AIR POLLUTION CLIMATOLOGY*
by
George C. Holzworth**
Division of Meteorology
Environmental Protection Agency
*
For presentation at the American Institute of Chemical Engineers,
64th. Annual Meeting, San Francisco, California, November 28-
December 2, 1971.
** On assignment from National Oceanic and Atmospheric Administration,
U. S.1 Department of rCatmerce.
-------
Air Pollution Climatology
Air pollution climatology in this presentation refers to the
average state and spatial distribution of meteorological variables
that influence the transport and diffusion of pollution. Interest
in this subject is mainly in cities where emissions are usually most
prolific, and where most people and valuable properties are located.
Transport and diffusion of air pollutants in urban areas is accom-
plished by atmospheric motion scales that depend variously upon
large or synoptic scale and mesoscale meteorological features
(e.g., the order of 1000 and 100 km, respectively). These features
are more or less influenced by the roughness elements and thermal
properties of cities themselves. Detailed information on atmos-
pheric transport and diffusion in most cities is unavailable. But
even if it were abundantly available, there would hardly be time
in this presentation to adequately describe the features for
different cities in different regions of the country. This paper
briefly describes the occurrence over the contiguous United States
of a few general but basic elements of air pollution climatology,
with particular reference to cities.
Atmospheric stability
The degree to which the atmosphere effects vertical motions
and mixing depends upon the vertical variation of temperature
(i.e., upon stability). Since the earth's solid (snow-free)
-------
surface is a comparatively good absorber of solar radiation and a
good emitter of infrared radiation, the lower layers of the atmos-
phere ordinarily display a large diurnal variation in stability.
The significance of this variation can be seen in the behavior of
plumes from an elevated stack. Figure 1-upper depicts the "fanning"
type of plume, whose name is derived from its fan- or ribbon-like
appearance in the horizontal plane. In the vertical plane the
plume undergoes little diffusion and often remains entirely aloft
for long periods. Fanning occurs when the plume is in a tempera-
ture inversion (i.e., when the temperature increases with height,
a very stable situation). As will be shown, low-level temperature
inversions are very common at night. Ordinarily they are formed
by radiational cooling of the earth's surface.
Figure 1-middle shows the vertical temperature profile that
often occurs after sunrise following a night with a low-level
inversion. In this situation the warming ground heats the adja-
cent air, creating instability and vertical air motions. The
result of such mixing is to establish a temperature decrease with
height in the mixing layer that is nearly dry adlabatic (i.e., the
temperature decrease with height due only to the decrease of
pressure with height, .01°C/m). As the surface temperature
increases, the thickness of the mixing layer increases. A
"fumigating" type of plume occurs when the top of the mixing
layer reaches the height of the relatively concentrated effluent
that previously was entirely within a stable layer (e.g., a
-------
fanning plume). Consequently, r««.
-------
Low-level inversions
Certainly low-level temperature inversions rank as one of
the more important elements of air pollution climatology. Their
occurrence throughout the United States' has been determined by
Hosier (3) from whom Figure 2 is taken. This figure shows that at
night low-level inversions are anything but rare; in fact they are
quite common. Especially in valleys of the desert and mountain
region of the West a night without such an inversion is the anomaly.
Only in a few regions, whose combined area is comparatively small,
do low-level inversions occur on less than half the nights annually.
These areas are the Texas Coastal Plain,, southern Florida, the mid-
Atlantic Coast, the eastern Great Lakes, and the immediate Pacific
Coast. In the latter area the California Coast deserves special
mention because, as reported by Neiburger, Johnson and Chien (4),
a very Intense subsidence Inversion occurs there with great
frequency, both night and day, but it is usually based around 1500
feet and was not counted by Hosier who defined low-level as 500
feet or less.
