EPA-450/3-77-022a
October 1977
RELATION
OF OXIDANT LEVELS
TO PRECURSOR EMISSIONS
AND METEOROLOGICAL
FEATURES -
VOLUME I: ANALYSIS
AND FINDINGS
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Air and Waste Management
Office of Air Quality Planning and Standards
Research Triangle Park, North Carolina 27711
-------
EPA-450/3-77-022a
Final Report October 1977
THE RELATION OF OXIDANT LEVELS TO
PRECURSOR EMISSIONS AND
METEOROLOGICAL FEATURES
Volume I: Analysis and Findings
By: F. L. LUDWIG
SRI International
E. REITER
Colorado State University
E. SHELAR and W. B. JOHNSON
SRI International
Prepared for:
ENVIRONMENTAL PROTECTION AGENCY
OFFICE OF AIR QUALITY PLANNING AND STANDARDS
RESEARCH TRIANGLE PARK, NORTH CAROLINA 27711
Attention: MR. PHILLIP L. YOUNGBLOOD
CONTRACT 68-02-2084
SRI Project 4432
Approved by:
R.T.H. COLLIS, Director
Atmospheric Sciences Laboratory
RAY L. LEADABRAND, Executive Director
Electronics and Radio Sciences Division
SRI International
Menlo Park, California 94025 • U.S.A.
-------
ABSTRACT
Published ozonesonde data, radioactive fallout measurements and al-
pine ozone observations have been used to estimate the stratospheric
contribution to observed ozone concentrations at ground level. Long
term average effects from the stratosphere over the U.S. are on the
order of 10 ppb, with a springtime maximum around 20 to 25 ppb. Short
term stratospheric intrusion events resulting in one-hour-average con-
centrations of stratospheric ozone in excess of 80 ppb in the lower tro-
posphere have a frequency of only about 0.2 percent. Still fewer (but
some) of these events lead to ground-level impacts of such a magnitude.
Tropospheric causes of high ozone concentrations away from cities
have been investigated by statistical analysis of meteorological condi-
tions and the precursor emissions occurring along air trajectories and
by comparisons of weather maps and large-scale 03 distributions.
Meteorological factors are statistically more strongly correlated with
ozone concentration than are emissions, with air temperature being the
most highly correlated. At sites well removed from cities, the upwind
emissions of oxides of nitrogen are more strongly related to ozone con-
centrations than are the emissions of hydrocarbons. Widespread viola-
tions of the federal oxidant standard are most likely to be found in as-
sociation with a stagnant high-pressure system or in the warm southwes-
terly flow in the western portion of a high pressure area, often ahead
of an approaching cold front.
The results of this and other studies suggest that not all viola-
tions of the federal oxidant standard are controllable and this fact
must be considered in the design of control strategies. Also, for areas
within about 125 km of large cities, control might be achieved through
the reduction of HC emissions. In more remote areas, control strategies
involving NOx control throughout large regions must be considered.
iii
-------
TABLE OF CONTENTS
ABSTRACT . iii
LIST OF ILLUSTRATIONS vii
LIST OF TABLES xiii
ACKNOWLEDGEMENTS xv
CONCLUSIONS 1
1. INTRODUCTION 5
1.1. Motivation and Objectives 5
1.2. Lines of Investigation 6
1.3. Brief Description of Methods 7
1.3.1. General 7
1.3.2. Stratospheric Intrusion Analyses . . 8
1.3.3. Studies of Tropospheric Transport Processes 8
1.3.4. Synoptic Scale Ozone Distributions and
Weather Patterns 11
2. ANALYSIS AND RESULTS 13
2.1. Stratospheric Sources 13
2.1.1. Ozonesonde Case Studies . 13
2.1.2. Evidence of Ozone Transport from the
Stratosphere as Deduced from Studies of
Radioactive Debris 46
2.1.3. Ozone Observations at Zugspitze 59
2.2. Tropospheric Sources 63
2.2.1. Trajectory Analyses 63
2.2.2. Studies of Synoptic-Scale Ozone Distributions
and Weather Patterns 101
2.2.3. Combining the Trajectory Approach with Synoptic
Scale Comparison 124
3. DISCUSSION 139
3.1. How Much Does Stratospheric Ozone Contribute to
Ground-Level Ozone Concentrations? 139
3.1.1. General 139
3.1.2. Long-term Average Stratospheric Contribution .... 140
3.1.3. Short-term Stratospheric Contributions 140
3.2. What Are the Tropospheric Causes of High Ozone Values
in Areas Removed from Emissions? 141
3.2.1. Meteorological Factors 141
3.2.2. Emissions Factors 142
3.3. Implications for Control Strategies 143
3.3.1. Interactions between Ozone of
Stratospheric and Tropospheric Origins 143
3.3.2. Strategies for Control of Ozone of Tropospheric Origin 144
3.4. Recommendations for Future Research 146
REFERENCES 149
-------
LIST OF ILLUSTRATIONS
Figure Page
1: Candidate and Selected Sites for Detailed Analyses
of Air Histories 9
2: Ozone Sounding at Goose Bay, Canada, 2 May 1963 17
3: Radiosonde Ascent at Goose Bay, Canada, 2 May 1963, 0000 GMT. . 18
4: North American Surface and 500 mb Charts, 28 April
to 3 May 1963 20
5: Montgomery Stream Function Analyses, 30 April -
2 May 1963 23
6: Composite 6-hour Trajectory Segments from 29 April, 1963, 1800
GMT, 2 May 1963, 1200 GMT 24
7: Radiosonde Ascent at Baker Lake, Canada, 1 May 1963,
1200 GMT 25
8: Ozone Sounding at Thule, Greenland, 1 May 1963 26
9: North American Weather Maps, 24-25 June 1963 28
10: Ozone Sounding at Bedford, Massachusetts, 26 June 1963 29
11: North American Weather Maps 10 August to 15 August 1963 .... 30
12: Ozone and Dewpoint Sounding at Tallahassee, Florida,
14 August 1963 32
13: Temperature and Dewpoint Soundings at Tallahassee,
Florida, 14 August 1963 33
14: Montgomery Stream Function Analyses, 12 August to
14 August 1963 34
15: Sounding at Montgomery, Alabama, 14 August 1963,
1200 GMT 35
16: Soundings for Omaha, Nebraska and Columbia, Missouri,
13 August 1963, 0000 GMT 37
17: Soundings for The Pas (11 August 1963) and Resolute
(12 August 1963), Canada, 0000 GMT 38
18: Composite 6-hour Trajectory Segments from 11 August, 1963, 1800
GMT to 14 August 1963, 1200 GMT 39
vii
-------
LIST OF ILLUSTRATIONS (Continued)
Figure page
19: Ozone Sounding at Seattle, Washington, for 8
April 1964 40
20: North American Weather Maps, 6-7 April 1964 41
21: Ozone Sounding at Seattle, Washington, 15 April 1964 42
22: Ozone Sounding at Seattle, Washington, 17 April 1964 43
23: North American Surface and 500 mb Charts, 13 April
to 17 April 1964 44
24: Mean Meridional Distribution of Ozone 48
25: Monthly Mean Total Ozone Amounts at Arosa, Switzerland. ... 49
26: Mean Ozone Distribution for March-April 1963 and
Strontium 90 Distribution for May-August 1963 50
27: Stratospheric Inventory of Sr90 in the
Northern Hemisphere 51
28: Atmospheric Ozone and Sr90 Distributions, Fall
1964 and Spring 1965 52
29: Sr90/03 Ratios as a Function of Time 53
30: Mean Fallout in 1963 56
31: Seasonal, Maximum and Daily Fallout Frequency Maps
for 1963 57
32: Maximum 24-hour Fallout in 1964 58
