speed and diffusion coefficient. Numerical
methods may be useful to give approximate but practical solu-
tions. One open question is the elevation of the effective ground.
Since ground concentrations theoretically vary inversely as the
square of the source height, the assumed ground elevation is im-
portant. Perhaps the source height ought to be measured from the
ZQ of the large-scale wind profile over the city.
The rooftop area source poses an interesting analytical
problem, since part of the pollutant will diffuse upward and part
downward into the space between buildings. This aspect is beyond
my competence and I leave it as a challenge to the analysts. Roof-
top diffusion downward into city streets can easily be measured in
model tests, however. A representative section of the city contain-
ing several typical repetitive block arrangements can be tested for
area emission or multiple-point-source emissions on a given
roof. Three-dimensional concentration gradients within the
street channels can be measured as far downwind as is necessary
for the concentrations to reduce to negligible values. A numerical
addition of displaced concentration patterns will give the in-
tegrated effect of emissions on all buildings.
The diffusion of pollutants from street-level sources should
be studied with models to determine the three-dimensional diffu-
sion pattern. Visual smoke studies to determine the region of
highest concentration, followed by quantitative concentration
measurements in this region offer assurance that the maximum
concentrations will be detected. Model concentrations can be
represented as non-dimensional coefficients, which may then be
used with full-size values of wind speed, source strength, and
building size to obtain full-size concentrations. ^
The validity of model testing of diffusion around buildings
has not been established as yet by direct comparison of full-scale
and model experiments in a variety of turbulence conditions.
There is theoretical support and some experimental evidence to
indicate that the procedure is valid for neutral atmospheres and
higher winds. Tests of gas bomb releases in a mock-up of a
full-scale village and a model of the village in a wind tunnel have
shown good correlation. 4' 5 Funds have been requested from
NIH for comparative tests to establish with greater refinement
the degree of discrepancy between model and full-scale tests in
the case of diffusion around a simple block-shaped building, and,
if such discrepancies are found to be large, to develop analytical
SYMPOSIUM: AIR OVER CITIES
-------
224 HALITSKY
or experimental techniques to improve correspondence.*
The three-mode source arrangement suggested in the previ-
ous paragraphs is, of course, an approximation. Even if all the
analytical and experimental procedures that were discussed should
prove effective, there is no assurance that the combined picture
would correspond to reality. Some experimental evidence must be
found against which the predictions can be weighed. There is no
substitute for full-scale field sampling for this purpose.
A study just beginning in New York City may provide the
type of information required. The New York City Health Research
Council has given a research grant to Cornell University to cor-
relate weather and pollution variables with the incidence of pul-
monary ailments in a normally healthy population. The portion of
the program concerned with weather and pollution documentation
has been subcontracted to New York University under the direction
of Dr. Wm. T. Ingram. I am involved in setting up the weather
and sampling stations. The medical and statistical analyses will
be handled by Cornell.
We intend to record continuously the concentrations of a
number of atmospheric pollutants at four levels of a 200-foot-high
building that stands fairly isolated in a more or less uniform area
of 5-story tenements. In addition to the contaminants, we will
measure the wind speed at the roof, the variation of temperature
with height, solar radiation, rainfall, mean temperature, and
pressure. We also intend to measure simultaneously, under
selected test conditions, the horizontal variation of specific pollu-
tants. To supplement our own measurements, the records of the
various weather stations at municipal airports and stations of the
New York City Department of Air Pollution Control will be used.
It is hoped that the data collected in this manner can be inte-
grated into a space-time picture of the pollution distribution in the
area under consideration, which measures about 1 by 5 miles and
is located on the west shore of the East River in midtown Manhat-
tan. This information will then be analyzed in conjunction with
records of the activity of sample populations in the area and their
state of health. Fortunately, this information will also be avail-
able for other studies, such as those suggested in this paper.
*The NIH has awarded Mr. Halitsky a 3-year research grant for
this purpose, effective June 1, 1962.
SEC TECHNICAL REPORT A62-5
-------
ASPECTS OF ATMOSPHERIC DIFFUSION 225
REFERENCES
1. Halitsky, James, Diffusion of Vented Gas Around Buildings,
APCA Journal, v. 12, no. 2, February 1962.
2. Halitsky, James and Jones, Herbert H., Wind Tunnel Tests
of Exhaust Recirculation at the NIH Clinical Center, Paper
No. 65, Amer. Ind. Hyg. Conf. , Washington, D. C. May
16, 1962. (To be published in AIHA Journal. )
3. Halitsky, James and Golden, Jack, Diffusion of Vented Gas
Around Simple Boxes (in preparation).
4. Kalinske, A. A., Jensen, R. A. and Schadt, C. F., Wind
Tunnel Studies of Gas Diffusion in a Typical Japanese
Urban District, NDRC Div. 10 Informal Kept. No. 10. 3A-
48, June 1945.
5. Kalinske, A. A., Jensen, R. A. and Schadt, C. F. , Correla-
tion of Wind Tunnel Studies with Measurements of Gas Dif-
fusion, NDRC Div. 10 Informal Rept. No. 10.3A-48a,
September, 1945.
DISCUSSION
FROM THE FLOOR: We frequently run into the problem of
an adequate sampling population. You mentioned using Welfare
Department cases. Are we to assume that everybody in Manhat-
tan is on welfare ?
MR. HALITSKY: The sample population was selected by
Cornell. It consists principally of Welfare Department cases,
who report weekly for their allowances and may thus be questioned
about their ailments. Some volunteers not on welfare are included.
Our building choice was based on physical suitability and prox-
imity to the center of the sample population. The economic
status of the population will have to be filtered out statistically.
MR. SMITH: Do you have any data at all from the Clinical
Center at Bethesda to correlate with the concentration measure-
ments you made in the tunnel? This is the kind of thing I think
would be most instructive.
MR. HALITSKY: There are no field-correlation measure-
ments available. I wish I could assure everybody that the tunnel
measurements can be converted to field concentrations simply by
inserting the appropriate known parameters into the scaling for-
mula. In tests of this type, where flow disturbances are aero-
dynamic rather than thermal in origin, there are indications
that diffusion is determined mostly by turbulence induced by the
building rather than turbulence in the approaching air stream.
The request for research support from NIH is aimed at resolv-
ing this problem.
SYMPOSIUM: AIR OVER CITIES
-------
226 HALITSKY
MR. SMITH: To finish my question, I think you are right
that in such cases the tunnel probably represents beautifully what
happens around a building. But I think that many of us are a little
worried about the scale-factor problem, about whether the model
representation is correct or not.
MR. HALITSKY: My tests were limited to regions very
close to the buildings because I, too, am concerned with the in-
creasing influence of atmospheric turbulence as wake turbulence
decreases. I don't think there is any information available on the
relative importance of atmospheric turbulence in the presence of
wake turbulence.
MR. COLLINS: For a good many years I worked with an
industrial concern that took advantage of some of the fine work
that Mr. Halitsky and his colleagues are doing at New York Uni-
versity. I would like to tell a little story and then ask him a
question.
We were consulted on the design of a chemical laboratory,
which had stacks for many chemical exhausts on the roof. The
question arose as to how high these stacks should be to prevent
the recirculation that Mr. Halitsky has demonstrated. After
studying some of the shapes in the wind tunnel, we decided that
if the stack were well above the surface of separation there should
be no trouble.
We passed our recommendations on to the design engineer
and the architect. We didn't hear anything more about it until
about a year later. The laboratory had been built at that time,
and we got an emergency call that about twice a week they had to
evacuate all the personnel, 250 people, because of recirculation
of H2S and other unpleasant fumes.
We went out to look at the laboratory, and this is what had
happened. True enough they put the stacks up to the height of 30
feet above the roof as we had suggested. But the architect didn't
like the looks of them. He built a penthouse around the stacks so
that the actual stack was about six inches above the roof of the
penthouse; furthermore, the penthouse had a 2-foot papapet on it,
and all the fumes were trapped in this little well on the top of the
building and they just sort of oozed out into the air intake.
My question is, how do you win an argument with an
architect ?
MR. HALITSKY: Well, actually it is pretty tough. In the
Bethesda work we tested not only the original conditions, but some
suggestions that had been made for alleviating the conditions.
There were severe restrictions on what could be done with the
SEC TECHNICAL REPORT A62-5
-------
ASPECTS OF ATMOSPHERIC DIFFUSION 227
building, because aesthetically they are very proud of it. I think
this is true in many of the modern types of construction.
FROM THE AUDIENCE: Your use of this simple model
for data reduction and inclusion of V in the determination of
scale factor seems to imply that there is a stationary wake form
with respect to wind velocity. Is this in fact so?
MR. HALITSKY: This is so if you use the term quasi-
stationary instead of stationary to indicate a time mean rather
than an instantaneous wake form. We have tested a number of
sharp-edged building models at different tunnel wind speeds
varying from 1 to about 15 ft/sec. The values of K in the wake
do not vary with wind speed above a speed of about 3 ft/sec. I
believe that this is due to the sharp edges of the building, which
prescribe the wake shape. A rounded building shape would have a
wake that is more sensitive to boundary layer formation, which
in turn depends on Reynolds Number and therefore on wind speed.
MR. MOOK: On the matter of scaling, I think the same
phenomenon of exhaust recirculation occurs when the wake is
not generated by a moving wind but by a moving building or car.
MR. HALITSKY: Yes. I have a station wagon with the
rear seat facing backward. My children like to ride in the back
with the rear window open. The exhaust recirculates into the
rear window, and out through the butterfly in my left front window.
It is not so bad on a level road with the throttle open, but when
going down hill while braking, I find the exhaust concentrations
up front intolerable. As a general rule, I do not permit the rear
window to be open more than 3 inches from the top.
FROM THE AUDIENCE: In your tests, to what height
above the building did you have to go to measure the free wind
speed?
MR. HALITSKY: The height varies with the distance down-
wind from the leading edge. On the cube the height of the wake
at the downwind edge of the roof is about half the height of the
cube. The speed just outside of the wake is about 15 to 20%
higher than the free wind speed.
FROM THE AUDIENCE: In other words, if we were mak-
ing a measurement for air pollution purposes, we ought to get up
a distance of half the height of the building?
MR. HALITSKY: It depends on wind direction. The wake
over the roof in a cornering wind is very small. The wake in a
frontal wind is large. To be absolutely safe in all wind directions
you would have to get above the largest wake, and even then you
must take into account a certain degree of overspeed. This
SYMPOSIUM: AIR OVER CITIES
-------
228 HALITSKY
figure of half the building height applies only to a cube. I believe
a tall building of-small plan cross section would have a wake over
the roof no larger than that of a cube of side area equal to the plan
area of the building.
MR. GOLDSTON: In reviewing the papers that have been
presented here in the past day and a half, I hope I leave here more
enlightened than confused. I can see the immensity of the prob-
lems we face here, and their complexity. I think we are facing a
problem of greater gravity than simply that of time and space,
per se. I just wonder whether, at some time in the future, we
would be able to take the mass of material that we have con-
sidered here, the heat, the various strata and layers of pollu-
tants, the precipitation patterns in cities, the suburban tempera-
ture variations, and make them amenable to a simple statement
that we could use as a guide to further our activities and studies.
MR. HALITSKY: I think that other members of the
Symposium would be better qualified to answer that question.
DR. HILST: Throughout what we have been talking about
this morning there seems to run a theme not stated, and I hope it
is not assumed, that there are unique solutions to the problem
which we face. I would like to disabuse anybody with that idea.
We recognize that we are working with highly variable fluid
quantities here. They vary with time and with space. We recog-
nize that, from a statistical point of view, if we are going to de-
fine over a large amount of time, we have to take a large sample.
But we also recognize that the atmosphere itself is guilty of taking
small samples. It has a large degree of freedom within which to
work and because of inhomogeneity and the rapid changes in the
wind and temperature, the atmosphere itself gets only a small
chance to act in the specific circumstances we prescribe. There
is no solution; we can only describe the distribution of possible
solutions.
CHAIRMAN NEIBERGER: I think that we can all see that
there are various approaches to this problem of dispersion in the
atmosphere. The approach of the statistical turbulence measure-
ments, the approach of the continuous diffusion concept, and the
approach of the model experiments; the use of computers both
analog and digital give some hope of solving the complex problems.
-------
Session 3: Present and Future Needs for Meteorological
and Air Quality Observations
J. J. SCHUENEMAN,* Chairman
Trie Relative Importance of Some
Meteorological Factors in Urban Air Pollution
ELMER ROBINSON, Stanford Research
Institute
Summary
Wind patterns and inversion or stability conditions are important meteorological
factors affecting urban air pollution. Popular attention is usually focused on inversion,
even though wind conditions can be shown to be more significant in many situations. The
relative importance of winds and inversions is examined in terms of theoretical, statis-
tical, and climatological considerations.
This .discussion is not a technical paper in the usual sense,
since it puts forth a particular point of view of the author and does
not include the examples of contrary fact or opinion that would be
necessary in a technical paper. Neither does the discussion in-
clude a comprehensive presentation of available research studies.
This type of presentation was specifically designed to take advan-
tage of the program chairman's remark that "controversial topics
should not be avoided. " The author hopes that this approach will
lead to a more careful appraisal of urban diffusion meteorology.
Not long ago an eastern city experienced a period of partic-
ularly severe air pollution. When a newspaper reporter asked
the cause, he was told that the pollution was due to a temperature
inversion affecting the area. On the West Coast it sometimes
seems that everyone believes that if there were no inversion over
Los Angeles there would be no smog in that afflicted area.
