United States
Environmental Protection
Agency
Office of Air Quality
Planning and Standards
Research Triangle Park NC 27711
EPA-450/4-80-001
January 1980
Air
Assessing the
Representativeness
of Ozone Monitoring Data
Final Report
-------
EPA-450/4-80-001
Assessing the Representativeness
of Ozone Monitoring Data
Final Report
by
F.L Ludwig and E. Shelar
Atmospheric Science Center
SRI International
Menlo Park, California 94025
Contract No. 68-02-2548
SRI Project 7863
EPA Project Officer: Warren P. Freas
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Air, Noise, and Radiation
Office of Air Quality Planning and Standards
Research Triangle Park, North Carolina 27711
January 1980
-------
This document is issued by the Environmental Protection Agency to
report technical data of interest to a limited number of readers.
Copies are available free of charge to Federal employees, current
contractors and grantees, and nonprofit organizations - in limited
quantities - from the Library Services Office (MD 35), U.S.
Environmental Protection Agency, Research Triangle Park, North
Carolina 27711; or, for a fee, from the National Technical Information
Service, 5285 Port Royal Road, Springfield, Virginia 22161.
This report was furnished to the Environmental Protection Agency by
SRI International, Menlo Park, California 94205, in fulfillment of
Contract No. 68-02-2548. The contents of this report are reproduced
herein as received from SRI International. The opinions, findings, and
conclusions expressed are those of the author and not necessarily those
of the Environmental Protection Agency.
Publication No. EPA-450/4-80-001
11
-------
ABSTRACT
This project has been directed toward defining the extent to which existing ozone monitor-
ing data can be used to represent conditions outside the area immediately surrounding a moni-
toring station. The most important concern has been with the National Ambient Air Quality
Standard (NAAQS) for ozone, seeking relationships between observed second-maximum hour-
average ozone concentrations and the area within which it is possible to state with reasonable
probability that NAAQS have or have not been exceeded.
Areas within which the National Ambient Air Quality Standard (NAAQS) for ozone is likely
to have been exceeded are shown to be related to the observed annual second-maximurn
hourly ozone concentrations. All pairs of SAROAD stations separated by 500 km or less that
had nearly complete annual data sets were considered for the years 1974 to 1977 to determine
the probability that the NAAQS would be exceeded at one station of the pair, given the
observed second-maximum hourly ozone concentration at the other and their separation.
Applying this relationship to current SAROAD data for 1977 and 1978, circles were drawn
around each SAROAD monitoring site to show the area within which it is 95 percent probable
that the 120 ppb NAAQS has been exceeded during 1977 or 1978.
The report describes meteorological situations and seasons during which high ozone con-
centrations are most likely in various parts of the country, so that special monitoring studies
can be designed to supplement existing data. An aircraft monitoring protocol is given that will
provide ozone observations that are representative of ground-level conditions.
This report was submitted in fulfillment of Contract 68-02-2548 by SRI International under
the sponsorship of the Environmental Protection Agency.
-------
CONTENTS
Abstract Mi
Figures v
Tables vii
Acknowledgments viii
Executive Summary S-1
1. Introduction 1
Purposes of the research 1
Approach to the study 1
Organization of the report 1
2. Major findings 3
Relationships among ozone observations at separate sites 3
Factors causing observed ozone concentrations to be unrepresentative 36
Relationships of ground-level ozone concentrations to those aloft 42
3. Supporting Material 46
Data survey and selection 46
Identification of conditions associated with widespread exceedances
of the ozone standard 51
Analysis of the conditions associated with the different ozone profile types 52
Station characteristics likely to result in unrepresentative readings 59
4. Recommendations 60
Areas where more data are required 60
Methods for obtaining additional data 60
References 68
-------
FIGURES
Number Page
S-1 Probability that the annual second-maximum ozone concentration exceeds 120 ppb
within 50 km of a station, given the observed value for that station S-2
S-2 Probability of observing an exceedance of the NAAQS for ozone as a function
of observed second-maximum ozone concentration and distance S-4
S-3 Areas where it is 95-percent probable that hourly ozone concentrations in excess
of 120 ppb occurred during 1977 or 1978 S-5
1 Scattergram of the distance between monitoring stations and the absolute difference
between the second-maximum ozone concentrations for stations in the Northeast states.... 5
2 Scattergram of the distance between monitoring stations and the absolute difference
between the second-maximum ozone concentrations for stations in the Midwest states 6
3 Scattergram of the distance between monitoring stations and the absolute difference
between the second-maximum ozone concentrations for stations in the Southwest states. ... 7
4 Scattergram of the distance between monitoring stations and the absolute difference
between the second-maximum ozone concentrations for stations in Southern California .... 8
5 Geometric means of difference in second-maximum ozone concentration for the Northeast ... 13
6 Geometric means of difference in second-maximum ozone concentration for the Midwest .... 14
7 Probability as a function of distance between stations that ozone will exceed 120 ppb
if the annual second-maximum ozone concentration has been observed to exceed
specified thresholds for all U.S. monitoring sites for 1977 15
8 Probability that the annual second-maximum ozone concentration exceeds 120 ppb
within 50 km of a station, given the observed value for that station 18
9 Observed probabilities of an exceedance of the ozone standard for the different station
separations as a function of the second-maximum ozone value 19
10 Probability of observing an exceedance of the NAAQS for ozone as a function
of observed second-maximum ozone concentration and distance 20
11 Areas where it is 95-percent probable that the NAAQS for ozone has been exceeded based
on SAROAD stations with more than 7000 hours per year for the years 1977 and 1978 ... 22
vi
-------
Number Page
12 Areas where it is 95-percent probable that the NAAQS for ozone has been exceeded
based on all SAROAD data for 1977 and 1978 23
13 Example of high ozone concentrations associated with the southwest quadrant
of a high-pressure system over the Texas-Louisiana Gulf Coast area, 4 June 1977 28
14 Example of high ozone concentrations associated with the northern part
of a high-pressure system in the northeastern U.S., 4 July 1977 30
15 Example of high ozone concentrations in warm air near a weather front
in the northeastern U.S., 16 July 1977 31
16 Example of high ozone concentrations in warm air near a weather front in Texas,
31 May 1977 32
17 Example of high ozone concentrations in the northwest quadrant of a high pressure system
in the Midwest, 28 July 1977 33
18 Example of high ozone concentrations in the southwest quadrant of a high pressure system
in the Midwest, 27 May 1977 34
19 Example of high ozone concentrations in warm air near a weather front in the Midwest,
25 May 1977 35
20 Example of a typical pressure pattern associated with high ozone concentrations
in Southern California, 7 September 1977 37
21 Estimated ozone depletion at distances from a roadway 38
22 St. Louis ozone concentrations on the morning of 19 July 1976 39
23 Average August 1977 afternoon ozone concentrations (ppb) in Houston 40
24 Six types of vertical profile 44
25 Scattergram of average mixing-layer ozone concentrations as a function
of surface concentrations for profiles A and D 45
26 Locations of SAROAD Northeast ozone monitoring stations for 1977 47
27 Locations of SAROAD Midwest ozone monitoring stations for 1977. 48
28 Locations of SAROAD Southwest ozone monitoring stations for 1977 49
29 Locations of SAROAD ozone and oxidant monitoring stations in Southern California for 1977 . 50
30 Histograms of ratios and differences between surface and mixing-layer ozone
concentrations for different atmospheric stability classes 57
31 Schematic diagram of aircraft ozone-measurement program designed
to estimate ground-level concentrations 65
vii
-------
TABLES
Number . Page
1 Cross-Tabulation of Distance and the Absolute Difference Between Second-Maximum
Ozone Concentrations for the Northeast States 9
2 Cross-Tabulation of Distance and the Absolute Difference Between Second-Maximum
Ozone Concentrations for the Midwest States 10
3 Cross-Tabulation of Distance and the Absolute Difference Between Second-Maximum
Ozone Concentrations for the Southwest States . . '. 11
4 Cross-Tabulation of Distance and the Absolute Difference Between Second-Maximum
Ozone Concentrations for Southern California 12
5 Joint Frequency of Occurrence of Annual Second-Maximum Ozone Concentrations
at Stations Separated by less than 50 km 17
6 Number of Days per Month in 1977 with Daily Maximum Ozone in Excess of 120 ppb
in Selected Regions of the U.S 24
7 Ozone Concentrations in Excess of 120 ppb and Associated Weather Features
in Various Parts of the Eastern United States During 1974 26
8 Selected Days with Widespread High Ozone Concentrations in Different Regions During 1977
During 1977 _ 27
9 Weather Features Associated with Widespread High Ozone Concentrations
in Selected Regions of the United States During 1977 29
10 Threshold of Spatially Averaged Peak-Hour Ozone Concentrations Used to Select Dates
for Case Studies 51
11 Frequency of Occurrence of Different Profile Types at Different Times of Day 54
12 Frequency of Occurrence of Different Profile Types at Different Locations
Relative to the City 55
13 Comparison of Ozone Concentrations at the Surface and Aloft by Stability Class 56
14 Profiles Observed During Unstable Conditions Outside Cities from 1100 and 1600 CST 58
viii
-------
ACKNOWLEDGMENTS
Many people and organizations have been helpful during this study. We are especially
indebted to the Project Officer, Mr. Warren P. Freas, for his useful suggestions. At SRI, R. Tru-
deau, S. Webster, A. Smith, and J.H.S. Kealoha assisted in various phases of the project. The
efforts of the publications group of the SRI Advanced Development Division are gratefully ack-
nowledged.
Ix
-------
EXECUTIVE SUMMARY
Introduction and Objectives
This project has been directed toward defining the extent to which existing ozone monitor-
Ing data can be used to represent conditions outside the area Immediately surrounding a moni-
toring station. The most important concern has been with the National Ambient Air Quality
Standard (NAAQS) for ozone, seeking relationships between observed second-maximum hour-
average ozone concentrations and the area within which it is possible to state with reasonable
probability that NAAQS have or have not been exceeded.
Another important objective has been to identify site characteristics that make data
unrepresentative of surrounding areas. The meteorological conditions associated with high
ozone concentrations In different parts of the country have also been investigated. Finally, all
the information has been combined to specify areas where further monitoring may be needed,
and to define methods by which special monitoring programs can be used to meet those needs.
The approach taken during this study has used only existing data. The emphasis has
been on Information available from the system for Storage and Retrieval of Aerometrlc Data
(SAROAD). Some data from special studies have also been Incorporated, especially to deter-
mine conditions under which aircraft measurements can be used to estimate ground ozone
concentrations.
Relationships Among Ozone Observations
at Separate Sites
The focus of this study has been on exceedance of the NAAQS for ozone. To this end, a
list of observed second-maximum hourly ozone (hereafter referred to as the second-maximum)
concentrations was compiled for all stations and years for which more than 7,000 hours of data
are available. By taking all possible pairs of such stations that were within 500 km of one
another, it was possible to determine the probability that the NAAQS would be exceeded at one
station of the pair, given the observed second-maximum ozone concentration at the other and
their separation. Figure S-1 shows the probability that the annual second-maximum ozone con-
centration would exceed 120 ppb within 50 km of a station, given the observed value for that
station. Similar presentations were prepared for other station separations. The curve In Figure
S-1 has the expected characteristics. It shows that at very low observed annual second-
maximum ozone concentrations, the probability of observing an exceedance of the NAAQS Is
quite small, but as expected, the probabilities increase within increasing observed annual
second-maximum ozone concentrations. For locations within 50 km of an observing station, it
becomes virtually certain that an exceedance of the standard will be found If the station
observed a sufficiently large second-maximum ozone concentration. Figure S-1 shows that the
probabilities reach 90 percent at a little less than 150 ppb, 95 percent at about 175 ppb and
99 percent at 275 ppb. These statistics are heavily Influenced by data collected In urban areas
S-1
-------
1/9
O
IU
X
0
O
oc
a.
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360
OBSERVED ANNUAL SECOND-MAXIMUM OZONE ppb
Figure S-1. Probability that the annual second-maximum ozone concentration exceeds 120 ppb
within 50 km of a station, given the observed value for that station.
S-2
-------
where ozone gradients are apt to be greater than in more remote locations. Therefore, the pro-
babilities of observing an NAAQS exceedance are expected to be greater in a remote area, for
any given observed second-maximum ozone concentration, than in the urban areas.
There is a relationship between the annual second-maximum ozone concentration
observed at a location and the distance from the station at which the probability of observing
an exceedance of the NAAQS for ozone falls below a specified level. Figure 8-2 shows the
relationships that were found for three different specified probabilities as derived from plots of
data such as that presented in Figure S-1. Qualitatively, the relationship shown in Figure S-2
is as expected. For example, if the concentration of 120 ppb is observed, it provides little
guarantee that the NAAQS will be equalled or exceeded anywhere but at the station itself. As
the observed value increases so does the distance within which it is likely that the standard
will be exceeded. However, it is not possible to achieve the higher probabilities beyond some
cutoff distance, regardless of how great the observed second-maximum ozone concentration
might be because the physical and statistical relationship between observations separated by
great distances are exceedingly tenous. This is discussed in Section 2-A-3. Thus, the dis-
tance within which one might assume a 95 percent probability of an exceedance should not be
more than 175 to 200 km regardless of the observed second-maximum ozone concentration.
With a 90 percent probability, the cutoff is about 225 to 250 km.
Areas of Probable Exceedance
of the NAAQS for Ozone
It is obvious that curves like those in Figure S-2 can be used in combination with
observed values of maximum ozone concentrations to define areas within which exceedances
of the NAAQS for ozone are likely to have occurred. Figure S-3 is a map of the United States
that indicates those areas where it is 95 percent probable, on the basis of observed annual
maximum ozone concentrations, that the NAAQS for ozone has been exceeded during 1977 or
1978. The annual maximum was selected as the design value for the data set based on the
procedures described by Curran (1979). Figure S-3 is based not only on data from those sta-
tions with nearly complete data sets, but also on data from sites that had fewer than 7,000
hours per year. Even though a data set is not complete, only one observation in excess of 120
ppb may be sufficient to define an area within which exceedances are likely. In general the
area defined by an incomplete data set will be smaller (it can not be larger) than it would be if
derived from a complete data set. In Figure S-3, the symbol 'L' is used to mark locations where
the observed maximum concentration was less than 120 ppb.
Data from special studies, or data not otherwise archived in the SAROAD, could be used to
characterize additional areas in Figure S-3 in regard to exceedance of the NAAQS. One of the
recommendations of this project is that an intensive search for such data be undertaken before
any special monitoring programs are planned. A search for, and use of, existing data would be
considerably more cost-effective than field monitoring projects.
Meteorological Factors Associated
with High Ozone Concentrations
If special monitoring programs are required, they can be planned more effectively if the
meteorological conditions associated with high ozone concentrations and exceedances of the
S-3
-------
350
325
300
275
250
225
§ 20°
K
ui
v>
as
°
O
IT
u.
u,
u
175
125
75
50
25
85%
90%
120 140 160 180 200 220 240 260 280 300 320 340 360 380 400
OBSERVED SECOND MAXIMUM ppb
Figure S-2. Probability of observing an exceedance of the NAAQS for ozone as a function
of observed second-maximum ozone concentration and distance.
S-4
-------
01
r\
Figure S-3. Areas where it is 95-percent probable that hourly ozone concentrations in excess of 120 ppb
occurred during 1977 or 1978.
-------
NAAQS are defined. It has been found during the course of this project that although there are
simiJarities in the relationships between synoptic-scale meteorological features and areas
where ozone concentrations are high from one region of the country to another, there are also
important differences. In most regions of the country, high ozone concentrations are frequently
associated with high-pressure systems, but the segment of the high-pressure system where
concentrations are highest tends to vary from one region to another. For example, the northern
part of high-pressure systems are most frequently associated with episodes of high concentra-
tion in the northeast part of the United States. In the central and northern Midwest, the
northwestern quadrant is most common. Along the Gulf Coast and in the Southeast, it is the
southwestern quadrant of high-pressure systems that is most frequently associated with high
ozone concentrations. In Southern California, the photochemical season is associated with the
presence of the semipermanent Pacific high-pressure system; Southern California is located
near the eastern margin of this system. However, the thermal low over the interior valleys and
deserts is the most persistent feature of the weather patterns observed on the highest ozone
days in Southern California.
Except in Southern California, the warm air in the vicinity of frontal systems also provides
a frequent site for ozone "episodes." Such areas could also be frequently classified as being on
the periphery of a high-pressure system, but it seems worthwhile to classify them separately.
There is a common tendency to think that frontal systems and air pollution episodes are
incompatible, but the prevailing conditions here are frequently atypical of high-pressure sys-
tems, with greater wind speeds.