33: Joint Frequency Diagram of Peak-Hour Versus 24-Hour
Average Ozone Concentrations at Zugspitze 59
34: Ozone Concentrations at Zugspitze 8-9 January 1975 61
35: McHenry Trajectories—Ozone Concentrations in Top 20
Percentile 64
36: McHenry Trajectories—Ozone Concentrations not in Top 20
Percentile 65
viii
-------
LIST OF ILLUSTRATIONS (Continued)
Figure Page
37: Queeny Trajectories—Ozone concentrations in top 20
percentile 66
38: Queeny Trajectories—Ozone Concentrations not in top 20
percentile 67
39: Wooster Trajectories—Ozone concentrations in top 20
percentile 68
40: Wooster Trajectories—Ozone concentrations not in top 20
percentile 69
41: Yellowstone Lake Trajectories—Ozone concentrations
in top 20 percentile 70
42: Yellowstone Lake Trajectories—Ozone concentrations
not in top 20 percentile 71
43: Scatter Diagram of Ozone Concentration Versus Precipitation
Index During Last 12 hours of Trajectory-Combined Data. ... 76
44: 95% Confidence Limits for Coefficients in the Regression
Equations Using Weighted Emissions Indices 83
45: 95% Confidence Limits for Coefficients in the Regression
Equations Using Emissions for Last 12 Hours 84
46: Scatter Diagram of Observed Ozone Concentrations versus
Those Estimated from a Regression Equation Using Indices
of Emissions and Temperature During the Last 12 Hours .... 86
47: Scatter Diagram of Observed Ozone Concentrations Versus
Those Estimated from a Regression Equation Using
Temperature and Weighted Emissions Indices 87
48: Scatter Diagram of Observed Ozone Concentrations Versus
Those Estimated from a Regression Equation Using
Temperature and NOx Emissions 89
49: Scatter Diagrams of Estimated Versus Observed Ozone
Concentrations for Two Piecewise Linear Regression
Expressions 91
ix
-------
LIST OF ILLUSTRATIONS (Continued)
Figure Page
50: Scattergram of Estimated Versus Observed Ozone
Concentrations for Piecewise Linear Regression Using
Temperature and Insolation 93
51: Scatter Diagram of Ozone Concentration Versus the Distance
to the Air Parcel Position 12-hours before Measurement ... 95
52: Scatter Diagram of Ozone Concentration Versus the Distances
to the Air Parcel Position 36 Hours Before Measurement. ... 96
53: Frequency Distributions of 12 and 36 Hour Travel
Distances 98
54: Scatter Diagram of Ozone Concentration Versus the
Direction to the Air Parcel Position 12 Hours
Before Measurement 99
55: Scatter Diagram of Ozone Concentration Versus the
Direction to the Air Parcel Position 36 Hours
Before Measurement 100
56: Frequency Distributions of "High" and
"Not High" Ozone Travel Directions 101
57: Locations of SAROAD Sites in the Eastern United States
Measuring Ozone During 1974 102
58: Example of High Ozone Concentrations Southeast of Lakes
Erie and Ontario and in the St. Louis-Ohio River Valley . . . 106
59: Example of High Ozone Concentrations in Western Kansas
and the New York-New England Area 107
60: Example of High Ozone Concentrations South of Lake
Michigan and in the New England area. 108
61: Example of High Ozone Concentrations Southeast of Lakes
Erie and Ontario and Along the Texas-Louisiana Gulf Coast • . 109
-------
LIST OF ILLUSTRATIONS (Continued)
Figure Page
62: Example of High Ozone Concentrations in Western Kansas,
The Florida Peninsula and the Washington-Philadelphia
Corridor 110
63: Counties with Average Annual NOx Emissions Greater than
75 t mi-2 yr-1 Ill
64: Prototype Pressure Pattern and the Ozone Pattern for the
Same Day 116
65: Prototype Ozone Pattern and the Weather Map for the Same
Day 117
66: Locations of Grid Points Used for Classifying Ozone and
Pressure Patterns and for Pressure-Ozone Correlations .... 119
67: Weather Map and Ozone Distribution for Sunday, 21 July 1974 . 121
68: Weather Map and Ozone Distribution for Sunday, 28 April 1974. 122
69: Weather Map and Ozone Distribution for Friday,
20 September 1974 123
70: Ozone-Pressure Relationships at McRae, Montana 125
71: Weather Map, Ozone Distribution and Trajectories for
7 July 1974 126
72: Weather Map, Ozone Distribution and Trajectories for
8 July 1974 127
73: Weather Map, Ozone Distribution and Trajectories for
10 July 1974 129
74: Weather Map, Ozone Distribution and Trajectories for
11 July 1974 130
75: Weather Map, Ozone Distribution and Trajectories for
12 July 1974 131
76: Weather Map, Ozone Distribution and Trajectories for
13 July 1974 132
xi
-------
- 12 -
LIST OF ILLUSTRATIONS (Continued)
Figure Page
77: Weather Map, Ozone Distribution and Trajectories for
14 July 1974 133
78: Weather Map, Ozone Distribution and Trajectories for
18 July 1974 134
79: Weather Map, Ozone Distribution and Trajectories for
21 July 1974 136
80: Weather Map, Ozone Distribution and Trajectories for
22 July 1974 137
81: Weather Map, Ozone Distribution and Trajectories for
26 July 1974 138
82: Areas Appropriate for Hydrocarbon Emissions Control
According to Meyer 147
xii
-------
LIST OF TABLES
Table Page
1: Frequency Distribution of Maximum Ozone
Concentrations in the Lower Troposphere as
Measured by Ozonesondes 15
2: Ozone Concentrations in the Lower Troposphere that
Exceeded 80 ppb 16
3: Correlation Coefficients Between Observed Ozone Concentrations
and Various Meteorological and Chemical Indices 72
4: Spearman Rank Correlations between Ozone and Meteorlogical
and Chemical Indices for the Combined Data for all sites. ... 78
5: Correlation between Pairs of Meteorological Indices 80
6: Correlation Between Hydrocarbon and NOx Emissions 81
7: Regression Constants Relating Ozone Concentrations to Weighted
Indices of NOx and Hydrocarbon Emissions, and to
Temperature During the Last 12 Hours 82
8: Regression Constants Relating Ozone Concentrations to
Temperature and Emissions Indices for the Last 12 Hours
of the Trajectory 85
9: Standard Error of Estimate of the Regression Equations 85
10: Frequency of Occurrence of Days when the Federal
Oxidant Standards were Violated in the Eastern
United States 104
11: Number of Cases for each Month with Daily Maximum 03 > 80 ppb
in Specified Regions of the Eastern United States 112
12: Winds Reported on Morning Weather Map in Areas
Where Peak-Hour Ozone Exceeded 80 ppb During the Day 113
13: Meteorological Features Associated with High Ozone
Concentrations 115
14: Dates Classified as Having Pressure and Ozone Patterns
Similar to the Prototypes 118
15: Frequency of Correlation Values Between Pressure and
Ozone at 20 Points in the Eastern United States 120
xiii
-------
ACKNOWLEDGEMENT S
We are indebted to Mssrs. E. L. Meyer, E. L. Martinez, W. P. Freas
and H. R. Richter of EPA's Air Management Technology Branch for their
valuable comments. D. H. Barrett and J. A. Tikvart of the Source Recep-
tor Analysis Branch have also provided many valuable suggestions.