Although public recognition of the smog-producing effect of
a temperature inversion seems quite remarkable, careful con-
sideration indicates a rather poor public understanding of urban
air pollution meteorology. The most unfortunate aspect of this
situation is that professional meteorologists are doubtlessly to
blame for this state of affairs. The inversion is often given spec-
ial treatment and emphasis, and the rest of the meteorology is all
too often glossed over or not mentioned at all. With respect to
meteorological air pollution factors, it would appear that a much
better case can be made for a weak wind pattern than for a low
inversion. In the following discussion the relative importance of
these two parameters, inversion and wind, will be considered in
some urban air pollution situations. The impact on air pollution
meteorology of "inversionless" thinking will be examined, and a
more rational approach to an explanation of city-wide air pollu-
tion will be outlined.
* Chief, Technical Assistance Branch, Division of Air Pollution,
U. S. Public Health Service.
229
-------
230 ROBINSON
AREA-WIDE CALCULATIONS
An exercise undertaken at some time by almost everyone in
air pollution is the calculation of a typical volume of air available
for the dilution of an area's daily emission of pollution. Table 1
shows the results of such calculations for the San Francisco Bay
area. On the first line are the calculated concentration figures
for two inversion heights with typical nonsmog day winds.
TABLE 1
CALCULATED AREA SO2 CONCENTRATIONS
San Francisco Bay Area (500 mi^ )
SO2 emissions 1147 T/day
Inversion Height
Ventilation
Rate 1500 ft. 1000 ft
8 mph 0. 07 ppm SO2 0.11ppmSO2
5 mph 0. 11 0. 17
2 mph 0. 29 0. 43
(Ref: Wohlers, Community Air. Pollution Sources)
The succeeding lines of calculations show the results of altering
the meteorological parameters in ways that are typical of moder-
ate and severe air pollution days. When the inversion is lowered
to a moderate level with constant winds, the result is a propor-
tional but moderate increase in concentration. When the inversion
is left at its average height and the average wind is dropped to
2 mph, the resulting calculated concentration change is still pro-
portionate but it is obviously much more pronounced than in the
inversion change. In the low-inversion, low-wind situation the
wind is still the dominant feature. It is generally much more
realistic to expect a several-fold change in wind speed than in
inversion height.
When the day's smog story is written, however, it will prob-
ably begin with: "A strong inversion layer sealed off the area
today and smog concentrations quickly climbed to irritation
levels. " Perhaps the wind pattern will be mentioned, perhaps not.
STATISTICAL RELATIONSHIPS
Another way of viewing the inversion and wind relationship is
to correlate their values with air pollution measurements. This
SEC TECHNICAL REPORT A62-5
-------
RELATIVE FACTORS IN URBAN POLLUTION 231
has been done many times and perhaps it is dangerous to present
individual examples; however, since the object of this discussion
is to look for over-emphasized inversions, some selection is
perhaps excusable. In the Los Angeles air pollution situation, the
temperature inversion has probably received its earliest and
widest attention. Thus a more careful look at some of the rela-
tions among inversions, wind, and smog in Los Angeles may pro-
vide a valuable example.
It should be pointed out that the Los Angeles inversion is
primarily a subsidence inversion associated with the subtropical
Pacific High. The height of the base averages between 1000 and
1500 feet during the summer and fall months. The inversion
base is below 3000 feet for 80 to 90 percent of the time from June
through October. Below the inversion base, thermal turbulence
will be a dominant factor in mixing emissions. The air below the
inversion is usually identified with the on-shore movement of
marine air, while the air above the inversion is dry superior air.
A number of years ago Stanford Research Institute correlated
oxidant data with various meteorological variables. These factors
included the afternoon height of the inversion base and the daily
wind movement in the Los Angeles area. Both simple and partial
correlation coefficients were calculated on the basis of data for
the period of July 1 to November 30, 1952. The results are
shown in Table 2.
TABLE 2
CORRELATION COEFFICIENTS FOR LOS ANGELES
(July 1-November 30, 1952)
Single Partial
Relationship-1 PM Observations Correlation Correlation
Coefficient Coefficient
Oxidant--Wind Speed (m = 130) -0. 70 -0. 41
Oxidant--Inversion Height (m = 130) -0.67 -0.29
The simple correlation coefficients between total oxidant
and either wind speed or inversion height were high. The nega-
tive result is expected with both variables. The partial correla-
tions do not show the same picture, however, and the correlation
for oxidant and wind speed is higher than that for oxidant and in-
version height. Other calculations have shown that in Los Angeles
there is a strong correlation between inversion height and wind,
and thus it is necessary to use partial correlations to separate
this interrelationship.
SYMPOSIUM: AIR OVER CITIES
-------
232
ROBINSON
Correlations such as these were carried out several times
with Los Angeles observational records for various periods. The
results in general showed less correlation between inversions and
smog than was shown here. Thus it seems that the subsidence
inversion is not as significant a factor in the Los Angeles smog
picture as many writers have claimed.
URBAN CLIMATOLOGICAL CONDITIONS
The statement has already been made that for diffusion of
urban pollution the effects of wind movement are more important
than are the effects of inversion. One apparent explanation for
the lesser role of the inversion is the heat island produced by the
city itself. This phenomenon would produce less stable surface
conditions within the city without its being apparent at the usual
neighboring airport observation point.
This factor of urban climate has been carefully studied by
Duckworth and Sandberg (Figure 1). Their study showed not only
that large temperature gradients occurred at night between open
and urban areas but also that this temperature effect frequently
caused instability up to about 3 times the roof height. The study
also related the magnitude of this heating effect to the size of the
city: the more developed the city, the more it modified its
700
58
so
68
62 64 66
TEMPERATURE, °F
Figure 1. San Francisco wiresonde data for 2210 PST, 26 March 1952, showing soundings
over built-up (B) and adjacent undeveloped (U) areas.
SEC TECHNICAL REPORT A62-5
-------
RELATIVE FACTORS IN URBAN POLLUTION 233
atmosphere; thus the larger the city, the less likely it is to have
surface inversion conditions.
The impact of this urban heating effect on pollution diffu-
sion seems quite clear. The polluted layers are unstable during
the day because of the solar heating of the buildings and pavement
areas, and at night the heat lost by these surfaces is enough to
prevent strong stable conditions from building up. There is prob-
ably not much change in the stability conditions over the central
urban areas. Thus it is not surprising that pollutant concentra-
tions are not highly responsive to stability parameters such as
inversion height or inversion conditions measured outside the
central area.
INTER-CITY COMPARISONS
Another way to consider these urban diffusion factors is to
compare similar pollution data from several different cities.
Data for three California areas are shown in Figure 2: Berkeley
is in the San Francisco Bay Area, Los Angeles is in the Southern
California Los Angeles basin, and Riverside is 50 miles inland
from the Los Angeles basin. The pollutant shown in these data is
the daily oxidant maximum concentration for 1960.
CONCENTRATION, ppn,
0 40 i — >
BERKELEY
0.30
0 20
0 60
O'.SO
0.40
0 30
0.20
0. 10
LOS ANGELES
MAY JUNE JULY AUG SEPT
Figure 2. Daily maximum oxidant concentrations in three California cities, 1960
(potassium iodide method).
If the potential air pollution hazard can be judged from the
yearly maximum value of oxidant, these areas did not differ
greatly in I960: the maximum value in Berkeley was 0. 30; in Los
Angeles, 0. 45; and in Riverside, 0. 37. The frequency of the high
concentration days, however, is considerably different for the
three areas. High values, though rare in Berkeley, were the
general thing in Los Angeles except for the winter months, and
SYMPOSIUM: AIR OVER CITIES
-------
234 ROBINSON
were the normal situation in Riverside during the summer and
fall.
This pattern is not explained by the inversion conditions,
because there is very little difference between those in Los
Angeles and in Berkeley. An important factor that does differ is
the wind. Berkeley has considerably more wind than either Los
Angeles or Riverside, and the wind is normally strongest in mid-
summer. When the wind blows, Berkeley has little pollutant
accumulation as shown by the data; on the few days when the wind
does not blow the pollutant peaks can occur.
A comparison of the Los Angeles and Riverside data indi-
cates the ineffectiveness of thermal mixing as an important factor
in establishing community pollution levels. Riverside is in an
area that regularly is considerably warmer than Los Angeles.
This warmer temperature causes the inversion base to rise
rapidly during the day and to exceed that reached by the inversion
over Los Angeles. Although it is difficult to compare pollution
values without studying the source areas, the data at hand still
might be appraised in the following manner: (a) Since the devel-
opment in Riverside is less than in Los Angeles, lower pollution
levels would be expected; (b) if increased thermal turbulence
were a major additional factor reducing pollution levels at River-
side, concentrations should be significantly lower than in Los
Angeles; (c) since concentrations are not markedly lower, it
seems doubtful that much reduction in air concentration values
ran be attributed to the midday inversion height.
DISCUSSION AND CONCLUSIONS
Several aspects of the stability-wind situation have been
described. First, pollutant concentrations were shown to be
better related to wind conditions than to inversion conditions;
second, it was pointed out that wind conditions normally vary
over a wider range than stability conditions; and third, it was
shown that the city itself is the cause of prevailing unstable con-
ditions in the polluted air. In spite of facts such as these, the
idea is still prevalent that pollution problems go hand-in-hand
with inversion conditions.
The Public Health Service Weather Bureau meteorologists
have made a valuable step in the proper direction with their large-
area air pollution studies. In these studies they have pointed out
that a stagnating anticyclone is the basic characteristic of large-
scale urban pollution incidents. The stagnating High brings to an
area both a weak wind pattern and low-level stability. The pre-
ceding discussion has tried to make the point that the wind pattern
is more important than the stability. Perhaps it would be even
SEC TECHNICAL REPORT A62-5
-------
RELATIVE FACTORS IN URBAN POLLUTION 235
more logical to argue that since a stagnating anticyclone is the
basic meteorological cause of air pollution troubles, then air
pollution meteorology should deal primarily with this weather
system rather than with various individual or local wind and in-
version conditions. The Public Health Service studies take this
more inclusive type of approach. It is to be hoped that their re-
sults will have the needed impact on air pollution thinking.
It seems clear that air pollution meteorologists have been
caught in a trap caused by the popular nature of air pollution
troubles. In an effort to supply meteorological explanations to a
large nontechnical audience, a "popular" description of the
weather factors was provided. A semi-mysterious warm air
layer, called an inversion, was a more interesting concept to pre-
sent to the public than either the prevailing winds or the stagnat-
ing anticyclone. Once the description was provided, it reproduced
itself many times, and repetition made it "authentic. " Progress
toward public understanding of air pollution meteorology depends
upon how rapidly this popularized erroneous description can be
replaced with real definitions.
DISCUSSION
MR. FIELD: I was just curious. How many stations do you
use when you arrive at an oxidant figure for the city of Los
Angeles? And when you say maximum concentration, is that an
average or the peak reached during the day?
MR. ROBINSON: These data were the peak values attained
by a recording instrument situated in downtown Los Angeles.
This is not necessarily an average value for the City of Los
Angeles, which would be more difficult to realize. These data
are probably representative of rather large areas within the city.
MR. FIELD: Well, I am curious. How do you select the
station? Is it simply put there because it is there?
MR. ROBINSON: Mr. Kauper has probably had more to do
with selecting the stations than anybody else here. All I did was
select the data.
MR. KAUPER: That was his first mistake.
I hope I am the reverse of the architect. We advise where
stations go, and then architect types choose stations. It depends
where there is space available, within rather broad limits. In
this case, I think that our downtown Los Angeles station has
varied from the top of our present building, which is in skid row,
six floors up, to the middle of town, considered as Pershing
SYMPOSIUM: AIR OVER CITIES
-------
236 ROBINSON
Square, sitting within a bunch of banana trees, and directly over
a rather large parking lot. The resulting oxidant values at these
various locations have varied considerably. Not because the
instrument was changed, but because the environment was changed.
At the time of this work, in 1960, the station was on the 6th floor,
in a rundown section of town, and slightly east of the main traffic
area.
MR. FIELD: I am just wondering how you can compare
values from three different stations if they are set up without any
regard to conditions you are trying to measure. I am sure if you
set up three stations in one city, you will get three different vari-
ations.
MR. KAUPER: That is what we have now.
DR. NEIBERGER: How many do you have now?
MR. KAUPER: I haven't looked lately but there are prob-
ably about seven -- not in the city, but in the county area. They
range from extreme west to east, and as a rule the oxidant values
seem to come out higher in the direction toward Riverside and
lower toward the coast. As a matter of fact, our downtown Los
Angeles readings are relatively lower than those in the eastern
suburbs region.
MR. ROBINSON: I did not put the data for Anaheim in here
because I didn't want to get into the argument as to how much
Disneyland might have done to the oxidant, but the data did show
that with the eastward transport of the marine air, it continued
to develop higher oxidant concentrations.
MR. HOLZWORTH: I think these data you showed were
from the publication by the State Department of Public Health.
MR. ROBINSON: Yes.
MR. HOLZWORTH: Having been with them for a time, I
am somewhat familiar with their selection of these data. In order
to make as much data available as possible and at the same'time
to cover as much territory of the state as possible, they include
one station for each county. They give consideration to the con-
tinuity of the data, the reliability of the data from each station
(such as breakdown of the instruments, and so on), and to the
representativeness of the stations. For instance at Riverside
County there is more than one instrument, but they chose one and
they hope that that station will remain in the same place for a long
time to come, so that they will have a long comparative record.
DR. NEIBERGER: First, I wanted to say something about
inversions and then I want to ask something about Riverside.
SEC TECHNICAL REPORT A62-5
-------
RELATIVE FACTORS IN URBAN POLLUTION 237
The role of the inversion is obviously a necessary one,
even though it isn't a sufficient one. That is shown by the fact
that even though you get the advection of the air from Los Angeles
over in the desert very often, somehow or other you never get
eye irritation. The same air manages to mix to still higher
levels. So I don't think that you can belittle the inversion; it is a
necessary part of the picture. The fact that winds are important
I think has been recognized. I don't know at whom the criticism
was aimed, but I was among the first to have written on this sub-
ject, about 15 years ago. In a little bit earlier work Leopold and
Veer, who, as far as I know, gave the first study of air pollution
in Los Angeles, mention the inversion and then devote most of
their paper (in the transactions for the AGU for 1945) to the wind
patterns over Los Angeles.
