-------
and the greater the lapse rate the greater the diffusion,
To summarize, one property of the atmospheric layer that determines
the rate of atmospheric diffusion within the layer is the vertical termperature
gradient. This is classified as stable, neutral, or unstable in order of
increasing potential for diffusing pollutants (see Figure 3-5).
1 C/lOOm
TEMP
Figure 3-5. RELATION OF TEMPERATURE AND
HEIGHT TO ATMOSPHERIC STABILITY.
3-19
-------
Of interest is th* moat stable configuration, the inversion, in
which temperature actually increases with height. Inversion layers are so
successful at suppressing vertical mixing (have such small diffusion co-
efficients) that in most cases they effectively halt the vertical transfer
of pollution. A combination of large-scale and local weather conditions
determines when and where inversions are formed. When these condition either
cool a layer of air from below, heat it from above, or both, they produce an
inversion in the layer.
For the lowest layer of the atmosphere, cooling froa below is
usually caused by nighttime radiation losses at or near the ground. These
are most pronounced when the sky is clear (no radiation back to the ground
from the clouds) and when the wind is light (a minimum of mechanical mixing
transferring heat from warm air above to cold air below). Under such conditions
very marked inversion layers, usually a few hundred feet in depth, develop
at the surface. Inversions that form in this way are called radiation
Inversions.
Another way in which air can be cooled from the bottom is by moving
horizontally from a warm lower boundary to a cool one, coastline to a
cool ocean. Eddy diffusion transports heat from the air to the ocean,
cooling the lowest part of the atmosphere.
Heating a layer from above invokes less obvious processes. Only one,
adlabatic compression, is of importance. It occurs when a layer of the
atmosphere becomes shallower. For example, If a volume of air next to the
ground spreads out horizontally, it must at the same time get shallower. The
3-20
-------
air at the top is heated by compression as it descends, but the air at the
surface remains at the same elevation and is unheated. This sort of deformation
of air volumes is called subsidence and tends to stabilize the layer;
if carried far enough, it will produce a temperature inversion in the layer.
As mentioned earlier, subsidence is associated with anticyclones: Air tends
to spiral out away from the centers of these high pressure areas, producing
a divergence of the air flow which is automatically accompanied by sinking
of the air over the anticyclone. The presence of an anticyclone over the
eastern United States for a protracted period set the stage for the well-known
air pollution disaster at Donora, Pennsylvania in 1948, and the large
semipermanent anticyclone off the coast of California provides Los Angeles
with its persistent summer inversion.
When the weather situation provides the opposite of the inversion-
producing processes (cooling from above and heating from below), an inversion,
if present, is destroyed. Cooling aloft occurs when the atmosphere is
stretched in the vertical (the opposite of subsidence). This is generally the
case in the vicinity of cyclonic storms and accounts, in large part, for the
accompanying comparative lack of low inversions. Heating from below as the
hot ground passes heat to the air is a regular daytime occurrence over land
in the absence of clouds, and accounts for the rapid weakening—if not complete
destruction—of low inversions during the day.
3. Transport
Horizontal air motions also play an important role in the dilution
and transport of atmospheric contaminants. For example, in a large metropolitan
3-21
-------
area on a day when a low, strong inversion exists, the inversion may be so
stable that it effectively damps out all vertical mixing motion at its level.
Although vigorous mixing motions may exist in the less stable—perhaps even
unstable—layer between it and the ground, the Inversion imposes an upper
limit on this eddy diffusion. Pollution introduced at the ground becomes
diluted with the subinversion air, but the absence of vertical mixing motions
in the inversion layer itself prevents any further penetration of the
contaminants in the vertical.
In the absence of any horizontal motion, the urban environment
would constitute a large area source continually pouring pollution into a
fixed volume delineated by the periphery of the source area and the depth of
the subinversion layer. As long as conditions remained unchanged, pollutant
concentrations would continue to mount in this fixed volume of air. The
obvious and most important effect of horizontal air movement is to bring new,
unpolluted air over the source area and at the same time remove an equal
volume of polluted air on the downwind side. Any such net transport of air
across the source area puts an end to the unlimited increase in pollutant
concentration in the cloud. For any given wind speed and time interval there
is a prescribed volume of air into which the area source can inject con-
taminants; an increase in wind speed Increases that volume and reduces the
pollutant concentration.
For a large city these considerations yield a rather simple picture
for the most important pollution situation for photochemical smog: the day with
the low, strong temperature Inversion. The cloud of pollution produced by
3-22
-------
the multitude of individual sources—some fixed, some moving, some continuous,
others intermittent—is one fairly continuous pall over the city, stretching
out downwind of it with a sharply defined upper boundary. A knowledge of the
mean wind speed, the height of the base of the inversion layer, the diameter
of the source area and the pollutant emission rate would permit a first appro-
ximation of the mean concentration of contaminants in the cloud, providing
that chemical or mechanical removal of the contaminants remained negligible
and the terrain was suitably flat. The local terrain, however, typically
impresses important diurnal fluctuations on air motions and a variety cf
other constratints that complicate the horizontal and vertical flow field.
These can lead to important distortions of the simple picture above.
4. Terrain Effects
The most obvious effect of terrain on horizontal air motions is
to direct and channel the air flow, much as mountain ranges prescribe the
course of rivers. Although this analogy is far from perfect, its applicability
depends on the stability or instability of the atmospheric layer in question.
!f the layer of air covering the terrain has an unstable lapse rate, air
encountering a mountain will more readily move up and over the obstacle than
deviate from its course horizontally, since vertical displacements aid positive
buoyancy forces. On the other hand, air within a stable layer develops
opposing buoyancy forces when displaced vertically, and does behave very much
likr a river in mountainous terrain.
ft. is popular to ascribe the accumulation of smog in the Lo.s
Anj/eLes Basin to this effect: "the damming or trapping of the smog by the
3-23
-------
encircling mountain ranges." But this cliche" resembles reality only part
of the time: during the hours of darkness, cooling takes place from below,
creating stability and thus enhancing the channeling effects of major terrain
features.
In the daytime, however, the layer of air next to the mountain is
heated from below, becomes unstable, and moves upslope along with any
entrained air contaminants. Far from being a dam, the slopes become a vent
for the polluted lower layer during the day.
Terrain also plays a part in determining the rate at which fresh
air is brought in over the source. Winds typically are not constant but have
a diurnal fluctuation—a cyclic variation geared to the coming and going
of the sun and instigated by differences in heating and cooling of various
terrain surfaces. Undoubtedly the largest disparity in terrain as regards
temperature variations is a coastline. Temperature contrasts across a
coastline during a clear day set in motion a large convection regime that
at ground level produces a sea-breese (wind directed from sea toward
land); at night the temperature gradient is reversed, resulting in a land-breeze
in the opposite direction.
Normally this fluctuating wind is superimposed on the large-scale
flow provided by pressure patterns seen on conventional weather maps. The
result is an oscillatory perturbation of this general flow. In Los Angeles
in summer, for instance, the general flow is from sea toward 1: .d. Du^-ug
the day the sea-breeze enhances this flow and at night the land-breeze
opposes it. The result is on-shore movement of around 10 mph during the warn
3-24
-------
parts of the day and very nearly calm or very weak land-breeze at night.
Figure 3-6 shows the halting paths taken by two arbitrarily chosen parcels of
air as they moved across the Los Angeles Basin, illustrating that the afternoon
, • '.
provides hours of good ventilation but the early morning is a period of near
*
stagnation. Under such conditions the source area should-produce a more
4t
dilute polluted wake in the afternoon under the benevolent influehce of the
sea-breeze than it does in the early morning hours, when virtually no new
air is brought in over the source area. This should cause an accumulation
of contaminants that stops only with the midmorning onset of the sea-breeze.
Diurnal variation in ventilation can also occur in the absence of
coastlines as, for example, in valley-mountain regimes. The thermal contrast
producing these winds is that between the alternately heated and cooled slopes
of mountains and the air at the same level out over (and well above) the
valley floor—air, out of contact with the ground, that is scarcely heated
or cooled at all. Figure 3-7 illustrates these daytime up-valley and up-slope
flows and nighttime down-valley and down-slope winds for a valley in western
Argentina. In those important instances where the large-scale pressure
patterns prescribe weak flow, these diurnal wind fluctuations provide whatever
horizontal ventilation an area receives.
D. TRAJECTORY ANALYSIS
The horizontal movement of air provides more than a dilution volume
for air contaminants: it defines the path, or trajectory, of a diffusing air
parcel according to the large-scale mean wind field. In the dynamic sense
che continuing summation, over all sources, of the contaminant loadings
3-25
-------
^^%^r--
r^.^c£*vQ>- «-*?
&&«' jWy*.... ^V>-; .:• .:. rf
IPir^¥(^V^' «-*? "V 1,,
•EM-<'•. S -4/« -., ^V^u ^* <-w ,_ • -4,
Effife' ;;k • :\. \SfcsC.-.-5 ^ ^ttr^ ^—-^ ^f-
P^-^A9^..~J> /^L.v-'sV^ ' ^r>^ «-^>.'-^ ^ rfe
U.^-^-AS/., |'C^-^"- ^,. ' ^---?/--1 x -vK ,?|N
te^lSff f^^f ^x-' ••- "I ^- *
J^^LV vv/^v 8v;fSr v-^>^\ %:j-f-.--- i-r^ ^\ l) ^r^* _^-,f^-- ^x x''->^'%c^iiirwr^V^ -TV -wc..\ ••r-^^-:Avt- ; i
i ^ ^^w ^
^*vv, TPi^.
^-Vv -WVtf
> _-^' ;fv,-
%-•' .-il^
^.w.^v^s .a s
,'i:-^-'v. \^_a^-co\
ib >'- j? 5»*aJe>*yV' '
Vii, .. ^' c3vw^v'^\
v :'*»v
^ ^ *****N*Li'*ir "> i'._-»" " '^ "•'* *.'—"« <
- XJL- r*\ ^-^^^y^--'-1 "-
tfil@ftS
••^N2°
,. vj -T^"*"!"*.
:-t o
'^x
4v;
1
-------
Figure 3-7. PREVAILING SURFACE WIND PATTERNS FOR 6 AM
AND 12 NOON IN THE TUNUYAN VALLEY, ARGENTINA.
3-27
-------
resulting from diffusion and large-scale transport of emissions sets the
air quality for a region having well-defined boundaries. Even here we neglect
losses due to reaction or escape from the system.
To account for the large-scale movement of air parcels some effort
has been devoted to methods of trajectory construction from readily available
wind speed and direction data. Both manual and computer-based techniq>*es
have been used. In practice, manual trajectory construction involves the
analysis of air flow over the area of interest by means of a series of
maps, each one representing the wind pattern at a specific time. Streamlines
of wind flow (lines everywhere parallel to the wind direction) are developed
on the basis of the wind direction plotted at each observational station
location . Wind speeds are also plotted. In most cases, hourly maps
are constructed. The trajectory is developed by moving the point representing
a simulated air parcel in the direction of the wind at the speed indicated
by the streamline map . This process is carried on over a series of maps,
so that each hourly wind pattern makes its contribution to the path of the
parcel, as would be the case if the parcel were actually a part of the air
mass under study. Figure 3-8 shows how this method is performed.
The process of trajectory construction can be carried on both
forward and backward in time. That is, air leaving a specific point is
tracked forward in time as it moves from source to receptor, or, starting at
a point of reception, the prior history cf a parcel of air is determined by
going backward in time via the appropriate series of wind streamline
maps. The example in Figure 3-8 demonstrates the first case, where air
3-28
-------
leaving a specific point is tracked as it moves with the wind, until it is
carried out of the area under study.
Figure 3-8a. T + 1 Figure 3-8b. T + 2
Path of air parcel due to air movement Path of air parcel due to movement
at time T + 1. at time T + 2.
Figure 3-8d. Resulting trajectory,
carried over a series of time steps,
Figure 3-8c. Trajectory movement,
average conditions, T + 1 and T + 2.
Location of parcel at point 2 is due to
graphical averaging of movement due to
wind-flow maps covering time period
T + 1 to T + 2.
Figure 3-8. METHOD OF CONSTRUCTING TRAJECTORIES USING STREAMLINE MAPS.
3-29
-------
A reasonably dense wind station network is required for construction
of air parcel trajectories using the analysis techniques described. Approximately
seventy stations are shown in the 60- x 60-mile grid system used by
8
Davidson, et al. in the New York City area, while about fifty stations
Q
are used by the Los Angeles Network' in an area of about 1500 square miles.
Questions regarding the accuracy of trajectories obtained through
streamline map techniques have been raised from time to time. For example,
how representative of the movement of air parcels is the surface wind?
Inasmuch as parcels are also being moved in the vertical, the net movement
may well be the result of winds aloft rather than just those at the surface.
Studies to verify trajectories have generally taken two forms.
The first measures the wind variation with height, to determine how much
difference exists in wind flows within the lower atmosphere's mixed layer.
If such winds do not vary greatly, the use of surface winds can be justified.
Such was the conclusion of a vertical wind study in the Los Angeles Basin.
Tail-method pilot-balloon observations were used to determine the winds
aloft in these studies.
The second method involves inserting tracer material into the air
and, through a network of sampling stations, determining its location at a
future time. This technique, widely employed to analyze the diffusive
character of an air mass, obviously also provides information about the actual
trajectory of the tagged air parcels .
Currently, fluorescent dusts are most often used for relatively
short-distance tracing. A mineral powder, usually a zinc-cadmium sulfide, is
3-30
-------
dispersed so as to produce an aerosol cloud. The finely divided material Is
carried with the wind, and thus fulfills the requirement of an air tracer.
Detection is accomplished by use of an ultraviolet light source shining on the
sampling surface. The tracer material, of known fluorescent properties, can
12
be identified even in the presence of other dusts . Another tracer system
using sulfur hexafluoride has been developed by the National Center for Air
Pollution Control . This method shows favorable results as compared with
the fluorescent particle technique. It is especially recommended for tracing
air movement over longer distances, since good results have been achieved
in the range of 100 miles or more from the release point.
Another form of an air tracer is the airborne balloon. So-called
"no lift" balloons may be released and their positions tracked visually or by
radar through use of reflecting surfaces attached to the balloons. An
assumption is made that these balloons have no excess buoyancy, and therefore
respond only to the motions imparted by the air mass in which they are
14
released
A variation of the balloon trajectory technique involves the use
of nonexpandable balloons, inflated with helium so that they will float at
a preselected density level. Because of their characteristic tetrahedronal
shape these balloons are known as "tetroons," and have been used to determine
trajectories at various locations -)>-LO>-1-'_
No matter how obtained, any trajectory represents only the path
taken by one parcel initially positioned at a specific time and place. ThcTe
is no assurance that another trajectory starting at the same place but at
3-31
-------
another time would be similar. Therefore, to make any judgment as to the
frequency of occurrence of air parcel movement of a particular nature, it is
necessary to replicate the trajectory-determining process many times.
The product of this process is a climatology of air parcel movements.
The most frequent path may be thought of as the most frequent or "prevailing"
trajectory, analogous to the concept of "prevailing" wind. To predict the
most probable path air pollution will take in going from source to receptor,
it is necessary to have sufficient trajectory determinations on which to
base a statistical judgment. Computer-derived trajectories make it convenient
9
to obtain such information in volume, as has been done in Los Angeles .
The single air pollution episode, however, requires only information related
to the actual cime of occurrence; in this situation, the single trajectory
is sufficient to relate source to receptor.
E. SOLAR RADIATION
A major element in the study of photochemical air pollution is the
effect of solar radiation on the initiation of photochemical reactions. These
•
reactions are discussed in Chapter A of this document.
That region of the solar spectrum important in the primary energy-
absorbing processes (i.e., 3000 to 4000A) varies regularly in Intensity at
the earth's surface according to season and time of day. It also varies
with latitude and presence of attenuating species in the atmosphere. Information
on this atmospheric factor is needed for any complete ambient photochemical
air pollution study.
3-32
-------
1 • Actinic Irradi ance
The light available in urban atmospheres for the promotion of
photochemical reactions includes, but is not limited to, the light transmitted
directly from the sun. An appreciable proportion of the light scattered out
of the. direct beam of sunlight still reaches the surface of the earth, and
this accounts for the brightness of the sky. Under some conditions, such
light from the sky may exceed that direct from the sun. Since scattering is
quantitatively more important for the shorter wavelengths, its effects on the
photochemical smog reaction system cannot be neglected.
Actinic irradiance can be defined as the photon flux through a
horizontal unit surface. In a weakly absorbing medium (such as a polluted
urban atmosphere), the rate of absorption for light of any wavelength is pro-
p, r'-'onal to the concentration of the absorbing material and to the actinic
irradvance for that wavelength. A useful estimate of actinic irradiance in
18
terms of the factors discussed above is given by Leighton
where
I = incident radiation intensity at the top of the atmosphi-re
oX
Ta,\= transmissivi ty (fraction transmitted) directly through absorbing layer
T = transraissivity due to combined scattering and diffusion
sX
g,i a semiempirical parameters representing particular aspects of the
geometry of the scattered light
z = solar zenith angle
3-33
-------
Table 3-1 shows values of J, calculated by this equation, for
100A (10 ran) wavelength bands through the ultraviolet and visible regions
of sunlight, for a series of values of z. Figure 3-9 displays values of J
for several wavelengths as a function of solar altitude, while Figure 3-10
shows the diurnal variation of J-^7nn as estimated for various latitudes at
the summer and winter solstices. Within the latitudes of the contiguous
forty-eight states, as Figure 3-10 shows, there is little variation with
respect to either the maximum actinic irradiance (at 3700A) or the integrated
24-hour value at the summer solstice. However, at the winter solstice,
the effect of latitude on these parameters is quite great. At other times
of the year, of course, the latitude effect is intermediate, with the
greatest variation between summer and winter at the higher latitudes. For
reference purposes, the three latitudes shown are approximately those of
London, Los Angeles, and Mexico City.
2. Effect of Smog on Atmospheric Transmission
21
Measurements by Stair , taken during a severe episode of smog in
Pasadena, California, showed clearly that transmission of the direct solar
beam was sharply reduced at wavelengths approaching 3000A. This effect
18
is illustrated by Figures 3-11 and 3-12, both from Leighton . The
effect at wavelengths between 3200 and 5000A was well represented by the
empirical equation
S =• 0.133\~lt5
PX
where
S = attenuation coefficient of the polluted layer
r A
3-34
-------
Table 3-1. ESTIMATED ACTINIC IRRADIANCE
IN THE LOWER ATMOSPHERE.
X(A)
2000
3000
3100
3200
3300
3400
3500
3000
3700
3800
3900
4000
4100
4200
4300
4400
4500
4000
4700
4800
, 4900
6000
6250
6500
6750
0000
6250
0500
0750
7000
7500
8000
<
0°
0.0014
0.12
0.05
1.17
1.83
1.88
2.00
2.10
2.4S
2.30
2.20
3.10
4.01
4.00
3.80
4.50
6.00
6.00
5.13
6.20
4.88
• 4.95
5.14
5.30
6.32
6.32
5.27
6.22
5.10
5.05
4. SO
4.55
/» X 10-", photons
20'
0.0009
0.10
0.00
1.10
1.75
1.82
1.98
2.02
2.40
2.28
2.13
3.01
3.00
3.95
3.70
4.39
4.88
4.89
5.02
6.11
4.78
4.85
5.05
6.21
6.23
6.24
5.19
5.10
5.10
6.00
4.75
4.51
(cm'1 sec"1
40*
0.0002
0.05
0.43
0.91
1.48
l'.W
1.72
1.77
2.11
2.02
1.90
2.70
3.61
3.57
3.42
4.02
4.48
4.50
4.04
4.72
4.44
4.51
4.73
4.90
4.92
4.95
4.94
4.92
4.89
4.82
4.59
4.37
100 A"') at « -
60°
—
0.01
0.13
0.53
0.90
1.05
1.18
1.24
1.50
1.46
1.38
2.00
2.03
2.71
2.02
3.11
3.61
3.56
3.08
3.79
3.59
3.08
3.91
4.09
4.14
4.21
4.25
4.30
4.32
4.29
4.10
3.98
80"
—
__
0.01
0.10
0.20
0.32
0.36
0.38
0.40
0.44
0.42
0.02
0.82
0.80
0.85
1.03
1.19
1.22
1.30
1.38
1.33
1.40
1.50
1.09
1.75
1.80
2.01
2.10
2.29
2.38
2.47
2.47
• Yfcluri HIVI-D *rc from rrj-lation (11-37), vith I-\ f;otn T»Me 1, T*v from Table 4, and T*i from equa-
tion 01-13). rtxtmetrrc rni| foycd ire P - 1000 tab. w - 2. d - 1; (O») -2V moi. 9 » 0-5. and • - 2.
3-35
-------
9.0
ft.O
30 60
Sffar Alliludf, 90-f
•o
Figure 3-9. VARIATION IN ACTINIC IKRADIANCE
WITH SOLAR ALTITUDE.
3-36
-------
IS
1.0
I I I
I I I I
66769 10 II 12 IZJ4
AM Ttut Solar Tint
$6
Figure 3-10. DIRUNAL VARIATIONS IN ACTINIC IRRADIANCE.
NOTE: Values illustrated are for the 100A interval of the solar spectrum
centered at 3700A. The designations of the individual curves: I,
20° N lat, summer solstice; II, 35° N lat, summer solstice; III, 50°
N lat, summer solstice; IV, 20° N lat, winter solstice; V, 35° N 1-jt,
winter solstice; and VI, 50° N lat, winter solstice.
3-37
-------
r DIFFUSION WITHIN THE POLMTTED LAYER
ie
!
e •
10 II 12 I 8
All. Pttllle Sltndtrt Timt P.M.
Figure 3-11. ATTENUATION OF DIRECT SOLAR ENERGY BY PHOTOCHEMICAL SMOG.
NOTE: Both lines represent the direct solar energy over a 10A bandwidth
centered at 3235 A, incident on a normal surface at 800 feet elevation.
The smooth line is I .Ts.Ta^, calculated from Table 1 and equations
(11-9), (11-11) and (11-13) with P-985 mb, w-2, d-0.5, and (Os)-2. The
irregular line is the observed energy, from Stair's data for 0,-.::3bar 18,
1954 at Pasadena21.
3-38
-------
38
1.0
H. SOLAR 1UDUTIO.V .VXD ITS AKSOHPTION
te
£06
0.4
«
0
I
JOOO SZOO
3500
4000
, A
1500
5000
Figure 3-12. VARIATION IN THE ATTENUATION COEFFICIENT
OF PHOTOCHEMICAL SMOG WITH WAVELENGTH.
3-39
-------
19
Similar measurements were made in Los Angeles during October, 1965
with a variety of UV sensors, including a filter phototube, filter
photocell, photochemical sensors, photosensitive plastic, and photochromic
glass. Overall attenuation from smog in the region of 3000 to 3800 A incidental
radiation was about 14 percent. Peak attenuation of 58 percent during the
study period occurred during moderate to heavy smog as determined by chemical
pollutant measurements.
Although the direct radiation from the solar beam may be reduced
at some wavelengths by more than 80 percent as shown in Figure 3-11, the
effect of smog on actinic irradiance is more difficult to estimate. The
angular distribution of the scattered light is not known, but is likely
19
to have a substantial effect on estimated values. Leighton , however,
studied estimates based on two postulates which were intended to bracket the
probable conditions. He concluded that for a solar zenith angle less than
60° (i.e., solar altitude above 30°), the average actinic irradiance in
smog should be greater—by perhaps as much as 25 percent—than that found in
the absence of smog. Since smog is more likely to be encountered when the
solar zenith angle is low, this conclusion suggests that the generation of
smog aerosols by the photochemical reaction system may further accelerate
the photochemical process by increasing the actinic irradiance within the
smog layer.
20
Nader and White , during a comparison study of horizontal
plate and volumetric photovoltair sensors, did find some evidence of a slight
increase in UV radiation in smoggy air as compared to clear air at solar
3-40
-------
zenith anglrrb. Much more information from field measurements of UV
radiation will be required to enable refined predictions of the effect of
<*
this parameter on photochemical smog formation.
3-41
-------
REFERENCES
1. Korshover, J., Synoptic Climatology of Stagnating Anticyclones,
U.S. Public Health Service, R. A. Taft Sanitary Engineering Center,
Tech Report A60-7, Cincinnati, Ohio, 1960. (Updated to 1965 by PHS
Pamphlet HEW AP-34, 1967).
2. Holzworth, G.C-, A Study of Air Pollution Potential for the
Western United States, Journal of Applied Meteorology, Vol. 1, No. 3,
366-382, 1962.
3. Holzworth, G. C., Mixing Depths, Wind Speeds and Air Pollution Potential
for Selected Locations in the United States, Journal of Applied
Meteorology, Vol. 6, No. 6, Dec. 1967.
4. Holzworth, G. C., Large-scale Weather Influences on Community Air
Pollution Potential in the United States, Journal of the Air Pollution
Control Association, 19, 4, 248-254, 1969.
5. Turner, D. Bruce, Workbook of Atmospheric Dispersion Estimates,
PHS Publ. No. 999-AP-26, 1967.
6. Saucier, W. J., Principles of Meteorological Analysis, University of
Chicago Press, Chicago, Illinois, 1955.
7, Petterssen, S., Weather Analysis and Forecasting, McGraw-Hill,
New York, pages 221-227, 1940.
3-42
-------
8- Davidson, B. and C-workers., Final Report on the Urban Air Pollution
Dynamics Research Project,New York University Report, December, 1969.
.9. Taylor, J. R., Normalized Air Trajectories and Associated Pollution
Levels in the Los Angeles Basin, Air Quality Report #45, Los Angeles
County Air Pollution Control District, Los Angeles, California, 1962.
10. Kauper, E.K., Homes, R. G., Street, A. B., "The Verification of
Surface Trajectories in the Los Angeles Basin by Means of Upper
Wind Observations and Tracer Techniques," Technical Paper #14,
Los Angeles County Air Pollution Control District, updated.
11. Slade, D. H., Ed., "Meteorology and Atomic Energy, 1968," U. S. Atomic
Energy Commission, pages 293-298, July 1968.
12. Leighton, P. A., Perkins, W. A., Grinnell, S. W., Webster, F. X.,
"The Fluorescent Particle Atmospheric Tracer," J. Appl. Meteorol., 4_,
(3), 334-348, June 1965.
13- Niemeyer, L. E., McCormlck, R.A., "Some Results of Multiple-Tracer
Diffusion Experiments at Cincinnati," J. Air Poll. Control Assoc., 18,
(6), June 1968.
14. Gifford, F. A., Jr., "A Study of Low Level Air Trajectories at
Oak Ridge, Tennessee," Monthly Weather Rev., 81, (7), 179-92.
3-43
-------
15. Angell, J..K. , "Use of Constant Level Balloone in Meteorology,"
Advances in Geophysics, Vol. 8, 137-219, H. E. Landsburg, J. Van Meighem,
Eds., Academic Press, New York., 1961.
16. Pack, D. H., "Air Trajectories and Turbulence Statistics from Weather
Radar Using Tetroons and Radar Transponders," Monthly Weather Rev. ,
90, (12): 491-506, 1962.
17. Pack, D. H., and Angell, J. K., "A Preliminary Study of Air Trajectories
in the Los Angeles Basin as Derived from Tetroon Flights", Monthly
Weather Rev., 9^(10-11): 583-604, 1963.
18. Leighton, P. A., Photochemistry of Air Pollution, Academic Press,
New York, 1961
19. Nader, J. S., "Pilot Study of Ultraviolet Radiation in Los Angeles,
October 1965" PHS Publ. No. 999-AP-38, 1967.
20. Nader, J. S. and White, N., "Volumetric Measurement of Ultraviolet
Energy in Urban Atmosphere", Env. Sci. , and Tech., 3., 848-854,
September 1969.
21. Stair, R., Proceedings National Air Pollution Symposium, Pasadena,
California, 1955, pg. 48.
3-44
-------
CHAPTER 4
ATMOSPHERIC REACTIONS
Page
A. INTRODUCTION 4-8
B. PRINCIPLES OF PHOTOCHEMICAL PRIMARY REACTIONS 4-9
1. The Initiation of Photochemical Reactions by the Absorption 4-9
of Solar Radiation.
a. Light and Quanta 4-9
b. The Laws of Photochemistry 4-11
c. Photochemical Primary Processes for Simple Molecules 4-14
2. Molecules that Absorb Light in Urban Atmospheres 4-18
3. Photochemical Primary Processes for Absorbers in Urban 4-25
Atmospheres.
a. Oxygen 4-25
b. Ozone 4-26
c. Nitrogen Dioxide 4-26
d. Sulfur Dioxide 4-27
e. Nitric Acid 4-28
f. Alkyl Nitrates 4-28
g. Alkyl Nitrites 4-29
h. Nitroalkanes 4-29
i. Aldehydes 4-30
j. Ketones 4-30
4-1
-------
Page
k. Peroxides 4-30
1. Acyl Nitrites, Nitrates, and Peroxynitrates 4-31
m. Particulate Matter 4-31
C. ATMOSPHERIC REACTIONS 4-31
1. Kinetics of Thermal and Free Radical Reactions 4-31
a. Rate Laws 4-32
b. Collision Theory 4-35
2. Thermal Reactions 4-38
3. The Generation of Free Radicals 4-40
4. General Treatment of Radical Chain Reactions 4-42
5. Free Radical Reaction Types in Photochemical Smog 4-45
D. A THESAURUS OF PROPOSED REACTIONS IN THE URBAN PHOTOCHEMICAL 4-48
SYSTEM.
1. Inorganic Reactions 4-48
2. Reactions of Atomic Oxygen and Hydrocarbons 4-53
3. Reactions of Ozone and Hydrocarbons 4-59
4. Reactions Between Free Radicals and Molecular Oxygen 4-63
or Ozone.
5. Reactions Between Free Radicals and NO 4-69
6. Reactions Between Free Radicals and NO,, 4-73
7. Reactions Between Free Radicals and Hydrocarbons 4~?6
8. Reactions Between Free Radicals 4-80
4-2
-------
Paj-e
9. Radical Decomposition Reactions 4-83
10. Summary of Important Reactions of Radicals 4-86
E. PHYSICAL AND CHEMICAL PROPERTIES OF THE PHOTOCHEMICAL REACTION 4-87
PRODUCTS.
1. Ozone 4-87
2. Peroxyacyl Nitrates 4-87
3. Oxides of Nitrogen, NO 4-90
4. Aldehydes and Ketones 4-90
5. Carbon Monoxide 4-91
6. Nitric Acid 4-91
7. Sulfuric Acid 4-91
8. Aerosols 4-92
9. Epoxides 4-93
10. Alkyl Nitrates 4-93
11. Alkyl Nitrites 4-93
12. Peroxy Compounds 4-93
13. Alcohols 4-93
14. Ketene 4-93
F. COMPUTER SIMULATION OF CHEMICAL KINETICS 4-94
1. Rationale for Kinetic Simulation 4-94
2. The Basic Simulation Process 4-97
4-3
-------
Page
- 'I *~*L
3. Acceptable Simplifications 4-100
a. Approximate Constancy of Concentration of Components 4-101
Present in Large Excess.
b. Steady State Concentrations for Extremely Short-Lived 4-102
Species.
c. Aggregation of Free Radical Termination Steps 4-104
4. Photo-oxidation Simulations 4-104
4-4
-------
t
*
CHAPTER 4
LIST OF FIGURES
Figure Page
4-1 Summary of Specific Absorption Rates. 4-20
4-2 Absorption Spectrum of SO (g) at 25° C. 4-21
4-3 Absorption Spectrum of NO- at 25° C. 4-21
4-4 Molecules A and B in Collision 4-J6
4-5 Time-Dependence of Concentrations in Simulated Photo- 4-108
oxidation of Propylene for 40-Step Mechanism, Compared
to Experiment.
4-6 Time-Dependence of Concentrations in Simulated Photo- 4-111
oxidation of Isobutene, Compared to Experiment. From
73
Curves Presented by Westberg and Cohen.
4-5
-------
CHAPTER 4
LIST OF TABLES
Table Page
4-1 Approximate Limits of Spectral Regions 4-12
4-2 Some Useful Units and Conversion Factors 4-13
4-3 Some Bond Energies of Molecules of Interest in 4-21
Photochemical Smog.
4-4 Species that Absorb Sunlight in the Region 3000A" to 700oA 4-22
4-5 Species Energetically Possible from 0_ + hv -»• 4-26
4-6 Conversion Factors 4-33
4-7 Some Free Radicals and Atoms Required by Photolysis 4-41
4-8 Free Radical Classes Formed in Irradiated Polluted Air 4-43
4-9 Inorganic Reactions 4-49
4-10 Rate Constants for Thermal Homogeneous Inorganic Reactions 4-50
4-11 Rate Constants of Atomic Oxygen-Hydrocarbon Reactions 4-55
4-12 Rate Constants for Ozone-Hydrocarbon Reactions 4-60
(£. mole sec ) x 10 .
4-13 Some Reactions of 0- with Free Radicals 4-70
4-14 Some Reactions of NO and N02 with Free Radicals 4-77
4-15 Some Free Reactions with Free Radicals 4-84
4-16 Some Properties of Photochemical Reaction Products 4-88
4-6
-------
Table
4-17 Illustration of Matrix Method of Formulating Reaction 4-
-------
A. INTRODUCTION
Reactions involved in the urban photochemical system fall readily
into two classes: (1) primary photochemical reactions, in which contaminants,
energized by light, rearrange or decompose to produce excited molecules or
decomposition products; and (2) secondary reactions, in which products of the
primary photochemical reactions react with other molecules, either contaminants
or normal atmospheric constituents.
Photochemical reactions are initiated by the absorption of light by
molecules. This interaction changes the configuration of electrons in a
molecule, placing it in an electronically excited state. The short-lived
excited molecule can lose its excess energy by collision with other molecules
or by spontaneous emissions of energy, thereby returning to its original
ground state. It may undergo internal rearrangement, may split up into two
or more fragments, or may react with other molecules. Dissociation and
rearrangement constitute photochemical primary reactions, whereas fluorescence
and collisional deactivation do not lead to further chemical reaction. Thus,
ordinarily, only a fraction of the light absorbed is effective in promoting
chemical change.
Products of the primary processes are available to react further by
entering into thermal (i.e., nonphotochemical) interactions with species
with which they collide. Some important photochemical reactions involve the
4-8
-------
dissociation of excited molecules into free radicals* or atoms which,
reacting with the components of polluted air, are responsible for most of the
unique characteristics of photochemical smog.
The secondary chemical reactions in urban atmospheres include
atom-molecule, radical-molecule, radical-radical, and molecule-molecule
reactions. The purpose of this chapter is to describe the processes of
photochemical excitation and the subsequent types of reactions which it
initiates.
B. PRINCIPLES OF PHOTOCHEMICAL PRIMARY REACTIONS
!• The Initiation of Photochemical Reactions by the Absorption of
Solar Radiation
a. Light and Quanta
Light as a form of electromagnetic radiation is commonly described
in terms of wave-like properties, especially the wavelength,**
A - £ (4-1)
in which C is the speed of light, 2.9979 x 10 cm/sec., A is its wavelength
*A radical is a group of atoms forming a coherent structure which may be part
of various molecules; a free radical is a molecular fragment consisting of
such a group, unattached to other groups or atoms, and carrying one or more
unpaired electrons which render it very reactive chemically, as illustrated
in this chapter. Free atoms, such as those of oxygen, are also very
reactive for the same reason.
**The wavelength of a wave is defined as the distance between two successive
peaks in that wave.
4-9
-------
and v is the frequency of the wave. However, certain properties of light, such
as the photoelectric effect, require a particle theory. For the purposes of
photochemistry, both theories are useful. In describing the absorption of
energy, it is convenient to recognize the particle of light, or photon, as an
irreducible bundle of energy, also called a quantum. The energy of a quantum
is given by
E - hv - ^ (4-2)
A
-27
in which h is Planck's constant, equal to 6.6256 x 10 erg-sec/quantum.
Light is emitted or absorbed by molecules in a quantum process that
results in the transition of an electron from one quantum state to another.
When light is absorbed, the electron is promoted to a higher energy level and
when light is emitted, the electron falls to a lower energy level. The
difference in energy between the two states is the same as the energy, hv, of
the quantum (photon). (In photochemical notation, the term hv is commonly
used in place of a chemical symbol to represent a photon.)
The energy of interest in photochemical processes is that necessary
to excite molecules and dissociate chemical bonds. Since most bonds are at
least 40 Kcal/mole or greater in energy, the wavelengths of interest here are
o
less than 7000A, which corresponds to energies of 40 Kcal/mole or more.
o o
Solar radiation at the earth's surface between 3000A and 7000A is in the
range that could excite molecules with sufficient energy to cause photochemical
reactions. The scientific relation between available chemical energy ana
4-10
-------
wavelength of absorbed light, derived from equation (4-2), is
E - 2.86 x 105/X (4-3)
where E is in Kcal/mole and X in Angstroms. Table 4-1 shows the characteris-
tics of various wavelength regions of the electromagnetic spectrum, named
according to common usage. Some useful energy units, their relationships and
the values of constants referred to in this chapter are given in Table 4-2.
b. The Laws of Photochemistry
Grotthuss (1817) and Draper (1843) stated that "Only the light which
is absorbed by a molecule can be effective in producing photochemical change
in the molecule." This statement constitutes the first law of photochemistry.
The second law, described by Stark (1908-12) and Einstein (1912-13), states
that each molecule taking part in a photochemical reaction absorbs precisely
one quantum of the radiation which causes the reaction.
In the light of these laws, a useful concept for quantitative
photochemistry is the quantum yield, which is the ratio between the extent of
a particular reaction or the amount of a particular reaction product observed
and the amount of light absorbed, in terms of molecules per quantum. Thus the
second law implies that the quantum yield for excitation is unity, since
every absorption event produces an excited molecule.
As another consequence, the sum of the primary process quantum
yields is unity, since each excited molecule is soon converted either to a
ground state molecule or to photochemical reaction products. These primary
processes may include dissociation, rearrangement, fluorescence and any other
4-11
-------
00
2-
u,{
CJJ
i
i*4
J
s
IT-*
8
Oi
CO
EM
C/3
^TJ
s
_1
H^
CQ
2
^
h-4
X
§
PH
%
—**
rH
|
o
^
o
C!
O
rH
<"•%
^
o
X
\^
o
4-1
in
*
CN
vrP
rH
O
rH
^
^^
i-H
O
4-1
m
•^
o
o
o
o
o
4J
0
o
o
en
4J
0)
rH
O
•rl
CO
^4
4J
rH
3
M
id
0)
rH
^
O
i-H
y-^
^>
O
rH
>*•/
m
XI
m
0
rH
X
cn
^j-
•
o
jj
o
rH
a)
XI
o
o
o
A
o
cn
v
o
XI
cd
1
cd
^4
M-i
£j
•H
M
Id
01
>
$
id
'•W
O
CO
4J
CO
01
rl
O
Pi
01
0>
5
4J
0)
XI
01
§
4J
co
iH
0)
X!
4J
CO
cd
•o
0)
C
fl
14H
rH
i-H
3
CO
xT
60
d
0)
rH
0>
>
rt
3
/-%
id
.
M
01
4J
1
4-1
S
U
CX>
1
0
rH
CO
•^
"
*
01
•
1-*
*£
4J
50
0
01
rH
01
cd
01
X!
4-1
14H
0
rH
cd
U
O
a
u
0)
0)
X!
4J
CO
•H
l>l
•>
0* 0)
i x>
4-1 3
c
d
a) 0)
> >
•r4 ft)
60 S
X^
3
CO
01
rH
3
O
01
^
M
O
ca
e
0
4-1
CO
M-t
o
c
o
•H
4J
Id
4J
•H
1
O)
^4
o
co
c
o
•rl
4J
rt
A
£4
O
CO
c
rl
4-1
rH
cd
O
•H
a>
u
rl
O
14-1
01
rH
•s
rH
•rH
id
>
id
00
tit
c
w
/— 1^
(U
J
£
1-1
rH
00
d
^
o
«s
XI
cd
4-12
-------
Table 4-2. SOME USEFUL UNITS AND CONVERSION FACTORS.
erg/molecule e.v
cm
-1
cal/mole
erg/molecule
6.242x10
11
5.036xl015 1.439xl016
e.v.
-1
~12
1.602xlO
1.986xlO~16 1.2396xlO~A
8.067
23,060
2.858
cal/mole
6.949xlO~17 4.338xlO"5
0.3499
_o
1 erg - 2.39x10 calories.
4-13
-------
processes chat lead to the deactivation or change in the molecule that has
absorbed the quantum of energy. Whatever the process, only one quantum of
energy has been absorbed by each molecule. Therefore, the quantum yield for
a primary process can reach a maximum of one if every absorption event results
in that process. The quantum yield for primary photochemical reaction is less
than one if some of the molecules that are excited are subsequently
deactivated and restored to their original states.
On the other hand, any values of observed over-all quantum yields
greater than unity indicate the existence of secondary (thermal) reactions.
This means that, following the primary processes, thermal reactions Involving
the products of these processes destroy more of the original molecules. Thus,
more molecules are lost than quanta of light absorbed. Detailed examples are
given in the following sections.
The Beer-Lambert law is also applicable to the systems encountered
in photochemical smog. The magnitude of the absorption coefficient serves as
an index of the relative amount of absorption to be expected at a given
wavelength of light. If the absorption coefficient of a given species is
o o
appreciable in the range of 3000A to 7000A, that species will be an important
absorber of sunlight and, therefore, must be considered in the discussion of
the primary processes in atmospheric photochemistry. Nonabsorbers, however,
may be important in subsequent thermal reactions.
c. Photochemical Primary Processes for Simple Molecules
The energy possessed by atoms and molecules may be considered to take
on various forms. In atoms, the energy can be nuclear or electronic, while in
4-14
-------
molecules consisting of more than one atom the energy can be nuclear, electronic,
rotational or vibrational, corresponding to the various modes of energy
absorption. Translational or kinetic energy of motion is common to all
molecules.
The absorption of electromagnetic radiation may excite any of these
modes of activity if the radiation is of the proper wavelength. Quantum
theory asserts that the energy can be absorbed or emitted only in discrete
packets called quanta.
Each time a quantum of energy of a given kind of absorbed, an
increment of activity of the appropriate kind is stimulated. Thus, when
8°
energy corresponding to wavelengths of about 10 A is absorbed, a molecule will
4°
rotate more energetically. When energy corresponding to 10 A is absorbed, the
molecule will also vibrate more energetically. When energy corresponding to
3°
wavelengths less than 7 x 10 A is absorbed, an electronic change is induced
in the molecule. These changes to higher energy take place in single quantum
Jumps and the molecules are said to be raised to higher quantum states or
energy levels. In the photochemical process, only those absorptions that
cause electronic change lead to photochemical activity.
A number of excited states are usually available for various values
of the excess electronic energy for a particular type of molecule. When it is
necessary to differentiate between these states, they are referred to by
quantum mechanical designations known as "term symbols." For example, the
3 1
ground state of the oxygen molecule is E , its first excited state is A ,
o o
4-15
-------
1 + 3
and its second excited state is 2. The ground state of the oxygen atom is P,
while the lowest excited state is D. Detailed explanations of these terms
can be found in standard texts on quantum chemistry.
For convenience, less specific indications of electronic excited ,-
states are often used; thus, 0~ ( A ) is often referred to simply as "singlet
oxygen" in contrast to the triplet ground state, Z . Again, the asterisk is *
often applied to Indicate an unspecified excited state; for example, 0* might
stand for any of the excited states of molecular oxygen, without discriminating
betwe&n them.
Processes in which molecules are changed from one electronic state
to another by absorption or emission of energy are called transitions and
often written in a manner analogous to chemical reactions. Thus
02( 2~) + hv -*• 02(1s+) (4-4)
represents the transition of oxygen from the ground state to the second
excited state of an oxygen molecule, caused by absorption of a photon. (Light
o o
appropriate for this transition is in the wavelength range from 6920A to 7660A.)
A description of all the possible energy states of any given
molecule and of the processes which may give rise to the various transitions
among them constitutes a very complicated exercise in theoretical photochemistry,
for which the reader is referred to standard texts. For an approximate
understanding of the photochemistry of urban atmospheres, it is sufficient
to recognize the following general considerations: *
4-16
-------
1. An electronically excited molecule is produced whenever a
molecule absorbs a photon.
AB + hv •»• AB* (excitation) (4-5)
2. The lifetime of the excited molecule is short, and can be
terminated by various processes: decomposition (4-6), reaction
(4-7), fluorescence (4-8), collisional deactivation (4-9).
AB* -»• A + B (dissociation) (4-6)
AB* + C -»• Products (4-7)
AB* ->• AB + hv (fluorescence) (4-8)
AB* + M -»• AB + M (collisional (4-9)
deactivation)
Some of these processes lead to disappearance of the original
molecule and formation of photochemical products, while others
do not.
3. Decomposition of the excited molecule is very rare unless the
quantum energy of the absorbed photon exceeds the bond energy
of at least one chemical bond in the molecule.
4. Energy can be transferred from the excited molecule to another
molecule with which it collides. The latter molecule then
4-17
-------
behaves as if it had absorbed light, while the original
absorbing molecule is deactivated. (This process is called
photosensitization. It seems to be rather rare in the
urban atmosphere, but may be responsible for some
production of singlet oxygen.)
The processes described in equation (4-5) followed by (4-6) or
(4-7) constitute possible photochemical primary reaction
paths, while those described by (4-5) followed by (4-8) or
(4-9) return the molecule to its original state and do not
lead to chemical change.
2. Molecules that Absorb Light in Urban Atmospheres
A very large number of compounds absorb some sunlight in the
atmosphere. However, the absorption will not promote photochemical reactions
if the energy is too low, nor will the reaction be of importance if the
intensity is too low. The energies necessary to dissociate certain types
of chemical bonds are listed in Table 4-3. An expression derived from the
Beer-Lambert law is
I = I (l-10~acl) (4-10)
a O
where I is the intensity of Incident light, a is the decadic absorption
o
4-18
-------
coefficient* of the absorber, c is its concentration, and 1 is the length of
the absorption path. Since the absorption coefficient is a function of
wavelength, to obtain total absorption values, equation (4-10) must be
applied for all the wavelengths absorbed, and the values of I summed (or
a
integrated) over the range of interest.
Absorption rates for most of these compounds in urban atmospheres
were estimated and presented by Leighton as a function of solar zenith
angle (Figure 4-1). All indications are that nitrogen dioxide is the
dominant absorber in terms of specific rates.
A summary of the molecules of interest in "smog" with their
absorption coefficient ranges and wavelengths is given in Table 4-4.
Most of the molecules shown in Table 4-4 absorb light over a very
o o
limited range within the 3000A to 7000A region of the spectrum. Absorption
o
is limited to a region near 3000A where the intensity in sunlight is low
because of absorption in the upper atmosphere. In addition, most of their
absorption coefficients are quite low. As a consequence, these molecules are
not considered important as principal absorbers in urban atmospheres.
*The notation used here is that of Leighton. It is to be noted that some
authorities2 call this coefficient the "extinction coefficient" and
denote it by e.
4-19
-------
I
I
0.01 r
f
•«
OOOll
ZO* 40* 60*
So/or Ztnith Angli, Z
80*
Figure 4-1. SUMMARY OF SPECIFIC ABSORPTION RATES.
4-20
-------
Table 4-3. SOME BOND ENERGIES OF MOLECULES
OF INTEREST IN PHOTOCHEMICAL SMOG.
Molecule
°2
°3
CO
co2
NO
N02
SO,
H20
HO
H02
H2
HCO
RCHO
CH3COCH3
HNO
HONO
HON02
RONO
RONO
RON02
RON00
Bond
0-0
°2-°
c-o
oc-o
N-0
ON-0
OS-0
H-OH
H-0
H-02
H-H
H-CO
R-CHO
CH3C-°-CH3
ON-H
ONO-H
02N-OH
RO-NO
R-ONO
RO-NO
R-ONO 0
Energy Kcal/mole
118.3
25.2-26
257
127
151
72.7
131
119
102
47
104.2
18
80
79
49
78
53
36-40
55-60
36-37
78-79
4-21
-------
Table 4-4. SPECIES THAT ABSORB SUNLIGHT IN
THE REGION 300oX to 7000A.
Substance
Maximum Absorption
Coefficients (liter
mole cm
Wavelength of Maximum
Absorptions within
Range (Angstroms)
°2
°3
N02
so2
HNO- (HONO,)
•J &
RON02
RONO
RON02
RCHO
RCOR1
ROOR'
0
RCONO
P
RC ONO^
7 x 10"6
110
171
220
121
1.2
0.9
40
8.4
5
16
12
2.3
—
—
6900
3000
4000
3000
3000
3000
3000
3600
3000
3000
3000
3000
3000
— —
—
PAN
0.5
3000
4-22
-------
Ozone and sulphur dioxide do have appreciable absorption coefficients
o
in the range of interest, but their main absorption is near 3000A where
little sunlight energy is available. Absorption by ozone is of sufficient
energy to dissociate the molecules and some photodissociation should take
place. However, in the case of SCL, even if absorption does take place,
o
the energy at 3000A is not sufficient to cause the molecule to dissociate (the
bond energy of OS-0 is about 131 Kcal/mole). The absorption spectrum of SO
is shown in Figure 4-2,
o o
Figure 4-3 shows the absorption spectrum of NO from 2500A to 5000A.
It can be seen that absorption takes place over the entire range. In addition,
o
it has been shown that photodissociation is possible below 4350A. It is
evident, therefore, that N0_ is the principal absorber of sunlight, and plays
one of the most important roles in the photochemical process that takes place
in the urban air. Its photodissociation yields free oxygen atoms that react
further with hydrocarbons and other components of polluted air. As discussed
below, the subsequent molecular and free radical reactions can explain a
great deal of the smog process.
Although some of the products of photochemical and thermal reactions
may be formed in appreciable quantities, their residual concentrations in the
air may be limited by photodissociation. This is more likely for those with
large absorption coefficients (Table 4-4), since they would absorb sunlight
more strongly. N0_ is an example of this type of behavior.
4-23
-------
0
2000 2200 2400 2600 2800 3000 3200 3400
Figure 4-2. ABSORPTION SPECTRUM OF SO, (g) AT 25° C.
o
296
T 148.
* 74
2500 3000 3500 4000 4500 5000
Figure 4-3. ABSORPTION SPECTRUM OF NO AT 25° C.
4-24
-------
3. Photochemical Primary Processes for Absorbers in Urban Atmospheres
a. Oxygen
3 -
The absorption by ordinary molecular oxygen ( £ ) is in general too
o
o o
weak to lead to dissociation. Absorption between 6870A and 6920A yields about
41.5 Kcal/mole. The energy required for dissociation is about 119 Kcal/mole.
Therefore, the direct dissociation of 02 is negligible. However, oxygen may
be excited by the spin-forbidden process,
02 (32J + hv f 0 (VS (4-11)
o o
The excited ( £ ) molecules might react directly with ozone or other species.
O
A new photo-oxidation mechanism based on singlet molecular oxygen
4 5
has been proposed by Pitts and coworkers. ' Singlet molecular oxygen has
a theoretical lifetime of only a few seconds, thus the concentrations produced
by direct absorption of sunlight can be expected to be quite small. This led
Leighton to conclude in 1961 that the role of 0 * was unimportant. Pitts,
4 5
et al. ' pointed out that a high yield of singlet oxygen might be obtained
in a system in which an organic molecule absorbs energy from solar radiation
and then, on collision, transmits it to normal molecular oxygen to produce
singlet excited oxygen. The overall mechanism is represented as
D(S )+hv - D*(S,)
o 1
D*(S1) -
D**(TX) + 02(V) - D(SQ) + 02( or
(where D represents the "donor" organic molecule). Subsequent reaction of 0
with olefinic substances produces thermally unstable hydroperoxides which may
be involved in the rapid conversion of NO to N0?.
4-25
-------
b. Ozone
The photolytic processes that are energetically possible are given
in Table 4-5.
Table 4-5. SPECIES ENERGETICALLY POSSIBLE FROM 03 + hv •*-.
Limiting
Wavelength Species
11,400 °2(3V * °(3p)
5,900 O.^A ) + 0(3P)
L 8
4,600 V"^ * °(3p>
4,100 °2<3lg) +
3,100 02(1A ) + O^D)
Most of these processes are spectroscopically forbidden and, therefore, do not
occur at appreciable rates. However, the 0 atoms and the excited 0» molecules,
to the extent that they are produced, would presumably contribute to secondary
reactions in urban atmospheres.
c. Nitrogen Dioxide
Nitrogen dioxide is probably the most important absorber. At
0 7
wavelengths below 4047A, it dissociates readily by the process,
3
N02 + hv -»• NO + 0( P) (4-12)
4-26
-------
The oxygen atoms produced in this reaction are probably responsible, directly
or indirectly, for the majority o/ the thermal oxidation reactions xn urban
pnotochenustry. The primary process probably involves oirei t excitation
2 2
from AI (the ground state for NO- to A for energies greater than the ON-'
o
bond energy of 71.8 Kcal/mole. However, at 4047A the molecule could be
2 2
excited to the B? state and then cross over into the A state and hence
dissociate according to equation (4-12).
d. Sulfur Dioxide
The energy required for dissociation into SO and 0 is aboi i ) Kcal/
0
mole corresponding to a wavelength of about 2180A. This energy is not
available in urban atmospheres. Sulfur dioxide may be excited by sunlight ii.
o
trie region of 3200A and lower wavelengths (Figure 4-7).