We are especially indebted to the Project Officer, Mr. Philip L.
Youngblood, for his many useful suggestions during the course of the pro-
ject and for his valuable comments concerning the final report.
At SRI the following people assisted in the analyses of data and
the preparation of the reports: J.H.S. Kealoha, A.H. Smith, L.J. Salas,
R. Trudeau, L. Jones, W. Ligon, R. Troche, and S. Gillen. Mr. Dale
Coventry of EPA and Mr. R. Haws of Research Triangle Institute provided
many useful data, as have the staff of the National Climatic Center.
xv
-------
CONCLUSIONS
General
The analyses presented here, when combined with other evidence from
the literature, provide a reasonably clear picture of the importance of
stratospheric and tropospheric processes in determining ground level
ozone concentrations. The findings can be grouped into the following
five categories.
1. The importance of stratospheric ozone sources
2. The importance of anthropogenic emissions to ozone concentra-
tions in rural areas
3. The identification of geographical and meteorological factors
conducive to the occurrence of high ozone concentrations in
rural areas
4. The likelihood of tropospheric/stratospheric interactions to
produce high ozone concentrations
5. The feasibility of oxidant control strategies that can be ap-
plied over extended geographical areas.
A discussion follows the conclusions under each of these ca-
tegories:
The Importance of Stratospheric Ozone
There is strong evidence for the following conclusions:
• Stratospheric contributions are greatest during late winter and
spring
* During these seasons, the long term average stratospheric con-
tribution to the tropospheric ozone burden amounts to about 20
to 25 ppb
• The greatest contributions occur near the mean positions of the
polar and arctic jet streams and perhaps in the lee of the Rock-
ies
* Ozone concentrations from stratospheric intrusions exceed the
federal standard only about 0.2 percent of the time in the lower
troposphere, and even less frequently at ground level.
• Ground level hour-average ozone concentrations in excess of 150
ppb can be caused by stratospheric intrusions that have been ob-
served under special circumstances.
-------
The studies described here have identified at least some of the
special circumstances that can lead to very srong stratospheric influ-
ences at ground level. Subsidence in the stratosphere, coupled with
tropopause folding can introduce large amounts of ozone into the tropo-
sphere which can be brought to the ground with minimal dilution by
strong convection in the lower troposphere. This is an uncommon combi-
nation of events. It is uncertain whether other combinations of events
also transfer ozone at high concentration from the stratosphere to
ground level. It is almost certain that such transfers are quite rare
and probably of limited duration (a few hours) and spatial extent
(several tens of kilometers).
The Importance of Anthropogenic Emissions
Meyer's (1977) analysis indicates that anthropogenic emissions,
especially hydrocarbons, are important determinants of ozone concentra-
tions for distances of up to 125 km from very large cities and to short-
er distances from smaller cities. This is quite consistent with ob-
servations of urban "plumes" at similar ranges. The importance of hy-
drocarbons is consistent with smog chamber results that show ozone con-
centrations to be a function of hydrocarbon concentrations when
hydrocarbon/NOx ratios are low, such as they usually are in and around
cities.
The results of the present study suggest that ozone concentrations
in rural areas are more dependent on NOx emissions than on hydrocar-
bon emissions. This too is consistent with smog chamber studies that
show ozone dependence on NOx very high at hydrocarbon/NOx ratios such
as are common in nonurban areas. The findings also agree with similar
results obtained by Meyer et al (1976) and by Singh, Ludwig and Johnson
(1977).
Meteorological and Geographical Factors Associated wih High Ozone Con-
centrations in Rural Areas
The one factor most highly correlated with the maximum daily,
hour-average ozone concentrations in rural areas is the air temperature
during the last 12 hours before the observation. It should be noted
that air temperature is closely related to other factors that are also
likely to be instrumental in the photochemical formation of ozone.
Solar radiation is probably the most important of these other factors.
High ozone concentrations are often, but not always, associated
with light winds and recent trajectories that are characterized by a
clockwise curvature. Both of these conditions are conducive to in-
creased buildup of precursor concentrations. Light winds encourage this
buildup because less air is available to dilute the emissions. Clock-
wise, or anticyclonic, motion is generally associated with subsidence
and a resulting atmospheric stratification which inhibits vertical mix-
ing and dilution. Subsidence also tends to suppress cloudiness, thereby
allowing for more solar radiation to drive the photochemical processes.
If temperatures are high, ozone formation is even further enhanced.
2
-------
This combination of conditions is frequently found near and to the west
of high pressure centers in the summertime.
Light winds and strong atmospheric stability are not essential to
the accumulation of precursors and photochemical production of ozone.
Extended travel over large emissions areas can also lead to substantial
concentrations of ozone if the air is warm and there is sufficient
sunshine. Several extended areas of high emissions are oriented gen-
erally southwest-northeast in the eastern United States. This confi-
guration undoubtedly contributes to the fact that a large portion of the
violations of federal ozone standards are accompanied by southwest
winds. Air movement from the southwest in these elongated regions would
allow the accumulation of precursors necessary for ozone formation. It
happens that southwest-toward-northeast is the generally prevailing air
movement over the eastern United States during the summer months and it
also happens that the occurrence of the meteorological conditions iden-
tified as being conducive to ozone formation often accompany such air
motions.