Just as there are geographical differences with regard to the synoptic-scale weather
features that are associated with high ozone concentrations, there are also some geographical
differences in seasonal effects. Throughout most of the country, the summer months have the
highest ozone concentrations. In New England and in the southern Great Plains, high ozone
concentrations appear to be frequent from May through September. However, in the Ohio River
Valley and the area south of the Great Lakes, the months of most frequent high ozone
concentrations are July and August. Further to the south, along the Gulf Coast from Texas to
Florida, June and July are the months of most frequent high ozone concentrations. In the
northern High Plains, some of the highest ozone concentrations occur in the Spring, however,
anthropogenic sources of precursor emissions, found in the northern Great Plains or the Great
Basin, extend the season of highest ozone from spring through summer.
Station Characteristics Likely
to Result in Unrepresentative Readings
The second-maximum ozone readings available from SAROAD were used as one basis for
identifying stations that appeared to have unrepresentative data. Readings from stations within
25 km of one another were compared, and those pairs whose readings differed by more than
100 ppb were identified. Maps like the one shown in Figure S-3 were also examined for sta-
tions reporting compliances with the standard near the center of areas that had a high proba-
bility of exceeding the standard. The characteristics of the stations identified were examined;
the most frequent contributors to unrepresentative readings were NO sources. If a station is
located close to an arterial with high traffic density, the NO from that traffic can react with the
ozone to remove most of it. Thus, the presence of nearby NO sources can reduce ozone con-
centrations from rather high levels to a few tens of ppb or less. Even more common are sta-
tions that are not close to any single NO source, but are located within a general area of high
S-6
-------
emissions. Data from St. Louis, Denver, Washington, among other cities, suggest that the
scavenging effects of NO may reduce ambient ozone levels by several tens of ppb in heavily
urbanized areas.
An analysis of available information suggests that well-operated, well-calibrated ozone
monitors should provide data that are accurate to within 15 pbb or less when the new ultra-
violet assay calibration techniques are used. Gas phase titration calibration techniques provide
similar accuracies. Some older methods of calibrating the instruments can introduce errors of
several tens of ppb.
Recommenda tions
Methods for Obtaining Additional Data
The most efficient means to obtain data to characterize the NAAQS status of additional
areas would begin with a survey of non-SAROAD data sources. Data may be available from
EPA regional offices, local agencies, and industrial sources, or from special studies conducted
in the different regions, especially those where current monitoring is sparse. Presumably, such
a survey would reduce the number of areas that may need further monitoring.
The availability of such data might also provide a basis for better determination of the
degree to which areas that are in compliance can be specified from observations already avail-
able. The present data base seems to contain too few data from remote and rural locations to
be used to develop methods analogous to those used here to specify areas of exceedance of
the NAAQS.
One of the methods investigated for obtaining data from inadequately monitored regions
was the use of aircraft surveys. Earlier studies of the characteristics of vertical ozone profiles
were used to define the conditions under which aircraft data could be used to specify ground-
level concentrations. In general, aircraft observations can probably be used to specify ground
level ozone concentrations when the measurements are made under sunny conditions, between
about 1100 and 1600 local standard time (LST), in areas well away from cities. This should
present few or no problems, because the greatest interest will be in those situations likely to
produce the highest ozone concentrations, and these tend to coincide with the conditions
necessary to obtain aircraft data that will be useful for specifying ground-level conditions. High
ozone concentrations occur outside cities during the late-morning and afternoon hours of
sunny days.
The aircraft monitoring protocol that is described in this report specifies that the observa-
tions should be conducted during the seasons with the highest probability of ozone standards
exceedances, in the appropriate segments of weather systems, and during the appropriate
periods of the day. As further insurance that the aircraft data are in fact representative of
ground-level conditions, it is also specified that periodic vertical soundings be made to meas-
ure ozone, temperature, dew point and aerosol concentration. These measurements can be
interpreted on board the aircraft, if suitable on-board display capabilities are available, to
assure that the measurements are being made within a well-mixed layer and that the vertical
gradient of ozone concentration is small.
S-7
-------
SECTION 1
INTRODUCTION
A. Purposes of the Research
The study described in this report was undertaken primarily to determine the degree to
which existing historical data could be used to define areas where the National Ambient Air
Quality Standard (NAAQS) for ozone was either met or exceeded. Three corollary purposes
accompanied this major purpose. First, we sought to determine those factors that affect
representativeness of the ozone monitoring data: meteorological factors, the siting of the moni-
toring stations and the methods of measurement have been examined. The study also defines
areas where additional ozone monitoring may be necessary to provide information on existing
ozone levels. Finally, the characteristics of an air/ground monitoring program that can be used
to characterize the United States with regard to areas of compliance with the NAAQS for ozone
are described.
B. Approach to the Study
It should be understood that this was a "paper study," and only existing data were used.
The major source of data has been SAROAD (Storage and Retrieval Of Aerometric Data). Data
from SAROAD have been examined for agreement between observations at one site and those
from other sites at varying distances. The data have also been used to identify times of
widespread high ozone concentrations, so that the meteorological conditions associated with
such events could be identified. We have also used the data base to attempt to identify a few
stations that are in disagreement with their neighbors, to see if they exhibit common siting
characteristics.
The results of these analyses have been used to define geographical areas where moni-
toring is not now adequate to determine whether the NAAQS for ozone is exceeded. The
results with regard to characteristics of nonrepresentative monitoring sites should be useful in
the future in siting monitors that will provide more widely usable data. The studies of the
meteorological conditions associated with widespread exceedances of the ozone standard have
been incorporated into some recommended strategies for surveying areas to determine whether
the standard is exceeded.
C. Organization of the Report
The next section of this report provides an extended summary of the major findings and
includes only enough background information so that the findings will be intelligible. The
remainder of the supporting material, from which the findings were derived, is discussed in the
third section of the report. The last section presents recommendations that have been derived
from the analysis.
-------
The authors hope that the organization used for this report will make it easier for a reader
to digest the conclusions as a whole. With the conclusions already stated, it should be easier
for a reader to interpret the supporting material. Furthermore, if only certain parts of the study
are of interest, it should be easier to identify the related supporting material.
-------
SECTION 2
MAJOR FINDINGS
A. Relationships Among Ozone Observations
at Separate Sites
1. Background
If ozone concentrations are spatially continuous, two locations that are geographically
proximate should have similar ozone concentrations, whether the ozone concentration under
consideration is for a particular hour, the highest hour-averaged concentration for a particular
day or the second-maximum hour-averaged ozone concentration for the year, so long as the
field for the type of observation is spatially continuous and relatively smooth. With this reason-
ing in mind, some important questions can be asked regarding the concept of representative-
ness. For example, one could examine the average difference between readings at separate
stations, and determine how that difference varies as a function of the distance between sta-
tions.
For the purposes of this study, it is more important to know how well exceedances of the
NAAQS for ozone can be determined from existing data. For example: If a particular concentra-
tion of ozone has been observed at one location, can that observation be used reliably to deter-
mine whether or not the NAAQS is exceeded at some other, nearby location? The answer to
that question depends in large part on:
The type of ozone concentration being consideredfor example, the answer may be
different for same day comparisons than for comparisons on an annual basis.
The magnitude of the observed ozone concentrationthe observation of a very high
ozone concentration indicates that the standards are more likely to be exceeded at
nearby locations. Similarly, very low concentrations are more likely to be associated
with compliance nearby.
Separation between the point for which an observation is available and the point of
interest in determining whether or not the standards are exceeded.
Location and siting factorsa station that is strongly affected by local and anomolous
effects can seldom be used reliably to estimate conditions in surrounding areas.
The effects of the first three factors listed above have been quantified by analyzing ozone
observations from pairs of stations. The details of the analyses are described in greater detail
later. Briefly, they included stratification according to the distance separating the two stations
of a pair (e.g. 0-50 km, 50-100 km, etc.). The subsets of data were then analyzed to determine
the average absolute difference between ozone observations at the two stations, for example,
and the conditional probability that the NAAQS was, or was not, exceeded at one of the stations
of a pair, given the observed ozone concentration at the other.
-------
2. Average Difference Between Observed
Ozone Concentrations as a Function
of Separation Between Stations
In order to study the possibility that there are differences in the relationships among
observations at different sites from one geographical region of the country to another, some
analyses have been performed on subsets of data from different regions of the country. The
four regions considered were:
The Northeast
- Connecticut
- Massachusetts
- Rhode Island
- New Jersey
- New York
- Pennsylvania
. The Midwest
- Kentucky
- Illinois
- Indiana
- Iowa
- Missouri
- Ohio
. The Southwest
- Arkansas
- Louisiana
- Oklahoma
- Texas
Southern CaliforniaCalifornia south of latitude 34" 22' (roughly south of Santa Bar-
bara).
Figures 1 through 4 show scattergrams relating the distance between stations to the abso-
lute value of the difference between their annual hourly second-maximum ozone concentration
(hereafter referred to as the second-maximum). Only stations where more than 7,000 hours of
data had been collected during the year were included in the samples. In these figures, each
asterisk indicates one case of that particular combination of station-separation and absolute
difference between observed second-maximum ozone concentrations. Numerals less than nine
indicate that a corresponding number of cases had occurred with that particular difference in
reading and spatial separation; the numeral 9 indicates nine or more such cases. Tables 1
through 4 are cross-tabulations of the data shown in the corresponding figures.
-------
661.6*
O
K
I-
oo
Z
uu
529.*>
«63.*0
3«T.»T
331
UJ
CD
§ 265.11
<
-------
l">2.*6
io«2.«o
967.94
8*2.08
in
O 721. »
u>
»«3.AS
U)
m
§ 483.-9
<
(E
<
ui )63.{3
23&36S32?23*33?**
3779(.696435323673*74223?3 6*23*2
1*2227532424*2* ^*4Z2**2*
I»«*2?2?»3 ?3 * 3 *
ti2 *42»5446*3233*3j*2 3**4*3*
2442*6*435423*2*23 *32
t *433*1*?2* *2?2*2 3
13*33 717* *7324 ?
U5*^H4*5* 44325**3 222* *2
<2S7253363623?24?M43**3
2
2
3 2
42*2 2
! 2 2
'
2
2
2
1
0 50.90 101. tO 15?. 70 203.60 254. 50 305.40 356.30 4Q7.20
ABSOLUTE DIFFERENCE BETWEEN SECOND-MAXIMUM OZONE CONCENTRATIONS
456.10
509.00
Figure 2. Scattergram of the distance between monitoring stations and the absolute difference
between the second-maximum ozone concentrations for stations in the Midwest states.
-------
1ST*.
u>
O
H
<
c/J
Z
UJ
III
T9?.»4
UJ
m
O
H 635.15
321.
164.A9
?
*
21.AO *3.«0 65.»0 87.20 109.00 130.8" liZ.60 I7».»0
ABSOLUTE DIFFERENCE BETWEEN SECOND-MAXIMUM OZONE CONCENTRATIONS
196.20 216.00
Figure 3. Scattergram of the distance between monitoring stations and the absolute difference
between the second-maximum ozone concentrations for stations in the Southwest states.
-------
00
?28.«7 »
I
?«-6.l9
183.41
160.O
o
< I3T.«8
z
UJ
01
z
UJ
to
119.«
92.10
69. «2
*6.T»
?3.<
2 2
3
12
1
2
2
-.
12
I
12
1
2
»7.0« 9*. 00 1*1.00 109.00 235.01' 282.00 929.00 3U.OO 423.00
ABSOLUTE DIFFERENCE BETWEEN SECOND-MAXIMUM OZONE CONCENTRATIONS jigm"3
4TO.OO
Figure 4. Scattergram of the distance between monitoring stations and the absolute difference
between the second-maximum ozone concentrations for stations in Southern California.
-------
TABLE 1 CROSS-TABULATION OF DISTANCE AND THE ABSOLUTE DIFFERENCE BETWEEN
SECOND-MAXIMUM OZONE CONCENTRATIONS FOR THE NORTHEAST STATES
ABSOLUTE DIFFERENCE BETWEEN SECOND-MAXIMUM OZONE CONCENTRATIONS ppb
51-lnn.N
O
fe
z
111
1LI
201-2SiK»
301-35AK"
JS1-401KM
401-45JK*
rUIIMT 1
CO* PCT li-2oP"
COL PCT T
TOI pry ! I.
1.1 S5
i »7.»
t 15.1
t S.n
J. 1 79
t «3.9
I (".»
I 2.6
>. j »5
I 44.1
1 12.4
T 4.1
4. i 47
I 35. J
I 12.9
! *-3
». 1 4?
i 3».l
' 11. 5
l 3.S
A. i 19
' i 17.3
i 5.2
i 1.7
T. j 19
1 27.9
I 1C. 7
i 3.*
«. T 2"
' T 22.5
T S.5
i i.e
9. I ?P
' i 25.7
! 7.7
I Z.?
)-. I 4'
*u I 34. R
l M.<-
I 3.6
COLUMN 30,4
TOT«I 33. r:
21.40°en
2.
37
31.9
f>.7
3.4
17
?«.«
(.3
1.5
27
2«..«
H'.C
?.4
2»
?l.l
ir.4
?.'
27
z?.o
i n.ii
'.4
27
24.5
)«.(>
?.«
23
IA.4
«.^
2.1
2?
?4.7
R.I
f.f
}>
'9.4
11.9
?.o
3-
?*.]
II. 1
7.7
?7n
?».«
41.60PPB
3.
10
8.6
*.?
.9
6
9.1
3.1
.5
1?
u.e
6.3
1.1
?5
u.e
13.1
2.3
?2
17.9
11. S
2.0
20
IB. 2
10.5
1.6
27
19.3
14.1
2.*
22
24.7
11.5
2.0
?*
23.9
13.6
2.*
?1
It. 3
11.0
1.9
191
17.3
61-80PPB
».
R
6.9
6.3
.7
5
7.6
«.C
.5
9
e.R
7.1
.»
17
12. t
13.5
1.5
IS
12.2
11.9
1.4
17
15.5
13.5
1.5
2P
1«.3
15.9
1.*
13
U.6
10.3
1.2
B
7.3
6.3
.7
14
12.2
11.1
1.1
12*
11. «
P1-100PP
f
5.
6
5.2
9.0
.5
3
4.5
4.5
.3
1 .0
1.5
.1
1"
7.S
14.9
.9
1)
A. 5
H.9
.7
11
10.0
!*.»
1.0
10
7.1
14.9
.9
7
7.9
1(1. «
.A
A
5.5
0.0
.5
^
4.3
7.5
.5
<,7
*.l
10I-1?OP
»-B
6.
(l
0
n
0
3
4.*
10. n
.1
o
n
n
n
1
.0
3.3
.1
*
3.3
13.3
.4
6
5.5
20.0
.5
9
f.,4
30.0
.»
2
2.?
6.7
.2
1
2.«
10.0
.3
2
1.7
6.7
.?
31
2.7
1?1-UO»
Ph
7.
0
0
n
n
i
1.5
6.7
.1
2
2.r
13.3
.2
1
.ti
6.7
.1
1
."
6.7
.1
4
3.6
26.7
.4
2
1.4
13.3
.2
n
0
(i
n
3
?."
20.0
.3
1
.9
6.7
.1
15
1.4
i6i>ie«p
PR
9.
0
0
0
0
n
0
0
0
1.0
25.0
.1
0
0
0
0
2
1.6
5H.O
.2
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
.9
25.0
.1
4
.4
490VE IB
fiPPB
10.
0
0
0
0
2
3.0
5.6
.2
>
».»
13.9
.5
4
3.0
11.1
.»
i
1.6
5.6
.2
6
5.S
16.7
.5
10
7.1
27. «
.«
3
3.*
8.3
.3
3
2.8
8.3
.3
1
.9
2.8
.1
36
3.3
B0«
mm
116
10.5
66
6.0
102
9.2
133
12.1
123
11.2
110
10.0
140
12.7
89
8.1
10*
9.9
115
10.4
1103
mo.o
-------
TABLE 2 CROSS-TABULATION OF DISTANCE AND THE ABSOLUTE DIFFERENCE BETWEEN
SECOND-MAXIMUM OZONE CONCENTRATIONS FOR THE MIDWEST STATES
ABSOLUTE DIFFERENCE BETWEEN SECOND-MAXIMUM OZONE CONCENTRATIONS ppb
c/i
2 151-21 UK-
UJ
m
O
201-7S**'-
301-35»«»
»H
35,8
4.1
1.7
44
40. C
1.9
44
57.*
5.
1.9
51
45.9
s.e
2.7
37.9
?!7
40.4
13.
5.3
21-40PP"
2.
45
7.1
7.0
25
3J9
1.1
33
31.1
1.4
37
33.6
l.»
U
?I?
2"
1R.O
3.7
41
74
11.7
3.7
1-1 T7
3«.4 1 ?f.7
11.* 1 17.1
4.4 1 1.4
317 i »6"
36. * | 31.'