S02 + hv -> SO * (4-1'J)
Q _ O
The first excited state may be a triplet with a lifetime of i.3 x LO to
2.2 x 10~ seconds. It is estimated that the natural radiative lifetime
would be about 2 x 10 seconds if the molecule were in a singlet state as a
o o
result of irradiation in the region of 3000A to 3200A.
The natural lifetime, or mean radiative lifetime, of an electronically
excited species is the average time zn seconds that the excited species retains
its excess energy in the absence of collisional deactivation. The mean
lifetime is equal to the inverse of the rate constant for the process by
which the molecule loses its energy by radiation (i.e., fluorescence,
phosphorescence, etc.). When a molecule has a relatively iong lifetime, say
4-27
-------
> 10 seconds, it Is an indication that it is in an electronic state from
which it is forbidden, spectroscopically, to fall into the ground state b>
radiation. The triplet state meets these conditions, whereas molecules in
the singlet state may radiate back to ground state readilv and have lifetimes
shorter than 10" seconds. The excited SO * may be able to react directly
*•
9
with components of smog such as hydrocarbons.
e. Nitric Acid
Although Reynolds and Taylor observed the format >>n of NO-
corresponding to the reaction in sunlight, the absorption spectrum 01 HNO~
0
indicates very weak absorption in the region above 3000A.
HN03 + hv •*• OH + N02<-53 Real) <4-14)
However, Jaffe and Ford have shown that HNO. will yield N02 in the presence
of NO- and light. Since a trace of NO,, will stimulate the reaction, equation
(4-14) probably results from
0 + HN0 + OH + N0 (4-l'>>
with subsequent formation of NO-,. The 0 atoms derive from the photolysis of
N02 as in equation (4-12).
f . Alkyl Nitrates
o
The photolysis of ethyl nitrate at 3130A serves as an example of the
alkyl nitrate primary process. The most probable photodissociatlon mode is
+ hv •*• RCH20 + N02
4-28
-------
g. Alkyl Nitrites
v The most generally ac :epted process for RONO is
RONO + hv -v RO 4- NO (4-17)
* Subsequent reactions of the RO radical are probably important in these systems.
Some results with ter_t-butyl nitrite and tert-pentyl nitrite have given
quantum yields, RCHO +NOH , (4-19)
h. Nitroalkanes
The only well-established primary process is the formation of the
alkyl radical and N02 according to the reaction:
RN02 + hv -»• R + N02 (4-20)
For nitromethane » » » j^ was shown that
CH3N02 + hv -* CH3 + N02 (4-21)
was more important than
CH NO + hv -*- CH20 + NOH (4-22)
*22 » 0.04
For nitroethane, in addition to process (4-20), a second reaction,
CH CH2NO + hv -»• CH2=CH2 + HONO (4-23)
O
was important. (^ -0.08 at 3130A.) These data Indicate that nitro compounds
4-29
-------
first formed in thermal reactions could photolyze further and be destroyed so
that no appreciable concentration is found in polluted air.
i. Aldehydes
Photodissociation of aldehydes is only possible energetically at
o
wavelengths of about 3400A and shorter. The principal dissociation reaction is
RCHO + hv -»• R + HCO (4-24)
At shorter wavelengths,
RCHO + hv •* RH + CO (4-25)
o
is possible. However, at 3130A, 25/4>24 * 0.001 -0.05, indicating that the
only reaction for consideration here is (4-24). These conclusions have been
verified for simple aliphatic aldehydes such as formaldehyde, acetaldehyde,
2
propionaldehyde, etc.
j. Kg cones
When alkyl ketones are photolyzed,
RCOR' + hv ->• RCO + R' (4-26)
-» RR' -I- CO (4-27»
o
at 3130A, equation (4-26) is favored, process (4-27) increasing at lower
wavelengths. Therefore, (4-26) should be the only important photodissociation
mode in urban atmospheres.
k. Peroxides
o o
Alkyl peroxides absorb weakly in the range of 3000A to 7000A. The
2
process most consistent with observation Is
ROOR' + hv -> RO* + R'O* (4-28)
o
The radicals are quite energetic even at 3130A, where they carry off about
4-30
-------
56 Kcal/mole. Subsequent fragmentation of the radicals is to be expected.
1. Acyl Nitrites, Nitrates, and Peroxynitrates
There is no experimental evidence of the primary photochemical
processes that may take place with acyl nitrites, R(CO)ONO, acyl nitrates,
R(CO)ONO- or peroxyacylnitrates, R(CO)OON02. Leighton1 has speculated on
possible photochemical reactions and the reader is referred to his book for
further possible applications of these compounds to smog.
m. Particulate Matter
Particulate matter such as metal oxides, i.e., ZnO, can absorb light and
give rise to photosensitized surface reactions. For example, with zinc oxide-
water-sodium formate systems, a reaction produces H~05 with a quantum yield
16 *
of 0.5 at 3130A. Since the urban atmospheres contain particles of metallic
oxides such as PbO and Pb_0,, particulate photosensitized reactions might be
of some importance.
For a more complete discussion of photochemical primary processes the
2
reader may refer to Calvert and Pitts' excellent text "Photochemistry," from
which most of the information cited above was obtained.
C. ATMOSPHERIC REACTIONS
!• Kinetics of Thermal and Free Radical Reactions
Reactions that involve products of the primary photochemical processes
(described in paragraph B above) are called secondary reactions. These include
the interactions of the products of photodissociation with each other and with
any other species in polluted air. The reactions that result from the collisions
4-31
-------
of various chemical species (not including light) are referred to as thermal
reactions. These include the interact .ons of normal molecules as well as the
interactions of free radicals with molecules or with themselves.
In some circumstances, small numbers of free radicals can induce a
very large number of reactions; they are therefore important in atmospheric
photochemistry. Free radicals in polluted air arise mainly from the dissociation
of excited molecules that have absorbed solar radiation (see paragraph B above)
or from reactions of such radicals with ordinary molecules. Free atoms and
free radicals are very reactive species because they have unsatisfied bonding
orbitals. That is, they have unpaired electrons that normally would interact
with other unpaired electrons and nuclei to form strong chemical bonds. Upon
collision with molecules and other atoms or free radicals, new stable molecules
and new free radicals can be formed. Thus, when free radicals are formed
in polluted air, they may rapidly initiate a chain of reactions that subsequently
lead to the formation of the many species found after irradiation.
A brief introduction to the principles of chemical kinetics is
presented in the following paragraphs in order to facilitate the understanding
of air pollution systems. The presentation consists of definitions and empirical
relationships and some theories that help to explain the observed phenomena.
The concept of rates of reactions and rate constants is especially helpful
in judging the relative importance of competing processes in photochemical
smog.
a. Rate Laws
It has been found experimentally that the rates of simple chemical
4-32
-------
reactions are a function of the concentrations of the reacting species,
Rate - k(A)n(B)m (4-29)
in which the rate may represent the amount of reactant being lost per unit
time per unit volume or the amount of products being formed per unit time
per unit volume. The units of rate may, for example, be moles per liter per
minute. Jk is called the specific rate constant and is the rate at unit
concentration of reactants. Some useful units and conversion factors for con-
centrations and rate constants are presented in Table 4-6.
(A), and (B) , etc., are the concentrations of the reactants; the
exponents, n and m, are small integers—0, 1 or 2, The over-all order of the
reaction is equal to the sum of n, m, etc. Most elementary reactions are of
order 2, although orders 1 and 3 are possible.
A complex chemical reaction takes place in a series of a simple
elementary steps. Although these steps may not always be known, the investigator
may postulate a reasonable set of elementary reactions in order to explain the
appearance of products. Such a set of reactions is called a mechanism. From
this set of simple reactions, one may derive a rate law and compare it with
experimental observations. If reasonable agreement is shown, the postulated
mechanism is a possible one. A test of this kind may be made on the postulated
reactions that take place in polluted air. For each elementary reaction, one
can write a rate expression. For example, for the process,
A + B -> C (4-30)
C + D -> E (4-31)
4-33
-------
Table 4-6. CONVERSION FACTORS.1
("Based on concentration in ppm and pphm relative
to air at 1 atm, 25°C.)
Concentrations
-1 9
moles 1 x 2.445 x 10 * pphm
-1 12
moles cc x 2.445 x 10 « pphm
pphm x 4.09 x 10 • moles 1
pphm x 4.09 x 10 - moles cc
Rate constants, k
Bimolecular
1 mole sec x 2.45 x 10 • ppm min
cc mole sec x 2.45 x 10 - ppm min
1 mole sec x 1.47 x 10 - pphm r»r
cc mole sec x 1.47 x 10 • pphm nr
Termolecular
2 -2 -1 -13 -2 -1
1 Mole sec x 1.005 x 10 - PP* min
2 -2 -1 -19 -2 -1
cc mole sec x 1.005 x 10 - ppm min
2 -2 -1 16 -2 -1
1 mole sec x 6.03 x 10 - pphm hr
2 _2 -i _22 -2 -1
cc mole sec x 6.03 x 10 " pphm hr
4-34
-------
the rate of formation of E is
d(E) = k (C) (D) (4-32)
___ 31
But C may be a reactive intermediate such as a free radical, whose rate of
formation may be equalled by its rate of reaction. That is, as soon as a C
radical is formed, it reacts with D to form E. In this case,
d(C) = 0 = k XA) (B) - k (C) (D) (4- 13)
dt JU Ji
from which, under so-called "steady-state" conditions,
Steady-State = k30(A) /k3l(D) (4'3A)
insertion of this result in (4-32) gives
7^ = VD) k3Q(A) (B) - k ,n(A) (B) (4-35)
k31(D) JU
The derived rate law will verify that the mechanism (4-30) and
(4-31) is appropriate if experiment shows that the rate of formation of E is
directly proportional to the product of (A) times (B).
b. Collision Theory
The most frequent chemical reactions in the gas phase involve
bimolecular interactions. In the simplified theory, the interaction is viewed
as one of the collision between two hard spheres. When two spheres collide,
they come together at a distance between centers called the collision
diameter
-------
£
Figure 4-4. MOLECULES A AND 8 IN COLLISION.
4-36
-------
result in reaction; this proportion is known as the collision yield and is
exponentially related to the temperature as expressed in equation (4-36)
—E/RT
y - pe ' (4-36)
where E is the activation energy of the reaction, II is the universal gas con-
stant, T is the absolute temperature and p_ is a probability factor which is
also characteristic of the reaction.
According to simple kinetic theory of gases, appropriate collisions
occur at a rate proportional to the concentrations of the colliding molecules;
thus
rc " ZAB(A)(B) (4~37)
when.' r is the rate of collision (in a unit volume of the gas phase) and Z
(~ A.JJ
is a frequency factor which can be estimated from the dimensions of the
colliding molecules. The product of this collision rate by the collision
yield y is the rate of reaction of A^ with IJ,
Comparison with equation (4-29) shows that, if m and n are both
unity, the rate constant k can be identified as the product of the collision
yield, y_, by the frequency factor, Z._. Thus
AD
Values of Z are normally of the order of magnitude of 3x10
Ajj
liter/mole sec., equivalent to about 4x10 ppm hr . Collision yields depend
strongly on the energy of activation, but cannot be larger than unity. Values
of k_, for all practical purposes, are thus constrained to be smaller than
3x10 liter/mole sec. Furthermore, the value of k is considered to be less
4-37
-------
11 —E/RT
than 3x10 times the exponential factor, e ; thus, If a lower limit for
£ can be reliably estimated, an upper Limit less than 3x10 liter/mole sec.
can be assigned for k_. This procedure can be very useful in estimating rates
of elementary steps for use in postulated mechanisms of complex reactions, as
shown in Chapter 6 in connection with the photochemical smog reaction system.
In summary, a chemical reaction results when two molecules collide
with sufficient energy and in the proper orientation to produce the activated
complex. The collision theory offers, for any specified elementary reaction,
a means of estimating an upper bound to its specific rate constant. As a
general explanation of chemical rate phenomena, the collision theory has
several weaknesses. First, the factor p_ may not be readily correlated with the
structures and properties of the reacting molecules. Secondly, abnormally high
rates cannot be interpreted on this basis. These faults can be overcome to a
large extent by the application of absolute rate theory. (For a brief
introduction to this more elaborate theory, refer to Calvert and Pitts (2,
p. 608ff).
2. Thermal Reactions
As a result of the photodissociation processes, atoms, radicals and
excited molecules are available to react thermally (that is, by collisions) with
otherwise stable species. The list of these compounds includes nitrogen,
water, carbon dioxide, carbon monoxide, nitric oxide, sulphur trioxide, sulphuric
acid, alcohols, organic acids, and hydrocarbons, as well as compounds th f do
absorb sunlight (Table 4-4). All such interactions are classified as secondary
reactions.
4-38
-------
The most important secondary reactions include the conversion of
NO to N0?, the formation of ozone, and the oxidation of hydrocarbons. An
extensive review of these reactions is presented in paragraph D. below. It is
very useful to compare the rates of these reactions in order to ascertain
which ones are more likely to predominate. If the rate constant is large and
the concentrations of reactants are high, reaction proceeds rapidly. If the
rate constants are too small or the concentrations of the reacting species are
too low, reaction takes place slowly and is not important.
By way of illustration, one could compare the rates of the very
important processes of olefin oxidation. The principal reactions are the
oxidation by 0 atoms,
0 + ^OC ->• Products
/ \
and the oxidation by ozone
0- + C=C •* Products
The range of specific rate constants for the oxygen atom reactions
is about 10 to 10 liter mole sec , while those of ozone are from 10 to 10
liter mole sec . These values apparently favor the oxygen atom reactions,
since the ratio
= 105 to 10?
However, the actual rates depend upon concentrations as well as rate constants,
and the steady-state concentration of oxygen atoms relative to ozone is
frequently about 10 to 10 . Using the rate laws,
4-39
-------
Rate Q m kglOHOlefin]
Rate Q kQ [03][01efin]
Therefore, Raten
U - 10
Rateo3
Under those conditions, oxygen atom oxidation of olefins might be only
10 times as fast as ozone reactions with the same olefins. Thus bot>> oxvgen
atom and ozone oxidation could occur at appreciable rates, neither being
negligible relative to the other. Such considerations are especially useful
in comparing rates of competitive reactions, which are not uncommon in the
development of photochemical air pollution.
3- The Generation of Free Radicals
Free radicals may be produced in photochemical smog in at least two
important ways. In the first, a neutral molecule absorbs light and is promoted
to an excited state. The excess energy in the molecule is eventually trans-
ferred into vibration of chemical bonds, and the molecule is split into two
or more fragments which may be free radicals. Examples of free radicals
produced in this manner by reactions discussed in paragraph B, above, are given
in Table 4-7.
The second process of free radical formation involves the bimolecular
reactions of molecules and radicals which produce excited intermediates which
dissociate, producing new radicals. The reactions of 0 atoms, 0» and
NO with hydrocarbons, especially olefins> give rise to this type of behavior.
4-40
-------
Table 4-7. SOME FREE RADICALS AND
ATOMS GENERATED BY PHOTOLYSIS
Absorber Molecules
°3
N02
HN03
RON02
RONO
RN02
RCHO
RCOR
Active Species Produced*
0
0
OH
RO
RO
R
R + HCO
RCO + R
oxygen atom
oxygen atom
hydroxyl
alkoxyl
alkoxyl
alkyl
alkyl + formyl
acyl + alkyl
ROOR
RO
alkoxyl
* R is any alkyl group.
4-41
-------
The thermal dissociation of molecules is also possible. This process
takes place when heat in excess of the bond energy is absorbed by a molecule
as a result of energetic collisions. However, at ambient temperatures only
a negligible fraction of molecules has sufficient energy to cause thermal
dissociation.
Free radicals are also prouuced by the reaction of other free
radicals with reactant molecules. Such reactions are discussed in paragraph 5
below, following the general introduction to chain reactions in paragraph 4.
Table 4-8 summarizes the kinds of free radicals that are probably
formed by the action of sunlight on polluted air. It should be noted that
NO, N07, NO. and even 0. possess unpaired electrons and behave as free
radicals in many instances. The general reactions of each of these are reviewed
in paragraph D.
4. General Treatment of Radical Chain Reactions
Free radical chain reactions are initiated by molecular collisions,
photolysis or any other method that can generate free radicals. The first
step can be represented by *
A •* mR + nB (4-40)
In one type of reaction chain, when the free radical chain carrier, R, collides
with another molecule, it can continue the process by a propagation step such as
R + C ->• R + D (4-41)
Here the radical R is reproduced so that it is free to repeat reaction (4-41)
many times before the chain process is broken or terminated.
*A, B, D, C, and E are reactant or product molecules, and m and n are integers.
4-42
-------
Table 4-8. FREE RADICAL CLASSES FORMED
IN IRRADIATED POLLUTED AIR.
Name of Radical
alkyl
alkoxyl
peroxyalkyl
hydroxyl
hydroperoxyl
formyl
acyl
peroxyformyl
peroxyacyl
formate
acylate
Formula*
R-
RO-
ROO-
OH
HOO-
HCO
RCO
*°
HC-00-
^0
RC-00-
0
ii
HCO-
0
ii
RCO'
*R is any alkyl group.
4-43
-------
R + R + M-»-R2+M (4-42)
In reaction (4-42), two free radicals recombine to form a neutral molecule
which cannot react further, thus breaking the chain.
An example of this kind of single- chain carrier process with second order
breaking is the ortho-para hydrogen conversion,
initiation M + H.(o or p) -»• 2H + M (4-43)
propagation H + H,,(p) -*• H2(o) + H (4-44)
(4-45)
termination M + 2H •*• H_(o or p) + M
A more important case occurs with two chain carriers and second-order
breaking,
initiation A -+-R + S (4-46)
C (4_47)
L R + B > S +
<
( S + D -». R +
propagation
E (4-48)
termination 2R + M -»• RZ + M (4-49)
Here R and S are two different free radical chain carriers.
A famous example of this kind of behavior is the formation of HBr from H. + Br-
initiation Br2 •* 2Br (4-50)
Br * H2 -> HBr + H (4-51)
propagation H + Br2 •* HBr + Br (4-52)
H + HBr -»• H2 + Br (4-53)
4-44
-------
termination 2Br •> Br2 (4-54)
Chain reactions of the kind that may occur in photochemical smog and involving
free radicals like those in Table 4-8, are postulated in what is called the
18
"Rice-Herzfeld mechanism." The thelnnal or photochemical dissociation of
acetaldehyde can be treated as follows,
initiation CH CHO -v CH + CHO (4-55)
CHO -»• H + CO (4-56)
H + CH3CHO -»• H? + CH^CO (4-57)
propagation
f CH3CHO -»• CH4 + CH3CO (4_58)
CH CO -»• CH + CO (4-59)
termination 2CH -»• C H, (4-60)
Using a steady-state approximation* on all free radicals, this mechanism
leads to the rate expression
\R /k59 \ •^ Ml
Rate- 58{^r-} [CH3CHO]J/^ (4.61)
5. Free Radical Reaction Types in Photochemical Smog
The free radical types shown in Table 4-8 react with virtually all of the
species in polluted air. The important reactions include those with 0? , NO,NO_,
0 and reactions among the free radicals themselves. A detailed description of
*The steady-state approximation is based on the assumption that the free
radicals are produced and lost at the same rates so that the concentrations
remain constant as long as reactants are available. This permits one to set
d[R] « 0 and solve for a steady-state concentration as was done in equation (4-33).
d[t]
4-45
-------
these reactions is given in paragraph D. A few general features of these
reactions, however, are of interest here.
Oxygen molecules are the most abundant reactive species in air; however, most
of the reactions of CL are relatively slow. Oxygen molecules can add to free
radicals producing more complex radicals:
R. + 02 •+ R02-
They can enter into metathetical reactions:
CH3 + 02 •*• CH20 + HO
Nitric oxide may add to free radicals quite readily, since it is itself an odd
electron molecule.
R- + NO -»• RNO
or 0
RCO -4- NO •* RCNO etc.
NO can be oxidized to NO.:
ROO + NO •+• RO + N02
Nitrogen dioxide may also add to free radicals, since it is an odd electron
molecule.
RO- + N02 •* RON02
It may also donate oxygen to some free radicals:
R«+ N02 -»• RO1 + NO
and NO. may abstract hydrogen from free radicals:
RCH26 + N02 •*• RCHO + HONO
4-46
-------
Ozone may react directly:
R-+ 03 -»• RO- + 0
In addition, the zwitterions formed as a result of the attack of 0» on oleflns,
may react further to yield free radicals, although this has not been experimen-
tally demonstrated.
Rates of these reactions depend on various factors determined in part
by the structures of the substituent radicals. Details of individual reactions
are discussed in paragraph D.
4-47
-------
D. A THESAURUS OF PROPOSED REACTIONS IN THE URBAN PHOTOCHEMICAL SYSTEM
Experimental and theoretical research has produced considerable insight
into the chemical reactions in photochemical smog. It is now widely accepted
that the process involves photo-oxidation of hydrocarbon vapors, with N0«
as the main effective primary absorber. However, detailed steps in the
photo-oxidation are still subjects of conjecture. This paragraph presents
a comprehensive collection of the chemical reactions which may be taking
place in smog formation. The reactions are grouped by appropriate categories,
e.g., inorganic reactions, free radical reactions, etc. A summary in each
category cites the most probable reactions occurring in that category.
Photochemical smog reactions have been studied by a large number
of investigators. Although references to original work are included where
appropriate, no attempt has been made to cite all the references to studies
of the chemistry of smog. In addition to this paragraph, the reader is
19 20
referred to the reviews of Leighton, Wayne, Altshuller and
21 22 23
Bufalini, Stephens, and Haagen-Smit and Wayne.
1. Inorganic Reactions
A number of reactions can occur when air containing oxides of
nitrogen is irradiated with sunlight in the absence of organic gases. The
reactions which can occur are summarized in Table 4-9 and 4-10.
4-48
-------
Table 4-9. INORGANIC REACTIONS.
N02 + hv « NO + 0 (Al)
0 + 02+M=03+M (A2)
03 + NO - 02 + NO., (A3)
0 + N02 + M = N03 -f M (A4)
0 + N02 - 02 + NO (A5)
0 + 03 = 202 (A6)
0 + NO + M = N02 + M ' (A7)
0 + 0 + M -' 02 -f- M» (A8)
N03 + NO - 2N02 , (A9)
N03 + N02 - N205 (A10)
N205 - N03 + N02 (All;
2NO + 02 = 2N02 (A12)
N03 + N02 - NO + 02 + N02 (A13)
N°2 + °3 " N°3 + °2 ^A14)
03 + hv - 02 + 0 (A15)
2N03 - 2N02 + 02 (A16)
NO -f N02 -I- H20 - 2HN02 (A17)
N205 + H20 i. 2HN03 (A18)
(A19)
2HNO - N20 + H20 (A20)
N03 + H20 - HN03 + OH (A21)
H20 + 0 = 20H (A22)
02+M = 0 + 0+M (A23)
03+M = 0 + 02+M (A24)
02 + 02 - 0 + 03 (A25)
4-49
-------
Table 4-10. RATE CONSTANTS FOR THERMAL
HOMOGENEOUS INORGANIC REACTIONS.
Reaction
Number
A2*
A3
A4
A5
A6
A7
A8b
A9
A10
All
A12
A13
A14
A16
A17
A23b
A24a
A25
Rate
Constant
k2
k3
k4
S
k6
k7
k8
k9
kio
kll
k!2
k!3
k!4
k!6
k!7
k23
k24
R25
Value*
(i, mole, sec.)
1.68 X 107 exp (2.1/RT)
1.17 X 107 - 2.8 X 107 (300°K)^
10U (300° K)
3.3 X 109 (300°K)
1.205 X 1010 exp (-4.79/RT)
1.8 X 1010 (300°K) [or 7.5 X 108//
, j , mole sec ]
1.379 X 10 T" exp (-.34/RT)
5.6 X 109 (300°K)
1.8 X 109 (300°K)
0.24 (300°K)
1.5 X 104 (300°K)
2.5 X 105 (300°K)
2 X 104 - 4.3 X 104 (300°K)
1.2 X 104 (300°K)
4.3 X 107 (300°K)
2.75 X 1016 i"1 exp (-118.7/RT)
9.94 X 1011 exp (-22.72/RT)
1.28 X 1010 exp (-100.6/RT)
Reference »
24
25
26
27
24
28
24
29
25
25
1
25
29,30,31
25
32
24
24
24
»
Activation energies are in Kcal mole x a. if M is 0_.
Range of values represents reported variations
D. if M IS
_.
4-50
-------
The main reactions which occur are (A1,A2,A3). The primary
19 —1
photolysis rate of (Al) was estimated by Leighton to be 0.4 min with
bright sunlight at a zenith angle of 20 degrees. Reactions (A2) and (A3) are
also very fast, resulting in a conversion of light to heat in the sequence,
and the establishment of a steady-state. .
Although. (Al to A3) are the most important inorganic reactions,
0
several other reactions may be of secondary importance. Reactions (A4 to A8)
involve reactions of oxygen atoms which must compete with (A2). Reaction (A4)
may be important in systems in which the concentration of N0_ is much higher
than in polluted air. Reaction (A7) may be important when NO is in great
excess. Reaction (A5) is too slow to compete with (A4), Reactions (A6) and
(A8) are too slow to be of any importance.
Reactions (A9) to (A14) involve other reactions of 0- and nitrogen
oxides. The most important processes consuming NO- are (A9) and (A10).
These two reactions are fast enough to prevent the accumulation of NO
from (A4). Of the two reactions (A9) is somewhat faster than (A1Q), pre-
venting the accumulation of N-O,. Reaction (All) is very slow and probably
not competitive with (A18) when water vapor is present. Reaction (A12)
accounts for the oxidation of NO to N0_ in the dark but is far too slow to be
of importance in sunlight. Reaction (A13) is negligible compared to (A10).
Reaction (A14) is slow relative to (A13) but is probably significant in
4-51
-------
systems with high concentrations of 0,. This reaction could be an important
source of NO,. Reaction (A15), the photolysis of 0,, is estimated to be
only about 1/20 - I/SO as fast as (Al) at similar concentrations. Reaction
(A16) is negligible compared to (A9) and (A10).
Reactions (A17) to (A21) could occur in the presence of water vapor.
19
Leighton has suggested that appreciable fractions of the oxides of
nitrogen could be hydrolyzed by (A17) and (A18) in a period of an hour or
less, which is sufficiently rapid to warrant consideration in atmospheric
systems. However, (A17) can be shown unimportant by equilibrium considerations.
The direct hydrolysis of N02 by (A19) is quite slow. Reaction (A20) was
33
postulated by Tuesday to account for small amounts of N-0 in the
photo-oxidation of trans-2-butene. The existence of the species HNO has not
•*
been confirmed, however, in photo-oxidation systems. Reaction (A21) is
endothermic and thus quite slow compared to (A9) and (A10). Reactions (A23),
(A24), and (A25) are unimportant.
The reactions can be separated according to their probable importance
in photochemical smog:
a. Primary Importance: (A1,A2,A3).
b. Secondary Importance: (A4,A5,A7,A9,A12,A14,A15,A18).
c. Negligible Importance: (A6,A8,A10,A11,A13,A16,A17,A19,A20,A21,
A22,A23,A24,A25).
4-52
-------
The reaction of 0 atoms with $()„ may play an important roio Ln tlir
production of SO. and subsequently H_SO, in smog. HoSO W1H form aerosols with
water. The reactions are
0 + SO •*• S03* (A26)
S03* + M -»• SO + M (A27)
(A28)
34 35
Reaction (A26) was studied by Jaffe and Klein and by Kaufman. It
10 2 —2 —1
yielded a rate constant equal to 1.4 x 10 liter mole sec for the over-all
reac L ion
S02 + 0 + M -»• S03 + M (A29)
2- Reactions of Atomic Oxygen and Hydrocarbons
Compounds containing carbon-hydrogen bonds are particularly susceptible to
attack by free oxygen atoms. Among the most reactive hydrocarbons are the
olefins and other compounds containing olefinic double bonds. The elementary
step is usually conceived as the addition of the oxygenation to the double
bond to form an activated complex which may decompose in various ways. The
21
general olefin-atomic oxygen reactions can be written as
K I\_ "T U "•> ..
3 \ /3 1 , ./ 3 081)
0(JP) + >C/ + V
4-53
-------
•v (B2)
A R *
'R3
0(*D) + * ^C=-C^ -^ ^ \ '
\
The asterisk indicates that the complex possessed excess energy and will
rapidly participate in further reactions or decomposition.
The reactions of oxygen atoms with paraffins are slower than with
olefins. The most likely elementary step seems to be
(B4>
RH + 0 -*• R- + OH
with a race constant of 10 to 10 1 mole sec .
Rate constants for the reactions of atomic oxygen and hydrocarbons
21 36 37 38
have been compiled by Altshuller and Bufalini and Cvetanovic. ' '
Rate constants (1 mole sec ) are shown in Table 4-11.
The probable reaction of aldehydes and atomic oxygen is
RCHO + 0 -»• RCO + OH (B5)
8 —1 —1
with a rate constant the order of 10 1 mole sec
Products from the decomposition of the activated complexes in (Bl)
to (B3) differ for the various olefins, but comprise principally the correspond-
A
inf. olefin oxides and aldehydes derived from them by slight rearrangement
The following product distribution has been observed from the reaction of
4-54
-------
Table 4-11. RATE CONSTANTS OF ATOMIC OXYGEN-HYDROCARBON REACTIONS.
r Hydrocarbon Rate Constant Reference
* (I, mole, sec)
Ethylene 3.1 x 108 39,40
9
I'ropylene 1.7 x 10 *
1- Butene 1.7 x 109 *
9
Isobutene 7.5 x 10 *
Cis-2-butene 7.1 x 109 *
9
Trans-2-butene 8.5 x 10 *
9
n-1-pentene 3.1 x 10 *
9
cis-2-pentene 6.8 x 10 *
1-Hexene 1.95 x 109 *
Cyclopentene 1.46 x 10 *
Cyclohexene 1.32 x 1Q10 36,37,38
2-Methyl-2butene 3.88 x 1Q10 36,37,38
2,3-Dimethyl-2-butene 5.12 x 1Q10 36,37,38
1,3-Butadiene 1.19 x 1Q10 36,37,38
Benzene 1.98 x 107 36,37,38
Toluene 6.93 x 10? 36,37,38
1,3,5 Trimethylbenzene 3.3 x 1Q8 21
9
Styrene 5.0 x 10 21
3-Methylheptane 6.5 x 10 28
* Calculated from the relative rates of 0 + X vs. 0 + C.H -iso, X=NO =0.44
91 9 4 8 -1 -5
from reference, i where k(0 + NO ) = 3.3 X 10 /liter/mole sec .
4-55
-------
19
from the reaction of ethylene and atomic oxygen:
.03 polymer
n*i P u A
• UJ V^_H,U
.05 H2 + CH2 + CO
CH2CO
.04 H2
.10 H + C2H30
.75 CH« + HCO
The activated complexes in (Bl) and (B2) formed in the propylene-atomic
19
oxygen reaction can participate in the following further reaction steps;'
CH.CHLCHO (propionaldehyde)
CH3-C-CH2
(M)
decomposition
CH-9-CH * (M)
H ——.
0
I \
H
(propylene oxide)
decomposition
0
I .
CH-C-CH2
CH. C-CH *
3 u 3
(M)
(CH )2CO (acetone)
0
/\
CH3-C — CH2* CM) CH3-C-CH2 (propylene oxide)
H " ^ "
N. 0
X . II
decomposition CH, + CH,C-
4-56
-------
When molecular oxygen is present the propylene oxide will react with 0-,
most likely according to Q
PU — p—.PH -*• n -*• PU Pun 4. i
\>nA v* v«*i«^ ~ o v/ti^v^nw i r
J H J Z
The reaction of 1-butene with atomic oxygen has been found to yield 57 percent
a-butene oxide and 43 percent butyraldehyde. ' ' The mechanism accounting for
these products is similar to that for propylene: n
C H-C-CH
H
decomposition
C3H7CHO*. (M") ^ C3H?CHO (butyraldehyde)
0 0
/ \ / \
2H5CHCH2* (M) i C2H5CHCH2 (a-butene oxide)
0
0
i
C3H_CHO*
> C H?CHO (butyraldehyde)
Cis-2-butene and atomic oxygen yield 25 each cis-and trans-3-butene oxide,
22 percent isobutyraldehyde and 28 percent methyl ethyl ketone (2-butanone) .
The probable mechanism for this reaction is again similar to those
above and is:
4-57
-------
CH3-H-
decomposition
HC-CH(CH3)2*—*(CH3)2CHCHO (Isobutyraldehyde)
—* decomposition .
H H * M H H (cis/and trans-
6 - butene oxide)
v A XH * M H\ A/H
V-C — V C^-c'
— •> decomposition
The reactions for trans-2-butene are similar to those for cis-2-butene. Iso
butene-atomic oxygen products include methyl ethyl betone, isobutene oxide and
isobutyr aldehyde .
Reactions of hydrocarbons and atomic oxygen can now be summarized. Only a
minor fraction of the ethylene consumed when reacting with atomic oxygen
yielded products with two carbon atoms. Fragmentation into methyl and formyl
radicals is apparently the principal process. In the propylene-atomic oxygen
reaction the main products are propylene oxide, propionaldehyde, acetone and
small amounts of carbon monoxide. The main fragmentation process is probably
to methyl and acyl radicals. The butenes yield butene oxides, butyraldehyde and
methyl ethyl ketone. The main fragmentation process is expected to yiexa either
propionyl and formyl radicals or ethyl and acyl radicals. In the presence of
0_ the products are typical of those in 0, attack.
4-58
-------
30
Recently, Klein and Sheer have proposed a new transition state
lor the 2-methyl-2-pentene reaction with 0( P). They suggest a concerted
rearrangement of the transition state rather than the formation of an actual
biradical to obtain the products of the reaction.
j. Reactions of Ozone and Hydrocarbons
0- is a strong oxidizing agent and will react with hydrocarbons and
other organics present in urban air. Olefins are readily attacked by 0_
in the vapor phase, the initial step of which is usually assumed to be addition
of 0- to the olefin molecule to form an energetic complex. The energetic
complex can decompose or switch to its isomer through loss of some of the
excess energy.
Rate constants for 0.,-hydrocarbon reactions have been tabulated
21
by Altshuller and Bufalini and are shown in Table 4-12 Each column
represents the results of particular investigators.
The rate constants for the 1-olefins with the exception of ethylene fall
20
closely together, while those for the 2-olefins may be considerably higher.
4-59
-------
Table 4-12. RATE CONSTANTS FOR OZONE-HYDROCARBON REACTIONS (i mole~1sec"1)xl03.
Hydrocarbon References
42 43 44 45 21
Ethylene 1.8 1.8 0.8 1.6
Tropylene 5.1 3.8 4.9
1-Butene 3.9 3.2
n-1-Pentene 3.9 3.2 4.5
u>
n-1-Hexene 4.6 6.1 6.1 5.5 6.8
1-Octene 4.9
1-Decene 6.6
3-Methyl-l-butene 3.0
isobutene 3.6 3.7 14
2-Methyl-l-butene 4.0
Cls-2-butene 13 29 200
Trans-2-butene 17 260
*
Trans-2-pentene 10 . 98
Trans-2-pentene 13
3-Heptene (cts-trana) 53
Cyclohexene 14 35
2-Methyl-2-butene 12 450
2,3-Dimethyl-2-butene 14
1,3-Butadiene 4.2 4.9
Acetylene 0.04
Xylene <0.01
Mesitylene <0.1
Styrene lb
Arrolcin 0.45
Crotonaldehyde . "0.45
Ketene <0.04
4-60
-------
The products of the 0,,-olefin reactions vary depending on the particular
olefin. Prominent products include aldehydes, ketones and carboxylic acids. The
most widely accepted reaction mechanism hypothesis is that the energetic intermediate
formed from initial attack of 0, on the olefin decomposes to produce a
carbonyl compound and an unstable intermediate compound, sometimes called a
zwitter ion. Thus, the initial reaction is
The zwitter ion is unstable and decomposes in various ways, among
which may be:
V-
;coo - R R + co (C2)
D /
R2
R. + -
v COO = ROH + CO (C3)
H
R ' II
xr ~ - ROCH or HOCR (C4)
H
X COO = RO- + CHO (C5)
H
4-61
-------
CH3
\+ -
COO - CH2-OO -f ELO (C6;
/
H
CH3CH2
NCOO - C2H4 + HCOOH (C7)
H ' °r
HN+ -
H
In addition to decomposition, the zwitter ion can participate in
reactions with other species. Possibilities include:
Olefins ^ R.J R$ 0 R3 6 RS
^ COO + ^C~C^ - RjCR. + ^C-C^ • (C8)
R2X RAX R6 4 6
Oxygen Rl + _
+ °2 ' R1CR2 + °3 (C9)
KO
Rl
^
00 - Rn. + C00 + HOO (CIO)
212
NO R, 0
COO + NO - R CR- + NO (Cll)
X 1 i /
R2
4-62
-------
R +_ R ,0S X3
Ketones R,CR, + J COO - C C^ (C12)
1 P ' R x R
R4 R2 0-0 K4
N00 R. 0
i 1^ + - IL
2 COO + N00 - R^R, + N0_ (C13)
^ i 1 i J
R2
In addition, two zwitter ions could react
x
2 COO - C Cv . (C14)
R R' 0-0 R
forming an alkylidene peroxide which, however, has not been observed. Another
possible reaction between two zwitter ions is
2R1R2CO + 02 (C15)
Mechanisms alternate to that of the zwitter ion have been proposed. A direct
dissociation of the original activated complex to two free radicals has been
suggested by Wayne. It is unlikely, however that a large proportion of
0 -olefin collisions would yield free radicals directly.
4. Reactions Between Free Radicals and Molecular Oxygen or Ozone.
There is substantial evidence for the existence of chain reactions
in the photochemical smog system. These chain reactions proceed with the aid
4-63
-------
of free radicals, the unpaired elections of which are preserved through the
various steps of the chain even though the species with the unpaired electron
may change from step to step.
Haagen-Smit, et al. concluded that 0, was formed in a chain
reaction during the photo-oxidation of 3-methylheptane. They suggested that
alkyl and peroxyalkyl radicals might be the chain-propagating species. Cadle
47
suggested that peroxyalkyl radical could be responsible for the conversion
of NO to NC>2 in photo-oxidation systems. Schuck, et al. compared the
observed rates of consumption of olefins in photo-oxidation systems with the
expected rates based on literature values for the reactions with 0, and 0.
They found the observed rates much faster than the expected rates. They
attributed the "excess rate" to reactions of the olefins with free radicals
33
in the system. Tuesday concluded that a free radical chain reaction
Initiated by reaction of oxygen atoms and olefin was important in the
photo-oxidation of trans-2-butene and NO.
Wayne has provided a clear explanation of the importance of
chain reactions in photochemical smog. Since oxygen atoms are known to react
very readily with olefins, it is feasible that the chain reactions be initiated
by this reaction. However, only a fraction of the oxygen atoms generated
can be assumed to react with olefins because of competition with the 0~ reaction.
Also, the olefin-oxygen atom reaction yields a variety of products, not all
of which are capable of propagating chains. For example, if only half of the
oxygen atoms generated react with olefins and only one olefin molecule in
five yields free radicals when it reacts with an oxygen atom, then a chain
4-64
-------
length of 60 would be necessary to account for an observed quantum yield of 6.
Thus, only small amounts of N0« are required to initiate the process. As the
small amount of NO- dissociates into NO and oxygen atoms, chains are initiated
and NO is produced from''NO in a self-accelerating process. This is precisely
the situation in early morning urban air, in which small quantities of N0« are
capable of initiating the daily smog cycle.
Reactions responsible for the generation of free radicals have been
outlined above, mainly from the hydrocarbon-oxygen atom and hydrocarbon-ozone
reactions. One of the most probable steps is the fragmentation of the
activated complexes in (Bl), (B2) and (B3) to yield two free radicals, e.g., the
production of methyl and formyl radicals from the ethylene-atoznic oxygen reaction.
Another example is the production of free radicals from atomic oxygen in (B4).
Another possible source of free radicals is the decomposition of the activated
complex from the olefin-ozone reaction as in (C5). In addition, reactions of the
zwitter ion might yield free radicals (C8), (CIO), as well as the photolysis of
various organic compounds as shown in Table 4-4.
In this and succeeding subparagraphs, possible reactions of free
radical species in polluted urban air are outlined. In each case a summary
lists the most important reactions of each type. In this subparagraph the
reactions of free radicals with molecular oxygen are outlined. For the
most part, these are chain-propagating reactions which conserve the number
of free radicals in the system. Reactions are listed by radical type.
4-65
-------
a. Alkyl Radicals R-
R- 4- 02 (+M) - R06 (4M) (D1)
e.g., CH3- + 02 (-W) - CH300 (4M)
b. Acyl Radicals RCO Q
H .
RCO + 0 «* RCOO (peroxyacyl) (02)
RCO + 02 - RO + C02
e.g., HCO + 02 - OH + C02
CH3CO
RCO -I- 02 - ROO + CO (D4)
4-66
-------
e.g., HCO + 0 = HOO + CO
^
RCO + 20. " R&6 4- 0, (D5)
'3
c. Peroxyalkyl Radicals ROO
ROO + 0 - RO + 0, (D6)
RCC
d. Peroxyacyl Radicals RCOO
0 0
ii . II.
RCOO + 02 = RCO +
e. Acylate Radicals RCO
0
RCO 4- 0 - RCO + 03 (D8)
0
RCO + 00 = ROO + C0_ (D9)
f. Alkoxyl Radicals RO-
RO- + 0,, =• R'CHO + HOO (D10)
RO' + 0 - R* + 03 (D11>
4-67
-------
g. Hydroxyl Radical HO
Does not react with 0. at ordinary temperatures.
h. Hydroperoxyl Radicals HOO
HOO- + 02 - 03 + HO- (D12)
1. Summary of Free Radical Reactions with 0.
Reaction (Dl) is well substantiated as the most probable fate of
alkyl radicals in an oxygen-containing environment. The rate constant is
8 -1 -1
about 10 liter mole sec . Of the various reactions of acyl radicals with 0_,
(D2) is probably most likely at ordinary temperatures. This reaction, the
direct combination of acyl radicals and 02 to form peroxyacyl radicals, is
supported by the occurrence of PAN-type compounds in organic photo-oxidation
49
systems. Taylor and Blacet suggested (D3) and (D4) aa possible
reactions to compete with (D2). Reaction (D5) was suggested by Schuck and
Doyle to account for 0. production.
Reaction (D6) has been often proposed to account for the
production of 0. in organic photo-oxidations. In the oxidation of methyl
radicals ac room temperature, (D6) has been suggested as the principal
route for destruction of peroxymethyl radicals. Since it is certainly endothermic
by at least 20 Kcal, the collision yield can hardly be greater than 10~ ,
unless the alkoxy radicals are available in a higher electronic state or a
highly excited vibrational state. Such radicals with sufficient energy may be
produced by reaction (Dl), but they are probably highly susceptible to
collisional deactivation. Reaction (D7) and (D8) analogous to (D5) have
been proposed to account for the production of 0_ in urban air. Their relative
4-68
-------
importance is difficult to assess but, like (D6), they must be quite
endothcrmic. Little evidence exists to support (D9), proposed by Stephens;
in contrast to (D8) it should be highly exothermic (ca. 50 Kcal) and thus
might have a high collision yield.
52
Reaction (D10) has been suggested by Hanst and Calvert to
explain the production of formaldehyde in the photolysis of dimethyl peroxide
49
in the presence of 0~. Taylor and Blacet proposed (D10) as a competing
reaction to (Dll) in the photo-oxidation of biacetyl. Schuck, et al.
also suggested (D10) could be responsible for acetaldehyde production.
Reaction (Dll), since it involves the rupture of a carbon-oxygen bond to form
an oxygen-oxygen bond, is even more endothermic than (D6) and is quite unlikely
to be of any significance in urban atmospheres.
Some actual examples and rate constants are given in Table 4-13.
The direct reaction of 0_ with radicals has been suggested by Calvert and
Hanst in the reaction
CH3 + 03 - CH36 + 02 (D13)
Q 11
The rate constant for (D13) is estimated to be 5 x 10 to 10 liter
mole sec . The rates of heavier alkyl radicals with ozone are not known,
but are probably of the same order of magnitude as that of methyl radicals.
A similar reaction would be expected with hydrogen atoms,
H- + 03 = OH + 0- (D14)
but no rate data are available.
5. Reactions Between Free Radicals and NO
Nitric oxide may add to free radicals quite readily, since it itself
4-69
-------
Table 4-13. SOME REACTIONS OF 0- WITH FREE RADICALS.
Reaction
Rate Constant Preexponen- Activation Collision* Refer-
at 3QO°K tial Factor Energy Yield ence
3x10
10
53
CH
+ HO ~6xlO
54
10
5500
55
10
8.8
53
-RO -f
10
8.5
3500
55
* k -F/RT
Collision Yield - - - pe '
4-70
IxlO"4 1
-------
is an odd electron molecule. The requirement is that the excited adduct be
stabilized by collisions, since the excited molecule formed would dissociate
due to the exothermicity of the reaction.
a. Alkyl Radicals
R' + NO =• RNO (El)
b. Acyl Radicals
RCO + NO - RCNO (E2)
c. Peroxyalkyl Radicals
ROO + NO « ROONO (peroxyalkyl nitrite) (E3)
ROO' + NO - RO- + NO, (E4)
d. Peroxyacyl Radicals
0 0
RCOO- + NO - RCOONO (peroxyacyl nitrite) (£5)
0 0
li H
RCOO-+ NO - RCO-+ N02 (E6)
e. Acylate Radicals
0 0
II- II
RCO + NO - RCONO (E7)
0
II.
RCO -I- NO - RCO + NO- (£8)
4-71
-------
f. Alkoxyl Radicals
RO- + NO - RONO (E9)
RO- + NO - R'CHO + HNO (E1°)
g. Hydroxyl Radicals
HO- -I- NO - HONO (Ell)
h. Hydroperoxyl Radicals
HOO 4- NO - OH + N02 (E12)
i. Summary of Free Radical Reactions with Nitric Oxide
(El) is insignificant when compared to the combination
of alkyl radicals and Q^. Also, (£2) is probably negligible
when compared to the corresponding reaction with 0?, (D2).
The peroxyalkyl nitrite product in (E3) has not been
observed with certainty. It may be a short-lived product
or a transition state for (E4). Such molecules probably
undergo rapid photolysis to the original reactants or to
the products of (E4), and are thus not important as
chain-terminating species. Reaction (E4) has been pro-
posed to explain the rapid conversion of NO to NO- in organic
19
photo-oxidation systems. Reaction (E4) is strongly
exothermic and would be a possible important reaction.
4-72
-------
The peroxyacyl nitrite product in (E5) has not been observed.
Reaction (£6), analogous to (E4), is strongly exothermic
and believed to occur. A plausible argument for the delayed
appearance of FAN in such systems is that (E6) successfully
competes with the PAN-forming reaction (F5) until all the NO
is consumed. The product in (E7) has not been observed.
Reaction (E8), similar to (E6), would be highly exothermic
and a possible route for conversion of NO to N0_.
Ri-action (E9) may well be occurring with a high collision yield.
However, the alkyl nitrite product is seldom observed because
of its high efficiency of photodissociation. Reaction (£10)
has been postulated to explain the decrease in oxidation rate in
the presence of excess NO. Little evidence of the importance
of (Ell), analogous to (E9), is available. Reaction (E12),
like (£4), is strongly exothermic and possibly important in
urban atmospheres.
6. Reactions Between Free Radicals and NO-
a. Alkyl Radicals
(Fl)
R- + N02 - RO + NO (F2)
4-73
-------
b. Acyl Radicals Q _
H II
RCO + N02 - RCONO or RCNC>2 (F3)
c. Peroxyalkyl Radicals
ROO + N02 - RO + N03 (F4)
d. Peroxyacyl Radicals
0 0
II . II
RCOO + NO- - RCOONO (peroxyacyl nitrate) (F5)
9.
- RCO + N03 (F6)
e. Acylate Radicals
0 0
II II
RCO + N02 » RCON02 (F7)
0
«.
RCO + N02 » RN02 + C02 (F8)
f. Alkoxyl Radicals
RO + N02 - RON02 (alkyl nitrate) (F9)
RO + N02 - R02 + NO (F10)
g. Hydroxyl Radicals
OH + N02 (+M) - HN03(-Ht) (Fll)
4-74
-------
h. Hydroperoxyl Radicals
HOO + N02 - N03 + OH (F12)
i. Summary of Free Radical Reactions with NO.
Reactions of (Fl) and (F2) probably cannot compete with
(Dl). Nevertheless, the nitroalkane product of (Fl) is
relatively stable to photolysis and, at low 0. concentrations,
could be a chain terminating step. Reaction (F3) is probably
negligible compared to (D2).
Reaction (F4) is considerably less exothermic than the
corresponding NO reactions, (E4) and (E6). It is probable
that this reaction will be slower than (E4) and (E6) by a
3 19
factor of 10 at equivalent concentrations. Similarly,
(F6) is not favored energetically by comparison to (E4)
and (E6). Reaction (F5) is probably quite important,
yielding peroxyacyl nitrates when the concentration ratio
(N02)/(NO) becomes large.
Reaction (F7) and (F8) are probably both unimportant
compared to the decomposition of acylate radicals. Reaction
(F9) may be an important chain-terminating step. It has been
generally postulated to account for the occurrence of alkyl
nitrates among the products of photo-oxidation systems. An
alternate reaction to (F9) is (F10), with a collision yield about
4-75
-------
twice as great.
Reaction (Fll) has been suggested as a chain-terminating step
for hydroxyl radicals, although nitric acid has not been commonly
recognized as a product in photo-oxidations. Reaction (F12) is
analogous to (F4) , but is energetically less favored.
Table 4-14 presents rate constants, activation energy and
collision yield data for some reactions of free radicals with
NO and N02 .
7. Reactions Between Free Radicals and Hydrocarbons
a. Alkyl Radicals
R- + XH - RH + X. (Hydrogen-Abstraction) (Gl)
R . -I- R2CHO - R H + R2CO (G2)
R- 4-
-------
Table 4-14. SOME REACTIONS OF NO AND N02 WITH FREE RADICALS.
Reaction
CH3 + NO— *-CH3NO
Rate Constant
at 300°K
6.3X10
8
Activation
Energy
Collision
Yield Reference
57
0
CH CO + NO—^CH CNO
~13.2 + 2.4
58
1.7X10
56
NO 3.3X10
56
4-77
-------
d. Peroxacyl Radicals
0
RCOO + Oi - (RCO) 0 (};0)OJl' (G17)
4-78
-------
OH + XH - H_0 + X' (G18)
OH + RCHO - H20 + RCO (G19)
h. Hydroperoxyl Radicals
HOO + QA - (HOO)OX,- (G20)
HOO + RCHO « RCO' + H_02 (G21)
i. Summary of Free Radical Reactions with Hydrocarbons
Of the three reactions (Gl), (G2), and (G3), (G3) is the most
favored, although none of the three can compete with (Dl) in
the urban atmosphere. Reaction (G4) is probably slow when
compared to (D2), but might account in part for the existence
of olefin polymers and co-polymers in liquid phase products
in the form of aerosol particles.
Reactions (G5) and (G6) are probably unimportant at room
temperature. Reaction (G7) has been proposed aa the gas
phase analog of the free radical reaction which initiates
polymerization in liquid-phase olefin reaction, and could be
part of a chain which yields aldehydes and ketones as
products.
4-79
-------
Reactions (G9), (Gil), (G14), and (G17) are similar
to (G4) and can account for the presence of higher molecular
weight oxidation products. However, (G9) and (G10)
probably cannot compete with (E6) and (F5).
Reactions (G15) and (G16) probably do not compete
successfully with (G14). Of the various hydrogen abstraction
reactions, (G8) is the least favored and (G19) the most
favored energetically. Reaction (G21) has been suggested
as part of a mechanism in which the hydroperoxyl radical
was a main chain-carrying species.
8. Reactions Between Free Radicals
a. Alkyl Radicals
R CH + RCHCH - R(CH)R (HI)
b. Acyl Radicals
See Reaction (Hll)
(H2)
c. Peroxyalkyl Radicals
2ROO - 2RO + 0 (H3)
4-80
-------
d. Peroxyacyl Radicals
0 00
ii n ii
2RCOO - RCOOCR + 0- (HA)
0 0
ii . . »
RCOO + ROO - RCOOR + 02 (H5)
0 0
ii . • ii
RCOO + RO - RCOR + 0 (H6)
e. Acylate Radicals
0 0
n . ii
RCO + RO + RCOOR (H7)
0
RCO + RO - RCOOH + R'CHO (H8)
f. Alkoxyl Radicals
2RO - ROOR (H9)
2RO - ROH + R'CHO (H10)
RO + RCO - ROR + CO (Hll)
RO + ROO - ROR + 02 (H12)
g. Hydroxyl Radicals
20H - H2 + 02 (H13)
h. Hydroperoxyl Radicals
2HOO - H00^ + 0, (HI 4)
4-81
-------
HOO + ROO - ROOH + 0 (1115)
HOO + ROO - RO + OH + 02 (HI6
0 0
« • n
HOO + RCO - RCOH + 0- (HI7)
HOO + RO - ROH + 0 (HIS)
i. Summary of Reactions Between Free Radicals
Reactions (HI) and (H2) are negligible compared to the
reaction with 0 , (Dl). Reaction (H3) is possibly
important but must compete with (D6), (E4), and (G7).