Most of Hosier's observational data are from sites in rural
surroundings and, therefore, it is reasonable to assume that almost
all of his nocturnal low-level inversions were the usual radiation
type and were actually based at the ground. At night in cities,
however, buildings and streets cool rather slowly, due mostly to
their large heat capacity, and this often results in relatively
-------
unstable atmospheric temperature profile extending through the lower
few hundred or so feet. But tlr-s shallow mixing layer is usually
topped by a more stable layer, as shown in the pioneering work of
Duckworth and Sandberg (5) and in later studies by DeMarrals (6)
and by Clarke (7). Figure 3 is Clarke's space cross-section of
temperature across Cincinnati, Ohio, based on special temperature
measurements near sunrise following a nearly cloudless summer
night. The cross-section is oriented roughly from west-southwest
on the left to east-northeast on the right, along the effective
wind direction. The thin lines are isotherms (°F) and heavy lines
represent the base or top of a main inversion*. Notice in this fig-
ure that the top of the upper inversion near 1200 feet is practically
horizontal although the topography is quite irregular. Above this
level the temperature decreases with height at a rate slightly less
than dry adiabatic. On the left side of the figure in rural
Kentucky, upwind of Cincinnati, the inversion is based at the
ground and the temperature increases from 67°F near the ground to
77*F at the top of the inversion, 400 feet above ground level.
Moving down the hill to the urbanized Kentucky side of the Ohio
River, the ground-based inversion is replaced by an unstable super-
ad iaba tic layer about 200 feet deep which continues through down-
town Cincinnati to the top of the first hill to the east. Hare
the building density begins to decrease and in the downwind
suburbs a very weak inversion occurs, but it is topped by the more
-------
intense main inversion. Finally, on the right side of the --.yu
in the rural area, what Clarke has called the "urban heat plume"
is embedded aloft within the main inversion. It should be noticed
that over downtown Cincinnati pollutants emitted at this time from
the top of the tallest building would be effectively trapped within
the inversion aloft, whereas emissions near ground level would be
nixed in the boundary layer below the inversion.
There is hardly any doubt that in most regions of the conti^-
uous United States, including cities, vertical mixing at night is
ordinarily confined to a rather thin layer near the surface, much
*
as depicted by Clarke. In spite of the fact that Hosier's L-c?or«-
tant paper is now ten years old, the erroneous notion persists
among much of the general public and among some air pollution
control authorities that nocturnal low-level inversions are unusual.
Afternoon mixing heights
Since inversions and shallow mixing layers are common at night,
the height of the layer through which relatively vigorous vertical
mixing develops during daytime is an important feature of air
pollution climatology. Because of international agreement on syn-
optic observation times, vertical temperature measurements in the
afternoon are not taken throughout the United States. However,
*
afternoon mixing heights have been estimated by Holzworth (8)
using morning vertical temperature me£.i.urements and af terr.oor.
-------
surface maximum temperatures under the assumption that the temper-
ature profile in the mixing layer is dry adiabatic. Figure A,
based on data for the 62 locations in Figure 8 or 9, shows that,
.annually, afternoon mixing heights over the United States range
from about 800 meters to over 2600 meters. The heights are com-
paratively low along coast lines and in the vicinity of the Great
Lakes, while, over the southern Rockies ar.d central Appalachians,
they are high.' Thus, the diurnal venation of heights through
which vigorous vertical mixing occurs is large over inland regions
compared to coastal regions. Figures 3 and 4 also suggest that
the meteorological condition conducive to the* fumigation process
occurs frequently throughout the United States.
Winds
Another variable of paramount importance in air pollution
climatology is the wind, both, speed and 'direction. Wind directions
are difficult to display over large regions because they are so
variable and influenced by local features (e.g., land and sea or
lake breezes, mountain and valley winds, topography, and, as
demonstrated by Pooler (9) and by Georgii (10), _c times a city
may induce its own air circulation). Figure 5, from Sladc (11),
shows some effects of irregular topography on wind direction around
the Oak Ridge National Laboratory in Tennessee. In this figure
the length ... direction bars of the w-.r.d .-cses arc y.roportional to
-------
the frequency of the direction. Numbers in circles give the
percent frequency of calms, which for these 12 stations vary
between 0% and 40%. Notice that for stations within the two main
valleys most winds blow roughly parallel to the axes of the
valleys (i.e., from northeasterly and southwesterly sectors). This
bimodal distribution suggests that during the daytime the winds in
the valleys are mostly southwesterly, upvalley, and during night-
time they are northeasterly, downvalley. In the gap between the
two valleys most winds are roughly perpendicular to the valley
axes. Wind roses over higher terrain are generally less skewed
than those in valleys.