Stratospheric/Tropospheric Interactions
It is important to understand that significant stratospheric intru-
sions are unlikely to occur concurrently with large tropospheric build-
ups of anthropogenic ozone. Stratospheric intrusion is most likely in
the late winter or spring when the sunshine necessary for photochemical
processes in the troposphere is relative weak. Stratospheric intrusions
are most likely to occur behind cold fronts where rain, cool tempera-
tures, high winds, and strong convection are likely to work against
the buildup of anthropogenic ozone from photochemical activity. Thus it
does not appear that these two sources of ozone (i.e., stratospheric in-
trusion, and photochemical formation from anthropogenic precursor emis-
sions) interact to the extent that control strategies for anthropogenic
ozone cannot be formulated.
Implications for Control Strategies
Although the frequency of violations of the federal ozone standard
caused by stratospheric intrusion appears to be small, the reality of
such events cannot be ignored. It is important that such uncontrollable
incidents be recognized so that the strategies will focus on those in-
cidents that are the result of controllable processes.
In addition to the unusual instances, the design of control stra-
tegies should also recognize that there exists an uncontrollable reser-
voir of "background" ozone arising in large part from an ongoing inter-
change between the stratosphere and the troposphere. In the spring this
reservoir may have concentrations of about one-half the federal stan-
dard (Singh, Ludwig and Johnson, 1977).
It appears that control strategies uniformly applied within res-
tricted areas are feasible for meeting the federal oxidant standard. As
Meyer (1977) shows, the major effects of urban areas are most often
3
-------
realized within some resonable distance of the city. That distance is
generally less than 125 km. Meyer's (1977) approach of defining hydro-
carbon control regions by a circle centered on the city might be re-
fined by introducing asymmetry. For instance, if certain wind direc-
tions are disproportionately associated with ozone incidents, the boun-
daries for control might be extended in those directions. Nevertheless,
the concept of a radius of influence seems useful in the design of gen-
eralized control strategies.
At greater distances from cities the control of ozone appears to
require the control of NOx emissions. It also appears that high ozone
concentrations in rural areas are influenced by emissions at greater
distances and over greater periods of time than is usually the case with
the high concentrations found near cities. Thus, the control of ozone
in rural areas poses much more of a problem than that in or near urban
areas. The regions to be considered are apt to be quite large so that
the definition of source-receptor relationships is likely to be quite
difficult. It may not be possible to identify the area where controls
should be applied in order to lower the ozone concentrations at a speci-
fied rural area. Furthermore, the strategies for controlling rural
ozone are apt to involve a large-scale curtailment of NOx emissions.
This could be a difficult feat. Finally, there is the distinct possi-
bility that the control of NOx emissions to reduce ozone concentrations
in rural areas could lead to a worsening of the ozone problem in and
near urban areas.
-------
1. INTRODUCTION
1.1. Motivation and Objectives
The United States has established ambient air quality standards
(AAQS) for several toxic pollutants, including photochemical oxidant. *
The rationale underlying these standards is that they should be met in
order to protect public health. The concept of the AAQS has been ex-
tended to include the formulation of plans for the control of emissions
so that the standards are met. These plans usually involve the curtail-
ment or rearrangement of some human activities, tacitly assuming that
human activities are responsible for the observed violations of the
AAQS. This thesis is hard to dispute for most pollutants in urban
areas, but there are frequent occasions when the oxidant standard of 80
ppb (measured as ozone) is violated in rural areas well removed from ur-
ban centers. Questions arise concerning the importance of natural
processes, such as the intrusion of ozone-rich stratospheric air to
ground level, and the importance of long-range transport.
The answer to the question of long-range transport directly af-
fects the formulation of strategies to prevent the violation of the oxi-
dant standard. If emissions affect oxidant formation over a wide area,
then the control strategies may well have to involve more than the im-
mediate surroundings where the violations occur. It becomes important,
therefore, to define the appropriate domain for the application of oxi-
dant control plans. In an effort to resolve these dilemmas, the
research described here seeks to answer the following specific ques-
t ions:
• What causes the high oxidant values that are frequently observed
in areas well removed from major sources of anthropogenic pre-
cursor emissions?
• How much does ozone of stratospheric origin contribute to
ground-level oxidant concentrations?
• Over what distances is oxidant traceable to upwind precursor em-
issions?
• What are the effects of synoptic and smaller-scale meteorologi-
cal variables on ground-level oxidant formation and transport?
• Is it possible, on the basis of relationships among oxidant con-
centrations, emission patterns, and synoptic-scale meteorologi-
cal conditions, to identify geographic regions for uniform oxi-
dant control strategies? If so, how?
The objective of the whole program is then to use the answers to
these questions to formulate emission control strategies that are con-
sistent with the physical processes governing the observed ozone concen-
* Throughout this report the terms "oxidant" and "ozone" are used
interchangeably except where the context of the discussion dictates
otherwise. c
-------
trations. Specifically, EPA hopes to determine if it is feasible to de-
fine geographic areas in which precursor control strategies can be ap-
plied to lower oxidant concentrations to acceptable levels.
1.2. Lines of Investigation
The approach to the problems outlined above was essentially
descriptive. Two possible mechanisms that may lead to high oxidant
concentration areas were examined:
• Intrusion of ozone from the ozone-rich stratosphere to ground
level
• The formation and transport of oxidant in the lower atmosphere
under meteorologically favorable conditions.
The intrusion of ozone is difficult to verify because it is not
possible to distinguish stratospheric ozone from anthropogenic ozone un-
less other trace constituents of either stratospheric or troposhperic
origin are examined at the same time. Such concurrent observations,
however, are rare. Therefore, it was necessary to examine the behavior
of a surrogate for stratospheric ozone. To this end the behavior of ra-
dioactive debris injected into the stratosphere during nuclear weapons
testing was studied. The transfer of this debris from the stratosphere
to the troposphere, and downward to ground level, was used to trace the
transfer processes that would move stratospheric ozone over the same
route. Another approach was to identify a few selected cases of high
ozone concentrations at low altitudes having conditions of dryness and
thermal stratification that suggested possible stratospheric origins for
the air. These cases were then carefully analyzed to determine whether
the observed ozone concentrations had actually come from the strato-
sphere.
The importance of photochemical and transport processes in the
lower atmosphere was first investigated through a detailed examination
of the history of the air arriving at several different locations.
Where the air had been during the preceding days, what meteorological
conditions prevailed, and what anthropogenic pollutants had been intro-
duced were determined and used as a basis for interpreting the observed
ozone concentration at the end of the period studied. This "Lagrangian"
approach of examining the effects of events along the trajectory on the
resulting ozone concentrations provides focus on processes and influ-
ences affecting ozone which may be generally applicable in all geograph-
ical areas. Thus, more than one location can be compared on a common
basis. In addition to providing a means of identifying the factors that
are important to ozone production, this approach also helps to identify
what might be called the "historical span of influence", within which it
is hypothesized that an existing ozone concentration is influenced ap-
preciably by the past emissions and meteorological history of the air.