So I think it has been recognized all the time that this vol-
ume of air through which the pollution can go is a function of these
factors: the horizontal dispersion due to the wind, the vertical
dispersion due to the wind, and the limitation to vertical disper-
sion due to the inversion. And I remember many times we said
that the only reason they didn't have this type of smog at all in
those days in San Francisco was that they have twice as much
wind as we do.
But I wanted to ask what specifically you had in mind about
Riverside, because we have at various times examined the trajec-
tories of the air and the behavior of the oxidant as it moves along
from the west coast. At this time of year when we get drifts
towards the coast the worst situation is out in the western part of
the area. But ordinarily, in most of the smog, we follow it from
the west coast, where there is practically zero oxidant, (8 or 9
ppm maximum) into the downtown area, where you get 20 to 30
ppm, then to Pasadena, where it might be 30 to 40 ppm, and then
continuing all the way through to the late afternoon, when the peak
value is reached at Riverside. And the fact is that most of the
time, even though the inversion just about gets wiped out at River-
side, the maximum is still reached there.
I was wondering what type of study you had in mind at
Riverside.
MR. ROBINSON: The point that I was trying to make on
that Riverside-Los Angeles comparison was that if you take the
surface temperatures and extrapolate them against the upper air
soundings you get an increased depth of turbulence, an increased
depth of mixing as the trajectory moves inland. In some of my
own thinking, perhaps, this has tended to play a more important
role than more careful consideration might have placed on it.
This thermal turbulence, which I think is a factor in the Central
SYMPOSIUM: AIR OVER CITIES
-------
238 ROBINSON
Valley of California --in Sacramento and Stockton --is almost
cutting away at the base of the inversion. Increased turbulence
does not seem to be as important a factor in all the air pollution
concentrations as one might expect from reading some of the
technical writings on urban air pollution studies.
SEC TECHNICAL REPORT A62-5
-------
Measurement Programs Required for
Evaluation of Man-Made and Natural
Contaminants in Urban Areas*
E. W. HEWSON, E. W. BIERLY, and
G. C. GILL, Meteorological Laboratories,
The University of Michigan, Ann Arbor,
Michigan
Summary
Existing studies of diffusion in transitional states emphasize the need for more
comprehensive research programs near shorelines, over lakes, over varying terrains, in
valleys, within forests, and over and within uncomplicated cities of various sizes. Be-
cause of the serious public health problems created by aeroal lergens, many more centers
are needed for aeroallergen research. Pollen samplers that minimize nonisokinetic
sampling errors must be developed. Television towers should be widely instrumented,
not only for lapse rates in typical varieties of terrain, but also for contaminant evalua-
tion. Tracer studies are needed to permit direct and accurate determination of diffusion
coefficients. Relatively simple methods are required to assess long-term trends of pollu-
tion by industrial waste products. Methods of calibrating electrical conductivity measure-
ments in terms of air pollution levels should be devised and used widely.
URGENT PROBLEM AREAS
Diffusion In Uniform And Steady States
It is clearly logical to begin the evaluation of atmospheric
diffusion by studying diffusion under the least complex circum-
stances. For this reason most of the comprehensive investigations
undertaken thus far have been confined to the simplest steady-
state conditions. Such studies have been made in England •"•"» '
28, in the United States 2, 4, 19, and in the U. S. S. R. 21, 24 for
uniform and level terrain and for limited time periods, over which
atmospheric conditions were effectively uniform and steady in
space and time.
Within cities, however, the presence of buildings and other
obstructions to air flow leads to substantially different diffusion
patterns. It is unfortunate that virtually no observational data on
diffusion within typical cities are available. A beginning should
be made by conducting a measurement program of atmospheric
'Research conducted under U. S. Public Health Service Grant AP-1 (Publication No. 29)
and under National Science Foundation Grant G-11404
Publication No. 57 from the Meteorological Laboratories of The University of Michigan
239
-------
240 HEWSON, BIERLY AND GILL
diffusion with a city under the simplest possible conditions. The
city should be one on a flat terrain, without topographic features,
with statistical uniformity of building type and size. The study
should be conducted near the center of the city, away from edge
effects. Appropriate times for the field investigations should also
be chosen. The most satisfactory periods of the day will be from
0 to 4 hours and from 12 to IS hours, and the best meteorological
conditions will have wind speed and direction most nearly constant,
so that steady-state conditions most nearly prevail. Later studies
would be made in succession in cities having increasingly more
complex terrain and structural diversity.
Some of the techniques and instrumentation required for such
studies will be discussed in later sections.
Diffusion In Transitional States
Most of the existing diffusion theories are based on the as-
sumption that turbulent diffusion is invariant with time and space,
a basis that requires exclusion of a wide range of situations that
are of the utmost importance in urban air pollution. The atmos-
phere is a dynamic entity whose diffusional characteristics may
and often do change radically within short periods of time. When
the field of atmospheric turbulence exhibits marked variations in
time or space or both, then the resulting atmospheric diffusion
may be referred to as "diffusion in transitional states. "
Diffusion in River Valleys The first investigation of diffusion in
transitional states was conducted in the Columbia River Valley
near Trail, British Columbia. 16 Two examples were discovered,
one associated with a space variation of turbulence and the second
with a time variation. The space variation resulted from a dry
adiabatic lapse rate in the lower portion of the valley surmounted
by an inversion aloft but in the valley. Although turbulence and
mixing were pronounced below, high concentrations of SO2 occurred
at the surface because the inversion aloft prevented upward diffu-
sion out of the valley. A similar condition has since been shown
to be the primary contributing factor in Los Angeles smog. Con-
tainment of pollutants beneath an inversion aloft even when terrain
does not limit horizontal mixing has been named "trapping. " ^ 15
The time variation within the Columbia River Valley oc-
curred during the morning hours in summer, when solar heating
caused a turbulent layer to develop upward from the surface, re-
placing an inversion. When this turbulent layer reached the smoke
SEC TECHNICAL REPORT A62-5
-------
MEASUREMENTS FOR URBAN STUDIES 241
plume aloft, it brought high concentrations of SO2 to the valley,
a process known as "fumigation" caused by "inversion breakup
diffusion. " H Many other examples of this process over flat
terrain have since been found.
Diffusion Near Shore Lines Because of the need for an adequate
water supply, river valleys and shore lines are preferred loca-
tions for cities and industries. Some of the aspects of the "diffu-
sion in transitional states" in river valleys have been mentioned
above, but relatively little research on such diffusion near shore
1 7 1 R
lines has been conducted, although a start has been made. >
The construction of nuclear power plants at shore-line lo-
cations on the Great Lakes has led to measurement programs that
will establish the air pollution climatology of these areas. The
authors are engaged in such measurement programs at the Enrico
Fermi Atomic Power Plant near Monroe, Michigan, at the west
end of Lake Erie and at the Big Rock Point Nuclear Power Plant
near Charlevoix, Michigan, on the east side of Lake Michigan.
The latter is shown in Figure 1, an aerial photograph showing the
plant in the foreground and the 250-foot instrumented meteorolog-
ical tower behind it, very near the water's edge. Figure 2 shows
the steel tower and wooden pole as viewed from the lake. Ane-
mometers and wind vanes are installed at 32 and 64 feet on the
pole and at 128 and 256 feet on the tower. Water temperature at
3 feet beneath the lake surface and air temperature at 10, 50, 100,
150, 200, and 250 feet on the tower are measured for lapse-rate
determinations. All winds and temperatures are recorded in the
small building at the base of the tower.
At a shore line the horizontal distribution of the diffusion
field is the important factor. Marked horizontal variations in
turbulence may be due to horizontal differences in the tempera-
ture of the underlying surface; these temperature differences in-
duce or suppress thermal turbulence. Alternatively, such horizon-
tal variations in turbulence may be due to differences in the rough-
ness of the underlying surface; these differences also can cause
mechanical turbulence to increase or to decrease in intensity. At
a shore line both surface temperature differences and surface
roughness differences may be pronounced and may combine to
produce highly complex patterns of diffusion.
Two main types of diffusion in transitional states should be
considered: one with the horizontal distribution of turbulence sta-
tionary; the second with the horizontal distribution varying with
time.
SYMPOSIUM: AIR OVER CITIES
-------
242
HEWSON, BIERLY AND GILL
Figure 1. Aerial view of the Big Rock Point Nuclear Power Plant near Charlevoix,
Michigan, with 250-foot instrumented meteorological tower near the shore line.
SEC TECHNICAL REPORT A62-5
-------
MEASUREMENTS FOR URBAN STUDIES
243
Fijure 2. View of 250-foot meteorological tower and instrumented pole along the lake
shore at the Sig Rock Point Nuclear Power Plant.
SYMPOSIUM: AIR OVER CITIES
-------
244 HEWSON, BIERLY AND GILL
1. Stationary Horizontal Distribution. There is a natural
subdivision in this category near shore lines. The most general
case is horizontal variation primarily along wind. In this case
the diffusing plume is entirely over land or entirely over water,
A particular but important case is horizontal variation primarily
across wind. Here the wind direction is parallel to the shore line,
so that one side of the plume has a land trajectory and the other
side a water trajectory.
2. Horizontal Distribution Varying with Time. This type of
variation may be associated with such factors as the diurnal vari-
ation of wind speed and lapse rate, a frontal passage, or a shift
in wind direction resulting in a change of upwind surface rough-
ness. The varying field of air flow associated with a regime of
land and the lake breezes near a shore line is an excellent example
of this type of diffusion in transitional states.
Three aerial photographs of smoke released from the mete-
orological tower at the Big Rock Point Nuclear Power Plant illus-
trate very well some of the complex diffusion patterns in transi-
tional states with a stationary horizontal distribution as specified
above. These photographs, taken in July 1961, are shown in Fig-
ures 3, 4, and 5. Figure 3 shows the behavior of the smoke plume
with a northwest wind from over the lake. The initial spreading
of the smoke is limited, but diffusion becomes rapid as mechanical
and thermal turbulence develop over land. Figure 4 illustrates
looping of the plume under unstable conditions with thermal turbu-
lence in southwest winds, with the plume being brought down to
the lake surface. Figure 5 is an especially striking photograph
which shows the behavior of the plume with a west to west-south-
west wind. Initial diffusion is limited but is followed by very
pronounced lateral spreading of the plume.
Detailed analyses of shore-line effects found at the site of
the Enrico Fermi Atomic Power Plant have been given by the
i n -| o o i/
authors. ^ '> ±0 The general findings to date may be summarized
as follows:
1. Values of Button's virtual diffusion coefficient GZ are
characteristic of those for a 3-minute sampling period,
2. Values of Button's virtual diffusion coefficient C are
characteristic of those for a 1-hour sampling period.
3. The wind at 96 feet is relatively strong, 12. 4 mph
averaged over a 3-year period, in comparison with that at the
same height over a uniform land surface.
4. A lake-breeze inversion occurs frequently in late spring
and early summer.
SEC TECHNICAL REPORT A6S-5
-------
MEASUREMENTS FOR URBAN STUDIES
245
Figure 3. Smoke plume behavior with a northwest wind from over the lake at the Big Rock
Point Nuclear Power Plant.
Figure 4. A looping plume caused by thermal turbulence with southwest winds at the Big
Rock Point Plant.
SYMPOSIUM: AIR OVER CITIES
-------
246
HEWSON, BIERLY AND GILL
Figure 5. Lateral shear in the smoke plume at the Big Rock Point Plant with west to
west southwest winds.
5. This lake-breeze inversion moves inland about 4 miles
on the average, and diffusion improves with distance inland from
the shore line.
6. Prolonged inversions caused by the advection of warm
air over the cold lake occur, but extend no more than 8 miles in-
land, with improving diffusion over that distance.
7. During stagnant anticyclones the difference in horizontal
air density from water to land results in local winds of sufficient
strength to provide substantial natural ventilation.
8. Measurements to date indicate that there is considerable
natural ventilation at the shore line of a large lake.
Although the researches mentioned above were not conducted
near a city, it is clear that a city situated on a shore line will be
subjected to a number of localized meteorological influences; as-
certaining their exact nature will require comprehensive measure-
ment programs.
SEC TECHNICAL REPORT A62-5
-------
MEASUREMENTS FOR URBAN STUDIES 247
Diffusion in Other Transitional States Diffusion in transitional
states near shore lines has been singled out for special attention
because the authors are engaged in field measurement programs
near shore lines. It is clear, however, that the study of diffusion
in many other types of transitional states is of fundamental im-
portance for an adequate understanding of the dispersive processes
at work in the air over cities. For example, diffusion within
wooded and forested areas has received relatively little attention.
Trees have a pronounced effect on local stability patterns in the
air below, around, and above them and in addition have a filtering
action in removing particulates. Although there have been a num-
q c 1 f^
her of studies of air flow in valleys'3' Oj , many more measure-
ments of wind speed, direction, turbulence, and temperature
lapse rate in and near cities situated in valleys are urgently
needed. The influence of valley cities in heating the air flowing
over them and the possible effects of this heating on air flow
patterns during very light winds should be investigated thoroughly
for such cities.
Atmospheric Pollution by Aeroallergens
It is estimated that some ten million Americans are seri-
ously inconvenienced by aeroallergens, airborne substances which
cause allergic reactions in sensitive individuals. Ragweed pollen
is probably the worst offender among the aeroallergens. There
is good evidence that man's land-use practices are leading to in-
creased production and wide aerial distribution of ragweed pollen.
It is surprising, therefore, that practically no attention has been
paid to this important public health problem until recent years,
especially since remedial measures can be taken and others un-
covered if adequate research is conducted.