35." 1 47.1
'3.9 I 11.7
3.
31
in. 3
7.9
1.4
11
13.3
2.P
.5
19
»!?
.«
IT
15.5
.7
10
11.'
7.6
.4
21.6
6.2
1.1
IB.O
7.4
1.3
39
12.9
10.0
1.7
2(1.4
IS. 4
7.6
150
17.4
3B.S
SB* (,34 390
3B.H 77. R 17.1
R
*.I i>.
4
j!«
.2
5
6.0
3.0
.7
9
5!5
.4
In
9.1
.4
'."
.4
7
6.3
is
9
5.6
.4
25
K.3
15.?
7?
7.5
13.3
1.0
7.7
40.0
2.9
1*4
7.?
7
1.7
7.9
.1
1
1.7
1.5
.0
,Q
l.S
.0
1
.9
1.5
.0
3
3.*
.1
4
i?
7
ids
11
.1.6
16.2
9
3.1
11.7
.4
79
1.4
47.6
1.3
IOI-170P
6.
3
1."
7.5
.1
1
1.7
7.5
.0
A
0
0
1
.9
7.5
.0
2
?.»
.1
1
,o
7.5
.0
5
3.1
17.5
.7
e
2.6
20.0
.4
R
2.7
70.0
.4
II
1.1
77.5
.5
AS 4rt
1.0 1 .R
171-UOP
Px
7.
3
n!f
.1
i
1.2
>.5
.r
3
2.0
13.*
0
0
0
r
1
1.2
.r
n
0
0
ft
1
.6
4.5
.r
6
2.r
77.3
.3
.'
9.1
.1
14I-160P
PR
«.
7
1.2
7.4
.1
2
2.4
7.4
.1
0
0
0
0
0
n
0
0
0
0
0
0
7!
2.
1*.
14.
7.
75.
5 I
. I
77.7 1 72.
.7 1
1
f
!
I61-1BOP
PR
9.
1
.6
6.3
.0
7
7.4
17.5
.1
1
.*
(.3
.0
(1
0
o
0
1
1.2
*.3
.0
0
0
0
0
0
0
0
0
5
1.7
31.3
.2
I
.7
17.5
.1
4
.5
75.0
.7
eovt ie
nopr
10.
5
3.
14.
.
0
0
2
1.9
i.9
.1
0
0
0
0
1
1.2
2.9
.0
2
I.I
!i
4
i.S
11. 1
.2
1
2.6
73.5
.4
6
17^6
.3
6
.7
17.6
.3
2? ?7 U 3«
1.0 1.7 .7 1.5
00.
707«l
169
7.4
B3
3.6
110
4.ft
e.
3.7
111
4.9
161
T.I
307
13.7
794
12.9
67
37.«
2207
100.0
10
-------
TABLE 3 CROSS-TABULATION OF DISTANCE AND THE ABSOLUTE DIFFERENCE BETWEEN
SECOND-MAXIMUM OZONE CONCENTRATIONS FOR THE SOUTHWEST STATES
ABSOLUTE DIFFERENCE BETWEEN SECOND-MAXIMUM OZONE CONCENTRATIONS ppb
0-50KM
I
o
<
. T 0 i n
T ft 1 o
T f. 1 ft
T ? I n
4. 1 r. I ?
T C 1 100. ft
1 P 1 «s,3
T ri ?.l
*. T 01 2
T 0 1 4ft. 0
T P 1 ?.3
T ft I 9.1
7. T " 1 1
T f 1 ?*.0
T " i 2.6
T 0 1 l.ft
«. T 51 3
T 50.0 1 3P.O
T 15.2 1 7.
1 ^3.6 1 71.1
.T. 1 ..
41-60PPR
3.
0
0
0
0
1
100.0
B.3
1.0
1
100.0
fl.3
1.0
0
0
0
0
2
4P.O
16.7
2.1
2
50.0
16.7
2.1
1
10.0
8.3
1.0
0
0
0
0
s
7.P
41.7
S.2
61-BoPPR
4.
P
0
0
0
0
3
n
ft
0
p
n
0
0
0
0
ft
n
n
n
c
1
25.0
10.0
l.P
1
10.0
1ft. n
1.9
0
0
n
0
R
12. S
BO.O
B1-100PP
R
I 5.
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
n
[ 0
[ 0
t o
t o
0
0
n
0
i
1.6
100. n
1 .0
101-120P
P?
[ 6.
ft
ft
ft
ft
n
ft
ft
ft
0
ft
ft
0
0
ft
ft
ft
1
20. ft I
33.3 I
1." T
ft I
0 I
0
ft
0
ft
0
0
6
n
o
0
9
3.1
frft.7
....'.
ROtt
TOTAL
1
1.0
l.c
TuTM
3*.
3«
30.?
1?.*
1"
t".3
I
1.0
3.1
5
5.2
4
4.1
10
1ft.3
6
6.2
64
97
IftO.n
11
-------
TABLE 4 CROSS-TABULATION OF DISTANCE AND THE ABSOLUTE DIFFERENCE BETWEEN
SECOND-MAXIMUM OZONE CONCENTRATIONS FOR SOUTHERN CALIFORNIA
ABSOLUTE DIFFERENCE BETWEEN SECOND-MAXIMUM OZONE CONCENTRATIONS ppb
I/)
o
ill
Ol
lil
CD
Z
O
Q_
UJ
(O
Sl-lr
tOi PfT
lol pry
1 .
?.
-l.
4
...
w
COLI1""
TCT«I.
1 .
17
53.1
7...H
I".!-
*
0
*
?
7.1
K.3
2.Z
3
20. 0
17. S-
3.3
IP.?
*.3
?.5
?6. 1
?.
f
1 H , P
7S»<»
A.*
1
1
(»
r
1
1 *
1 ? ^
1 1
1
* 7
1? *
1 1
A
r
'»
A
H.T
1.
3
0.4
30,u
3.3
?
33.3
20 0
f.z
i
1.6
in.o
l.l
l
»>.7
in.o
1.1
3
?7.3
in."
1.3
10
11.9
4.
h
1B.H
1S.1
h.S
4
6t>.7
? J .^
4.3
3
1C. 7
17. A
J.3
1
(-.7
S .9
I.I
3
?7.3
17. A
3.1
17
1P.S
f I - 1 no^f1
h
S.
n
0
.1
u
fl
c
(
n
s
17.9
loc.n
'.4
r
r.
c
r
r
o
r
it
S
*<*
101-1 ?!»**
» ^
*
n
n
n
n
n
n
n
"
1
"***
jfs.T
1 .'
4
?6.T
ft^.7
*.T
t
9.1
IA.7
1.1
A
6.*
1 1? 1 ) 4 QP
PH
7.
r
t,
f
0
i.
r
r
c
i
3."
?b. *
1.1
?
11.1
Sn. c
?.?
1
9.1
?*!"
i.i
»
4. .1
\ 4 ) - | h**P
PH
p .
I)
n
n
n
0
0
n
0
c,
17.9
fS.6
%.4
3
?n.n
33.3
3.3
1
9.1
11.1
1.1
9
9.K
AHOVf |b
OPPH
10.
U
U
0
(i
n
0
!
U
v
3?.l
100.0
0
0
0
'
,.
0
n
0
V
9,b
lnl«L
32
0.5
10.4
IS
16.3
II
1C.O
100.0
It is evident from the figures and the tables that very large differences in the observed
hourly second-maximum concentrations have occurred even at stations that are separated by
only a few kilometers. Surprisingly, differences in readings appear to be greatest for locations
that are separated by distances of a few hundred kilometers and less for greater separations.
This may be an artifact, however, reflecting that more pairs of stations in each data set are
separated by distances of a few hundred kilometers than are separated by somewhat greater or
lesser distances. Tables 1 (Northeast states) and 4 (Southern California) show a tendency for
greater representation of the intermediate distances.
The contingency tables (or cross-tabulations) have four numbers within each square; the
absolute number of cases that meet the two criteria that define the square is given by the first
figure. The second figure within each square defines the percentage of the total number of
cases in that row that falls within that particular column; the third figure shows the percentage
of the total number of cases in that column that falls in that particular row. The last figure
shows the percentage of the total data set that falls within that square. For instance, the
upper-left square of Table 1 shows:
There were 55 instances when the absolute difference between second-maximum
readings was 0-20 ppb and the station separation was less than 50 km.
47.4 percent of all stations separated by less than 50 km had second-maximum read-
ings that differed by less than 21 ppb.
12
-------
15.1 percent of the instances when the second-maximum values differed by less than
21 ppb were associated with stations separated by less than 50 km.
5.0 percent of all cases had second-maximum values that differed by less than 21 ppb
and were separated by less than 50 km.
Row and column totals of absolute and percentage frequencies are shown at the ends of the
rows and the bottoms of the columns.
The tables indicate that most differences in second-maximum observations are less than
40 ppb, except in the Southern California data set, where more than half the differences exceed
60 ppb. The larger differences associated with the Southern California data reflect the effect of
the Southern California summer weather, which changes little from day to day and which
results in strong, persistent pollutant gradients in the Los Angeles basin.
Figures 5 and 6 show how the average differences between second-maximum ozone read-
ings vary with the separation between stations. These figures were compiled from the same
data used to prepare Figures 1 and 2. The data were stratified according to separation
between stations using 50-km bins. The absolute values of differences in the second-
maximum readings were transformed by adding 0.5 ppb and then determining the geometric
mean and the standard geometric deviation for each data subset. The geometric means
(reduced by 0.5 ppb) and the range indicated by the standard geometric deviations have been
plotted in Figures 5 and 6: The transformations avoid the problems associated with determining
geometric means for data sets that include zero values. The data were too few to prepare
diagrams of this type for the Southwest or Southern California regions.
5
D
5
X
If
8 1
£ 2 80
UJ <
UJ Z
CD UJ
ag
58
5^ 40
tS
Q 0
UJ
D
0
in
DD
n
-
-
i
<
i
'
i
(
I
I
I
1
<
;
i
i
i
-
.
25 75 125 175 225 275 325
DIFFERENCE BETWEEN STATIONS km
375
425
>425
Figure 5. Geometric means of difference in second-maximum ozone concentration for the Northeast.
13
-------
Z a
O i
u
ui '
Z
z2
UJ H
ui 425
Figure 6. Geometric means of difference in second-maximum ozone concentration for the Midwest.
There is an obvious relationship between station separations and differences in the
observed second-maximum ozone concentrations. In the Northeast, there is a slight tendency
for the absolute values of the differences to increase with separation out to about 250 km, and
then to actually decrease with greater separations. However, even if this apparent tendency is
not an artifact, it is too weak to be useful. The data from the Midwest in Figure 6 show virtually
no change with station separation in the average difference between observed second-
maximum ozone readings. Figures 1 through 6 suggest that other types of relationships must
be sought in order to determine the degree to which existing data can be used to characterize
those areas where the NAAQS is met or exceeded.
3. Conditional Probabilities
of Observing Exceedances of the NAAQS
As noted in the preceding section, the average difference between second-maximum con-
centrations observed at two stations is not closely enough related to station separation to be
used to determine areas of noncompliance with the standards. For this reason, another
approach is needed. The approach described in this section is based on the following ques-
tion: "How does the probability of observing an exceedance of the NAAQS vary with distance
from a station where a given concentration has been measured?" If that question can be
answered for a variety of observed second-maximum ozone concentrations, then it should be
possible to define an area around each existing station where the probability of observing an
exceedance of the NAAQS is either very high or very low (and that area is reasonably certain to
be in exceedance or in compliance).
14
-------
The question of whether or not the standards are exceeded within an area can be asked
for a given day, but that is not the question of greatest interest for this particular study. In
essence, the form of the question to be addressed here is whether or not a particular location
can be reliably designated in exceedance of the standards on an annual basis. The annual
second-maximum hourly concentrations were selected as the appropriate design values. The
reader is directed to Section 3-A for a more detailed discussion on this selection.
Figure 7 shows the conditional probabilities that the NAAQS was exceeded more than
once during a year at different distances from a location where the second-maximum ozone
concentration exceeded different thresholds. There is considerable scatter evidenced in the
data, because many of the subsets representing certain increments of station separation had
sparse data, but the major features of the expected behavior are apparent. For example, the
curves do converge at the smaller separation distances, as is to be expectedany threshold
above 120 ppb is certain to be associated with an exceedance when the location for which the
probabilities are being estimated is coincidental with the observing station, i.e. zero separation
distance. At the other extreme, when the location for which the probabilities are being
estimated is very distant from the observation site, one might expect the two to be unrelated.
Thus, the probability of finding a second-maximum ozone concentration above the NAAQS is
independent of the observed value and equal to the probability of finding such an exceedance
at a location selected randomly from the total data set. In this instance, 72.5 percent of the sta-
tions had second-maximum ozone concentrations in excess of 120 ppb. Thus, it is not surpris-
ing that at distances greater than about 250 km, the probability of observing an exceedance of
the NAAQS is about 72 percent regardless of the threshold.
100
80
60
H
_l
m
o
cc
a.
40
20
...Q"- 240 ppb
D 200 ppb
A 160 ppb
50
100
150 200 250 300
STATION SEPARATION km
350
400
450
>450
Figure 7. Probability as a function of distance between stations that ozone will exceed 120 ppb
if the annual second-max!mum ozone concentration has been observed to exceed
specified thresholds for all U.S. monitoring sites for 1977.
15
-------
4. Identification of Areas Where Exceedances
of the NAAQS for Ozone are Probable
The probability that a station will have exceeded the NAAQS for ozone, given that it is
within a certain distance from a station where some threshold ozone concentration has been
exceeded, has already been discussed. These conditional probabilities can be expressed as a
function of the distance between the observing stations and the threshold that was exceeded.
If the probability is fixed, then it should be possible to express-as a function of observed
annual second-maximum, hour-averaged ozone concentrationthe radius of the area for prob-
able NAAQS exceedances. Each station in the United States that has observed an exceedance
of the NAAQS for ozone can then be used to define an area within which such exceedances
were probable. Those areas can be considered to have an established status, with regard to
the NAAQS for ozone. In principal, a similar approach could be used to establish areas of prob-
able compliance with the standards, however, the present data are not sufficient to establish
such a relationship. The status of other areas within the country might then be considered
undefined, so these areas would be candidates for the establishment of some sort of monitor-
ing.
Contingency tables like Table 5 show the relative frequency at which the standards were
exceeded at the second station of a pair, given that some concentration had been observed at
the first station. For example, Table 5 shows that for station pairs separated by 50 km or less,
there were 20 instances when the second-maximum ozone concentration was between 40 and
60 ppb at one of the stations. Of those 20 instances, a second-maximum ozone concentration
in excess of 120 ppb was observed on only three occasions at the other station of the pair.
Thus, there was only about a 15-percent likelihood that two or more exceedances would be
found within 50 km of a station whose second-maximum ozone concentration for the year was
between 40 and 60 ppb. As another example, Table 5 shows 28 instances with second-
maximum concentrations between 260 and 280 ppb at one of the stations in the pair. Of those
28, the observed second-maximum at the other station exceeded 120 ppb in 26 cases. Thus,
the annual probability of exceeding the ozone standard twice or more within 50 km of a station
where the second-maximum was observed to be between 260 and 280 ppb was about 93 per-
cent. Similar probabilities can be determined for each of the concentration intervals shown in
Table 5.
Contingency tables were prepared showing the number of cases of exceedances and
nonexceedances versus observed second-maximum ozone values in 40 ppb bins. Tables were
prepared for station separations of 0 to 50 km, 50 to 100 km, and so forth. The Chi-square test
showed significant relationships at better than the 1-percent level for stations separated by as
much as 250 km. The tables for stations separated by 250 to 350 km were also significant at
the 1 percent level, but most of the relationships were opposite to the expected, i.e., there were
fewer than the expected number of exceedances associated with the higher second-maximum
observations. This is further evidence of the anomolous relationships discussed earlier in con-
nection with Figures 5 and 6. At separations greater than 350 km, the relationships are not
significant at the 1 percent level.