Events in the more distant past are probably less influential than simi-
lar events that occurred more recently. The concept of historical span
of influence can be translated into one of spatial span of influence by
6
-------
examining the distances associated with the time intervals over which
the processes are effective.
A second approach to the question of tropospheric photochemistry
and transport was to study the relation between ozone patterns and
synoptic scale weather patterns. The trajectory and meteorological his-
tory of the air are the direct products of synoptic scale meteorological
patterns. So it was natural to examine weather and ozone patterns in
concert. This is largely the Eulerian equivalent of the trajectory ana-
lyses. Although the two facets of investigation were conducted sequen-
tially, they are closely related. In presenting the results there is,
of necessity, some separation between the discussion of these two facets
of the tropospheric work. Nevertheless, it should be remembered that
the trajectory analysis and the synoptic map comparisons are simply
Lagrangian and Eulerian attempts to answer the same general questions
concerning the process of ozone formation and transport in the tropo-
sphere.
1.3. Brief Description of Methods
1.3.1. General
The details of the methods used in the data analysis are given in
the appendices to this report, included in Volume III. These detailed
descriptions should be consulted by readers who would provide their own
interpretations to the results, or who might wish to apply the methods
to other data sets. However, proper understanding of the results re-
quires at least a brief description of the methods. This section is in-
tended to provide that understanding.
As noted above, the research was divided into three parts:
(1) Stratospheric intrusion analyses
(2) Tropospheric transport studies
(3) Studies of synoptic-scale ozone distributions and weather pat-
terns.
The purpose of the stratospheric intrusion analyses was to evaluate
the degree to which the stratosphere affects ground-level ozone concen-
trations. The objective of the tropospheric transport studies was to
examine relationships between ozone concentrations and the emissions and
meteorological conditions to which the air had been subjected. The
synoptic analyses identified patterns of ozone distribution in the
eastern United States that could be related to the emissions and
meteorological features in the same area. Each part of the research had
its own requirements. These are described briefly in the following sub-
sections.
-------
1.3.2. Stratospheric Intrusion Analyses
The determination of the magnitude of ozone intrusion from the
stratosphere was based largely on the study of other identifiable stra-
tospheric constituents, especially radioactive debris from nuclear tests
or cosmic radiation processes. These serve as tracers of stratospheric
air, and hence of stratospheric ozone. The use of the radioactive
debris data as a surrogate for stratospheric ozone allows a tracing of
the processes that transfer ozone from the stratosphere to ground level.
Case studies of specific ozone observations were also undertaken to es-
timate the frequency with which stratospheric ozone might be brought to
the lower troposphere at high concentration as a result of the intense
meteorological processes that have been proposed in the literature (e.g.
Danielsen and Mohnen, 1976).
1.3.3. Studies of Tropospheric Transport Processes
In studying the importance of tropospheric formation and transport
of ozone, the meteorological and emissions history of ozone-containing
air was examined in detail by means of trajectory analysis. This part
of the research was undertaken in six steps:
• Selection of suitable sites for study
• Selection of suitable cases for study at each site
• Construction of the trajectory for each case
• Specification of meteorological conditions prevailing along each
trajectory
• Specification of the amounts of ozone precursors introduced into
the air along each trajectory
• Statistical analysis of relationships between the ozone concen-
tration and the preceding meteorological and emissions condi-
tions.
Four sites were selected for study after the review of data sources
described in Volume II of this report. The review identified 38 candi-
date sites, but many were unsuitable for reasons such as insufficient
data or proximity to urban areas. Figure 1 shows the locations of the
38 sites originally considered and the four finally selected. Those
selected were McHenry, Maryland; Queeny, Missouri; Wooster, Ohio; and
Yellowstone Lake, Wisconsin. These sites are circled on Figure 1. They
provide widespread geographical coverage of the northeastern quarter of
the United States, a fact that influenced their selection.
-------
(O
LU
OC
I
I
cc
LL
O
CO
LU
UJ
o
tr
O
LL
ta
UJ
c/5
o
UJ
LU
CO
a
o
<
o
LU
OC
-------
Thirty cases from each location were chosen in such a way that a
broad spectrum of ozone concentrations would be represented, but with an
emphasis on the higher concentrations. Half the cases represent days on
which the peak one-hour concentrations are in the top 20 percent of all
days in the period considered (June - October, 1974). The remainder are
among the lowest 20 percent or are days with peak-hour ozone concentra-
tions near the median (between the 40th and 60th percentiles). Within
these constraints the cases were selected to represent a variety of
synoptic meteorological conditions, hours of the day when the maximum
ozone concentration occurred, and days of the week. The selection of
the cases is described in detail in Appendix A (Volume III).
The next step in the research was to construct air trajectories for
the 60 hours preceding each of the 120 selected ozone observations.
This was done using a version of Heffter and Taylor's (1975) trajectory
model. The trajectory determinations were based on observed winds in
the layer between 300 m and the average height (generally, 1500-2000 m)
of the summer afternoon mixing layer near the selected sites (as report-
ed by Holzworth, 1972). The model determined the location of the air at
3-hour intervals.
Once the trajectories had been calculated, the meteorological con-
ditions prevailing during each 3-hour segment were determined from con-
current National Weather Service weather maps. Specifically, the fol-
lowing parameters were obtained from the weather maps:
* Temperature
• Dew point
• Cloud cover
• Current and recent precipitation (rain, snow, etc.)
• Cloud types
• Wind
Some of these parameters were used to derive indices of conditions
that are more directly relatable to ozone formation and dilution
processes than are the primary data. Specifically, cloud conditions were
used in combination with solar elevation to provide a measure of incom-
ing solar radiation (insolation). This insolation index was also used,
along with surface wind speed, to determine atmospheric stability. Tem-
perature and dewpoint were used to calculate relative humidity.
Hydrocarbons (HC) and oxides of nitrogen (NOx) are known to be im-
portant to the formation of ozone. Therefore, emissions of these two
pollutants were determined for the trajectory segments. Estimates of
annual-average countywide emission rates of HC and NOx were obtained for
each U.S. county from the National Emissions Data System (NEDS). These
data, corrected for diurnal and seasonal changes, were used to determine
the HC and NOx emitted into each trajectory segment.
10
-------
In summary, indices of those emissions and surface meteorological
parameters thought to affect the eventual ozone concentration were
determined for each 3-hour segment of each of the 120 trajectories that
were examined. The indices represented the following variables:
• NOx emissions
• HC emissions
• Insolation
* Temperature
* Dew point
• Relative humidity
* Precipitation
• Atmospheric stability
To reduce the number of indices that had to be treated in subsequent
statistical analyses, the 3-hour values for the indices were averaged
over 12-hour periods. Thus, there were five indices for each of the
above variables for each of the 60-hour trajectories.