Reactions (HA), (H5) and (H6) are, for the most part, negligible
compared to the other reactions of peroxyacyl radicals.
Reactions (H7) and (H8) are probably unimportant, as
evidenced by the lack of esters in urban air. The
disproportionation reaction (H10) is much faster than the
combination reaction (H9). Reaction (H13) has been
suggested as a chain termination step in mechanisms involving
hydroxyl and hydroperoxyl radicals, but is probably un-
important in urban air. Of the reactions (H1A) to (HIS),
the reactions most likely r.o be of importance arc (H1A)
and (HIS).
A-82
-------
It should be noted that radical-radical reactions, although
in general very fast, are restricted by the low concentrations
of radicals, which lead to low collision frequencies.
Table 4-15 presents rate constants for some reactions
between free radicals.
9. Radical Decomposition Reactions
a. Alkyl Radicals
No decomposition reaction.
b. Acyl Radicals
RCO(+M) = R- + CO(4M) (11)
c. Peroxyalkyl Radicals
RCH200 - RCHO + OH (12)
RCH200 - RCO + H20 (13)
d. Peroxyacyl Radicals
0
ii
RCOO - RO + C00 (14)
4-83
-------
Table 4-15. SOME FREE REACTIONS WITH FREE RADICALS.
Reaction
Rate Constant
Reference
.-10.34, . . -1 -1
10 liter mole sec
59
.-10.2 9.4
10 or 10
55
10
9.5
55
Cll 0 + CH 0—»-CH OOH + CH20 10
9.2
55
CH 0 + OH "CH-OH +
55
-jO + R "-CH.OOH + olefin 10
9.5
55
CH 0 + CH
-CH OCH
10
9.8
55
+ CH,
-0 + CH.
I 4
10
10.0
55
-f C
10
9.8
55
2C1M)
10
8.8
55
4-84
-------
e. Acylate Radicals
0
RCO - R- + CO- (15)
f. Alkoxyl Radicals
RO - R'CHO + R"- (16)
g. Hydroxyl Radicals
No decomposition reaction.
h. Hydroperoxyl Radicals
No decomposition reaction.
i. Summary of Radical Decomposition Reactions
The activation energy for the decomposition of acyl radicals
by (II) is quite high. It is probably not fast enough to
exclude the competing reaction of acyl radicals with 0 ,
(D2). Reaction (II) might be favored by the larger acyl
radicals. The role of reactions (12) and (13) in urban
air is uncertain, as is the role of reaction (14).
There is a strong possibility that (15) is the major process
for acylate radicals in air, particularly if they are formed
with excess energy. The rate of reaction (16) may be
competitive with the rates of the reactions of alkoxyl radicals
with NO and NO (E10) and (F9).
4-85
-------
10. Summary of Important Reactions of Radicals
The most important reactions by radical type are:
1. Alkyl radicals react almost exclusively with 0 in (Dl).
2. Acyl radicals probably react exclusively with 0_ in (D2).
3. Peroxyalkyl radicals
a) (D6) may be important if the radical has access emergency.
b) Otherwise, (E4), (G7, G8, G9) and (F5) are likely to be
important.
A. Peroxyacyl radicals probably participate in (D7), (F4),
and (F5).
5. Acylate radicals probably undergo decomposition as in
(15).
6. Alkoxyl radicals
a) (D10) can occur from photolysis of the alkyl nitrite
from (E4).
b) (£10) might help to explain the oxidation decrease
with excess NO.
c) (F9) is probably occurring in most systems to some extent.
d) (G14) may help to explain excess rates.
e) (16) may be competitive with (E9), (E10), and (F9).
f) Little is known about u~9) to (H12).
7. Hydroxyl radical reactions are not well elucidated as
to relative importance.
8. Hydroperoxyl radicals probably participate in (Fll),
(G20), (H14), and (HIS).
4-86
-------
E. PHYSICAL AND CHEMICAL PROPERTIES OF THE PHOTOCHEMICAL REACTION PRODUCTS
The major products ' that result from the irradiation of NO,
NO.., and various olefins in air are ozone, peroxyacyl nitrates, aldehydes,
ketones and carbon monoxide. The minor products include epoxides, nitric acid,
alky! nitrates and nitrites, peroxy compounds, methanol and ketene. In the
presence of sulfur dioxide, sulfuric acid is also formed.
Some of the chemical and physical properties of these substances
are listed here to help in their detection and identification or analysis.
Some of these compounds can be recognized in the atmosphere by characteristic
odors and chemical effects even though they are present in very low concentrations.
A short description of each compound follows, and some relevant
properties are summarized in Table 4-16.
1. Ozone
Ozone is the most abundant oxidant in photochemical smog. It is
thermodynamically unstable and highly reactive. It is a strong oxidizing
agent, attacking rubber, textiles, plants and animal tissue. Ozone may be
recognized by its pungent odor which resembles hay at low concentrations (pphm
range) and the characteristic electrical arc welding odor at the ppm range.
2. Peroxyacyl Nitrates61' 62> 63> 64
The peroxyacyl nitrates are a series of thermodynamically unstable
oxidants presumably formed by the combination of NO with various peroxyacyl
radicals. The pure compounds decompose slowly on standing at room temperature,
or explosively, under some circumstances. Ultraviolet light also accelerates
decomposition. They are very reactive and cause plant damage and eye irritation.
4-87
-------
)F PHOTOCHEMICAL REACTION PRODUCTS.
<*_,»
CO
W
H
OH
O
oi
OH
H
3t
*— *
\O
1
^
^
ii«
vu
H
t~.
J3
•H
rH
»H
D
cr
W
U)
c
•H
rH
•H
O
PQ
OC
C
i_i
s
t-H
to
u
•rH
tn
.n
ClH
en
^
3
CJ
0)
O
z:
rH
13
U
•H
E
0)
u
Ul
n)
1
01 U
M o
OH m
0. 4-1
ctJ *0
>
•H
o oc
OH 01
Q
4-1 (J
(2
iH •
O OC
OH 01
Q
U
QJ O
4-> in
nj CM i
1 ^""'^
I/)
u) tn
I/) Ol
0) rH 13
rH U) -H .H
p cj O 3
o O rH cr
TJ O -H
0 0 rJ
rti
r* r— t
toO O
•H S
0) ""*•••
3 0(
On rH
^
O 0- S5
-Red - 9.3 21 >;i atm Yellow liquid as N^
No true B.P. or M.P.
less - 92 - 21 >1 atm. Polymerizes readily
to paraformaldehyde
C rH
3 0
o to rH en
rH « O (D
PQ O O O
\jO O
0 *^ rH
O tt!
S 0 g
^H "* O
z Q fa
less - 88 53 -300mm Hg Well known lacrimatoi
d
rH iH
0 3
rH CT
O -H
U rJ
V£)
in
O
33
O
X
o
II
CM
33
C
•H
Ol
rH
O
rH
O
<
00
»
m
vD
to
to
01
r^ *O
M -H
0 3
rH CT
O -H
C.) rJ
oo
m
PI
33
0=8
CO
33
U
01
c
0
4J
01
u
<
rH
EH"
•H
CO
o
01 01
rH 13
-------
^~\
T3
Ol
3
a
•H
4-)
a
o
u
v«x
03
f~4
U
*U
a
o
tf:
OH
S-;
o
hH
H
0
<
W
erf
j
5
o
h~4
•A
o
o
H
O
EC
rVl
fH
O
CO
M
t-H
H
Crf
W
A*
O
r"v'
P-4
V*
%
o
co
.
\o
rH
1
-t
OJ
rH
.£>
fd
H
cx
E
*iH
CO
U-l
o
01
cn 01 -a
-vi
l-i
CO
g
01
a;
s •
3 cn
•H cn
U 01
J3 M C_>
-H a, o
rH in
-H • CN
3 CX
cr cd
rH -rW
Cu X
E 0
CO V4
X 0>
M a
00
33
g
E!
00
J
f\fl
uu •
C 4J U
•H G
rH -H •
•H o oo
O (X. 01
M O
rH
,
CN
in
rH
GO
a iJ u
•H C
U -H •
f-H O 60
ai cw oi
S Q
0\
00
1
n) cj
U 01 0
•H 4-1 m
in co CM
>, 4J
nC CO 4J
(X, Cfl
)-<
CO 4J
rH JZ 01
3 00 rH
O -H O
01 o) B
rH » -C
£ B
£ 00
T3
•H
Ol 3
3 cr
rH -H
BQ J
<^
ro
rH
CO (0
O rH
•H 3
B B
01 U
J3 O
O PL,
CN
O
rH 4-)
CX Ol
X iw
u o
8
4-1
CO
oo
o
II
o
II
CN
X
u
o
•H cn
S O
co %
rH O
•H O
O 01
« a
o
I
o
n
c*1
r-
o>
1
U)
0)
a>
rH
M
O
rH
0
O
CNI
m
T3
•H
3
cr
•H
rJ
rH
m
rH
1
CO
in
01
rH
)-l
O
<-( cn
o rt
o o
CN
3u
01
G
Ol
4J
Ol
X
TD
•H
CJ
<
CJ
•H
l-l
4-1
•H
7.
o
•H
r4
3
<4-J T3
rH -H
3 CJ
co <;
4-89
-------
3. Oxides of Nitrogen, NO
Seven known oxides of nitrogen are listed in Table 2-18. Of these
NO compounds, only NO and NO. are abundant in the air and of major importance
X i-
in photochemical smog. N 0 may also play a minor role by its reaction with
water to form nitric acid (HNO ).
Nitric oxide is a colorless, odorless gas which is not a photochemical
reaction product nor an oxidant, but is a precursor in formation of nitrogen
dioxide.
Nitrogen dioxide is an oxidizing gas that is strongly toxic to
plant and animal tissues. It can be recognized by its brownish color and its
characteristic odor. Although N0? itself reacts slowly as an oxidant, its
photodissociation produces oxygen atoms which, along with ozone, produce most
of the photo-oxidation products found in smog.
4. Aldehydes and Ketones
Aldehydes and ketones are among the principal carbon-containing
products resulting from the oxidation of hydrocarbons by oxygen atoms and
ozone. They are dissociated by ultraviolet sunlight with the production of
tree radicals. Formaldehyde and acrolein, among the aldehydes, are known to
cause eye irritation in smog. They can be oxidized slowly, mainly to
organic acids; with or without oxidation, they show a strong tendency to
polymerize and to be absorbed on particular, matter.
4-90
-------
5. Carbon Monoxide
Carbon monoxide is a colorless, odorless gas that is extremely toxic
at high concentrations. It is a product of incomplete combustion in internal
combustion engines as well as a product of photochemical smog. It is
rather inert chemically, oxidizing to CO- very slowly upon reaction with oxygen
atoms or certain free radicals. It is generally considered a primary
pollutant, since only relatively small concentrations can be photochemically
generated.
6. Nitric Acid
Nitric acid vapor may be present in smog as a result of the reaction
of NO with wafer. Pure nitric acid is a colorless liquid at room temperature.
It slowly decomposes to form NO-, which in turn absorbs sunlight and hastens
decomposition. The pure gaseous acid is a strong oxidizing agent, attacking
metals and many organic substances. Ultraviolet photolysis of HNO- may be a
source of OH radicals in the atmosphere.
7. Sulfuric Acid
Sulfuric acid, H.SO , is an oily, exceedingly corrosive, nonvolatile
liquid which results from the oxidation of sulfur dioxide in the presence of
wdter vapor. It may play an important role in the formation of aerosols in
photochemical smog. It is miscible in all proportions with water and is very
highly hygroscopic. Droplets of sulfuric acid suspended in air, therefore,
grow or shrink as the relative humidity of the air increases or decreases.
4-91
-------
8. AorosoIs
Atmospheric particulates are mainly in the range of 0.05 to about 10A*
in size. They are capable of acting as condensation centers. They consist of
solid matter, droplets of solution, insoluble smokes and mineral dusts. The
•f = - -
presence of Nli and SO , H_SO , Cl , N0_ and other minor components in the
solution phase have been reported, the composition varying with size and
geographical location. Hygroscopic salts may be the origin of the ions in the
part icles.
The relation of aerosol loading to photochemical air pollution has
21 66
In en reviewed by Altshuller and Bufalini. Mader, et al. found
high organic loadings in atmospheric aerosols collected in Los Angeles.
Studying synthetic aerosols from irradiated mixtures containing gasoline vapor
,md nitrogen oxides in air, they found similar organic components. Infrared
and chemical analysis of both types of aerosols showed the presence of
c.trbonyl, hydroxyl, nitrate ester, peroxy, and carboxy groups. Renzetti and
Doyle showed that in irradiated mixtures of auto exhaust in air, large
amounts of organic aerosols were produced. Schuck, et al. showed
that aeiosoJs were produced more copiously in such experiments when the exhaust
K.ISCH were derived from fuels of high olefinic content. It thus appears that
at least some fraction of the aerosols found in ambient photochemical smog
i generated by the photochemical reactions.
4-92
-------
9. Epoxides
Cyclic ethers such as ethylene oxide may be formed to a small extent.
Ethylene oxide is a colorless gas at 25°C. It is a fumigant and insecticide but
not dangerously toxic to humans.
10. Alkyl Nitrates
Alkyl nitrates, such as ethyl nitrate, have pleasant fruity odors.
Th
-------
toxic and highly reactive compound. It reacts readily with such mild agents
,10 ethunol, water, ammonia, and acetic acid. It has been suggested, but not
detected, as a product in photochemical smog.
V. COMPUTER SIMULATION OF CHEMICAL KINETICS
1. Rationale for Kinetic Simulation
The classical task of chemical kinetics is the discovery of the
mechanism of n complex reaction—that is, the formulation of a set of elementary
le.ictiotiK, postulated as occurring in a reaction system, such that the observed
kinetic behavior of the components of the system can be shown to be a con-
sequence of the simultaneous occurrence of all the individual reactions of
tin' set .
The photochemical origin of ozone and other oxidants in polluted
urban atmospheres was first recognized over twenty years ago. Since then,
hundreds of experiments have been performed with synthetic smog, and dozens
of elementary reaction steps and partial kinetic mechanisms have been discussed.
As a result, substantial agreement has emerged with respect to most of the
chemical principles involved, and many important aspects of the reaction
mechanism.", have been clarified.
Nevertheless, the task of theoretical interpretation of the
photochemical smog phenomenon is incomplete, largely because of the chemical
complexity of the systems that have to be studied. Most of the complex
reactions whose kinetic behaviors have bten successfully explained by
tradition.il methods have been found to require mechanisms consisting of only a
few elementary reactions, and involving only a few molecular species, including
4-94
-------
, the reactants and observed products. Indeed, it is only for
such relatively small sets of elementary reactions that the algebra involved
could be. analytically solved or conveniently manipulated.
+ For example, Shepp , who studied the photolysis of acetone, was
nble to demonstrate the applicability of a mechanism involving five elementary
reactions and eight molecular species, including one reactant, four observed
products and three postulated free radical intermediates. When the experimental
conditions were restricted to temperatures above 100°C and relatively high
light intensities, observed product concentrations were consistent with the
algebraic expressions analytically derived from the mechanism. Experiments
involving intermittent light, pulsed at various intervals, permitted him
to estimate rate constants for the individual elementary reactions, even
though the concentrations of the free radical species were not independently
known. Although other reactions besides the five postulated are known to
occur tn such systems, the investigator was able to minimize their effect
in the system by a judicious choice of experimental conditions.
By contrast, dealing with the photo-oxidation of even a single
hydrocarbon with oxides of nitrogen in air requires postulating a mechanism
of dozens of elementary reactions. Since each postulated reaction furnishes
one additional differential equation to the collection, the mathematical
* analysis needed for their simultaneous solution becomes impractical.
In effect, this situation has been solved by the use of the
• high-speed numerical analysis capability of the modern digital computer,
together with the accumulated body of experimental kinetic information and
4-95
-------
validated theory concerning elementary reactions of molecules, atoms and free
radicals. Briefly, the system is to utilize the high-speed capability by
postulating a large number of elementary reactions which may be thought
important in the chemical system; to provide the computer with known or
theoretical rate constants for each, together with initial concentrations
tor all main components, and to generate therefrom (by numerical integration)
t IK- expected kinetic behavior of all the reactants, products and intermediates
known or .suspected to be present in the system.
A particular advantage of such computerized analysis is that, in
j.rii>cipl<-%, it is unnecessary to make a priori judgments as to the relative
importance of alternative reaction sequences in accounting for utilization
or iormation of the participating molecular species. Since the computer does
not develop analytical solutions to the simultaneous differential equations,
the inclusion of unnecessary elementary reactions does not interfere with the
solution process. Consequently, if there is uncertainty regarding the
importance of a proposed elementary reaction, it may be investigated by
including the reaction in the set to be tested. If the rate of occurrence
is ne^ii'^ible relative to that of other reactions in the set, this will
In-come apparent on examination of the behavior of the specific rates as a
function of simulated time, and the reaction in question can then be deleted
from the set. Thus the numerical integration routine furnishes a sort of
automatiL correction procedure for the elimination of unnecessary steps from
the postulated set.
Nevertheless, it is undesirable to expand the set of elementary
4-96
-------
rvacLions unnecessarily, since this would consume extra computing time.
2. The Basic Simulation Process
When the elementary reactions which can take place in a chemical
system are known, together with their corresponding rate constants, it is a
straightforward procedure to calculate rates of change of concentration
of all reactant and product species, provided the concentrations are known
for some specified instant of time. This is done for each species by
calculating (using equation 4-29) the rate of consumption for each elementary
reaction in which it is consumed and the rate of production for each elementary
reaction in which it is produced. A negative sign is affixed for each of the
consumption rate terms, and the net rate of concentration change is then the
total of all the terms (with their appropriate signs).
A systematic procedure is to list and number the individual
components of the reaction system as well as the individual elementary reactions,
then use the stoichiometric coefficients of the reactions in the form of a
matrix. An illustration of this procedure is shown in Table 4-17. A zero
entry at the intersection of a row and a column indicates that the component
corresponding to the column is not involved in the reaction corresponding
to the row of entry. An entry with a positive sign indicates that the
component is produced by the elementary reaction step, while an entry with a
negative sign indicates that it is consumed. A numerical value of j_ indicates
that only one molecule of the component in question is produced on each
occurrence of the reaction in question; if two or more such molecules arc
produced, the corresponding digit is inserted. Similarly, -_!_ indicates that
4-97
-------
Table 4-17. ILLUSTRATION OF MATRIX METHOD OF FORMULATING REACTION RATE EQUATION
(Mechanism by Shepp for photolysis of acetone)
CH,COCH, + hw - „ CH.CO + CH. ^ . k/ ,
3 J .> j a 1
CH CO - »-CH + CO k
CH + CI
et component i = CH COCH , 2 - CH.CO , 3 - CH3, 4 - CO , 5 - CH^ , 6 * Cl^COCH,.,
7 = C-H COCH., 8 - C.H,
i 5 3 2 o
Reaction
Number 1
1 -1
2 0
J -1
4 0
5 0
2
1
-1
0
0
0
System Components
3456
1
1
-1
-1
-2
0
1
0
0
0
0
0
1
0
0
0
0
1
-1
0
7
0
0
0
1
0
8
0
0
0
0
]
rl = dCl/dt - - VR3 ' -k/lCl - VlC3
r, = dC,/dc =
Cl + R2C2 -
4-98
-------
Table 4-17. ILLUSTRATION OF MATRIX METHOD OF FORMULATING REACTION RATE EQUATION
r4 - dC4/dt - ^ » k2C2
rr>
r, - dC,/dt - R,-R. - k,C C - k C C,
6 6 34 313 436
r? - dC7/dt
rg - dC8/dt
one molecule is consumed, and -2_ that two molecules are consumed on each
occurrence of the given elementary reaction step.
From the matrix (Table 4-17) the expressions for the rates of the
L'lfmoniary reactions are formulated by reading the corresponding row to
identify the reacting components (corresponding to the columns containing
negative stoichiometric coefficients), then multiplying their concentrations,
each raised to the appropriate power, together with the rate constant for
the reaction. The rate of change of concentration of each component is then
formulated by reading down the appropriate column for non-zero entries and
adding the individual rates of the indicated reactions, each multiplied ty
the indicated stoichiometric coefficient (including the sign).
Mathematically this process can be expressed as
R -fl (S: C )Sij , J-l,...n (4-62)
J 1=1 1J
4-99
-------
where
if Sy < 0, (4-63)
S' - 0 if S^ > 0. (4-64)
and n
r
i S R , i-1, ...m. (4-65)
In these expressions R. is the instantaneous rate of the jth
elementary reaction, while r, is the net instantaneous rate of production of the
ith component; C is the concentration of the ith component, and S . . is the
.stoichiomet ric coefficient for component i in reaction j. S' , as defined in
equations (4-63) and (4-64) is the absolute value of S. for components
consumed by the reaction only.
In principle, the procedure within the computer is to calculate
the rates of change of all components, use these rates to estimate the
expected changes of concentration during a specified minute interval of
simulated time, then accordingly increment each of the concentrations pre-
viously calculated (or specified as input). This procedure is then repeated
as often as necessary to reach the total desired simulated time. In
practice it is necessary, however, to apply all possible simplifications and
aggregations in the computation routine, to reduce computation time and _ost,
Consistent with reasonable accuracy in the output.
j. Acceptable Simplifications
Some of the standard simplifications ased in chemical kinetics
4-100
-------
in deriving the mathematical consequences of assumed chemical mechanisms
are also useful in abridging the necessary input, output, or computations
required in computer simulation.
a. Approximate Constancy of Concentration of Components Present in
Large Excess.
The primary example of a component present in large excess in the
Atmospheric photochemical system is molecular oxygen, with a concentration near
200,000 ppm. Even high concentrations of hydrocarbons in the ambient
atmosphere, say 20 ppm as carbon, if completely converted to carbon dioxide
and water, would consume far less than one thousandth of the oxygen available.
Thus the concentration of oxygen cannot change appreciably due to photochemical
reactions, and it may be treated as a constant factor in those elementary
rc,K:tions in which it occurs. For example, in the case of the reaction
2NO + 0 = 2ND , (4-66)
the regular rate expression is
R = k (N0)2(0 ). (4-67)
Assimilating the essentially constant value of (0~) into the rate constant
by t'ie device
k/ = k(02), (4-68)
the rate of the elementary reaction can be more briefly expressed as
R = kx (NO)^. (4-69)
4-101
-------
The same considerations can be applied to recombination reactions
of atoms and small radicals, where a third molecule is necessary to carry away
the t'xcess energy; for example,
20H + M - H202 + M (4-70)
In this case, the formal rate expression is
R = k (OH)2(M) (4-71)
where (M) is the concentration of all unreactive molecules in the system. Here
the rate is more simply expressed as
R - k' (OH)2, (4-72)
where
k' = k(M). (4-73)
S i not- it is of little interest to compute the relatively minute changes of
con contrat ton of oxygen, water vapor and carbon dioxide which occur as the
photochemical reactions proceed, these substances may be omitted from the
list of components in the reaction matrix, provided that appropriate
pseudo constants are calculated for the reaction steps which involve them.
b. Steady State Concentrations for Extremely Short-Lived Species
Free radicals, atoms, and various other highly reactive molecular species
tend to be consumed so rapidly after they are formed that their average
period of intact existence is very short. More importantly, under these
conditions their concentrations remain always negligible relative to the
i unceiit rat ions of the less reactive primary contaminants. They remain, in
most ij.ises, so low that ordinary experimental methods fail to detect them;
4-102
-------
yet, at the same time, the reactions in which they are produced and consumed
may proceed as rapidly as, or more rapidly than, the reactions of the less
reactive components.
To simplify the computations involving these species, whose
exact concentrations in any event have yet to be experimentally tested, the
principle approximation used in standard kinetic analysis is that of the
"pseudo-steady state." It is assumed that the rates of change of concentration
of siu-h species are negligible. Then the concentration of each such component
can be derived from the rates of the individual steps. In fact, it is found
by dividing the total rate of production by the total relative rate of con-
sumption. Mathematically, the rate of production of component i_ is
rpl - S"y V (4-7A,
where
S " S lf
S" =0 if S < 0. (4-76)
The rate of consumption is
r = I S' Rj, (4-77)
J-l J
4-103
-------
with s'. defined as above (Equation 4-63) and 4-64), while the relative
rate of consumption is
r . - r ./C. - Z 01 (S' C. )°ki)/C,.
ri ci i jml k-1 kj k i
(4-78)
Tims the approximate value of the desired concentration is
C, - r ,/r .. (4
i pi ri
c. Aggregation of Free Radical Termination Steps
Since free radicals, when they recombine, cause termination of
radical chains, the aggregate effect of such steps may have an important
effect on chain lengths and, thus, on rates of product formation. However,
the absolute rates of these reactions are relatively small because of the
very low concentrations of the radicals; hence the formation of recombination
products is not important relative to other routes of product formation. To
avoid expanding the reaction set to include a large number of recombination
reactions, it is convenient to assume that all such reactions have the same
rate constant, and thus aggregate the corresponding elementary rate terms
Into a single term for each listed free-radical component.
4. Photo-oxidation Simulations »
Mechanisms of the type discussed here have been successfully
ut i 1 izi'il in digital computer simulation of photo-oxidation kinetics by Wayne and *
'/' 73 »
Ernest " and by Westberg and Cohen . In the first example, a list of forty
4-104
-------
proposed elementary reactions served as the hypothetical mechanism, involving
as initial reactants three species (nitrogen dioxide, nitric oxide and propylene)
in addition to atmospheric oxygen; nine products, in addition to carbon
dioxide and water; and fifteen intermediate species, mainly free radicals.
This list of reactions is shown in Table 4-18.
For these forty reactions, rate constants were either taken from
the literature or estimated on the basis of modern methods in chemical kinetics
Tht- degree of success attained in simulating, with this mechanism, several
experiments in the photo-oxidation of propylene, is illustrated in Figure 4-">.
and Table 4-19. Even without any substantial effort at fine tuning, all
simulated concentrations were within 20 percent of the experimental points,
not only for the species shown in the figure, but also for the aldehydes
produced by the photo-oxidation.
4-105
-------
Taolc 4-18. 40-STEP MECHANISM FOR PHOTO-OXIDATION OF PROPYLENE
(27 participating species underlined, excluding
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
02, H20, C02)
N02 -f hv — > NO + 0
0 -f 02 + M > 03 + M
03 + NO > N02 + 02
2NO + 02 > 2N02
N02 -f 03 > N03 + 02
N03 + N02 > N205
N03 + NO * 2N02
NO + H02 > N02 + jOH
C2H303 + NO > C2H302 + N0?
C2H3°2 + N0 ^ C2H3° + N02
CII^O + NO >-CH3 + N02
CH 0 + NO * GILO. + N00
-. J J J / 2.
CH302 + NO > CH30 + N0?
C2H4°2 + N0 ^ CH3CHO + NO,
CH30 + 02 + NO * CH302 + N02
CH3 + N02 > CH30 + NO
CH302 + N02 j» NO + 03 + CH
° + C3H6 *CH3 +C2H3°
°3 + C3H6 *• HCHO + C2H402
CH30 + C3H6 ^2CI!3 + c2H30
4-106
-------
Table 4-18. 40-STEP MECHANISM FOR PHOTO-OXTDATTON OF PROPYLENE (continued)
(27 participating species underlined , excluding
°2' H2°' CCV
21. CH 0 + C H - > CH. + CH 0 + C.H-0
J Z J D J J 2 J
22. C2H3°2 - ^ HCHO + CHO
23. C2H30 + 02 - -> C2H303
2A . CH3 + 02 + M — » CH302 + M
25. C2H30 + OH — * CH3CHO -1- 0
26. CH302 -f 02 — > 03 + CH30
27. .0 -»- HCHO - > OH -I- CHO
28. 0 + CHO — > CO 4- OH
29. OH + CHO — > CO 4- H20
30. CH3CHO -f OH - > CH3CO + H20
31. 2CH3CO - > CH3CHO + C^O
32. CH30 + 02 - > CH303
33. CH30 + 02 - > HCHO
34 . _H + 02 - > H02
35. CO + OH - -> H + C0
36. 20H > 0 + H20
38 • CoHoOo ~i~ NO'j """"*""?
39. CH3 + N02 - > CH3ONO
40. CH3 + NO - > CH3NO
4-107
-------
O O 0100
>• ?<: u o
D O
03 i
4-1
Ci "O
Q) QJ
S "^
•H (13
0) 3
X' -H
W en
CO
w
H
LO
I
60
•H
En
4-108
-------
r
rf
Table 4-19. COMPARISON OF SIMULATED RESULTS WITH EXPERIMENTAL
PHOTO-OXIDATION RUNS
Initial
Concentrations,
pphm
HC NO N02
100
100
200
100
100 10 sim.
exp.
100 2 sim.
(10) exp.
100 10 sim.
exp.
200 10 sim.
exp.
N02
79
85
72
85
79
80
140
150
Maxima ,
pphm
03 CH20
35
37
35
37
42
45
26
30
44
48
44
48
85
80
38
40
Half-time,
minutes
NO C3H6
30
40
30
40
25
30
63
80
115
115
115
115
90
95
145
150
4-109
-------
A similar degree of success has been reported by Westberg and
Cohen in simulating the kinetics of photo-oxidation as measured in an
experiment by S chuck and Doyle. The comparison of measured and calculated
i.ite.s tor several components of the system is shown in Figure 4-6. The
me eh, -ui i MII ,»s presented by the authors comprises 42 equations, but since
sev. 11! of tlu'ue represent general reactions of a series of radicals, the
lot.i! ii.n:.i)er of reactions involved is 68, with more than 40 participating
Exploration of the effects of variations in the rate parameters
which were subject to estimation showed only four to which the simulated
ivites wen' significantly sensitive. These were as follows:
R()7« + NO = RO. + NO
H09* f RO • = ROOH + 0
(CH1CO)0? + N02 * (CH3CO)OON02
(A:, between the last two of these reactions, only the ratio of the rates was
. ip,ii i 1 i can t . )
The calculations further indicated a strong possibility that
carbon monoxide could play an important role in converting nitric oxide
• :i i 11•.',,'.( :i dioxide and in generating c^cne ^n a polluted atmosphere.
4-110
-------
Experimental
Simulated
20
40
60
80
100
TIME, MINUTES
Figure 4-6. TIME-DEPENDENCE OF CONCENTRATIONS IN SIMULATED PHOTO-OXIDATION
OF ISOBUTENE, COMPAREQ TO EXPERIMENT. FROM CURVES PRESENTED
BY WESTBERG AND COHEN
73
4-111
-------
REFERENCES
I. L.'itfhton, P. A., "Photochemistry of Air Pollution," Academic Press,
Now York, 1961.
divert, J. G., Pitts, J. N., Jr., "Photochemistry," John Wiley and
Sous, Inc., New York, New York, 1966.
II.ill, T. C., Jr., Blacet, F. E. , J. Chem. Phys. , 20, 1745, 1952.
it. Khun, A. V., Pitts, J. N. , Jr., Smith, E. B. , Environ. Sci. Tech. , 1,
8, 656, 1967.
S. PUts, J. N. , Jr., Khan, A. V. Smith, E. B., Wayne, R. P.,
Knviron. Sci. Tech., 3^ 3, 241, 1969.
h. MrNf.shy , .1. R. , Okabe, H. , "Vacuum Ultraviolet Photochemistry,"
Advaiu-fs in Photochemistry, Vol. Ill, ed., by W. A. Noyes, Jr.,
(',. S. Hammond, and J. N. Pitts, Jr., Interscience Publishers, a
division of John Wiley and Sons, New York, 1964, page 157.
7. Pitts, J. N., Jr., Sharp, J. H. , Chan, S. I., J. Chem. Phys., 30,
238, 1963; Ibid. 40_, 3655, 1964.
8. Metropolis, N., Beutler, H., Phys. Rev., 57, 1078, 1940.
9. Dainton, F. , S., Ivin, K. J., Trans, Faraday Soc., 46, 374, 382, 1950.
10. Reynolds, W. C. , Taylor, W. H., J. Chem. Soc. , 1912, page 13.
11. ,lu!fe, S., Ford, H. W. , J. Phys. Chem. , 71, 1832, 1967.
12. Gr.iv, P., Yoffe, A. D. , Roselaar, L. . T'rans. Faraday Soc. , 51, 1489, 1955.
i <. iUiwn, li. W. , Primetel, G. C. , J. Chem. Phys. , 29, 883, 1958.
i ,. NI.no I sou, A. J. C., Nature, 190, 143, 1961.
4-112
-------
15. Rebbert, R. E., Slagg, N. , Bull Soc. Chim. Beiges^ 71, 709, 1962.
16. Rubin, T. R., Calvert, J. G., Rankin, G. T., MacNevin, W. M.,
J. Am. Chem. Soc. , _?!» 2850, 1953.
17. Frost, A. A., Pearson, R. G., "Kinetics and Mechanism," 2nd edition,
John Wiley and Sons, Inc., New York, p. 241-259, 1961.
18. Rice, F. 0., Herzfeld, K. F., J. Am. Chem. Soc. , 5jj, 284, 1934.
19. Leighton, P. A., "Photochemistry of Air Pollution," Academic Press,
New York, 1961.
20. Wayne, L. G., "The Chemistry of Urban Atmospheres," Technical Progress
Report, Volume III, Los Angeles County Air Pollution Control District,
December, 1962.
21. Altshuller, A. P., Bufalini, J. J., Photochem. and Photobiol.. 4., 97, 1965.
22. Stephens, E. R., Int. J. Air Water Poll., 1CI, 649, 1966
23. Haagen-Smit, A. J., Wayne, L. G., "Air Pollution," Volume I, 2nd ed. ,
A. C. Stern, Ed., Academic Press, New York, 1968.
24. Johnston, H. S., NSRDS-NBS 20, Sept. 1968.
25. Schott, G., Davidson, N., J. Am. Chem. Soc., 80, 1841, 1958.
26. DeMore, W. B. International Journal of Chemical Kinetics, l^, 209, 1969.
27. Klein, F. S., Herron, J. T., J. Chem., Phys., 41, 1285, 1964.
28. Ford, H. W. , Endow, H. , J. Chem. Phys.. 2_7, 1156, 1277, 1957.
29. Johnston, H. S. , Yost, D. M. , J. Chem. Phys. , 17., 386, 1949
30. Klein, R., Scheer, M. D., J. Phys. Chem. , _7_3» 1598, 1969
31. Ford, H. W., Doyle, G. J., Endow, N., J. Chem. Phys.. 26, 1337, 1957.
4-113
-------
\JL. Wayne, L. G., Yost, D. M. , J. Chem. Phys. , l^_, 41, 1951.
jt. Tuesday, C. A., "Chemical Reactions in the Lower and Upper Atmosphere,"
R. D. Cndle, Ed., Interscience, New York, 1961.
Vt. .laffo, S. , Klein, F. S. , Trans. Faraday Soc., 62, 2150, 1966.
r,. Kaufman, F., Proc. Roy. Soc. A., 247, 123, 1958.
it,. Cv. i.movir, R. J., Can. J. Chem. , 38_, 10, 1681, 1960.
(/. f.v.-t auovie, R. J., J. Chem. Phys. , 33, 4, 1063, 1960.
) 191, 1969.
'i 1 . CvL'tanovic, R. J., Advan. Photochem. , JL., 115-182-1963.
.'»:.. Vrbaski, T. , Cvetanovic, R. J. , Can. J. Chem. . 28, 1053, 1063, 1960, Ozonolysis
.'( i. CadU>, R. D. , Shadt, C., J. Chem. Phys., 21, 163, 1953.
.'(4. llaiiht , 1'. L. , Stephens, E. R. , Scott, W. E. , Doerr, R. E. , 136th Meeting
of iho Amer. Chem. Soc., Atlantic City, September 1958.
AS. Saltzm.-m, B. E. , Gilbert, N. , Ind. Eng. Chem. , 51, 1415, 1959.
A6. Haagtm-Smit, A. J., Bradley, W. E., Fox, M. M., Ind. Eng. Chem., 45,
208b, 1953.
A 7. Cladle, R. D. , "Proceedings of the Conference on Chemical Reactions in
Urban Atmospheres," Report No. 15, Air Pollution Foundation.
San Marino, California, November 1956.
,^. IK huek, h. A., Doyle, G. J., Endow, N., "A Progress Report on the
l'h..i orhcmi.st ry of Polluted Atmospheres," Report No. 31, Air Pollution
1 .Mniu.it i on , San Marino, California, December 1960.
4-114
-------
49. Taylor, R. P., Blacet, F. E., Ind. Eng. Chem., 4£, 9, 1505, 1956.
50. Schuck, E. A., Doyle, G. J., "Photooxidation of Hydrocarbons in
Mixtures Containing Oxides of Nitrogen and Sulfur Dioxide," Report No. 29,
Air Pollution Foundation, San Marino, California, October 1959.
51. Stephens, E. R., in "Chemical Reactions in the Lower and Upper Atmosphere,"
R. D. Cadle, Ed., Interscience, New York, 1961.
52. Hanst, P. L., Calvert, J. G. , J. Phys. Chem.. j>3, 1, 104, 1969.
53. McMillan G. R., Calvert, J. G., Oxidation Combust. Rev., !_, 83, 1965.
54. McKellar, J. F. , Norrish, R. G. W., Proc. Roy. Soc. , London, A263, 51, 1961.
55. Heicklen, J., "Reactions of Alkylperoxy and Alkoxy Radicals,"
International Oxidation Symposium, San Francisco, California,
August 28 - September 1, 1967.
5(>. Calvert, J. G. , Hanst, P. L. , Can. J. Chem. . _37, 1671, 1959.
57. Sleppy, W. C., Calvert, J. G., J. Am. Chem. Soc. . 81, 769, 1959.
58. Birss, F. E., Danby, D. J., Hinshelwood, C., Proc. Roy. Soc. (London),
A239, 154, 1957.
59. McMillan, G. R. , Wijnen, M.H. J. , Can. J. Chem. , J16_, 1227, 1958.
60. Leighton, P. A., "Chemical Reactions in the Lower and Upper Atmosphere,"
Interscience Publishers, N. Y., 1960, pp 1-14.
61. Stephens, E. R., Darley, E. F., Taylor, 0. C., Scott, W. E.,
"Photochemical Reaction Products in Air Pollution." Proc. Am. Patrol.
Inst., ^0, III, 1960.
62. Stephens, E. R., Anal. Chem., 36, 928, 1964.
63. Darley, E. F., Kettner, K. A., Stephens, E. R.. Anal. Chem.. 35. 589, 1963.
4-115
-------
64. Stephens, E. R., Burleson, F. R., Cardiff, E. A., APCA Journal, 15,
87, 1965.
dri. Mason, B. J., in "Chemical Reactions in the Lower and Upper Atmosphere,"
Interscience Publishers, N. Y., 1961, pp 197-218.
(if). Mader, P. P., MacPhee, R. D. , Lofberg, R. T. , and Larson, G. P.
I'l'l- 'in_8jL Chem. 4^, 1352 (1952)
67. Rt-nzftti, N. A., Doyle, G. J., Int. J. Air Water Poll. , 2., 327, 1960.
68. Schuck, E. A., Ford, H. W., and Stephens, E. R. , Report No. 26,
Air Pollution Foundation, San Marino, California (October 1958).
69. "The Oxides of Nitrogen in Air Pollution," State of California,
Department of Public Health, Bureau of Air Sanitation, January 1966.
70. Schuck, E. A., and Doyle, G. J., "Photo-oxidation of Hydrocarbons in
Mixtures Containing Oxides of Nitrogen and Sulfur Dioxide,"
Report No. 29, Air Pollution Foundation, San Marino, California, 1959.
71. Shepp, A., J. Chem. Phys.. 24_, 939, 1956.
II. Wayne, L. G., and Ernest, T. E., "Photochemical Smog Simulated by
Computer," paper 69-15, presented at 62nd Annual Meeting of the Air
Pollution Control Association, New York, June 1969.
73. Westberg, K., and Cohen, N., "The Chemical Kinetics of Photochemical
Smog, as Analyzed by Computer," Report No. ATR-70 (8107)-!, The
Aerospace Corporation, El Segundo, California, December 1969.
4-116
-------
CHAPTER 5
AIR QUALITY APPRAISAL
Page
A. INTRODUCTION 5-10
B. GENERAL SAMPLING AND STANDARDIZATION METHODS 5-11
C. NON-NITROGENOUS OXIDANTS 5-14
1. "Net Oxldant" Concept 5-14
2. Standards 5-15
3. Total Oxidant Measurement Techniques 5-17
a. Colorimetric KI 5-17
b. Coulometric 5-18
c. Coulometric vs. Colorimetric Methods 5-18
d. Other Methods 5-20
1) Alkaline KI 5-20
2) Acid KI 5-20
3) Phenolphthalin 5-21
4) Miscellaneous 5-21
D. SPECIFIC OZONE METHODS 5-21
1. Chemiluminescence 5-21
2. Rubber Cracking 5-22
3. Ultraviolet Photometry 5-22
4. Other Methods 5-23
5-1
-------
ii. OXIDES OF NITROGEN
1. Standards
2. Colorimetric Methods
3. Other Methods
K. OASIXHIS ORGANIC POLLUTANTS
I. Total Hydrocarbons
a. Flame lonization Analyzers
1). Spectrostopic Methods
2. Hydrocarbons by Type
a. Gas Chroraatography
I). Spcctrometric Method
i-. Miscellaneous Methods
3. Oxygenated Gaseous Organic Methods
a. General Method
!.. Bisulfate Method
i'. Other Condensation Reagents
G. PKROXYACYL NITRATES
1. Long Path Infrared Spectroscopy
2. Gas Chromatography
II. AEROSOLS AND ATMOSPHERIC TURBIDITY
1. Particulate Collection Methods
5-23
5-23
5-24
5-25
5-25
5-25
5-25
5-27
5-27
5-27
5-28
5-28
5-28
5-28
5-29
5-29
5-30
5-30
5-31
5-31
5-31
5-2
-------
a. General Properties of Photochemical Aerosols 5-31
b. Filtration 5-32
c, Inertial Collection 5-33
d. Electrostatic Precipitation 5-34
e. Thermal Precipitation 5-34
2. In Situ Aerosol Analyzers 5-34
a. Photometric 5-34
b. Condensation Nuclei Counters 5-36
I. BIOLOGICAL INDICATORS 5-36
1. Bacterial Response 5-36
2. Plant Damage 5-37
3. Sensory Irritation 5-37
J. AIR QUALITY DATA 5-38
1. Methods of Expressing Air Quality 5-40
2. Typical Concentration Patterns in Photochemical 5-45
Air Contaminants.
a. Diurnal Variations 5-46
b. Seasonal Variations 5-55
3. Ozone vs. Oxidant Measurements 5-58
4. Observed Contaminant Concentrations 5-72
a. Oxidant 5-72
b. Nitrogen Oxides 5-78
5-3
-------
Page
1) Series Flow 5-78
2) Parallel (Mod. 1) 5-78
3) Parallel (Mod. 2) 5-79
c. Hydrocarbons 5-84
d. PAN 5-92
c. Aldehydes 5-96
5. Trends in Photochemical Air Contaminants 5-99
K. DATA ACQUISITION REQUIREMENTS FOR DETERMINING REGIONAL 5-109
AIR QUALITY
1. Suggested Measurements 5-109
2. Sampling Network Design 5-110
J. Station Siting 5-115
4. Data Processing and Validation 5-116
5-4
-------
CHAPTER 5
LIST OF FIGURES
Figure Page
5-1 Apparatus for Dynamic Calibration 5-16
5-2 Comparison of Data Resulting from Three 5-42
Sampling Approaches.
5-3 Diurnal Oxidant Concentrations, Pomona, California, 5-44
September 1966.
5-4 Concentrations of Pollutants—Los Angeles, California 5-47
5-5 Diurnal Variation in Oxidant and NO,, Concentrations 5-'jO
5-b Diurnal Variation of Total Oxidant Levels— 5-5)
Philadelphia, Pennsylvania, August 6, 1966.
5-7 Oxidant Monthly Mean Hourly Average Concentrations— 5-52
October 1965.
5-8 Carbon Monoxide Monthly Mean Hourly Average 5-54
Concentrations—October 1965.
5-9 Oxidant Concentration by Month (Mean of all Hours) 5-56
5-10 Oxidant Concentration by Month (Mean of Maximum Hours) 5-57
5-11 Mean Monthly NO Concentrations 5-59
5-12 Mean Monthly NO. Concentrations 5-60
5-13 Comparison of Ozone and Oxidant Monthly Mean of Hourly 5-^4
Average Concentrations—Los Angeles and Pasadena,
1964-1965.
5-5
-------
Vigurc Page
'j-14 Comparison of Ozone and Oxidant Monthly Mean of 5-65
Maximum Hourly Concentrations—Los Angeles and
Pasadena, 1964-1965.
5-15 Comparison of Ozone, Oxidant and Oxidant Adjusted for 5-66
N0« and SO,, Response—Los Angeles and Pasadena, July 1964.
5-Lh Average Ozone Concentrations by Hour 5-70
5-17 Diurnal Variation of Ozone Concentration in Philadelphia 5-71
and Denver.
!>-18 Nonmethane/Methane Hydrocarbons Ratios—Cincinnati and 5-86
Los Angeles (1964).
r>-19 Oxidant and PAN Concentrations by Hour of Day—Downtown 5-93
Los Angeles—(1965).
r>-20 Oxidant and PAN Concentrations by Hour of Day Air 5-94
Pollution Research Center—Riverside, California—(1966).
S-21 Oxidant and PAN Concentrations Air Pollution Research 5-97
Center—Riverside, Califorria—(June 1966 Through
June 1967, Inclusive).
•>-22a Monthly Mean Daily Maximum NO , Civic Center, 1957-1961. 5-102
X
Significant Trend, +8.6 pphm/Year.
r)-22b Trend Curves Fitted to "Deseasonalized" Monthly Mean 5-102
Daily Maximum NO Data, Los Angeles Civic Center, Stations
X
I and 58, 1957-1961.
5-6
-------
Figure
5-23a Monthly Mean Daily Maximum NO , Burbank, 1957-1961. 5-103
A
No Significant Trend.
5-23b Trend Curves Fitted to "Deseasonalized" Monthly Mean 5-103
Daily Maximum NO Data, Burbank, Station 9, 1957-1961.
5-24 Oxidants 5-104
5-25 Nitrogen Dioxide 5-105
5-26 Annual Variation in Number of Days Oxidant and N07 5-107
Exceed Stated Levels Together with Number of Days
Conducive to Accumulation of Air Contaminants — Los
Angeles Basin.
5-27 Factors to Consider in the Development of an Overall 5-113
Air Quality Appraisal Implementation Plan.
5-7
-------
Tabit
CHAPTER 5
LIST OF TABLES
Effect of NO- Oxidant Determination in Absence of Ozone 5-19
5-2 Oxidant Concentrations Adjusted for Nitrogen Dioxide 5-62
Response Monthly Means of Hourly Average Concentrations,
ppm.
5-3 Oxidant Concentration Adjusted for Nitrogen Dioxide 5-63
Response Monthly Means of Daily Maximum Hourly Average
Concentrations.
5-4 Cumulative Hourly Average Oxidant Concentrations (1964- 5-73
1965).
5-5 Oxidant Concentrations (1964-1965) 5-74
5-6 Highest Monthly Mean Oxidant Concentrations (1964-1965) 5-76
5-7 Total Oxidant Concentrations Exceeding Selected Levels 5-77
(1967-1968.
5-8 Cumulative Frequency Distributions of 5-Minute Values — 5-80
Nitric Oxide and Nitrogen Dioxide (1966 CAMP).
5-9 Cumulative Frequency Distribution of Hourly Average 5-81
Concentrations of Nitrogen Dioxide and Nitrogen
Oxides (1967 SCAN).
5-8
-------
Table Page
5-10 Cumulative Frequency Distribution of Hourly Average 5-82
Nitrogen Dioxide Concentrations (1967 New York-
New Jersey).
5-11 Average Hydrocarbon Composition—218 Los Angeles 5-87
Ambient Air Samples (1965).
5-12 Average and Highest Concentration Measured for Various 5-88
Aromatic Hydrocarbons.
5-13 Average Atmospheric Light Hydrocarbon Concentrations— 5-89
Downtown Los Angeles—(Fall 1967).
5-14 Comparison of Results from the Ultraviolet Irradiation 5-91
of Ambient Air Samples.
5-15 Average and Maximum Concentration of PAN—Riverside, 5-95
California—(August 1967-April 1968).
5-16 Range of Maximum Concentrations of Aldehydes and 5-98
Formaldehyde—Los Angeles County—(1951-1957).
5-17 Summary of Total Oxidant Concentrations Recorded at 5-106
Camp Sites, 1964-1967.
5-9
-------
A. INTRODUCTION
Air quality appraisal consists of the measurement of contaminants
in the atmospheric photochemical pollution system and/or their indicators to
assess temporal and spatial air quality profiles in a given region. Measure-
ment of air quality is constrained by costs and technical feasibility in
terms of the design, performance and geographical deployment of sensors,
operational and maintenance requirements and the representative sampling
that may be required to give valid results in any region. Because of these
considerations, an "optimized," i.e., cost-effective, strategy to appraise
photochemical (as well as nonphotochemical) air quality is required. The
considi-rations involved in the development of an implementation plan are
discussed in paragraphs J. and K. of this chapter.
A cost-effective air quality appraisal plan is based, initially, on
an analysis of current and prospective technical requirements, and costs and
benefits are compared in order to select the sensors and sensor systems that
are most preferred in meeting these requirements. An optimum plan would
probably lead to applying measurements to carefully selected key contaminants
which, when coupled with the measurement of selected environmental parameters
(winds, mixing height, solar radiation), yield the greatest amount of
interpretive information and the greatest degree of verification and
predictability within the resources available.
5-10
-------
The interaction of numerous reactants under'irradiation in the
photochemical system leads to a large number of intermediates and products,
many of which are themselves capable of entering into the reaction sequence to
some degree. Even in the most simplified synthetic smog systems the estimation
of concentrations of all products as they vary with time has been an immense
task. Only certain components can be practically measured in the atmosphere,
some of these with great effort. Nevertheless, this assay is possible and is
necessary to show clearly the nature of any given photochemical air pollution
system. Suitable techniques for estimating the concentrations of contaminants
in terms of the state-of-the-art and current consensus are described in the
following paragraphs.
B. GENERAL SAMPLING AND STANDARDIZATION METHODS
The data necessary to determine the air quality in a region should
be adequate enough to cover the region spatially and frequent enough over
sufficiently protracted time periods to show normal fluctuations. The number
of observations that may be required, therefore, is large enough to make
automatic analysis preferable to manual, where available. Although automatic
equipment is more expensive initially, its much lower requirements in man-hours
per sample period soon compensate for initial costs. Manual methods are still
needed, however, for (1) calibration, (2) management of special short-term
sampling problems, and (3) situations in which automatic methods exist. These
problems have been recently reviewed.
5-11
-------
Sampling an air pollutant usually involves moving a measured amount
ot air through or into a scrubber or collector. The inlet tube is sized
a<. cording to flow rate and must not absorb or react with the pollutant to be
measured. Usually glass, stainless steel or certain plastics such as
polyethylene or Teflon are used, depending on the pollutant. In a grouping
oi several analyzers, a common manifold with a tube to each analyzer is
customary. The air is moved by means of a pump. To avoid contamination by
pump oil or other pump components, a vacuum pump is ordinarily placed down-
stream from the collector. If the pump must be used upstream with positive
pressure, it should be of a type in which components coming into contact with
the sample are inert towards the pollutant.
Measurement of the sample is done either by timing the flow at a
measL-red flow rate or by measuring the total sample volume. Flows are
conveniently measured by pressure drop across a constriction in the inlet
2
(such as a capillary, hypodermic needle, or nozzle) or by a rotameter. Either
ot those types should be calibrated against a volume-measuring device such as
a wet or dry test meter, and the flow checked frequently for steadiness during
the sampling period. Automatic analyzers with liquid reagents also require
steady flow in one or more liquid streams, usually gauged with rotameters.
Calibration at periodic intervals is required for either manual or
automatic methods. It may be dynamic or static. Dynamic calibration involves
passing the pollutant in air into the system at realistic concentrations as if
it were an actual sample. This may be fairly difficult because of the very
5-12
-------
low concentrations of pollutants, and the reactivity or condensability of
some of them. Specific cases are discussed below as they arise.
Various gas dilution systems have been described. A fairly simple
4
method recently developed is the permeation tube, suitable for pollutants
which may be passed through the walls of a tube of a specific plastic at a slow rate
which depends on the surface area and the temperature. To generate a known
dynamic concentration, air is passed over the tube in a thermostatted vessel.
The tube may be calibrated-gravimetrically if held at .constant temperature,
and the permeation rate remains constant as long as appreciable liquid remains
inside. Another technique is to add a measured amount of pollutant to a known
fixed volume of air in a large vessel. The addition may be made by syringe
injection or by crushing a weighed glass ampoule containing the pollutant. If
the vessel is rigid it must be large relative to the amount of sample to be
withdrawn. Bags made of plastics such as Mylar or Tedlar may be used for this
purpose, and have the advantage of collapsing as the sample is withdrawn.