Figure 6, after Georgii (12), illustrates a small scale
effect of buildings on air flow and carbon monoxide concentrations
(from autos) over city streets. When the wind at the building top
was perpendicular to the street with speeds less than 2 meters/sec,
(Figure 6-upper), the space above the street was poorly ventilated.
Carbon monoxide concentrations varied most in the vertical with
highest concentrations at street level and only a slight concen-
tration gradient across the street. However, when the wind speed
exceeded 2 meters/sec (Figure 6-lower), a vortex formed which
reached the «treeI level and rcwultod in the conco.ii era Lion incruar,-
ing by a factor of two from the windward to the leeward side of
the street. Gaorgii stated chat complete ventilation of the street
required a wind speed of 5 meters/sec or more. Such effects as
thaso arc undoubtedly critically dependent upon wind direction,
-------
building heights, roadway widths, atmospheric stability, etc., and
obviously are not revealed by ordinary wind measurements.
Wind speeds, like directions, are often highly variable in
time and space, but average values for unobstructed locations
yield a fairly simple pattern over the United States as shown in
Figure 7 after Holzworth (13). Annual speeds averaged through the
afternoon mixing layer exceed- 8 meters/sec over much of Montana,
the middle tier of states north from Texas, and along the north
Atlantic Coast. Speeds less than 5 meters/sec only occur in
Pacific Coast states where the slowest speed is a little less
«.
than A meters/sec. The 6 meters/sec-isopleth extends inland to
the second tier of Pacific Coast states, and also occurs over the
central Rockies and from most Gulf Coast states to the central
Appalachians. The pattern of wind speeds through the relatively
shallow mixing layer height estimated for about mid-morning is
like this one for afternoons, except the morning speeds tend to
be 1 to 2 meters/sec slower.
Combined effects
While there is little doubt that the variables mentioned so
far are interesting and important elements of air pollution clima-
tology, their combined effects are of even greater interest. How-
ever, in such climatological considerations, it is not yet feasible
to include small scale details but only so:r.e features for which
-------
10
large-scale generalizations can be more appropriately inferred
(e.g., mixing height and wind speed). Such an appraisal has been
made recently for the.contiguous United States by Holzvorth (14).
This appraisal is based upon a rather simple diffusion model, much
like that of Miller and Holzworth (15), in which the average pol-
, /
lutant concentration over the city (X), normalized for a uniform
area emission rate (Q), is a function of the mixing height, wind
speed, and distance along the wind across the city. The main
assumptions in the model are:
1. Steady state conditions prevail.
2. Uniform emissions occur over the city at ground-level.
3. Pollutants are non-reactive and remain airborne.
4. Lateral diffusion can be neglected,
i
5. Vertical diffusion from each elemental source conforms to
unstable conditions and concentrations in the vertical
follow a Gaussian distribution out to a travel time which
is a function of the mixing height. Thereafter, a uniform
vertical distribution of pollutant occurs.
Obviously, this model greatly simplifies transport and diffusion
as they really occur over cities. But it does provide a means of
quantitatively evaluating some general and important features of
urban air pollution climatology.