Figure 8 shows the empirical probabilities associated with the annual second-maximum
ozone concentrations at stations separated by less than 50 km. A smooth line has been drawn
approximating the trends of the points. The curve in Figure 8 has the expected characteristics:
At very low observed annual second-maximum ozone concentrations, the probability of observ-
ing an exceedance of NAAQS within 50 km of a station is quite small, and the probabilities
increase with increasing observed annual second-maximum ozone concentrations. For loca-
tions within 50 km of an observing station, it becomes virtually certain that an exceedance of a
16
-------
TABLE 5 JOINT FREQUENCY OF OCCURRENCE OF ANNUAL SECOND-MAXIMUM OZONE
CONCENTRATIONS AT STATIONS SEPARATED BY LESS THAN 50 km
Second-Maximum
Ozone Concentration
(ppb)
0-20
20-40
40-60
60-80
80-100
100-120
120-140
140-160
160-180
180-200
200-220
220-240
240-260
260-280
280-300
300-320
320-340
340-360
360-380
380-400
Column Total
Second-Maximum Ozone Concentration (ppb)
0-20
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
20-40
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
40-60
0
0
2
3
4
8
3
0
0
0
0
0
0
0
0
0
0
0
0
0
20
60-80
0
0
3
0
5
9
2
1
0
0
0
0
0
0
0
0
0
0
0
0
20
80-
100
0
0
4
5
6
13
8
6
2
1
0
0
1
0
0
0
1
0
0
0
47
100-
120
0
0
8
9
13
22
44
15
11
3
0
1
0
2
0
0
0
0
0
0
128
120-
140
0
0
3
2
8
44
58
45
40
22
4
4
4
3
1
0
2
0
0
0
240
140-
160
0
0
0
1
6
15
45
66
52
38
8
6
5
4
2
2
3
1
0
0
254
160-
180
0
0
0
0
2
11
40
52
64
28
13
4
4
3
0
0
0
0
0
0
221
180-
200
0
0
0
0
1
3
22
38
28
28
14
10
1
3
1
2
3
2
0
0
156
200-
220
0
0
0
0
0
0
4
8
13
14
20
2
1
0
0
0
0
0
0
0
62
220
240
0
0
0
0
0
1
4
6
4
10
2
2
2
0
1
1
3
1
0
0
37
240-
260
0
0
0
0
1
0
4
5
4
1
1
2
0
0
2
0
2
0
0
0
22
260-
280
0
0
0
0
0
2
3
4
3
3
0
0
0
2
1
4
2
1
1
2
28
280-
300
0
0
0
0
0
0
1
2
0
1
0
1
2
1
0
2
6
1
0
0
17
300-
320
0
0
0
0
0
0
0
2
0
2
0
1
0
4
2
2
8
3
2
4
30
320-
340
0
0
0
0
1
0
2
3
0
3
0
3
2
2
6
8
20
6
1
2
59
340-
360
0
0
0
0
0
0
0
1
0
2
0
1
0
1
1
3
6
0
1
2
18
360-
380
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
2
1
1
0
2
7
380-
400
0
0
0
0
0
0
0
. 0
0
0
0
0
0
2
0
4
2
2
2
2
14
Row
Total
0
0
20
20
47
128
240
254
221
156
62
37
22
28
17
30
59
18
7
14
1380
-------
standard will be found if the station observes a sufficiently large second-maximum ozone con-
centration. Figure 8 shows that the probabilities reach 90 percent at a little less than 150 ppb,
95 percent at about 175 ppb, and about 99 percent at 275 ppb.
v>
a
1.0
0.9
UJ
X 0.8
u 0.7
ill
u
< 0.6
O
LU
Ul
O 0.5
UJ
0.4
0.3
m
2 0.1
0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360
OBSERVED ANNUAL SECOND MAXIMUM OZONE ppb
Figure 8. Probability that the annual second-maximum ozone concentration exceeds 120 ppb
within 50 km of a station, given the observed value for that station.
One would expect that at sufficiently large distances from a station, the likelihood of
observing an exceedance of the NAAQS twice or more during the year (based on the observed
second-maximum ozone concentrations at that station) would not reach such high levels,
regardless of the observed ozone concentrations because the physical factors that cause the
relationships are generally limited in spatial scale. This expectation is generally confirmed by
the data. Figure 9 shows the conditional probabilities calculated for data from pairs of stations
separated by other distances. While there is considerable scatter, it is still possible to draw
reasonable curves to approximate the trends indicated by the points. At the greatest separa-
tions, the probabilities of observing an exceedance of the NAAQS (given an observed second-
maximum ozone concentration) do not generally reach 99 percent. Furthermore, as expected,
the observed concentration required for a specified probability level increases with increasing
separation. For example, when stations are separated by 50 to 100 km, an annual second-
maximum ozone concentration of 175 ppb or greater must be observed at one of them before
the probability of observing an exceedance of the NAAQS at the other exceeds 90 percent.
When the stations are separated by 100 to 150 km, the required concentration climbs to about
235 ppb before the probability that second-maximum ozone concentration will exceed 120 ppb
reaches 90 percent. Finally, for stations separated by 200 to 250 km, an observation of around
290 ppb is required.
18
-------
1.0
.9
.8
.7
.6
.5
.4
.3
.2
.1
0
I
I
(a) STATION SEPARATION - 50 - 100 km
(b) STATION SEPARATION - 100 - 150 km
1.0
.9
.8
.7
.6
.5
.4
.3
.2
.1
0
30 60 90 120 150 180 210 240 270 300 330 360 0
ANNUAL SECOND MAXIMUM CONCENTRATION - ppb
(c) STATION SEPARATION 150 - 200 km
30 60 90 120 150 180 210 240 270 300 330 360
ANNUAL SECOND MAXIMUM CONCENTRATION - ppb
(d) STATION SEPARATION = 200-250 km
Figure 9. Observed probabilities of an exceedance of the ozone standard for the different station separations
as a function of the second-maximum ozone value.
-------
There is an obvious relationship between the annual second-maximum ozone concentra-
tion observed at a location and the distance from that station at which the probability of
observing an exceedance of the NAAQS falls below a specified level. Figure 10 shows the
relationships for three different, specified probabilities as derived from the data presented in
Figures 8 and 9. Qualitatively, the relationship is as expected. For example, an observed con-
centration (annual second-maximum) of 120 ppb provides little guarantee that the NAAQS will
be equaled or exceeded twice or more anywhere but at that station. As the observed value
increases, so does the distance range for likely exceedance of the standard. As noted before, it
is not possible to achieve the higher probabilities beyond some cut-off distance, regardless of
how great the observed second-maximum ozone concentration might be. Thus, the distance
range for a 95 percent probability of exceedance is unlikely ever to be more than 175 to 200
km, regardless of the observed second-maximum ozone concentration. For 90 percent proba-
bilities, this cutoff is 225 to 250 km, and the 85 percent threshold can be achieved to distances
around 250 to 275 km.
350
120 140 160 180 200
220 240 260 280 300 320
OBSERVED SECOND MAXIMUM ppb
340 360 380 400
Figure 10. Probability of observing an exceedance of the NAAQS for ozone as a function
of observed second-maximum ozone concentration and distance.
20
-------
It is obvious that the curves in Figure 10 can be used in combination with observed values
of second-maximum ozone concentrations to define areas within which exceedances of the
NAAQS for ozone have probably occurred twice or more during a year. Figure 11 is a map of
the United States that shows those areas where it is 95 percent probable, based on the rela-
tionship shown by the straight line in Figure 10 and on the observed annual maximum ozone
concentrations, that the NAAQS for ozone has been exceeded once or more during years 1977
or 1978. The annual maximum ozone concentrations for each year and station for which more
than 7,000 hours of data were available were considered. Stations with more than 7,000 hours
per year with observed concentrations less than 120 ppb are marked with the symbol "L". The
largest radius used was 200 km, and the smallest, for graphical reasons, was 4 km. The
appropriate size circles were drawn centered on the location of the various observing sites,
resulting in Figure 11.
It is obvious from the figure that available data are sufficient to characterize much of the
country as exceeding the ozone standards. However, many large areas have not been con-
clusively identified as exceeding or being less than the standard on the basis of data in the
SAROAD.
It should be noted that the approach taken above can also be used to define areas that
are not likely to have exceeded the standards. Such classifications are more limited in area.
For example, Figure 8 indicates that an observed annual second-maximum ozone concentration
less than about 50 ppb is required before the probability of an exceedance within 50 km is
reduced to 10 percent or less (corresponding to a probability of compliance of 90 percent or
more). The kind of analysis used here does not appear to be very sensitive for determining
compliance with the standard; it does not appear possible to establish compliance with the
standards to distances greater than about 50 km from an observing site. There are some
regions of the United States shown in Figure 11 where the standards have not been exceeded
at any of the sites from which data are available and it is probable that the ozone standards
have not been exceeded anywhere within those general areas. However, this cannot be
quantified from presently available information.
Figure 11 shows some inconsistencies. For instance, a circle centered on Denver encom-
passes an "L" symbol denoting compliance. Such inconsistencies have been used to identify
some stations whose characteristics were studied for common traits related to unrepresenta-
tiveness. These have been discussed in a technical memorandum prepared as part of this pro-
ject (Ludwig and Shelar, 1979).
Figure 11 shows the areas that can be characterized with regard to probable exceedance
of the NAAQS for ozone on the basis of reasonably complete data from the SAROAD, i.e., data
sets of more than 7,000 hours per year. The characterized areas can be expanded when less
complete data sets are included; only one or more observations in excess of 120 ppb are
needed to be reasonably certain that an area is one within which exceedances are likely. In
general, the area defined by an incomplete data set will be smaller (it cannot be larger) than
would be obtained from a complete data set, but the inclusion of incomplete data sets can
reduce the area that is uncharacterized as shown in Figure 12.
Presumably, data from special studies or data not otherwise archived in the SAROAD
could be used to reduce the area uncharacterized with regard to compliance or exceedance of
the NAAQS. An intensive search for such data sources was beyond the scope of this project.
However, a search for existing data that could be used to characterize some of the regions left
blank in Figure 12 would probably be much more cost effective than special monitoring of some
kind. (This will be discussed further in Section 4.)
21
-------
ro
ro
Figure 11. Areas where it is 95-percent probable that the NAAQS for ozone has been exceeded
based on SAROAD stations with more than 7000 hours per year for the years 1977 and 1978.
-------
Figure 12. Areas where it is 95-percent probable that the NAAQS for ozone has been exceeded
based on all SAROAD data for 1977 and 1978.
-------
5. Factors Accompanying Widespread
Exceedances of the NAAQS
a. General
For practical reasons, it is important to be able to recognize those situations where
exceedances of the NAAQS are most likely to occur. The essential ingredients for the buildup
of high ozone concentrations are: an accumulation of precursor emissions, sunshine, and rela-
tively little ozone removal at the surface or by chemical reactions. The first two items depend
largely on meteorological conditions. The last item depends more on location.
Sunshine is not usually measured and reported but is closely related to temperature: It is
not surprising that ozone concentrations have been found to be closely related to temperatures.
An examination of the analyses of ozone concentrations prepared by Ludwig, Simmon et al.
(1977) for each day of 1974 shows that there were no instances in the eastern United States
when ozone concentrations exceeded 120 ppb on a day when the maximum temperatures did
not exceed 70°F (21 °C). Meyer et al. (1976) and Ludwig, Reiter et al. (1977) provided further
evidence of the importance of temperature in determining ozone concentration. Both groups
found significant correlations between observed ozone concentrations and ambient tempera-
tures measured in the air parcel during the preceding several hours. Ludwig and Martinez
(1979) recently found in an examination of 1977 summer data from Houston that ozone con-
centrations in excess of 120 ppb were not observed with temperatures less than 75°F (24°C).
TABLE 6 NUMBER OF DAYS PER MONTH IN 1977 WITH DAILY MAXIMUM OZONE
IN EXCESS OF 120 ppb IN SELECTED REGIONS OF THE U.S.
Month
January
February
March
April
May
June
July
August
September
October
November
December
Region
Northeast
0
0
1
5
14
15
17
17
3
0
0
0
Midwest
1
4
3
10
22
19
26
13
8
0
0
0
Texas-Louisiana
Gulf Coast
2
2
3
8
11
9
15
13
8
13
0
0
Southern
California
0
10
4
16
8
27
30
29
25
22
12
10
Data source: SAROAD
24
-------
b. Seasonal Effects in Different
Geographical Areas
With the dependence on temperature described above, it is not surprising that there are
strong seasonal variations in the incidence of exceedances of the ozone standard. The
SAROAD data for 1977 from the selected regions were examined to see how the frequencies of
occurrence of ozone standard exceedances changed from month to month and from region to
region during 1977. The results are shown in Table 6. A similar tabulation was prepared from
1974 data for the eastern U.S. by Ludwig, Reiter et al. (1977) using the old NAAQS of 80 ppb.
That tabulation also showed summer and fall to be the seasons when high ozone concentra-
tions are most probable. There are differences in probability from one part of the country to
another and there are also some differences in the seasonal distribution of ozone exceedances
from place to place.
6. Effects of Large-Scale Weather Systems
in Different Geographical Areas
Several authors have pointed out that relatively high ozone concentrations are frequently
found in association with high-pressure systems (e.g. Bach, King and Vukovich, 1976; Ludwig,
Reiter et al., 1977; Wolff, Ferman, and Monson, 1979). It is usually stated that the preferred
part of the high-pressure system is the "backside" or western part. Ludwig, Reiter et al. (1977)
examined the frequency that high ozone concentrations (in excess 80 ppb) were observed in
association with different meteorological features in the eastern United States during 1974.
They found that the higher ozone concentrations occurred most frequently in the western parts
of high-pressure systems. If cases that they classified as being in the warm air near a front
an area usually located at the western extreme of a high-pressure systemare added to those
cases that they classified as being in the western parts of anticyclones, then, the high ozone
concentrations were found in the western half of high-pressure systems in sightly more than
half of the cases. About 25 percent of the time, the high ozone concentrations were associated
with the central or eastern parts of a high-pressure cell.
Ludwig, Reiter et al. (1977) presented another table which hinted that the meteorological
features associated with high ozone concentrations may differ from one part of the country to
another. That table related the winds reported on the morning weather map to areas where
relatively high peak-hour ozone concentrations (in excess of 80 ppb) occurred during the day
for the months June through August. The differences In the wind direction that they found from
one geographical region to another suggest different locations relative to the high-pressure
system. Since winds are calm in the center of the anticyclone and spiral clockwise outward,
the different winds can be interpreted as being associated with different parts of the anticy-
clone. Based on the tabulations of wind direction associated with relatively high ozone concen-
trations given by Ludwig, Reiter eta I. (1977), the Texas-Louisiana Gulf Coast is most frequently
in the center or southwestern part of a high-pressure system when the relatively high ozone
concentrations occur, while in New England the more northerly parts of a high-pressure sys-
tems are more often involved. The analyses of Ludwig, Simmon et al. (1977) were used to
prepare Table 7, which shows a breakdown of the occurrence of exceedances of the 120 ppb
NAAQS in the eastern U.S. during 1974 according to geographical area and meteorological
feature.
25
-------
TABLE 7 OZONE CONCENTRATIONS IN EXCESS OF 120 ppb AND ASSOCIATED WEATHER
FEATURES IN VARIOUS PARTS OF THE EASTERN UNITED STATES DURING 1974
Synoptic-Scale
Weather Feature
Near a front
Cold air
Warm air
High-pressure
system
Northwest
Southwest
Southeast
Northeast
Center
Other
Region
Texas/
Louisiana
Oklahoma
1
9
1
18
1
0
1
8
*f
Northeast
6
11
8
4
2
5
2
5
Southeast
1
6
1
6
2
2
2
2
Midwest
1
13
9
6
2
1
2
3
Table 8 was prepared from analysis of 1977 ozone data from the four geographical
regions that served as a focus of this study. Threshold ozone concentrations were selected for
each area such that observed ozone concentrations, averaged over the area, exceeded the
threshold on ten to fifteen days during the year. In this way, cases of widespread, high ozone
concentrations were identified for each of the geographical areas. A computer analysis of the
ozone pattern was prepared for each selected case and compared with the corresponding
morning weather map (Nat. Oceanic and Atmos. Admin., 1977).
Table 9 summarizes the nature of the meteorological conditions associated with those
widespread, high ozone-concentration cases. It is apparent from Tables 7 and 9 that incidents
of high ozone concentration are associated with somewhat different meteorological features in
different parts of the country. Both Tables 7 and 9 show that high ozone concentrations in the
Texas-Louisiana Gulf Coast area are most frequently associated with the southwestern qua-
drant of a high-pressure system. Figure 13 gives an example4 June 1977of such a system.
The pressure gradients cause light airflow onshore, or nearly parallel to the shore. This, in turn,
allows an accumulation of precursors, with subsequent ozone formation.
Ludwig and Shelar (1977) pointed out that the accumulations of precursors and ozone
occur in the northeastern part of the United States when the airflow parallels the alignment of
the urban areas in that region. In contrast to the Texas Gulf area, such airflow in the
northeastern U.S. tends to be associated with the northwestern quadrant of high-pressure sys-
tems. Tables 7 and 9 reflect this fact. An example of such a situation is shown in Figure 11 for
4 July 1977. The example shows that the high concentration area was in the northern part of
the high. This is not unusual; in fact, as Table 7 shows, the high concentrations sometimes fall
in the northeast quadrant of a high-pressure cell.