In static calibration a standard solution of the pollutant, a
substance which will generate the pollutant, or the final reaction product may
be used. Quantitative dilutions to various levels are made and the appropriate
later stages of the determination are made, omitting the gas-contacting step.
Static methods are inherently capable of better control and precision than
dynamic methods, but give no indication of gas absorption or r action
efficiencies. Occasionally samples must be returned to a central laboratory
for analysis. The form in which they are transported must be chosen so that
5-13
-------
they do not deteriorate in storage or transit. Some may be absorbed in the
appropriate reagent, others must be kept in gas form. For the latter purpose,
evacuated glass vessels or plastic bags are sometimes used.
For condensable pollutants a freeze-out technique may be used. The
air is passed through a trap immersed in a cold bath (ice, dry ice, or liquid
air) suitable for condensing the materials of interest. Sometimes the trap
is packed with an inert solid absorbent. Later the trap is connected to the
detector and warmed up to volatilize the sample. In this method the normal
large quantities of water vapor in air also condense. The water can
sometimes be removed by a drying agent if the pollutant of interest is not also
at fee ted as a result.
C. NON-NITROGENOUS OXIDANTS
•!•• "Net Oxidant" Concept
Total oxidants in air may be defined as compunds that will oxidize
a reference material which is not capable of being oxidized by atmospheric
oxygen. Methods which measure oxidants, therefore, involve exposing a
reference substance to a sample of ambient air, and determining the degree of
oxidation which has occurred. This is done in different ways, depending on
the reference substance chosen. The following factors should be considered:
a. Ambient air contains a mixture of oxidizing and reducing
compounds—ozone, nitrogen dioxide, peroxaycetyl nitrate (PAN), sulfur dioxide,
hydrogen sulfide, aldehydes, unsaturated hydrocarbons, and others.
h. The concentrations of these compounds may be constantly varying
within relatively wide limits—from 0-2 rag/m (0-3 ppm as ozone).
5-14
-------
Since reducing compounds in air have an opposite effect on the
reference material—that is, a decrease in the degree of observed oxidation—
the result obtained by potassium ioxide (KI) methods (see paragraph 3. below),
unless empirical corrections are made, is a "net" oxidant value rather than
a total one. The net value thus describes a condition of the air, rather
than a specific compound concentration. For this reason, concentrated
efforts are being made by researchers to obtain measurements for each specific
oxidant present. Hopefully, these efforts will enable air pollution
scientists to better define this atmospheric "condition."
Most oxidant data which have been collected are net determinations,
and a high correlation between these values and other pollutant levels has
been observed. Consequently, until the more promising methods become
routinely available, "net" determinations must serve as indicators of total
oxidant levels.
2. Standards
For dynamic calibration ozone may be generated in the air stream
either by glow discharge or by irradiation with a germicidal ultraviolet lamp.
Since neither method offers a precisely known ozone concentration, determination
o
by manual absorption train is required. Usually potassium iodide reagent is
used and is assayed colorimetrically or by titration with thiosulfate. Electric
discharge in air may produce some oxides of nitrogen. These should be
measured by a separate method, since the oxidant method is sensitive to them.
The influence of these contaminants can be negated by generating the ozone in
a stream of pure oxygen instead of air. Figure 5-1 illustrates this procedure.
5-15
-------
II
E
m o
D <
CD >
u
.
CC C-
1
o
u.
U.
UJ H
§§
N ^
O u-1
cc -r
UJ
O o
OC tM
am
-------
Except for nitrogen dioxide and peroxyacyl nitrates, which are
treated separately below, there is little experience in dynamic calibration
with other oxidants. They are calibrated with static procedures. Hydrogen
peroxide is available in reagent grade and may be standardized by conventional
titrimetric methods. Certain other peroxides such as t-butylhydro-peroxide,
di-t-butyLperoxlde, or peracetic acid are available in reasonable purity and
aie used to represent the whole class of their homologs, many of which an.-
unstable and not readily available.
3. Tola I Oxidant Measurement Techniques
a, Co Lor imejtr ic KI
Continuous analysis of total oxidants is most commonly accomplished
with instrumentation using neutral buffered potassium iodide as the
absorbing medium. In devices using a coloriraetric method, sample air is
passed at a known rate through a liquid-absorbing reagent with a vacuum pump
arrangement. Oxidants in contact with the KI react to free iodine and the
tri-iodine ion. These are determined colorimetrically in a continuous flow
colorimeter at 352 mu. The continuous instrumental method is extremely
.sensitive to oxidizing substances and is suitable for determinations in the
3
0-05 ppm (0-1 mg 0,/m ) range, with a sensitivity of about 0.005 ppm
(.Olmg/m ) as ozone. The negative interference caused by SO,, is often reduced
by passing the air sample over CrO^ scrubbers, but this technique has not been
wholly successful, primarily because of the effects of humidity on the scrubber
9
.•>vsicn)s, oftentimes rendering them ineffective. Better methods for
eliminating SO. interference are critically needed.
5-17
-------
b. Coulometric
In these devices, which are also widely used for measurement of ;
oxidants, sample air is passed in a manner similar to the colorinetrie type
of instrument into an electrolytic detector cell containing potassium
iodide. The free iodine liberated by the oxidants is reduced at the cathode
of the cell, causing a current flow through an external circuit. The current
flow is proportional to the amount of iodine liberated and, in turn, to the
oxidants entering the solutions. The current is measured with a microammeter
3
usually calibrated directly in pphm or mg/m of ozone.
c. Coulometric vs. Colorimetric Methods
Since these general types of instruments are the most widely used
for oxidant measurements, numerous studies have been conducted comparing the
results obtained with the two approaches. Field comparisons of colorimetrie
and Coulometric oxidant analyzers Indicate differences would be no greater
than those of two colorimetric analyzers working side-by-side. Comparative
data show a correlation coefficient of 0.87 between readings of the two
different instruments, both calibrated with respect to known ozone streams. In
this study, however, this coefficient was the result of a least squares fit
and in some Instances, the agreement betw >n the two methods was poor.
12
In another field study the two methods agreed within their respective
degrees of precision after correction for NO. responses.
Analyzers utilizing other principles of detection vary !.t their
sensitivity to NO,. As shown in Table 5-1, the Coulometric cell analyzers
5-18
-------
Table 5-1. EFFECT OF NO OXIDANT DETERMINATION IN ABSENCE OF OZONE.
NO- Concentration
Range Tested
ppm* in air
> 0 to 36
1.5 to 5.5
2.0
> 0 to 1
0.24 to 0.43
2.0
2.0
Conditions
1% KI, pH 7
bubbler
2% KI, pH 7
bubbler
10% KI, pH 7
contact column
20% KI, pH 7
contact column
20% KI, pH 7
contact colurcn
coulombic cell
U. V. photometer
Response
Percent
. of N0'2
Concentration
8 to 11
6.4
21
30
12 to 47
average : 25
10
2
Reference
13
14
15
13
16
15
15
*Based on 4-liter sample,
5-19
-------
register about 10 percent of the NO. concentration as ozone and the UV
photometer device shows the least interference from NO-.
Perhaps the largest variations between the methods are obtained
because of the differences in reagent formulation. Generally, 10- or 20-
percent KI solutions are used in colorimetric instruments, and at these
concentrations, response to NO. is higher than at the 2- percent level used
with coulometric instruments. Studies comparing the two methods indicate that
both must be checked and calibrated often and that variations are usually
caused by changes occurring in air or reagent flow, and by interfering
compounds.
d. Other Methods
1) Alkaline KI
This method is intended for the manual determination of oxidants
3 3
in the range of a few pphm (mg/m ) to about 20 ppm (40mg/m ). The advantage
of this procedure over the neutral iodide technique is that a delay between
sampling and completion of analysis is allowed. Sampling is conducted in
midget impingers containing 1 percent KI in 1 N sodium hydroxide. Because of
the preferred reference neutral-buffered method, however, the alkaline
procedure is not widely used.
2) Acid KI
This procedure was reported recently to give good color stability,
no SO ? interference, and good agreement with results by neutral-brffered KI
methods. It has not yet been widely tested.
5-20
-------
3) Phenolphthalin
18
This method is based on the oxidation of the colorless phenolphthalin
to phenolphthalein, which is pink in alkaline solution by atmospheric oxidant.
Because the method is sensitive to pH and temperature variation, it is no
longer widely used.
4) Miscellaneous
The specificities of several colorimetric reagents to various oxidant
species have been evaluated. They offer the means, when needed, for
distinguishing among most of the oxidant classes to be expected in photochemical
air pollution. Methods specific for ozone are discussed in the next paragraph.
D. SPECIFIC OZONE METHODS
1. Chemiluminescence
A chemiluminescent method originally developed by Regener for use in
balloon sondes * has been adapted for use in surface measurements. A small
pump aspirates air in the dark across the surface of a fluorescent substance
(rhodamine B absorbed on silica gel) while the material is scanned with a
photomultiplier tube. An inlet trap of desiccant eliminates moisture. The
reaction is highly specific to ozone. Responses to N0_, S0~ and PAN appear
to be negligible. However, the reaction is not stoichiometric and requires a
timer to alternate flows of ambient air and clean air which has passed over
a calibrated ultraviolet ozone generator lamp.
5-21
-------
21
For several weeks the chemiluminescence method was compared with
colortmetric and coulometrlc KI analyzers In measurements on Los Angeles smog.
It generally (but with some exceptions) read lower than the colorimetric
analyzer. This very promising method is still being evaluated.
2. Rubber Cracking
The earliest technique used to identify ozone in ambient air was
22
based on the rapid cracking of stressed rubber strips. A strip of a
suitable rubber is bent double, tied with a thread, and exposed to the air
sample to be measured. Either the time until the rubber begins to crack
or the degree of cracking after a specified time is then related to the ozone
concentration. This procedure has the following advantages:
a. The equipment required is extremely simple.
b. The rubber is apparently sensitive only to ozone and not to other
oxldants.
There are several disadvantages:
a. The simplicity of the equipment is offset by the necessity of
calibrating the rubber and training the operator.
b. Continuous monitoring is extremely laborious.
3. Ultraviolet Photometry
Another method specific for ozone is based on ozone's absorption
of ultraviolet lightwaves between the lengths of 3000 and 4000 A. Renzetti
described an ultraviolet photometer using this band, and later a mercury
23 24
vapor detector was developed. '
5-22
-------
The strong mercury line at 2437A matches the peak absorption in the
ozone band very closely. It is believed that there are no serious
interferences. The Kruger photometer was available commercially for several
years. Because of electronic instability and temperature dependence, it
required frequent attention and is not in extensive use.
4. Other Methods
A variety of oxidant and ozone methods were critically evaluated. '
' Several spectrometrie methods have been reported to be specific for
9 (\ 97 9 ft 90
ozone. ' ' ' One of these involves ozone addition to l,2-di-(4-pyridyl)
28 29
ethylene. ' There is one recent reference, one in which dihydroacridine
is oxidized by ozone to acridine, another in which N-phenyl-2-naphythylamine
31
with o=dichlorobenzene gives a color with ozone.
32 33
In a gas-titration technique ' the incoming air is injected
periodically with trans-2-butane, then held in a mixing chamber. This
intermittent removal of ozone is quantitative and may be detected with a total
oxidant analyzer such as the Mast detector. There is also a galvanic ozone
34
method. None of these methods has yet been used much in air monitoring.
E. OXIDES OF NITROGEN
1. Standards
The nitrogen oxides important in air pollution are nitric oxide, NO,
4 and nitrogen dioxide, N0?. Calibration with the relatively stable NO is made
much as it is with othei gases and only presents problems in that its
35
+ oxidation at high concentrations in air is quite rapid. (The rate for
-------
oxidation of NO to N02 by oxygen Is proportional to the square of the NO
concentration.) To avoid oxidation during dilution, one technique is to
dilute it initially with nitrogen alone, then rapidly mix with air to the
desired concentration. Once diluted to near atmospheric levels, the reaction
is very slow. The fast oxidation may be used to advantage in some cases
where small amounts of gaseous N02 are needed. Gaseous NO and a slight
stoichiometric excess of air or oxygen are drawn successively into a gas-tight
syringe, held for 30 to 60 seconds to complete NO- formation, then injected
36
into the vessel where the NO, is wanted.
Preparation of accurate low concentrations of NO. starting with the
liquid is not a simple task, because at high concentration it tends to
dimerize and condense in gas lines. The system must be heated beyond the
point in the lines where adequate dilution has occurred. A precision gas
dilution system, an electrolysis system and permeation tubes have all
been used for NO, dilution. One or another of these systems is needed for
dynamic calibration. Static calibration is made with standard solutions of
nitrite ion.
2. Colorimetric Methods
39
The Griess-Ilosvay reaction as modified by Saltzman is the most
popular method for determining NO,. The NO, in acid solution in this reaction
forms a colored dye complex with an aromatic sulfonlc acid and an aromatic
amine. There are numerous variations on this basic, reaction. 'ine method is
used for both manual and automatic analysis and is quite sensitive (to 0.01
5-24
-------
ppm NO-, or better). By using an oxidation scrubber to convert it to N0«,
40 41 42
NO may also be determined. ' '
Although the original Saltzman reagent included sulfanilic acid
and ft~naphthylamine, the latter has been replaced by N-(l-naphthyl)-ethylene-
diamine hydrochloride. The medium is 14 percent (volume) acetic acid. (These
reagents are used in the tentative method adopted by the Intersociety. Committee
43
on Methods for Ambient Air Sampling and Analysis.) A similar formulation
is used in automatic analyzers, where the coupled complex may be removed on
a charcoal column and the reagent recirculated. Other variations are used.
3. Other Methods
A gas chromatographic procedure exists for the analysis of NO and
44
N0_. These contaminants possess characteristic infrared absorption patterns
and with long path length optics may be assayed spectrometrically, although the
NO absorption is rather weak. Remote-sensing correlation spectrometry,
similar to that used for S0_ determination, ' has been applied to NO
detection and may be a useful procedure when fully tested. These and other
47
remote sensing methods have been reviewed recently.
F. GASEOUS ORGANIC POLLUTANTS
1- Total Hydrocarbons
a. Flame lonization Analyzers
Originally developed as a detector for gas chromar.^graphy, me
flame ionization technique was then adapted for total hydrocarbon analysis.
In this technique the increase in ion intensity upon introduction of a sample
5-25
-------
Into a hydrogen flame is observed on an electrometer. The response is
approximately in proportion to the number of organically bound carbon atoms
in the sample. Carbon atoms bound to oxygen, nitrogen, or halogen give
reduced or no response. There is no response to nitrogen, carbon monoxide,
carbon dioxide, or water vapor, but there is an oxygen effect which can be
minimized by appropriate operating conditions. The response is rapid and,
with careful calibration, fairly sensitive (to a fraction of a ppm carbon,
as methane). Some variations in response to various hydrocarbons occur.
49
These must be accounted for in data evaluation. The instrument also
i
responds to hydrocarbon derivatives approximately according to the proportion
of carbon atoms bound to carbon or hydrogen.
In typical polluted air samples the largest hydrocarbon component
by far is methane, usually more abundant than all other hydrocarbons combined.
Methane, however, is generally considered to be virtually inert in photochemical
reactions. In effect, it serves as an inert dilutent, reducing the precision
of reactive hydrocarbon measurement. This fact has led to attempts to
measure methane and other hydrocarbons separately. In one technique a
carbon column is treated with methane until it "breaks through," i.e., no
more methane is absorbed, although other hydrocarbons are still retained.
This column, along with the flame ionizatlon analyzer, constitutes a "methane
only" analyzer. Run in parallel or in alternation with the conventional
analysis, it affords a measure of methane and total hydrocarbons ana, by
difference, the nonmethane or close approximation of the reactive fraction. ' '
5-26
-------
h. Spectroscopic. Methods
These methods are usually applied to a sample concentrated by
freeze-out or other technique. The principal problem is calibration. In
many cases hexane is used as a calibration compound to represent the whole
uydrocarbon class. In infrared methods the reading is made at about 3.4p
(the carbon-hydrogen bond stretch wavelength). This tends to give more
weight to saturated hydrocarbons rich in carbon-hydrogen bonds than to
unsaturates. If the absorbance at other infrared wavelengths can be read,
some correction can be made for certain principal components such as methane,
acetvleae and ethylene.
Nondispersive infrared instruments have similar limitations,
although they may be calibrated with other hydrocarbon mixtures. For
atmospheric analysis without freeze-out, the cell path lengths required are
. 1
very long.
2. Hy d rocar bon s by Type
a. Gas Chromatography
"GC" is almost the ideal method for atmospheric hydrocarbon
auaiysas. ' With flame ionization detection, it has sensitivity to the
ppb range; with judicious choice of columns and temperatures, almost any
ili-sii fd separation of components can be effected. Its principal drawback is
tlie luck of qualitative identification. This is usually alleviated by a
.standardized operation with both sample and reference materials, or by use
01 a qualitative detector such as an infrared or mass spectrometer to identify
i ndiv Ldua L components.
5-27
-------
For type separations with GC there arc eubtractlve columns to be
used before or after GC. Good examples are the mercuric parchlorate or silver
sulfate-sulfuric acid columns which remove unsaturates and pass paraffins.
This simplifies the total analysis problem where individual components of
different hydrocarbon types overlap or interfere in the chromatogram.
b. Spectrometric Method
Both infrared and mass spectrometric methods are capable of
considerable discrimination among hydrocarbons, but their sensitivity levels
necessitate a freeze-out or concentration step. Infrared spectra can show ,
the proportions of various olefin types, the various aliphatic carbon-
hydrogen types (primary, secondary or tertiary) and some aromatic types.
The mass spectrometer can differentiate paraffin, olefin plus naphthene, and
aromatic groups, and in restricted narrow fractions may permit analysis for
56
individual components.
c. Miscellaneous Methods
A number of methods for olefin determination by colorimetric or
coulometric techniques are available. The colorimetric reagents Include
57 58
phosphomolybdate and p-dimethylaminobinzaldehyde in sulfuric acid.
Among various methods based on addition i •" bromine, a coulometric method
based on time for bromine generation was used successfully in several
studies.
3.- Oxygenated Gaseous Organic
a. General Method
5-28
-------
Only a few specific types of oxygenated organics have been shown to
be present in the atmosphere, although others are strongly suspected. Except
60
for one formaldehyde procedure, there are no automated methods for these
contaminants, only manual (mostly colorimetrie). The contaminants definitely
identified in the atmosphere include formaldehyde, other aldehydes, and formic
acid. Ketones and alcohols are expected but not yet demonstrated. The
methods for carbonyls are based largely on condensation reactions. Sampling
of oxygenates is difficult because of water solubility and ready condensation
or adsorption in the system.
b. Bisulfite Method
This aqueous reagent forms moderately stable complexes with lower
62
molecular weight aldehydes and methyl ketones. Heavier aldehydes are too
insoluble and ketones form unstable complexes. After complex formation the
excess reagent is destroyed and the complex broken up. The released bifulfite
is analyzed by iodimetry. The method is only moderately sensitive. The
reagent is also used as a collection absorber for other methods because the
caihouyls are readily liberated by acidification of the complex.
c. Other Condensation Reagents
There are numerous reagents in this class which will react with
carbonyl compounds. They include cyanide, hydroxylamine, phenyhydrazine,
6Q
2,4-diraitropheylhydrazine, Schiff's reagent, 2-hydrazinobei.
-------
and sensitivity for formaldehyde when recent procedures are used. The MBTH
is good for aliphatic aldehydes (including formaldehyde).64'65'66'67
68
2,4-dinitropheylhydrazine is one of the more nearly general carbonyl reagents,
since it reacts with both aldehydes and ketones; however, used colorimetrically
there are shifts in wavelength with carbonyl type, and the gravimetric
procedures are lengthy and insensitive for atmospheric work.
Formaldehyde has generally been found to be the predominant carbonyl
compound found in the atmosphere. It is also water soluble and in most
derivatives is not typical of the rest of its class. For this reason it is
often desirable to use a specific procedure for formaldehyde (such as the
chromotropic acid method), then determine "total" aldehydes or carbonyls by
another method and attempt to allow for the formaldehydes found independently.
The unsaturated aldehyde acrolein has also received special
attention because, like formaldehyde, it is a known lachrymator found in
measurable concentration in amoggy air. The method in common use employs
69
4-hexylresorcinol reagent and is specific and sensitive. It is a
colorimetric procedure.
There is still neither a general method which really measures
"total" carbonyls for aldehydes completely nor a proven automatic analysis
for any of them.
G. PEROXYACYL NITRATES
!• Long Path Infrared Spectroscopy
Peroxyacyl nitrates were discovered by means of long path IR, but
characterized several ways. Several aliphatic homologs have been synthesized
5-30
-------
72
and show similar but not identical IR spectra. The acetyl and propionyl
73
compounds have been observed in the atmosphere. The butyryl and benzoyl
homologs have only been found by synthetic means, but the latter is reported
74
to be an extremely potent lachrymator. Peroxyacetyl nitrate is the most
abundant in atmospheric samples thus far analyzed and is used for calibration.
It has several characteristic absorption bands in the infrared which serve
72
to identify and measure it.
2. Gas Chromatography
Increased sensitivity for the detection of PAN is provided by gas
enromatography using an electron capture detector. The advantage of this
detector is its relative insensitivity to other compounds which might
interfere.
Short columns (9 to 18 inches by 1/8-inch diameter) packed with
carbowax or an inert substance are used for the separation. Retention time
for PAN is 1 to 2 minutes. Two ml samples containing 1 ppb of PAN give
detectable peaks. A cycling automatic GC method has been described recently.
H. AEROSOLS AND ATMOSPHERIC TURBIDITY
1. Particulate Collection Methods
a. General Properties of Photochemical Aerosols
In addition to the usual proportion of solid particulates (found
even without a photochemical reaction svstera), the participate material
observed in a photochemical aerosol is partially liquid in character. There
i.s generally a certain amount of moisture and other evaporable liquid present,
5-31
-------
which makes weight and volume estimates difficult and dependent on recent
sample history. Prolonged passage of sample air over the aerosol may
evaporate part of the collected material. The weight of a collected
photochemical aerosol depends on the relative humidity in which it has been
kept between collection and weighing. The extent to which gaseous
photochemical reaction products condense to liquid and perhaps solid
materials, or condense into solid particulate material, is very uncertain.
Thus aerosol collection, size separation, and characterization are still
very difficult. Most of the conventional methods described below are
collection methods which are suitable for solid particulates but have very
uncertain effects on liquid aerosols. The methods based on light scattering
do not affect the particles appreciably, but are hard to relate to known
particulate sizes and densities.
b. Filtration
The ordinary fibrous filters are unsuitable for airborne particulates
because the particles penetrate the filter and cannot be readily observed,
counted, or sized. The preferred filter media are those which are not
penetrated by the particles, e.g., membrane filters such as Milllpore or
Nuclepore. These are efficient collectox into which the particles penetrate
little, and are available in a range ol pore sizes suitable for collecting
given size ranges of particles. Their pressure drops are generally not
excessive. The membrane material may be selected so as to be soluble in
chosen solvents, to allow suspension or solution of the collected material
in a liquid for further assay.
5-32
-------
c. Iner t tal Co1lee t ion
This method directs the air sample stream against a solid surface
at high velocity so as to adsorb particulatea on the surface. Very small
particulates require high air velocities, since otherwise they will follow
the airstream paths and not settle on the solid surface. Impingement
usually refers to the process as carried out within a liquid; impaction is a
dry process. The cascade impactor uses several sequential impactions made
to occur at serially increasing air velocities by reducing in steps the size
of the Jets or orifices through which the air flows towards the impaction
surface. These collectors are suitable for particles down to about 1/3 to 1
micron in diameter. The Anderson Sampler'"''' and the Lundgren Impactor
are examples of this type. The Anderson device has six to eight stages in
which material is collected on flat dishes or plates. The size distribution
at each stage has been accurately determined. In the Lundgren device, at
each of four stages a slit faces a rotor covered with a collecting film or
foil. Varying rotor speed allows for variable time resolution or collection
density. These and other stage impaction devices are commercially available.
An example of centrifugal inertial collector is the Goetz Aerosol
n i
Spectrometer, a device used mainly for research work. In it the air passes
through a helical path against a rotating cone upon which particles are
deposited in a graded mass sequence. The simpler cyclone is C.XSQ a
centrifugal collector but has no moving parts, using only the tangential
5-33
-------
velocity of the air stream. It is a simple, reliable device capable of many
variations and moderate efficiency, but has been used more in industrial
exhaust control than as an atmospheric analyzer.
d. Electrostatic Precipitation
This method is highly efficient for both solid and liquid particulates
in the size range 0.01 to 10y, but is limited by the stability of the
particles at the operating temperature and the rate of gas flow. It is used
more in industrial waste control than in atmospheric analysis. For further
examination after collection the material must usually be transferred from
the apparatus to another site, such as a microscopic slide. Commercial
versions are available.
e- Thermal Precipitation
The efficiency of this technique is very high for particles below
10u, virtually 100 percent for particles below 5y, both liquid and solid.
Its limitation is principally the very slow rate of air flow and sample
collection. It is particularly advantageous in not subjecting the particles
to much force, so that they are collected without much alteration from their
82
state in the gas phase. Commercial-sized units are available.
2. In Situ Aerosol Analyzers
a. Photometric
In general, photometric methods require use of a light beam of
wavelength that is an order of magnitude of the diameter of particles to be
82
assayed. With visible light, particles of about 0.2 to around lOy are
5-34
-------
measurable. In one type of analyzer the light scattered at some selected
angle by each particle is detected and counted to give an estimate of
83
particle concentration. If a pulse height analyzer is added, the
particles may be counted in size ranges to determine the size distribution.
In another type of system, used more in smoke or plume measurement, the
attenuation of a standard light source is used to estimate the aerosol
concentration.
A promising application of light-scattering by aerosols is the
nephelometer of Charlson. In this device a photomultiplier detector at
one end of a tube is shielded by fixed annual rings from direct view of a
llashing xenon lamp at the side of the tube. Calibration is made with clean
nitrogen (or even helium) for downscale readings, a series of heavier gases
up to Freon 112 for higher molecular scattering levels, and a small fixed
spot of reflective material behind a shutter for field calibration. The
scattering intensities have been related to meteorological visibility
0 C Of.
estimates in a precise way. ' The instrument is rugged and suitable for
field use. A commercial version is under evaluation.
87
In the Lidar instrument a tightly collimated beam of light (a
Laser is used) is flashed through the atmosphere. A telescope focused down
the same path is.used to collect back-scattered light for photomultiplier
detection and display on an oscilloscope. There is a known intensity decay
curvy in clear air which is increased by scattering from any particulates.
The time to any given response point can be related to the distance. Pointed
5-35
-------
upward, for instance, the presence of an inversion layer has been shown by
discontinuities in the decay curve. The aerosol density may be estimated
from scattering intensity.
b. Condensation Nuclei Counters
If submicron aerosol particles are present in an air sample in
which the relative humidity is brought to supersaturation by adiabatic
expansion, vapor condenses on the aerosol particles to give droplets detectable
by optical methods. The expansion for any existing relative humidity may be
varied so that aerosol particles of various size ranges are selectively
effective in nucleation. The concentration of droplets under given conditions
is a measure of aerosol concentration. Commerical Instruments are
82
available.
I. BIOLOGICAL INDICATORS
1. Bacterial Response
Photochemical air pollution and some of its specific constituents
have been shown to have bacteriostatlc or bactericidal effects (in fact,
ozone used as a germicide). The effects of gaseous smog components on
i
bacteria apparently depend strongly on the medium in which the bacteria
occur and possibly on the humidity. There uave been efforts to use
bacterial responses as smog Indicators. In one example the bacteria were
•
i
exposed to the impaction of photochemical aerosols, then growth inhibitation
5-36
-------
was estimated. A more recent technique utilized luminescent bacteria. ' '
The inhibitory effects of a synthetic smog on the luminescence was measured
by a photodetector.
2. Plant Damage
Many types of plants are sensitive to photochemical air pollution.
Plant damage Is often the first indication of deteriorating air quality. Under
laboratory conditions the types of damage to selected species are relatable
to specific air pollutants. In ambient air, the damage is apt to be less
readily diagnosed and it is suspected there are phytotoxicants in smog which
have not yet been identified. The known agents include ozone, PAN and
nitrogen dioxide.
3. Sensory Irritation
Photochemically polluted air usually has an odor—often described as
pungent—somewhat akin to that of ozone itself at low concentrations. Heavily
polluted air causes some people to feel discomfort in breathing, such as
nose or throat irritation or chest pain. Probably the most universal
92
irritation, however, is eye irritation. In severe smog episodes a
large fraction of those exposed suffer eye-smarting or burning, sometimes
accompanied by lachrimation. Estimation of intensity or prevalence of
eye irritation has been used as a smog indicator. The sensation is so
subjective and variable among subjects that panels of at least five or
six persons are ordinarily averaged. This makes the measurement clumsy and
expensive as a routine measure.
5-37
-------
J. AIR QUALITY DATA
The great bulk of available air quality data goes back little more
than IS years. In the middle 1950*0, the Los Angeles County Air Pollution
Control District (LACAPCD) established the first large-scale air monitoring
93
network based upon the use of continuous automatic analyzers. This network
included potassium iodide oxldant records, Saltzman reagent dual nitric oxide
and nitrogen dioxide analyzers, ozone photometers (for several years beginning
in 1958) and, at a later date, flame ionlzation type hydrocarbon analyzers.
At its peak this network comprised 14 stations.
In 1961, the State of California Department of Public Health organized
a statewide cooperative air-monitroing network (SCAN) of 16 stations. Six
of these were equiped and operated entirely by the Department; the remaining
nine were selected stations from the existing Los Angeles County network.
The Public Health Service opened the first station of its Continuous
Air Monitoring Project (CAMP) in Cincinnati in October 1961.95 By early
1962 five additional stations were opeating in Chicago, Philadelphia, San
Francisco, New Orleans, and Washington, D.C., in cooperation with local air
pollution control agencies. The station in New Orleans was moved to St. Louis
in 1964, and in 1965 the San Francisco St. '-.ion equipment was moved to Denver.
Air monitoring in the San Francisco area was continued and expanded by the
Bay Area Air Pollution Control District.
The Public Health Service has conducted numerous other tir sampling
activities, principally as a part of abatement or research efforts. In
the last few years there have been major expansions in monitoring efforts
5-38
-------
by local, State and Federal agencies. The cities of Chicago and New York, the
Puget Sound area, the States of New Jersey, New York, and Pennsylvania have
all embarked on such efforts incorporating telemetering capabilities.
In California alone, oxidants are continuously measured at about 40 air
monitoring stations. Many other local agencies, states and even industrial
organizations and universities are engaged in programs that are almost daily
adding to the available air quality data.
Air quality as related to the photochemical air pollution problem
must be described in terms of a whole series of interrelated compounds. Some
of these are of direct interest because of their effects on humans, animals,
plants, materials, or visibility, and some because they are primary reactants
which form substances responsible for the aforementioned effects. Several
art- or can be important from both points of view.
It is necessary then to present information on atmospheric concen-
trations for at least the most significant photochemical reactants and products.
Of those compounds discussed earlier in this chapter, data are presented on
total oxidant, ozone, nitrogen oxides, hydrocarbons, and PAN.
Even though substantial quantities of data are being collected, much
of it is still relatively inaccessible. Substantial differences exist in
reporting parameters, methods of data processing, computer software and hardware
(where used) and hard copy output. In many cases collect^ ig organizations
are not equipped to provide substantial outside access to their data.
Most of the above problems are the result of rapid growth, changing
demands on daia format, and advancing technology, rather than lack of concern
5-39
-------
about the problems. National data banks have been proposed such as the Storage
and Retrieval of Air Quality Data (SAROAD) system97 of the Public Health
Service. No single source exists, however, for even a major portion of the
data collected nationally.
The contaminant data presented in this report are from the continuous
air monitoring program (CAMP) stations and the New York-New Jersey Abatement
project of the U.S. Public Health Service; several stations in California
(two from Los Angeles area, one each from San Francisco and San Diego, one
smaller coastal city, and one smaller Inland city; Bayonna, New Jersey
(State of New Jersey Network); New York City; Phoenix, Ariiona; and, In the
case of hydrocarbons and PAN, from special sampling activities of the
USPHS, State of California, Los Angeles County APOD, and the University of
California at Riverside.
1. Methods of Expressing Air Quality
The initial form of the basic contaminant data is dependent on the
measuring system. This ranges from discrete samples taken over a given time
interval to the record produced by a rapid response continuous analyser.
In either instance only a sample of the total atmosphere is available:
and in the second, our continuous record still represents a single point
i
in space. ' y
['
The problem in selection'Jof data reporting parameters i* to provide
sufficient information to enable
complete system and to Judge the
o£ inferences about the real and
Ce» against known criteria or standards
5-40
-------
relating to its effects. Figure 5-2 provides an example of the sampling results
that might be obtained from the same system by three different approaches.
Although the inference about the time of peak would be offset in the case
of the grab samples, it is evident that 4-hour mean calculated from all three
approaches would be very close.
Still other problems surround the task of relating distinct episodes
to mean values over longer time periods. Here we first encounter "descriptive"
statistics, or those used to determine position and dispersion. These
statistics are useful in making comparisons among locations, seasons or
years, and in describing the past or expected frequency of pollution events.
A few of these common statistics are defined below.
The first is the arithmetic mean. This value is described as the
t>um of thf observations divided by the total number of observations. If
we let the n observations of some value x be denoted as x, ,x,,,x-,.. .x , and
1 / j n
the mean as x, then we can write the formula
win-re 2__, it- the usual symbol for the sum of the values. A useful statistic
2
describing the dispersion of these individual values is the variance, s ,
or the standard deviation, s. The variance is defined as the sum 01 -vc
.squares of the deviations from the mean, divided by one less than the number
of observations. The formula expressing this value is
n-1
5-41
-------
o
§
a.
o.
3
S
i
u.
o
Ul
oo
.8
!•«.
O
a
i—*
DC
2
o
evj
i
in
(Wdd) '3NOO
5-A2
-------
In some cases Che distribution of a large number of values about
the arithmetic mean is symmetrical. In these cases the distribution is described
as "normal." In many situations in nature the distribution of the logarithms
is normal. In these cases interest lies in the geometric mean illustrated
somewhat indirectly in the formula below:
n
log (geometric mean) - ]P log x
where the logs are distributed normally, we have a "lognormal" distribution.
To illulstrate the effect of averaging, Figure 5-3 includes an hourly
plot of the average oxidant values for 2 different days in a month, along
with the mean monthly trace of these values. All are known as diurnal plots.
The mean monthly curve is smoother than either of the other 2 days shown.
Although differences in data processing procedures and formats
exist among the organizations collecting air quality data, there are several
basic reporting parameters used by most for data acquired by continuous
analyzers. These would include (1) peak concentration for each day (highest
instantaneous value reached), (2) hourly average concentrations by each hour
of the day, and (3) the maximum hourly concentration for each day. From these,
»
basic day mean values over various time periods may be calculated and
frequency distributions of the individual values or the means compiled.
Most organizations determine hourly average values by a simple
graphical integration procedure using the recorder charts. In the U.S.
Public Health Service CAMP network, the use of analog-to-digital converters
5-43
-------
.50
.40
1 E-30
a.
o
.20
.10
SEPTEMBER 22, 1966, THURSDAY,
SEPTEMBER 11, 1966, SUNDAY,
MEAN MONTHLY TRACE.
00 03
06 09 12N 15
TIME OF DAY
18
FIGURE 5-3. DIURNAL OXIDANT CONCENTRATIONS,
POMONA, CALIFORNIA, SEPTEMBER 1966.
21
00
5-44
-------
also provide machine-calculated or/ -machine determined hourly averages.
Most of the data presented in the remaining portion of this chapter
deals with hourly averages as the basic unit of data. Those cases in which
i
instantaneous values are used are clearly defined.
2. Typical Concentration Patterns in Photochemical Air Contaminants
In preparing an environmental appraisal of the photochemical air
pollution problem, we must be concerned not only with specific time-concentration-
frequency values for the individual contaminants for which effects criteria
are available, but also with the interrelated temporal patterns exhibited
by these substances. These patterns give form to the data and provide a
preliminary insight into the character of the phenomena, i.e., possible
differences resulting from location, from meteorological influences, and from
emissions of air contaminants. More rigorous mathematical analyses and simulation
modeling may extend or confirm this preliminary knowledge, but the visual
impact of typical pollution patterns is an important first step in under-
.sLandiug the problem.
It is for this reason that most of the information on diurnal and
season.-jJ variations in photochemical reactants and products is presented
in the following subparagraphs. Furthermore, most information of this type
is in graphical form. Hydrocarbon data are not generally included here
because most such data from continuous analyzers provides cuiy carbon atom
concentrations with no separation of natural methane. Thus the patterns
of reactive hydrocarbons are somewhat indistinct. Hydrocarbon data are discussed
in section J4c.
5-45
-------
a. Diurnal Variations
Inasmuch as air pollution is observed initially by incidents
or episodes, Figure 5-4 is presented to show the diurnal variation of oxidant,
nitric oxide, and nitrogen dioxide for a single day in the urban center of
Los Angeles. The time of year is during the period that frequent photochemical
air pollution episodes occur.
Before daylight, the primary contaminants NO, hydrocarbons, and
CO remain fairly stable at concentrations somewhat higher than the daily
minimums, which usually occur during daylight hours at times of maximum
ventilation and mixing. N0_ also fits the former pattern. Ozone is near
zero. As human and traffic activity increase in the hours Just after dawn,
primary contaminants increase. As ultraviolet energy from the sun becomes
available, nitric oxide peaks and nitrogen dioxide increases, often with a
nearly complete conversion occurring. Finally, as the nitric oxide concentration
(and thus the nitric oxide to nitrogen dioxide ratio) reaches very low levels,
ozone begins to accumulate, reaching a peak about midday.
Before the ozone peak is reached, the primary contaminants and N0»
normally have been declining for some time. Although not shown in this
figure, other reaction products such as ^AN peak at a time more closely
corresponding with ozone than with the primary contaminants. In the usual
situation, all contaminants decrease rather rapidly as the ventilation
capacity of the atmosphere increases. This results from surface heatit:j, which
causes increased instability of the atmosphere, and from rising winds.
5-46
-------
0.50 i
0.40 -\
§
ZE
Q.
3
O
CJ
00
o
*
CXI
o
o
TIME OF DAY
FIGURE 5-4. CONCENTRATIONS OF POLLUTANTS—LOS ANGELES, CALIFORNIA.
5-47
-------
As solar intensity decreases, stability increases and winds decrease.
The late afternoon and evening traffic increase results in increased concen-
trations of primary contaminants, but solar energy is no longer available to
pause the photo-oxidation of nitric oxide and the production of ozone and
other products.
It is obvious, but necessary, to point out that the concentration
patterns shown in Figure 5-4 are for a single day at a single loc&irion.
There are many factors which could result in absolute and relative levels of
the concentrations and in their time distribution during the day. Of particular
importance are (1) variations in time and nature of emissions of photochemical
product-forming contamlnatns, (2) variation in available solar irradiation and
other photophysical variables, and (3) variations in transport and diffusion
of contaminants in the atmosphere. All of the above influences may vary over
very short time periods (hours), while the latter two categories also
exhibit significant seasonal variations.
In general, the time of year having highest concentrations of
oxidant and other photochemical products is between early summer and fall.
If the zenith angle of the sun were the only consideration, the time of the
summer solstice would produce the higher concentrations. However, periods
of maximum atmospheric stability and lew horizontal transport do not coincide
in all locations of the country, thus producing a range in time of seasonal
highs. The mean daily time of oxidant peak is near noon, although even hire,
cloud cover, thermal stability and the advection of pollutants may result in much
later peaks or even in multiple peaks (commonly, two).
5-48
-------
Figure 5-5 shows average diurnal traces for oxidant and N02 for
Bayonne, New Jersey; Philadelphia, Pennsylvania; and St. Louis, Missouri.
A peak oxidant month from the available data for each location was selected
for this illustration. In each case the familiar pattern is followed: an
early NO peak, followed by an oxidant maximum near midday, and finally a
late afternoon and evening rise in NC^.
Figure 5-6 presents the diurnal variation in oxidant for Philadelphia
on August 6, 1966 during a 3-day period (August 6-8) for which unusually high
concentrations were observed. This illustrates the situation in which acute
episodes in differing geographical locations can be quite similar, although
the means show distinct differences. In fact, caution must be used in
interpreting mean diurnal curves because statistical smoothing results in
.i trace which may not accurately represent any one single day. This is
particularly true when comparisons with other contaminants are made. The
ratio of means for any given period could likely be different from the
moan of the individual ratios.
The specific time at which the oxidant value reaches its peak
deponds not only upon meteorological parameters governing accumulation and
dispersion of the contaminants reacting to form this group of compounds
(mainly ozone), but upon the spatial relationship of the point of measurement
with the source area. To illustrate the effect of transport a..id souxx-- area,
Figure 5-7 shows the monthly mean hourly average concentrations for October,
1965 for West Los Angeles, Los Angeles (center city), Azusa, and Riverside,
California.
5-49
-------
(BAYONNE, PHILADELPHIA, AND ST. LOUIS
MONTHLY MEAN OF HOURLY AVERAGE
CONCENTRATIONS.)
.20-
.18-
,16
.14-
.12-
.10
.08
.06H
.04
.02
FIGURE 5-5a. OXIDANT
ST.LOUIS, JUNE 1966
BAYONNE, JULY 1966
*xl — S ^PHILADELPHIA, AUGUST 1966
00
03 06 09 12N 15 18 21
00
.20-
~ .18-
I -16-
- .14-
J .12-
I .10-
t .08-
§ .06-
2 .04-
X
o n:
FIGURE 5-5b. NO,
.BAYONNE, NOVEMBER 1968
.PHILADELPHIA, AUGUST 1966
xrnu.Auci.rn in
ST.LOUIS, JUNE 1%6
00 03 06 09 NOON 15 18 21 00
TIME OF DAY
FIGURE 5-5. DIURNAL VARIATION IN OXIDANT AND NO-, CONCENTRATIONS.
5-50
-------
CNJ
OO
a I
o:
•s
o <:
u. >-
o oo
z
z z
O LU
o: 1-1
< T:
> o-
O Q.
VO
I
UJ
a:
PO
o
O
CO
o
CM
O O
(Wdd) 'DNOD 1NVQIXO
5-51
-------
0.120
0,0»0
0040
0
0.160
0.120
E
«. o.oao
^ 0.040
1 °
ft 0.160
8 0.120
0.080
0.040
0
0.120
O.000
0.040
0
/ X
WEST / X
- IDS ANGELES / \
""""%""" ^Y^^**" i j i ii ' i r i ""**
DOWNTOWN / \
- LOS ANGELES / \
„_ _/ X
1 1 "™"l III 1 1 1 1 1
- AZUSA / \
i i i i i i ii iii
t -A
- RIVERSIDE /"^ . \
r-; j-^, ,,,-,,, ""*,
2 4 6 0 10 12 2 4 6 8 10 I!
« A M k.
-------
The station at West Los Angeles is about 10 miles west, Azusa about
20 miles east, and Riverside about 60 miles southeast of the central Los Angeles
station. As shown, the time of peak oxidant concentration progressively
becomes later from west to east. This coincides with the daytime ocean-to-
land movement of air regularly observed in California coastal regions from
spring to fall months. There is indication of another but earlier peas in
Riverside which is generally attributed to local pollution sources.
The supposition that the rather high oxidant concentrations in less
densely populated Azusa and Riverside are at least in part due to pollution
transport is given strength by the carbon monoxide traces for the same areas
as shown in Figure 5-8. The times of peak CO values are much the same for
all four locations, but the concentrations are lower in Azusa and Riverside
as compared to West and central Los Angeles. The conclusion reached by confirm-
ing inspection of these figures with Information gained from Chamber studies
is that under the conditions present, the rate of oxidant formation in the air
mass moving from west to east over the Los Angeles basin and inland valleys
exceeded the dilution effect.
In almost all metropolitan areas of the United States outside of
California, oxidant data are presently available from only one,—or at most
a few—locations. However, particular combinations of source strength variations
in hydrocarbons and nitrogen oxides and in patterns of air PASS ttti^^ort
may contribute to the same type of contaminant concentration and time
variation as just shown for the Los Angeles area. Thus we should not be
surprised at any deviations from present patterns as information from augmented
air monitoring networks becomes available.
5-53
-------
2*0"
20.0
16.0
12.0
/\ WEST x^.
/ \ LOS ANGELES X X
^ .^" *x ^
• nl l i it til i l l l j
i
«
ff
20.0 -
camuu.
LOS ANGELES
16.0
12.0
ft
.. — — -^ ^— ^^^^"^ — •-"
-
1 1 1 1 1 1 1 1 1
1 1 1
I
0
u
12.0
e.o
4.0
o
AZUSA
__— _ ^.•-p*~"— """»—.»
-
.
l i l l i i l l J I l l
12.0
8.0
4.0
ft
R
*^^ ^^.
- ^^ X
l i i l t i
RIVERSIDE
i I I I ' '
A.M.
8 10 12 i 4
HOUR OF DAY, PST
.
6 8
P.M.
10 12
Figure 5-8. CARBON MONOXIDE MONTHLY MEAN HOURLY
AVERAGE CONCENTRATIONS—OCTOBER 1965.
5-54
-------
b. Seasonal Variations
The seasonal variation by month for the major contaminants involved
in or resulting from atmospheric photochemical reactions depends upon whether
the contaminant is consumed or produced on a net basis during daylight hours.
For nitric oxide and hydrocarbons, as with most primary contaminants,
higher aiean values are observed during late fall and winter months. IViere
is la,;y overall atmospheric mixing during these months in most cases and less
consumption to form products.
Total oxidant, as indicated previously, typically exhibits a seasonal
peak during the period from early summer to early fall. This is, of course,
the time of year when more intense short wavelength solar irradiation
reaches the earth's surface. On the west coast of the United States a
persistent late summer and early fall subsidence temperature inversion and
less cloud cover enhance the potential for high oxidant levels.
The pattern for nitrogen dioxide is not quite so distinct. The more
efficient photochemical oxidation of nitric oxide in the summer is somewhat
offset by the slower nitrogen dioxide disappearance rate in the winter.
Thus varied and somewhat differing patterns are observed from city to city.
Figure 5-9 shows the seasonal variations by month in average total
oxidant concentrations for the cities of Los Angeles, Denver, Phoenix and
Bayonne. The month-by-month variation in the maximum hourly mean it, . "iown
for the same cities in Figure 5™lu. In all locations the overall mean and
maximum hourly mean concentrations are higher during the summer months, although
the peak period for Los Angeles is somewhat later. A number of factors includ-
5-55
-------
r<=>
•O
O» 00
9) C^
UJ *
IS
"3 00
VO O
2^
I
LD QC oe
ys *^ '*t
r- CO Z 2
en x uj
>- p£i-i Z
UJ Z Z
• >UI O
< zo >•
•Ul 3C <
_l O SZ OQ
"I
III
• rs
i
OQ
•—i
1
O
X
O
to
t—t
u.
00
CNJ
§
Wdd NOI1VU1N30NOO
5-56
-------
-Q
Of.
=>
o
-o
-co
o
o
•z:
o
o
Si
>—«
X
o
o
I
in
00
-------
ing leas cloud cover, lower wind speeds and stronger inversions account in
part for this difference.
Figure 5-11 shows the seasonal patterns in nitric oxide by present-
ing the mean values by month of year during the time period indicated for the
cities of Chicago, Denver, Bayonne, and Los Angeles. The distinctly higher
winter time levels are evident.
The seasonal variation in nitrogen dioxide shown by monthly mean
values for the same cities is shown in Figure 5-12. Here the pattern is less
distinct and varies from location to location.
i
3. Ozone vs. Oxidant Measurements
Almost from the beginning of concern about photochemical air pollution,
questions have been raised about the true ozone concentrations in the atmosphere,
and the extent to which this compound comprises so-called total oxidant values.
Numbers of methods have been proposed which either theoretically or empirically
are claimed to give results specified for ozone. These are treated more fully
in paragraph D. of this chapter.
The principal problems associated with measurement of ozone with an
oxidation-reduction reaction such as that used in the potassium iodide total
oxidant method are positive interference fr.m nitrogen dioxide and negative
interference from sulfur dioxide. In the case of nitrogen dioxide, the positive
Interference is variable depending upon specific instrument configuration and
potassium iodide concentration used. ' ' ' For the color1'^.eric
type analyzers used in the CAMP network, and many state and local systems,
a reasonable estimate is that nitrogen dioxide gives a response equal to
5-58
-------
u>
VO
<*
ss
\
\
/
o
. LU
Q
>
•O
• O
o
Q.
• uu
1/1
\\i
\\:
/ I
/ i
/ /
' /
\
\|
/
/
\
\
\
•3
<
(X
<
Ul
_J >-
- fe
3:
5 S
-Z> O
•-a ae
s
UJ
o
o
i
in
oe
(Udd) NOIi\/aiN33NOD ON
5-59
-------
vD
«O
e>
i— •
1
*>
rv.
vo
•
co
UJ
UJ
o
CO
o
t
>
/
/
/
/
<— o 55
r— r— O
• • •
(Wdd)
/ h-
/ 1 \ .
/ | \ to
/ ' / £
/ / /
/ / / OL
* £ M LlJ
< / ' I '
>--J \
/a
/ /
/ / 1
'
H \ 1 \
' * \
I\ I \
%
NJ « \
\ / 2 /
\ /! tf /
V ' ' i /
Xv X / 2 f
X »' <
x Aw ?
^Ns^
\ ! ^
» • •
\ 1 /
•A
-'-"'! V
I y
l- r\
\ \
\ \
0
UJ
o
5»
g
§
0.
UJ
co
§3
s
• ri ul
0 O
-c
K-
• 5 §
o
>-
• <
oc
• O-
f.
Oi
$
•£9
u.
5
oor^^Lrt^ocvj^-
oo ooooo o
» * ••*•••
NOIlVblN3DN03 ZQH dV3A JO H1NON
to
H- 1
»—
S
£
UJ
<_)
S
o
c<
o
>-
_J
I
2-
£
§
z
cj
LA
UJ
g
CO
•— 1
u.
5-60
-------
20 percent of its concentration.
Other positive interferences are possible (such as from PAN), but
under normal circumstances are of minor concern. It is true, however, that
in some locations oxidant concentrations found are higher than would be
expected for the time of day or year. An example is the St. Louis CAMP
station, which is located near a large chemical complex. Occasional hJgh
oxidant concentrations, including one at 0.85 ppm, do not coincide with the
usual diurnal end seasonal pattern. The National Air Pollution Control
Administration has found it necessary to exclude high levels occurring
before 11:00 A.M. and after 4:55 P.M. in listing maximum concentrations for
its CAMP stations.
A potentially more serious interference in the measurement of ozone
by the potassium iodide method is the negative response caused by sulfur dioxide.
It acts in the reagent by reducing the iodine to the colorless iodide form.
The negative response is equal to the sulfur dioxide concentration, up to
the limit of the total oxidant originally present. As an example, the response
of an oxidant analyzer to a stream containing 0.5 ppm ozone and 0.5 ppni sulfur
dioxide would be zero. The response to a strema containing 0.25 ppm ozone and
0.5 ppm sulfur dioxide would also be zero. This is important to remember
when considering some of the adjusted oxidant data presented in the next
few pages.
Of all the approaches to obtaining true ozone concentrations only
a few have been used extensively enough to provide much data. These would
15 98
ii.clude the ultraviolet photometric and the chemiluminescent methods.
5-61
-------
An approach to eliminating the sulfur dioxide but not nitrogen dioxide
interference is the use of chromium trioxide scrubbers which remove sulfur
dioxide. These were incorporated in all CAMP stations except San Francisco
9
at the beginning of 1964. They are not used in the Los Angeles County network.
One slight disadvantage particularly in areas with little sulfur dioxide
results from the conversion by the scrubbers of nitric oxide to nitrogen
dioxide, thus increasing the positive interference.
Efforts have also been made to correct the total oxidant concentrat-
ions by adjustment of the data to remove the effects of nitrogen dioxide and
sulfur dioxide. This assumes that such data are available. It also is done
on a fixed and arbitrary basis, and thus may not represent reality. The
concentrations of oxidant for a summer and a winter month for each of four
stations are adjusted for nitrogen dioxide in Tables 5-2 and 5-3.
Table 5-2 lists the monthly mean concentrations, while Table 5-3 lists
the monthly means at daily maximum hourly average concentrations. The adjust-
ment was performed by subtracting a sum equal to one-fifth of the nitrogen
concentration for the same period from each total oxidant concentration.