Figure 8 shows isopleths of average normalized concentration
-------
11
(i.e., x/Q values) for an across-city distance (i.e., a city size)
of 50 km* As used here the dimensions of x/Q are sec/meter, but
multiplication by the emission rate (Q) in micrograms/square meter-
. sec yields concentration in micrograxns/cubic meter. It should
also be mentioned that at each data location the variation of- X/Q
values with city size is practically linear down to a size of about
I T- '
10 km. The figure depicts X/Q values that are exceeded on one-
tent,h of all mornings, annually. Data are based on the model,
which has been given by Holzworth (3-6)» daily estimates of the
urban mixing height a few hours after sunrise, and the average
wind speed within the mixing layer. Thus, ail other things than
mixing heights and wind speeds being the same, the upper decile
morning x/Q value for a 50-km size city
at New York City is 63 sec/meter
at Huntington, W.Va. is 375 sec/meter
at Chicago, 111. is 225 sec/meter
at Oklahoma City, Okla. is 77 sec/meter
at Salt Lake City, Utah is 95 sec/meter
at Los Angeles, Calif. is 188 sec/meter
at San Francisco, Calif. is 237 sec/meter
at Portland, Oregon ' is 300 sec/meter
X/Q data for other city sizes, for other percentile values,
for afternoons, and for the seasons have been given by Holzworth
(17). The x/Q isopleth pattern of Figure 8 for morning upper
-------
12
decile values is significantly different from those for morning
median values. Also, the x/Q values are generally very much less
for afternoon than for morning mixing height and wind speed values;
there are also seasonal variations. Thus, even in terms of a
rather simple and general approach, the quantitative evaluation
of air pollution climatology can be a complicated matter.
Stagnation episodes
Since particularly objectionable levels of air quality often
occur in episodes lasting over periods of days, air pollution
climatology is also concerned with episodes 6f limited dispersion
or episodes of atmospheric stagnation. In his well-known study
Korshover (18) found that such episodes frequently occurred in
connection with slow-moving, warm-type anticyclones. Korshover
examined 30 years of United States daily weather maps and tabulated
occurrences of anticyclonic conditions with geostrophic wind speeds
of 15 knots or less lasting at least four days. He found a maximum
of over 400 episode-days, about one day in 27, centered over north-
eastern Georgia, tapering off to almost zero just north of the
Great Lakes and zero west of the longitude through about central
Texas. However, Korshover's study was limited to the United
States east of the Rockies because sea-level pressure gradients
over high mountains are usually unrepresentative.
More recently Holzworth (19) has objectively determined the
-------
13
occurrence of limited dispersion episodes throughout the contiguous
United States, based on daily estimates of morning and afternoon
mixing heights and wind speeds. Various limiting criteria and
episode-durations have been tabulated but critical conditions
.likely to lead to undesirable air quality in large cities are
believed to be as follows:
1. All mixing heights 1500 meters or less.
2. All mixing layer average wind speeds 4.0 meters/sec or less,
3. No significant occurrence of precipitation,
4. Above conditions satisfied continuously for at least two
days.
Similar criteria to these have been used in the national air pol-
lution potential forecasting program, as described by Gross (20).
Figure 9 shows the total number of these episode-days in five
years. Zn the East, there is only a small area where the total
episode-days exceeds 100, about one day in 18, but in the West 100
days are exceeded at most stations and 200 days are exceeded over
a large area. It is interesting that in the five years examined,"
the specified, conditions did not occur once at two stations near
the middle of the country. The figure also shows that at most
stations in the West the season with the greatest number of episode-
days is winter, whereas in the East it is autumn.
-------
.14
Summary
Low-level temperature inversions at night are the rule in the
United States rather than the-exception. Even in cities where the
nocturnal urban heat island produces a relatively unstable tempera-
*
Cure profile through the lowest 200 feet or so, an inversion often
occurs immediately above. In the daytime the annual average height
of relatively vigoroup vertical mixing is estimated to range from
over 2600 meters over the southern Rockies to about 800 meters
along the California Coast; comparatively low values occur near
major shorelines. Wind speeds and directions are highly variable
.
in time and space, especially near ground level. But speeds ave-
raged through the afternoon mixing layer display a fairly simple. '
pattern. Annual average values range from 4 to 9 meters/sec;
%
speeds less than 5 meters/sec occur only in Pacific Coast states
but 6 meters/sec or less also occur over the Great Basin of the
central Rockies and around the area centered on Mississippi,
Alabama, and Tennessee. Speeds averaged through the comparatively
shallow mixing layer around mid-morning generally follow the
afternoon isopleth pattern but are about 1 to 2 meters/sec slower.