26
-------
TABLE 8 SELECTED DAYS WITH WIDESPREAD HIGH OZONE CONCENTRATIONS
IN DIFFERENT REGIONS DURING 1977
Northeast
Date
17 May
4 Jul
15Jul
16 Jul
17 Jul
19 Jul
20 Jul
21 Jul
4 Aug
5 Aug
Regional
Average Ozone
Concentration
(ppb)
119
111
112
152
150
108
123
132
111
105
Midwest
Date
16 May
24 May
25 May
27 May
28 May
4 Jun
15Jun
14 Jul
15 Jul
28 Jul
30 Jul
Regional
Average Ozone
Concentration
(ppb)
91
94
95
95
110
94
95
94
98
93
91
Texas/Louisiana/
Arkansas
Date
26 Apr
11 May
12 May
31 May
1 Jun
2 Jun
4 Jun
5 Jun
6 Jun
9 Jun
16 Jul
23 Jul
26 Jul
27 Jul
Regional
Average Ozone
Concentration
(ppb)
93
91
99
111
98
144
103
111
99
108
92
98
93
94
Southern California
Date
4 Jun
25 Jun
26 Jun
11 Aug
4Sep
5 Sep
6 Sep
7 Sep
11 Oct
140ct
15 Oct
17 Oct
18 Oct
Regional
Average Ozone
Concentration
(ppb)
157
146
157
148
146
155
162
178
154
166
164
154
154
Data source: SAROAD
-------
MAP __l_\ \ \ \_^
AND STATION WEATHER L-J \ ._x >^_
AT 7:OO A.M.. E 5 T
(a) SURFACE WEATHER
35
33
| 31
<
-1 29
27
25
12080
107 105 103 101 99 97 95 93 91
LONGITUDE
(b) PEAK-HOUR OZONE CONCENTRATIONS
Figure 13. Example of high ozone concentrations associated with the southwest quadrant
of a high-pressure system over the Texas-Louisiana Gulf Coast area, 4 June 1977.
28
-------
TABLE 9 WEATHER FEATURES ASSOCIATED WITH WIDESPREAD
HIGH OZONE CONCENTRATIONS IN SELECTED REGIONS
OF THE UNITED STATES DURING 1977
Synoptic-Scale
Weather Feature
Warm air
near a front
High-pressure
system
Northwest
Southwest
Southeast
Northwest
Center
Other
Region
Texas/
Louisiana
4
0
7
1
0
0
2
Northeast
3
7
0
0
0
0
0
Midwest
4
3
2
0
0
0
1
Southern
California
0
0
0
0
0
0
13 (see text)
Ludwig and Shelar (1977) also noted that the orientation of fronts in the northeastern
United States is frequently such that the airflow ahead of them parallels the alignment of urban
areas and allows the accumulation of precursors. The importance of this mechanism is indi-
cated by Tables 7 and 9, which show an appreciable number of cases of high ozone concentra-
tions in the warm air near fronts. Figure 15 shows an example that occurred on 16 July 1977.
Tables 7 and 9 suggest that this same meteorological feature is also frequently associated with
high ozone concentrations in Texas and Louisiana. Figure 16, for 31 May 1977, provides an
example of one such instance for that area. Unlike the Northeast, where there is usually well-
organized air flow over a succession of urban emission areas in such synoptic situations, warm
air ahead of a front in Texas is frequently rather stagnant, allowing an accumulation of precur-
sors and ozone to occur. Certainly, as can be seen from the example shown in Figure 16, the
air behind the front tends to be cleaner. (This is also true in the Northeast.)
Tables 7 and 9 show that the northwest quadrant of a high-pressure system is the favored
location for episodes of high ozone concentrations in the Midwest, but the southwest quadrant
is also a frequent site of such events. Other parts of the high-pressure system are less fre-
quently associated with high ozone concentrations. Figures 17 and 18 provide examples of
instances when the areas of high ozone concentrations in the Midwest were associated with
the northwest and southwest quadrants of high-pressure areas, respectively. Tables 7 and 9
also suggest that the warm air near fronts is a not-infrequent locale for high ozone concentra-
tions in the Midwest. Figure 1925 May 1977shows an example where the warm air just
ahead of an advancing cold front contains rather high concentrations of ozone.
29
-------
(a) SURFACE WEATHER
48
46
1U
O
t 44
<
42
40
79 77 75 73
LONGITUDE
71
(b) PEAK-HOUR OZONE CONCENTRATIONS
Figure 14. Example of high ozone concentrations associated with the northern part
of a high-pressure system in the northeastern U.S., 4 July 1977.
30
-------
(a) SURFACE WEATHER
48
46
ui
o
?44
<
42
40
79 77 75 73 71
LONGITUDE
(b) PEAK-HOUR OZONE CONCENTRATIONS
Figure 15. Example of high ozone concentrations in warm air near a weather front
in the northeastern U.S., 16 July 1977.
31
-------
ACE LEATHER MAP
STATION WEATHER
T 7 00 *.M E S T
(a) SURFACE WEATHER
35
33
§ 31
29
27
25
80
160= '
A.'.:; I--J >V..=..
j f A. A-f.-? * , ...-<... A f.f
107 105 103 101 99 97 95 93 91
LONGITUDE
(b) PEAK-HOUR OZONE CONCENTRATIONS
Figure 16. Example of high ozone concentrations in warm air near a weather front in Texas,
31 May 1977.
32
-------
(a) SURFACE WEATHER
45
43
S
D
37
35
95 93 91 89 87 85 83 81
LONGITUDE
(b) PEAK-HOUR OZONE CONCENTRATIONS
Figure 17. Example of high ozone concentrations in the northwest quadrant
of a high pressure system in the Midwest, 28 July 1977.
33
-------
(a) SURFACE WEATHER
45
43
41
39
37
35
95 93 91 89 87 85 83 81
LONGITUDE
(b) PEAK-HOUR OZONE CONCENTRATIONS
Figure 18. Example of high ozone concentrations in the southwest quadrant
of a high pressure system in the Midwest, 27 May 1977.
34
-------
SURFACE WEATHER HAP ) \ I \_
ID STATION WEATHER - \ \ \ 3*
»T 7:00 A.M.. E.S.T. J \ \
(a) SURFACE WEATHER
35
95 93 91 89 87 85 83 81
LONGITUDE
(b) PEAK-HOUR OZONE CONCENTRATIONS
Figure 19. Example of high ozone concentrations in warm air near a weather front
in the Midwest, 25 May 1977.
35
-------
The 1974 analyses of Ludwig, Simmon et al. (1977) were used to provide some estimate
of where the highest ozone concentrations were found in the southeastern United States,
because this area has been mentioned as being potentially in need of supplemental monitoring.
However, the need for supplemental monitoring of itself makes the analyses less reliable.
Nevertheless, interpolation between the few data points available for the region has given some
indication of the weather features that are associated with high ozone concentrations there.
Table 7 shows that concentrations in excess of 120 ppb were less frequently observed in the
Southeast than in the Texas-Louisiana area, but the relative distribution with regard to location
in high-pressure systems is similar. The southwest quadrant of the high-pressure system is the
most frequent location of the high ozone concentrations. Next, as in the Texas-Louisiana
region, comes the warm air near a weather front. This information should provide some gui-
dance for planning field programs.
The classification of weather systems shown in Tables 7 and 9 was derived from the work
of Ludwig, Reiter et al. (1977) which focused only on the conditions in the eastern United
States. This classification system does not fit the conditions in Southern California very well,
and all the cases selected for study in that region are in the "other" category. Figure 20 is an
example of the pressure pattern that prevailed in all the selected Southern California cases.
Basically, the semipermanent Pacific high-pressure system is located well to the west of the
California coast. In that sense, Southern California was in one of the eastern quadrants. How-
ever, this large-scale pattern was consistently modified by the presence of a thermal low-
pressure trough over the interior deserts and the Central Valley of California. A trough was
found over the interior in all cases, but in one instance (1 7 October 1977) the Daily Weather
Map analysis showed a cold front through the southeast part of California and Nevada. Other-
wise, the pressure patterns all look very much like that shown in Figure 20, except that at the
hour of the analysis (0400 PST), stratus clouds or fog were frequently present along the coast;
the coastal clear skies in Figure 20 at this hour of the morning are rather unusual.
The low pressure over the interior, combined with the high pressure offshore, produces
onshore flow in Southern California, transporting the precursors and the resulting ozone east-
ward and southeastward, following the lower-altitude paths through the topography. The areas
of high ozone concentration can be seen in Figure 20 to stretch out along paths that parallel
the airflow.
B. Factors Causing Observed Ozone Concentrations
To Be Unrepresentative
1. Local Siting Effects
The most serious errors in the measurement of ozone concentrations that are caused by
improper siting generally arise from the destruction of ozone by nitric oxide. Ludwig and Shelar
(1978) have used a very simple model to describe the destruction of ozone in the vicinity of
roadways: The change in ozone concentration caused by traffic interference is estimated to be
about numerically equal to the locally added NOX when the added NOX is less than about 50
ppb. Figure 21 provides estimates of the maximum interference likely at different distances
from roadways of differing average daily traffic (ADT). Of course, this maximum, which is based
on light (1 m/s) winds paralleling the roadway at rush hour in a relatively stable atmosphere
(Paquill-Gifford Class E), will not commonly occur. Furthermore, the effects are limited by the
ambient ozone and NO concentrations. Obviously, the amount of ozone removed by NO cannot
be greater than the amount of ozone already present. Therefore, the effects tend to be asymp-
totically limited by the total amount of ozone present. Also the effects tend to be less for the
36
-------
(a) SURFACE WEATHER
34
UJ
Q
33
32
118 117 116 115
LONGITUDE
(b) PEAK-HOUR OZONE CONCENTRATIONS
114
Figure 20. Example of a typical pressure pattern associated with high ozone concentrations
in Southern California, 7 September 1977.
37
-------
higher ambient NO concentrations. Figure 21 is based on Ludwig and Shelar's (1977)
diagrams, using the assumptions shown in the figure. The K,/K2 ratio refers to the ratio of the
reaction rate for the photolytic decomposition of NO2 to the reaction rate of NO with O3 to form
NO2.
According to Ludwig and Shelar (1978) the worst effects can be Kept below about 40 to
50 ppb by staying more than about 20 m from small streets (about 1000 ADT) and about 250 m
from larger streets (10,000 ADT). Under more nearly ordinary circumstances, the effects can
be kept to the same magnitude even when the probes are much nearer to the street, as
reflected in recent regulations promulgated by EPA (Federal Register, 1979b). Those regula-
tions specify that the minimum distance to roadways from ozone monitors should be from 10 m
(for 10,000 ADT) to 250 m (100,000 ADT).
1000
a.
a.
O
z
UJ
o
o
o
n
O
Z
o
o
100
10
ASSUMPTIONS: PEAK-HOUR TRAFFIC EQUAL TO
10% OF ADT
SLIGHTLY STABLE ATMOSPHERE
INITIAL VERTICAL DISPERSION
EQUAL TO 1.5 m
NOX EMISSION RATE EQUAL TO 4 g/mile
AMBIENT NO = 30 ppb
K,/K2 = 2
10 20 40 100 200 400 1000
DISTANCE FROM HIGHWAY m
2000
4000
10.000
Figure 21. Estimated ozone depletion at distances from a roadway.
2. Urban Area Siting Effects
The effects of a city's NO emissions are not limited to points that are within the immediate
vicinity of NO sources, although the effects are greatest at such locations. In general the
highest concentrations of ozone in an area will be away from areas of greatest NO emissions.
Figure 22 shows how the NO emissions in St. Louis on the morning of 19 July 1976 reduced
ozone concentrations by about 40 to 50 ppb. The map in the figure shows the Mississippi and
Missouri River; the urbanized area of St. Louis is outlined by the dashed line that runs through
the center of the closed 40-ppb isopleth near the center of the map. Vertical ozone profiles
measured during the same hour around the area are also shown along with nephelometer read-
ings. Arrows indicate surface wind direction; their length equals one hour's travel distance at
the observed speed.
38
-------
S 600
S
i «--
1 2:>;
c
1 1 1 j - :',-.-
TIME 0839
JL
- f "Ss>~- -
> \"~
TIME 0849 1
TIME 0830
TIME 0820
50 100 IM 2C: 250 JOO
OZONEIloicl°°B BSCiTlt*fq« o»»hlI/IOKM
SOURCE: LUDWIG, 1979
Figure 22. St. Louis ozone concentrations on the morning of 19 July 1976.
The vertical profiles taken outside the city (along the right side of Figure 22) show nearly
uniform concentrations of about 80 ppb for the first several hundred meters. The profile taken
over the center of the city shows the same concentrations aloft, but near-surface concentra-
tions were reduced by NO emissions to about 40 ppb. The vertical profile at the top of the
figure shows similar conditions at the northern edge of town, but the reduction of ozone by NO
was somewhat lessonly 20 to 30 ppb.
Figure 23 from Ludwig and Martinez (1979) shows average August 1977 afternoon ozone
concentrations classified according to wind direction observed at a site north of Houston. (The
Houston urban area is in the middle of the map.) For most of the wind directions, the average
urban concentrations were 10 to 20 ppb less than concentrations outside the urban area. It
appears that the presence of a relatively large city can reduce ozone concentrations over fairly
large areas by a few tens of ppb or more. In general, the highest ozone concentrations in an
area will be outside the city limits (Ludwig and Shelar, 1977). Specific siting examples,
identified during the analysis of SAROAD data undertaken as part of this project, tend to
confirm this conclusion. Those cases are discussed in a technical memo (Ludwig and Shelar,
1979).
39
-------
(a) NORTHEAST
(b) EAST
(c) SOUTHEAST
SOURCE: LUDWIG AND MARTINEZ, 1979
(d) SOUTH
Figure 23. Average August 1977 afternoon ozone concentrations (ppb) in Houston.
3. Instrumental and Operational Factors
Miller (1978) has pointed out at least five possible sources of error in an overall monitor-
ing system, including the error caused by unrepresentativeness of the site, errors arising when
samples are randomly collected, and errors arising in data handling. Neither of the latter two
errors will be addressed here, but errors due to measurement method and calibration must be
discussed to some extent, if the importance of unrepresentativeness is to be kept in
perspective.
40
-------
The calibration procedure for ozone monitoring has been described in the Federal Register
(1979a). Basically, a stable ozone concentration must be generated, and that concentration
determined by ultraviolet (UV) photometry. According to the regulations the photometer must
have a precision within 5 ppb or 3 percent of the concentration, whichever is greater. Thus, the
uncertainty arising from currently prescribed calibration procedures must lie between about 5
and 10 ppb, depending on concentration. Similar errors might be expected to arise from the
use of a UV instrument for monitoring; in fact, the monitoring errors might be somewhat larger,
because of the special care and maintenance prescribed by the regulations for the instrument
used in establishing the standard concentrations. Assuming that the errors add as the square
root of the sum of their squares, the total uncertainty in this type of measurement is 7 to 15
ppb.
Judging by manufacturers' specifications (Lawrence Berkeley Laboratory, 1973) chemi-
luminescent ozone monitors have a noise level of 1 or 2 ppb, reproducibility of 5 or 10 ppb, and
similar uncertainties in linearity. Thus, an overall uncertainty of 10 ppb for carefully operated
chemiluminescent instruments does not seem to be an unreasonable estimate. In combination
with the recommended calibration procedures, a well-operated chemiluminescent instrument
should be able to measure ozone concentrations within 10 to 20 ppb.
Before 1979, calibration procedures other than the UV assay technique were used. Beard
(1978) found that one of these procedures, gas phase titration of excess ozone with nitric
oxide, tended to indicate higher ozone concentrations, by 2 to 7 percent, than were obtained by
UV photometry. Hogeson (1976) found "excellent agreement between O3 determination of UV
or GPTE" (gas phase titration). Methods that used iodometric procedures to determine ozone
concentrations for calibration purposes, however, did not generally agree well with the UV pro-
cedures according to Hogeson (1976), and the different variations of the iodometric procedure
did not agree well with each other.
Hogeson (1978) compared three different iodometric techniques and gas phase titration
with ozone concentrations determined by UV calibration. The three iodometric techniques
were: the EPA technique using 1 percent neutral-buffered potassium iodide reagent, the Cali-
fornia Air Resources Board method using 2 percent neutral-buffered potassium iodide, and the
Los Angeles Air Pollution Control District method using 2 percent unbuffered potassium iodide.