Meteorological aspects of the production, transport, and
deposition of ragweed pollen are being investigated at two centers:
the University of Michigan and Brookhaven National Laboratory.
The importance of aeroallergens as a national public health prob-
lem requires the establishment of other centers for the study of
the many complex problems involved.
Among the techniques developed at Michigan is that of grow-
ing preseasonal ragweed plants that pollinate in June, well before
the regular pollen season, which extends from mid-August through
September. Since the preseasonal plants are the only ones pollin-
ating in the area, the source of pollen in the air is clearly estab-
lished as the experimental ragweed plot. Such a plot, containing
over 3000 ragweed plants developed from seedlings in the green-
house and located on the farm lands of the State Prison of Southern
SYMPOSIUM: AIR OVER CITIES
-------
248 HEWSON, BIERLY AND GILL
Michigan at Jackson, is shown in Figure 6.
Figure 6. Ragweed pollen plot at the State Prison of Southern Michigan, Jackson,
Michigan.
Determining the sources of pollen found within cities is an
important problem that awaits solution. Does the pollen come
from ragweed plants within the city, which might be destroyed by
spraying, or from sources well outside the city, which would be
less amenable to control? Some of the features of an experimental
program designed to find answers to such questions are illustrated
in Figure 7. The map shows the route followed by an instrumented
automobile each day during the regular ragweed season as it trav-
eled through Ann Arbor into the adjacent farm lands and back to
the city. 25 The shaded areas along the route give the vegetative
characteristics of the farm lands. The density of ragweed stands
was estimated by the botanists associated with the project. The
results of this study are being prepared for publication.
The production and release of ragweed pollen by the plants
occurs mainly from 7 to 11 a.m. on clear mornings. Precipitation
or prolonged fog in the morning will delay and reduce substantially
the pollen emission. Our research has shown that meteorological
conditions during May and June have a significant influence on the
subsequent pollen crop. Objective forecasting methods that are
SEC TECHNICAL REPORT A62-5
-------
MEASUREMENTS FOR URBAN STUDIES
249
now being developed hold promise of yielding forecasts of pollen
concentrations on both a daily and a seasonal basis.
;>SSX RURAL, TILLED
|§g§S RURAL, UNTILLEO
'•-f.^f URBAN
Figure 7. Route of mobile sampling program through Ann Arbor, Michigan, and surround-
ing areas.
There is evidence that the sensitivity of many individuals
to ragweed pollen is a function of the current meteorological con-
ditions. This phase of the problem has been barely touched, and
many measurement programs will be required before significant
progress is made.
Since source strength, atmospheric dispersion, and receptor
reaction are all dependent on fluctuating atmospheric conditions,
it may be that the pollen dispersion problem represents the tran-
sitional state problem in its ultimate meteorological complexity.
SYMPOSIUM: AIR OVER CITIES
-------
250 HEWSON, BIERLY AND GILL
Although ragweed pollen is, medically speaking, the most
important of the aeroallergens, there are many others, including
spores, rusts, and smuts. Carefully devised measurement pro-
grams are needed if we are to make progress in combating this
growing public health problem.
RECOMMENDED MEASUREMENT PROGRAMS
The following outline of recommendations is not meant to be
inclusive. It represents only those programs that are urgently re-
quired if continued progress is to be made. Comprehensive cover-
age of the instrumentation currently available and in use, its ad-
vantages and disadvantages, methods of mounting, etc., is avail-
able elsewhere. 14 For this reason only general outlines of the
needed programs are given here.
Standardization Studies
Such a variety of measurements have been proposed and
undertaken from time to time that a serious attempt to achieve
even limited standardization of requirements is an important next
step. Valuable progress in that direction has recently been made,''
but even more fully detailed specifications are now required, It
may be necessary to set up a specific project for the purpose at
some appropriate center, with provision for consultants to aid in
the essential decision making. Examples of specifications needed
are as follows: the height and exposure of wind sensors, permis-
sible limits in the response characteristics of wind sensors used
to measure atmospheric turbulence, and minimum accuracy re-
quirements for lapse-rate temperature measurements.
Air Pollution Climatology Studies
It is essential to establish air pollution climatologies for a
number of cities of various sizes subject to various climatic
regimes and with a variety of terrain features. Many of the
problems described in earlier sections will be solved most ex-
peditiously by an intensive program of installing appropriate
meteorological and sampling equipment on existing and future
television towers.
Meteorological and sampling programs that utilize TV towers
have been proposed for a number of years, 12 but little progress
has been made in implementing the proposals. H. W. Baynton and
associates were responsible for instrumenting the first tower,
WJBK-TV, located in northwestern Detroit, as part of the Detroit-
Windsor International Air Pollution Investigation. 20 This
SEC TECHNICAL REPORT A62-5
-------
MEASUREMENTS FOR URBAN STUDIES 251
installation was later placed under the care of the Meteorological
Laboratories of The University of Michigan. Temper ature-
measuring units are installed at heights of 20, 300, 600, and 870
feet. The top portion of the 1050-foot tower consists of the radi-
ating antenna. Figure 8 is a photograph of the tower showing the
platforms at 300, 600, and 870 feet, where the artificially venti-
lated temperature sensors are mounted. Figure 9 shows the
lowest sensor mounted beneath the platform at 20 feet. It is pro-
posed that all the sensors be relocated on booms extending hori-
zontally from the tower to minimize the possibility of errors due
to radiational heating or cooling of the tower during the day or at
night.
A recent detailed study of air pollution in Nashville, Tennessee,
led to the installation of wind speed and direction transmitters at
251-and 501-foot levels on the WSM-TV tower. 30
Until more definitive standards for tower instrumentation
can be established, it is proposed that temperatures or temper-
ature differences be measured at 200-foot intervals if possible,
with an accuracy of 0. 20 . Because of the positions of tower plat-
forms, instruments may have to be spaced at greater intervals.
Anemometers and wind vanes should be mounted at the same
heights, and the response characteristics of the vanes should be
sensitive enough to justify their use as accurate sensors of the
horizontal component of atmospheric turbulence. A high-grade
turbulence sensor, such as a rapid response bivane without ex-
cessive overshoot for all ranges of turbulent eddies, should be
maintained at the station at a level of 200 or higher. All data
should be recorded at the base of the tower.
Installations of this type should be established at a number
of cities of various sizes and with various climatic and terrain
features, With such installations in operation it will be possible
to study diffusion over cities in uniform and steady states and in
a wide variety of transitional states.
Supplementary wind and turbulence data would be obtained
from wind sensors on 30-foot (10-meter) masts at a limited
number of other stations in and near the city and in the streets on
horizontal cables stretched between buildings, well above the im-
mediate influence of street traffic. Precipitation measurements
should be taken by means of one or more recording rain gages in
locations with good exposures.
It must be emphasized that well-designed programs of at-
mospheric sampling must supplement the meteorological measure-
ments if results are to be significant.
SYMPOSIUM: AIR OVER CITIES
-------
HE\VSON, BIERLY AND GILL
Figure 8. View of the WJBK-TV tower, Detroit, Michigan.
SEC TECHNICAL REPORT A62-5
-------
MEASUREMENTS KOH URBAN STUDIES
253
Figure 9. Detail of the WJBK-TV tower showing the lowest sensor beneath the platform
at the 20-foot level.
Aeroallergen Studies
The prime requirement for advance in aeroallergen studies
is the development of a simple and accurate sampler for pollens
such as those of ragweed. An isokinetic sampler 9 is required
for accurate sampling of ragweed pollen (18/u), but such a sampler
presents severe design problems. The sampler should be able to
operate reliably without attention for a minimum period of 24 hours
if it is to displace the simple but highly inaccurate gravity slide
sampler.
A number of other measurement programs are needed, such
as those designed to determine the variation of ragweed pollen
source strength as a function of meteorological conditions, the
distance traveled by such pollen, the role played by precipitation
in scavenging, and the filtering action of natural vegetation . .11 pol-
lens and oilier aeroallergens.
SYMPOSIUM: AIR OVER CITIES
-------
254 HEWSON, BIERLY AND GILL
Tracer Studies
An important technique in evaluating the dispersive prop-
erties of the atmosphere is the use of tracer substances such as
ZnCdS as a fine powder or liquid uranine dye. Such tracers may
be used to follow air motion and turbulent diffusion over a wide
range of distances from a hundred feet or so, representing the
distance between a source and an air conditioning intake, to a
hundred miles or so, representing diffusion over a megalopolis
such as that along the east coast.
Model Studies
Model studies in wind tunnels have been most useful in the
past in estimating the amount of aerodynamic downwash of plume
gases occurring in the lee of buildings. ' A second important
application is in delineating the details of air flow around buildings,
Much remains to be done in using wind tunnel model studies to
predict the effect of nearby terrain features on the movement of
contaminants in their vicinity. Existing programs should be ex-
tended and new programs should be established to permit wide
and more effective use of this powerful tool.
Turbidity and Conductivity Studies
Certain types of studies are required to assess long-term
trends in urban pollution levels. A network of turbidity stations
has been established, but the areal distribution of this network
o o
should be increased. It has been known for some time that the
electrical conductivity of the atmosphere is related to its particu-
late pollution content. 2" The use of this method to assess long-
term trends in pollution was strongly recommended a number of
years ago, 12 but no action has resulted. Calibration methods
should be developed and an active program for measurement of
electrical conductivity in the atmosphere and study of its relation
to particulate pollution should be established in a number of cities,
Carbon Dioxide Studies
The possible influence on our climate of increased CC>2 in
the atmosphere resulting from our combustion of fossil fuels
should be thoroughly studied. For an adequate program, CC>2
should be measured on a routine basis year after year at a number
of points far from sources, such as large cities. Routine observa-
tions of CC>2 on a number of oceanic weather ships would be par-
ticularly helpful.
SEC TECHNICAL REPORT A62-5
-------
MEASUREMENTS FOR URBAN STUDIES 255
REFERENCES
1. Bierly, E. W. , and E. W. Hewson, 1962: Some Restrictive
Meteorological Conditions to be Considered in the Design of
Stacks, Submitted for publication.
2. Bowne, N. E., 1961: "Some measurements of diffusion
parameters from smoke plumes, " Bull. Amer. Meteor.
Soc., 42, 101-105.
3. Buettner, K. J. K. , and N. Thyer, 1959: On Mountain and
Valley Winds, Final Report, Contract AF 19 (604) 2289,
AFCRC-TR-59-283, Department of Meteorology and Clima-
tology, University of Washington.
4. Cramer, H. E., F. A. Record, and H. C. Vaughan, 1958:
The study of the diffusion of gases or aerosols in the lower
atmosphere^ Cambridge, Mass. Inst. of Tech. , AFCRC-
TR-58-239.
5. Davidson, B., 1961: "Wind valley phenomena and air pollution
problems," J. Air Poll. Control Assoc., 11, 364-368, 383.
6. Dingle, A. N., G. C. Gill, W. H. Wagner, Jr., and E. W.
Hewson, 1959: "The emission, dispersion, and deposition
of ragweed pollen, " Adv. Geophysics, 6, 367-387.
7. Gill, G. C. , H. Moses, and M. E. Smith" 1961: "Current
thinking on meteorological instrumentation for use in air
pollution problems, " J. Air Poll. Control Assoc. , 11,
77-82, 96.
8. Halitsky, J. , 1962: "Some Aspects of Atmospheric Diffusion
in Urban Areas, " Symposium: The Air Over Cities,
9. Harrington, J. B. , Jr., G. C. Gill, and B. R. Warr, 1959:
"High-efficiency pollen samplers for use in clinical allergy, "
J. Allergy, 30, 357-375.
10. Hay, J. S. , and F. Pasquill, 1959: "Diffusion from a con-
tinuous source in relation to the spectrum and scale of tur-
bulence, " Adv. Geophysics, 6, 345-365.
11. Hewson, E. W., 1945: "The meteorological control of at-
mospheric pollution by heavy industry, " Quart. J. R.
Meteor. Soc., 71, 266-282.
12, Hewson, E. W. , 1952: Meteorology Panel, in Air Pollution
L. C. McCabe, Ed., New York, McGraw-Hill, 10-13.
13. Hewson, E. W. , 1955: "Stack Heights Required to Minimize
Ground Concentrations, " Trans. Amer. Soc. Mech. Engrs. ,
77_, 1163-72.
14, Hewson, E. W. , 1962: "Meteorological measurements, "
Air Pollution, A Comprehensive Treatise, Vol. I, A. C.
Stern, Ed., New York, Academic Press, 528-567.
SYMPOSIUM: AIR OVER CITIES
-------
256 HEWSON, BIERLY AND GILL
15. Hewson, E. W. , (in press): "Meteorological measuring
techniques and methods for air pollution studies, " Industrial
Hygiene and Toxicology, Vol. 3. L. Silverman, Ed. , New
York, Interscience.
16. Hewson, E. W. , and G. C. Gill, 1944: "Meteorological inves-
tigations in Columbia River Valley near Trail, B. C. , "
Report submitted to the Trail Smelter Arbitral Tribunal by
R.S. Dean and R.E. Swain. Bull. U. S. Bur. Mines No.
453, 23-228.
17. Hewson, E. W. , G. C. Gill, and E. W. Bierly, 1960:
Atmospheric diffusion study at the Enrico Fermi Nuclear
Reactor Site. Ann Arbor, Univ. of Michigan Research
Institute, Report No. 2728-3-T, 19 pp.
18. Hewson, E. W. , E. W. Bierly, and G. C. Gill, 1961:
"Topographic influences on the behavior of stack effluents, "
Proceedings of the American Power Conference, 1961, 23,
358-370.
19. Hilst, G. R. , and C. L. Simpson, 1958: "Observations of
vertical diffusion rates in stable atmospheres, " J. Meteor.
15, 125-126.