The statistical analyses are described in more detail in subsequent
sections. In general, this study was limited to standard techniques in-
volving single and multiple linear regression and correlations. Simple
factor analysis was also employed.
1.3.4. Synoptic Scale Ozone Distributions and Weather Patterns
The influence of meteorological factors on ozone formation was stu-
died from still one more point of view. Over the eastern two-thirds of
the U.S. the large scale patterns of ozone were examined to determine
what relation they bore to concurrent synoptic weather patterns. The
weather patterns for the study were taken directly from the National
Weather Service analyses published in the "Daily Weather Map" series.
The basic data for the ozone analyses came from the SAROAD (Storage and
Retrieval of Aerometric Data) files of EPA. The highest hourly average
ozone concentration was determined for each station for each day on
which reasonably complete data were available. Isopleth maps of
maximum hour-average ozone were then constructed for each day. The
method for drawing the isopleths is explained in detail in Appendix D
(Volume III).
This part of the study treated those states east of, or traversed
by, 100 degrees W. Meridian. Ozone observing sites are distributed rea-
sonably well so that the large scale patterns could be derived. It is
also an area where meteorological systems are more clearly definable
than they are in the more mountainous western states. The ozone pat-
terns were used to identify those parts of the eastern United States
where ozone concentrations most frequently exceeded the federal standard
11
-------
during the study year 1974. The various weather patterns most conducive
to ozone formation were also identified. Pattern classification
methods were applied for those days when ozone concentrations exceeded
the federal standard over an appreciable area in the eastern U.S.
12
-------
2. ANALYSIS AND RESULTS
2.1. Stratospheric Sources
Two approaches have been used to examine the contributions of stra-
tospheric ozone to the concentrations observed at the surface. The two
approaches reflect the two different time scales of the stratospheric
contribution: 1) the average, long-term contributions to the ground-
level ozone burden, and 2) the short-term impact of individual intru-
sions of stratospheric ozone to ground level.
As discussed in Volume II of this report, the following mechanisms
are known to be important in the exchange of mass between the strato-
sphere and the troposphere:
• Seasonal adjustment of tropopause level
* Mean meridional circulation
* Stratospheric exchange between hemispheres
* Small-scale eddies.
• Large-scale eddies
The first four of the above transport processes proceed rather slowly
and allow enough time for turbulent dilution of ozone in the tropo-
sphere. Thus, these four mechanisms relate to the average, long-term
contribution. On the other hand, the large-scale eddy phenomenon is a
mechanism of potentially high, short-term impact of stratospheric ozone
on ground-level concentration.
The transport of radioactiive debris from the stratosphere to the
troposphere was studied in detail during the years of atmospheric nu-
clear testing and for some time after the test-ban treaty went into ef-
fect in 1963 (for summaries of research findings see Reiter, 1972;
1976a). These studies have provided a valuable basis for understanding
the processes affecting both the short-term and the long term inter-
change of air between the stratosphere and the troposphere. Also, ex-
tensive ozonesonde measurements were made over North America during the
early 1960's. These data were analyzed to estimate the impact of stra-
tospheric intrusion on short-term ozone concentrations. These two ap-
proaches have proved quite useful in estimating the impacts of stratos-
pheric ozone. The results are discussed in the following sections.
2.1.1. Ozonesonde Case Studies
2.1.1.1. General
Studies were made of ozonesconde observations which had been taken
at 13 stations on the North American continent for the period between
December 1962 and December 1965 (Hering, 1964; Hering and Borden, 1964;
1965a; 1965b; and 1967). Hering (1964) has alluded to possible incon-
sistency in the data due to problems with instrument sensitivity, cali-
13
-------
bration, and stability. However, the data were assumed to be correct
for purposes of this analysis. From a total of 1477 soundings the max-
imum ozone mixing ratios encountered below the 800-mb surface (below the
750-mb surface for the stations at Albuquerque, New Mexico and Fort Col-
lins, Colorado) were determined (800 tab corresponds roughly to 7000 feet
above mean sea level, 750 mb to 8500 feet). Table 1 presents a frequen-
cy distribution of these maximum mixing ratios.
As shown in Table 1 , concentrations of 80 ppb (federal oxi-
dant standard for one-hour average) were exceeded in this data sample 25
times; i.e., in less than 2% of all sounding ascents. Table 2 summar-
izes the dates and places of these occurrences. In most of these cases,
the high ozone concentrations were associated with high relative
humidities—a clear indication that the air masses in which these ozone
concentrations were observed came from the troposphere rather than from
the dry stratosphere. Also, subsidence of stratospheric air into the
troposphere results in a temperature inversion in that air. Thus, stra-
tospheric intrusions will be characterized by dry, stable layers.
Numerous cases analyzed by Reiter and other authors (for references see
Reiter, 1972) show that stable layers of air of stratospheric origin em-
bedded in the lower troposphere are so dry that the radiosonde humidity
sensor does not respond.
Using low relative humidities and the presence of a stable layer as
criteria for the possible involvement of air masses of stratospheric
origin, all but the following soundings were excluded:
Goose Bay, Canada 2 May, 1963
Bedford, Massachusetts 26 June, 1963
Tallahassee, Florida 14 August, 1963
Seattle, Washington 8 April, 1964
Seattle, Washington 15 April, 1964
Many of the soundings that did not meet the selection criteria were
characterized by major ozone maxima very close to the ground. Such
vertical gradients of ozone are characteristic of a ground-based, rather
than a stratospheric, ozone source. The ground-based source is most
likely anthropogenic. A discussion follows of the five cases which were
selected.
2.1.1.2. Goose Bay, Canada, 2 May 1963
Figure 2 shows the ozone partial pressures and mixing ratios as
measured by the ozonesonde ascent at Goose Bay on May 2 (1117 GMT). The
Weather Bureau radiosonde ascent, plotted on a tephigram, is reproduced
in Figure 3. A comparison of the two figures reveals that the highest
ozone concentration in the lower troposphere appears to be centered near
14
-------
CO
H
c
IH >
41 L.
E? *
* i
M i.
0 L
S
3
CO
CM
cn
CM
i-t
g
^
r*
CO
0
I-t
CM
O
O*>
i;
s
00
CN
r-
2
2
in
00
CM
cn
o
ncentration
ppb
Station
uO
cn
-
o
CO
*
cn
CM
.