As mentioned earlier the nitrogen dioxide interference is known to be
variable, particularly from analyzer to analyzer, so the correction cannot
be assumed to be perfect in any sense. There is a further complication in
that the nitrogen dioxide analyzer tends to "smooth" the true concentrations
more than does the oxidant analyzer because it responds more slowly. Since
hourly averages are calculated over fixed time periods, there is more likeli-
hood that a nitrogen dioxide concentration occuriing in the atmosphere at a
5-62
-------
Table 5-2. OXIDANT CONCENTRATIONS ADJUSTED FOR NITROGEN DIOXIDE RESPONSE
MONTHLY MEANS OF HOURLY AVERAGE CONCENTRATIONS, ppm
Station
Month
Concentration, ppm Percent Due
Unadjusted Adjusted
Summer
Months
Winter
Months
Lo& Angel ns
Sacramento
Denver
St. Louis
Los Angeles
Sacramento
Denver
St. Louis
July 1964
July 1965
July 1965
July 1964
Jan 1965
Jan 1965
Feb 1965*
Jan 1965
0.055
0.042
0.048
0.035
0.021
O.G20
0.029
0.024
0.044
0.036
0.041
0.029
0.005
0.015
0.019
0.020
20
14
15
17
76
25
35
17
* 12 days of data only
5-63
-------
Table 5-3. OXIDANT CONCENTRATION ADJUSTED FOR NITROGEN DIOXIDE RESPONSE
MONTHLY MEANS OF DAILY MAXIMUM HOURLY AVERAGE CONCENTRATIONS
Summer
Months
Winter
Months
Station
Los Angeles
Sacramento
Denver
St. Louis
Los Angeles**
Sacramento
Denver
St. Louis
Month
July 1964
July 1965
July 1965
July 1964
Jan 1965
Jan 1965
Feb 1965*
Jan 1965
Concentration, ppm Percent Due
Unadjusted Adjusted to NO,
1
0.166
0.085
0.110
0.078
0.037
0.035
0.073
0.046
0.154
0.075
0.098
0.071
0.020
0.028
0.060
0.040
7
12
11
9
46
20
18
13
* 12 days of data only
** 9 days of data when NC>2 measured at time of maximum hourly average oxidant
concentration
5-64
-------
particular time period will be partially reflected at a later instrument
time than will the reading on the oxidant analyzer. This effect would tend to
disappear when averaging results over a month as in Tables 5-2 and 5-3,
but could be of consequence in individual values.
Nevertheless, the tables show that the adjustment is generally
greater during the specified winter months than during the specified summer
months. They further show that adjustment is less for maximum hourly averages
than for overall averages. It is also interesting to note that greatest
change in percent effect from winter to summer is for Los Angeles. The absolute
adjustment in all cases varies little from .01 ppm.
For several years the Los Angeles County Air Pollution Control
District operated ultraviolet photometers as an approach to specific ozone
measurement. In most cases they were operated in parallel with potassium
iodide total oxidant analyzers. Although there were substantial operational
and maintenance problems, a substantial quantity of what is believed to be
valid data were produced.
In Figure 5-13, the monthly means of the hourly average concentrations
of ozone are compared to concentrations of oxidants in Pasadena and in Los Angeles.
In Figure 5-14, the monthly means of the maximum hourly averages are compared.
In Figure 5-15 hour-by-hour values for ozone and oxidant concentrations for
Los Angeles and Pasadena are shown for the month of July, 1%4.
The term "adjusted" oxidant in the three foregoing figures is calcu-
lated from the following expression
Adjusted oxidant - OX-1/5NO +SO
5-65
-------
z z
Z3 i—
O
Z<
U.Z
O LU
O
^ <:
>• Q
_J Z
co
UJ
H- Cfl
i-l CO
X O
O-J
O I
z co
<£. Z
LU •—<
ZH-
O
-------
z
o
<£ UJ
O o
•-« z
X O
O o
o
5-67
-------
'V
o
£'
a r~
UJ >-
i— _i
t/l 3
o •
I— Llj
z o
O I/O
<£ UJ
Q CJ
CO
•o
LU _J
Z I
O i
M LU
O LO
oo
Z UJ
O QC
1/1
»-t C
CC O
-------
These quantities are, respectively, the measured oxidant, nitrogen dioxide
and sulfur dioxide concentrations. There is a possibility of an overcorrection
for sulfur dioxide if it exceeds the real ozone concentration. However, this
is very unlikely in daylight hours for the locations used because of the
routinely low sulfur dioxide concentrations experienced.
The close agreement between adjusted oxidant and ozoa concentrations
indicated that the contribution to the oxidant measurement by PAN or other
oxidants is very low at the locations involved. This is as expected as
far as photochemical sources of these other oxidants is concerned. Agreement
between the unadjusted oxidant and ozone is also good, but this may be partly
becuase interferences by N0_ and SCL compensated for one another.
In a recent study made by the Research Triangle Institute under
contract with the U.S. Public Health Service, the chemiluminescent method for
99 IQQ
ozone was used in several cities. ' The mean hourly ozone concentrations
for each of these cities during the period of sampling are shown in Figure 5-16.
The diurnal patterns on selected days for the Denver and Philadelphia CAMP
sites are shown in Figure 5-17. The physical appearance of these curves
tends to indicate a high degree of similarity to typical total oxidant traces.
The data so far available would seem to indicate that, by using caution,
total oxidant data can be used to judge air quality effects related both to
those based upon ozone and oxidant, there are situations where p^irive
interference from sulfur dicxl.1 - .v,J rcsii^.3 interference from other oxidants
eouJd cause difficulties in data interpretation. For this reason it would be
desirable to continue both laboratroy and field studies on more specific methods
ior ozone.
5-69
-------
I I
f-s. !*•*. • — » f*-**
• to to r**. 10
0 -x ">% <£> ^
O r— \ •—
5E r— 1 >~> O\ UJ r—
O 1 ^ >— J —II
^f~* 00 VjO *^C r*1^ i-i-iJ r^v
Ol^->^«a=<^'t3t£> l>l
Z -^. CXt*°- Z^ Z«^ 1 »
HI || Ld i— i— i IT) < CM I
OOOZOZ^.tOr— 1 L
3ec. Sc. G2. 3~ ^ i
/ '*
-&
— -— fc^
^ — --* — ^^y^
^-*^ >" yA
x^" / <[
~ o
§ °
UJ
U. Z
0 0
.co uj o
VO
in
CM"— o2Jcor~vom'*cocoi—
,-•— r=-ooooooooo
(Hdd) bO -
*
*
5-70
-------
w
w
o
w
o
w
H
w
u
z
o
CJ
w
a
o
N
O
fa
O
S
a
r^
i—(
I
3
00
•H
5-71
-------
4. Observed Contaminant Concentrations
a. Oxidant
Information in the previous subparagraphs has shown that the shape
of the diurnal and seasonal oxidant concentrations are generally similar
over a broad geographic spectrum of locations. However, on the basis of
presently available data, the frequency of occurrence and the severity of ozone
and other oxidant episodes varies considerably among even the larger metropolitan
areas.
Table 5-4 gives the cumulative distribution of hourly mean oxidant
levels for the six CAMP cities and five locations in California for the period
1964-1965. One fact immediately apparent is that the yearly average concen-
trations vary over a much smaller range than does the value which is exceeded
1 percent of the time at each location. In fact, the range over which the
yearly averages vary is little different from that observed in remote, sparsely
populated areas (see Chapter 2). A prime reason for this apparent anomaly
is that the yearly averages are unduly weighted by nighttime values. In
locations having substantial photochemical air pollutions problems, ozone-
scavenging compounds (principally NO) are injected into the atmosphere with
the late afternoon traffic with t. result that nighttime ozone values are
often very low.
Table 5-5 shows, for the same locations as in Table 5-4, the number
of days with maximum hourly oxidant averages exceeding the. levels 0.15, 0.10 and
0.05 ppm. The number of days with at least one hourly average above 0.15 ppm
oxidant is an order of magnitude greater for Pasadena and Los Angeles as
I
5-72
-------
Table 5-4. CUMULATIVE HOURLY AVERAGE OXIDANT CONCENTRATIONS (1964-J965).
City
Pasadena
Los Angeles
San Diego
Denver*
St. Louis
Philadelphia
Sacramento
Cincinnati
Santa Barbara
Washington, D.C.
San Francisco
Chicago
Percent of Hours with Concentrations
Equal to or Greater than Stated
Concentrations , ppm
90 70
.01
.01
.01
.01
.01
.01
.01
.01
.02
.01
.01
.01
.01
.01
.02
.02
.02
.02
.01
.02
.02
.01
.01
.01
50
.02
.02
.03
.03
.03
.02
.02
.02
.03
.02
.02
.02
30
.04
.04
.04
.04
.04
.03
.04
.04
.04
.03
.03
.03
10
.12
.10
.08
.06
.06
.06
.06
.06
.06
.06
.04
.05
5
.18
.14
.10
.08
.07
.08
.08
.07
.08
.07
.05
.06
2
.23
.18
.12
.10
.09
.11
.10
.08
.09
.09
.06
.08
1
.26
.22
.14
.12
.11
.14
.12
.10
.10
.10
.07
.08
Yearly
Average
(1964-1965)
0.042
.036
.036
.036
.031
.026
.030
.030
.036
.029
.019
.028
* Eleven months of data beginning February, 1965
Source: See Reference 38 and 39
5-73
-------
Table 5-5. OXIDANT CONCENTRATIONS (1964-1965).
Station
Pasadena
Los Angeles
San Diego
Denver*
St. Louis
Philadelphia
Sacramento
Cincinnati
Santa Barbara
Washington, D.C.
San Francisco
Chicago
Days with Maximum Hourly Average Equal to Maximum Peak
or Greater Than Concentration Specified Hour Average Concentrations
0.15 ppm 0.10 ppm
Days Percent
299 41.1
220 30.1
35 5.6
14 4.9
14 2.4
13 2.3
16 2.3
10 1.6
11 1.5
7 1.2
6 .9
0 0
Days Percent
401 55.1
354 48.5
130 20.9
51 17.9
59 10.1
60 10.9
104 14.6
55 9.0
76 10.5
65 11.3
29 4.5
24 4.5
0.05 ppm
Days Percent
546 75.0
540 74.0
440 70.6
226 79.3
362 62.2
233 41.9
443 62.3
319 52.0
510 70.5
313 54.2
185 28.6
269 50.8
(ppm)
0.46
0.58
0.38
0.25
0.35
0.21
0.26
0.26
0.25
0.21
0.18
0.13
(ppm)
0.67
0.65
0.56
0.31
0.85
0.25
0.45
0.32
0.28
0.24
0.22
0.19
* Eleven months of data beginning February, 1965
5-74
-------
compared to the other ten cities. The unusually high peak value of 0.85 ppm
for St. Louis has been referred to earlier. This statibn is located near a
large chemical complex, with the distinct possibility that this'high concentration
—as well as others—may be due to interferring emissions from some unidentified
source.
Other oxidant information is presented for the same twelve cities in
Table 5-6. The month having the highest mean oxidant concentration, the overall
mean for the month, and the mean of the maximum hourly concentrations are shown.
Data for the CAMP locations for the years 1966-1967 plus several
locations reporting to the USPHS New York-New Jersey Abatement Project for
1967 or 1968 are shown in Table 5-7. This information is in the same form
as Table 5-4, showing number of days with at least one hourly average equal
to or exceeding the state levels. While some year-to-year variation is
Indicated for the CAMP stations, it is not ususual. The three locations in
the New York Metropolitan Area show oxidant levels that fit within the range
of the CAMP cities.
Data for a number of contaminants were recently assembled by the
California Air Resources Board prior to Hearings on State Air Quality Standards.
This included a report by month and year for 1967 on the number of days the
maximum hourly oxidant concentration at each of 39 stations exceeded 0.10 ppm.
Every station in the state except Berkeley, Richmond, and San Franci'
exhibited maximum hourly averages above this level for at least 5 percent of
the days. In many locations in Los Angeles County this level is exceeded
on more than 50 percent of the days. One or more stations in Orange, Riverside,
5-75
-------
Table 5-6. HIGHEST MONTHLY MEAN OXIDANT CONCENTRATIONS (1964-1965).
Station
Highest Month
(all hours)
Mean of Hourly
Average Concentrations
(ppra)
Mean of Maximum
Hourly Concentrations
(ppmi
Pasadena
Los Angeles
San Diego
Denver*
St. Louis
Philadelphia
Sacramento
Cincinnati
S.mta Barbara
Washington, D.C.
San Francisco
Chicago
July
August
October
July*
May
July
June
July
May
May
April
0.075
0.056
0.050
0.050
0.042
0.054
0.040
0.048
0.042**
0.041
0.031
0.044
0.24
0.17
0.11
0.11
0.072
0.11
0.075
0.098
0.064 and .072
0.072
0.046
0.070
* Eleven months of data beginning February, 1965
** 1964-1965 average for month of May the same as for September
5-76
-------
00
vO
ON \O OO O\ rH cs) ul f>*O vO vO 00 tX) O
•H *H .H «N * CM »-t CM ,-4 •-* CS
oo od do do do oo o o o
sa
A*
IA
0) fH
00 •
n o
i*
I
ifN CN CO moo t^-r^ O
-.-4 m CN roro O
lA rH *O CN 00 t-.
m O o
00 CO r^.
ON ON ON
i
ai
c
o
u
•!»
00
5-77
-------
San Bernardino, and San Diego Counties exceeded 0.1 ppm for more than 100
days during 1967.
b. Nitrogen Oxides
Air quality data for these contaminants are limited to nitrogen
dioxide (NO-) and nitric oxide (NO). The method used, whether manually or
instrumentally performed, is based upon the formation of a colored, organic
reaction product of NO-. Most data are from continuous analyzers which deter-
mine both NO. and NO on the same instrument, the NO being determined as NO
* 2
after passing through an oxidation unit. There are several flow schemes in
102
use which may produce information not strictly comparable for NO. These
schemes are diagrammed below.
1) Series Flow;
Air
Stream
N02
Column
NO
Converter
NO
Column
2) Parallel (Mod. 1);
NO
>"
2
Column .
NO
Converter
N0+N02
Column
5-78
-------
3) Parallel (Mod. 2);
Column
Scrubber
NO
Converter
NO
Column
OnLy the first two schemes are used to any extent, with the series approach
predominating. In scheme 2) the NO must be determined by difference.
Of the two compounds, NO is the most concern in ragard to direct
effects. However, by far the greatest portion of nitrogen oxides emissions
is in the form of nitric oxide (see Chapter 2), which is therefore of prime
importance in the photochemical reaction process. Data are presented
below for both NO- and NO, and for their sum expressed as nitrogen oxides.
Table 5-8 shows nitric oxide and nitrogen dioxide data from the
CAMP network in the form of cumulative frequency distributions of 5-minute
values for the year 1966. The same type of information from selected cities
in the State of California SCAN network is shown in Table 5-9, except
hat the concentrations represent 1-hour averages and the data arc sepals, sd
according to summer and winter. Please note that nitrogen-dioxide and nitrogen
oxides (sum of NO and NO ) are reported.
Data on nitrogen dioxide only were available from several stations
5-79
-------
U
o
U vO
H O\
33 W
i a
m M
x
u. o
o M
e»
w
i'. yi
o w
r:8
§g
M h-l
nd %
H
W Q
M S
Q -2
tw tJ
U O
X M
W X
§.°
W u
fi
a!
H
W M
> «
M I
H I
*t in
rJ M
^
sd
u >
oo
I
,a
ca
H
I» Q
15
o «
> r.
U
O 41
U
a o
at U
u
r* "9
« V
a, u
i tg
i u
C (A
O
•H q
** a
5 P
W (A
W ID
(0 0)
Q
>» O
O
a o
3
23
(*4 0*
U]
5
u
u
H
Jflj -j «^r o vf> so-* •• ••
0 00 0» mrx (^m
O\ rHO oo oo oo oo
O OCN (N(*> rHCN O«N
H OO OO OO OO
r-
•H
•H .-
4J 1.
(0 C«l tN . 4 CM
gig s>§i x : ,,gg
•H flj PI OJ
O U r >
a -H •»-. c
•H j: jz
-------
Table 5-9.
CUMULATIVE FREQUENCY DISTRIBUTION OF HOURLY AVERAGE CONCENTRATIONS
OF NITROGEN DIOXIDE AND NITROGEN OXIDES (1967 SCAN).
I City Frequency Percent
No Sunimrr
Winter
NO Summer
* Winter
Pasadena
NO Summer
Winter
NO -Summer
x Winter
NO- Summer
Winter
NO Summer
x Winter
San Diego
NO, Summer
Winter
NO Summer
* Winter
Riverside
NO. Summer
Winter
NO Summer
x Winter
Oakland
NO Sununer
Winter
NO Summer
x Winter
10
.02
.05
.02
.04
.01
.00
.01
.01
.02
.01
.02
30
.03
.04
.10
.05
.02
.02
.02
.02
.02
.02
.04
.04
.02
.OJ
.04
50
.06
.08
.16
.05
.05
.07
.11
.03
.03
.05
.02
.04
.04
.03
.06
.08
.02
.03
.04
.oa
70
.06
.08
.11
.27
.06
.08
.10
.18
.03
.05
.04
.10
.01
.01
.08
.06
.04
.09
.12
.03
.05
.14
90
.09
.15
.20
.49
.09
.14
.15
.32
.04
.07
.07
.22
.02
.05
.04
.22
.09
.08
.13
.21
.06
.08
.09
.28
95
.11
.21
.24
.63
.11
.19
.17
.39
.05
.09
.09
.32
.03
.07
.07
.30
.10
.12
.15
.26
.08
.11
.12
Maximum
.43
.54
.86
1.34
.20
.37
.34
.71
.15
.30
.32
1.14
.34
.24
.34
.84
.25
.31
.31
.67
.26
.33
.48
1.06
Arithmel^
Mean
.05
.08
.10
.22
.06
.07
.09
.15
.03
.04
.04
.10
.01
.02
.02
.08
.05
.04
.08
.10
.03
.05
.05
5-81
-------
CM
a
v
u
tj
0)
IT)
U"> in
o o
rH O
a
lA
> a) M
rH O M
D 0) (U
*~) Q r^
M
01
,0
a x oi M
d I-H u n>
O 3 41 o»
5-82
-------
cooperating in the New York-New Jersey Abatement Study for the year 1967.
Freqeuncy distribution data were available by the month. This information
is shown in Table 5-10.
Examination of Tables 5-8 and 5-9 indicate that for many cities
the long term (annual) mean nitrogen dioxide and nitric oxide concentrations
are nearly equal. Nitric oxide, however, exhibits higher maxima and lower
minima. This results from its injection into the atmosphere as a primary
contaminant, often during times of reduced mixing capacity. During photo-
chemical reaction it may nearly disappear. Nitrogen dioxide, on the other
hand, is formed photochemically and is subject to simultaneous atmospheric
mixing processes. Its rate of disappearance is slower than nitric oxide,
but this too is subject to specific reaction and mixing conditions.
Peak values of nitric oxide above 1 ppm are experienced on a rather
widespread basis. Nitrogen dioxide concentrations have rarely reached this
level, with most of the known such occasions having occurred in the Los Angeles
area. In most major urban areas, peak NO concentrations are under 0.5 ppm.
Recalling Table 5-5, the two stations in the Los Angeles area had
higher oxidant levels more frequently than the other cities listed. The
same situation does not hold in the case of the nitrogen oxides. Although
exact comparisons cannot be made in each case, examination of Tables 5-8,
5-9, and 5-10 shows that Chicago, Bayonne and Newark have nicrogen oxide
patterns not unlike those of Los Angeles and Pasadena.
5-83
-------
c. Hydrocarbons
In examining atmospheric hydrocarbon concentrations in relation to
photochemical smog, several factors must be considered. First, as discussed
in Chapter 2, there is an enormous variation in the tendency for different
hydrocarbons to enter into the photochemical smog reaction sequence, some
(like methane) being virtually inert. Second, the simpler and less laborious
atmospheric hydrocarbon detection methods are likely to respond most strongly
to methane and other alkanes. Third, methane is very often more abundant
than all other hydrocarbons combined; there is, in fact, apparently a
"geophysical" minimum worldwide level of methane of about 1.0-1.5 ppm. ' *'
This makes it important in assessing photochemical air pollution to be able
to discriminate between methane and other more reactive hydrocarbons.
Unfortunately this type of analysis is expensive and time-consuming, so that
such data are not abundant. Nevertheless it is available, and in this sub-
paragraph such observations are discussed.
Although on occasion all other hydrocarbon concentrations drop
to unmeasurably low levels, methane does not. There have been numerous
measurements in remote, sparsely populated areas and far out on the ocean
which all suggest a worldwide minimum mi ^ne concentration of about 1.0 to
1.5 ppm. "Natural" sources of methane a:c mostly vegetation, especially
1n swampy areas. In inhabited areas methane levels are normally much higher;
values of 6 ppm or more have been observed. This is attributed to escape
from petroleum and natural gas, although small amounts also result from combustion
processes.
5-84
-------
Ratios of nonmethane hydrocarbons (as carbon) to methane have been
estimated for urban areas, after subtracting 1 ppm from methane values to
allow for estimated biogenic background levels . The nonmethane/methane
hydrocarbon ratios for several weeks averaged 0.6 in Cincinnati and 1.9
in Los Angeles, although methane values were similar (Figure 5-18). The
higher Los Angeles ratios evidently reflect the higher traffic densities.
In Table 5-11, hydrocarbons up to C, to C, as averaged in over 200
D /
samples in the Los Angeles atmosphere (1) are tabulated on a molar basis.
Note the great preponderance of methane. Even on a weight or carbon atom
basis, methane would constitute about half the total hydrocarbon fraction;
of the remainder, the saturated hydrocarbons (also relatively unreactive
photochemically) also predominate. These samples were generally taken just
before or during the morning traffic rush.
The proportion of aromatic to aliphatic and some of the proportions
108
among the aromatics have been studied . Table 5-12 shows (again for
Los Angeles) averages for aromatic concentrations observed in the atmosphere
over several weeks sampling.
In Table 5-13 the diurnal patterns are given for the C~ - C,.
hydrocarbons, showing the hour-by-hour variations averaged over some weeks
in the Los Angeles smog season. All species listed reached a maximum in the
morning, then declined through the midday (although there are irrc;t,,Parities
V99
in certain types)
5-85
-------
Sourcc-ttn nQih rutiot lor non
-------
Table 5-11. AVERAGE HYDROCARBON COMPOSITION--218 LOS ANGELES
AMBIENT AIR SAMPLES (1965).
Methane
Ethane
Propane
Isobutane
r»-Butane
Isopentane
n-Pentane
2,2-Dimethylbutane
2-Methylpentane
2,3-Dimathy1 butane
Cyclopentanu
3-Methylpen Cane
n-Hexane
Alkanes (beyond methane)
Ethyiene
Propone
1-lluLane -f Isobutylene
trans-2-Butane
cJhs-2-Butane
1-Pentene
2-Methyl-l-Butane
trans-2-Pentene
cls-2-I'entene
2-Melhyl-2-Butene
Propatllene
1,3-Bntadlene
Alkenes
Acetylene
Mctliylrfcctylene
Ai elyIt-no
Brnzonc
Toluene
Aromatics
TOTAL
Concentration, ppm Molar
3.22
.098
.OA9
.013
.064
.043
.035
.0012
.014
.004
.1)08
.012
ppm in Methane
3.22
.3412
.060
.018
.007
.0014
.0012
.002
.002
.003
.0013
.004
.0001
.002
.039
.0014
.032
.053
.]020
.0404
.0850
3.7886
5-87
-------
Table 5-12. AVERAGE AND HIGHEST CONCENTRATION MEASURED
FOR VARIOUS AROMATIC HYDROCARBONS.
Average Concentration Highest Measured
Aromatic Hydrocarbon ppm by Volume Concentration, ppm by Volume
Benzene 0.01S 0.057
Toluene 0.037 0.129
Ethylbenzene 0.006 0.022
£-Xylene 0.006 0.025
m-Xylene 0.016 0.061
o-Xylene 0.008 0.033
1-Propylbenzene 0.003 0.012
£-Propylbenzene 0.002 0.006
3- and 4-Ethyltoluene 0.008 0.027
l,3,i-Trimethylbenzene 0.003 0.011
1,2,4-TrimethyIbenzene,
i-Butyl- and Sec-Butylbenzene 0.009 0.030
tert-Butylbenzlne 0.002 0.006
Total Aromatics 0.106 0.33
5-88
-------
I
I
CO
25
O
W
o
§
u
o
o
Q -I
H
O
ya
P< W
w p
D-, o
4(NrHrHrHFHrHrH>H>-H
•H
•H
fl
O
CO CO *A sO 00 00 O 00 00 F"* rH rH C r^*» sO Csl -J" iA \D C>| 00 r^ lA iA \O
rHrHtNirMO>rHooor*-osooo nJ oOf^OOOvOsO^o^sOr^r^f^.
rHrHrHrHrHrH rH t/1 rH
w
2
f* "^ n C>4 C*4 rH rH rH fS (*4
oooooooooooo oooooooooooo
OOOO'j'yOOC OOO OOOOOOOOOOOO
OOOOO—*rHrH»-
-------
Very few comparison studies have been conducted of hydrocarbon
concentrations among various geographical locations. The study of Altshuller,
Ortman, et al. compared methane and nonmethane hydrocarbons in Cincinnati
1
and Los Angeles. Stephens and Burleson have determined the hydrocarbon
compositions for a number of air samples from widely scattered locations:
Hawaii, Denver, New York, and Monterey-Salinas (California), as well as Riverside
in Southern California. There are apparently substantial differences in
proportions of individual hydrocarbons which may sometimes, but not always, be
related to known differences in source contributions and extent of atmospheric
reaction. However, there is much not yet fully explained in such data.
Laboratory work shows that in the photochemical reaction system there
is a wide difference in the rate of disappearance br reaction of various
hydrocarbons. These differences, for single hydrocarbons at least, may be
related to structure: olefins and most aromatics are reactive, higher
alkanes not very reactive, while benzene, acetylene and lower alkanes are
virtually inert. In the atmosphere these differences have been harder to
observe. Since the general effects of meteorological dilution plus continuing
source emission obscure the effects of reaction, Stephens and Burleson
studied the reaction of trapped a""•'•> ric samples, artifically irradiated,
and found that reactivities were much as expected from laboratory results.
See Table 5-14.
The same authors as well as others ' have also attempted
to correlate observed atmospheric hydrocarbon concentrations with known
emission sources, In general three sources—auto exhaust, natural gas, and
5-90
-------
Table 5-14. COMPARISON OF RESULTS FROM THE ULTRAVIOLET
IRRADIATION OF AMBIENT AIR SAMPLES.
Compound s
1' 111 ane
htham'
I'ropune
Prupcne
Pro p. me
Propone
Isobutam*
ii-bulani-
Aretyl ene
I butane
i ->obut t* ni1
1 r.in-.-2-hu t fnr
1 soprn t anf'
i:is- J -huti-nr
n-pt>n tarn**
1 , )-|>llt .III 1 I'lll
Mr 1 In. 1 .uvt v. li iu-
.', .'-I1 IllU'tllV 1 1)11 1 .1111'
J -pi'p t rat*
2-mi'tlivl buti'HP-1
trnns-2-penti'iii'
2 , 3-ilii'H'thyl butnnu
2-mc'ihvl pcnt.im-**
i-mt-i hvl pi'ntjrx ***
Cycli p-'iitani-
T-lli- .llli'
Cyc l.^-'TiCt-'no
Concentration tn ppb
12/22/65 3/3/66
Percent
0 hour 2h hour Remain 0 hour 24 hour
38.5
134.2
17.4
35.4
21.6
48.'.
25..'
119.0
310.5
6.4
11. ti
(..li
93.6
-', . 5
60.0
t.'parated on chromatograph No. 2; remainder on
chronuit i<^' r'ipb No. 1 -
( . >ni .1 ins i-. u-i n v I hut me- 1
** ( -in t.» ins t ; -,
***(,., n t.i ins .'-IM
thyl
5-91
-------
gasoline vapor—, will account qualitatively for the observations. The
quantitative agreement shows some discrepancies, however. (There is often
an apparent excess of alkanes in the propane range over known source contributions.)
These are not thought to be of great significance in atmospheric photochemistry,
since these low-molecular weight alkanes are only very slightly reactive in
the system, compared to unsaturated hydrocarbons.
Although details remain to be explained, the role of various hydro-
carbons in atmospheric photochemistry is generally clear.
d. PAN
Utilizing gas chromatographic techniques with an electron capture
detector, PAN concentrations were measured in Los Angeles during September
and October 1965 by the California State Department of Public Health.
Seven measurements per day were made for each of the 16 weekdays in September
and 19 weekdays in October. The average concentrations by hour of day for
these periods are shown in Figure 5-19. The measurements were made near the
central Los Angeles air monitoring station, and the oxidant concentrations
are shown for comparison with the PAN concentration^.
Beginning in June 1966, measurements of PAN have been made on the
campus of the University of Calif., ai;.. it Riverside using the gas chromatograph
and electron capture method describ m ^aragraph G. of this chapter. Samples
are usually collected once each hour between 7:00 A.M. and 4:00 P.M. PST.
Other pollutants are also measured at the station operated by .ae Riverside
Air Pollution Research Center and supported by the U.S. Public Health Service.
In Figure '>-20, the average oxidant concentrations, measured with a Mast
5-92
-------
0.20r
0.18
0-.16
0.14
0.12
c
o
0.10
u
o 0.08
0.06
0.0-1
0.02
Averages -.
19 weekdays,
October
16 weekdays,
September
oi—
I I
8 9 10 11 12
_^1_-..._T_^,_.I,_ A I ( _ -
" -"^-----r-™--,.- ^ . n, •• *~
123
P.M. *
Hour of day, PST.
Figure 5-19. OXIDANT AND PAN CONCENTRATIONS BY HOUR OF DAY—DOWNTOWN
LOS ANGELES—(1965).
5-93
-------
Oxirtitnt
pp::i
0 . 1 6 r-
0. 1 1
0.12
M<>n:hl\ '.run hourly
ii.coat rat i
,JVIDA::T A.\U PAX COXCLNHATIONS BY HOUR OF DAY AIR POLLUTION
CEXTER—RIVERS LDi., CALIFORNIA—(1966) .
5-94
-------
Table 5-15.* AVERAGE AND MAXIMUM CONCENTRATION OF PAN-
RIVERSIDE, CALIFORNIA—(AUGUST 1967-APRIL 1968).
Month
August, 1967
September
October
November
December
January, 1968
February
March
Apr i I
August
(24-hour)
5.9
5.1
7.0
6.9
0.9
0.8
1.4
3.4
3.1
August3
(10-hour)
7.9
6.7
7.4
8.1
1.2
1.0
1.3
3.5
4.0
Maximum
(Month)
28
34
43
58
12
8
25
38
21
a Average for 10-hour period, 8:00 A.M. to 6:00 P.M.
* From J. APCA, L9, 348 (May 1969).
5-95
-------
analyzer (see paragraph C.)> and the average PAN concentrations are shown by
hour for the month of September, 1966. Similar data for October 1966 are
also shown in Figure 5-20. The monthly mean oxidant and PAN concentrations and
the monthly mean of the daily maximum hourly averages are shown in Figure 5-21.
In Figure 5-20 there are two daily maxima for the oxidant and PAN
concentrations. As previously discussed, the second maximum may be due to the
transport of pollutants from Los Angeles to Riverside. The PAN concentrations
in Riverside are an order or magnitude lower than those in Los Angeles, while the
concentration of oxidants are the same order of magnitude.
Additional data have been reported recently on PAN concentrations
114
in Riverside during late 1967 and early 1968. Included was an 18-day period
in November 1967 when an extended atmospheric stagnation episode resulted
In the highest concentration of PAN recorded to that time. This level,
53 ppb, is still substantially lower than levels which have been observed in
Los Angeles. A summary of the 1967-68 data is given in Table 5-15.
e. Aldehydes
Of all the principal reaction products formed in the atmosphere by
photochemical processes, the aldehydes are among the most poorly quantified.
By far most of the available data are from Los Angeles, where for the period
1951-1957 aldehydes were regularly T... i using manual techniques.
Total aldehydes were determined by the bisulfite procedure and reported
as formaldehyde, while formaldehyde was determined by the chromotropic acid
procedure. Table 5-16 shows the rang»i of maximum concentrations obtained
over this period. Typical maximum concentrations were nearer the low end of
5-96
-------
••; P O 00 U3 O
e o
O vO
O rH
I
«o
4J O
tn vn
nj o
c c
(0 a)
'O 3
X ai
O a'-
ai
u
o
a
hH
co
U4
H
a!
O
M
H •
!-J ^^
nJ W
,-J >
O H
Cl< CO
M O
< v.
i-t
V)
^ «
O r-.
hH »^D
f—t O^
^C «~H
H W
W 3
O *~j
•z
O X
U 0
3 °
H
Q
•-i VD
ON
H ^-i
Q X
rH LJ
X "I
O ^
ri
I
00
•H
u,
5-97
-------
Table 5-16. RANGE OF MAXIMUM CONCENTRATIONS OF ALDEHYDES AND
FORMALDEHYDE—LOS ANGELES COUNTY™(1951-1957) .*
^ear Concentration Range (ppm)
Formaldehyde Total Aldehydes
1951 .05 - .12 .26 - .67
L952 .20 - .27
1953 .25-1.20
1954 .39 - .80
1955 .47 -1.28
1956 .51 -1.30
1957 .27 - .47
* From "Air Quality of Los Angeles," J. E. Dickinson, Technical Progress
Report, Volume II, Los Angeles County Air Pollution Control District (1961)
5-98
-------
the range, although ,-ivailable data are insufficient to provide a complete
characterization.
118
A cooperative study jointly sponsored by the Air Pollution Control
District of Los Angeles County and the Air Pollution Foundation during the
period July-November 1960 provided limited data on total aldehydes, formalde-
hyde and acrolein using the bisulfite, chromotropic acid, and 4-hexybesorcinol
procedures, respectively. Total aldehydes ranged up to 0.36 ppm for a 10-
mlnute sample, while formaldehyde did not exceed 0.10 ppm. Typical aldehyde
concentrations were near 0.10 ppm on many days. The maximum acrolein value was
0.011 ppm, with most values being less than half that amount.
Aldehyde concentrations tend to peak near the time of the oxidant
maximum but this is not always the case, particularly in the winter season.
Tlu'y are produced both as primary contaminants from incomplete combustion and
in the photochemical process, so that a varied diurnal and seasonal pattern
is not unexpected. Because their importance has not been fully assessed and
bec.-iuso limited sampling data are available, additional effort is necessary
to determine the influence aldehydes will have in setting air quality—and
hence emission—standards for photochemical reactants.
5. Trends in Photochemical Air Contaminants
The time serves analysis of air quality data involves at 1ft_iSt four
components of variation: (1) eC'cuai trends, (2) cyclical variations, (3)
seasonal variations, and (4) random fluctuations. The ability to determine
possible trends in air quality rests upon the extent to which cyclical,
5-99
-------
seasonal, and random variations can be accounted. The assumption generally
made is that these latter three sources of variation are due to measurement
and meteorological variables and that any remaining indication of trend
is likely due to a real change in contaminant loading in the atmosphere.
It seems almost certain, based upon the experience of those who
119
have examined precipitation data and other atmospheric phenomena , that
there is an insufficient time record of photochemical air contaminant data
to arc-ount for any possible long-term cycles. On the other hand there is
little, or no evidente of any truly cyclical behavior of any of the meteorological
parameters which might affect the concentration of air contaminants.
Seasonal variations in air contaminant levels in the atmosphere
are present for many contaminants and have been referred to earlier in this
section. Several approaches may be used to account for this factor, the
most common being the 12-month moving average. When data for several years
is available, an average "seasonal index" may be calculated and applied, for
example, to monthly means to obtain "deseasonalized" monthly values.
The remaining and most troublesome source of difficulty in determining
possible trends is that of random variation. In the evaluation of visibility
trends in Los Angeles over the period 1 ',1 - 1958 reported by Holzworth and
PO
Mag.i *~ a trend of reduced visibilities was shown over the period from
1932 - 1948. Visual examination of the data suggests that the trend was
not clearly established until 1940, or a period of eight years. Further
evidence of the difficulty in evaluating possible trend information relating
to photochemical air contaminants is shown in several figures reproduced from
5-100
-------
121
a report by Ingels. eL al on atmospheric trends of nitrogen oxides.
Flgun- 5-22., shows a "Jeast squares" linear trenc* line fitted to raw monthly
dat i fiora di.wmtowii Los Angeles while Figure 5-22b shows a subjectively fitted
trend curve applied to "deseasonalized" monthly data. Figures 5-23a and 5-23b
stun; t.lie sai e information for data from an air monitoring station in Burbank
(about 10 miles northwest of downtown Los Angeles). Even though there was
a siight im rease in nitrogen oxides emissions in the County of Los Angeles
during this ;.ime period, no ea.^y interpretation of this data seems possible.
Ti"ud data on oxidant and nitrogen dioxide for several cities in
122
New Jersey ;.re shown for the period 1966 - 1968 in Figures 5-24 and 5-25.
Although a v>-ry slight trend downward seems evident for oxidant, no significant
i:oiuidunce rould bo attached to that statement. Table 5-17 presents on a
yearly basis, the maximum hourly average concentration and the number of
davs wlit-n the maximum hourly average for oxidant exceeded specified valjes
123
at each of t ue CAMP sites, from 1964 through 1967. Again, no evidence
of (.ri'i-.ds is seen for any of the sites over the relatively short period
considered.
Probably the longest time series of air quality data relating to
photochemical contaminants is that from the Los Angeles County Air Pollution
Control District. Figure 5-26 presents for the period 1955-1968, the yearly
numl'er of days exceeding specified levels of oxidant and nifrogen c~ >xi
-------
IX
CX,
w
H
M
S5
M
Q
M
X ,0
ilUUJJLLiU.1. lillilLUJliJ.UiJ..1 iOi-UJiL
iLUlLlL
IM7 '
UUULLI1
1961 '
MONTH
Figure 5-2Ja. MONTHLY MEAN DAILY MAXIMUM NO , CIVIC CENTER, 1957-1961.
TREND, +8.6 PPHM/YEAR. X
SIGNIFICANT
»<> a —
Figure 5-2.'h.
-'• J' ; * J M ' ' J ' H ' ..'uJ-LL.' ' w.' ! ' ] - ' ' ' '
Jii'.H ' )•,,.> ! I'l'" ' I'.f.l '
MONTH
TREND CURVES FITTED TO "DESEASONALIZED" MONTHLY MEAN DAILY MAXIMUM
NO DATA, LOS ANGELES CIVIC CENTER, STATIONS 1 AND 58, 1957-1961.
b-102
-------
LJlLLUlL'JJJJiUllUil .UiIiHJ.ll>1 J1LJ.U '.111 "
t'JS7
JJJJiU
'
.
' r<(.o '
'
MONTH
Fl«uru V;>;.,a. MONTHLY MEAN DAILY MAXIMUM NO , BURBANK, 1957-1961. NO
SIGNIFICANT TREND. X
'-23b. TREND CURVES FITTED TO "DESEASONALIZED" MONTHLY MEAN DAILY MAXIMUM
:ION 9,
5-103
NO DATA, BURBANK, STATION 9, 1957-1961.
fi
-------
&
PL,
W
§
a
M
X!
O
ijjyor.ne
Ancora
A Camdcr.
Oil *•< fik Apr J-« A.j On D.i f.k Ap, Jwn A.« O<|
N.. J.« Mw Mo, J.I S.pl N,. J.n tfe, Mn Jul Snl I
1965
1966
1967
1C F«b Apr JwH Attf, O(f 0
J.« Mv Ufa, J.I i.,1 N..
1960
Figure 5-24. QXIDANTS.
5-104
-------
Monthly Avi_rj^
J<< ».k A»< J-. A.< O« 0.. f.k A,, J,. A.. Ot
1965
1966
1967
Figure 5-25- NITROGEN DIOXIDE.
5-105
-------
Table 5-17- SUMMARY OF TOTAL OXIDANT CONCENTRATIONS RECORDED AT CAMP SITES,
1964-1967.
City
Chicago
Ommiuli
Denver
I'lul.ulcli'liia
St l.Ollll
W.(.-.liii-.;i.,ii. IVC.
You
1964
1965
1966
1967
1964
1965
1966
1967
1965
1066
1967
1964
I'M 5
19^6
l')67
»9M
IVOS
1966
1967
l%»
1965
1966
1967
Days of
vihJ iUti
254
275
235
255
303
MO
">nn
22*
?S5
298
166
2f,9
266
315
2X2
253
129
292
2S9
293
284
125
3>2
Number of days with at least 1 hourly
aveujje equal to or exceeding
0.05 ppm
149
120
52
113
137
182
54
122
226
187
76
124
109
145
1 M
156
206
174
1S5
163
150
IJ1
1 17
O.lOppm
IS
9
6
16
36
19
1
24
51
46
12
37
23
52
28
26
33
33
38
•10
25
27
27
O.OlSppm
0
0
3
1
5
5
0
1
14
9
4
9
4
19
3
6
g
5
4
4
3
2
5
Maximum
hourly
ppm
0.13
0.12
0.19
016
026
0 17
0.10
020
02S
0 19
0.21
020
0 31
O.i2
017
.126
0 3$
022
0 20
• 120
0 '1
0 16
0 >6
5-106
-------
DAYS MAX. OXIDANT CONC. EXCEEDED
NO. DAYS N02 EXCEEDCJ
0.25 PPM FOR 1 HOUR
70
NO. OF DAYS
CONDUCIVE TO ACCUM.
OF AIR CONTAMINANTS*
*DAYS WITH EARLY MORNING
INVERSION S 1500 FT. MAX.
MIXING HT. * 35CO FT.
0600-1200 HP.S WIND SPEED
•< 5.0 MPH.
55 b£ 57 S3
60 61 fi2 63
YEAR
65 65 67 68 69
b-26 ANNUAL VARIATION IN NUMBER OF DAYS OXIDANT AND N02 EXCLEU STATED
LEVELS rOGETHER WITH NUMBER OF DAYS CONDUCIVE TO ACCUMULATION OF
ONTAMINANTS —LOS ANGELES BASIN.
5-107
-------
1. Inversion base ^ 1500 feet.
2. Maximum mixing height • ')SOO feet.
3. 0600-1200 hours average.* wind speed <_ 5.0 mph.
A very preliminary assessment would seem to indicate a decreasing
trend of days exceeding the stated oxidant level and increasing trend
in number of days exceeding the given NO level. It is obvious that there
is substantial random, or unaccounted for, year-to-year variation. Taking
the meteorological parameters as an example (for which a trend does not seem
present) the standard deviation in the number of days meeting the high air
pollution potential criteria is 19. This is approximately 20 percent of
the mean annual value of 94 days. Several other precautions should be stated
regarding the interpretation of these data. First, even though the data
presented are "Basin" data (a day is included if the specified level is
exceeded at any air monitoring station), there were changes in the number
and location of air monitoring stations over the time period involved.
Secondly, the possibility of localized influences on concentrations cannot
be completely eliminated. Thirdly, it is accepted that the meteorologocal
criteria used to define high air pollution potential are at best rather crude.
In summary it seems evident that really clear-cut trends in the
occurrence and concentrations of ph 'tocl.emical air contaminants are either
not present or require more powerful tools of analysis to be discerned.
124
Serial correlation techniques normally used or special toe1 3 such ,.,„
125
power spectrum analysis as described by Brier may have to be brought into
play. Finally a substantial effort is still needed to account for the effects
5-108
-------
of m.'ti.orc.logical nnu other photophysical influences on tue reaction , ud
aceuiiulatiou of photochemical contaminants in the atmosphere.
K. DATA ACQUISITION REQUIREMENTS FOR DETERMINING REGIONAL AIR QUALIT
1. Suggested Measurements
The establishment of air quality criteria for e;ny one of tht possible
air contaminants occurring as a result of photochemical reaction has t gnifleant
implications insofar as atmospheric measurements are concerned. Not i. ,ly
must that particular contaminant be characterized so as to enable inte.pretation
of significant expected effects, but the primary contaminants must be described
in terms of those concentration parameters to permit structuring and e*.aluation
of control implementation plans. The potential number of possible meetureraents
is relatively great in number. This is particularly true if an atmospheric
research capability is required.
Fcr those areas in which the major objectives of an air quality data
acquisition program are (1) ability to evaluate air quality against criteria
and standards, and (2) ability to plan and evaluate contaminant control
implementation plans, the list may be narrowed.
As a minimum, a program should involve the capability of obtaining
hourly mean values for oxidant, nitrogen dioxide, nitric oxide, and hydrocarbons.
It would be highly desirable to make at least some nonroutine measurements
using one of the specific ozone methods to enable a more precise evaluation of
the oxidant data. The potential effect of sulfur dioxide should be judged
and, if necessary, compensating measures taken. If prescrubbers are nu.-essary,
the possibility of enhanced positive interference from oxidized nitric oxide
should be acknowledged.
5-109
-------
The major concern in the measurement of nitrogen oxides lies not in
problems of specificity, but in development of a capability to properly
calibrate and maintain presently available analyzers.
The measurement of hydrocarbons is best made on the basis of obtaining
information on the reactive classes. However, this 9bjective still remains a
challenge even for research-oriented organizations having substantial laboratory
resources. In lieu of obtaining detailed information on specific hydrocarbons
by full gas chromatographic procedures, the method used should at least separate
or remove methane from the remaining hydrocarbons, ihe total hydrocarbon
analyzer based upon flame ionization and incorporating the accessories enabling
I
a specific measurement of methane is, for the present, the best routine approach.
Wind speed and direction measurements will be necessary to interpret
air quality data in terms of sources and to facilitate construction or estimation
|
of overall community pollution patterns. It will be desirable in many cases
to supplement the wind data available from ESSA or other sources with that
collected specifically for air quality evaluation purposes.
Other measurements of photophysical parameters such as solar irradiation,
vertical temperature structure and other stability parameters will be useful.
It is particularly in this area that rr-wional, State and Federal cooperation
can be helpful in optimizing the util.ty rnd cost of the data.
2. Sampling Network Design
•»
In theory, the establishment of an air sampling network should be
based upon (1) criteria for meeting program objectives; (2) desired confidence
in output data; (3) ability to relate sampling site measurements to sources
and receptors; (A) optimizing cost/effectiveness ratios; (5) site specifications;
arid (6) overall program balance.
5-110
-------
I i practice, economic factors and the relative complexity of at least
continuous uonitoring equipment have set the pattern for the acquisition of air
quality dat.i. There has been an expansion of air sampling activity in recent
years, of course, particularly in some of the major metropolitan areas mentioned
in the data discussions of this chapter. But here, most of the expansion has
been related to particulate and sulfur oxides measurement.
It seems prudent to assume in long-range planning of resourct allocation
that economic and manpower pressures will not permit the acquisition of
detailed information about every point in space and time in the atmosphere.
It is not, in the strictest sense, even possible to do so. It is incuirbent
on us then to plan future networks with strict attention to objectives, to take
advantage of our knowledge of atmospheric behavior, and to apply interpretive
analysis tei hniques to the point that our expenditures for air quality data
actually contribute to the improvement in the quality of our environment.
Insofar as community air pollution is concerned we can list several
prim ipal ol jectives of air monitoring, as follows:
I. Establishment of background data
2. Evaluation of criteria and standards
3. Development of control strategies and plans
'4. Evaluation of control programs
). Development of acute incident warning
(i. F.stnM Lshment of source identification
7. Study of ntmo£-|ilse objt/i i Lves create varying requirements for measurement time
basi-s, sampling location density and distribution, for communications, and for
mtal information systems. A review of a number of these factors has recently
appeared. '•"^
5 111
-------
A few examples may help illustrate the applicability of some of the
above factors. First, let us assume we are Interested -i obtaining mean
geophysical contaminant levels (worldwide background). If 1000 stations were
to be established, the proportionate share for the United States (based upon
area) would be about 20. We would be concerned with having great measurement
sensitivity and field suitability, but not necessarily with communications or
immediate use of the data. Time response of the analyzer would be of little
i
importance and sampling could very well be intermittent.
Second, we would consider the requirements I for providing an early
warning system against an acute incident involving the possible release of
toxic material from a point source. In this case we would (1) use a very
concentrated geometric sampling grid; (2) require only moderate sensitivity;
!
(J) need rapid analyzer response; (4) use immediate communication links, and
(5) demand high reliability.
The more typical community air monitoring network would fall some-
whore between these extremes. The sampling grid will be affected by location of
sources and receptors and by local wind patterns; some—but not all—measure-
Hunts will necessitate the time response availability only from continuous
analyzers. The need for systematic analysis of air sampling instrument require-
ments with a view towards best meeting future programs has been recognized
by many. One particular scheme is illustrated by the systems chart in Figure
5-27-
In determining whether a community or eir quality control region
meets or exceeds air quality standards, a basic question exists as to how well
any given air sampling system represents a true situation in the atmosphere.
Assuming no problems with methodology, the question can be expressed in terms
5-112
-------
1
is
15
< *f
0 0 ~"
_ Ci- JJ
Q u_ x/i
^. ^ -) <
W H
> a
M W
Q £
H 2
M
a
M3
pi to
w M
o <
o
H M
CO <
o a-
H
U Oi
< M
5-113
-------
of the accuracy expected at a stipulated confidence level in using air quality
data from a network to estimate the contaminant concentration over a given time
base at a randomly selected point in time and space.
One approach to solving this problem is exemplified by the analysis
by Stalker * of data collected as part of a large-scale epidemiological study
conducted In Nashville. Up to 119 sampling site locations were used enabling
the analysis of predictive approaches based upon varying numbers of stations.
As a result of this work the following equation was developed to estimate the
minimum number of stations required to obtain the confidence and accuracy
j
levels specified:
where N « minimum number of stations required, t - student "t" for specified
confidence interval,
CV - coefficient of variation in percent
« standard deviation x 100 (for normal distribution)
mean
- (antilog sgxlOO))-100 (for lognormal distributions, where sg « standard
geometric distribution), and
p - allowable departure in percent from true mean.
As an example, it was determined that 245 stations would be required
(four per square mile) to estimate the daily mean concentration of SO^ at any
given point with 95 percent confidence of + 20 percent accuracy. On the other
hand it was estimated that one central and one peripheral station woulf '>*>.
required to calculate seasonal means at. any point in the area. This was based
upon the finding that the geographic cross-section of values gave an approximate
tit to the normal distribution curve.
5-114
-------
1 28
i'y c inducted by the Los Angeles County Air Pollution
Control District ut.ing one mobile and several fixed stations showed t'..it .sampling
locations up to 5 miles apart exhibited instantaneous values within + > pphm
of one anotlii-r at sll concentration levels.
Otlier a^i-roacheb to estimating the number of sampling stations to
provide' a satisfactory discription of the air quality of an area are lused
129
upon atmosp'.cric simulation models. Most of these are btatir diff-ision
models and thus an- not adequate to handle contaminants exhibiting significant
rates of reaction over the times used in the models. Neverthele s, initial
130
results of mode ling for Connecticut and several of the newly designated
air quality control regions indicate the possible utility of more advanced
models.
Oiu1 conclusion common to both the use of statistical analysis and
simulation model inc. techniques is that substantial field experimentation
iKTtleJ to develop (.he techniques and validate results. The potential savings
aiui heuefits possible should encourage further work in these areas.
3. Station Si tins
A usual and sometimes unwarranted assumption made about air quality
data Js tnai it f i^rly represents a reasonable area surrounding the site.
Even though we know, as discussed in the previous paragraph, that there are
limits tu tills assumption, precautions in siting must be taken to obtain results
not unduly influenced by highly localized factors.
Because physical requirements would usually preclude a purely
randomized site selection procedure, the establishment of some guidelines are
5-115
-------
desirable. Factors most commonly considered can be classified as (1) external:
(a) height above ground level of sampling point, (b) influence of surrounding
structures or terrain on air flow, and (c) localized sources; and (2) internal:
(a) length and type of sampling lines, and (b) analyzer environment.
Generally, sample inlets should be 10 to 12 feet above ground to
avoid street dust from low-level turbulence and unmixed auto exhaust. Higher
inlets should be evaluated in terms of atmospheric stability and representativeness
of principal receptor locations. Sample locations closer than two building
heights from taller structures should be avoided to preclude stagnation areas
formed in the lee of such structures. Strict rules about localized source
Influences are difficult to define but a contribution of more than 5 percent
to n measured concentration from any one source, as estimated from point source
diffusion calculations, would be undesirable for a station expected to provide
valid results for the surrounding area.
4. Data Processing and Validation
The usefulness of air quality data is related to its reliability and
availability. The former cannot be assured only by proper method and sample
location solection. Continuous attention to calibration, operation and maintenance
Is particularly important in continuous analyzer systems.
Calibration (the relationship of measured value to contaminant)
should be carried out as specified in the description of the method being
used. Neglect here can reduce to useless the value of the data collected.
Operational measures such as routine zeroing and spanning, adjusting flow controls,
5-116
-------
Servicing and cleanin.', components, noting cnange in parts, and describing
unusual events also contribute to producing ^formation approximating . lie
accur.iey theoretically attainable by the method.
Following tuo field procedures taken to produce valid inforn..1 ..ion,
a formal editing proo.'.ss is desirable for data collected with record!'
devac.es. A trained tvlitor can detect many symptoms of.' malfunction tiucii as
electronic noise and component failure, which may result in drift or lack of
sensitivity. Further quality control measures can b«- taken by establi.suing
criteria on the acceptance or rejection of data exhibiting unusual patterns—
whatever the cause—or for evidence of proper operational procedures su> h as
entry of zer», span, flow rate and other pertinent information.
The validation procedure should continue following the reduction of
dat.i to the form usud for storage. Steps would include random checks ol" reduced
datn vs. original charts and examination for rationality. Wherever computer
processing i;; utilised some of these checks can be built into the programming.
Example mean values over a given time period could be tested to insure that they
be less than shorter interval means or instantaneous values over the sane
olapsod time. Deviation from typical patterns could be checked by flagging
unusual peak-to-mean ratios or times of peaks.