The data were compared with the UV measurements and linear expressions derived by
regression as follows:
Gas phase titration (GPT):
[O3]GPT = 1.09lO3]uv - 3ppb
California Air Resources Board (CARB):
(OSCARS = 1.29[O3]UV - Sppb
. EPA:
1.24[O3]UV - 35ppb
41
-------
Los Angeles Air Pollution Control District (LAAPCD):
- 32ppb
The above results were obtained with air humidified to 50 percent relative humidity. Hogeson
(1976) concluded that the neutral buffered potassium iodide procedures (i.e., CARB and EPA)
indicated ozone concentrations that were 18 to 30 percent higher than those observed by UV
procedures in the presence of moist air. Thus, there appears to have been a systematic bias
toward higher readings with the older calibration procedures. This can be corrected for, but a
residual uncertainty might amount to 5 to 30 ppb, depending on concentration. It appears that
older data, obtained using iodometric calibration procedures, might be subject to uncertainties
of a few tens of ppb.
C. Relationships of Ground-Level
Ozone Concentrations to Those Aloft
1. Background.
A monitoring network may not be adequate to determine whether or not the NAAQS are
exceeded everywhere in the country: It may be necessary to supplement existing data. The
aircraft-survey, approach requires a knowledge of the conditions under which aircraft measure-
ments can be used to estimate ground-level observations. This section addresses that prob-
lem; recommendations are given in Section 4 for incorporating the results described here into
an aircraft monitoring program to obtain data quickly and efficiently over a large areas.
Most of the analysis required to relate ozone concentrations aloft with those of the surface
has been performed by Ludwig (1979). Effort on this project has been directed toward adapt-
ing his findings to the design of aircraft monitoring programs. Ludwig examined 268 ozone
profiles measured in different parts of the country and classified them according to their shape.
This kind of analysis is obviously useful to the solution of the problem of estimating surface
concentrations from data measured aloft. Aircraft measurements can be easily used to esti-
mate surface concentrations whenever the vertical ozone-concentration profile is characterized
by small vertical gradients. The problem is knowing when it can be reliably assumed that such
vertical profiles of ozone concentrations are present.
An other factor that needs to be considered is the compatibility of the aircraft data with
the hour-averaged surface measurements. At typical wind speeds of a few meters per second,
an hour-average ozone value is derived from air that has traveled about 10 km. In an aircraft
traveling about 50 m/s, this distance will be covered in about 3 minutes. The response time of
an ozone monitor will generally be measured in seconds, so the aircraft measurements will tend
to provide somewhat finer spatial/temporal detail than the ground measurements. However, in
practice the aircraft data are reported with separations of a few kilometers between observa-
tions (see e.g., Westberg et at., 1977). When such data are analyzed, the result is a spatial
smoothing comparable to that for one-hour averaging.
The remainder of this section describes the classification system developed by Ludwig
(1979), identifies the types of profile suitable for aircraft-survey determinations of surface con-
centrations, and describes the conditions under which those profiles are most likely to be
found.
42
-------
2. Types of Vertical Profiles
of Ozone Concentration
The reader will better understand the classification system used for vertical ozone profiles,
especially the later discussions of the conditions under which those profiles are most likely to
be observed, after a brief discussion of factors governing the three-dimensional distribution of
ozone in the atmosphere. Basically, the distribution of ozone is determined by the chemical
processes that lead to its formation and destruction and by the meteorological processes that
encourage or inhibit vertical mixing.
The chemical processes that cause the formation of ozone from the primary emissions of
hydrocarbons and oxides of nitrogen operate only during the daylight hours. A substantial
period of time (hours) is usually necessary before the resultant ozone is produced. However,
destruction of ozone can occur at any time and generally proceeds rather rapidly. The most
rapid process for destroying ozone is its reaction with nitric oxide; ozone can also be destroyed
by contact with surfaces. Destruction proceeds more slowly in the free atmosphere.
Significantly, most destruction of ozone occurs near the ground.
The other important governor of ozone distribution in the lower atmosphere is mixing.
Atmospheric mixing determines the depth to which precursors (which are usually emitted near
ground level) are mixed, during or before the ozone-formation process. Atmospheric mixing
also governs the amount of pre-existing ozone that is mixed downward to replace ozone des-
troyed at, or near, the surface. Most vertical profiles result from combinations of these
processes and fall into a limited number of categories. These categories and their causes are
discussed below.
Ludwig's work categorizing ozone profiles (1979) evolved from earlier work by Johnson
and Singh (1977a) and from the analysis of a large amount of data collected by others in the
following geographical areas: St. Louis area (Mage et al., 1978), the northeastern United States
(Environmental Monitoring and Support Laboratory, 1975; Spicer et al., 1976; Washington State
University, 1976 and Wolff et al., 1975), Washington D.C. (Fitzsimmons et al., 1978), Houston
(Westberg et al., 1977), Los Angeles (Blumenthal et al., 1974; Johnson and Singh, 1977b),
Indianapolis (Lovelace et al., 1975), and Toronto (Weibe et al., 1975).
Figure 24 shows the six commonly found ozone profile types. The height and concentra-
tion scales shown in the diagram are used for illustrative purposes only; observed conditions
frequently differ substantially. It is obvious that measurements made anywhere within the
boundary layer will do a good job of characterizing surface concentrations when the profiles fall
in either the A or D categories.
Figure 25 shows the relationship between the hour-average concentration measured at
the surface and the average concentration through the mixing layer as derived from observed
aircraft profiles that were classified either as Type A or Type D. The unit slope and essentially
zero intercept of the least-squares regression line (correlation = 0.98) allows the assumption
of equality between mixed layer and surface values to be made with considerable confidence.
The close agreement should not be surprising, since one of the criteria for classifying a profile
as either a Type A or Type D is that the vertical gradient of concentration within the mixing
layer be small. Nevertheless, the findings are quite useful because they demonstrate that such
conditions are frequently found.
43
-------
3
2
1
E
JC
; °
X
I 3
2
1
o
~
(a)
-
_
if | |
0 40 80
r-
(b)
_
_
/
1
1 (0
1
1
L X*
MIXED LAYER 4
3 40 80 120 (
I
w
1
1
V,
^
MIXED LAYER t
^ 1
».. . « . t » y j
(e)
-
V
x\
STABLE LAYER ^1
) 40 80 120 160
1
I (f)
I
*""~"^
^J
MIXED/ LAYER
I .It . t . )
40 80 0 40 80 120 0
OZONE CONCENTRATION - ppb
Figure 24. Six types of vertical profile.
40
80
120 160
Section 3 discusses the times of day, locations relative to urban areas and meteorological
condition that have been found to be associated with Types A and D profiles. The application
of these findings to the design of an aircraft sampling program is discussed in Section 4.
44
-------
Wl
. t + + » 1 1 + + » + f » + *
V = 1.015x - 0.33
CORRELATION <= 0.98
O
Z 117.00
oar' 3«
10.00 »/
o.oo 40.oo ao.oo 120.00 leo.oo 203.00 240.00 200.00 320.00 aea.oo 400.00
SURFACE OZONE CONCENTRATION ppb
Figure 25. Scattergram of average mixing-layer ozone concentrations as a function of surface concentrations
for profiles A and D.
-------
SECTION 3
SUPPORTING MATERIAL
The preceding section contained a discussion of the findings that were most relevant to
the project objectives and certain supporting material necessary for an understanding of these
findings. Some results from the project did not lead to any major conclusions (e.g., some of the
analyses of the relationships among daily peak-hour ozone readings at different stations) or
were not necessary to the understanding of those major findings presented earlier (e.g., the
decisions and logic that went into the selection of data to be used for detailed studies and for
the selection of case studies). This section is provided for the sake of completeness. It
includes some discussion of the logic underlying decisions and some of the less conclusive
results.
A. Data Survey and Selection
The project specified that the data from at least four geographical areas be used for the
statistical analysis. These regions were to have surface-monitoring data sufficiently extensive
that statistical comparisons among sites would be possible. It was also important that the
chosen geographical areas cover a variety of climatological conditions. With these considera-
tions in mind, four areas were selected in consultation with the Project Officer. The four areas
that were chosen are shown in Figures 26 through 29. (The locations from which data were
available through the SAROAD for the year 1977 are also shown in the figures.) The four areas
include:
The Northeastern statesPennsylvania, New York, New Jersey, Connecticut, Rhode
Island, and Massachusetts.
The Midwestern statesIowa, Illinois, Missouri, Indiana, Kentucky, and Ohio.
Texas and the adjacent states of Oklahoma, Arkansas, and Louisiana.
Southern California stations, south of the Tehachapi Mountains.
These states encompass a variety of climatic regions, and the figures show that they have rea-
sonably dense air quality monitoring networks.
After the study areas were selected and the data obtained from the Project Officer, the
hourly data was processed to make it more manageable for further analysis. The hourly
SAROAD data were processed to extract the daily maximum-hourly concentrations, using the
following approach:
All units were transformed to parts per billion (ppb). Daily maxima were selected from
the hourly values whenever there were at least 16 observations for the same day and
at least 9 observations from the period 0800 to 1800 local standard time (LST).
Peak-hour values of zero were deleted.
Excessively high values (greater than 300 ppb in the winter or greater than 500 ppb in
the summer) were deleted.
46
-------
38
-80
-78
-76 -74
LONGITUDE
-72
-70
Figure 26. Locations of SAROAD Northeast ozone monitoring stations for 1977.
47
-------
92 90 88 86 84 82
36 -
94
Figure 27. Locations of SAROAD Midwest ozone monitoring stations for 1977.
48
-------
108 106 104 102
100 98
LONGITUDE
Figure 28. Locations of SAROAD Southwest ozone monitoring stations for 1977.
49
-------
35.0
34.5 -
34.0 «
O
D
11.8 11.8 11.7 11.7 11.6 11.6
11.9 11.9
11.5
Figure 29. Locations of SAROAD ozone and oxidant monitoring stations
in Southern California for 1977.
50
-------
The result of this processing was a computer tape that contained the peak-hour values for
all the stations and for all the suitable days of the years 1975, 1976. and 1977. With the data
in computer-compatible form, cross-tabulations of observed peak-hour ozone concentrations
were prepared for various station-separation intervals.
In addition to the hourly data from the four selected geographical regions, the Project
Officer also supplied a data tape of the maximum and second-maximum observed hourly ozone
concentrations for all the stations available from the SAROAD for the years 1974 through 1977.
This data tape also contained information regarding the number of hours of observations from
which the maximum and second-maximum values were determined. This data base was con-
structed prior to promulgation of the new statistical form of the ozone standard (Federal Regis-
ter, 1979a), so estimated values in terms of the daily form of the standard were not readily
available from SAROAD. Procedures for determining compliance with the new standard and for
estimating design values have been published by EPA (Curran, 1979). The daily form is only
slightly less stringent (Federal Register, 1979a) than the originally specified interpretation.
B. Identification of Conditions Associated
with Widespread Exceedances of the Ozone Standard
The 1977 daily peak-hour ozone data from the four different geographical regions were
used as the basis for selecting a limited number of "episode days," when high ozone concen-
trations were widespread in the areas. Each of the four geographical regions was considered
separately. The goal was to identify about a dozen days for each region and to examine those
days to determine the shared meteorological characteristics. Those characteristics were then
compared, from region to region and with earlier studies, particularly those of Ludwig, Reiter, et
al. (1977). The results of these comparisons were discussed in Section 2. Recommendations
derived from the analyses are presented in Section 4.
For each day of 1977 all the observed peak-hour ozone concentrations were averaged for
each region. A threshold was selected, against which the spatially averaged peak-hour con-
centrations could be compared, such that ten to fifteen days during the year of 1977 exceeded
the threshold in that particular region. Table 10 shows the threshold used for each region and
the number of days when the spatially averaged peak-hour ozone concentrations in the
corresponding geographical area exceeded that threshold. Table 10 shows the selected dates
for each region and the corresponding regional average peak-hour ozone concentrations.
TABLE 10 THRESHOLD OF SPATIALLY AVERAGED PEAK-HOUR OZONE
CONCENTRATIONS USED TO SELECT DATES FOR CASE STUDIES
Region
Northeast
Midwest
Texas/Louisiana
Southern California
Threshold
(ppb)
104
90
90
145
Number of Days
Over the Threshold
10
11
14
13
51
-------
In addition to the selected cases from 1977 discussed above, the daily weather maps and
peak-hour ozone concentration analyses for 1974 that were presented by Ludwig, Simmon et
al. (1977) were reexamined. These same analyses were used as the basis for an earlier study
(Ludwig, Reiter et al., 1977), where the areas with ozone concentrations in excess of 80 ppb
(the NAAQS in effect at the time the analyses were prepared) were examined to determine what
kinds of weather features were associated with such high concentration areas in the eastern
United States. These earlier analyses made no attempt to determine if the areas of high ozone
concentration were associated with different meteorological features from region to region.
The reexamination of the earlier analyses undertaken as part of this project focused on
geographical areas where ozone concentrations were in excess of the higher, newer NAAQS of
120 ppb. Furthermore, an attempt was made to determine whether these high concentrations
were associated with different meteorological features in the Northeast states, the Midwest
states, and the Texas-Louisiana-Arkansas area and the Southeast states. The Southern Cali-
fornia area was not included in this reinterpretation, because it falls outside the area covered
by the maps prepared by Ludwig, Simmon et al. (1977). The results of the analyses were sum-
marized in Table 7.
It should be noted that the automated analysis procedures that were used to prepare the
ozone analyses introduce considerable smoothing into the field. The methodology is discussed
in detail by Ludwig, Simmon et al. (1977). Basically, the method of analysis begins by deriving
ozone concentrations for a number of grid points. These values are based on weighted aver-
ages of peak-hour ozone concentrations observed in the vicinity. The weighting of the observa-
tions is an inverse function of the distance from the grid point. In essence, if there are a
number of observations close together, the analysis is drawn as though their average was used.
An extrapolation algorithm is also employed to extend the analysis beyond the area from which
observations are available.
C. Analysis of the Conditions Associated
with the Different Ozone Profile Types
Practical consideration in the design of an aircraft monitoring program requires that the
experimenter be able to identify when the aircraft measurements are suitable for representing
ground-level ozone concentrations. This is tantamount to requiring a knowledge of when the
vertical profile will be of either Type A or D (see Figure 24). It will also be necessary to deter-
mine when the aircraft is within a well-mixed layer with small vertical gradient concentrations.
This will be discussed at greater length when recommendations are given for the design of an
aircraft monitoring program.
The profile studies of Ludwig (1979) provide the necessary guidance for recognizing when
the profiles are apt to be Type A or Type D. As a general rule, Type A profiles evolve from Type
B profiles during the morning hours as the nocturnal surface inversion is destroyed by heating
and mixing. Usually, the transition is complete by about 1000 LST. Type D profiles sometimes
evolve from Type E profiles in a similar way, but the establishment of the Type D profiles in
such cases may not occur until the mixing layer reaches its maximum height for the day. The
Type D profiles also evolve from Type C profiles on occasion, after the air has traveled farther
from the city. The evolution of a Type D profile from a Type C requires that there be enough
time for the precursors emitted near ground level (and the ozone formed from them) to mix uni-
formly through the depth of the boundary layer. It should be apparent from the preceding dis-
cussion that the establishment of Type A or Type D profiles, with their relatively uniform distri-
bution of ozone through the mixing layer, requires fairly strong vertical mixing. Vertical mixing
depends on atmospheric stability, which in turn depends on time of day, cloud cover, and wind
speed (Turner, 1964).
52
-------
The objectives of this study are such that not all locations and conditions are of equal
interest. Thus, when an attempt is made to determine what conditions are associated with the
occurrence of Type A or Type D profiles, not all cases are of equal interest. The following con-
siderations are of greatest importance:
Conditions most likely to cause exceedances of the federal ozone standards.
Regions without an adequate ground-level monitoring network.
Areas away from cities (where ground-based monitoring would be most difficult to sup-
port).
Thus, it is necessary to identify Type A and D profiles in areas away from cities during periods
conducive to the occurrence of high ozone concentrations, i.e., in strong sunshine and light
winds. Strong sunshine and light winds are also associated with unstable atmospheric condi-
tions (see, for example, Pasquill, 1961 or Gifford, 1961) and vigorous mixing. Furthermore, des-
tructive processes that cause strong concentration gradients are generally more pronounced in
the urban areas, with their nitric oxide sources, than in the more remote areas. Thus, the condi-
tions and locations of greatest interest to this study are those that are frequently associated
with the occurrence of Type A and 0 ozone-concentration profiles.
1. Relative Frequency of Different Profile Types
at Different Times of Day
Table 11 is a contingency table showing the frequency of combination of profile type and
time of day in the data analyzed by Ludwig (1979). The first number in each square is the
number of cases having the combination of profile and time of day represented by the square.
The second number shows the percentage of the total number of cases in that row (time inter-
val) that had the characteristics of that column (profile). The third figure in the square shows
the percentage of the total cases in that column (profile) that occurred with the characteristics
represented by that row (time interval). Finally, the last number in a square shows the percen-
tage of the total cases that have the combination of characteristics represented by the square.