20. International Joint Commission, 1960: "Relation between
pollution levels and meteorological factors, " Report of
the International Joint Commission, United States and
Canada, on the pollution of the atmosphere in the Detroit
River area, 164-206^
21. Kanzanskii, A. B. , and A. S. Monin, 1957: "On the shape
of smoke plumes, " Trans, from Bulletin (Izvestiya) of the
Academy of Sciences of the U. S. S. R., Geophysics Series,
No. 8, 56-70.
22. Lowry, P. H. , 1951: "Microclimate factors in smoke pol-
lution from tall stacks, " Meteor. Monogr. , 1, No. 4, 24-29.
23. McCormick, R. A., 1960: (Personal communication).
24. Monin, A. S. , 1959: "Smoke propagation in the surface
layer of the atmosphere, " Adv. Geophysics, 6, 331-343.
25. Sheldon, J. M. , and E. W. Hewson, 1959: Atmospheric
pollution by aeroallergens. Ann Arbor, Univ. of Michigan
Research Institute, Report No. 2421-3-P, 103pp.
26. Sherlock, R. H. , and E. J. Lesher, 1955: "Design of
chimneys to control down-wash of gases, " Trans. Amer.
Soc. mech. Engrs., 77, 1-9.
27. Stewart, N. G., H. J.~Gale, and R. N. Crooks, 1958: "The
atmospheric diffusion of gases discharged from the chimney
of the Harwell reactor BEPO, " Int. J. Air Poll. , _!, 87-102.
28. Button, G. G. , 1947: "The theoretical distribution "of air-
borne pollution from factory chimneys, " Quart. J. R.
meteor. Soc., 73, 426-436.
SEC TECHNICAL REPORT A62-5
-------
MEASUREMENTS FOR URBAN STUDIES 257
29. Wait, G. R., 1946: "Some experiments relating to the
electrical conductivity of the lower atmosphere, " J. Wash.
Acad. Sci. , ^6, 321-343.
30. Zeidberg, L. D. , J. J. Schueneman, P. A. Humphrey, and
R. A. Prindle, 1961: "Air pollution and health: general
description of a study in Nashville, Tennessee, " J. Air
Poll. Control Assoc. , 11, 289-297.
DISCUSSION
DR. SCHMIDT: What sort of thermometers do you use in
your television masts?
PROFESSOR HEWSON: I can tell you about our own
installations, We have used thermocouples more widely.
Ted Munn, what was on the TV tower?
MR. MUNN: A Honeywell Resistance Thermometer.
DR. SCHMIDT: We used thermometers also on television
masts, but we had difficulties as a consequence of the large elec-
trical field. It was unmanageable.
MR. MUNN: We have had this problem with radio towers
but not television towers.
PROFESSOR HEWSON: There was some difficulty with the
upper levels, but I think that was the shielding that was developed
at the point. This is an important aspect to keep in mind to make
sure that it is covered adequately.
SYMPOSIUM: AIR OVER CITIES
-------
The Representativeness of Local
Observations in Air Pollution Surveys
MAYNARD E. SMITH, Meteorology Group,
Brookhaven National Laboratory,
Upton, Long Island, N. Y.
Summary
The need for appropriate meteorological data in the evaluation of air pollution
problems is now generally conceded. The natural tendency is to use existing observa-
vational data whenever the material seems pertinent. In investigations by the Brookhaven
Meteorology Group, data from a number of sources, including tower and surface observa-
tions, have been compared. Depending on the circumstances, both the differences and
similarities are striking. Enough information is now available to indicate situations in
which data may be validly transferred from one site to another and those in which transfer
should not be attempted. Careful estimates of wind and stability distributions for sites
with complex terrain may often be far superior to an inappropriate transfer of data.
With the constantly increasing interest in air pollution,
there is an associated need for appropriate meteorological data
for evaluation of the problems. Obtaining such information in
its original form is time-consuming and expensive, and the
question of the suitability of existing information naturally arises.
In this paper, some of the problems involved in transferring data
from one site to another are considered, and examples of suita-
ble and unsuitable transfer are shown.
In the meteorological program at Brookhaven National
Laboratory, the opportunity has arisen for the examination of
data from various locations, and in this study two sets of records
are used. The first includes wind data obtained from three ele-
vated installations along the east coast of the United States. The
maximum separation of these sites is about 150 miles, but all are
in rather flat terrain and located near the coast. The locations
of these three sites are shown in Figure 1, and are designated
as Brookhaven, South Norwalk, and Delaware City.
The meteorological instruments from which the wind data
were obtained were all mounted on structures above the ground.
The instruments at Brookhaven were supported on a meteorolog-
ical tower, those at South Norwalk on a tank, and those at Dela-
ware City on a refinery structure. Heights of the instrumenta-
tion ranged from 150 to 350 feet above ground. The other group
of records was obtained from three local airports in western New
York State, all in rolling but not rugged terrain, and located
259
-------
260
SMITH
Figure 1. Coastal Sites
Figure 2. Inland Sites
somewhat closer to each other than the coastal stations. Figure
2 shows the relative positions of the inland stations: Dansville,
Jamestown, and Olean. At these sites the meteorological equip-
ment was mounted about 30 feet above ground, so that all of the
data are low level.
SEC TECHNICAL REPORT A62-5
-------
REPRESENTATIVENESS OF OBSERVATIONS
261
In comparison of the data, only wind directions are used,
both because the height differences of the coastal stations are con-
siderable, and because at least one of the anemometers in the
inland group probably was defective.
COASTAL STATIONS
Figures 3 and 4 represent wind roses at the three coastal
stations during winter and summer, respectively. Although the
figures show a distinct seasonal change, the wind roses of the
y<
*i.o*Y_ ( °-0% ]
/r^
Irookhaven, N. Y.
Wilmington, Del.
S. Norwalk, Conn.
Figure 3. Coastal Sites, December/ January and February
0.0%
/r
Srookhaven, N.Y.
Wilmington, Del.
S. Norwalk, Conn.
10 19 20 %
Figure 4. Coastal Sites, June, July and August
SYMPOSIUM: AIR OVER CITIES
-------
262 SMITH
three stations are quite similar. In the winter, each of them
shows a definite NW maximum with important contributions from
SW, W, N, and NE as well, but none of the stations is particularly
affected by winds ranging from E through S. In the summer wind
roses, Figure 4, similarities are also apparent, although the
South Norwalk record shows a little more frequency of S and NW
winds as opposed to a south-westerly flow, which is predominant
at Brookhaven and Delaware City. The harbor orientation favors
a more southerly sea breeze at South Norwalk, and a drainage
wind down a shallow valley at night contributes to the northwest-
erly maximum.
Figures 3 and 4 show that calm winds (entered as percent
in the center of the circle) are exceedingly rare at all stations.
INLAND HILLY STATIONS
Similar wind roses for winter and summer are presented in
Figures 5 and 6 for the hilly inland stations: the direction distri-
bution for these stations is shown to 16 rather than 8 points. One
would be hard pressed to identify these wind roses as coming from
the same general area of the country. The differences are marked,
with Dansville showing a strong tendency for NW-SE flow, James-
town generally recording winds from S through NW, and Clean re-
flecting sharp peaks from the SW and E, especially in the summer.
The Jamestown station shows a very high percentage of
calms, which may reflect a defective instrument, since no other
reason is apparent.
These marked differences cannot be explained on any gen-
eral meteorological grounds, but the reasons for them become
clear on inspection of terrain maps of the local areas, Figures
7, 8 and 9. Dansville, shown in Figure 7, lies in a pronounced
NNW-SSE valley, which channels the flow very sharply. Pre-
sumably there would be a strong diurnal variation in this valley
with drainage winds from the SSE at night and up-valley winds
from the NNW in the daytime. The map of Olean, Figure 8 shows
quite clearly the reason for the predominance of east and west
winds at the site, since the main configuration of the valley favors
such flow. It is not obvious on a map of this scale, but closer
inspection of detail would show a sharp cut directly south of the
station; presumably the strong tendency toward S winds represents
a drainage wind down the slope. The map of Jamestown, Figure 9,
is not quite so easy to interpret, since the terrain features are
not as sharply marked as those of the other two sites. Particularly,
the very high percentage of calms is difficult to explain, and for
this reason it is thought that the anemometer may have been
defective.
SEC TECHNICAL REPORT A62-5
-------
REPRESENTATIVENESS OF OBSERVATIONS
263
2.7%
Jamestown, N.Y.
Clean, N.Y. Dansville, N.Y
Figure 5. Hilly Sites, December, January and February
Jamestown, N.Y. Clean, N.Y.
Dansville, N.Y.
Figure 6. Hilly Sites, June, July and August
SYMPOSIUM: AIR OVER CITIES
-------
264
SMITH
Figure 7. Terrain map — Dansville, N.Y.
Figure 8. Terrain map — Olean, N.Y.
SEC TECHNICAL REPORT A62-5
-------
REPRESENTATIVENESS OF OBSERVATIONS
265
Figure 9. Terrain map — Jamestown, N.Y.
CONCLUSIONS
This brief review indicates that prospects for the use of
existing meteorological data for air pollution studies are both
encouraging and discouraging. Essentially the main conclusion of
the study is that in each case that requires data, the decision as
to the means of acquiring it should be made by a person familiar
with micrometeorology. If the terrain and altitude considerations
warrant it, transfer of data among similar sites, even over a
fairly wide area, is a perfectly reasonable procedure. Certainly
from a site-survey point of view, the wind roses of Brookhaven,
South Norwalk, and Delaware City are substantially identical.
On the other hand, it seems equally clear that there is little
point in devoting any study to data transfer in rough terrain. A
careful inspection and evaluation of the site may produce an
estimate of the wind distribution which would be superior to that
obtained from a misguided attempt to transfer data.
SYMPOSIUM: AIR OVER CITIES
-------
266 SMITH
DISCUSSION
PROFESSOR HEWSON: There is one type of variable that
shows up in that Jamestown wind rose that I don't think you men-
tioned. Is there not a possibility that since the peaks in the wind
rose are from west, southwest, south, and so on, and you had
minima in between these points -- may this not indicate a bias on
the part of the person who was reading the observations? I
noticed this, and my interpretation was that it was perhaps easier
to write southwest or west rather than south southwest.
MR. SMITH: I think that anyone who has ever done any
reading of such data would have to agree with you. This is one
of the points that I wanted to make really, that when you select
information of this kind you accept the problems as well as the
data. I think any one of us will show a bias, particularly some-
one who isn't scientifically curious about it. That certainly could
be what happened in that case.
We performed a study on the problem some time ago; we
had two girls read two full years of wind-speed data, and they
were to read this information to a tenth of a meter per second.
The records were perfectly satisfactory for this. If you look
through the data and make a frequency distribution, of course,
0. 5 and 0. 0 show up far more frequently than anything else. There
is no question about it.
MR. KALPERN: What type of stations were these actually?
Were they 24-hour observations?
MR. SMITH: Yes, they were.
MR. KALPERN: Were they Weather Bureau stations?
MR. SMITH: They were Weather. Bureau stations or
Airways stations.
MR. KALPERN: What period of time ?
MR. SMITH: The records I showed you covered a period of
two or three years.
MR. KALPERN: How long ago?
MR. SMITH: I think the earliest date was about 1956 and
the latest date was very recent.
SEC TECHNICAL REPORT A62-5
-------
Present and Future Needs for
Meteorological and Air Quality
Observations in Canada*
R. E. MUNN, Meteorological Branch,
Department of Transport, Canada
Summary
Air pollution meteorological data can-
not be assessed without knowledge of the ob-
jectives of urban air pollution studies. A set
of objectives and requirements for meteorolog-
ical support is outlined. Some typical Canadi-
an air pollution surveys now in progress are
discussed. Field data are assessed and rec-
ommendations made. The advantages of na-
tional or international uniformity in methods
of measuring, reporting, and storing air pollu-
tion meteorological observations are presented.
*To be published in full by the UNESCO Conference on Science and Technology for
Underdeveloped Countries, Geneva, February 1963.
267
-------
Problems Associated with Forecasting Air
Pollution Over an Urban Area
ERWIN K. KAUPER, Air Pollution Control
District, County of Los Angeles
Summary
The need for a more specialized observational program scaled for air pollution
studies is discussed in terms of the operational air pollution forecasting service in the
Los Angeles Basin area of Southern California.
Any air pollution study requires an accompanying body of
meteorological data. This holds true both for the simple survey
that determines whether there is air'pollution, and for the all-
out effort of continuous measurement designed to detect pollutant
levels and alert a community to the approach of pre-determined
danger levels.
In none of these situations is one likely to find the meteor-
ological data in the form and at the location desired. This is a
characteristic of such data as presently gathered. The location
of weather stations and the type of observations reflect a preoccu-
pation with aviation problems. In former days, agricultural
interests received all the attention. That was the period of empha-
sis on temperature and precipitation reports gathered by a multi-
tude of cooperative stations. The city office was in its ascendency,
although it served more as a climatological clearing house than a
source of detailed weather information about the city itself.
The airport weather station has now replaced the city office
at most urban centers. What the study of air pollution needs,
though, is a return to the city office, a city office that conducts
an observational program geared to urban problems.
In this paper, we will discuss these problems in the light of
the forecasting program conducted by the Los Angeles County Air
Pollution Control District.
GENERAL DISCUSSION
Instead of forecasting the general weather of whole states or
sections of the country, the Air Pollution Control District concen-
trates on predicting the air pollution conditions over one coastal
basin. Through inter-agency agreements, Los Angeles County
provides meteorological forecast service to two of the three other
counties that share the region included in the term, Los Angeles
-------
270 KAUPER
Basin. Altogether, then, the forecast area encompasses about
2400 square miles, running from the Pacific Ocean on the west
and south, inland 77 miles to the fringing mountains to the east,
and 31 miles to the mountain boundary on the north.