Insofar as the analysis of air quality uata is concerned, a prime
objective is the de-termination of how the various indices of that data i.^Jate
to criteria and standards. Most commonly these standards are exps't-Tscd in terms
of frequency distributions of concentration ms-ans taken over various avtraging
periods. Study of existing air quality data indicates that there are c'aracteristic
5-117
-------
lognormal probability distributions functions for each contaminant at a given
locution. This knowledge enables the estimation of a wider range of values than
132
actually measured. Larsen and Zimmer demonstrated that 26 random
24-hour batch model nitrogen oxides samples served as a reasonably good predictor
of the frequency distribution of 70,000 5-minute values taken over a period
of a year with a continuous analyzer. Given the proper input, computer techniques
can be used to calculate a wide variety of distribution statistics in the form
required for evaluation of data against criteria and standards.
Still another use of air quality data is the development and evaluation
of abatement implementation plans. One approach involves the use of atmos-
pheric simulation models, which in turn involve emission and atmospheric
diffusion and reaction submodels. Reference was made to several of these in
tho previous paragraph. Because these models are still in beginning stages
i i development, somewhat more use has been made of multivariate analysis
I -j-j
iorhuiques. Larsen" has used this approach in calculating reduced emission
toqulremenis to meet certain air quality goals for a hypothetical case.
134
A factor analysis approach was used by Bifford and Meeker in
producing a number of independent factors characteristic of certain types of
pollution sources from a large amount of data on particulate subclasses from
the National Air Sampling Network. A regression of these factors on the cities
used in the study gave information on the importance of some of these factors
in characterizing the types of pollution, and thus likely source, in each location.
5-118
-------
REFERENCES
!• Stern, A. D., Ed., "Air Pollution," Academic Press, New York, 1968,
Vol. II, 2nd ed., Chapters 16, 17, 27.
2- Lodge, J. P., Jr., Pate, J. B., Ammons, B. E., Swanson, G. A., "The
Use of Hypodermic Needles as Critical Orifices in Sampling,"
J. Air Poll. Cont. Assoc.. 16, 197-200, 1966.
3. Thomas, M. D., Amtower, R. E., "Gas Dilution Apparatus for Preparing
Reproducible Dynamic Gas Mixtures in any Desired Concentration and
Complexity," J. Air Poll. Cont. Assoc.. 16, 618-623, 1966.
4. O'Keefe, A. E., Ortman, G. C., "Primary Standards for Trace Gas
Analysis," Anal. Chem., 38, 760-763, 1966.
5. Schuette, F., "Plastic Bags for Collection of Gas Samples,"
Atm. Environ., i, 515-519, 1967.
6. Renzetti, N. A., Romanovsky, J. C., "Comparative Study of Oxidants
in the Los Angeles Atmosphere," Arch. Ind. Health. 1.4, 458-467, 1956.
7. Renzetti, N. A., Ed., "An Aerometric Survey of the Los Angeles Basin,
August-November 1954," Air Pollution Foundation, Los Angeles, 1955.
8. Pierce, L. B., Nichikawa, K., Fansah, N. 0., "Validation of Calibration
Techniques," 8th Conference on Methods in Air Pollution and Industrial
Hygiene Studies, Oakland, California, February 6-8, 1967.
9. Interbranch Chemical Advisory Committee, "Selected Methods for the
Measurement of Air Pollutants," PHS Publication No. 999-AP-ll,
Cincinnati, 1965.
5-119
-------
10. Mast, G. M., Saunders, H. E., "Research and Development of Instrumentation
of Ozone Testing," ISA Transactions, JL., No. 4, October 1962.
11. Potter, L., Duckworth, S., "Field Experience with the Mast Ozone
Recorder." J. Air Poll. Cont. Assoc.. L5, 207-209, 1965.
12. Siu, W., Ahlstrom, M. R., Feldstern, M., "Comparison of Coulometric
and Colorimetric Oxidant Analyzer Data," 10th Conference on Methods in
Air Pollution and Industrial Hygiene Studies, February 19, 1969,
San Francisco.
13. Byers, D. H., Saltzman, B. E., "Determination of Ozone in Air by
Neutral and Alkaline Iodide Procedures," Amer. Ind. Hyg. Assoc.. 19.
251-257, 1958.
14. Sanders, G. R., State of California Department of Public Health,
Personal Communication, June 1965.
15. Cherniack, I., Bryan, R. J., "A Comparison Study of Various Types of
Ozone and Oxidant Detectors Which are Used for Atmospheric Air
Sampling," J. Air Poll. Cont. Assoc.. 15. 351-354, 1965.
16. Cholak, J., Shafer, L. J., Younker, W. J., Yeager, D., "Air Pollution
in a Number of American Cities," Arch. Ind. Health. 11. 280-289, 1955.
17. Deutsch, S., "Acid Potassium Iodide Method for Determining Atmospheric
Oxidants," J. Air Poll. Cont. Assoc.. 18. 78-83, 1968.
18. Haagen-Smit, A. J., Brunelle, M. F., "The Application of Phenolphthalin
Reagent to Atmospheric Oxidant Analysis," Int. J. Air Poll.. JL,
51-59, 1958.
5-120
-------
*' ReRener, V. H. , "On a Sensitive Method ''or the Recording of Atnu spheric
o^ne," J. «;>iOHiys. Res.. 65., 3975, 1960.
R«'gt'iK;r, V. H., "ieasurement of Atmospheric Ozone wit.h tho
Ctu-mi lumine&-.eu(.:e Method," J. Geopnys. Kes. , 69. 3795, 1964.
21 • Kuhtir, H., s;,itM, J. K., Tabor, K. C., "Ozone Determination: .'
Comparison Wutvoi:n Chemi Luminescence and Kl Techniques," 9th Coi>; erence
01. Methods in A.r 1'ollution and Industrial Hygiene Studies, nHs..ana,
C.ili£< rnia, Kebru.iry 7-9, 1968.
^- Haagrin -Smit, .\. I., i;ox, M. M. , "Ozone Formation in i'hotocheiruca ,
Oxidation ot Organic Substances," Ind. Eng. Ghem.. 48, 1484-1487. 1956.
') '\
*• Renzetui, N. A., "Ozone in the Los AngeJes Atmosphere," Ozone
Chemistry and Technology, ACS Advances in Chemistry Series, .21 ,
465-, 1959-
2't. Bravo, ii. , Lotlgt-, J., "Specific Spectrophotometric Determination )f
Ui'Un^ in Lht /^Lirjoaphere," Anal . Chcm. , 36, 671-673, 1964,
25. i:ol!'!ii, I. 'i. . Purcell, T. C., Altshuller, A. P., "Analysis of th
<>\] i. in" in I'hoLo-oxidatiou Reactions," Environ. j>_ct_. and Tech. ,
/u. Mash, T., "Color metric Determinatiun ul 0 -i.ne by I)i. t etyl-lunvv
LutiUine," Atoms. Environ. , _!, 679-6B7, 1967.
27. Cohen, 1. K,, Bufalini, J. J., "Further Observations on v)4a Ftrr us
Ammonium Thiocy;.uate Reagent for Ozone," Enviroa . be . and Tec 'a. 1 ,
1014, 1967.
5-121
-------
28. Hauser, T. R., Bradley, D. W., "Specific Spectrophotometric Determination
of Ozone in the Atmosphere Using l,2-Du-(4-Pyridyl)ethylene,"
Anal. Chem. .38, 1529-1532, 1966.
29. Hauser, T. R., Bradley, D. W., "Effects of Interfering Substances and
Prolonged Sampling on the l,2-Di-(4-Pyridyl)Ethylene Method for
Determination of Ozone in Air," Anal. Chem.. 39, 1184-1186, 1967.
30. Manita, M. D., Rumyantseva, M. V., Egite, M. E., "Spectrophotometric
Determination of Ozone with Dihydroacridine in the Atmosphere of
Populated Areas," Hygiene Sanit.. 3.2, 219-225, 1967.
31. German, A., Panouse-Perrin, J., Quero, A. M., "Dosage Colorimetrique
de I1 Ozone," Ann. Pharm. Fr. . (Paris). 25., 115-120, 1967.
32. Bufalini, J. J., "Gas Phase Titration of Atmospheric Ozone,"
Environ. Sci. and Tech.. ,2, 703-704, 1968.
33. Dudley, J. E., Lozano, E. R., Bufalini, J., Yunghans, R. G.,
"Continuous Monitoring for Atmospheric Total Oxidant and Simultaneous
Determination of Ozone by Gas Phase Titration," Paper No. 69-87,
presented at Air Poll. Cont. Assoc. Meeting, New York, June 24, 1969.
J'». Harsch, P., Dcuringer, R., "Galvanic Monitoring of Ozone in Air,"
Anal. Chem.. 15, 897-899, 1963.
3'j. Leighton, P. A., "Photochemistry of Air Pollution," Academic Press,
New York, Pp. 184-186, 1961.
M. Bonamassa, F., (California Air Resources Laboratory), personal
communication.
5-122
-------
J7- Saltsra.in, B. I., Wartburg, A. F., Jr., "Precision Flow Dilution i
for St.nulard Low Concentrations of Nitrogen Dioxide," Anal. Chem. . _37_,
1261-1.'o4, 1%5.
W- Hersch, P., and Deuringer, K., "Trace Addition of Nitric Oxide and
Nitrogtn Dioxide to Air by Electrolysis," Paper 63-36, 56th Annua]
Meeting, Air I'oll. Cont. Assoc., Detroit, 1963.
39. Saltznun, B. F.., "Colorimetric Microdetermination of Nitrogen
Dioxide in the Atmosphere," Anal. Chem.. 26, 1949-1955, 1954.
40. Riplcy, 1). L., Clingenpul, J. M., Hurn, R. W., "Continuous
Determination of Nitrogen Oxides in Air and Exhaust Gases,"
Int. J_. Air and Water Poll. , 8., 455-463, 1964.
•'•>!• Wilson, 1)., Kopxzynski, S. L. , "Laboratory Experiences in Analyst •; of
Nitric Oxide with 'Dichromate' Paper," J. Air Poll. Cont. Assoc.. 18,
160-161, 1968.
42. Calhoun, J. D., Brooks, C. R., "A Solid Oxidant for Oxides of
Nitrogen AnaJyzur," 7th Conference on Methods in Air Pollution and
Industrial Hygiene Studies, Los Angeles, January 25, 1965.
43. Saltzman, B. K., Chairman, Subcommittee 3, "Tentative Method of
Analysis for Nitrogen Dioxide Content of the Atmosphere (Griess-
S.iltzman Reaction) 42602-01-68T," in Health Lab. Sci.. 6, 106-113, 1969.
44. Morrison, M. E., Corcoran, W. H., "Optimum Conditions and Variability
in Use of Pulsed Voltage in Gas-Chromatographic Determination oJ
l'art.s-P«-r-Million Quantities of Nitrogen Dioxide," Anal. Chem., _3j!»
:">5-258, 1967.
5-123
-------
45. Barringer, A. R., Newburg, B. C., "Remote Sensing Correlation
Spectrometry for Pollution Measurement," 9th Conference on Methods
in Air Pollution and Industrial Hygiene Studies, Pasadena, California
February 7-9, 1968.
46. Kay, R. B., "Absorption Spectre Apparatus Using Optical Correlation
for the Detection of Trace Amounts of S0»," J. Appl. Opticsr .6,
776-778, 1967.
47. "Remote Monitoring Instrumentation for Source Emissions and Urban
Atmospheres," Clean Air Quarterly (State of California), 11. pp 9-12,
December 1967.
48. Androatch, A. J., Feinland, R., "Continuous Trace Hydrocarbon Analysis
by Flame lonization," Anal. Chem.. 32. 1021-1024, 1960.
49. Bruderreck, H., Schneider, W., Halasz, I., "Quantitative Gas
Chromotographic Analysis of Hydrocarbons with Capillary Columns
and Flame lonization Detector," Anal. Chem.. 36. 461-473, 1964.
50. Altshuller, A. P., "An Evaluation of Techniques for the Determination
of the Photochemical Reactivity of Organic Emissions,"
J. Air Poll. Cont. Assoc.. 16., 257-260, 1966.
51. Ortman, G. C., "Monitoring Methaae in Atmosphere with a Flame
lonization Detector," Anal. Chem.. 38_, 644-646, 1966.
5-124
-------
52. Altshuller, A. P., Ortraan, G. C., Saltzman, B. E., Neligan, R. K.,
"Continuous Monitoring of Methane and Other Hydrocarbons in Urban
Atmospheres," J. Air Poll. Cont. Assoc.. 16, 87-91, 1966.
53. Altshuller, A. P., "Atmospheric Analysis by Gas Chromatography," in
"Advances in Chromatography," Vol. 5, Giddings, J. C., Keller, R. A.,
Eds., Marcel Dikker, New York, 1968.
54. Pierct, L. B., Mueller, P. K., "Analysis of Atmospheric Hydrocarbons
by Gas Chromatography," 10th Conference on Methods in Air Pollution
and Industrial Hygiene Studies, San Francisco, February 19-21, 1969.
55_ Mader, P. P., Heddon, M. W., Lofberg, R. T., Koehler, R. H.,
"Determination of Small Amounts of Hydrocarbons in the Atmosphere,"
Anal. Chem.. 2A, 1899-1902, 1952.
55 Shepard, M., Rock, S. M., Howard, R., Stortnes, J., "Isolation
Identification and Estimation of Gaseous Pollutants in Air,"
Anal. Chem.. 23, 1431-1440, 1951.
Weavers E. R., Hughes, E. E., Gunther, S. M., Schuhmann, S.,
Redfeam, N. T., Gordon, R., Jr., "Interpretation of Mass Spectre
From Condensates from Urban Atmospheres," J. Res. Natt. Bur. Std.t 59,
383-404, 1951.
57 MacPhec, R. D., "Use of Molybdates for Estimating Amount of Olefinic-
Type Hydrocarbons in Air," Anal. Chem.. 26. 221-225, 1954.
<^g Altshuller, A. P., Sleva, S. F., "Spectrophotometric Determination
of Olefins," Anal. Chem.. 33., 1413-1420, 1961.
5-125
-------
59. Nicksic, S. W., Roatenbach, R. W., "Instrumentation for Olefin
Analysis at Ambient Concentrations," J. Air Poll. Cont. Assoc. 11.
417-420, 1961
60. Lyles, G. R., Dowllng, F. B., BIanchard, V. J., "Quantitative
Determination of Formaldehyde in the Parts per Hundred Million
Concentration Level," J. Air Poll, Cont. Assoc., 15, 106-108, 1965.
61. Altahuller, A, P., Cohen, I. R., Myer, M. E., Wartburg, A. F., Jr.,
"Analysis of Ahphatic Aldehydes in Source Effluents and the Atmosphere,"
Anal. Chim. Acta. 25_, 101-117, 1961.
62. Wilson, K. W., "Fixation of Atmospheric Carbonyl Compounds by
Sodium Bisulfite," Anal. Chem.. 30, 1127-1129, 1960.
63. Altshuller, A. P., Miller, D. L., Sleva, S. F., "Determination of
Formaldehyde in Gas Mixtures by the Chromotropic Acid Method,"
Anal. Chem., 33, 621-625, 1961.
64. Sawicki, E., Hauser, T. R., Stanley, T. W. , Elbe'rt, W. ,
"The 3-Methyl-2-benzothiazole Hydrazone Test," Anal. Chem.. 33,
93-96, 1961.
65. Altshuller, A. P., Leng, L. J., "Application of the 3-Methyl-2-
benzothiazole Hydrazone Method for Atmospheric Analysis of Aliphatic
Aldehydes," Anal. Chem.. 3_5, 1541-1542, 1963. j
66. Hauser, T. R., Cummins, R. L., "Increasing Sensitivity of
3-Methyl-2-benzothiazole Hydrazone Test for Analysis of Aliphatic
Aldehydes in Ari." Anal. Chem.. 36., 679-681, 1964.
5-126
-------
67. Sawicki, E., Hauser, T. R., Fox, F. T., "Spectrophotometric
Determination of Aliphatic Aldehyde 2,4-Dinitrophenylhydrazones with
3-2-benzithiazolr Hydrazone." Anal. Chim. Ada. 2&, 229-234, 1962.
68. Jonr
the Determination of Acrolein in Combustion Gases and in the Atmos;> ;ere,"
Anal. Cham.. 33, 726-733, 1961.
70. Stephens, E. R., Darley, E. F., Taylor, 0. C., Scott, W. E.,
"Photochemical Products in Air Pollution," Int. J. Air and Water J^'l.,
4., 79-100, 1961.
71. Nicksic, S. W., Harkins, J., Mueller, P. K., "Some Analysts for IV.:
and Studies of Its Structure." Atmos. Environ.. _!, 11-18, 1967.
72. Stephens, E. R., "Absorptivities for Infrared Determination ai
IV-roxyacyl Nitrates," Anal. Chem.. 36, 928-9^9, 1964.
/J. Darley, I. F., Kittner, K. A., Stephens, E. K., "Analysis ot
Peroxyacyl Nitrates by Gas Chromatography with Electron Capcuie
Detection," Anal. Chem.. 35. 589-591, 1963.
74. Heuss, J. M., Glasson, W. A., "Hydrocarbon Reactivity and Eye IrriciLiun,"
Environ. Sci. and Tech.. 2., 1109-1116, 1968.
75. Altshuller, A. P., Kopczynski, S. L., Lonneman, W. A., Beck«r
Slater, R., "Chemical Aspects of the Photp-oxidation of Prcpylene-
Nitrogeri Jxide System," Environ, Sci. and Tech.. 1_, 899-914, 1967.
5-127
-------
76. Taylor, B. C., Stephens, E. R., Cardiff, E. A., "Automatic
Chromatographic Measurement of PAN," Paper 68-70, 61st Annual Meeting
Air Poll. Cont. Assoc., Saint Paul, Minnesota, June 23-27, 1968.
77. Lundgren, D. A., Cooper, D. W., "The Effect of Humidity on
Light-Scattering Methods of Measuring Particulate Concentrations,"
J. Air Poll. Cont. Assoc.. 19, 243-247, 1969.
78. Anderson, A. A., "A New Sampler for the Collection, Sizing, and
Enumeration of Viable Airborne Particles," J. Bact.. 76. 471-484, 1958.
79. Friedrichs, K. H., "Experience with Impactors in Dust Measurements,"
Staub-Reinhalt. Luft. 2B, 19-21, 1968.
80. Lundgren, D. A., "Determination of Aerosol Composition as a Function
of Particle Size and Time," 10th Conference on Methods in Air
Pollution and Industrial Hygiene Studies, San Francisco,
February 26-28, 1969.
81. Goetz, A., Stevenson, A. J. R., Preining, 0., "The Design and the
Performance of the Aerosol Spectrometer," J. Air Poll. Cont. Assoc..
10. 378-383, 414, 416, 1960.
82. Lieberman, A., Schimpa, P., "Air Pollution-Monitoring Instrumentation.
A Survey," NASA SP-5072, IIT Research Institute, 1969.
83. Whitby, K. T., Vomela, A., "Response of Single Particle Optical
Counters to Nonideal Particles," Environ. Sci. and Tech., _!,
801-814, 1967.
5-128
-------
84. Charl.«4on, R. J., Howath, H., Pueschel, R. F. , "The Direct
Measurement of Atmospheric Light-Scattering Coefficients for Stu
of Visibility and Pollution," Atmos. Environ.. 1., 469-478, 1967-
85. Charls'm, R. J., Ahlquist, N. C., Howath, H., "On the Generality of
Correlation ot Atmospheric Aerosol Mass Concentration and Light
Scatter1," Atmos. Environ.. .2, 455-404, 1968.
86. Noll, K. E., Mueller, P. K., Imada, M., "Visibility and Aerosol
Concent,ration in Urban Air," Atmos. Environ.. J2, 465-475, 1968.
87. Barriiti, E. W. , Ben-IJov, 0., "App ication of the Lidar to Air
Pollutiom Measurements," J. Appl. Moteorol.. .6, 500, 1967.
88. (Joetz, \., Tsuneisht, N., "Bacteriologic Analogue for Eye Irriataon
by Aerosols," Arch. Ind. Health. 20. 167-180, 1959.
89. Serat, W. K., Budinger, F. E., Jr., Mueller, P. K., "Evaluation cf
Biologu-al Eftects of Air Pollutants by Use of Luminescent Bacceria,"
J. Bact.. 90, 832-833, 1965.
90. Serat, W. F., Bridinger, F. E., Jr., Mueller, P. K., "Toxicity
Evaluation of Air Pollutants by Use of Luminescent Bacteria,"
Atmos. Environ., .1, 31-32, 1967.
91. Serat, V. F., Kyons, J., Mueller, P. K., "Measuring the Effect of /iir
Pollution on Bacterial Luminescence: A Simplified Procedure,"
Atmos. ''.nviron. . 3[, 303-309, 1969.
92. Wayne. L. G., "Eye Irritation as a Biological Indicator of Photochemical
Reactions in the Atmosphere," Aimos. Environ. . !_, 97-104, 1967.
5-129
-------
93. Bryan, R. J. and Romanovsky, J. C., "Instrumentation for Air Pollution,"
Instruments and Automation, 29, 2432-2438 (1956).
94. Perry, L., "Methods Used in the Statewide Cooperative Air-Monitoring"
Project," Proceedings of the Sixth Conference on Methods in Air
Pollution Studies, California Department of Public Health, Los Angeles
(January 25-26, 1965).
95. CAMP in Cincinnati 1962-1963; Public Health Service Publication
No. 999-AP-21, Cincinnati, 1965.
96. Technical Report, New York-New Jersey Air Pollution Abatement
Activity, Sulfur Compounds and Carbon Monoxide, USPHS, National Center
for Air Pollution Control, Cincinnati (January, 1967).
'^7. Fair, D. H., Morgan, G. B., Zimmer, C. A., "Storage and Retrieval
of Air Quality Data (SAROAD). System Description and Data Coding
Manual," PHS, Publication No. APTD-68-8, Cincinnati, Ohio, National
Center for Air Pollution Control (June, 1968).
98. Regener, V. H., "Measurement of Ozone with the Chemiluminescent
Method," J. Geophys. Res.. 3795-3800 (1964).
99. Richter, H.G., Smith, J. R., and Ripoerton, L.A., "Chemiluminescent
Ozone Measurement Program-Urban Atmosphere," Research Triangle
Inst. Final Report. Contract PH-27-68-26 (December, 1967).
5-130
-------
100. Richter, H.G., et al, "Chemiluminescent Ozone Measurement Program—Ozone
Total Oxldant Relationship in Ambient Air," Research Triangle Ins;..
Final Report, Contract PH22-68-30 (May, 1969).
101. "Recommended Ambient Air Quality Standards," State of California Air
Resources Board (May 21, 1969).
102. Mueller, P.K., Fansch, N.O., Tokiwa, Y., Kothing, E.L., "Series vs.
Parallel NOx Analysis," 9th Conf. on Methods in Air Pollution Stmties,
Pasadena, California, February 7-9, 1968, California Department i J
Public Health (February, 1968).
103. Cavanagh, L.A., Schadt, C.F., and Robinson, E., "Atmospheric
Hydrocarbon and Carbon Monoxide Measurements at Point Barrow, Alaska",
Env. Sri, and Tech. 3_ 251-257 (1969).
104. Ehhalt, D.H., "Methane in the Atmosphere", J.Alr Poll. Con t. As so c.
12 518-519 (1967).
105. Swlnnerton, J.W., Tinnenbom, V.J., and Cheek, C.H., "Distribution of
Methane and Carbon Monoxide Between the Atmosphere and Natural
Waters", Env. Sci. and Tech. 3_ 836-838 (1969)
106. Altshuller, A.P., Ortraan, G.C., and Saltzman, B.E., "Continuous
Monitoring of Methane and Other Hydrocarbons in Urban Atmospheres"
J. Air Poll. Cont. Assoc. 16 87-91 (1966).
107. Altshuller, A. P., and Bellar, T.A., "Gas Chromatographic Analysis
of Hydrocarbons in the Los Angeles Atmosphere", J.Air.Poll. Cont.Assoc.
13 81-87 (1963).
5-131
-------
108. Tonneman, W. A., Bellar, T.A., and Altshuller, A.P., "Aromatic Hydrocarbons
in the Atmosphere of the Los Angeles Basin," Env. Sc i. and Tech. £ 1017-1020
(1968).
109. Gordon, R. J. Mayrsohn, M., and Ingels, R.M., "C.-C, Hydrocarbons in
the Los Angeles Atmosphere", Env. Sci. and Tech. j2 1117-1120 (1968).
110. Stephens, E.R., and Burleson, F.R., "Distribution of Light Hydrocarbons
in Ambient Air", Paper No. 69-122 presented at 62nd Annual Meeting of
the Air Poll. Cont. Assoc., New York, 1969.
111. Stephens, E.R., and Burleson, F.R. "Analysis of the Atmosphere for Light
Hydrocarbons", J.Air Poll. Cont.Assoc. 17 147-153 (1967)
112. Neligan, R.E., "Hydrocarbons in the Los Angeles Atmosphere",
Arch.Environ. Health, _5 581-591 (1962).
113. Mayrsohn, H. and Brooks, C., "The Analysis of PAN by Electron Capture Gas
Chromatography," presented at Western Regional Meeting of the American
Chemical Society, November 18, 1965.
114. Taylor, O.C., "Importance of Peroxyactylinitrate (PAN) as a Phytotoxic
Air Pollutant," J. of the Air Poll. Contr. Assoc. 19, 347-351 (May, 1969)
115. Technical Progress Report, Vol. 2, "Air Quality of Los Angeles County,
pp. 26,29,43,54 County of Los Angeltj Air Pollution Control
District (February, 1961).
116. Method 5-46 "Aldehydes, Total," Laboratory Methods of the Los Angeles
County Air Pollution Control District (1.968).
5-132
-------
117. Method 8-53, "Formaldehyde", Ibid
118. Renzeti-i, N.A. , and Bryan, J.F., "Atmospheric Sampling for Aldehvdes and
Eye Irritation," J.of the Air Poll. Contr. Assoc.^ 11, 421-424,
427 (September, 1961).
119. Brier, G.W., "Some Statistical Aspects of Long-Term Fluctuations in
Solar and Atmospheric Phenomena," Annals of the New York Acad.
of Science, 95: 173-187, (1961).
120. Holzworth, G.C. and Maga, J.A., "A Method for Analyzing the Trem: In
Visibility," Journal of the APCA, 10, 6, 430-435, (December, I960?.
121. Ingels. R.M., Holmes, R.G., Hamming, W.J., Chass, R.L., and
Griswold, S.S., "Trends in Atmospheric Concentrations of Oxides of
Nitrogen, 1957-1961, Los Angeles County Air Pollution Control District
Report, (August, 1962).
122. Technical Bulletin A-69-1, New Jersey Department of Health, Division of
Clean Air and Water (July, 1969).
123. "Air Quality Criteria for Photochemical Oxid.-jits," U.S. DHEW, PHS, EHS,
NAPCA, (March, 1970).
124. "Statistical Analysis in Chemistry and the Chemical Industry,"
C.A. Bennett and N.L. Franklin, John Wiley and Sons, Inc., New York, (1954),
125. G. W. Brier, in Symposium on Environmental Measurements—Valid Data and
Logical Interpretation, 265—72, PHS Publ. No. 999-AP-15, (July, 1964).
126. Bryan, R. J.,"Air Pollution", Vol. il, 2nd Ed., A.C. S.era, Ea.,
Academic Press, Inc., New York, New York, 1968, Chapter 26.
5-133
-------
127. Stalker, W. W., Dickerson, R. C. and Kramer, G. D., J. Air Poll.
Contr. Assoc. 12, 361 (1962).
128. Holland, W. D., Fisher, E. L., Brunelle, M. F., and Bryan, J. R.,
"Relationship Between Fixed Station and Mobile Air Sampling in Los Angeles
County", paper, 68-41, 61st Annual Meeting of the Air Pollution Control
Association, St. Paul, Minnesota (June, 1968).
129. Martin, D. 0., "A General Atmospheric Diffusion Model for Estimating
the Effects on Air Quality of One or More Sources", paper 68-148,
61st Annual Meeting in Pollution Control Association, St. Paul, Minn.
(June 1968).
130. N. E. Bowne, "A Mathematical Diffusion Model for Connecticut", paper
69-17], 62nd Annual Meeting, Air Pollution Control Association,
New York, New York (June, 1969).
HI. "Report for Consultation on the Metropolitan Chicago Interstate Air
Quality Control Region (Indiana-Illinois), PHS, Nat. Air Poll. Control
Admin. (Sept. 1968).
H2. Zimmer, C. E., Larsen, R. L., "Calculating Air Quality and Its
Control", J. Air Poll. Contr. Assoc.. 15 565-72 (January 1965).
133. Larsen, R. I., "Determining Reduced Emission Goals Needed to Achieve
Air Quality Goals—A Hypothetical Case", J. of the Air Poll. Contr.
Assoc., r7, 823-9, (December, 1967).
134. Bifford, Jr., I.H., Meeker, G.O., "A Factor Analysis Model of Large
Scale Pollution," Atm. Environment. 1, 147-57 (1967).
5-134
-------
'- CHAPTI.R 6
MULTIVARIATE ANALYSES OF AIR
* QUALITY AND ENVIRONMENTAL DATA
Page
A. INTRODUCTION 6-5
B. DATA ACQUISITION AND MANAGEMENT 6-5
1. Air Quality Data 6-5
2. Meteorological Data 6-7
3. Data Management 6-8
C. THE STATISTICAL PROGRAM SYSTEM 6-10
1. Rank Order Analysis 6-12
2. Freauency Distribution and Simple Statistics 6-12
3. Correlation and Regression Analysis 6-15
4. Auto-correlation and Cross-correlation 6-17
D. RESULTS AND DISCUSSION 6-22
1. The SCAN Stations as a Set 6-23
2. The CAMP Cities as a Set 6-36
3. Comparison of the SCAN Stations and CAMP Cities 6-40
4. Additional Analysis Possible 6-41
6-1
-------
CHAPTER 6
LIST OF FIGURES
Figure Page
6-1 SDC Air Quality Data Management and Analysis 6-9
System (AQ/DMAS).
6-2 Sample of Total Merged Data Base, Denver, 1967. 6-11
6-3 Rank Order Analysis, Anaheim, 1967. 6-13
6-4 Frequency Distribution for Total Oxidant. 6-14
6-5 Descriptive Statistics. 6-16
6-6 Regression Analysis of Episode Days Selected on 6-18
Basis of Rank Order Analysis.
6-7 Regression Analysis of Episode Days Selected on 6-19
Basis of Rank Order Analysis.
6-8 Auto-correlations. 6-21
6-9 Cumulative Frequency Distributions of Hourly 6-26
Average Oxidant, 1967.
6-10 Cumulative Frequency Distributions of Hourly 6-27
Average Carbon Monoxide, 1967, 5 SCAN Stations.
6-11 Cumulative Frequency Distributions of Hourly 6-28
Average Total Hydrocarbons, 1967, 4 SCAN
Stations.
6-2
-------
Figure Page
6-12 Cumulative Frequency Distributions of Hourly 29
Average Nitric Oxide, 1967, 5 SCAN Stations.
6-13 Cumulative Frequency Distributions of Hourly 30
Average Nitrogen Dioxide, 1967, 5 SCAN
Stations.
6-3
-------
CHAPTER 6
LIST OF TABLES
Table Page
6-1 Photochemical and Meteorological Data Elements. 6-6
6-2 Sample Statistics, All Stations, 1967. 6-24
6-3 Comparison of Five SCAN Stations by Ranking 6-31
Relative to Five Contaminants (99th Percentile).
6-4 Results of Correlation and Regression Analyses, 6-34
Six SCAN Stations, 1967.
6-5 Results Auto- and Cross-correlation Analyses, 6-35
Six SCAN Stations, 1967.
6-6 Results of Correlation and Regression Analysis, 6-38
Six CAMP Stations, 1967.
6-7 Results of Auto- and Cross-correlations, Six 6-39
CAMP Stations, 1967.
6-4
-------
A. TNTHTT, •.ION
l.jo purpose of tlut. statistica. analysis vaa to explore the Inter-
relationships among commonly measured photochemical contaminants and
meteorological variables in > urban regions. The yea.. 1967 was selected for
study as this was the year when the most complete data for contaminant
concentration and meteorological variables were available for 12 stations in
the CAMP and SCAN networks. Total oxidant concentration, an Indicator of
photochemical pollution, was selected as the key dependent variable. Both
instantaneous daily maximum and hourly average values formed two statistical
sample populations over which several types of analyses ranged.
A first iteration of the multivariate analysis was completed in four
sequentially interdependent phases: (1) rank ordering of the? days of t'.ie year
according t'o the values of their peak instantaneous total oxidant concentra-
tions; (2) fcaqueney distribution and descriptive statistics on all hourly
averse photochemical contaminant concentrations and meteorological variable
hourly • \Tnra 'e values; (3) correlation and regression analysis; and (4) auto
and cross-correlation analysis.
B. DATA ACQUISITION AND MANAGEMENT
1. Air Quality Data
Th<: statistical studies were concentrated on CAMP (Cooperative Air
Monitoring Program) data for six cities and six Los Angeles SCAN (State of
California Air Monitoring Network) stations for the year 1967, as shown in
Table 1. Tl.rse data were acquired from the National Air Pollution Control
6-5
-------
00 4->
d a
•rt 00
X 1-1
•H V
X 93
2
H
9)
O jd
H W
CO
-------
Administration and the California Air Resources Board in the form of computer
taped. The physical and logical structure of the SCAN tapes were well
described in covering documentation and were compatible to the IBM 360 series.
The CAMP information arrived on Honeywell 400 3/4 inch tapes which
necessitated a complex, multistage conversion process to achieve IBM 360
compatibility.
A significant number of logical and physical tape anomalies such as
variations in card image overpunching codes and an interleaving of 1968 data,
not accounted for in the received covering documentation, were uncovered. A
unified data storage structure and processing strategy was thus necessitated
to facilitate preparation and management of CAMP and SCAN data sources. The
hourly average concentration for six photochemical-related pollutants,
CO, NO, NO., hydrocarbons,methane-free hydrocarbons (where available) and total
oxidant was selected as the basic statistical unit. These data were merged
with corresponding data on six meteorological variables (wind speed, wind
direction, pressure, temperature, relative humidity and total sky cover).
The SCAN station variable sets were augmented by the inclusion of
estimates of mixing height. The mechanics of this estimation are described
below.
2. Meteorological Data
For each CAMP city and for a number of weather bureau sites in the
Los Angeles basin, a directly compatible WBAN 144 deck was .icquii-.I frozi
ESSA (Environmental Sciences Services Administration). Augmenting the
6-7
-------
Los Angeles meteorological data, the California Air Resources Board supplied a
special "dense" network data tape for six Los Angeles weather-measuring stations.
These recorded the diurnal fluctuations of wind speed, direction and temperature.
Computer subroutines were applied to merge these data with the pollutant data
bases, thus, in effect assigning the values to the pollutant stations. For the
estimation of mixing heights a series of temperature versus height radiosonde
curves was constructed from Los Angeles Airport (LAX) data. Interpolated
temperature is used in a computer subroutine in connection with an appropriate
radiosonde curve and lapse rate to estimate the mixing height at the pollutant
station. These routines are similar to those Incorporated in the photochemical
pollution model, described in Chapter 7.
3. Data Management
To manage the data, implement the above data transformations and
place the data in a form suitable for statistical analysis, a system of original
data management routines was devised. This system of programs, referred to as
an Air Quality Data Management and Analysis System (AQ/DMAS), is illustrated in
generalized form in Figure 6-1. AQ/DMAS, which operates on IBM 360-50-67
systems, accomplishes the following:
1. Reads Honeywell 3/4 inch tape, reproduces to 1/2 inch tape and
rearranges bits to be readable on IBM 360 equipment.
2. Deciphers unique characters In Honeywell environment.
3. Transforms CAMP 5-minute-readings to correspond with other
events recorded by hour or other time increments.
6-8
-------
L.A. STATIONS
MERGt SUbROUTINE
I
I
,1
HEIGHT
SUBROUTINE
INTERPOLATION
MET-TO-POLLUTAHT STA.
SUBROUTINE
CAMP STATIONS
MERGE SUBROUTINE
OATA ASSEMBLY
RAW OATA BASE
BY/STATION BY/HR
STANDARD STAT
SUBSET PROCESSING
STANDARD STAT DATA BASE
PP/TAPE - PP/FILE
360/67 BATCH 8 TIME-SNARING
DATA ANALYSIS
EXECUTIVE
"CRISP"
STO. STAT DATA BASt
SUBSEJ7IHG
FIGURE b-1. SDC AIR QUALITY OATA MANAGEMENT AND ANALYSIS SYSTEM (AQ/DHAS)
6-9
-------
4. Sorts and/or reformats data to order data appropriately for
station data analyses and to compensate for overpunching due to
use of obsolete EAM procedures (application of machine assembly
programs); places data in a guaranteed, standard format
compatible with FORTRAN processing; and provides diagnostic tests
and checks.
5. Interpolates meteorological data to the nearest pollutant
measuring station.
6. Computes mixing height from radiosonde data.
7. Herges ESSA-WBAN and other dense meteorological data with pollutant
station data (CO, HC, NO , SO., oxidant etc.), as shown in Figure 6-2.
A fc
8. Provides data retrieval, extraction, subsetting and format
capabilities.
9. Provides standard statistical data base files, and performs
statistical analysis desired. Programs used include rank order,
descriptive statistics, frequency distributions, auto- and
cross-correlations, multiple regression, contingency tables and
others.
C. THE STATISTICAL PROGRAM SYSTEM
The format of the merged pollutant/meteorological pre-processed data
file tapes was designed to be directly accessible, via appropriate structured
control card decks, to a library of statistical programs. These included
the following:
6-10
-------
w — Xtt^i". tt^O'ii' c *** r*f
», r< 4» * >
O O ^ <. C t
*
u. t. c t o i c. u C- o w w c i' c <• f o o c o 0 t c 0 o r u
t- *.. o c- f o «' c c- r» c ft O o p o ft o o o o o r, . c- o c> *-
^ C- 4lr ft C O C? t,1 C C' C W C1 *^ *•* C- r- C ' *- O O '
• *?•*? J,*^
*- ts, tj W * * *- C * V C
S-£H :: j
V i it
•- 1 C X « - Oft4.«****ft
«* t«-C'CL,O*
C*«CcOC4-*'
3 g. O LJ • • *
UO*< *,OCt^tw
o o « * *• o 4 . c o e c-
w. O C- — • *»
X 3 ^ * *
Oc»Oc« • ocr>c o
*" "c?i'n^".r>
< ar *,' r »• *^ **
*j C t —
( w J t ^J V ^' O 3 0 — *0-'0'*--»U'M^O
oooooooc o o c
o» jj c c o t. c^ o o o o o
« sss?£ssssegj
5 B v'-s^.^1:^.^
» « «• v «
« f . V- C,
i* t :» >-. o
4.
* i *• e a •
0 <^ C' C o «
v o c c* , *>
*
•
C C C t O o *
O O o t O j C
» 0 C 0 V 0
(, t. n r> w *-'
!* *
i
f r. o w O O
e c. o j e- o c
o o o c o c c<
« h> N m c fl 0
!8S5i=888
» C. -» « i C
''Of O C i_ «. C
4>
t*
-•
c
c
c
c
c
n
i
o
§
c-
4»
u
4
o
o
*
«
«
ft
i
c
c
o
^
c
•
fc
K
§
& 1
O (
o <
u <
f , b
( «
*
< ,
*
c <
o *
o <
o<
o '
s<
•^ t
c-- 1,
ft.*
c* c*
>c *
4
*
.• O •
i o *
A *
i *
*
* • • r
?ss
• • *
"4 ^4 N
?*£
,£&
1 4 *"
> 4.' K
« *
e. «
c <
.
t* &
00
eg
Oo
o rt
c r
S8
•J O
C' O
SB
ti
0
o
o
c
c
o
r
c>
o
c
o
CJ
cv
r
C
*»
«
*.
t
-,
O
o
o
c
u
c-
o
?
c
o
0
*
«
£
~
f
C-
o
c
c
f-
r
0
4
«•*
Cj
r*
*
.rr-e» *»***«
*
CO |T, O
•^ j- »—
c
c o c-
o c
o c -^
c o
O C' <-
^, f J *.
o r '3
C (- c
000
•J iT> IN
O O t
O • J O
o r* c> ,
e o f s
O C C1 I
4 IT « ^
•00********
»cc** ******
?* * N
c c o
0 C 0
o o o
C' O O
Cl C O
f kT f•^
c i «.•*•
> . i O
0^§
C «4 fVl
030
o o o
• • *
f- f- f
,.',»?
one
^ o o o
^ -c h-
6-11
-------
1. Rank Order Analysis
A rank order subroutine was developed and applied to identify "high
oxidant" episode days. In order to reduce the possibility of dealing with data
which do not reflect actual photochemical activity. The subroutine ranked days
by decreasing levels of average oxidant, peak oxidant, average N0« and peak N02,
and gave the month and day of all values and the hour the peak values occurred.
This subroutine was applied to each of the 12 air monitoring stations being
studied and served, along with the frequency distribution, as the criteria for
the selection of days for more detailed statistical analyses. A sample Rank
Order Analysis is shown in Figure 6-3.
Days were selected primarily on the basis of the peak instantaneous
value of total oxidant concentration which was greater than or equal to a
prespecified value. The value selected was 7 pphm for the CAMP Cities (Chicago,
Cincinnati, Denver, Philadelphia, Washington, D.C., and St. Louis) and 14 pphm
for the Los Angeles SCAN Stations (Anaheim, Azusa, Burbank, Downtown Los Angeles,
La Habra and Pasadena.)
2. Frequency Distribution and Simple Statistics
Frequency distributions and simple statistics were computed for each
data element defined in Table 6-1 at each station. ( Each such frequency
distribution consists of a table of observed frequency versus the numerical
value of the variable. The sample statistics are, for each variable, sample
number, mean, standard deviation, low and high. Figure 6-4 is an example of the
frequency distribution table for hourly average total oxidant concentration
6-12
-------
O1 CO
^ x a
< a.
u> a
(X **
*-, rg r-» »-i P-* *^ fv
-
OX I >
* X. •f uj
ue a
•i> a
C! >
UJ
in TJ
X >
o -H .H
1-1 U
J3 O G
D 4J 0)
<0 3
M O*
OT ai a)
•H T3 W
X M ,0
H O 3
0)
c
• i-l 0)
r- ^H
SO -
o\ cMtn
rH O -rt
2 W
B-g1?
w « c
5
O
<" -^ o
o- -c ^ o* o •-« •*
u.0
*J tJ
c a)
e
91
O
C
O
o
a.
I >
M
C—
51 u.
i >
O 0. J- 0.
« O IM _J
st-
oo
x a a
fej i-l O
^ CO
3"«
M PS
• O (0
o-i ai -a
I T>
vo a>
XT)
OJ ^1 O
U 00
3 o) v*
M >» a.
•rt (0 «
tu -O =
itO-C1
<• m -t
6-13
-------
*
n
o
IM
•
u>
I -4
7
a
UJ
a.
_1
3
t—
c
-" •*
»
z
a
te
o
l~
!>-
•0
in
o
a
-4 m 00
OO
oo
OJOO
-lOo
£000
OOOOOOOOOOOOOOoOOOOoOOOOOOOOOOOOOOOO
oooooooooooooooooooooooooooooooooooo
OOOOOOOOOOOOOOoOOOOOOOoOOOOOOOOOOoOO
OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO
000000000000000000000000000000000000
a
o
O
^
0*
2
<
L
«j Q
<*» X'
6-14
-------
measured at the Pasadena SCAN station. Figure 6-5 is a statistical strnna/y of
all pollutant and meteorological variables measurements (.;r computed v^timates)
at the SCAN station located at the U.S.C. medical school.
3. Correlation and Regression Analysis
A multiple regression analysis program was applied to the dal.i for
high-oxidant days for each station (high oxidant days defined by the episode
criteria mentioned previously, namely, oxidant 14 pphm or higher ...l SCAN a tat Long
and 7 pphm or higher at CAMP stations). In this application, the dependent
variable was related to maximum hourly oxidant, while the independent variables
were based on concurrent and immediately preceding values of contaminants
(CO, NO, N0?, total hydrocarbon, non-methane hydrocarbon) and of weather
parameters (wind speed, pressure, temperature, relative humidity, sky cover and
mixing height).
Logarithmic indices were used, rather than the directly observed
values, on the premise that frequency distributions for these variables tend to
be lognormal rather than normal. Further, in empirical studies relating to
variables of this type it is common to use power-law relations and multiplicative
combinations, both of which are compatible with linear regressions on the
corresponding logarithmic indices, but not with linear regressions on the direct
values. The dependent variable was taken as the common logarithm of the daily
maximum hourly average oxidant, while each of the independent variables was the
mean logarithm of three observed values, viz., the hourly average v- Vic for the
hour of maximum oxidant and the hourly average values for the two hours
6-15
-------
.» IM c. o o o <> o o o o o o
~OOOOOoOOC.iOO.oOO
z .o r- o o <:• o C* n o <•• o f- o o
O
o
2
o
X
o
o o o
a a a oaaaoouaoo o
I O O O OOOOOOOOOO O
OOOO OCOOOOOOOO O
-jooo o o o 10 o o o r- o o o
O O O OO'">OOOO'">>no O
OOOOOOOOOOOOOOC3OOO
I/)
a
z
<
o
in
eg
-^ ^ O
^- (*1 9-
;g§,
0»
OB
o o
-I 1*1 0^ •* 1
r.
U U
t-
u
Ul
T
O
oc
Q.
fj m in o M< in CM -..-. — ., wl.
oomomo^f t<\ -s 13 •$ & & w
5OOOOOO10 f*\ m ^« .f fM M tf\
Orgrsjomm-t ovMr^^in (M
in
U
o
i
a
.
-0 ^ -C -0 O m -i <>-t
^iwr»r>mM^4 iNa-
05 m oo co « ^- p- ar-
coo
< -i
M &
0<
00
•t
ui
>
O
«5 «
o.a.o.CLO.^x.oLUJatZ2«* o
o.a.aa.Q.c.-Q.rc3i:— tuM —
• at ac
. • vl 3 iu
j ac o i^ t- • >
< H- • — Q. uj < r a
o < i
>- a .
o.
a . a. •
3C»
i o
M co j: u a o
H U -U cfl » M
eu -H cd u w jo
M 4J O CO gO O W
SO 4J W M-l
cx to r o
0) -H -HO)
IH u • Ja co co
3 o «fl w a G
M co w v cd
•H CU cd O T3 Q)
ti, Q -o u : s
6-16
-------
immediately preceding the hour of maximum oxidant. In effect, the question
being tested was whether maximum oxidant as observed on days of high oxidant
could be related in a statistical sense to other contaminants and weather
parameters taken as multiplying factors.
As illustrated in Figure 6-6 and 6-7, the program generated regression
coefficients and related statistics for the entire set of independent variables,
then identified the best subsets of the independent variables for all possible
subset sizes and performed equivalent computations for these best subsets.
Coefficients of correlation between the dependent variable and each of the
independent variables were also generated.
Results are discussed in more detail in subsequent paragraphs. They
showed, however, that in every case a subset of three independent variables
could be found for which the regression was significant at the 5 percent level
or below.
^• Auto-correlation and Cross-correlation
In auto-correlation, a set of values of a variable is arranged in
chronological sequence, and two series of values are obtained by taking initial
values for two different starting times, each followed by all successive members
of the same sequence. Auto-correlation is the determination of coefficients of
correlation between such series of values, and the difference in starting times
of the two series is termed "lag". A value near +1 indicates that the difference
betwoen values in the sequence, separated by the specified lag, ia -"'dinarily
small relative to the range covered by all the values in the sequence; that is,
6-17
-------
r >-
-J Z
• u.
X C
4 2
Z UJ
x a
C uj
c
y
O
c
UJ
1- I/I
O IX
UJ «!
U'
^ ^~
• z
lu
r- C
•0 Z
I U»
r o.
o o
< c
•J ?
^* •••
X
u u.
C
0
(
uj
K.
•M
I/I
U
a.
9
«j
V
Z 2
C. uj ••« •"<
— ~ 0 0
"* ° ' JL
Otju^-^'n.ni^c ««"i
uj o m o> o J tr « f-
CC O * fM I" a 1*1 • fi
O O OO O 1 U
II II
i
_i :
< a. i
— U. d
h- Ul O
a O 1
4 o a
CB *fl* *O *f ^ rv *t ^
Ci •M^fOl^f^^'t^
c^.**^ (Ntn^TMO
f^ ft» UJ <"! f^J 0* i^ V f>i *4
•f (/•£,''-<«'-«»P^d4-
t 1 O O u C
1 III
1
T
t- of
" J 1
X >
ft' Q IT m r-
f" ."^ f* pxl 1ft CT1 ij> (O f+
U 1, ^ ISJ Ct- -O C —
t o c c c c c
.
,
c c-
d at - o
< »- 1
c « o
<>U •^^^•tfs^O
>- 111 <0 (\j in O TO »•' rg
^1QODf\iXf«>9'^«^
(» ^ ••< »l d CSJ d
IO C O O OO O
i i
! z1 ;
• ivj o "^ *o cr* co *^ en
I C^Mf-KN*
*t or K r ^ ^ p
IT r- P* • eo • •
! • . • -• • M t-i
o o «. o
0- j '
1 J
; OP
au 1
5 1 :
-*i
'
! UJ • V^ L> •
1 uj' c or i- a CD
r •- nr r- 2 •
— X < i- IA o
i » u. c. C y >/> >
7) C •- C • u u,
r •- C ef C- a u
. c^ x r ui c D
' x o ? > ui t-
cc u. 3 a * «/
in i' i_i « 4>. C1 a
•- •- r j — uj
or a u < C »- o.
r- t- i- »• r < *
«i •- »- C — »- w
o r x t- x v.t-
•»«.
CO
2
«» "-4 ^
O <^
00
1
n
t-
o.
Uw
o
nr
r-* ^ ^
f-4 lO Z
^ O* d
l»l«>
-
o o
1
••4 r4
OC
i i
c c
<*\ o
IM in
•* o-
r~ co
00
1 l
o
o
o
o
<00- ,4
m m »
O03 O
«• in
" o-
•J ac .
0>Oo
• rxj
d •
O
1
cc o e
i
• a
c w
— > K
* C Z
TO «t
T C
>. ,-.
U • il! "»C
> v» C
^ -* *t
oe
X « •
OO
U. » CO
^•cr IN
O- O C'
•o m n
i/t<->m in K
cc c m f-
co CD * «t
C *f IT
• • • •
C OC C
UJU. p- V-
I- U. I- if
i r o s —
I «- O •— O
S.5E
I UJ rr u. u,
n t- uj
u r^ v C"1
| C OO U
1 « • u- •
C »- r a
d. _j o.
a r K O
u'y ^ o
UJ
C Z — UJ
a uj u J
? su"-
z r u, h-
< o- "• -i ;
t-ios '
(/• I/)O 1
6-18
-------
tn **1 w> <« ! 1 i/»
If a a o i • a ;
ui ' U* 1 1 UJ
J. a »~ o i i- t
_- ' »*
•a . ; 'i- . .
rf> ' •{? 1 O , » O !
N 0-5 I •» , ft.00 , 00 •
*^ ' * ;
' « • ill
a ! 0 • !
d ' -
**: ** '
f. lf\ ,O -O 1 IM
i 4> 1 . <«• 1
-• I (7P 1 [ • _< j
.* 1 i IM *
• 1 ' • ' ' • ' '
' 1 *
o •_ o . o
N 00 1 o (M
t * I O f- 1
f\f- ^^ i |
r«o • ! ir> o
* ma o n ' t/io o ; n
Jt • « T « • ' 1
>• 00 i~ 0 O | 1- '
a. c. ' 1 a i
UJ 1 ' UJ UJ 1
U 1 U l U i
* J °" «
UJ' «• 'UJ !.. Ujl'
t- U.IMIT. ^ • ij.fr> ^ K l
r C ir> z , Q if, 1 z |
*" ! *-• i ' *H I
> ' >• i »J
I 1 , i 1 .'
if> K IM ' I op xf (X j
! •*• IA w ' '*->aOO>' •
T ? £ moo I
rv N •!• IM 1- IT „ f.
V\J ift Irt^^lT. V) IS-** *} at 1/1
u >/»••• u, vn • • • u, '
u OOO U. iQOC. u'
Ut , IJJ 1 UJ
!o "!?"" Sr- «c ! UC. J f-0
NIM«"* -IT. Nr4 ITSmC'GC
,*£*"» ^a»IVJ 1 r_4Nvj-g
V/jfiiMTi . ^,Ooro.r- ,,j Ctat^,j.rg
~"<'ri TiC;/ -.-(•(. « 7^.-J ' — — O-CINJ
i^»« C. <»*t ifl... --?•*• I/I....
!fV" '''^"^ L' ^oro UP— ni- ,o v^or-oc
U' 1 o>" — C' »* UJI | Ol/ — O 'M luT 1
^,-f y^p-i- —y eiii-i^- — ry
, O 3U-«» IX« O 3u. < t'. c">
« «J C t> — uj r jf — iu
JJ oc >^f > t- af wit" > t- .1
^ U -u U tu IU U- I
•J ' tX C- (A a c .1^
; • a.. IT
1/1 1 =• ' i 3
•^ v>
^ . I *" trt 1
"• : i "• u. ,
"• ! u. u
w ' PJJ ! PJJ !