Thule,
Greenland
K
CM
«
»
-
-
*
"•*
o
-
Fairbanks ,
Alaska
s
CM
-
CM
-
2
2
CM
—*
2
0
-
Churchill,
Canada
S
fM
-
„
-
CO
m
O
CM
tM
S
-
«
X
CQ
OO
,H
^
-
-
CM
-
cn
•*
U1
2
cn
CM
CO
CM
CM
Seattle,
Washington
cc
CM
-
«
vO
CO
-
in
~
m
»
Green Bay,
Canada
s
CM
cn
-
-
i-i
cn
«
00
2
2
00
CM
cn
m
cn
3
CM
9)
Bedford ,
M.issaehusett
5
-
N
-
CM
«
00
CM
CM
in
cn
CS
Ft. Collins,
Colorado
cc
-
CM
•
-1
M
-
-
-
cn
-
£ o
~r
-
cn
en
cn
fs.
CO
CM
CO
CM
01
cn
in
*n
CM
Albuquerque ,
Sew Xc-xico
-
^
-
CM
-
cn
-
-
r-
o
o
CM
2
i-t
""
CM
CO
Tal lahassee,
Florida
S
*
CM
-
*
2
n
1-4
m
Grand Turk 1
Bah an as
»
-
-
CM
•*
cn
-
r-
i-i
CM
CM
cn
V
c
o
(TJ
-H C
m «
03 U
rv O
"
-
CM
^
( :
•*
i-l
tn
cn
«
cn
>0
s -
N
r* cn
cn
oo m
m oo
CM
O O
in f-i
oo
Total Cases
Percentage
15
-------
Table 2
OZONE CONCENTRATIONS IN THE LOWER TROPOSPHERE
THAT EXCEEDED 80 PPb
Maximum Ozone
Concentration
Date (ppb) Station
5-2-63
6-26-63
7-3-63
8-7-63
8-14-63
9-11-63
9-11-63
9-18-63
1-20-64
4-8-64
4-15-64
4-17-64
4-20-64
7-15-64
7-15-64
8-26-64
1-22-65
7-14-65
10-6-65
10-7-65
10-8-65
11-3-65
11-10-65
12-1-65
12-1-65
90
114
102
90
114
102
102
90
102
108
96
120
120
102
90
114
90
96
114
144
144
84
150
84
84
Goose Bay
Bedford
Bedford
Bedford
Tallahassee
Tallahassee
Seattle
Tallahassee
Bedford
Seattle
Seattle
Seattle
Seattle
Bedford
Goose Bay
Bedford
Tallahassee
Bedford
Pt. Mugu
Ft. Mugu
Pt. Mugu
Pt. Mugu
Pt. Mugu
Pt . Mugu
Tallahassee
16
-------
-25
1000
50 100 150 200
PARTIAL PRESSURE OF OZONE
250
300
FIGURE 2 OZONE SOUNDING AT GOOSE BAY, CANADA, 2 MAY 1963
17
-------
OJ
•a
1, •
2
350
340
330
320
310
300
290
280
270
260
-50 -40 -30
AIR TEMPERATURE
-20 -10
degrees (C)
FIGURE 3 RADIOSONDE ASCENT AT GOOSE BAY, CANADA, 2 MAY 1963, 0000 GMT
(Numbers along the sounding curve indicate relative humidities at significant points)
18
-------
the 282°K isentropic surface * within a stable layer with a low rela-
tive humidity of 28%.
The sequence of surface weather maps and 500 mb (about 18,000 feet
above sea level) charts in Figure 4 reveals the formation of an inten-
sive outbreak of arctic air over Hudson Bay on 28, 2y, and 30 April.
The outbreak is associated with a deep low pressure center over Baffin
Island. The leading edge of this cold-air is marked by the trough axis
on the 500-mb surfaces for those dates (Figure 4). The frontal system
delineating the cold outbreak had slowed down considerably by 1 May.
The ozone sounding taken at Goose Bay on 2 May was launched directly
into this frontal zone.
To assess the possibility that stratospheric air was responsible
for the relatively high ozone concentrations, isentropic trajectories
were constructed on the 282°K surface, starting over Goose Bay on 2 May
at 1200 GMT, and running backward in time. (The procedure used in com-
puting isentropic stream functions is outlined in Appendix C (Volume
III). The Montgomery stream function analyses between 30 April, 0000
GMT, and 2 May, 1200 GMT surface are shown in Figure 5. The temperature
of the isentropic surface used for the analysis was changed with time to
reflect the radiational cooling of the descending air mass. For air un-
dergoing rapid descent from the tropopause, such cooling is usually
about l°C/day. The stream function diagrams in Figure 5 show 12-hour
trajectory segments that are centered on the time indicated. The com-
posite trajectory is reproduced in Figure 6.
In the course of 2-1/2 days the air mass containing high ozone con-
centrations had moved from Baffin Island across Hudson Bay to Goose Bay,
descending from the 600-mb to the 820-nb level. The sparsity of upper-
air synoptic observations made it impractical to trace the air trajecto-
ry further back in time. It should be noted, however, that tropopause
heights inside the deep cyclone over Baffin Island were very low. As
shown by the 1 May sounding for Baker Lake (reproduced in Figure 7), the
tropopause was near 560-iab at a potential temperature of 281°K. Thule,
Greenland, also reported high ozone concentrations below the 500-mb sur-
face on 1 May at 1117 GMT. Thule was also influenced by the Baffin Is-
land low. The high ozone concentrations observed there, and shown in
Figure 8, can be taken as characteristic of the lower stratosphere in
polar latitudes.
Thus, even though the trajectory construction could not be complet-
ed for lack of upper air data, all evidence indicates that the high
ozone concentrations observed in the planetary boundary layer over Goose
Bay on 2 May 1963 were of stratospheric origin.
* An isentropic surface is one of constant potential temperature.
Air will move along such surfaces in the absence of diabatic processes.
Since most processes occurring in subsiding air are nearly adiabatic,
tracing air along these surfaces provides a reliable method for deter-
mining 3-dimensional motions.
19
-------
SURFACE MAP. 28 APRIL 1963, 1800 GMT
SURFACE MAP, 29 APRIL 1963. 1800 GMT
500 MB MAP. 29 APRIL 1963, 0000 GMT
500 MB MAP, 30 APRIL 1963, 0000 GMT
FIGURE 4 NORTH AMERICAN SURFACE AND 500MB CHARTS, 28 APRIL TO 3 MAY 1963
20
-------
SURFACE MAP, 30 APRIL 1963, 1800 GMT
SURFACE MAP, 1 MAY 1963, 1800 GMT
500 MB MAP, 1 MAY 1963, 0000 GMT
500 MB MAP, 2 MAY 1963, 0000 GMT
FIGURE 4 NORTH AMERICAN SURFACE AND 500MB CHARTS, 28 APRIL TO 3 MAY 1963(Continued)
21
-------
co
8
>
5
CO
0."
s
ffl
5
T3
O
(J
n
(T
CL
00
£
QC
O
03
O
<
CM
O."
u.