Based on morning mixing height and wind speed data, the iso-
pleth pattern of annual upper decile theoretical normalized con-
centrations for large cities (50 km across) vary in the United
States by a factor of nine. For smaller cities the concentration
-------
15
values and their range would be less. Highest values occur in
central Oregon with a gradual decrease in values southward through
the southern border of California. In Mississippi, the indicated
potential concentrations are almost as high as in Oregon, and the
axis of fairly high concentrations extends roughly northeastward
along the Appalachians. Moderately high concentrations are also
found just west of Lake Superior. Comparatively low values occur
along much of the Atlantic and Gulf Coasts and over Texas and
Oklahoma. In general, corresponding computed concentrations for
afternoons are considerably less than for mornings. The objective
determination of two-day or longer episodes with limited disper-
sion conditions reveals that the total of such episode-days in
five years is less than 25 over most of Texas and generally over
the tier of states northward to Canada. In the East 100 episode-
days is barely exceeded only in a small area around southern West
Virginia. In the West 100 episode-days are.exceeded at most
locations and 200 episode-days are not unusual. Episodes in the
West are most frequent in winter and in the East, in autumn.
-------
16
Literature cited
1. Smith, M.E., ed., "Recommended Guide for the Prediction of the
Dispersion of Airborne Effluents," 85 pp., see p.15-35,
American Society of Mechanical Engineers, 345 East 47th.
Street, New York, N.Y. 10017 (1968).
2. Slade, D.H., ed., "Meteorology and Atomic Energy 1968," 445 pp.,
see p. 56-61, available as TID-24190 from Clearinghouse
for Federal Scientific and Technical Information, Spring-
. field, Va. 22151 (1968).
3. Hosier, C.R., Mon. Wea. Rev.. 89. 319-O39 (1961).
4. Neiburger, M., D.S. Johnson, and C.U. Chien, "Studies of the.
Structure of the Atmosphere over the Eastern Pacific Ocean
in Summer," Univ. of California publications in meteoro-
logy, vol. 1, no. 1, 94 pp., Univ. of Calif. Press.,
Berkeley and Los Angeles (1961).
5. Duckworth, F.S., and J.S. Sandberg.'Bull. Amer. Meteor. Soc..
35, 198-207 (1954).
6. DeMarrais, G.A., Bull. Amer. Meteor. Soc.. 42, 548-554 (1961).
7. Clarke, J.F., Mon>. Wea. Rev.. 97, 582-589 (1969),.
-------
17
8. Holzworth, C.C., "Mixing Heights, Wind Speeds, and Poccntial
for Urban Air Pollution throughout the Contiguous United
States." 156 pp., see'Figs. 6-10. To be published by
EPA (1971). Note: A brief summary of this work was publi-
shed* by the Air Pollution Control Association, Pittsburgh,
Pa., as paper no. ME-20C in the Proceedings of the 2nd.
International-Clean Air Congress, Washington, D.C., Dec.
6-11, 1970.
9. Pooler, ?.. 'J. Appl. Meteor.. 2, 446-456 (1963).
10. CeorgU. H.W..'Bull. Wld.Hlth. Orp.. 40,'624-635 £1969).
11. Slade, p.30,
12. Georgli.
13. Holzworth,(Fig. 16}
14.' Ibid.
15. Miller, M.E,, and C.C. Holzworth, J. Air Poll. Control Asspe..
II, 46-50 (1967).
16. Holzworth.
17. Ibid.
18. Korshover, Jules, "Climatology of Stagnating Anticyclones East
-------
18
of Che Rocky Mountains, 1936-1965," Public Health Service
Publication No. 999-AP-34, Cincinnati, Ohio, 15 pp. (1967)
19. Holzworth.
20. Gross, E.M., "The National Air Pollution-Potential Forecast
n
'Program, ESSA Tech. Memo WBTM NMC 47, U.S. Wea. Bur.,
Washington, D.C., 28 pp. 0-970).