° 0 • 0 ,
" 1 ' ° ' " i
* i oc ; ct ;
<•" ' (.•> ' e
'•'?'/• i i uiir 0 * ' a. 0 01 rnivj
n "" <-• i uf r< O T I c. (>• 1 i
'• .... ,» ].... ^
c . oo co , c. 'c.o'n. Q
'
£' ! o c
K 1 p— 1 (_
1/1 ^ ' I/1 t ^ «X1
* ** . f
U. U. i. t- 1 U u. >. »-
i- " •• r ! >- 11. P- r ;
* UJ . I. -J |, ..j i, |
•»rjr— icr- —
" — u — i ' >^ — l__l) w
L (-5— Ui >-7^ u
-JIN^- ^.^|IL _J(M*-tf^ U'.OU. _jCV-r»*'^l"-
"•• UJ" '•- "• 'ri u.(v uj u , er.
": f* * L- *" u P- i* i -i ,
u r u r — in- c ui r- i _ 1
•', • c \- <.i >~' 1 u • • ' r o c w 1 n i. . •
J " ' •• a n i, , , a a,
f' cr«n» >/>-«/>. u* >•-!*
" r t- . f *' r i~ i. •- n x '•/ f ~
! " <-< u — i •> u in*. »:._!,- u 11 r <_ »
' f c ar^r tori. .r-^t- ri---cu
— o u. s . • i_> — o [. i : .- L> — t. ir u
•• »• o n » x r. ^ ui »- x t- -
[Vr'>- C^—u UJOXl- C.I- — U. Ut7>»-
^1 CX li. v.^ ^ 1/1 I ^| u u. C- „* */, t . T «1
u -Ci o u u *• — ii_ p. «i (t u n ^
' * » «•»'_! r •- t- n •• ,, u _j 7 i- > i- >
• • .. r i- i '. r — — 3 u p j ' j r •• — — r> in
"»» i/ii/iii» a >r •- i_ i/ 1/ i.. r U.7J-1-K
IM |
0
1 !
c
•~*
m
CO
O IM
•" 1
, '
O '
2 i
*M
M
U. •
in
Q O
a. • •
Ci 0'
J
u -4" i^
Q ifi
IM O' IM
O O' O
—> o- u
•O 0 !*•
bO *-4 4 b .
«/)•••
O C^O
1
1
t
«? l/i -J
r. z «
U -, C 1"
l_. ^, «- c.
n l/> »- *—
T> -iJ <
OOC ~
f!U >
UJ UJ
or Ci
i
U" f*- 01 O*
i, L <•; -.r
f** <"NI ?**»•«
« r t- <«>
C' -4 fv IT
C f C' t1
i
_ J
It U —. i>-
»-- u *- r*
'* L ' < L
^•- C.' -• • l '
•V V "••
I/ • f, II
t (l U U
O h I
1) •— \i f~
C1 1 r '--1
•t • u h
*< fi ti- _j
>- i f r
i/1 «j/i c. s
r
o
9
H
U
W
nJ
U]
w
CO
t-1
^;
a
M
a
o
w
M
Ou
w
b-.
O
to
M
M
>H .
p-4 M
5 w
j
§1
M •<
C/5
ui oi
M M
gg
s°
tw
J22
»>. 2
i
^O fi-,
o
OJ
M CO
3 i-l
00 CO
•H <
Cb M
-------
that a high value is relatively more likely to be followed by another high value
than by a low one. A coefficient near zero, on the other hand, indicates that a
high value is as likely to be followed by a low value as by a high one, while a
value near -1 suggests that a high value is much more likely to be followed by
a low value than by a high one.
When applied to hourly contaminant data or weather data, high auto-
correlation coefficients show that hour-to-hour changes in the measuzsmeat
under study are usually of smaller amplitude than the diurnal cycle as a whole.
To provide sequences of hourly contaminant data for auto-correlation
analysis, the output of the rank-order analysis for each station was scanned
and periods of three or more successive days of high maximum oxidant were
identified. As an example, the four days from June 2 through June 5 provided
such a period for the Chicago CAMP station; the output of the auto-correlation
program is shown in Figure 6-8. Lags of one hour to five hours were studied,
for each of four contaminants. At lag of one hour, coefficients for nitrogen
dioxide and oxidant were near 0.9, while those for nitric oxide and hydrocarbons
were near 0.7. At lag of three hours, nitrogen dixide and oxidant coefficients
were near 0.5, while nitric oxide and hydrocarbons were near zero. This sort of
behavior is probably to be attributed to the status of nitric oxide and hydro-
carbons as primary contaminants, more subject to rapid change because of
relatively small shifts in wind direction or in community activity from hour to
hour, while the accumulation or dissipation of the secondary contaminants,-,
nitrogen dixoide and oxidant, is relativfejy insensitive to these shifts.
6-20
-------
'
,11
0
.J
*
II
u
J
II
0
3
eg
H
0
*
-
ii
(J
*
O
II
o
-
ir»
•o
^
o
o
1
o
CO
0
CO
<-»
o
1
"
OJ
o
<-»
"
PI
r-
tr
•—«
PI
0
o
PI
UJ
•£>
sC
"
CJ
0
0
Cȴ
I.)
*
UJ
LJ,
•— 1
>v
cj
0
•~
'X
«w
•o
•"
^.
UJ
SJ
**"
2
p-
_
•^
INJ
n«
^
_L
-O
f^»
s^
— «
(NJ
UJ
^,
n>
P>
»jr
O
in
O
O
««4
»J-
IT
-o
o
j.
OJ
V
u\
cu
"
C,)
u
o
^
t^>
-
UJ
CJ
«l^
X.
O
»••«
U
*z
0
u
Cf
^
P»
~
^^
!«•
r~
*^
~
^~
*••»
«~
UJ
r-
«.
^^
(VI
UJ
-
«^
UJ
JO
-_
r\
•••4
(k.
•O
o
o
c;
<_j
c^
o
•— i
UJ
a
«^
t_j
LJ
a.
t.)
>
J.
_J
*^
a>
00
•*
^ .
o
cr
**
~
0-
^^
^.
c\j
o»
«.
m
u*
_
.~*
a ^
cr
_
eg
HJ
^p
CO
rg
u
rn
a)
UJ
u>
f»
^
r-
o
•g
»T
u>
«
<—•
»•*
u*
SO
o
u
o
c.
o
0
-
H*
^_
*4.
LJ
»M
X
0
_J
«
j_i_
*•<
gj
_^r
**
V
>o
^,
I
bo
rt
O
00
vO
oo
6-21
-------
(In fact, records of continuous monitoring of these contaminants frequently show
rapid changes in nitric oxide concentrations near peak traffic hours.)
In six other episodes at various CAMP and SCAN stations, the same
auto-correlation behavior was found for oxidant, with coefficients for lag 1 hour
ranging from 0.85 to 0.93.
Cross-correlation refers to the determination of coefficients of
correlation between levels of different pollutants taken in corresponding time
sequence, with or without the introduction of lag. Thus with hourly values for
two contaminants, the existence of a coefficient near +1 suggests that the
diurnal variation of the two is quite similar, while a coefficient near -1
suggests that they vary in reciprocal fashion.
D. RESULTS AND DISCUSSION
The statistical analysis completed in this project represents, in
effect, a first exploratory Iteration on the data. The primary objectives were
to: (1) identify photochemical pollution episode days; (2) explore the ability
of the contaminants and meteorological variables to account for photochemical
pollution episodes, as measured by oxidant; and (3) establish the relationships
that exist among oxidant, hydrocarbons, NO and NO.. Analyses (1) and (2) were
undertaken first in order to identify and validate the sample populations of
interest. Analyses were also conducted for (3) in the form of correlation,
regression and some contingency studies, the results of which are reported in
this chapter. However, in view of the extensive analyses which are now possible
on the established data bases, this analysis is less than exhaustive. It is
6-22
-------
also incomplete with respect to the SCAN network, as five of the eleven .stations
were not included in this study.
The atation-by-station results of these analyses are described in the
Appendix I of this document. A summary of the results are shown in Table 6-2
and the following.
1. The SCAN Stations as a Set
With respect to information obtained by rank-order analysis, the six
Los Angeles SCAN stations studied exhibit a good deal of similarity. For
SCAN stations the modal months of peak instantaneous oxidant concentration
episodes are July, August, and October. Peak instantaneous oxidant concen-
trations are most apt to occur during the time interval between the 10th and
15th hours with probabilities of around .90. For all the stations the modal
value of observed oxidant total hourly average concentration is 1 pphm.
Yearly mean values range from a low of 2.84 pphm for La Habra to 5.4
for Azusa. The range of interstation observation data variability may be
characterized by a sample standard deviation low of 3.94 pphm for Anaheim to
a sample standard deviation high of 7.12 pphm for Azusa. More detailed study
of the relative levels of the contaminants it the SCAN stations yielded
information bearing on some proposed hypothesis or generalizations as to the
relations between contaminants, and their geographical variations.
Frequency distributions of the hourly values of all variables were
generated for six of the eleven SCAN stations: Anaheim, Azusa, Burbank,
Los Angeles (DOLA), La Habra, and Pasadena. Non-blank values were available for
oxidant, carbon monoxide, nitric oxide, nitrogen dioxide for more than 85
6-23
-------
a Is s
ll'sl I I
U U > *H j;
s a i s j s
0 o a K « »
I ?3.
3 i "A
I I '£
8 A| .Ǥ}
5 8 "SS
«a JK
s 2 J:-s
r-
vO
co
§
H
I
H
to
M
a
W
CN
VO
(U
fH
X)
6-24
-------
percent of the 8,760 hours of the year (1967) for all stations except La Habra;
at the latter station, nitric oxide was known for less than 15 percent of the
year. Total hydrocarbon levels were available for more than 85 percent of the
year for Anaheim, Los Angeles and Pasadena, and for 25 percent of the year for
Azusa, but were missing throughout the year for Burbank and La Habra. These
frequency distributions (except for La Habra) are depicted in Figures 6-9
through 6-13, showing values of the 50, 75, 90, 95 and 99 percentiles, and the
highest value, of the cumulative distribution of each contaminant, derived from
computer output tables of the type illustrated in Figure 6-4.
As illustrated in Figure 6-9, oxidant levels at Azusa exceeded those
at the other four stations at all percentiles above the 75th. Oxidant levels at
Burbank were appreciably lower, and those at Pasadena still lower. Lowest were
levels at Los Angeles and Anaheim, which were nearly equivalent throughout.
(From the fragmentary data of La Habra, not shown, levels there appeared slightly
lower than those of Anaheim and Los Angeles). Figures 6-10 through 6-13 show
that Azusa levels for the other contaminants were in each case either lowest or
near the lowest among the five stations. From this it may be concluded that
a relatively hinh frenuency of occurrence of high| levels of the primary traffic
related contaminants is not required in order to produce a relatively high level
of development of photochemical oxidant at a given location, even within a
single air quality region such as the Los Angeles Basin.
A concise overview of the relative levels of thest. contaminants at
the five stations is shown in Table 6-3, which shows the rankings of the stations
in order of descending values of the 99th percentlie, for each contaminant.
6-25
-------
vo
X
o
1
Ul
>
>•
_l
I
2
LU
oe
I
VO
cs
'NOIitfaiNBDNOO
6-26
-------
vo
x
o
o
o
CO
o:
-------
5 «*
*t —i
-------
SE
o
CO
vo
en
x
o
o
H-t
Qi
CtL
LU
-
_1
a:
o
U.
O
GO
O
t—*
h-
CO
K—I
ac
O
LU
CD"
LU
CM
OC.
iNOUVHiN33N03
6-29
-------
6-30
-------
Table 6-3.
COMPARISON OF FIVE SCAN STATIONS BY RANKING
RELATIVE TO FIVE CONTAMINANTS (99th PERCENTILE)
Ox
CO
HC
NO
NO,
Azusa
Burbank
Pasadena
Los
Angeles
Anaheim
145
2 1 (1)
324
433
552
5 4
2 1
3 3
1 2
4 5
6-31
-------
(With respect to hydrocarbons, Burbank is tentatively assigned the first
ranking on the basis of its high levels of traffic gas constituents as indicated
by carbon monoxide; this assignment is uncertain, however, in view of the lack
of complete agreement in rankings on these two contaminants at the remaining
stations).
When these rankings are compared, it is seen that there is reasonable
agreement (parallelism) between carbon monoxide and nitric oxide; between carbon
monoxide and nitrogen dioxide, and between nitric oxide and nitrogen dioxide.
On the other hand, there is poor agreement or none between other pairs of
contaminants, and there is substantial disagreement (antiparallelism) between
oxidant and nitric oxide rankings.
Some particularly interesting examples of disagreement (i.e., anomalies
in terms of a hypothesis of parallelism in contaminant rankings) are the
following:
(1) Azusa had highest oxidant but lowest HC and NO, and low CO and NO..
(2) Anaheim had high hydrcarbons but lowest CO, Ox and N02, and low NO.
(3) Burbank had highest CO, N02 and (probably) HC, but only second
highest Ox.
(4) Los Angeles and Anaheim, although having low and very similar oxidant
levels, had widely different leve's of NO and NO..
This brief listing of important discrepancies, in effect, suffices to
demonstrate that annual air quality with respect to photochemical oxidants at a
selected location within the Los Angeles Basin cannot be readily deduced from a
6-32
-------
knowledge only of the annual levels of the primary contaminants at that point.
As a corollary, predicting the oxidant levels at a selected location requires
consideration of the daily history of air arriving
at that location, that is,
history in terms of the sources of primary contaminants that accumulate in it,
and the reactions that take place between them while the air is en route to the
selected location. For this it is necessary to consider the real paths of
motion of the air - an approach which is implemented in the photochemical
environmental simulation model, described in Chapter 7.
Correlation and regression results are summarized in Table 6-4. Of
the 9 or 10 independent variables available, carbon monoxide yielded the highest
individual correlation with the dependent variable (maximum hourly average
oxidant) at two stations, Azusa and Burbank; coincidentally, these are the two
stations for which oxidant levels were highest, as discussed above. For two
other stations, nitrogen dioxide yielded the highest correlation; for the
remaining two stations mixing height was more strongly correlated with oxidant
than was any other contaminant. In the best subset of three variables, mixing
height appeared for five of the six stations, with Los Angeles the only station
for which this was not true. Temperature appeared in three of the six best
subsets of three, and carbon monoxide appeared in the same three. Nitrogen
dioxide appeared in two, as did relative humidity; nitric oxide, pressure, and
sky cover appeared in one each.
It is interesting that the best subset of three independent variables
proved to be identical (carbon monoxide, mixing height, and temperature) for the
6-33
-------
vO
01
u
c
91 -H
•0 01
•H >
U «
a -i
3
t 3-var table
subset
a
£
87
•H a
U iH
(4 4
-3 S
S «
M W
S3
«
u S
s-s
*3
33 ^
M M 14 14 M
OH ON ON ON ON
ON ON ON ON ON
ON OS ON ON ON
43 r*. f»* s
00 W £ J= 1- *J
•H -H 00 M 3 -^t
v •o *H *H n * Kit Kit CM >>
8 -4 -3 M •P W O lia-rl O
u B» B J Z ouu
• a a a, -H «j
tsu . g . B - M-HX
Soi 08 08 Q -3 ai ji
U UU UU Z XklM
x%
4J
X
s
£
~ ^ ?
O4 ^S *«* (N| «H
§ § § § i
•**
Soi oo o
in ^
-------
tft
W
^d
3
3 u
•3
u
Is
o a
82
*•£
•
4) «
j: «
5j o
a
o
u
-t to «•
S? 9 5 „ S 3
i-t r-l <0
u w ig a a n
<9 Iv O
•p -g u. u u u
S « 3 U 3 3
a« o. 4 u nj «
01 Tf CJ 0)
T> "p 5» a. o*
c 2 e -g a a
•rt -i S c 5 S
> 5 w .* u «
3*0 00 * u"i Ov
oo oo m oo r»
00
<8
0 « ^ - - S
00 it) 00 00 00
(0 ^4 <9 R) « U
*J
10 « (B 10 <0
H W
i i s s s 2
r»» i-T (N O sO *^t !
st *Tt U^ *ff ^ **
r i* r r r r
i
•H 00 \O i* O *O
in r- (N r*. *n 00
00 00 &t 00 00 00
u
V
r-t
00
I
S J! § 2 §
>r4 G i . ^ 0}
O « «9 *- 01 T3
£ in ja c I n
« 3 u 5 «
3 n 3 o n) n
< •< oa o i-i a.
6-35
-------
three stations (Azusa, Burbank and Pasadena) having the highest frequency of
high oxidant levels. For Azusa the shrunken multiple correlation coefficient
was 0.77, indicating that these three variables could account for about 60
percent of the variability in maximum oxidant for that station. For La Habra,
a different set of three variables accounted almost completely for the observed
variation, but this is probably an accidental result of the extreme paucity of
data for that station.
Table 6-5 summarizes the main results of auto-correlation and cross-
correlation analysis. Oxidant auto-correlations exhibit the behavior described
above, ranging between 0.85 and 0.92 for lag 1 hour. No noteworthy consis-
tencies appear in the cross-correlation results, except that the highest coeffi-
cients for oxidant against other contaminants are always negative, while those
for oxidant against wind speed and temperature are positive. No explanation is
known. However, in the case of high correlations of oxidant against wind speed
(with lag) at Anaheim and Azusa, the two SCAN stations farthest from the center
of Los Angeles traffic, it may be suggested that the daily arrival of the sea
breeze at the station foreshadows the arrival of photochemical oxidants in
polluted air transported from centers of more intense emission of the primary
photochemical contaminants.
2. The CAMP Cities as a Set
The photochemical pollution potential of the CAMP cities appears to
be rather similar if the statistical properties of peak instantaneous and hourly
average values of total oxidant concentrations are taken as criteria. For all
6-36
-------
CAMP cities high prak oxidant levels are most likely to occur during the summer
months between the 9th and 15th hours. Yearly mean values for the hourly
average total oxidant concentration range from a low of 2.57 pphm for Washington
to a high of 3.44 for St. Louis. Standard deviations ranged from a low of 2.02
for Chicago and Denver to a high of 2.3 for Washington. All CAMP cities recorded
a modal hourly average value of 2 pphm. Fewer thin 10 percent of all hourly
values in each city reached 7 pphm or more.
Regression analysis produced predictive formulas yielding, at the
worse, between -37 percent and +58 percent relative error with 95 percent
probability. Of the 8 or 9 independent variables available, which did not
include mixing height, nitric oxide occurred in five of the six best subsets of
i
thr<->e independent variables; carbon monoxide appeared in two, total hydrocarbons
and non-methane hydrocarbons each in one. Again, temperature occurred in five
of these best sets of three variables, wind speed in two, and sky cover in one.
No single, set of three variables was best for more than one station. No best
subset yielded a higher shrunken multiple correlation coefficient than Denver
(0.57), for which the regression accounted for about 30 percent of the
variability in hourly oxidant values.
Auto-correlation analysis showed uniform high correlation between
successive hourly total oxidant averages. Cross-correlation analysis showed a
strong negative serial correlation between total oxidant and NO as expected.
Table 6-6 and 6-7 summarize the findings.
6-37
-------
>*4 Itf
§5
O J
Oi
ON
»<
<7<
01
e tj -i
«i « o
J< »-4 •*
e u w
3 U «M
p U »
JB O O
en o y
VJ
§
I
vO
"3 B
U -3
(Q *J
T3 tO
B U
(0 t
U U
W) CO
0)
i-H
jQ
nj
•~4
Ui
O
1 0)
.£>
w 3
M cn
ftj
&
a
•H
C
O
iJ
flj ,-K
iH 01
0) 3
1^ «-*
S 3
U
tA 3
•£ O
00 «
CO
0)
W
3
«J
g
*
8
^
41
O.
0)
•g
•J
o«
Csl
o
iH
£
3
U
-------
no u no
a a t
o
eo
00 » 00
« ns n)
Oi
O
V
CO
O •
OS r~
U vo
Q rH
1 ^
O O
H
P-. CO
C
W
x
ffs
•H O
O 'H
00 O
•-4 Wl
3: o
O (Q
4-1 rH
O 0)
a, u
O
sv
-------
3. Comparison of the SCAN Stations and CAMP Cities
Comparative information from simple statistics on oxidant hourly
values for the SCAN and CAMP networks were given in Table 6-3.
The following inferences may be drawn by comparison of the results
shown in the tables:
• In general, the oxidant highs at the SCAN stations are one to
three times as large as the largest CAMP city oxidant high.
• The SCAN stations exhibit much larger interstation and intra-
station data variability than the CAMP stations.
• Mixing height plays the role of the most frequently appearing
variable in the "3-variable best fits" at the SCAN stations,
while temperature plays the corresponding role at the CAMP
stations, where mixing height was not available.
• The shrunken correlation coefficients corresponding to the
3-variable regressions are higher at the SCAN stations.
• Total oxidant auto-correlation functions are very similar for
both networks.
• There are significant cross-correlations of total oxidant
with other meteorological variables for both networks.
• Hourly average oxidant during selected episodes at some SCAN
stations showed high positive cross-correlation with wind
speed, whereas this was not true for the episodes investigated
at CAMP stations. In the latter, cross-correlation with wind
speed were mainly low or negative.
6-40
-------
4. Additional Analysis Possible
The initial analyses that were planned were completed within the
resources available for this purpose. In general these analyses do not con-
i
stitute a full exploitation of the analytical potential inherent in the special
data bases that were constructed. Further analysis, especially discriminant
analysis, appears to be indicated, with emphasis on the SCAN data. In
particular:
1. Frequency distributions for high oxidant days should be compared
with those for low oxidant days, to determine where the significant
differences in the variables occur.
2. Frequency distributions should be obtained by hour of the day, and
the values of all variables of interest should be retrieved and
studied.
3. Data on non-methane hydrocarbons for Los Angeles should be collecte
and utilized in this study.
4. More hypothesis formulation and testing should be performed. In
particular, the relationship among values of contaminants in the
morning and varies of oxidants during the day should be explored,
as suggested _n i. pter 5, Air Quality for Hydrocarbons, (OHEW,
PHS, EHS, NAPCA, March, 1970).
5. The degree of agreement (cross-correlation) between stations in
the same air quality region, and the degree of dependence or
independence of the variables should be established.
6. The extent of the smog cloud should be established.
6-41
-------
CHAPTER 7
DIGITAL SIMULATION OF PHOTOCHEMICAL POLLUTION
Page
I
A. INTRODUCTION 7-4
B. THE REACTIVE POLLUTION ENVIRONMENTAL SIMULATION MODEL (REM) 7-6
1. The Control Program j 7-7
2. Meteorological Module 7-9
3. Ultraviolet Module 7-9
4. Transport Mechanism 7-10
5. Source Emissions 7-10
6. Diffusion Module 7-10
7. Chemical Kinetics Module 7-11
8. The Numerical Integration Module ' 7-1]
C. CURRENT STATUS AND DEVELOPMENT HISTORY OF THE MODEL 7-12
D. MODEL VALIDATION 7-13
1. Thirty-One Step Mechanism 7-13
2. Chemical Species Utilized 7-17
3. Trajectory Simulation 7-20
K. APPLICATION OF THE MODEL 7-26
7-1
-------
CHAPTER 7
LIST OF FIGURES
Figure Page
, 7-1 . . ., General Model Architecture 7-8
7-2 Station Locations and Simulation Trajectories 7-14
7-3 Contaminant Concentrations vs Simulated Time Lapse, 7-24
Runs 1 and 3.
7-4 Contaminant Concentration vs Simulated Lapse Time, 7-25
Run 2.
7-2
-------
CHAPTER 7
LIST OF TABLES
Table Page
7-1 31-Step Reaction Mechanism, Reaction 7~15
Rates and Species Numbers.
i
7-2 Species Involved in 31-Step Mechanism. 7-18
i
7-3 Initial Conditions for Simulations. 7-21
7-4 Photochemical Model Calculated vs Observed Air 7-22
Pollutant Concentrations.
7-5 Calculated Photochemical Contaminant Concentrations 7-28
for Lennox-Pasadena Trajectory, with Varied
Emissions Input.
7-6 Calculated Contaminant Levels from Simulation Runs 7-30
with Modified Emissions Input, Arranged for
Easy Comparison.
7-3
-------
A. INTRODUCTION
Air quality models are tools that can be used in decision-making,
particularly in the areas of planning, implementation and evaluation of air
pollution control programs. Because computer-based models can handle many
parameters and data files and rapidly perform computations, they can be
conveniently used to analyze and predict pollution trends and episodes, assess
the effects of pollution damage and alternative control plans, explore contaminant,
environmental and source interactions, and train students and workers in the
mechanisms and control of the air pollution system.
Acceptance of models in actual decision-making will depend on the
degree of confidence that can be placed in the validity and logic of the
mechanisms employed and the amount and quality of data available to test and
use the models. Model building, therefore, should be looked upon as an
evolutionary process—one that grows with the normal processes of analysis,
control and evaluation. In particular it should interact with and contribute
to the design of data collection systems and the management and analysis of
aerometric, source emission and other needed data.
Models will have practical utility to the degree they reflect
important aspects of physical and chemical reality. This requirement is
best satisfied by the dynamic simulation model in contrast to statistical
(e.g., regression) or limited empirical (e.g., diffusion) models which are
based on patterns and which resort to artificial expedients to fit parameters.
7-4
-------
Most air quality models in current use may be characterized as
diffusion models. These are based on an empirical relationship, the
diffusion equation, which describes a pattern of contaminant concentrations
in an atmosphere surrounding a point source of emissions. The pattern described
by this equation is static, i.e., does not vary with time. The diffusion
equation is intended to apply only to situations of constant emission rate,
constant wind direction and constant wind speed, and it would require that
these conditions remain constant for an adequate period prior to any test of
applicability.
An air quality model for photochemical contaminants is less amenable
to the diffusion model approach, since the diffusion equations contain no
factors to represent the interaction of contaminants. Such interactions are
in no way related to spatial position, wind velocity, or distances from
emission points. The development of important photochemical secondary con-
taminants, such as ozone, normally requires that the period of residence of
the primary emissions in the atmosphere be more than an hour, even in bright
sunlight. Experience has shown that existing diffusion models are highly
unreliable in predicting concentrations of even non-reacting contaminants at
distances from their sources corresponding to an hour's air movement.
A fully scientific air quality model for photochemical contaminants
would clearly require a mathematical framework related to the chemistry of
the photochemical contaminants. It would be able to predict concentrations
of a large number of primary and secondary contaminant species as a function
of time at any point in the region of interest, and auxiliary computer programs
7-5
-------
could derive, from these detailed predictions, predicted values of any desired
air quality indices. A reactive model may represent the more general case of
air quality modeling since relatively non-reactive contaminants, such as carbon
monoxide, SO. and particulates can also be followed in the system.
B. THE REACTIVE POLLUTION ENVIRONMENTAL SIMULATION MODEL (REM)
A simulation model embodying current state-of-the-art principles of
photochemical smog has been developed for the National Air Pollution Control
Administration by the System Development Corporation under Contract CPA 22-69-108.
The environmental mechanisms employed reflect those described in Chapter 3,
Atmospheric Physics and Meteorology and the kinetics of chemical reactions are
derived from those reviewed in Chapter 4, Atmospheric Reactions.
The structure of the computer program is basically that of a
detailed episode model for chemical kinetics, with subroutines for estimating
the location of the air parcel trajectories and the associated values of
ultraviolet irradiance, mixing height, input of emissions from vehicular
traffic and diffusion of contaminants from stationary sources*. The geometric
setting is a standard two-dimensional urban grid system. Coded in Fortran IV
for IBM Systems 360/50 and 67, a control program outputs parcel positions
and pollutant concentrations at previously specified instants in simulated
time. A meteorological subroutine interpolates wind and other data from
input values at weather stations to points along the trajectory. A traffic
subroutine generates emissions as a function of location and time of day, based
on generalized or specific assumptions about traffic density and emission
factors. An ultraviolet irradiance subroutine computes the relative rate of
* A mathematical description of the model is contained in Appendix II.
7-6
-------
photolysis of nitrogen dioxide as a function of time of day, and a diffusion
subroutine estimates contributions from upwind stationary sources to the parcel
being followed. !
The general architecture of the model is shown in Figure 7-1. The
functional elements of the model may be summarized as follows:
1. The Control Program
The Control Program accepts input data (the conditions specified
for a given simulation), controls the logical sequence and temporal flow of
the simulation, and outputs desired information. The inputs and outputs
currently include the following.
Input
Desired trajectory start position
and time.
Wind speed and direction.
Barometric pressure.
Temperature.
Cloud Cover.
Relative humidity or dew point.
Station or receptor point position
and elevation.
Solar zenith angle.
Radiosonde curves or mixing heignt.
List of chemical species and initial
concentrations.
Output
Pollutant concentrations.
Mixing depth (along the trajectory).
Emissions (along the trajectory), HC, NO.
Trajectory time and coordinate
positions.
Reaction Rates.
Rates of change of the pollutants.
7-7
-------
UJ
BS
0
o
oc:
Q.
§
3
o
oo
o
O
S
o
o
*•—I
0£
O Z
>-> O
u. •—
U-
s
QC
UJ
8
S S
S5
UJ
DC
CO
UJ
oc
o.
UJ
oc
UJ
&
5
§
5
^•4
8
SEt
2
UJ
DC
O
UJ
I
s
CD
UJ
oc
ta
o u-
2: o
S3
UJ
7-8
-------
Input (Continued)
Rate constants.
Traffic emissions (vehicle miles and distribution)
Stationary source data, stack height, diameter,
temperature, exit velocity, emission rate.
With minor program modifications, any of the inpvit datu can be standard-
ized as "fixed" parameters, or parameters can be modified to reflect inputs
which are desired to alter the design characteristics of the photochemical
system. Thus the inputs can be made limited and simple or extensive and complex
by presetting certain values within the code itself and reading in others at
run time.
2. Meteorological Module
This consists of a series of subroutines which accepts actual wind
velocity, temperature, pressure, humidity, cloud cover data, radiosonde and
height above sea level for spatial interpolation to pollutant column position,
e-stimation of mixing depth, and for calculation of time derivatives. Temperature
and pressure inputs are required for the uy_ module, for estimations of horizontal
diffusion, and for the source emission inputs in order to apply gas laws to
estimating concentration-time derivatives in the column.
3. Ultraviolet Module
This subroutine calculates a diurnal uv function based , • neasure-
ment of cloud cover, latitude and local calendar time. The module computes
as a function of time and as a subset of the meteorological topographical
variables the specific ultraviolet absorption rate of NO .
7-9
-------
4. Transport Mechanism
This mechanism is an integral part of the Chemical Kinetics Module
as described below. The transport of air parcels is determined by an inter-
polated trajectory based on actual wind direction and speed and other meteorological
data derived from a "dense" network of meteorological stations. The present
version operates from a data base of 22 wind stations. It has a capcity to
handle 32 stations.
5. Source Emissions
The subroutine computes as a function of time the perturbative
time rate contributions of NO, CO and propylene by application of emission factors
to vehicle miles determined on the basis of grid distributed percentages of
total vehicle miles, and diurnal mileage variations. The Los Angeles Basin is
subdivided into a grid system of 784 equal 4-square-mile squares corresponding
to a 56-mile by 56-mile square total area. The model also includes characterizing
data for 12 large power plant sources. It has the capacity to accept other
source emission input on demand. Stationary source emissions are computed by
the Diffusion Module.
6. Diffusion Module
The model calculates mixing height and assumes instantaneous vertical
mixing in the air parcel as a simplifying assumption. The trajectorized air
parcel is geared to accept variable area source (grid) emissions, e.g., traffic
emissions. Since stable conditions apply, lateral losses and g*ins from the
air parcel are assumed to be equal. The model contains an optional horizontal
diffusion subroutine for estimating downwind concentrations from large
7-10
-------
stationary sources, where stack height and other stack parameters are to be
taken into account. This subroutine computes the perturbative time rale
concentrations of pollutants from large upwind sources that contribute to the
trajectory air parcel of interest.
7. Chemical Kinetics Module
This is a "reactor" mechanism, i.e., chemical reaction;;- are Caking
place continuously within a moving parcel of air. This module has operated on
13-41 steps and is based on a sequence of stoichiometrically valid elementary
reaction steps, previously validated against chamber data. The current
version contains a 31-step mechanism (see Table 7-1). This mechanism preserves
the flexibility of chemical detail necessary to relate to source emission
species (primary contaminants) of interest on the one hand, and to permit
following of the spatial and temporal distributions of secondary contaminant
content rations of interest on the other. The reaction steps, reaction rates
and chemical species are easily modified and expanded on demand. Chemical
muchan isms corresponding to different hypotheses can be input on short notice
and tntL-gratcd with the other environmental and meteorological mechanises
of the model.
This module computes the total time derivative of each of the
photochemical pollutants (or reactants). This is accomplished by adding the
perturbative derivatives input from the other modules to the con••-•onding
right-hand side terms of a set of standard chemical kinetics ordinary
diitorential equations. (The module also utilizes the wind velocity outputs
from the Meteorological Module as the time derivatives of the position vectors
i;.at lit the air parcel trajectory.)
7-11
-------
8. The Numerical Integration Module
On the basis of derivatives computed from the Kinetics Module, the
Numerical Integration Module integrates all parcel pollutant concentrations
and parcel positions forward in simulated time. This module is a standard
numerical integration program based on Hamming's linear, multistep predictor-
corrector method.
C. CURRENT STATUS AND DEVELOPMENT HISTORY OF THE MODEL
A prototype version of the model has been programmed and is in the
validation and optimization phases. To arrive at this stage (1) each submodule
was coded for operation and debugged under time-sharing; (2) a module-by-module
maximum time-sharing usage approach was utilized to simultaneously maximize
program development speed and minimize computer time costs; (3) only after
each submodule was reviewed for computational correctness was the totality
of modules linked, interfaced, and compiled together with the main control
program.
The model is designed to be flexible and expandable. For example
•
any reaction mechanism, chemical species of interest, reaction rates, traffic
and stationary source information, and solar zenith angle diurnal curve is
readily inserted. The first versions of the model delivered to NAPCA
include the following optional features.
1. A 31-step reaction mechanism.
2. A mixing height determination routine (the model provides for both
an A.M. and P.M. curve).
3. One bolar zenith angle curve.
7-12
-------
4. Areal traffic ratios based on I960 data a., obtained from Teagu< and
on a diurnal traffic volume distribution curve. Los Angeles daily vehicle
mileage for 1969 has been estimated at 120,000,000 miles from data supplied
by the State of California.
5. Characterizing data for 12 stationary sources. These include exit
velocity, height of stack, diameter of stack, height above sea level, temperature,
and emission rate in pounds per hour, as obtained from local power plant sources.
D. MODEL VALIDATION
Preliminary validation runs of the model have been conducted through
a .series of .simulations representing conditions in the Los Angeles Basin on
September 30, 1969, a day on which elevated levels of oxidant were recorded
at various monitoring stations in the Air Pollution Control District network.
The trajectories generated by computer runs for this day are shown in Figure
7-2 and postulated rate constants in Table 7-1.
Input for these simulations consisted of weather data collected from
22 meteorological stations for September 30 and values of nitrogen dioxide,
nitric oxide and ozone concentrations as observed at a chosen station at a
chosen time of day on September 30. In addition, values for concentrations of
hydrocarbons were arbitrarily assigned for the input, inasmuch as hydrocarbon
readings were not available. For carbon monoxide a starting level of 5 ppm
was assumed. Contributions of oxides of nitrogen from stationary sources were
.ssumed negligible.
*•• Thirty-One Step Mechanism
The mechanism incorporated in the kinetics module for these simulations
was a set of 31 elementary (or quasi-elementary) steps, derived by simplification
7-13
-------
CO
W
H
CO
2
O
1
g>
•H
Pb
7-14
-------
<•> o O O O -• •» ir O o O <
PJ f* t-4
lc\jrn>-<to
-4 o * r- r- r- r- «
r- r- o<
CD Q. ^
— < m -i
z X O C i/>
o a a
r uj O uj
c i ui —
*- ct
cc
O
C' O
o c/
C O
O C>
O O
o c
e -4
O (Vf i
O o I
i. C.
o c
c o
i.) o
'-> O
C f
f c
—1 (Vj I
• C C1 i
IT C.
o o
O O
o c
C' O
0 O
o
o O
o c1
O '">
L C
c o
o O
o >
C' C
3 O
O C
o o
c c
C I
o o
C- o
O O
I o
O C
•
C o
i O C
i O o
• o o
' <-. C
' O ^
I (M (M
00
r c
' C o
i -J o
I C C'
I C- 'J
o o
OoOOC't~OOOOOCl
I
o c c
. o o
f • c o i. o O o
o o o o o c <->
o o o c c, o r>
O o d o C) C O
O o C. O < i O O
<"• C C i • o v C
c c c
o o c>
O C o
O C C
o c o
J O O
D r o
C> C O
000
C C' O
O '3 O
o & o
3 C O
C Oo
rg p» rv O X
c c r j r
X +
+ p
z o
r*% l\J Hi
X O X
5Om rr-i"f\j» r>^ip*MC'-ic»v****xc'
*Crv** *LJ + r*%*ccoOrn'^XLJ*c:c CCo*
rgm7C'v^rvj + roC'"*JU.iirf*\jx»-l^''^ITr r *fo
ocizcosooza + xiifvjpuri i.Xo'jXx'"!
ii ii ii ii ii "it ii ii ii ii n n n ii H ii n o n n n n ii ii c; ii
C C II II
oj i O *; fsi PJ
PJ *iPjC'l**PJ P*TCXJCC-^O
rOOC p«T 7-3 P+C'+i^jrvOCCcOTLJ*^**
cjOi^ic*rj**r"cc^i + r -3tr*» »** + *r+c<_
*p«i<»ti»i + +op4*p ++("<*. n iii fi iii 3 ;rxmxi
fCCCae *ODOXXXXX«MP. Mir «• r r>l»-r f\.XP,n
aZZZZ^XZ^'ZoOuJUUoC.tJOV'l^l-'U^O^C'OO
I
r-.
CO
H
7-15
-------
of a 40-step mechanism previously validated in simulation of irradiation
chamber experiments (see Chapter 4, Section F-4). Details of this mechanism
(see Table 7-1) are as follows.
Of the 31 steps listed, the first three are the universally recognized
set constituting the basic nitrogen dioxide photolysis cycle, discussed in
Chapter 4, Section D-l. Reaction 4 is the well-recognized oxidation of
nitrogen dioxide by ozone, which produces the standard form of nitrogen trioxide.
Reaction 5 disposes of the nitrogen trioxide by formation of nitric acid
vapor. (This is an example of the use of a pseudoelementary reaction to simplify
the required computations: the rate listed is actually that of the bimolecular
association of the trioxide with the dioxide to form N-0., nitrogen pentoxide
(or nitric anyhydride). It is assumed that nitric acid is formed rapidly,
and that it is the only important product derived from the anhydride. This
being the case, it is unnecessary to compute the concentration of nitric anhydride
as an intermediate, and unnecessary to specify the concentration of water
vapor.
Reactions 6 through 13 are inorganic reactions. All occur by
bimolecular collisions except reactions 9 and 11, in which the presence of a
third molecule is required for stabilisation by removal of kinetic energy.
(In such cases, the rate factor provided is a pseudoconstant, which incorporates
the concentration of inert molecules as discussed in Chapter 4, Section F-3.)
Reaction 14 is provided to test whether the presence of rei.'.":i"/ely
large concentrations of carbon monoxide could be expected to have detectable
7-16
-------
effects on the rate of consumption of reactants in the photo-oxidation i.ystem.
Reactions 15 through 30 provide for the consumption of hydrocarbon,',
(typified by propylene) and their degradation to, molecules of smaller carbon
number through chains of free radical reactions. Reactions 17, 18, and 30 are
postulated not to require trimolecular collisions; these are pseudoelementary
steps in which it is assumed that a virtual equilibrium is established between
a free radical and molecular oxygen as reactants and an oxygenated free radical
intermediate. Reactions 24 and 25, on the other hand, might be considered to
require essentially trimolecular collisions. Reaction 27 is a speculative
chain branching step, incorporated for exploratory purposes.
Reaction 31 allows for effects of less reactive hydrocarbons. For
the sake of simplicity, these hydrocarbons (labeled DUMHC) are assumed to have
the same molecular formula as propylene, and to yield the same products
as propylene on reaction with oxygen atoms. The rate constant of this reaction,
however, is assumed substantially smaller than that for propylene.
Termination reactions between all free radical species present are
also assumed to be part of the mechanism. Although these reactions are not
explicitly included in the list of elementary steps, their effect is incorporated
in the programming of the kinetics module. They are assumed to have a uniform
3 —1 -1
rate constant of 1 X 10 pphm min . Products from these termination reactions,
however, are not included among the products tabulated in the simulation.
2. Chemical Species Utilized
Species tabulated as reactants or products in the 31-step mechanism
are listed in Table 7-2. Of these, the first four represent the components
7-17
-------
Table 7-2
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
N02
NO
°3
C3H6
HNO
CH CHO
HCHO
CO
CH3ON02
C2H3N°5
Fixed 02
DUMHC
N03
H
OH
H02
CH3
CH30
CH302
C2H3°
C2H3°2
C2H4°2
0
SPECIES INVOLVED IN 31-STEP MECHANISM
nitrogen dioxide
nitric oxide
ozone
propylene (or reactive hydrocarbon)
nitric acid vapor
acetaldehyde
formaldehyde
carbon monoxide
methyl nitrate
peroxyacetyl nitrate
fixed oxygen parameter (see text)
less reactive hydrocarbon
nitrogen trioxide
hydrogen (free atom)
hydroxyl (free radical)
hydroperoxyl (free radical)
methyl (free radical)
methoxyl (free radical)
peroxymethyl (free radical)
acetyl (free radical)
peroxyacetyl (free radical)
acetaldehyde oxide (or "zwitter ion")
oxygen (free atom)
-------
most often monitored in connection with photochemical smog, nitrogen dioxide,
nitric oxide, ozone and hydrocarbons, respectively. Nitric acid (species 5)
is auwpected of being a major product in the photo-oxidation, but is almost
never quantitatively measured in the atmosphere. Aldehydes, represented by
acetaldehyde (species 6) and formaldehyde (species 7) , are known to be
formed, but quantitative information on atmospheric levels is scarce. Methyl
nitrate (species 9) and peroxyacetyl nitrate (species 10) are assumed to be
the principal organic nitrogen-containing products.
"Fixed 0 ", listed as species 11, is not a real chemical species
but n chemical parameter included here for convenience. It represents the
amount of molecular oxygen incorporated into products of the reaction complex;
as such it is not readily subject to experimental validation. Molecular
oxygen produced in steps 3, 4, 10 and 13 is registered negatively in this
parameter, which may therefore legitimately acquire negative values under
some > Lrcumslances.
Species 12, "DUMHC," is provided as a less reactive hydrocarbm,
^anticipating only in Reaction 31. Species 13 through 23 are all intermediate
compounds, assumed not to accumulate to appreciable concentrations. Species 13
through 21 are all odd-electron species. Nitrogen trioxide (species 13) is
the first product of oxidation of nitrogen dioxide by ozone, but reacts very
rapidly with either nitrogen dioxide or nitric oxide. Hydrogen atoms
(species 14) and seven free radicals (species 15 through 2JN alro *, j.1 ia
the odd-electron category. The remaining intermediate species utilized In
this mechanism are a hypothetical acetaldehyde-oxide or zwitter ion (species 22),
7-19
-------
and the free oxygen atom (species 23).
3. Trajectory Simulation
Initial conditions imposed in the various simulations are indicated
in Table 7-3 and calculated vs. observed values at terminal points for five
trajectory runs are shown in Table 7-4. In the first run, the starting
location and time were set to downtown Los Angeles at 1:00 P.M., and a simulation
time of one hour was specified. The trajectory generated in the simulation
moved toward the northeast, for a total distance of about six miles;
computer running time was about 4 minutes, for a ratio of simulation time to
run time of about 16 to 1.
The second run computed a trajectory of nearly 5 hours, starting
in West Los Angeles at 10:00 A.M. This showed the original air parcel
moving almost due north across the Santa Monica mountains and eventually
arriving in the San Fernando Valley. Run time was nearly 40 minutes in this
case, for a ratio of simulation time to running time about 8 to 1.
A third run was undertaken, using the same starting location and time
as the first run, but allowing for three hours of simulated time lapse rather
than one, to enable the trajectory to develop closer to the air monitoring
station in Pasadena, thus generating simulated contaminant concentrations
which could be reasonably compared with observations from that station. This
run yielded a simulation time ratio of sixteen to one, in agreement with
run one.
A fourth run was computed.; starting at the Los Angeles International
Airport at 11:00 A.M., and terminating after four hours at Pasadena. This
7-20
-------
Table 7-3. INITIAL CONDITIONS FOR SIMULATIONS.
Starting Point
Starting Time
Initial NO,,, pphm
Initial NO
Initial 0_
Initial C.H..
J 0
Initial DUMHC
Rate constant, step 13
Rate constant, step 21
Rate constant, step 22
Rate constant, step 25
Rate constant, step 27
Rate constant, step 28
Rate constant, step 29
Rate constant, step 30
Simulation
No. 1
DOLA
1300
13
1
15
50
50
100
10
140
10
10
10
10
10
No. 2
WLA
1200
14
1
9
50
50
100
10
140
10
10
10
10
10
No. 3
DOLA
1300
8
5
17
50
50
20
5
70
1
1
5
1
1
No. 4
LAX
1100
18
7
2
50
50
20
5
70
1
1
5
1
1
No. 5
LB
0800
22
38
1
50
50
20
5
70
1
1
5
1
1
7-21
-------
CO
U
I
r^ co
M
0) O
rH
43 CO
CO
H CO
CJ
P3
C?
55
0
Z
en
0
^
§
5
o
^
§
0
5
8
o
^
S
5
o
|
S
2
M
H
l •]
|j
M
y
H
H
0
Z
H
3
g_|
CO
O
S3
M
H
58
EH
CO
^
P
W
a
o
s
i* *
M
P9
O
*
1
o
CO
M
O
1J
CO
PQ
O
prl
5
H
H
S3
O
f-|
*
CM
rH
00
rH
^
O
CO
rH
^^
m
o
0
sf
rH
g
O
4J
0
O
Q
O
O
rH
rj
O
44
O
Q
rH
4g
^
CO
01
01
oo
(3
CO
0
CO
CD
•H
00
a
w
,3
•K
iH
H
i-l
rH
O
in
-^
vO
0
O
m
rH
CO
T3
0)
CO
01
oi
o
0
o
rH
3
o
P
CN
^
^
•K
•k
CM
CO
0~
H
H
rH
m
^^
CN
rH
O
O
m
i-l
(0
0)
•a
l
rH
0
o
CO
I-l
§
0
4J
g
&
o
0
rH
rH
^
I
Ol
-*
PQ
«
*
•K
CO
01
•H
01
00
C
00
J
CO
•3
CO
01
oo
CO
0)
CO
IH
3
O
43
01
M
CO
CO
3
rH
CO
^
•o
CD
£j
01
CO
fN
o
»
CO
c
o
•rl
44
CO
3
0
i-l
CJ
CO
3
O
01
a
CO
44
(3
4J
CO
a
•H
• 0)
"-
cO
CO
01
^
CO
^
•o
01
44
3
(X
e
0
CJ
*
•
CO
44
CO
•a
01
•a
^i
o
CJ
0)
o
Z
•K
•K
q;
0)
a
0)
o
o.
IH
01
44
0
•o
a
CO
01
3
•u
01
O
o
a
a
0
iH
Cfl
bO
01
a
•H
i-H
<0
CJ
- 1
0)
P
c
•a
0)
N
•H
B
•H
44
Pu
0
-------
trajectory showed the parcel moving towards downtown Los Angeles and paralleling
run three to Pasadena. The ratio of simulation time to running time was about
fourteen to one.
Finally, a run on a starting location in Long Beach at 8:00 A.M.
yielded a much smaller simulation time ratio, near unity, with relatively
short travel distance toward the north. This timing anomaly may be explained
by the large magnitude of traffic pertubation derivatives induced by small
computed mixing heights. Indeed, it has been subsequently observed that the
smaller these traffic pertubations, the greater is the simulation time ratio.
Contaminant concentrations calculated from the simulations are shown
as functions of simulated elapsed time in Figures 7-3 and 7-4. Values for
nitrogen dioxide, propylene and ozone for runs 1 and 3 appear in Figure 7-3.
The obvious differences in simulated behavior of the contaminants in these
two runs result mainly from changes in selection of rate constants for
certain elementary steps. These changes, indicated in Table 7-3, were intended
to ieduce the expected rates of consumption of the primary reactants, as those
rates seemed substantially too high in runs 1 and 2. With the new choices, the
values predicted for ozone, nitrogen dioxide and nitric oxide were all within
5 pphm of the observed values at the Pasadena air monitoring station at the
time of closest approach of the trajectory to that station. These validating
points are also shown in Figure 7-3. (Validation data for the hydrocarbon
concentration ar«; not available.)
7-23
-------
RUN 3
60 120
MINUTES
180
Figure 7-3. CONTAMINANT CONCENTRATIONS VS SIMULATED
TIME LAPSE, RUNS 1 AND 3.
7-24
-------
vo
O CO "ch
zoo
o a
O
in
~T
o
•
1
1
1
1
I
1
1
1
1
1
1
1
t
1.
I
1
t
1
1
T
j
|
x'j
X I
^ 1
jv""' i
/
/
^--r""
o
-o
f
1
1
co
1
fl
+
!
1
1
4
o
CM
-g
(
1
t
O
- CM
. o
vo
Ml
/ II
t
«-
- <—»
to
LU
H-
K-t
z:
O
CM
w
CO
w
w
2
o
g
u
o
§
CJ
8
3
60
•H
NOIlVaiN33NOD
7-25
-------
E. APPLICATION OF THE MODEL
After optimum operation and adequate validation and sensitivity
analyses have been achieved, the model will be useful in many ways for application
to practical problems in the control of air quality. Thus, for regions where
weather variables and air quality are well documented, the model will predict
the effect on air quality of postulated trends in emissions rates, either
uniform or geographically biased. With air quality models based on standard
types of diffusion equations, typical contaminant concentrations tend to
increase proportionally to increases in emission rates. However, with the
secondary contaminants generated as a result of photochemical reactions in
the atmosphere, no such relation holds, nor is any simple relation adequate
for describing the changes effected.
Similarly, an adequately validated model will predict the air
quality effects of proposed new stationary sources of oxides of nitrogen or
hydrocarbons.
For regions without air monitoring data, but with good information
about weather and emissions, the model can be used to produce detailed
•
estimates of air quality. For regions with only a single air monitoring station,
or with an inadequate number of stations, the model will provide estimates
of the variation of air quality with location within the region.
Even if no detailed information regarding emissions and weather
exists for the region of interest, potentials for serious photcchealca, -i.r
pollution can be projected from various reasonable assumptions which can be
readily tested by application of the computer model.
7-26
-------
A preliminary study of the application of the model to a control
strategy exercise was conducted. In this study the existing weather for
September 30, 1969, total vehicle mileage and distribution of vehicle mileage
were maintained, but the emission rates for hydrocarbons (propylene) and NO
were modified by 6 sets of relative factors, as follows:
RUN
NO
HC
1
1
.25
2
1
.05
3
.5
.25
4
.5
.05
5
.1
.25
6
.1
.05
Each set represents a conceivable control strategy test. The objective
of the tests is to determine which combination of emission factor reductions
(HC and NO) best minimize contaminant concentrations, particularly ozone.
Results of these runs are presented in Table 7-5 in terms of calculated con-
centrations of the contaminants (nitrogen dioxide, nitric oxide, ozone and
hydrocarbons) at the terminus of the trajectory, near Pasadena, and in
terms of the maximum calculated concentrations of the contaminants (which
may be found at other points of the trajectory.) Also listed is the "fixed
oxygen", a measure of the1 cumulative degree of reaction at the end of the
trajectory.
From Table 7-4 it will be noted that hourly average tonr-;.: ^rations
observed in Pasadena for 1500 to 1600, the hour of the trajectory approach,
were 5 pphm N0_, 1 pphm NO, and 15 pphm oxidant, with no reading on hydrocarbons.
7-27
-------
Table 7-5
Calculated Photochemical Contaminant Concentrations for Lennox-Pasadena
Trajectory, with Varied Emissions Input
Run Number
Initial NO, pphm
Initial HC, pphmc
Input NO, relative
Input HC, relative
NO., pphm: Pasadena
maximum
NO, pphm: Pasadena
maximum
0-, pphm: Pasadena
maximum
HC, pphm: Pasadena
maximum
Fixed oxygen, pphm
oa
7
50
1
1
1
21
0
7
21
21
42
81
155
1
8
12.5
1
0.25
38
A6
5
19
7
7
18
27
45
2
8
2.5
1
0,05
39
39
18
37
2
3
4
6
15
3
4
12.5
0.5
0.25
20
28
2
7
9
9
18
28
37
4
4
2.5
0.5
0.05
26
26
10
18
2
4
4
5
10
5
0.3
12.5
0.1
0.25
6
(18)b
1
3
11
11
17
26
34
6
0.8
2.5
0.1
0.05
14
(18)b
3
5
4
5
4
6
7
a. Run 0 is the reference run, Table 7-4.
b. Maxima of 18 pphm NO- occurred at the starting point of the trajfr.tory
because the assumed initial value wa~ 18.
c. "HC" is assumed as propylene in the mechanism used.
d. Factors applied to emissions input in the reference run.