Beyond these mountains lie the California deserts that to-
gether with the Pacific Ocean control the ebb and flow of the air
over the Basin, a transport mechanism that directs the movement
of the Los Angeles smog cloud from its first formation to its final
removal.
The forecast problem in this rather limited area basically
is the prediction of two meteorological conditions - the horizontal
wind flow and the stability of the air mass. Knowledge of these
conditions and of the amount of solar radiation that will be avail-
able at the surface should theoretically produce a good forecast
of the resulting air pollution situation.
Some rather formidable obstacles stand in the way of the
achievement of such a perfect forecast. These are, as would be
expected, forms' of ignorance: ignorance regarding the future
meteorological doings of nature, and further, ignorance of the
actual conditions that have led to smog attacks in the past.
In the latter situation, the air pollution forecaster finds him-
self in a position similar to that of the early Weather Bureau men
who had to make predictions without the benefit of an understand-
ing afforded by the frontal theory of storm formation and growth.
This analogy applies specifically to the photochemical type of
smog. To the extent that low-level, mesoscale, meteorological
measurements are not made synoptically with air quality measure-
ments, it applies also to the air pollution problem in general.
PROBLEMS INVOLVING THE ATMOSPHERIC-CHEMICAL
REACTIONS
The cause-effect relationship between hydrocarbon-produc-
ing sources and photochemical reactant products forming smog has
been firmly established since 1952 by Haagen-Smit. The inter-
mediate steps in this reaction, the way variations in initial pollu-
tion conditions may alter the course of the reaction, have re-
mained unknown, however. Within the past year, experiments
with large controlled-environment chambers have begun to pro-
duce results that should increase our knowledge in this regard.
In the meantime back at the forecast office, the forecaster is en-
deavoring to apply his knowledge of the reactions. This is most
difficult, to use as colorless an understatement as possible, in
our uncontrolled natural atmospheric environment.
SEC TECHNICAL REPORT A62-5
-------
FORECASTING URBAN AIR POLLUTION 271
Experience has taught the Los Angeles smog forecasters
that some of the most apparent smog attacks (apparent to the tax-
paying citizens' sense of smell and sight) are not accompanied by
correspondingly high concentrations of oxidants or ozone. Other
situations reveal the reverse - high concentrations of pollutants,
but no noticeable smog.
Then, too, there are differences in smog effects that de-
pend on the particular section of the Basin that is involved. For
example, in downtown Los Angeles the eye-irritating properties
of smog are most pronounced, but ozone values are relatively low.
It is at outlying stations to the north and east that alert-level
ozone values are usually measured. In these suburbs eye irrita-
tion may or may not accompany the ozone. Further east, in San
Bernardino and Riverside Counties, elevated ozone or oxidant
values may be recorded, but little or no eye irritation is noted.
These high values have been accompanied by visibility observa-
tions ranging from 30 miles to 1/2 mile.
Plant damage is another manifestation of air pollution on
which the forecaster must give advice. Here more unknowns ob-
scure the forecaster's crystal ball. Plant-damaging attacks of
smog have been noted everywhere in the Basin and even across
the fringing mountains in the Mojave Desert. Apparently the con-
centration of smog need not be very high to produce plant damage,
since such effects are noted in areas that so far have not reported
eye irritation.
THE OPERATIONAL FORECAST
Operational forecasts by the Air Pollution Control District
are designed for two general uses: public and internal. The
latter is the information on which the air sampling operation is
conducted, and probably closely resembles the type of service
given any organization by its meteorological supporting unit. The
public dissemination of smog forecasts is somewhat unusual and
for that reason is more fully discussed.
The issuance of smog forecasts to the Los Angeles public
began in 1952. Originally the public release of these forecasts
was made in order to educate. Smog in its varying degrees of
severity was not recognized by the general public as being mete-
orologically controlled. Rather it was thought that the source of
pollution varied in strength. It was to combat this idea that the
forecasts were issued, always relating the expected pollution in
terms of pertinent weather conditions. A secondary consideration
involved in the issuance of a public forecast was the District s
overloaded switchboard on bad days. Complaints came thick and
SYMPOSIUM: AIR OVER CITIES
-------
272 KAUPER
fast whenever the weeping-eyed populace could grope its way, via
telephone dial, to the District's complaint desk. The forecast, it
was hoped, would cut out that portion of the public that called just
to tell the District that it was smoggy. If it was already in the
forecast, the reasoning went, they all could save their telephone
message units.
The air pollution forecast has been the basis for various
systems of limiting emissions when poor dispersion conditions
were expected. For example, open fires are not allowed in the
Los Angeles Basin, with but few exceptions. These exceptions,
however, are also forbidden whenever stagnant air flow is fore-
cast.
At one time, a voluntary system of restricted driving was
put into effect whenever bad smog was anticipated. This took the
form of a request that no unnecessary driving be done on smoggy
days. Since no one, it turned out, did any unnecessary driving,
this use of the forecast could not be considered a huge success.
However, there is evidence that on days forecast to be smoggy,
fewer than the usual number of shoppers came into town. At least,
that is the complaint from the downtown merchants' association.
During the cool portion of the year, when the electricity-
generating steam plants are normally burning fuel oil rather than
natural gas, the forecast is used to cause these plants to switch
to gas when an air pollution attack is expected. This takes the
combined efforts of the power-generating systems and the gas
companies. Gas, being in great demand for domestic use in
winter, is provided to industry on an as-available basis. When
the fuel-switch forecast is made, the gas companies have to de-
pend on their storage facilities to provide service to all. For-
tunately heavy smog attacks, while they occur in winter, do not
last very long - usually only a day or two.
A major segment of the air sampling activity of the District
is that done in response to the legally required monitoring service.
Various pollutants, known to have toxic effects, are constantly
watched through use of automatic recording instruments. A sys-
tem of warning levels has been established. The first alert level
is in the nature of a preliminary warning. Industries are notified
to prepare for possible shutdown, if the pollution level goes higher.
A second-stage alert triggers the shutdown procedure. If third-
stage conditions are encountered, then concentrations are such
that a danger to the public health exists and the emergency powers
of the State Governor come into play.
The meteorologist is involved in this alert procedure be-
cause pollution levels exceeding the alert criteria may occur at
any station in the Basin. He is then asked to predict how long the
SEC TECHNICAL REPORT A62-5
-------
FORECASTING URBAN AIR POLLUTION 273
values will remain high, to what maximum they will go, and
finally, what other stations may be expected to measure high con-
centrations of this pollutant.
If the meteorologist knew for the entire area the wind flow
pattern and the depth of the polluted layer, then he would be able
to answer these questions successfully. Since he doesn't know
these things precisely, he does the best guessing he can. Cer-
tainly some type of telemetering system, both for meteorological
and air monitoring data, would be very useful here.
Obtaining the depth of the polluted air - a sounding - pre-
sents one of the more important difficulties. At the present stage
of technology the best that can be done is to obtain more frequent
temperature soundings by balloons or other airborne equipment.
What is really needed is an instrumented tower about a mile high,
but that does seem a bit impractical even for California. Three-
hundred-foot towers, as used elsewhere, would not meet the need
in Los Angeles because the restraining inversion base is as much
as 1000 feet or more even on the worst smog days.
FORECASTING WEATHER PATTERNS
CONDUCIVE TO SMOG
Turning to the strictly meteorological portion of the fore-
cast problem, one discovers the same lack of basic knowledge of
the existing large-scale weather conditions that exists regarding
the very local conditions associated with pollution concentrations.
For a forecast of the Los Angeles smog conditions 24 to 36 hours
in the future, it is necessary to consider past and current weather
patterns and how they are changing. The basic charts used to ob-
tain this picture are those for 500 and 850 millibars and the cor-
responding surface chart.
Since upper-air charts are made only every 12 hours, with
considerable space between network stations, the forecaster is
severely limited in his efforts to extrapolate. This would not be
such a problem if we were dealing with dominant features of a
weather map - the deep lows and the prominant highs - but air
pollution is a child of weak weather systems. These are often
barely detectable, even after the fact. Any changes at all in con-
ditions either at the surface or aloft, may be important to the
forecast. With present weather observing schedules, these subtle
changes are almost bound to be noted too late to do any good.
In an effort to detect small changes in weather conditions
the Los Angeles forecaster resorts to a close watch of surface
pressure gradients between coastal and inland stations. Generally,
an increasing onshore gradient results in a rising inversion and
SYMPOSIUM: AIR OVER CITIES
-------
274 KAUPER
a lessening of the smog. An increasing offshore gradient usually
indicates the reverse, until the change is too boisterous. In this
latter situation, the air mass lying on the desert side of the moun-
tains bursts its bounds and sweeps down over the coast. This is
the "Santa Ana", a wind that carries the smog far out to sea and
in certain sections is strong enough to carry off part of the real
estate as well.
The forecaster feels most confident about his forecast for
the occurrence of smog when the pressure gradients are near
zero and show no great change during the preceding hours. The-
oretically, the low-level wind flow that is responsible for the
accumulation and subsequent movement of smog is caused by these
pressure gradients. When strong gradients exist, this relation-
ship of pressure and wind flow is evident. But with weak gradi-
ents the winds appear to be uninhibited by such theoretical con-
siderations. It must be assumed that even in this situation the
wind does indeed flow from high- to low-pressure areas. The
apparent contradictions encountered must be the result of inac-
curate or imprecise pressure measurement.
Even in the low-lying coastal area of the Los Angeles Basin,
the first-order weather stations (military and Weather Bureau)
sometimes report as much as one, and often as much as one-half,
millibar difference in pressure between stations only 6 miles
apart. This, it would seem, represents a basic inaccuracy in
pressure readings at one or the other station. A thorough check
on pressure-observing practices seems indicated so that this un-
certainty can be minimized.
Even with these more obvious errors in pressure readings
corrected, there still remains the problem of preciseness. In a
situation of weak gradients, it is not at all unlikely that definite
wind flows will develop, indicating a response to some gradient,
weak as it seems to be. Possibly, a more refined pressure ob-
servation network, refined both from the standpoint of instrumen-
tation and geographical spacing, would reveal the cause of the
observed wind flows.
UNTAPPED SOURCES OF WEATHER DATA
One possibility of getting an enlarged picture of the basic
pressure patterns is to utilize temperature. Temperature is re-
lated to pressure and there are many more temperature than
pressure observations available for study.
In the Los Angeles Basin, in addition to the regular network
of meteorological stations, there are a great number of locations
at which temperature is measured. Within the past several years
SEC TECHNICAL REPORT A62-5
-------
FORECASTING URBAN AIR POLLUTION 275
it has become fashionable for commercial concerns to display tem-
perature readings on lighted signs. Preliminary reconnaissance
indicates that as a whole these thermometers have a rather uni-
form exposure. They are usually mounted on the roof of a one-
story building in a louvered "birdhouse". The indicators are
visible both from the street and inside. Telephoned in to a fore-
cast office, these temperature reports would provide a basis for
analyzing the existing temperature field. A series of hourly ob-
servations would show the change of temperature with time.
Hopefully, these data could be correlated with the various types
of morning air flows found in the Los Angeles Basin.
This is one method of trying to achieve full utilization of
existing meteorological data. Any study in depth of a geographical
area such as the Los Angeles Basin will uncover an unsuspected
number of meteorological data gathering organizations. Much
information could be made available to interested parties provided
some means could be found for communication. No one organiza-
tion appears able to afford the monetary and labor costs of setting
up and running such a local meteorological network. A proposal
for such a system, in which mesoscale analyses would be per-
formed at a local analysis center, has been recently made by
Todd.
SUMMARY
Problems facing the air pollution forecaster, as seen from
the midst of the Los Angeles smog cloud, fall into two categories.
These involve the lack of knowledge of, (1) the pollution reaction
itself, and (2) the meteorological variations.
The first is being worked on by air pollution chemists and
thus is outside the province of meteorologists. The second,
though, must be faced by the meteorological profession. An in-
creased emphasis must be placed on measuring the atmosphere on
a scale suitable to the air pollution problem. This includes not
only a relocation of observational sites back into the cities but also
an increase in the density of the weather-reporting network sur-
rounding a city. It further includes making the proper kinds of
meteorological measurements.
Once this basic meteorological data is available the fore-
casters of air pollution may have some hope of success in pre-
dieting precisely the extent and duration of a srnog episode. It is
conceivable that, if necessary, the future air pollution forecast
may be used to save lives even as flood and severe weather fore-
casts do presently.
SYMPOSIUM: AIR OVER CITIES
-------
276 .KAUPER
DISCUSSION
MR. WILLIAMS: What has the batting average been on pre-
dicting smog?
MR. KAUPER: You can always get an 80 percent accuracy
if you manipulate right. But we do a monthly verification on var-
ious parts of our forecast, and our verification ranges between 60
and 70 percent most of the time.
MR. LICHTBLAU: I feel that I have to say a few words in
defense of the Weather Bureau. Thirty years ago we were under
somewhat different circumstances, and practically all of our sta-
tions were at urban locations. I think that we have as many urban
stations at the present time as we had 30 years ago, but we also
have many, many airport stations that did not exist 30 years ago.
So that is the way the picture has changed. And probably we will
get more urban stations as time progresses.
You mentioned about the pressure observations. It is un-
fortunate that these inaccuracies do exist and we will try to cor-
rect this situation. But there are many agencies involved, such
as the military, the CAA, and the FAA, and sometimes it is rather
difficult to get uniform measurements.
There are a couple of other remarks I would like to make.
I just wanted to state for the record that meteorologists are cer-
tainly interested in morbidity and health. Otherwise there would
be no reason for our existence.
Yesterday the question was asked about how long it would
take for Weather Bureau people to make air pollution forecasts.
I think the figure was given as 45 minutes or an hour. At a
Weather Bureau Forecast Center the forecaster on duty often does
make informed air pollution forecasts. At certain times of the
year the city piles up the collected trash instead of burning it
every day as they usually do. Then the fire department will call
us and ask, "Can we burn our trash today, or shall we burn it
tomorrow? " And usually we can tell them within a minute or two
whether they should burn it today or tomorrow.