-------
19
List of Figures
1. Plume types and associated vertical temperature profiles.
2. Isopleths of annual percent frequency of nights with a low-
level temperature inversion (after Hosier).
3. Cross section of temperature structure through Cincinnati,
Ohio, based on measurements near sunrise. Thin lines are iso-
therms <°F) and heavy lines depict top or bottom of main
temperature inversion (after Clarke).
4. Isopleths of estimated annual afternoon mixing heights (hun-
dreds of meters) above ground-level (after Holzworth).
5. Direction wind roses for stations about Oak Ridge National
Laboratory, Tennessee. Shaded areas represent elevated
terrain (after Slade).
6. Cross section of air circulation and carbon monoxide concen-
trations (parts per million) over a street with tall buildings
on both sides for two different wind speed categories (after
Georgii).
7. Isopleths of annual wind speed (meters/sec) averaged through
the afternoon mixing height (after Holzworth).
8. Isopleths of upper decile values of theoretical city-wide
average morning concentration (x) normalized for average
area emission rate (Q) for an across-city distance of 50 km.
Units of average normalized concentration (x/Q) are sec/meter.
-------
20
9. Isopleths of total number of episode-days in 5 years with
mixing heights < 1500 mo.ters, vind speed i 4.0 meters/sec,
and no significant precipitation for episodes lasting at
lease two days. Numeral gives total number of episode-days
and season with greatest number of episode-days-indicated as
Winter, SPring, Slfamer, or Autumn (after Holzworth).
-------
I
I
UJ
I
FANNING*
TEMP
//
(M\GHT)
/
i
I
X
TEM?
\
FUM\GAT\OM
^Slilliiiip/fff
^itt^f|i||i!'
f f
TEMP
Figure 1. Plume types and associated vertical temperature profiles.
-------
Figure 2. Isopleths of annual percent frequency of nights with a low-level temperature inversion
(after Hosier).
-------
SUNRISE
13 JUNE 1967
WIND DIRECTION
*» 77
Figure 3. Cross section of temperature structure through Cincinnati, Ohio, based on measurements near
(°F) ^ ^ -^ -*«< -P -
-------
Figure 4. Isopleths of estimated annual afternoon mixing heights (hundreds of meters) above ground-
level (after Holzworth).
-------
'r>:^-fw'--
; "... (/'.''
i _ I. i.i|- x^'^jv.^ »"'!. "
i.-i ii.i< '.fW«t»,:'^*aC» * '
.ft ^ t*i _ -:"j
£ ;K&{?V-~:-:-: '-. .\iV .:'t"''i», >!:';
jijjim fHij
CALM
10 20 30
5=e=
% FREQUENCY
SPRING 1950
^
ss
MILES
», Figure 5. Direction wind roses for stations about Oak Ridge
National Laboratory, Tennessee. Shaded areas represent
elevated terrain (after Slade).
-------
TT-W \1SttTSPE E-IX '
Windward
Windward .
« Figure 6. Cross section of air circulation and carbon monoxide
concentrations (parts per million) over a street with
tall buildings on both sides for two different wind
speed categories (after Georgii).
-------
5 5
Figure 7. Isopleths of annual wind speed (meters/sec) averaged through the afternoon mixing
height (after Holzworth).
-------
/o
200
200
loo
100 200
Figure 8. Isopleths of
upper decile values of
theoretical city-wide average
morning concentrations (x) normalized
for average area emission rate (Q) for an
across-city distance of 50 km. Units of
average normalized concentration (x/Q)
are sec/meter.
2oo loo
-------
NOTE: ISOPUTHS FCS DATA AT SAN DIEGO. CALIFORNIA
ARE ttCUWLEU fOH CLARIIT
Figure 9. Isopleths of total number of episode-days in 5 years with mixing heights < 1500 meters,
wind speeds < 4.0 meters/sec, and no significant precipitation for episodes lasting
at least two days. Numeral gives total number of episode-days and season with greatest
number of episode-days indicated as Winter, SPring, Summer, or Autumn (after Holzworth),
------- |