7-28
-------
For the reference run, the corresponding calculated values were 1 pphm
NO,,, zero NO, and 21 pphm ozone. From the results in Table 7-5, it is clear
that computed values of NO. for Pasadena are drastically increased by reducing
the assumed hydrocarbon input without altering the oxides of nitrogen inputs
(cf. Runs 1 and 2), while simultaneously the calculated values for NO are
increased, and those for ozone and hydrocarbons strongly reduced. At lower
input levels of nitric oxide, the same trends are evident (Runs A vs 3,
and 6 vs 5), although not to as great an extent. Only in Run 5, with NO
input one-tenth and hydrocarbon one fourth of that in the reference run, do the
calculated values of NO., NO and oxidant in Pasadena approach the observed
values: NO,, 6/5, NO 1/1, Ox 11/15. (pphm observed/pphm calculated).
Comparison of the calculated contaminant levels for Pasadena with
the calculated maximum contaminant levels shows that the calculated ozone level
at the Pasadena terminus was at the maximum for all runs with hydrocarbon
input 0.25 of that in the reference run, but was less than the maximum for
those runs with hydrocarbon 0.1. On the other hand, calculated N0? maxima
occurred at the Pasadena terminus only for runs 2 and 4, with hydrocarbon
input 0.1. Thus the model, with its assumed chemical mechanism, predicts
that hydrocarbon control in the Los Angeles Region would have opposite effects
on ozone and oxides of nitrogen as measured in Pasadena, at least for conditions
like those of September 30, 1969.
Further exploration of the implications of these results may be
facilitated by the presentation given in Table 7-6. Here the contaminant levels
7-29
-------
Table 7-6
Calculated Contaminant Levels from Simulation Runs with Modified
Emissions Input, Arranged for Easy Comparison .
Pattern (Run No.):
Relative
NO
Input
Nitrogen Dioxide, pphm
at Pasadena
1 38 39
20 26
6 14
Nitric Oxide, pphm
at Pasadena
0 5 18
2 10
1 3
Ozone, pphm
at Pasadena
21 7 2
9 2
11 4
Relative HC Input
1.0
0.5
0.1
1.00
oa
0.25
1
3
5
0.05
2
4
6
Trajectory maximum
21 46 39
28 26
(18) (18)b
Trajectory maximum
7 19 37
7 18
3 5
Trajectory maximum
21 7 3
9 4
11 5
7-30
-------
Table 7-6 (Continued)
Hydrocarbons, pphm
at Pasadena Trajectory maximum
42 18 4 81 27 6
18 4 28 5
17 4 26 6
a. Run 0 is the reference run (Table 7-3)
b. Maxima of 18 pphm NO,, occurred at the beginning of the trajectory because
of constant assumed initial value.
7-31
-------
taken from Table 7-4 and 7-5 are arranged in a pattern such that, for each
contaminant, the values predicted on reducing hydrocarbons at constant NO
input are entered across rows, the the values predicted on reducing NO at
constant hydrocarbon input are entered down columns. Within the range tested,
the principal implications of the model with the assumed chemical reaction
mechanism appear to be the following:
1. Decreasing hydrocarbon input should cause decreases in ozone and
hydrocarbon levels at Pasadena, with increases in nitric oxide and nitrogen
dioxide levels.
2. As to the maximum contaminant levels at points between Lennox and
Pasadena, decreasing hydrocarbon input should cause first an increase, then
a decrease for nitrogen dioxide. For the other contaminants, the maximum
levels would behave in generally the same way as the Pasadena levels.
3. Decreasing nitric oxide input should cause decreases in nitric oxide
and nitrogen dioxide levels at Pasadena, but an increase in ozone levels.
Hydrocarbon levels would be little affected.
4. Decreasing nitric oxide input should have effects on maximum
contaminant levels similar to those described for Pasadena levels.
The application of the Reactive Pollution Simulation model
requires an orientation to the modeling of air pollution that differs con-
siderably from the diffusion model approach. The latter provides for a
capability to deal with multiple point, line or area sources, multiple
receptor points, and to display averages of pollutant concentrations on a
grid system which represent seasonal or annual time periods of interest.
7-32
-------
The diffusion approach lends itself to a gross picture of relative
pollution levels over a geographic region of interest. It has the advantage
of displaying values simultaneously. It has the disadvantage of poor accuracy,
of not being able to dynamically simulate meteorological conditions, reactive
contaminants or episode conditions.
The Reactive Model, on the other hand, is an episode model and simulates
conditions as they occur on a diurnal basis and outputs "instantaneous" concen-
trations along a calculated air parcel trajectory.
In general the differences in the two approaches can be seen in
terms of isopleth construction (lines of equal concentration of a single
contaminant), and trajectory plots which output many contaminant concentrations
for an unlimited number of receptor points along the trajectory. The contaminant
concentrations are governed by time and position and therefore are not simultaneous.
The values output are "quasi-instantaneous" in character, i.e., the rates of
change of the concentrations of contaminants are comparatively small over any
5-minute interval. Thus model calculated values can be reasonably compared with
5-minute average observations. For validation purposes, these can be considered
to fall with the range of observed hourly value observations.
Given these characteristics, the use of this model for analysis of
seasonal, annual and other long term conditions based on the distribution
of receptor points throughout the air quality region of interest requires the
following approach:
7-33
-------
1. Multiple Receptor Points - The determination of air quality at
multiple receptor points (other than those output along a single trajectory)
requires the use of a backward trajectory plotting technique. Points are
selected by coordinate position and time. A backward trajectory is plotted
to determine the trajectory start point. The trajectory is then run forward
with kinetics routine to establish the air quality at the desired receptor
point. Air quality at an unlimited number of receptor points can be so calculated.
The data generated can be used as a basis for isopleth plotting.
2. Representative Air Quality - The depiction of representative conditions
(monthly, seasonal, annual time periods) must be arrived at through characteri-
zation of meteorological conditions which represent days which are typical of
the time periods of interest for the given air quality region under study.
This requires that meteorological data of the given region be statistically
characterized particularly in terms of wind speed, direction, mixing height and
other meteorological and environmental parameters of interest. These can be
grouped into days which are conducive to acute, moderate and light episode
potentials. Meteorological data describing these representative days can be
input to the model.
3. Artificial Trajectories and Multiple Trajectories - Where complete
knowledge of the wind regime of a region is available, trajectories can be
artificially plotted from streamline or other summary data and corresponding
wind data can be assigned to the coordinate points. This tecLai.-;-^ has <.'.,c
advantage of hypothesizing and testing meteorological conditions which
maximize the build-up of contaminants to produce acute episodes. Also this
7-34
-------
technique provides the capability of standardizing sets of representative
[
meteorological conditions for mass-testing many urban regions for their
i ,
reactive pollution potentials, and for testing a variety of urban-source-
configurations and emission rates.
4. Non-Photochemical Contaminants - The model readily handles non-reactive
contaminants—CO, SCL, particulates. The model can operate either with all
i
contaminants simultaneously—photochemical and non-photochemical—or the chemical
i
kinetics mechanism can be suppressed in order to operate with only non-reactive
i
contaminants. \
7-35
-------
APPENDIX I
RESULTS OF STATISTICAL ANALYSIS
I. INDIVIDUAL LOS ANGELES SCAN STATIONS
A. ANAHEIM
I
1. Rank Order Analysis I
The peak instantaneous value of A3 pphm for total oxidant concen-
tration was recorded on August 28, 1967 during the 13th hour. On 113 occasions
the daily instantaneous peak reached a value of 14 pphm or greater. The 14th
is the modal hour of occurrence of peak oxidant concentration with 27 out of
113 such events taking place in that hour. Eighty-five percent of all peak
instantaneous values occur between the llth and 14th hour. October is the modal
month of peak instantaneous oxidant levels with 20 events out of a total of
113. During the 9th hour on August 28, 1967 a peak instantaneous value of
17 pphm of N0_ was recorded. This is the 12th highest reading in the N0»
instantaneous daily peak population.
j
2. Frequency Distribution and Sample Statistics
The frequency distribution of hourly average oxidant concentrations
is strongly skewed to the left with a mean of 3.45 pphm, a standard deviation
of 3.93 pphm and a mode of 1 pphm. More than 96 percent of the hourly average
oxidant concentration readings fall below 14 pphm. The value 14 pphm is 2.68
sample standard deviations from the mean.
3. Correlation and Regression Analysis
The largest correlation of .30 among dependent and independent
variables was found to be between the total oxidant concentration logarithmic
variable and the N02 logarithmic variable. The next correlation (in absolute
value) of -.28 was lor mixing height. In a 7-variable regression (CO, NO ,
mixing height, wind speed, temperature, relative humidity and sky cover) a
standard error of estimate of .102 and a shrunken correlation coefficient of
.55 were attained at the 99.9-percent level of confidence. From the above
standard error of estimate the following 95 percent relative error confidence
interval was computed:
) £ (CXR) 1 1.6 (OXM)
1-1
-------
4. Auto and Cross-Correlation Analysis (August 26 to August 30)
As with the CAMP cities it can be seen that during a string of high
oxidant days total oxidant is highly sequentially correlated. CO and N02 and
NO and total hydrocarbons auto-correlation functions are, in general, quite
similar. Meteorological variable auto-correlation functions are included for
comparison. In the cross-correlation analysis the highest cross-correlation
calculated between oxidant and other photochemical contaminants was -.47 at
lag 0 for NO. Among the meteorological variables, there was a cross-correlation
of .84 at a lag of 3 hours for total oxidant concentration versus wind speed.
This result seems to be clearly indicative of a wind-transport phenomenon.
Total oxidant was well cross correlated positively (.78 at lag 1) with temperature
and fairly cross correlated with mixing height negatively (-.45 at lag 1).
B. AZUSA
!• Rank Order Analysis
The peak instantaneous observed concentration of 73 pphm was recorded
on August 30, 1967 during the 12th hour. On 204 days a peak instantaneous value
of 14 pphm or greater was recorded. The 14th hour is the modal peak oxidant
hour with 57 out of 204 peaks taking place in that hour. 87.5 percent of all
peaks are attained between the 12th and 15th hours. August is the modal month
of peak oxidant level occurrence with 34 incidents out of 204.
On the highest peak oxidant day of the year the peak instantaneous
NO reached its 15th highest recorded value of 15 pphm during the 8th hour.
2. Frequency Distribution and Sample Statistics
The frequency distribution for average hourly oxidant concentrations
is left skewed with sample mean of 5.4 pphm, standard deviation of 7.12 pphm
and modal value of 1.00 pphm, with 2833 out of total population of 8040 read-
ings recorded at this level. The value 14 pphm exceeds 89.99 percent of
population values and is 1.23 standard deviations from the mean.
3. Correlation and Regression Analysis
In a sample of 204 points the lag of the maximum total oxidant
concentration had the highest correlation of .6 with the 3-hour logarithmic
average of CO concentration. This was followed by correlation of -.55 and .54
with mixing height and temperature. A 7-variable subset (CO, mixing h>inht,
wind speed, pressure, temperature, relative humidity and total a*y cover)
yielded a regression significant at ':he 9S percent level of confidence, with a
standard error of estimate of .092 and a shrunken multiple correlation coefficient
of .81. The standard error of estimate gives &8 the corresponding 95-percent
confidence relative error band:
1-2
-------
•65<°V 1
4- Auto- and Cross-Correlation Analysis(August 26 toAugust 30)
I
Characteristically, during episodic lag sequences total oxidant is
very highly sequentially correlated. The corresponding auto-correlation
function takes on values of .87, .65, .42, etc., through lags 1, 2, and 3.
In this particular case it is noted that the mixing height auto-correlation
function is similar to that of total oxidant. Generally, most pollutant and
meteorological auto-correlation functions indicate a temporal persistence.
The exceptions here are CO, NO, and N0?. In the cross-correlation analysis a
very high cross-correlation of .86 between total oxidant concentration and wind
speed is exhibited at a lag of 2 hours, with significant cross-correlation
values down to a lag of 5 hours. This result is a|very strong indication
of the existence of a wind transport phenomenon.
C. BURBANK
1. Rank Order Analysis
The peak instantaneous oxidant concentration of 51 pphm was attained
on July 21, 1967 during the 12th hour. On 182 occasions the peak instantaneous
recorded concentration equaled or exceeded 14 pphm. The modal hour of peak
oxidant occurrence is the 12th hour, with 61 out of 182 peak oxidant incidences
in that hour. Ninety percent of all peaks are observed between hours 11 and
14 inclusive. July and August are the modal months of peak oxidant cocurrence,
being together responsible for 72 out of 182 such episodes. On July 21, 1967
the 9th highest instantaneous peak level of 42 pphm of NO. concentration was
recorded during the 8th hour.
2. Frequency Distribution and Sample Statistics
The frequency distribution of 7992 hourly average total oxidant
concentrations is skewed to the left with mean of 4.64 pphm, standard deviation
of 6.17 pphm and mode of 1.00 pphm. Ninety percent of all observed concentra-
tions are exceeded by 14 pphm. The value 14 pphm is 1.52 sample standard
deviations from the sample mean.
3. Correlation and Regression Analysis
The total oxidant logarithmic variable was most highly correlated
with the CO logarithmic average with corresponding correlation coefficient of
.48, with the logarithmic s.emperature correlation at .47, ana the logarithmic
mixing height at -.40. A 3-variahIe sucsec (CO, mixing height, and temperature)
gave regression results significant at the 99 percent level (99.999 9
percent) with a corresponding standard error of estimate of .11 and a shrunken
1-3
-------
multiple correlation coefficient of .62. The above standard error of estimate
produces the following 95-percent confidence band for the oxidant concentration,
(OX_), predicted from the regression equation.
.6(03^) <. (OX^ i 1.66(03^)
4. Auto- and Cross-Correlation Analysis, (July 18 to July 22)
All contaminant (CO, NO, N02, NO , and total oxidant) auto-correlation
functions exhibited the persistence phenomena previously described for Azusa.
Total oxidant auto-correlation had the highest lag 1 value at .93, while total
oxides of nitrogen had persistently high auto-correlations ranging from .90
at lag 1 down to .51 at lag 4. The highest cross-correlation of total oxidant
with another photochemical pollutant was -.52 at lag 5 with NO. Cross-cor-
relations of total oxidant with the meteorological variables was characterized
by a cross-correlation of .88 for temperature at a 1-hour lag and .74 for wind
speed at a 4-hour lag—again indicative of wind transport.
D. DOWNTOWN LOS ANGELES
!• Rank Order Analysis
The peak instantaneous oxidant concentration of 45 pphm was recorded
on October 18, 1967 in the 13th hour. On 144 days peak levels of 14 pphm or
more were attained. The llth hour is the modal episode hour with 39 out of
144 such events occurring in the interval between 11:00 A.M. and 12 noon.
Ninety-two percent of all peak oxidant events take place between the hours of
11 and 14, inclusive. August is the modal month of high oxidant events, with
27 out of 144. On October 18, 1967 during the 23rd hour a peak instantaneous
concentration of 41 pphm of N02 was recorded. This was the llth highest peak
instantaneous value.
2. Frequency Distribution and Sample Statistics
The total hourly average oxidant concentration population of 8202
observations has the characteristic left skewedness with sample mean 3.34 pphm,
standard deviation 4.15 pphm and modal v :lue of 1.00 pphm. Ninety-six percent
of the population values are exceeded by 14 pphm. This latter value is 2.57
sample standard deviations from the mean.
3. Correlation and Regression Analysis
The highest correlation in a sample of 144 between cu.pendent and
independent variables was .30 for ths 3-houk* logarithmic average of NO.. A
6-variable subset (CO, NO, NO., pressure, relative humidity and sky cover)
yielded significant regression results with corresponding standard error of
1-4
-------
estimate at .098 and shrunken multiple correlation coefficient of .55. From
the above standard error of estimate is computed the following 95 percent confi-
dence interval:
•6*<°*M> 1 (°V 1 1-57(0^)
4. Auto- and Cross-Correlation Analysis, (October 10 to October 20)
The auto-correlation functions for N0« and total hydrocarbons followed
by total oxidant represented the strongest "persistence" phenomena. Again
total oxidant was the most strongly serially correlated time series with an
auto-correlation value of .87 at lag 1, 64 at lag 2, etc.
There was no striking cross-correlatidn between total oxidant and
either the other photochemical contaminants or meteorological variables; -.41
with NO concentration at a lag of 2 hours, -.46 with mixing height at lag 0
and .5 wind deviation at lag 3. There was no significant indication of trans-
port, the wind speed cross-correlation taking on a maximum value of .2 at lag
0. ' !
E. LA HABRA
1. Rank Order Analysis
The peak instantaneous concentration of 31 pphm was recorded on
October 11, 1967 during the 14th hour. On 54 recorded occasions a peak instan-
taneous total oxidant concentration of 14 pphm or more was reached. The 13th
hour is the modal oxidant episode hour, with 13 out of 54 occurrences of high
oxidant levels (>_ 14 pphm) recorded in that hour. Forty-eight out of 54 high
oxidant events occurred from the 12th to the 15th hours, inclusive. October
is the modal month of high oxidant. The 6th largest peak instantaneous NO.
concentration of 24 pphm was observed during the 16th hour on October 11, 1967.
2. Frequency Distribution and Sample Statistics
The sample population of 1350 total oxidant concentration was left
skewed with mean 2.84 pphm standard deviation 3.95 pphm and mode 0.00 pphm.
Over 97 percent of the population values were less than 14 pphm. The value 14
pphm lies 2.82 standard deviations from the mean.
3. Correlation and Regression Analysis
In a small sample size of 20, limited by missing data, tbe highest
correlation (in absolute value) between dependent variable aux independent
variables was -.5 for the 3-hour logarithmic average of mixing height. The
correlation with the 3-hour logarithmic average of sky cover was .3. A 3-
variable regression (mixing height, relative humidity, sky cover) gave results
1-5
-------
significant to the 10th decimal place with standard error of estimate .019
and shrunken multiple correlation coefficient .99. The 95 percent confidence
interval for the regression formula predicting total oxidant (OX_) is given
by:
1 (OXR) 1 1.04 (OXM),
where it is recalled that (OX.) is the observed value of total oxidant concen-
tration. These results are ox little importance because of the small sample
size.
In the auto- and cross-correlation analysis results similar to those
previosuly discussed for SCAM stations are generally obtained: high serial
correlation of total oxidant and other contaminant concentration, fairly high
cross-correlation of oxidant concentration with wind speed indicative of trans-
port, etc. There is a high cross-correlation of .85 at a lag of 1 hour for
temperature. There is a fairly high temporally persistent cross-correlation
for sky cover. Since the number of samples in the cross-correlation analysis
is about six times greater, higher confidence may be placed in the results of
the auto- and cross-correlations, than in those for correlation and regression.
F. PASADENA
!• Rank Order Analysis
The peak instantaneous concentration of 53 pphm of total oxidant
was recorded on August 29, 1967. On 189 occasions daily peak values equaled or
exceeded 14 pphm. The modal hour of peak instantaneous total oxidant concen-
tration was 13, with 52 12th hour incidences out of 189 total. Ninety percent
of all peak total oxidant concentrations occurred between the 12th and 14th
hours, inclusive. July was the modal month for peak oxidant levels with 31
out of 189. On August 29, 1967 during the 7th hour a peak instantaneous NO.
concentration of 13 pphm was attained. This was the 21st highest level recorded.
2. Frequency Distribution and Sample Statistics
The sample population of 821^ observations of total oxidant hourly
average concentrations is skewed toward 0 ipnm with sample mean at 4.57, sample
standard deviation of 5.92 and modal value of 1.00 pphm. Ninety percent of the
population values are less than 14 pphm. The value 14 falls 1.60 standard
deviations from the mean.
3. Correlation and Regression Analysis
The highest correlation (in absolute value) of -.59 corresponded to
the mixing height logarithmic average. This was followed by the value .5 for
the CO logarithmic average. An 8-variable subset (CO, NO, mixing height, wind
1-6
-------
speed, pressure, temperature, relative humidity and sky cover) yielded a regres-
sion with a standard error of estimate of .085 and a shrunken multiple correlation
coefficient of .78. The above standard error of estimate translates itself
into the following 95-percent confidence band:
.68(0)^) <_ (OXj^) i 1.48 (OXM)
A. Auto- and Cross-Correlation Analysis
All contaminant concentrations auto-correlation functions excepting
total oxidant exhibited a marked decrease in the aforementioned persistence
phenomenon. The subsequent cross-correlation analysis exhibited little serial
cros.s-correlation between total oxidant concentration and the other pollutaats,
the largest in magnitude being -.43 for total hydrocarbons at lag 3. There
was, however, good cross-correlation between total oxidant concentration and
some of the meteorological pollutants: -.48 for mixing height at lag 2, .79
for temperature at lag 2, and .74 for wind speed at lag 3.
II. INDIVIDUAL CAMP CITIES i
A. CHICAGO \
1. Rank Order Analysis
The peak instantaneous level of total! oxidant concentration of 16
pphm was reached on September 8, 1967 in the 12th hour. On 58 days a peak instan-
taneous concentration value of 7 pphm or more was measured. The hour in which
the peak level occurred has a bimodal frequency distribution with primary mode
at hour 15 and secondary mode at hour 13. Fifty-seven percent of all peak
instantaneous total oxidant recordings fall between the hours of 12 and 16.
June is the modal month of occurrence of high total oxidant levels, with 11
of 58 occurrences of oxidant levels of 7 or more pphm occurring in that month.
On September 8, 1967 at hour 7 an NO- concentration of 52 pphm
was observed. This peak value of N0? concentration was the.second highest
measurement of the year, exceeded only by peak level of 63 pphm recorded on
September 6, 1967 during the 6th hour.
2. Frequency Distributions and Sample Statistics
The frequency distribution of hourly average oxidant concentrations
is strongly skewed to the left with mean of 2.93 pphm, modal value of 3 pphm
and standard deviation of 2.02 pphm. In a total sample popul *itiou ^" 5597
hourly average total oxidant observations, 95 percent of the observed values
were less than 7 pphm. This value of 7 pphm is 2.04 standard deviations from
the mean.
1-7
-------
3. Correlation and Regression Analysis
A correlation analysis based on a restriction of the sample space
to the 58 points where (OX..) >_ 7 reveals:
• The highest correlation between dependent and independent
variable is .389, corresponding to total hydrocarbon concen-
tration.
« The next highest correlation of .326 corresponds to N02
concentration .
• The highest correlation in absolute value among the dependent
variables and those corresponding to independent meteorological
variables is -.261, associated with wind speed.
Utilizing information computed from the correlation analysis a
regression analysis was carried out with the following results:
• The results of the regression were significant at the 99-
percent level of confidence.
• The standard error of estimate for the complete regression
was .088. This can be translated to mean that, with approx-
imately 95-percent confidence, the regression estimate (OX_)
compiled by substituting observed values of the independent
variable in the regression equation and taking the antilog
of the result will satisfy:
where (OX.) is the observed peak hourly average total oxidant concentration.
• A smaller number of Independent variables (NO, N02> total HC,
temperature, and total sky cover) yields the smallest standard
error of estimate (.0861) and largest shrunken multiple
correlation coefficient (.523).
4. Auto- and Cross-Correlation Analysis
An auto-correlation analysis of 5 lags over a maximum of 96
possible hourly average rankings for four photochemical pollutants was carried
out. The particular time sequence (June 2 to June 5) involved ?- che ; i /Lysis
was selected on the basis of high oxidant levels. Scanning the computer output
it can be seen that:
1-8
-------
• The auto-correlation functions of NO and total HC are very
similar.
• The auto-correlation functions of NO. and OX match closely.
I
• There is fairly good correlation df approximately .7 between
observed successive hourly values of NO and HC concentrations
and the auto-correlation functions drop sharply for higher lags.
• There is very good correlation of approximately .9 between
successive hourly values of NO. and OX. The auto-correlation
functions drop gradually for higher lags.
The results of subsequent cross-correlations analysis were incon-
clusive. The highest (in absolute value) cross-correlation measured was .69
at lag zero for NO versus total HC. This result was in agreement with the
correlation of .79 between the 3-hour logarithmic averages as observed in the
correlation study.
b. CINCINNATI
1. Rank Order Analysis
The peak instantaneous level of total Oxidant concentration of
19 pphm was reached on July 31, 1967 in the 13th hour. On 68 days a peak
instantaneous concentration value of 7 pphm or more was measured. The hour
in which the peak occurred has a unimodal frequency distribution with a modal
value at 15 pphm. Sixty-six percent of all peak observed instantaneous values
fall between the hours of 12 and 16. September is the modal month of occurrence
of h.gh total oxidant levels, with 15 out of 68 occurrences of observed high
oxidant levels.
2. Frequency Distributions and Sample Statistics
The frequency distribution of hourly average oxidant concentrations
is strongly skewed to the left with mean of 3.14 pphm and standard deviation
2.24 pphm. Over 94 percent of the observed sample population of 5310 hourly
average total oxidant concentrations values are less than 7 pphm. This value
of 7 pphm is 1.72 standard deviations from the mean of 3.14.
3* Correlation and Regression Analysis
The correlation analysis on samples for which max.--.uffl hc^t^y average
oxidant levelh were greater than or eqnrl to 7 was inconclusive. The highest
correlation in absolute value was -.214 between oxidant and relative humidity.
1-9
-------
The results of the subsequent full 10-variable regression analysis
(9 independent variables) were significant at the 92-percent level of confidence.
The best predictive results were obtained with a 5 independent variable subset
(CO, NO, wind speed, temperature, relative humidity) regression. This regression
was significant at the 99-percent level with a corresponding standard error of
estimat of .097 and a shrunken multiple correlation coefficient of .40. Analogous
to our resulcs for Chicago, this "best" standard error of estimate yields, with
about 95-percent confidence:
.61(02^) <. (OJ^) <_ 1.63(0^),
where (OX_) is predicted from the regression equation and (03O is the correspond-
ing observed peak hourly average total oxidant concentration taken from the
restricted population of values greater than or equal to 7 pphm.
4. Auto- and Cross-Correlation Analysis
In the auto-correlation analysis a general similarity to the Chicago
case can be observed, i.e.,
• The shapes of the NO and total HC auto-correlation curves agree
fairly well.
* The same findings apply to NO. and total OX.
• Successive hour-to-hour correlations are high, with successive
total oxidant values very highly correlated.
As for the cross-correlation analysis, it can be seen that total
OX concentration is fairly negatively correlated with present and previous
total hydrocarbons concentration. This last result seems, at first glance, to
contradict the results of the correlation study. It must be remembered, however,
that in that study logarithmic relationships were under investigation.
C. DENVER
1. Rank Order Analysis
The peak instantaneous level of total oxidant concentration of
20 pphm was recorded on August 14, 1967 during the 10th hour. On 48 days a peak
oxidant value of 7 pphm or more was observed. The hour of peak instantaneous
oxidant occurrence has a blmodal distribution with primary mode between 9 and
13. August is the modal month of occurrence for high oxidant levels, with 24
percent of such occurrences of peak vsluas greater than or equal to 7 pphm.
1-10
-------
2. Frequency Distribution and Sample Statistics
Consonant with results from Chicago and Cincinnati the frequency
distribution of total OX concentration is strongly skewed to the left with mean
of 2.83 pphm, a standard deviation of 2.03 pphm and a modal value of 2 pphm.
The value 7 pphm exceeds 97 percent of the population of 36-7 values and is
2.06 sample standard deviations from the mean.
3. Correlation and Regression Analysis
The "best" correlations of the dependent variable were with the
3-hour logarithmic averages of NO and temperature. These were -.31 and .27,
respectively. However, fairly good results were obtained in the regression study.
A subset of 5 independent variables corresponding to NO, nonmethane HC, wind
speed, temperature and total sky cover yielded a standard error of estimate of
.089 and a shrunken multiple correlation coefficient of .66 at the 99.9-percent
level of confidence. These figures correspond to the 95-percent confidence
interval:
•67<°V 1 (O^R) - lt5 <0V
4. Auto- and Cross-Correlation Analysis
Included in the auto-correlation analysis for comparison purposes
are auto-correlations of three meteorological Variables, wind speed, temperature,
and relative humidity. As previously seen, among the photochemical contaminants
total oxidant concentration is best correlated time-sequentially.
In the cross-correlation analysis a high negative correlation of
-.82 is seen between present total oxidant levels and levels of N0_ 1 hour
previously. Noting the small number of observations in the N02 sample, there
is justifiably little confidence in this figure.
D. PHILADELPHIA
!• Rank Order Analysis
The maximum instantaneous value of total oxidant concentration of
17 pphm was attained on June 24, 1967 during the 16th hour. On 81 days a peak
value of 7 pphm or greater was observed. The hour of peak occurrence has a
unimodal frequency distribution with mode between 11 and 12. Fifty-five percent
of all peak instantaneous oxidant concentration were recorded between the hours
of 11 and 15. July is the modal month of incidence of hi^'.i total aidant
concentration: 20 out of the total of 81 incidents were recorded in that month.
1-11
-------
2- Frequency Distributions and Sample Statistics
The frequency distribution of hourly average oxidant concentrations
is strongly left skewed with a mean of 2.63 pphm, mode of 2 pphm and standard
deviation of 2.15 pphm. In a total sample population of 6134 observations,
94.46 percent of the population values were less than 7 pphm. The value
7 pphm is 2.00 sample standard deviations from the mean.
3. Correlation and Regression Analysis
The highest correlation between dependent and independent variables
of .25 was attained for N0_. This was followed by a value of .14 correaponding
to the temperature variable. A subset of 8 variables (CO, NO, NO., nonmethane
HC, wind speed, pressure, temperature and sky cover) yielded the Best standard
error of estimate of .095 and shrunken multiple correlation coefficient of .36.
From this standard error of estimate the following 95-percent confidence level
is derived:
.645(03^) <. (OXR) <. 1.55(0^)
4. Auto- and Cross-Correlation Analysis
Case 1 (June 23 to June 25)
In this auto-correlation analysis it is seen that total oxidant
and temperature auto-correlation functions are quite close in structure. Both
exhibit very high correlations of over .9 at lag 0, .8 at a lag of 1 hour and
thereafter dropping off slowly through values of approximately .8 for lag
2, .65 for lag 3, etc.
Case 2 (June 15 to June 18)
Once again it is noted that the hour-to-hour correlation of oxidant
remains high for lags of 1 and 2 hours. Here the resemblance between the
oxidant and temperature auto-correlation functions diminishes as the lag number
grows.
Case 3 (August 26 to August 30)
Here the total oxidant correlation curve falls rather more rapidly
than in the previous cases. The higher sample numbers indicate that these
results may be viewed with somewhat higher confidence.
In the Case 3 cross-conciatior; analysis, the case with the highest
sample numbers involved in computation of the cross-correlation, it can be
observed that total oxidant is fairly well correlated (negatively) with present
1-12
-------
and preceding hourly values of NO and positively correlated with wind .sp-eed.
This latter result may be indicative of contaminant transport phenomena.
li. ST. LOUIS
!• Rank Ord.; r_ Analysis
The highest instantaneous value of total oxidant concentration of
20 pphm was measured on June 30, 1967 during the 14th hour. On 119 days a
peak value of 7 pphm or greater was measured. This hour of peak occurrence
has a bimodal frequency distribution. The primary mode is at hour 12 with
22 occurrences. Fifty percent of all peak instantaneous averages occur between
the hours of 10 and 15. July is the modal month of high oxidant levels with
24 and 119 or 20.2 percent of such recordings being recorded in July. It can
be observed that a broad interval of months starting in April and ending in
October contain almost 75 percent of the high oxidant days.
2• Frequency Distribution and Sample Statistics
The frequency distribution of hourly average oxidant concentrations
is M rongiy left skewed with a mean of 3.44 pphm, mode of 2 pphm and a sample
standard deviation of 2.29 pphm. In a total sample population, 94 percent of
the values are less than 7 pphm. The value 7 pphm is 1.55 sample standard
deviations from the mean.
3. Correlation and Regression Analysis
Case 1
The highest correlation in absolute value of -.46 in a sample of
39 corresponded to total oxidant and wind speed. The next highest value of
-.26 was attained for wind speed. A subset of A variables (CO, NO, N0£ and
wind speed) yielded a regression significant at the 99.6-percent level in which
the standard error of estimate was .062 and the shrunken multiple correlation
coefficient was .54. Corresponding to the above sample standard error of
estimate, the following 95 percent confidence interval was computed:
(.7500XM) <_ (OXR) £ 1.33 (03^)
This inequality is equivalent to saying that with probability of
approximately .95 the relative err^r in predicting oxidant concentration from
the regression formula will 2Le between -.25 and +.33.
1-13
-------
Case 2
The sample of Case 1 was a 39 point subset of Case 2. In Case 2
there are 119 sample points corresponding to the 119 days of peak oxidant
levels. For this case, the highest dependent-independent variable correlation
of -.29 was between those corresponding to oxidant and wind speed. The next
highest, .23, corresponded to N0_. In the regression study a 4-variable
subset (NO, NO., wind speed, and total sky cover) yielded a regression significant
at the 99.40-percent level with a standard error of estimate of 0.92 and
shrunken multiple correlation coefficient of .36. Corresponding to the above
standard error of estimate is the 95-percent regression confidence interval:
.65(0^) <_ (OXj^) <_ 1.53(0]^)
4. Auto- and Cross-Correlation Analysis (June 29 to July 2)
For lag numbers 0 through 5 based on samples of 89, 84, 82, 81,
and 80, respectively, the corresponding auto-correlation of oxidants were 1, .91,
.76, .58, .41, and .27. Here again is an example of the strong linear serial
dependence of total oxidant concentrations.
The highest cross-correlation obtained was .54 between hourly average
total oxidant concentration and NO- concentration 5 hours earlier.
F. WASHINGTON
1. Rank Order Analysis
The peak instantaneous value of 25 pphm for total oxidant was
recorded on November 27, 1967 during the first hour. This also corresponds
to the time of occurrence of the third highest instantaneous N0« level of
76 pphm. Obviously, the time of this particular event precludes an association
with any photochemical mechanism. Further, the underlying data quality
becomes highly suspect. On 79 days a peak instantaneous level of 7 pphm or
greater was observed. The hour of peak occurrence has a bimodal shape with a
questionably small tertiary mode at 23. The primary mode is at 12. The
secondary mode is at 15. Seventy-two percent of all instantaneous total oxidant
peak values occur between the 12th and t,.r> 16th hours, inclusive. July is the
modal month of peak oxidant occurrence, with 18 out of 79 such occurrences
taking place in that month.
2. Frequency Distributions and Sample Statistics
The frequency distribution of hourly average oxidant concentrations
is skewed to the left with a mean 01 2.57 pphia, a mode of 2 pphm and a
standard deviation of 2.30 pphm. In a total sample of 7080, 94 percent of
the recorded values are less than 7 pphm. The value 7 pphm lies 1.93 standard
1-14
-------
deviations to the right of the mean.
3. Correlation and Regression Analysis
The highest correlation of .48 was 'for the CO variable. This
was followed by a correlation of .33 for NO and -.25 for total sky cover.
A 6-variable subset (CO, NO, pressure, temperature, relative humidity, and
total sky cover) yielded a regression significant at the 99.999-percent level
in which the standard error of estimate was .099 and the shrunken multiple
correlation coefficient was .56. Correspponding to the above sample standard
error of estimate, the following 95 percent confidence band was derived:
.63(0^) <_ OXj^) <_ 1.58(03^)
4. Auto- and Cross-Correlation Analysis
Case 1 (first high oxidant episodic sequence; (July 26 to 28)
The total oxidant auto-correlation function exhibits the previously
discussed properties. That is, very high correlation of .93 at lag 1, gradually
decreasing through values of .82 at lag 2, .63 at lag 3, etc. The other
photochemical pollutants for which data are available have fair to good corre-
lations at lag 1 but drop rapidly for higher lag numbers. The largest total
oxidant cross-correlation (in absolute value) of -.58 occurs at lag 1 for NO.
Case 2 (low oxidant sequence; (November 17 to 21)
The total oxidant autocorrelation function has values of .85,
.66, .47, .27, and .09 for lags 1, 2, 3, 4, and 5, respectively. The highest
auto-correlation of .93 at lag 1 is for N0«. The highest cross-correlation of
.54 occurred between total oxidant concentration and wind speed.
Case 3 (second high episodic sequence; (November 25 to 27)
This appears to be a L' s>hly atypical episode. It is in this sequence
of days that several instantaneous ^ks occur during the evening: that is,
three peak oxidant levels are observed during the 23rd hour. Without a minutely
detailed investigation of the environmental circumstances, the performance
status of the instrumentation, etc., it would be unwise to put forth as valid
even the most cursory interpretation.
1-15
-------
APPENDIX II
MATHEMATICAL DESCRIPTION OF THE MODEL
METEOROLOGICAL MODULE EQUATIONS
1.0 INTRODUCTION
The meteorological module computes the values of six trajectory point output
variables and their time derivatives as a function of time and trajectory
position inputs. Wind velocity x and y components, temperature, pressure,
relative humidity, weather cover and mixing depth and their time derivatives
are estimated via weighted inverse distance squared interpolations on time-
correspondent values among all "permissible" meteorological stations. Per-
missible stations are those which lie together with the trajectory point on
one side of each line segment of a set of predefined polygonal barriers.
Given height above sea level, temperature, and the x and y components of wind
velocity, the module derives a mixing depth and a mixing depth time derivative.
The mixing height is computed by passing a straight line of slope -.00986
(the dry adiabatic lapse rate) through the trajectory height-temperature point
and determining the abscissa of its intersection with one of two polygonal
temperature-versus-height radiosonde curves. The appropriate curve is selected
on the basis of input time.
The mixing height derivative is computed as a linear function of the temperature
and hiiight-above-sea-level derivatives, which in turn are derived from inter-
polating equations for the temperature and height and the interpolating
equations for the x and y components of velocity. The mixing depth and mixing
depth derivative are computed by subtracting interpolated height-above-sea-
Level and derivative from mixing height and its derivative
2.0 DERIVATIONS
2.1 Definition of the Barrier and Permissible Stations
Each barrier is defined by an integer N , which is one greater than the number
of barrier sides, and a sequence of points (x., y.), i*l, 2, . . t N , with
each consecutive pair of points defining a sicle or the barrier. Let (x , y )
and (x , y ) be, respectively, the coordinate of the trajectory point and a°
given station. Let m=(y -y )/(x -x ), b=y -mx be the slope and y the intercept
of the line joining the trajectory point to the station. For each i=l, 2, . . .,
NB-1 let
-------
X VmiXJ-bi (2-1)
(2-2)
A station located at (x ,y ) is said to be permissible if, and only if, for
each 1.1, 2 ..... NB-l8either ^iP.X^l.jj^O or y^l.
fl. INTERPOLATION OF METEOROLOGICAL VARIABLES
Let (x . »y.) be the coordinates of each permissible station (j"l,2,...N ).
Let, d^.d^.d-,... ,d be the distance from these stations to the trajeBtory
point and let x.. »x_,*). . . . ,£.. be, for example, the corresponding set of
observed x components of winfl velocity. The x component of wind velocity,
x , at the trajectory point is estimated according to:
N
Each of the five remaining trajectory point variables is estimated by the
corresponding analogue to equation (2-2.1).
The time derivative, p , of, say pressure, is computed according to
, N
P°"2 { BP°" J-l J
where N N
~ >2 „ -
and
J-l J-l
Each of the remaining trajectory time derivatives is estimated analagous to
(2-2.2).
II-2
(2-2.3)
-------
C. COMPUTATION OF MIXING HEIGHT AND MIXING HEIGHT DERIVATIVE
Civen the interpolated values (z ,T ) of height al>ovc sea level and temperature
at our trajectory point, the module selects the appropriate A.M. or P.M. radio-
sonde curve depending on the value t . liach of these curves is defined by an
integer N or N and a sequence of points (z. ,T. ) k=l,2,...N , N.
a.m. p.m. ^ r k k a.m. (p.m.)
The mixing height, h , is computed according to:
T -T. 4n> z.-az
h = o k K. k o
V
where k is the first integer such that l anda=--00986 °C/meter.
Tho mixing height derivative, h , is computed according to:
111. Svmbol Glossary and Units
x [meters/minute] = x component of wind velocity
| motors/minute ] = y component of wind velocity
yo
T [degree centigrade (°C)] = temperature
p [millibar] = pressure
H [ (iimens Ion less ] = relative humidity
w [ dimension less ] = cloud cover
/. [meters] = height-above-sea level
o
h [meters] = mixing height
h [nu'tcrs/miuutc ]
in
II-3
-------
T [°C/minute)
o
P [millibars/minute]
o
H [I/minutes]
o
w [I/minutes]
0
z [meters/iainute]
o
11-4
-------
ULTRAVIOLET IRRADIANCE MODULE EQUATIONS
1 .0 INTRODUCTION
The basic physical hypotheses underlying this module's computations have been
taken almost directly from [1]. Our departure is in the formulation of a
total irradiance equation which takes into account albedo and weather effects
and which reduces to a standard form (see equation 11-29 of [1]) when the
cloud cover and albedo are both zero. Thus, our final formulation (see equation
2-7) will differ from any developed on the basis of equation (11-38) of [1].
The input to the module is the time t , the height above sea level of the
trajectory point, z , the relative humidity at the trajectory point H , the
corresponding atmospheric pressure p , the mixing depth, h , the weather
o m
cover, w , and the average concentration of N07, C^Q , in the column. The
module's output is the specific absorption rate of N0_, k^ .
£• N\)j
2.0 DERIVATIONS
As a function of the time of day, t (Pacific Standard Time), a value of the
solar zenith angle 0, is interpolated from a table derived from the equinox curve of
Figure 2. in Chapter II of [1], The air mass, m, is interpolated as a function
of 9 in Table 2 of [1].
Depending on an input cloud-type parameter, a value of average fraction of
radiation transmitted, 0, is interpolated in Tabl* 10 of [1], For each
wavelength band of 100A in a wavelength region of 3000A £ X £ 4000A, the
molecular scattering coefficient (s .) is selected from Table 3 of [1], From
IDA o
equation (II-9) of [1], the transmissivity relative to molecular scattering,
T , may be computed from the formula
ttlA
10«10 TmX a -(SmX>
where p is the interpolated pressure in millibars.
The transmissivity relative to particle scattering, T ., may be computed from
the formula (11-11) of fl] (with X in microns) pX
log1Q T x - -(3. 75x10" 3X~2W+3.5xlO~2X~*75d)m (2-2)
II-5
-------
where W is the amount of precipitable water vapor In the atmosphere, measured
vertically above the trajectory point, computed as in equation (II-2) of [1],
in units of centimeters, according to:
W - 1.722xlO-VJ10~(Z°/22'000) (2-3)
W
Where PW is the partial pressure in millibars of water vapor at the trajectory point.
PW is interpolated from a vapor pressure table in [2], and d is a dimensionless
parameter of preset nominal value 1.
The ozone absorption coefficient a. [meters ] is interpolated from Table 5 and
A
the transmissivity of ozone, T . , as in equation (11-13) of [1], aay fca computed
from
lo*10 TaX ° -QX(03>m <2-4>
where (0.) is a preset parameter of nominal value 2.2x10*" [meters] (of ozone).
The mean solar spectral irradiance, IQ , is selected from Table 1 and the
direct and sky radiation I,, and 1 . are computed from equations (11-15) and
(11-16): dA 8A
- TsX TaX 'OX COs6
(2-6)
where T , =• T ,T , and g is a preset dimensionless parameter of nominal value .5.
SA mA pA
Following the development from equations (11-17) through (11-28) of [1] and
taking average weather cover and an average albedo effect into account, we
obtain the formula for the average rate of absorption per unit pollutanto
column volume for direct and sky radiation over a wavelength band of 100A
centered at wavelength X, I .:
Iax
h
m
where_ a is the albedo (preset nominal value of .235), wo is the weather cover,
and C.. is the average column concentration.of NO., a. now refers to N0_.
N02 4 A *•
Finally, the specific absorption rate of M>2> k», derived from equations (11-31)
and (11-32) of [1] is computed as:
II-6
-------
. 4000A
lfi/'273+T\« T
•!• ... in~ibi 2.1 > I ,
X iU I - I / . 3A
7.2
I \f ^»»^ I —
»300oX
3. UNITS
All lengths are in meters.
Time Is in minutes.
Temperature is in C.
AJ1 pressures are in millibars.
-2 -1 ° -1
I are in photons meters minutes (100A)
CNO l3 i
k is in minutes
w is dimensionless.
o
Is dimensionless.
REFERENCES
[I] P. A. Leighton, Photochemistry of Air Pollution, Academic Press, New
York, 1961.
12] Handbook of Physics and Chemistry.
II-7
-------
KINETICS MODULE EQUATIONS
1.0 INTRODUCTION
We first characterize an abstract k step -N species constant mass, constant
volume chamber photochemical mechanism. As a consequence of this character-
ization, using standard principles of chemical kinetics, we can write, in
compact notation, the differential equations of species concentration for
the constant volume, constant mass situation. Next, we expand our conceptual
framework to the chemical kinetics of the contents of a pollutant column whose
motion over a time-varying field of urban emission sources induces variations
in average column temperature, volume mass, etc. This is accomplished by
adding a series of appropriate first order pertubation terms to the species
concentration time derivatives. It is to be emphasized that the step number
and number of species characterizing our imbedded kinetic mechanism are limited
only by computer storage and computation time considerations. For this reason
we are utilizing the IBM 360-67, a high-speed, large storage computer, In con-
junction with an auxiliary FORTRAN system whose function is to derive a set
of differential equations from a corresponding set of photochemical equations.
Thus, we have complete flexibility in step number, species number and, more
Importantly, the photochemical equations themselves.
The Inputs to the module are the outputs of the weather, diffusion, and
ultraviolet irradiance modules.
2.0 DERIVATIONS
An abstract k-step, N -species, constant-volume, constant-mass photochemical
mechanism is characterized by (i) a binary valued kxN matrix B whose (1, J)
element, b , is 1 if and only if the jth species appears as a reactant In the
1 (elementary) step; (11) an integer valued kxN species conservation matrix,
P, whose (1, J) element, P ., is the number of molecules of species j produced
(p.,.i>0) °r consumed (p..<0) in the ith step; (ill) a K-element vector F of
reaction rate factors k., j " 1, 2....K; and (iv) a N - element vector, C ,
of initial concentrations, c»(0), £•!, 2,...,N . From the above definitions and
elementary considerations of chemical kinetics, we may formulate the following
set of differential equations:
II-8
-------
n
\
>/
(2-1)
c.co) - cio
or, in matrix-vector form:
C(t) - P R(t)
(2-2)
C(0) - CQ,
where
c(t)
c2(t)
c3(t)
c (t)
No
(2-3)
II-9
-------
R(t)
n
n ca(t)
(2-4)
and P is the transpose of P.
Taking equation (2-2) as our starting point, we permit F to be time-varying
(i.e., k. are replaced by k.(t)) and add first order perturbations due to
J , J •
stationary sources, C0(t), traffic sources, C_(t), and the rate of change of
O 1
column volume, C.(t), to derive the pollutant column concentration differential
equations:
C(t)
(2-5)
C(0)
where
0 if h <0
m-
-h
•r0- C(t) if h >0
n_ m
m
(2-6)
3.0
UNITS
All concentrations are in parts per hundred million. Concentration rates are in
pphm/minute. h (mixing depth) is in meters h (mixing depth time derivative)
tu m
Is in meters/tninute.
11-10
-------
TRAFFIC MODULE EQUATIONS
1 . INTRODUCTION
We first note that the present module represents a "first pass" attempt
to provide working traffic emission inputs for the photochemical simulation
model. As such a prototype it is necessarily based on a set of simplifying
assumptions leading to a compact formulation which, nevertheless, takes
Into account such parameters as total daily vehicle mileage, area
population ratios, emission rates, diurnal traffic variations and local
meteorology.
In characterizing the module, the Los Angeles Basin is subdivided into a
#rid system of 784 equal 4-square-mile squares corresponding to a 56- by
56-mile square total area. To each of these grid areas is assigned a
daily vehicle mileage ratio representing the ratio of the daily mileage
in this subarea to the total daily mileage. The module is further
characterized by a set of pollutant emission rates given in grams per mile,
a traffic density function, and a total vehicle mileage figure. The emission
rates are computed from published emission factor data by an auxiliary
preprocessing program. The inputs to the module are the time, t, the position
of the pollutant column, (x,y), the temperature, T , the pressure, p , and
the mixing-depth, h , at (x,y) and their corresponding time derivatives,
T , p , and h .
0*0 m
The module's output is the vector of concentration rates, c_, due to traffic.
2 . DERIVATIONS
Given the position (x,y) (in miles) such that 16
where [z] = integer part of z.
Let I. (a preset constant of nominal value 1.2x10 ) be the daily vehicle
mileage. Let E. and m be, respectively, the average emission rates
(grams/mile) and the gram molecular weight of the jth pollutant (j-i, <_,... ,N )
11-11
-------
Then, assuming uniform areal distribution, the number of moles of the jth
2
pollutant emitted by traffic in a 1 meter subarea of the ita grid up
to time t, n (t), is given by:
^ t
- r LE f
n4(t) - 9.64x10 -i—i / f(s)ds (2-2)
J m4 J
J 0
where (see Figure II-l)
a if 0
-------
DIURNAL TRAFFIC VOLUME CURVE
.14-
.12-
.10-
.08-
o
.06-
.04-
.02-
0 2
—1 r-
4 6
—T~
8
10
12 14
HOURS
16 18 20 22 24
FIGURE II-I.
11-13
-------
UNITS
h [meters]
in
p [millibars]
w
h [meters/rain.]
m
o [millibars/min]
G
r [dimensionless]
L [miles]
E [grama/mile]
f(t) [1/hours]
11-14
-------
STATIONARY STACK SOURCE (DIFFUSION) EQUATIONS
1.0 INTRODUCTION
The.1 Stationary Stack Source or Diffusion module computes the contribution
to the concentration derivatives of NO and NO in a pollutant column with
position (x ,y ) and velocity (x ,y ) due to presence of upwind stationary
stack sources. The set of such sources is characterized by an integer N
s
and a N x9 matrix of stack parameters (x,y,z,H,D,V,T,P,Q),
s e s J s' s s s s s s xs '
s-l,2,..,N , where (x ,y ) is the position of the stack, z the height of its
o S S S
base above sea level, H its height, D its diameter, V the exhaust speed of
ia S S
its gas, T its inside temperature, P its ambient atmospheric pressure, and
S S
Q its emission rate of NO. Module computations are based on the assumption
of small variation in wind velocity both spatially and temporally for "nearby"
upwind stack sources. This supposition allows us to use the standard stationary
source diffusion equations to be found, for example, in reference (1). It
is to be noted that such a hypothesis serves to emphasize the "credible episode"
nature of the photochemical modle.
The inputs to the module are column position X, column velocity X_, pressure
p , temperature, T , mixing depth, h , height above-sea-level, z , the correspond-
ing time derivatives and the time t . The output is total stationary source conce
concentration derivatives for NO ana NO . '
2 . 0 DERIVATIONS
Lot X- (x ,y ) and X=x ,y ) be the position and velocity vectors of a pollutant
column. For each stack source s=l,2,...,N whose position vector X =(x ,y )
S o S S
s (X-X ).X>0, the module determines the corner of the column base
--
nearest the stack from among the four candidates: X + 1 (x ,y ) + 1 (y ,-x )
___ 0 o _ _ o o
Denoting tins nearest corner position as j( , the downwind distance from the
stack to the nearest corner, C . is computed according to:
s
=(XS-X )-X/s
\ / '
s = ,!XM =x x + y
(2-J)
11-15
-------
The cross-wind distance from the centerline, n > i« computed according to:
Following (1\ from equation (3.1) on page 5 we may compute the mass
contribution of NO , M from stack s to a column of unit square cross
X S
sectional areas and height h (mixing depth) according to:
(2-2)
n+1 h
f f f
J J J X
x«C y«n Z»Z
(x,y,z)dzdydx
+H
(2-3)
where K »h +z (mixing height) and
m m o
Qg i r
]}.
Denoting the molar contribution of NO and NO. from stack s to the column as
n^^ and n.^, respectively, assuming that these molar contributions are
proportional to the existing column concentrations Cun and CMrt , and taking
HU HU-
into account the conservation of mass, we haver
CNOn2s
or, solving for n , and n :
11-16
-------
nls " 30(CN04CHO ) "a (2-7)
So.
_ £ _ u
n28 * •»*'-•~ r «-
From the Ideal gas law, we compute Che number of moles of air in the
column, n, according to:
pohm
1013.25 RT '
o
and we compute the total upwind stacks to column concentration contributions
of NO and NO., C..n „ and Cu. c» respectively, as:
i NU|S NO-,*
io8^ 2
CNO, S " T { s: (X-XJ -X>0 ^ i8 (2-9)
or
CNO,S ' 2-77xlOJUNOVW (2-10)
CN02,S ' 2
where
UN02
11-17
-------
V - (T +273)/p h
o o m
£ "•
W -
Differentiating with respect to time:
So •
where
- 2.77xlO:'(U
Mn
9 2 2
"NO -
VW),
(2-11)
(2-12)
U
N0
V -
W -
{8:o} * * '
5+1
.
o sv
x- C
2cr (X) z
11-18
-------
where "h" is the mixing height derivative, z is the time derivative of height
m o
• •
above-sea-level, T is the column temperature derivative, p , is the column
° fu / 3 \
pressure derivative, erf (u) • j. lexpl-u ) du.
-------