DR. HEWSON: In 1949 a paper was published by Dr. Jones
in which certain rules and definitions of chemistry were set forth.
It was indicated that these should be used in air pollution fore-
casts. Have these rules been proved to be of value when used
over a long period of time?
MR. KAUPER: The answer is that we don't use those rules.
SEC TECHNICAL REPORT A62-5
-------
The Need for More Meaningful Meteorological
and Air Quality Observations for Mortality
and Morbidity Studies
F. FIELD, Albert Einstein College of
Medicine of Yeshiva University
J. K. McGUIRE, Office of Climatology
United States Weather Bweau
Summary
The relationships between air pollution, meteorology, morbidity, and mortality are
being studied in many countries, especially Great Britain and the United States. These
studies show that more meaningful meteorological and air pollution data must be devel-
oped. Relating air pollution to health effects is difficult because the network of record-
ing stations is sparse and because not enough is known about the diffusion, dispersion,
and concentration of pollutants in the atmosphere of urban areas. Difficulties encoun-
tered in health studies of this type are discussed, and the need for descriptive meteor-
ological and air pollution data is indicated.
The existing information concerning the effects of toxic
substances upon animals and man is considerable and is derived
mainly from laboratory investigations and industrial exposure
studies. These data generally are concerned with above-normal
concentrations and as a rule, though not always, with acute ef-
fects. It is from these data that the industrial hygienists derive
the terms "maximum allowable concentration" and "lethal dose. "
In considering the problem of air pollution, however, we
are interested not only in the effects on man of the above-normal
concentrations that may occur at times, but also in the effects
of the levels of air pollution to which he is subject during the
course of his daily life.
The chief characteristics of air pollution are that it is
generally low in concentration and long in duration of exposure.
Air pollution varies temporally and spatially in a given geograph-
ical area and also differs from area to area. These character-
istics make it exceedingly difficult to arrive at meaningful average
values of air pollution over a long period of time.
Moreover, the residents of urban areas, who are exposed
to the effects of air pollution during their lifetimes, have great
mobility. A person employed in a section of the city where he is
exposed during an 8-hour day to high levels of air pollution may
reside in a relatively non-polluted suburban atmosphere. Another
person in the course of changes of employment may be subject to
277
-------
278 FIELD AND McGUIRE
various levels of air pollution. This mobility in a population adds
to the difficulty of evaluating the effects of air pollution.
Three approaches may be used to relate air pollution to
health. The first is to study man in the laboratory in experimen-
tal fashion. Despite moral and legal implications, volunteer work
with human subjects has produced some basic information. But
this type of experimentation has of necessity been limited. We do
not reject this direct approach -- it is the soundest physical at-
tack on the problem -- but by its very nature this work must ad-
vance slowly and cautiously.
The second source of information is the controlled exposure
of animals in laboratories to various concentrations and combina-
tions of air pollution and meteorological factors. Much knowledge
concerning physiological changes has been gained this way, but
carrying over the results of these experiments to human beings
is difficult.
The third method of evaluating the influence of air pollution
and meteorology upon morbidity and mortality is to study man in
his normal environment. This is the epidemiologic approach in
which we are working, under the direction of Dr. Leonard Green-
burg, at the Albert Einstein College of Medicine with support by
the U.S. Public Health Service. We are grateful indeed for being
permitted to attend this meeting and express our thoughts on the
matter.
The epidemiologic method consists of a study of the con-
ditions in a population, the distribution of these conditions, and
the factors that lead to such a distribution. The condition to
which we refer is the health status of the individual, which in-
cludes susceptibility to disease, illness, or death. The factors
that influence these conditions and with which we are concerned
are the contaminants in the atmosphere and the meteorological
variations.
Among the questions which we may ask are: Does exposure
to normally low concentrations in the air cause direct impairment
of health? Does such low level exposure make the individual
prone to infection? How important are the predisposing conditions
or susceptibility of the individual when he is exposed to low levels
or acute periods of air pollution ? What is the combined effect of
weather and air pollution on the health of New York City residents?
The several sources from which morbidity data may be ob-
tained include the records of emergency clinics, hospital admis-
sions and discharges, union and group medical plans, civil
service groups, private physicians, and clinics. The problems
SEC TECHNICAL REPORT A62-5
-------
MEANINGFUL DATA FOR HEALTH STUDIES 279
of differential diagnosis enter into the collection of such infor-
mation, and we would not like to mislead you into thinking that
morbidity statistics are clear cut.
Absenteeism and illnesses records are available from
large business firms, but in our experience they have been found
to be influenced by too many external factors to be of value.
Mortality information, on the other hand, provides more
definitive numbers, which include precise times of occurrence
and cause. We are fortunate in New York City that the Depart-
ment of Health's Bureau of Statistics, under the direction of Dr.
Carl Erhardt, has for the past 15 years compiled mortality data
by date of death and has recorded these on punch cards. These
are two invaluable assets in our study. Only New York City,
Chicago, and Los Angeles, we are reliably informed, report
mortality by date of death. * Everywhere else the data are by
date of report of death or grouped by week or month. Date of the
report of death may be any time up to several days afterwards.
This uncertain time element seriously weakens any attempted
correlation with air pollution and weather.
Turning now to air pollution and weather, the epidemiologist
asks: What indices may we use to represent the air pollution and
weather to which an individual is exposed? What liberties may
we take with the data collected by monitoring, sampling, and
meteorological measurements in order to make such data usable
in our intended statistical analyses? For through these statis-
tical associations we will be able to develop hypotheses concerning
the etiological factors and test the hypotheses developed in clinic
or laboratory.
Air pollution may be considered as an independent variable,
subject only to the magnitude of its sources. On the other hand,
air pollution is conveyed by the atmosphere, so that it is a de-
pendent variable with respect to meteorological conditions. The
body of knowledge that has accumulated in recent years on the
subject of air pollution has already given satisfactory evidence of
the qualitative relationships between weather parameters and air
pollution levels in the atmosphere.
But from this point on our information is weak. The more
detailed knowledge of the atmospheric structure over given cities,
parts of cities, given urban or rural localities -- this knowledge
plus knowledge of the effects of such small-scale atmospheric
conditions in varying the dispersion and diffusion of pollutants is
still to be acquired.
We have become very much aware in our epidemiological
studies in New York City of the lack of information regarding the
SYMPOSIUM: AIR OVER CITIES
-------
280 FIELD AND McGUIRE
variability of both air pollution and meteorology, not only spatial-
ly in the horizontal and vertical, but also chronologically. This
variability points up one of our problems: how to obtain valid in-
dices for the contaminants that may be present in the atmosphere.
We would like to offer an example of a valid index. In a
survey designed to investigate the relationship between the prev-
alence of pneumoconiosis and dust exposure in coal mines, we
might define the exposure to the air pollutant as the concentra-
tion of dust times the length of exposure. If the concentration
were measured at one end of the coal shaft, our dust exposure
would be invalid. First, the concentrations which were measured
could not be assumed to be the same as those that had existed in
the past and that may have produced the disease. Second, in the
course of employment in the mines, the men would probably have
worked in other parts of the mine where the measurements taken
would not be applicable. Lastly, the dust measurements may
prove to be irrelevant in such a study because such dust measure-
ments in terms of weight are unduly influenced by the larger
particles, which do not enter the lung at all.
A valid index is therefore one that measures what it is in-
tended to measure, and it can be judged only in the light of pre-
vious knowledge and investigations.
What then is a valid index of air pollution that the epidemio-
logist may employ in his study of health effects ?
How valid are the air pollution data that the epidemiologist
has at his command ? At the outset it should be noted that pol-
lutant measurements are often taken 15 feet above street level.
If the object of the sampling network is to meet some specific
need, however, then the station locations are adjusted to the
particular requirements of the study. Few sampling stations
have been established primarily for the purpose of gathering data
for epidemiological studies. It should also be noted that the great
variety of sampling and analytical techniques makes it difficult to
compare or interpret air quality data collected from different
sources.
Although the measurement of air pollution levels has pro-
gressed rapidly in the past few years and continuous monitoring
is now possible, the air pollution levels reported in our cities are
generally estimates based upon readings taken at one or two sta-
tions. We are therefore in a rather precarious position as to the
significance and interpretation of the air quality readings now
available. We must ask just how much of an area, both horizont-
ally and vertically, these numbers represent. The question of
chronology arises also, since in many instances the readings are
not continuous.
SEC TECHNICAL REPORT A62-5
-------
MEANINGFUL DATA FOR HEALTH STUDIES 281
For example, data on the levels of various air pollutants
are collected in the City of New York, where we are fortunate in-
deed to have a growing Department of Air Pollution Control. This
department has established a series of stations about the city to
provide continuous information on smoke shade. Since we are
dealing with a collection of morbidity and mortality statistics
drawn from various section's of New York City, we are interested
in ascertaining the variability of the records from these stations.
An analysis of variance was carried out for four of these
stations, with the following interesting results:
1. The four stations differ significantly in their absolute
levels with a P value of less than 0. 001. We had of course ex-
pected variations between the stations and thought that possibly
these variations between all of the stations might not prove sig-
nificant, but this was not the case.
2. The two stations located at the Department of Air Pollu-
tion Control Laboratory at the 59th St. Bridge report significant-
ly higher readings than the stations at Central Park in midtown
and Brooklyn Tech High School, which is located in the borough of
Brooklyn. The distances between the four stations are approxi-
mately 10 to 15 miles.
3. There is a significant difference in the smoke shade
readings for different days of the month.
The purpose of our analysis was to enable us to determine
whether one station might possibly be utilized to give us a repre-
sentative reading for the city as a whole, or whether some mathe-
matical manipulation might give us an index. It appears offhand
that we cannot presume to use one station for such a purpose; the
analysis also emphasized the question of just how much of an area
each of our current stations represents.
Similar to the problem encountered in smoke shade is the
preliminary comparison of the carbon monoxide readings measured
in New York City. Continuous observations are made at two sta-
tions: the Department of Air Pollution Control Laboratory at 121st
Street and the station at the 59th Street Bridge. The following in-
formation was obtained through a statistical analysis:
1. The absolute levels of carbon monoxide differ signifi-
cantly for the two recording stations. As might be expected, the
values registered at the Bridge are as much as 5 times higher
than those at the laboratory.
2 The range of concentrations is much greater at the
Bridge, as might be expected also, because of the proximity and
density of traffic. The onset of high levels of carbon monoxide
SYMPOSIUM: AIR OVER CITIES
-------
282 FIELD AND Me GUI RE
starts earlier and the high levels last for a longer period of time
at the Bridge.
In comparing these two stations we allow for the fact that
the laboratory measurements are made 20 feet above street level,
while those at the Bridge are recorded at the ramp near street
level.
But before we can make any assumptions or simplifications
in working with carbon monoxide, we must have a better picture
of the distribution of this fugitive gas. What index can we use
with carbon monoxide? Is it possible to measure this pollutant at
some level in the lower atmosphere that may provide a reading
applicable for health studies and yet be free from the undue in-
fluence of the source? These are the data we sorely need.
The question of whether carbon monoxide has any ill effects
at chronic low levels has been reopened recently. European in-
vestigators are of an affirmative mind. But since our knowledge
of the distribution of carbon monoxide in urban air is incomplete,
it is difficult to infer the effects upon individuals. Sampling of
carbon monoxide at street levels has disclosed values as high as
100 ppm, depending upon density of traffic and meteorological
conditions. The validity of our carbon monoxide readings must
be probed further.
To this point our discussion has been concerned with the
location of sampling stations. We have pointed out the lack of
standardization in the collection and analysis of data. Before we
can truly judge the validity of the numbers we are collecting and
use them more meaningfully, we must improve and standardize
the basic methods of measurement.
We find a weakness also in our knowledge concerning the
interactions between pollutants in the atmosphere and the inter-
actions between pollutants and weather elements. Observational
and experimental evidence indicates that the toxic effect of carbon
monoxide, a pollutant, is enhanced by higher temperature, a
meteorological element. Or in reverse, the loss of biologically
active ultraviolet radiation, a meteorological element, is brought
about by increasing air pollution. It is also possible that the
direct relationships we seek do not exist, but instead synergistic
reactions between pollutants and weather may play a serious role.
A recent statistical investigation by Holland indicates that
both atmospheric pollution and low temperature have an effect on
acute respiratory admissions to London hospitals. In this study
it was not possible to decide which had the greatest effect, the
temperature or the air pollution, or the combination. 3
SEC TECHNICAL REPORT A62-5
GPO 82511 1-10
-------
MEANINGFUL DATA FOR HEALTH STUDIES 283
The interactions of the various substances in the air must
be determined in order to study more specific forms of illness.
We must investigate further the role played by weather in trans-
forming or altering pollutants in the photochemical reactions, in
the absorption of gases and vapors on particulate material, and in
catalytic oxidations and other chemical and physical changes.
As for the variable of meteorology itself, the effect of
weather has been the subject of intensive investigation in European
countries and to a much lesser degree in this country. Physio-
logical studies have been carried out on high and low temperatures,
temperature and humidity combinations, high and low barometric
pressure, and other meteorological elements.
Extreme weather conditions, such as severe and prolonged
heat waves, have been shown to increase mortality. This has
been demonstrated for our area of interest by Kutschenreuter. ^
Human bioclimatology is a vast and largely uncharted field.
We can only hope that the forecast made by Dr. H. E. Landsberg
will come true:
"The greatest advances of climatology are destined
to lie in the border field of biology, provided an
adequate cooperative research program is started.
The interactions between the physical changes in
the atmosphere and living organisms are too great a
challenge to scientific curiosity to remain in a
relatively unexplored state.