For example, the square at the second row and second column of Table 11 shows that the
sample contained 12 Type A profiles that occured between 0600 and 0900 LST. This
represented 20.7 percent of all the data collected between 0600 and 0900 and 15.0 percent of
all the type A profiles. Finally, 4.5 percent of all the observations were both taken between
0600 and 0900 and of Type A. The first column of the table shows that the sample includes
one case whose profile classification was uncertain.
Table 11 shows that the A and D profiles were very uncommon during the nighttime hours,
and represented less than one quarter of the observations between 0600 and 0900 LST. The
frequency of Types A and D increased to nearly 60 percent of the cases between 0900 and
1200 and to more than two-thirds of the afternoon cases. Thus, the chances of encountering
conditions where aircraft measurements could be used to estimate ground-level ozone concen-
trations are quite good during the late-morning and afternoon hours.
53
-------
TABLE 11 FREQUENCY OF OCCURRENCE OF DIFFERENT PROFILE TYPES
AT DIFFERENT TIMES OF DAY
BEFORf
600-900
f 900-1200
8
I
I 1200-lbOO
z 1500-1800
ui
I
H
1 6UO-2IOO
2100-2400
COLNT
HOW PCT
COL PCT
TUT PCT
1.
9
z»
3.
4.
a.
6.
7.
UNCERTAIN
0
3
0
0
0
0
0
0
I)
0
0
o
0
0
0
0
1
2.0
IOJ.O
.4
0
o
0
0
0
0
0
o
*
o
J
o
J
l£
20.7
lb.0
43.8
13.0
Z'j
3J .u
31 .3
4.3
T
13.7
a. a
.i.6
o
o
J
0
1
bO.O
1 *2
.4
B
5
£0.0
lb.4
.7
t,
10.3
46.2
2.2
4
5.i!
3O.e
l.b
0
0
0
0
0
0
0
0
0
o
0
0
1
bO.O
7.7
.4
PROFILE TYPE
C
0
0
0
J
0
0
0
0
4
5.<£
19. O
1.3
9
12.2
42.9
3.3
7
13.7
JJ.3
2.6
1
. 33. J
«.e
.4
0
o
0
0
10
3
3
0
)
2
3.4
2.9
.7
1 3
1 J.O
14.7
3.7
27
36.5
34.7
10.0
27
52. >
J.7
2.-:
.7
0
J
0
o
E
2
!>0.0
19.5
27.8
5.C.
5
6.8
V.J
1.9
t>
11 .8
II .1
2.2
0
0
0
o
0
3
0
0
COLUMN
TOTAL
80
2
-------
TABLE 12 FREQUENCY OF OCCURRENCE OF DIFFERENT PROFILE TYPES
AT DIFFERENT LOCATIONS RELATIVE TO THE CITY
en
ui
u
ui
x
o
ui
1C
o
p
u
3
COUNT
ROW PCT
CUL FCT
TOT PCT
'
UPWIND OR
TO ONE SIDE
ABOVE
CITY
DOWNWIND
UNCERTAIN
UNCERTAIN
1
.8
100. O
.4
0
0
0
0
0
0
0
0
0
o
0
0
A
bl
3U.3
02.8
10.0
14
20.9
17.5
5.2
9
17. t
11 .2
3.3
6
33.3
7.5
2.2
E
5
3.U
38.*:
1.9
4
o.O
30.8
1.5
4
7.e
30. a
1.5
0
0
0
0
PROFILE TYPE
C
9
t.e
42.9
3.3
1
l.S
4.8
.4
9
17.43
42.9
3.3
2
1 1. 1
9.5
.7
10
24
1 8 . J
35.3
H.9
13
26.4
27.9
7.1
19
37. 3
27.<>
7. 1
6
33.3
a. 4
2.2
IE
14.3
b9.4
7.1
9
13.4
2ft. 1
3.3
4
7.3
12. b
1 .5
0
0
0
o
IF
24
1R.O
44.4
8.V
20
29.9
37.0
7.4
6
11 .B
11 .1
2.2
4
22.2
7.4
1 .b
COLUMN
TOTAL
1
4
BO
13
4.e
21
7.8
b8
25.3
32
11.9
54
20. 1
HOW
TOTAL
133
49.4
67
24.9
51
19.0
18
6.7
269
100.0
-------
2. Relative Frequencies of Different Profile Types
in the Environs of Urban Areas
Table 12 shows the relative frequency of occurrence of the different profile types in
different locations relative to the nearest city. The key to the entries is analogous to that dis-
cussed for Table 11. For example, the entries in the second column of the first row show that
there were 51 observations of Type A profiles taken upwind or to one side of urban areas,
representing 38.3 percent of all the observations that were made upwind or to one side of the
city and 63.8 percent of all Type A profile observations. The first row of Table 12 shows that
over half the observations made upwind or to one side of the city revealed a profile of Type A
or Type D. Downwind of cities the total frequency of Type A or Type D is about the same as
upwind, but their relative proportions are reversed. That is, Type D profiles are found about
twice as frequently downwind of cities as Type A. The frequency of occurrence of Types A and
D profiles above cities is somewhat less than in areas outside the city.
3. Relationship Between Surface Ozone Concentrations
and Those Aloft for Different Stability Classes
It should be apparent from the above discussion that the vigor of atmospheric mixing
plays a large part in determining the nature of the ozone profile. Thus, the most unstable
atmospheric conditions should be associated with the most nearly uniform vertical distributions
of ozone. Figure 30 gives strong indications that this is true. For those measurements taken
during extremely unstable conditions (A stability), the ratios of the surface to mixed-layer ozone
concentrations cluster around one, and the differences between mixed-layer and surface con-
centrations cluster around zero. The stability increases toward Class D (neutral stability), as
does the frequency of ratios that are much greater and much smaller than one. Large negative
and larger positive differences also are more common under the less unstable conditions.
Table 13 presents information similar to that given in Figure 30; it shows the averages and
the standard deviations for the differences and ratios as stratified according to the stability
class that prevailed when the measurements were made. The average ratio of surface to
mixed-layer concentrations varies with stability class, but the standard deviation shows a
monotonic increase from the most unstable (A) to the neutral (D) conditions. Similar, but much
less pronounced, tendencies appear in the tabulation of average differences according to sta-
bility class.
TABLE 13 COMPARISON OF OZONE CONCENTRATIONS AT THE SURFACE
AND ALOFT BY STABILITY CLASS
Stability
Class*
A
B
C
D
Number of
Cases
27
81
68
50
Difference
(ppb)
Average
6.5
-1.1
10.1
3.5
Standard
Deviation
12.5
16.0
19.4
16.5
Ratio*
Average
0.92
0.99
0.86
0.90
Standard
Deviation
0.18
0.24
0.25
0.31
*A = extremely unstable; B = moderately unstable; C = slightly unstable; and D = neutral.
Mixing layer concentration - surface concentration.
Surface concentration/mixing-layer concentration.
56
-------
STABILITY I
CLASS
I .4 TO .6
.6 TO 6
i a TO i.
i
I 1 TO I 2
I
MBMM ( ;
I 1 2 TO 1.4
1
I I . .
FREQUENCY
1 - TO 1.2
I LESS THAN -60
C 2)
-60 TO -40
1 -20 TO 0
I 0 TO 20
I 20 TO 40
1 I .
0 10
FREQUENCY
B ( 1 )
I UNDER .2
I
1 .2 TO .4
I
M^» (
1 .4 TO .6
I
I .6 TO .0
I
t | 1 )
I -«0 TO -40
I
I -40 TO -80
I
t -20 TO 0
I
1 0 TO 20
I 20 TO 40
» ( 2)
I 40 TO 60
1 60 TO 80
0 10
FREQUENCY
. . . I .
40
. . I
90
I UNDER .2
I .4 TO .6
I .6 TO .6
( 1 .2 TO 1 .A
. ,. I
12
RATIO surface'mming l
( II
40 TO 60
DIFFERENCE average mixing lave'
mmus turiace czonf concenuaron
.. i
20
Figure 30. Histograms of ratios and differences between surface and mixing-layer ozone concentrations
for different atmospheric stability classes.
57
-------
4. Most Favorable Conditions for Obtaining
Aircraft Observations Representative
ol Ground-Level Ozone Concentrations
As noted above, the most favorable conditions for Types A and D ozone profiles occur out-
side urban areas during the late morning and early afternoon when the atmosphere is unstable.
Ludwig's (1979) tabulations of ozone profiles were used to determine the frequency of
occurrence of the Types A and D profiles under these favorable conditions. Table 14 summar-
izes the findings. There were 65 instances when ozone profiles were measured outside cities
between 1100 and 1600 LST during unstable conditions (Pasquill-Gifford stability classes A, B,
or C). The table shows that about 77 percent of such cases are typified by either Type A or D
profile.
Table 14 shows that the data analyzed by Ludwig (1979) included an appreciable number
of instances where Type C profiles were observed under conditions that are most favorable for
the occurrence of Type A or D. This is understandable; in general, the same conditions that
produce a Type D profile also produce a Type C. Differences between the two profiles arise
from differences in the time elapsed since the precursors left the source area. As noted earlier,
a Type D profile is produced when sufficient mixing has taken place to cause ozone and its
precursors to be uniformly distributed through the mixing-layer. Before the mixing is complete,
the profile will be of Type C, suggesting that Type C profiles are apt to be found closer to the
urban areas than Type D. (The need for aircraft monitoring to fill gaps in the existing network
is greatest away from the cities, where Type C profiles are less likely to occur.) Table 14 shows
the relative frequency of occurrence of the different profile types when Type C occurrences are
removed from the list: Nearly 90 percent of such cases are of Type A or D. The percentage
would be even higher if we assume that the Type C profiles evolved to Type D at locations
farther downwind.
TABLE 14 PROFILES OBSERVED DURING UNSTABLE CONDITIONS
OUTSIDE CITIES FROM 1100 AND 1600 LST
Profile
Type
A
B
C
D
E
F
Total
Number of
Cases
28
0
9
22
3
3
65
Percentage of
All Cases
43.1
0
13.8
33.8
4.6
4.6
100
Percentage of
Cases When
Type C Profiles
are Neglected
50.0
0
-
39.3
5.4
5.4
100
58
-------
D. Station Characteristics Likely to Result
in Unrepresentative Readings
Two methods were used for identifying cases where the data from stations were likely to
be unrepresentative. In one approach, the annual second-maximum values of ozone concentra-
tion were examined and compared with values from other nearby stations. All pairs of stations
that were separated by 25 km or less and that recorded second-maximum ozone concentra-
tions that differed by 100 ppb or more were identified. The second method used maps similar
to Figure 11, but for individual years. A subjective search was made for "L" symbols that were
well within circles that defined areas of probable exceedance of the NAAQS for ozone. When
the potential cases of unrepresentativeness had been identified, the characteristics of the indi-
vidual stations were examined after preliminary screening. The preliminary screening included
elimination of duplicate pairsin opposite order or for different yearsand obviously spurious
data.
Three important factors have been identified in connection with the pairs of sites that
were examined. These are:
Meteorological factors and location relative to the urban center and local wind patterns.
General urbanization of the surroundings.
Local traffic effects.
The greatest concentrations are most likely to be observed outside cities in a direction
that is downwind during optimum photochemical conditions. The highest concentrations are
found a few tens of kilometers from the city. Monitors located in these areas provide data that
are representative of the worst conditions. Monitors upwind or within the city can also provide
data that are representative of relatively large areas, but these are not the areas that have the
highest concentrations and hence, are not of as great interest for purposes of determining com-
pliance with the NAAQS as are the extraurban downwind locations.
The effects of heavy local traffic can cause the data to be totally unrepresentative of
larger areas in most instances, although some of the data suggest that high ozone concentra-
tions have been observed at locations with heavy daily traffic under special circumstances.
Such special circumstances might include nonpeak traffic hours or wind directions that miti-
gate effects of the local emissions.
The nature of the air quality standard is such that extreme observations are emphasized,
and extreme concentrations tend to be produced by atypical conditions. Thus, factors that
influence the representativeness of means or even high quantile values are not always opera-
tive when conditions are extreme. Thus, high ozone concentrations can be found near strong
NO sources or in the heart of urban areas on occasion, but it is much more probable that they
will be observed more frequently elsewhere.
59
-------
SECTION 4
RECOMMENDATIONS
A. Areas Where More Data are Required
Figure 12 can be used as a guide to determine the areas of the country that may not be
adequately defined in terms of exceedance of the NAAQS for ozone. It is obvious from the
figure that some rather large areas cannot be characterized, using only the ozone concentra-
tions available from the SAROAD. Figure 12 shows that the greatest need for additional infor-
mation is in the Southeast, the Northern Great Plains, and the Great Basin. Most of the
northeastern quadrant of the United States is adequately covered, but there is a conspicuous
gap in Maine. However, according to Gordon (1979), a monitoring program now underway
should remedy some of the need for additional data. This monitoring is generally limited to
those months when photochemical activity is greatest.
Based on an analysis of the nearly complete data sets, as shown in Figure 12, the state of
California and parts of the West Coast are covered; coverage in Kansas, Oklahoma, Texas, and
along the Gulf Coast tends to be marginal. Our ability to define the air quality status of areas in
much of the West from these relatively complete data sets is quite limited.
Numerous data sets were not included in Figure 12 because they were not complete
enough for some of the analyses that were performed. For example, the joint frequency
analysis used to define probabilities of standards exceedance required that the data sets be
comparable and nearly complete. A nearly complete data set is necessary to define an area as
"in compliance" because higher values, in excess of the standard, might have occurred during
the unmonitored hours. However, an incomplete data set that contains one or more values
above the standard can be used to determine that the standard has been exceeded. In such a
case the observed second-maximum ozone concentration places a lower bound that can be
used to define a radius of probable exceedance which is also a lower bound.
The preceding discussion, which was an extension of earlier arguments, has been
included in this recommendation section because it provides a basis for defining geographical
areas of the continental United States that should be studied, and because it suggests that
greater use of existing data sources would be profitable. For one reason or another, not all
monitoring data are archived in the SAROAD, so other sources should be actively explored.
B. Methods for Obtaining Additional Data
1. Survey of Non-SAROAD Data Sources
As noted above, it should be possible to expand the data base, and enhance the areas
that can be characterized with regard to exceedance of the standards for ozone, by obtaining
as much non-SAROAD ozone-monitoring data as possible. This should certainly be the first
60
-------
step, before undertaking any special aircraft- or ground-monitoring program. The expense of
special monitoring programs should make them a last resort; acquisition of data already col-
lected could be relatively inexpensive when compared to the cost of large-scale monitoring pro-
grams.
Since no single person is likely to be familiar with all the uncataloged ozone data sources
in the United States, it would seem most logical to contact many knowledgeable people. The
most systematic approach would be to contact EPA Regional Offices to identify known data
that had not been entered into the SAROAD. Data from any special studies that had been
sponsored by that Region, or that the Region was aware had been undertaken by industries
should be obtained if possible. The regional EPA offices might also supply names of
knowledgeable individuals in industry, state and local agencies. These sources should be
explored, with the focus on those geographical areas of Figure 12 that are largely
uncharacterized.
Contacts with various agencies and organizations should result in a catalog of potentially
useful ozone data. The catalog, as a minimum, should contain information concerning the dates
when the data were collected, the instrumentation and calibration methods used, and at least
the general locations of the monitoring sites. Some screening should be done to eliminate
unsuitable data. It might also be efficient to eliminate redundant data that would not add
significantly to the available coverage. More detailed screening should be undertaken to elim-
inate unreliable data and to ensure that the monitors were reasonably well-sited.
It should be noted that special aircraft data might prove useful. The relationship between
ozone data collected from aircraft and the ground-level concentrations has been discussed at
length. That discussion should help decide which aircraft data are suitable. In a subsequent
section, a protocol for aircraft sampling that will obtain suitable data is given. That protocol can
also help to determine the suitability of aircraft data.
2. Selecting Times and Areas for Study
To a large extent, the areas in need of further study will be defined by the analysis of
existing data. Areas that cannot be characterized with regard to exceedance of ozone stan-
dards must be considered as candidates for special studies. However, not all of those unmoni-
tored areas will be of equal importance: Priority should probably be given to those regions that
are suspected of being in exceedance of the standards. It is well established that areas with
warm temperatures, plentiful sunshine, and appreciable anthropogenic emissions are candi-
dates for the formation of photochemical ozone. Using these facts as a guide, unmonitored
areas in the southern tier of states should probably be given high priority. The northern Great
Plains should also be high on the list, followed by the Great Basin states.