In this area we are again fortunate, since in New York
City we have abundant data from the U. S. Weather Bureau. In or
near the city we have hourly surface weather observations from
four locations: Central Park, LaGuardia Field, New York Inter-
national Airport, and Newark Airport. We have more data, though
less detailed, from the Bureau's cooperative climatological net-
work of about 30 stations. We even have an upper-air sounding
record, from N. Y. International Airport, so we have low-level
inversion data, etc.
Researchers in other areas may not be so fortunate. Most
cities have only one Weather Bureau station, usually at an airport,
away from the population from which the morbidity and mortality
data are derived. The network of atmospheric sounding stations,
adding the necessary third dimension, is much sparser than the
network of surface weather observations. In short, there is a
problem in the availability of the meteorological information.
Beyond this, we need more research on urban microclimatology
- the variations of climate between, for example, the airport
SYMPOSIUM: AIR OVER CITIES
-------
284 FIELD AND McGUIRE
where the meteorological data come from and the various parts of
the city where the people live and die. We need more research-
on the structure of the lower atmosphere over our cities, with
respect to both turbulence and microclimatology. As Wexler has
mentioned, the routine meteorological data are not suitable for
short-term time and spatial studies of air pollution. °
But even the possession of abundant routine data poses its
own problems. One of the first questions we had to settle in our
study was whether we might use the meteorological observations
from ore station to correlate with the mortality data for the city
as a whole. We were dealing with a population of 8 million in an
area of 365 square miles. The residents are living in environ-
ments ranging from semi-suburbs to stone canyons, in a climate
where, for example,-' sea breezes may be cooling Staten Island,
Lower Manhattan, and southern Brooklyn and Queens, while the
residents of the Bronx, Upper Manhattan, and the rest of Queens
and Brooklyn are sweltering in the heat.
With the assistance of the Weather Bureau's National
Weather Record Center, we compared a year's series of hourly
temperature, pressure, humidity, and wind observations from
the four locations previously mentioned, plus Battery Place, the
former site of the Weather Bureau City Office. We found that the
data from the five stations vary significantly with regard to ab-
solute values, but that with regard to changes in the values the
differences were not meteorologically important. In short, there
was a spatial but not a chronological difference. This result was
satisfactory, since it was the frequency and magnitude of the
weather changes that we felt all along represented the best key
to the interpretation of the mortality data. There is too much
uncertainty about the applicability of the values of the elements to
do much with the absolute values except in terms of large depar-
tures from "normal. "
While our experience has been confined to New York City
data, we believe that the same needs apply elsewhere, not only
in urban areas but in rural as well. Therefore, before summariz-
ing the foregoing in the form of conclusions as to present and
future needs for meteorological and air quality observations, we
must consider the important phase of rural investigations that we
have so far not mentioned. These needs are just as important as
those we have cited for cities. Much of what we have said applies
to rural areas, so we need not repeat ourselves. We do want to
bring out a few important points.
The first is the comparative lack of knowledge about rural
environments as regards the atmospheric, the air pollution, and
the medical information required. Weather Bureau stations are
SEC TECHNICAL REPORT A62-5
-------
MEANINGFUL DATA FOR HEALTH STUDIES 285
near or in the population centers; except for military airfields,
climatological information in the rural areas is limited to the
temperature and precipitation reports of the Weather Bureau's
cooperative observers. The stations of the solar radiation ob-
servation network are mainly in urban areas. There is a need
for more observations of solar radiation, surface winds, and
low-level inversions from the rural areas. There is an equal
need for more air quality observations from such areas. And,
just as the morbidity and mortality data for our cities need to
be improved and systematized, so do the same data for non-urban
locations.
There are several factors that support the foregoing state-
ments in terms of the needs of human bioclimatology. First, we
should have some idea of the ordinary air pollution levels away
from the major sources, in order to find out what our supposedly
healthier country cousins are breathing and to compare their
medical histories with those of us who gasp in the big cities.
Secondly, before we can decide how urban climates modify
air pollution and how air pollution modifies urban climates, we
should know more about non-urban climates and microclimates.
For example, how are we to compare New Yorkers with the
Westchester suburbanites and the Suffolk County farmers ?
Thirdly, we know that reduction of solar radiation is an
important health aspect of city climate and air pollution; and we
know that solar radiation itself is a major factor in certain forms
of morbidity.
Finally, we know that agricultural research specialists are
becoming increasingly concerned about the effects of air pollu-
tion on plants and animals. We know, too, that agricultural
meteorology and agricultural climatology are experiencing a
renaissance in many sections of our country, thanks to the
cooperation of the U. S. Weather Bureau and State Agricultural
Experiment Stations.
We suggest that we need the help of the agricultural in-
terests, and they need ours. For example, in the Midwest anet-
work of Model Agricultural Weather Stations has been proposed.
and some have been established. Shouldn't these include air
quality observations? Also, shouldn't more meteorological sta-
tions everywhere have air sampling instruments and vice versa?
Conclusions:
The following are the conclusions at which we have arrived:
SYMPOSIUM: AIR OVER CITIES
-------
286 FIELD AND McGUIRE
1. We need better morbidity and mortality data, and more
laboratory investigations, to assess the physiological meaning of
our meteorological and air quality observations.
2. We need to standardize the sampling and analytical
techniques for continuous air monitoring to match the rather high
degree of standardization that exists in the meteorological obser-
vations.
3. We need more joint meteorological and air sampling
stations in cities and towns and in rural areas.
4. We need to have these stations, primarily the rural
ones, instrumented to furnish more information on surface winds,
low-level atmospheric conditions, and solar radiation.
5. We need to encourage and support the U. S. Weather
Bureau so that more observations of the weather and climate can
be made in the places where the people live, in addition to where
the airplanes take off and land.
6. We need research on the meteorological and air quality
data to translate them into models for obtaining meaningful indices
of air pollution that may be interpreted in terms of health, both
7 Q
in urban and rural areas. '< °
7. We need more cooperation and more teamwork from
everybody -- the general public; the local, state, and federal
medical, public health, meteorological and climatological,
agricultural and industrial organizations --to obtain the obser-
vations needed and to achieve the goal the Surgeon-General has
defined: "To determine the conditions under which toxic substan-
ces in the community atmosphere affect human health adversely,
and the measures which must be applied to prevent adverse
effects. "9
REFERENCES
1. Dr. Richard Prindle, personal communication.
2. Castrop, V. J., Stephans, J. F., and Patty, F. A., "A
Comparison of Carbon Monoxide Concentrations in Detroit
and Los Angeles," Amer. Ind. Hyg. Assoc. Quart., vol. 16,
1955, p. 225.
3. Holland, W. W., Spicer, C. C., and Wilson, J. M. G., "in-
fluence of the weather on respiratory and heart disease, "
Lancet (London). Vol. 2, No. 7198, August 12, 1961, p. 338.
4. Kutschenreuter, P. H., "A study of the effect of weather on
mortality in New York City, " M. S. Thesis, Rutgers U.,
January 1960.
SEC TECHNICAL REPORT A62-5
-------
MEANINGFUL DATA FOR HEALTH STUDIES 287
5. Landsberg, H. E., "Trends in climatology, " Science, Vol.
128, No. 3327 (Oct. 3, 1958), pp. 749-758.
6. Wexler, H., "The role of meteorology in air pollution, " in
Air Pollution, World Meteorological Organization (Colum-
bia University Press, 1961) pp. 49-61.
7. Karplus, W. J., Berkey, G. A., and Pehrul, P. J., "At-
mospheric diffusion of air pollutants, Analog computer
study, " Ind. Eng. Chem., Vol. 50 (Nov.. 1958), pp. 1657-
1660.
8. Vaughan, L. M., "The prediction of atmospheric diffusion by
using an eddy diffusivity based on the vertical transfer of
heat, " Journal of Meteorology, Vol. 18, No. 1 (Feb. 1961),
pp. 43-49.
9. Burney, L. E., "Status Report to the Nation, " Proceedings
National Conference on Air Pollution, Washington, D. C.,
Nov. 18-20, 1958, U.S. Dept of Health, Education and
Welfare, PHS 1959, pg. 3.
DISCUSSION
DR. LODGE: What specifically, do you mean by measuring
air pollution? Air pollution is an abstract concept. You can't
measure everything in the air. This is some thousands of sub-
stances.
DR. FIELD: Well, let us put it this way, then; there are
many air pollutants. We are interested in a certain number of
pollutants. Let's take, for example, carbon monoxide. We are
interested in the effects upon health. We feel that air pollution,
while it may change the weather pattern, may bring the tempera-
ture up in our cities, may increase rainfall. All these are excellent
research projects which we highly subscribe to, but what we are
interested in is trying to learn what happens to a population such
as that of New York City, when the amount of sulfur dioxide in
the atmosphere rises to a certain level. What is the effect on
respiratory disease? Does this in any way affect morbidity?
Does this in any way hospitalize people?
What we require are more adequate measures of sulfur
dioxide that will tell us what these readings do mean in terms of
area and population effects. Right now we are at a loss. We have
five or six stations that we compare and we find that they vary
greatly. Since we would like to compare any change in specific
pollutants as they are measured now, carbon monoxide, sulfur
dioxide etc. we would like to know just how much of an area these
pollutants actually embrace, what mortality and morbidity data
SYMPOSIUM: AIR OVER CITIES
-------
288 FIELD AND McGUIRE
must we confine ourselves to. For example, it has been mentioned
that stagnation, large high-pressure systems moving into an
area, will create certain stability, inversions, and all the weak
wind patterns that go along with this, and that this would increase
air pollution.
Well, taking our cue from such a meteorological incident,
which occurred in New York City in 1953, we went back into the
records and studied the mortality during this period of air pollu-
tion and compared it to similar periods during 3 years prior and
3 years subsequent to the incident. At that time the only pollu-
tion data we had as a measure were smoke shade and sulfur diox-
ide collected at one station. Despite the fact that this was a gross
air pollution estimate, we were able to define a striking increase,
which proved to be statistically significant in the number of
deaths in New York City.
DR. LODGE: In a more precise study, though, do you feel
that you will uniquely characterize sulfur dioxide intoxication by
a knowledge of sulfur dioxide pollution?
DR. FIELD: Yes, I would say so. There are different
ways of grading the effects on human physiology of these various
pollutants. This would be more of a laboratory type of approach.
I had mentioned to Dr. Landsberg that the Russians, for ex-
ample, in setting up maximum permissible levels, will test an
individual through visual stimuli. They will seat an individual in
a chamber, subject him to certain stimulus of light or other stim-
uli, and introduce at the same time an amount of gaseous material
into the room at extremely low levels. They keep testing on
this basis until they find out at what level these various incre-
ments, these various changes take place. Then they will remove
the initial visual stimulus, after the subject has been conditioned
and introduce the secondary effect at lower and lower levels and
find the threshold where he no longer feels or elicits a response.
There are various ways of grading this.
Now, we feel that respiratory disease and sulfur dioxide go
hand in hand. We would like very much to be able to correlate
the two, but when we find that the measurements are made com-
pletely differently from city to city and techniques vary so great-
ly, that the measurements are made at stations which are arbi-
trarily set up at certain areas, without any regard to representa-
tion, I don't feel that such pollution data are adequate for our use.
I was curious when I asked the question before about the oxide,
whether somebody actually went out and did some surveys and
then decided this was a fairly representative area, unaffected by
local sources that would contribute in any way to the change.
SEC TECHNICAL REPORT A62-5
-------
MEANINGFUL DATA FOR HEALTH STUDIES 289
After all, we are dealing with the numbers that you provide us
for health studies and these numbers at this point are not satis-
factory for our purpose.
MR. STERN: I would like to commend to Dr. Field the
series of papers discussing sampling station density and place-
ment, as a result of the Nashville study. There the city was
literally saturated with sampling points in order to determine the
necessary number and location of sampling points. And I think
the Nashville data in the series of papers that have been published
go a long way to answer some of the questions you have raised as
to the meaning of any one station with relation to others in a net-
work, and how many are needed to characterize it.
DR. FIELD: Thank you.
MR. WOHLERS: I have a simple question. So far as
analytic techniques are concerned, they are physical measure-
ments and if you want analyses at this and that location, we can
provide them for you. But then, after you get these data how do
you then judge whether my liver was affected or my lungs, or
which has affected my head?
DR. FIELD: It has been mentioned previously that Mr.
Halitsky and another health study group were entering into a
study in New York City on welfare patients and the possible ef-
fects of air pollution. I have been at one or two of the meetings
and I understand they have quite a problem there in trying to as-
say that. However, there are certain diseases we may go into
and see whether we can correlate with any of the particular pol-
lutants. For example, one of the interesting things would be to
be able to compare cities. Does New York have more air pollu-
tion than Boston? I don't know whether this is possible. Is there
some level that we might get, or background level of air pollution
that would enable us to compare the morbidity and mortality and
air pollution among cities. This has been tried in many other
ways through fuel consumption and economic indices and so on.
But for the actual morbidity and mortality studies underway, are
pollutants as measured today, sufficient? And I feel that they are
not. I don't dispute Mr. Stern's statement. I just bring out what
our experience has been, the fact that stations are set up in an
area and we have no prior knowledge as to whether those stations
are adversely affected by the local topography or the local in-
stantaneous release of pollutants that might occur because of
factory sites and so on. Measuring carbon monoxide at a bridge
will give you a good idea of traffic. Measuring carbon monoxide
one story above street level in another part of town may give you
something else. But I hope it is possible through your research
and through what I feel is much-required research, for you to
SYMPOSIUM: AIR OVER CITIES
-------
290 FIELD AND McGUIRE
come up with numbers for us to use more meaningfully in the
morbidity and mortality studies.
If we are given more precise air pollution data to work with,
which you say you can provide, I am sure we will be able to study
such data and those diseases which we suspect are aggravated or
brought about by air pollution and weather.
SEC TECHNICAL REPORT A62-5
GPO 8251 1 1-1 I
------- |