Table 6 can serve as a useful guide for deciding what months would be most profitable for
conducting special monitoring programs. It is assumed that the months when the highest con-
centrations are most likely are the months that would be best for monitoring, because they
would be most likely to produce decisive evidence of exceedances. Futhermore, that no
exceedances were observed during these months, when the probabilities were highest, could
be taken as reasonably strong evidence of compliance with the standard.
/
Ludwig, Reiter et al. (1977) found that relatively high concentrations were most frequently
observed in June and July over the Florida Peninsula and along the Texas-Louisiana Gulf
61
-------
Coast. In the Ohio River Valley, the frequencies were greatest in July and August. This sug-
gests that the month of July would probably be best for a monitoring program in uncharacter-
ized areas in the southeastern United States. Judging by the observed frequencies of relatively
high concentrations in 1974 in the Texas-Louisiana Gulf Coast Area and the western parts of
Oklahoma, Kansas and Nebraska, monitoring programs in Texas or the southern Great Plains
regions could be undertaken almost equally well in June or July; favorable conditions extend
into August as well.
Little information is available from the northern High Plains. Singh, Ludwig and Johnson
(1977) summarized the seasonal variability of measured ozone concentrations in several
remote locations. Their analyses suggest that the highest concentrations are most likely to be
found in the late winter and spring months, a fact that suggests non-photochemical sources.
Some of the High Plains sites also had relatively high ozone concentrations during summer. It
appears that the locations analyzed by Singh, Ludwig and Johnson (1977) in the northwestern
High Plains were all subject to influences from stratospheric ozone sources; some of the loca-
tions were also subject to occasional influxes of anthropogenic precursors and the consequent
photochemical ozone.
In summary, it appears that monitoring in the northern Great Plains and the Great Basin
should be done in both the spring and summer months. Data from Great Basin locations in
Utah and western Colorado that were analyzed by Singh, Ludwig, and Johnson (1977) showed
increases in the average ozone concentrations and daily peak-hour concentrations during the
late winter or spring that remained at relatively high levels through July and August.
For purposes of planning more-or-less routine ground-monitoring programs, recommenda-
tions for seasonal monitoring are probably sufficient. For less-routine measurement programs,
however, especially those involving aircraft, some more detailed specifications of the measure-
ment program would be useful: in particular, the synoptic-scale weather features that were
most likely to yield high ozone concentrations. Routine National Weather Service forecast
material could thus be used in field planning to identify promising periods for conducting tests.
Tables 7 and 9 show the frequency of association of high ozone concentrations with cer-
tain kinds of synoptic meteorological features in various parts of the United States. In inter-
preting the tables, it should be recalled that high ozone concentrations are associated with
plentiful sunshine and are generally positively correlated with temperature. Therefore, field
monitoring would best be conducted during periods with clear skies and high temperatures,
especially when the area of interest is under the influence of the appropriate synoptic-scale
weather system.
In the southeastern United States, the highest ozone concentrations are most likely to be
found in the southwestern parts of high-pressure systems. Warm, sunny days when the south-
ern Great Plains are under the influence of the western or southwestern portions of the high-
pressure system should have the greatest potential for observing high ozone concentrations.
The northern Great Plains and the Great Basin are frequently associated with high ozone con-
centrations in the northwestern region of high-pressure systems.
According to Tables 7 and 9, nearly all occurrences of high ozone concentrations are
associated with warm air masses, rather than with the cold air behind weather fronts. Fre-
quently, high ozone is found on the periphery of the warm air mass, just ahead of a cold front or
in the vicinity of a warm front. (For purposes of classification, we have separated these regions
from the more central regions of the high-pressure system.) Tables 7 and 9 show that high
ozone concentrations can frequently be found in the vicinity of fronts, usually in the warm air,
so such regions should not be ignored in planning field-monitoring programs.
62
-------
3. Protocol for Aircraft Monitoring
When the appropriate areas and seasons have been identified, the field project has been
mounted, and conditions appropriate for measuring high ozone concentrations are forecast,
measurements must be conducted in such a way that they are representative of conditions
near ground level. Earlier sections of this report discussed the criteria for identifying situations
when concentrations aloft and at ground level were about the same. This section describes a
protocol for aircraft monitoring to meet those criteria.
The aircraft observations should be undertaken whenever a Type A or Type D ozone
profile (as shown in Figure 24) is likely to exist. In general, the flight should take place after
1000 or 1100 LST and before 1700 or 1800 LST, consistent with the requirement of sampling
when ozone concentrations, especially those from photochemical processes, are apt to be
greatest. The likelihood of encountering a Type A or Type D profile is further enhanced if con-
ditions are sunny and the atmosphere is unstable. It is assumed that the aircraft monitoring
would be conducted during periods and at locations (relative to synoptic weather features) that
are most likely to be associated with exceedances of the ozone standard. Furthermore, it is
assumed that the region will be without an extensive ground-level monitoring network. Finally,
it is also assumed that the areas of greatest interest will be outside cities, where ground-based
monitoring is more difficult to support and where higher ozone concentrations are more likely to
occur. As noted earlier, conducting the aircraft programs outside cities increases the likelihood
of encountering Type A or Type D ozone profiles.
It will be necessary to recognize, while in flight, when the aircraft's ozone measurements
represent ground-level conditions. It is therefore essential that the following parameters be
measured from the aircraft:
Ozone concentration
Temperature
Dew point
Altitude or atmospheric pressure.
Nephelometer measurements, to indicate atmospheric aerosol content, would be highly desir-
able. Other pollutant measurements (e.g., carbon monoxide, oxides of nitrogen, nonmethane
hydrocarbons) would be of some use, but by no means essential. Continuous ground monitor-
ing of ozone at one or two locations would be useful.
To determine when the aircraft is within the mixing layer (so that the measurements will
be representative of ground-level concentrations), the equipment should have simple, on-board
data processing and display capabilities to provide graphical plots of recently measured param-
eters as a function of altitude or pressure. Subsequent interpretation will require that the data
be recorded on board in computer-compatible form. Information must be recorded that can be
used to determine the location of the aircraft at any specified time.
Recent aircraft field operations (e.g., Johnson, et al., 1979) have shown that a large field
team is unnecessary when only daytime operations are required. Two persons, in addition to a
pilot, are required. One of the two team members must be capable of field maintenance of the
equipment. The other member of the team must be an experienced air quality meteorologist.
Both must be able to operate the equipment and recognize when the aircraft measurements
represent ground-level ozone concentrations. While one of the team is flying, the other gathers
corollary information (e.g., weather conditions and ground-level air-quality measurements) and
interprets data from earlier flights.
63
-------
Once areas for the monitoring program have been identified, an airport suitable for the
operations should be selected and the necessary arrangements made. Satellite airfields or
other areas where vertical profiles might be taken should also be selected before operations
begin.
To ensure that the aircraft monitoring is conducted when the vertical ozone profile is of
Type A or Type D, and that the observations are made in areas of likely high ozone concentra-
tions, a sampling protocol similar to that described below and shown schematically in Figure 31
should be undertaken.
Sfep 7 Begin sampling operations after the predicted time for breakup of ground-level
inversions, probably after 1000 or 1100 LST. An acoustic sounder would be useful to verify
that the break-up of the surface inversion had occured; acoustic sounder data would also pro-
vide information on the depth of the mixing layer.
Sfep 2Fly a vertical spiral near the airport through the top of the mixing layer or to 1.5
km, whichever is the lower. Typically, a vertical spiral to 1.5 km can be flown up-and-down
within 7 to 10 minutes in a twin-engine aircraft. During this spiral, the location of the top of the
mixing layer should be determined from on-board displays of the vertical profiles of ozone, tem-
perature and dew-point. (Figure 32 shows some profiles taken near Washington, D.C. when the
top of the mixing layer was very clearly defined by a temperature inversion and an abrupt drop
in pollutant concentrations.) The top of the mixing layer should be identified on board the air-
craft, based on the available data displays. This initial vertical profile serves two purposes: It
ensures that the ozone profile is of the proper type and that the measurements are being made
within the mixing layer.
Sfep 3Fly horizontally, at an altitude as far below the top of mixed layer as regulations
permit, to a location upwind of the city, and, as shown in Figure 31, fly another vertical spiral. If
the study is not focused on a specific city, this step can be omitted, and the flight would
proceed to the next step.
Sfep 4Fly downwind in a "zig-zag" path, such as that shown in Figure 31. This flight
should be at an altitude that is as far below the estimated top of the mixed layer as regulations
allow. If a "plume" of ozone is identifiable in the record, the "zig-zag" should be centered on it.
Sfep 5Obtain vertical profiles by spiral ascent/descent at 50-100 km intervals, depend-
ing on the spacing of suitable locations. These profiles will verify that the measurements are
being made within the mixing layer and that the nature of the vertical distribution of ozone is
still of Type A or D.
Sfep 6Patterns similar to those described in the preceding steps should be flown until
late afternoon (about 1600 LST). (Patterns can be flown in alternate directions.) If the study is
focused on an individual city, one or two round-trip flights should reach distances of 150 or
200 km downwind. If more than one city is included, the areas 100 to 150 km downwind of all
of them should be surveyed. If no urban areas are of interest, a general "search pattern" would
be appropriate. In all cases the horizontal flight should be interrupted for periodic determina-
tion (50 to 100 km intervals) of the vertical distribution of ozone, temperature, and dew point,
preferably near a ground level monitor, if available.
The above description of the required steps is incomplete, as it does not describe pre-
flight and post-flight instrument checks and calibrations, the importance of notebook entries,
and other commonly observed practices. Although a detailed discussion of such matters is
beyond the scope of this project, it should be understood that good monitoring and measure-
ment practices will be essential.
64
-------
CT>
(71
*- SPIRAL
Through mixing layer or 1.5 km
Ground Truth
Ozonu Monitor
Acoustic Sounder:
Mixing depth
Figure 31. Schematic diagram of aircraft ozone-measurement program designed to estimate ground-level concentrations.
-------
o>
O)
15 JO n 50 u 50 100 150 JOO
DEW POINT (°C)
TEMPERATURE ("O
BSCAT (10-6 m-1)
OZONE (ppb)
SOURCE: LUDW1G, 1979
Figure 32. Vertical profiles of ozone, temperature, dew point and BSCAT taken near Washington, D.C.
-------
4. Ground-Level Monitoring
Aircraft monitoring has the advantage of covering large areas, but it is expensive and will
generally be possible for only a limited time period. Ground monitoring at fixed locations will
be limited in spatial scope but can used over extended periods of time. Fixed, ground-based
monitors are indicated in locations with data gaps where some identifiable source of precursor
emissions is suspected of causing ozone exceedances.
Ludwig and Shelar (1978) give detailed instructions for locating sites where high ozone
concentrations are probable. It is recommended that their directions be followed in those
instances where identifiable precursor emissions from urban areas have not been adequately
monitored in the past. As noted before, special measurements may not be required on a per-
manent basis: Data collection during a few months when high ozone concentrations are prob-
able should suffice. Table 6 provides information that can be used to identify such months.
Ludwig and Shelar (1978) give some guidance for determining the "downwind" direction for
instances of probable high ozone concentration. The information concerning locations relative
to synoptic-scale weather systems given in Tables 7 and 9 and the information concerning
surface-wind direction associated with high ozone concentrations could be used for this pur-
pose.
If a program of monitoring at ground level is undertaken in more remote areas, such as the
northern Great Plains, the site selection will be governed more by the character of the entire
area than by location relative to specific anthropogenic sources, and sites should be chosen so
that their surroundings are characteristic of those of the area. This should not be difficult in
the northern Great Plains, but may not be possible in mountainous areas. In relatively flat
areas, low-lying ground should be avoided: Singh, Ludwig, and Johnson (1977), among others,
found that sites elevated above the surrounding terrain generally give readings characteristic of
the lower layers of the atmosphere throughout the day. Low-lying sites tend to be isolated from
conditions aloft by cold-air drainage, especially at night. As before, Table 6 provides informa-
tion concerning the best seasons for monitoring.
67
-------
REFERENCES
Bach, W.D., W.J. King, and P.M. Vukovich, 1976: "Non-Urban Ozone Concentration in Transient
High Pressure Systems," Paper presented at the Non-Urban Tropospheres Symposium,
sponsored by the American Geophysical Union and the American Meteorological Society,
Hollywood, Florida, 10-12 November.
Beard, M.E., 1978: "Evaluation of the Gas Phase Titration of Excess Ozone with Nitric Oxide
Calibration Procedure for Ozone Analyzers," Environmental Monitoring and Support
Laboratory, U.S. Environmental Protection Agency, Research Triangle Park, North Carolina,
32pp.
Curran, T.C., 1979: "Guideline for the Interpretation of Ozone Air Quality Standards," U.S.
Environmental Protection Agency Report No. 450/4-79-003, 37 pp.
Environmental Monitoring and Support Laboratory, 1975: "Meteorological Data for the Northeast
Oxidant Transport Study," U.S. Environmental Protection Agency, Las Vegas, Nevada
89114 (Xerox draft).
Federal Register, 1979a: "National Primary and Secondary Ambient Air Quality Standards," Vol.
44, No. 28 (8 February 1979), pp. 8202-8237.
Federal Register, 1979b: "Ambient Air Quality Monitoring, Data Reporting, Surveilliance Provi-
sions," Vol. 44, No. 92 (10 May 1979), pp. 27558-27604.
Fitzsimmons, C.K., K. Zeller, and M.J. Pearson, 1978: "Analysis of Aerometric Data Collected by
Aircraft During a Stagnation Period in Washington, D.C., August 1978," Presented at APCA
meeting, Houston, Texas, Paper 78-10.7.
Gifford, F.A., 1961: "Use of Routine Meteorological Observations for Estimating Atmospheric
Dispersion," Nuclear Safety, 2, pp. 47-51.
Gordon, N., 1979: Personal communication, State of Maine, Department of the Environment.
Hogeson, J.A., 1976: "A Survey of Calibration Techniques for Atmospheric Ozone Monitors,"
National Bureau of Standards Report NBSIR 76-1191, 29 pp.
Johnson, W.B. and H.B. Singh, 1977a: "The Origin and Significance of Ozone Maxima Aloft,"
Interim Prog. Report for Coordinating Research Council, Contract CRC-APRAC
CAPA/12/72 (1-76), SRI Project 2092, Stanford Research Institute. Menlo Park, California.
Johnson, W.B. and H.B. Singh, 1977b: "Vertical Profiles from Aircraft Soundings Conducted
During the Los Angeles Reactive Pollutant Program (LARPP)," Appendix A of Interim Prog.
Report for Coordinating Research Council, Contract CRC-APRAC CAPA/12/72 (1-76), SRI
Project 2092, Stanford Research Institute, Menlo Park, California.
68
-------
TECHNICAL RETORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
2.
3. RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE
ASSESSING THE REPRESENTATIVENESS
OF OZONE MONITORING DATA
5. REPORT DATE
December 1979
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
F.L. Ludwig and Eugene Shelar
8. PERFORMING ORGANIZATION REPORT NO.
SRI Project 7863
9. PERFORMING ORGANIZATION NAME AND ADDRESS
SRI International
333 Ravenswood Avenue
Menlo Park, California 94025
10. PROGRAM ELEMENT NO.
Task Order 2
11. CONTRACT/GRANT NO.
68-02-2548
12. SPONSORING AGENCY NAME AND ADDRESS
Office of Air Quality Planning and Standards
Research Triangle Park, North Carolina 27711
13. TVPE OF REPORT AND PERIOD COVERED
FINAL
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
16. ABSTRACT
Areas within which the National Ambient Air Quality Standard (NAAQS) for ozone is likely
to have been exceeded are shown to be related to the observed annual second-maximum ozone concentrations.
All pairs of SAROAD stations separated by 500 km or less that had nearly complete annual data sets were considered for
the years 1974 to 1977 to determine the probability that the NAAQS would be exceeded at one station of the pair, given
the observed second-maximum ozone concentration at the other and their separation. Circles were drawn around each
SAROAD monitoring site to show the area within which it is 95 percent probable that the 120 ppb NAAQS has been
exceeded for two or more hours during one or more years.
The report describes meteorological situations and seasons during which high ozone concentrations are most likely in
various parts of the country, so that special monitoring studies can be designed to supplement existing data. An aircraft
monitoring protocol is given that will provide ozone observations that are representative of ground-level conditions.
Conditions that cause ozone data to be unrepresentative are described as are uncertainties associated with instrumental
and calibration factors.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
Air Pollution Standards
Ozone
Meteorology
8. DISTRIBUTION STATEMENT
Release to public
19. SECURITY CLASS (This Report/
U
21. NO. OF PAGES
78
20. SECURITY CLASS (Thispage)
U
22. PRICE
EPA Form 2220-1 (R«v. 4-77) PREVIOUS EDITION is OBSOLETE
------- |