DC
Q
Z
ill
O
<
LL
cc
CO
1
x
DC
O
LLI
OC
22
-------
(a)
04-30-63,00 GMT
285 °K
(0
2840
(e)
05-02-63.00 GMT
282' K '
2864
2882
05-01-63, 12 GMT
282 "K
(f)
05-02-63,12 6MT-S
282 'K
\
FIGURE 5 MONTGOMERY STREAM FUNCTION ANALYSES (106cm25-2)30 APRIL-2 MAY 1963
23
-------
FIGURE 6. COMPOSITE 6-HOUR TRAJECTORY SEGMENTS FROM 29 APRIL 1963, 1800 GMT,
TO 2 MAY 1963,1200 GMT (Pressures in millibars are indicated next to dots corres-
ponding to synoptic observation times.)
24
-------
0>
0)
!H
(tf
330
320
310
300
290
280
270
260
250
-60
A 2;:
A 21
-50 -40
TEMPERATURE
-10
FIGURE 7 RADIOSONDE ASCENT AT BAKER LAKE, CANADA, 1 MAY 1963, 1200 GMT
(Numbers along the sounding curve indicate relative humidities at significant points.
'A' signifies that ambient humidity less than the value shown rendered the sensor
inoperative.)
25
-------
-25
-20
1000
50 100 150 200
PARTIAL PRESSURE OF OZONE
250 300
FIGURE 8 OZONE SOUNDING AT THULE, GREENLAND, 1 MAY 1963
26
-------
2.1.1.3. Bedford, Massachusetts, 26 June 1963
The weather conditions prevailing in this area are shown in Figure
9. The anticyclonic conditions and absence of frontal activity over the
northeastern United States are inconsistent with ozone transport from
the stratosphere. The confinement of the high ozone concentrations to
the lowest level, as shown in Figure 10, also suggests low tropospheric
origins. The low humidities that caused this case to be included among
those when ozone from the stratosphere was suspected are probably the
product of offshore winds and diurnal warming. Thus, stratospheric in-
trusion was ruled out in this case.
2.1.1.4. Tallahassee, Florida, 14 August 1963.
As can be seen in Figure 11, a cold front passed Tallahassee short-
ly after 0600 GMT on 15 August 1963. The Tallahassee ozone sounding
shown in Figure 12 measured exceptionally high ozone concentrations in
the lower troposphere around 1212 GMT on 14 August. The temperature
sounding in Figure 13 shows stable layer that corresponds with the high
ozone concentrations. Very low relative humidities observed between
about 700 and 850-mb suggest the presence of pronounced subsidence. The
surface weather map of 14 August at 0000 GMT (not depicted here) showed
a squall line northwest of Tallahassee with thunderstorm and cumu-
lonimbus development. Subsidence ahead of this squall line could con-
ceivably have carried air from the stratosphere toward the ground. To
test this possibility, the isentropic trajectories shown in Figure 14
were constructed backward in time, starting with 14 August at 1200 GMT,
on the 304°K isentropic surface over Tallahassee. This surface was
chosen because the nearest upwind sounding (Montgomery, Alabama), for
that time revealed a frontal inversion at 304°K potential temperature
(Figure 15).
It is important to remember that active cyclogenesis had been asso-
ciated with the front in question a few days earlier (11 August). Mahl-
man and Reiter (for references see Reiter, 1972) have identified periods
of strong cyclogenesis as prerequisites for the transport of stratos-
pheric air into the troposphere and toward the ground. Usually such
stratospheric air is found in stable layers above the planetary boundary
layer, especially in the anticyclonic area developing behind an advanc-
ing cold front. In the present case it appears likely that the squall-
line development ahead of the front disrupted the normal course of
events, allowing some of the stratospheric air to intrude ahead of the
cold front. It is somewhat analogous the occasional transport of stra-
tospheric ozone to the ground by the action of strong waves in the lee
of the Rocky Mountains (Reiter, 1975).
It is reasonable to assume that the major transport of stratospher-
ic air was associated with the frontal inversion centered on the 304°K
isentropic surface on 12 August, at 1200 GMT. The Montgomery stream
function and trajectory analyses presented in Figure 14 show that the
trajectory ending at Tallahassee on 14 August at 1200 GMT had come from
the northwest and was subjected to strong subsidence during its course.
27
-------
1-
O
111
z
LU
o
<
u.
DC
CO
CO
in
CM
QC
UJ
LLI
I-
Z
D
M
oT
5
LLI
O
CC
CC
LU
S
<
X
cc
o
0>
LU
OC
D
a .,
28
-------
10
P
R
E 20
S
S
U
R
50
(mb)
100
200
500
1000
^35
I*
-25
-20
-15
-5
50 100 150 200 250
PARTIAL PRESSURE OF OZONE (>umb)
300
FIGURE 10 OZONE SOUNDING AT BEDFORD, MASSACHUSETTS, 26 JUNE 1963
29
-------
SURFACE MAP, 10 AUGUST 1963, 1800 GMT
SURFACE MAP, 11 AUGUST 1963, 1800 GMT
SURFACE MAP, 12 AUGUST 1963. 1800 GMT
SURFACE MAP, 13 AUGUST 1963, 1800 GMT
FIGURE 11 NORTH AMERICAN WEATHER MAPS, 10 AUGUST TO 15 AUGUST 1963
30
-------
• W^-i-'. 2^S ,/,
rasraMsfi'-Y. V ^~~^ty^-^L^__ v-ikxsr^
""*'" E"-V tow ^- .' ,/ ' • ~~*&> "*
SURFACE MAP. 14 AUGUST 1963. 0600 GMT
SURFACE MAP, 15 AUGUST 1963, 0600 GMT
SURFACE MAP, 14 AUGUST 1963, 1800 GMT
FIGURE 11 NORTH AMERICAN WEATHER MAPS, 10 AUGUST TO 15 AUGUST 1963 (Concluded)
31
-------
(mb)
15
500
1000
10
~ 5
50 100 150 200 250
PARTIAL PRESSURE OF OZONE (,umb)
300
FIGURE 12 OZONE SOUNDING AT TALLAHASSEE,
FLORIDA. 14 AUGUST 1963
32
-------
35-
30-/
P
R
E
S
S
U
R
E
(mb)
200
500
-60
-40 -20 0
TEMPERATURE CO
20
40
1000
FIGURE 13 TEMPERATURE AND DEWPOINT SOUNDINGS AT
TALLAHASSEE, FLORIDA, 14 AUGUST 1963
33
-------
3094
08-12-63, 00 GMT
308 • K ''
(b)
3102
08-12-63,12 GM
308°K ' .. - ..
(0
(d)
08-13-63, 00 GMT
305° K
08-13-63,12 GMT
305 °K
(e)
(f)
08
304 "K
j
\ SS7
-14-63, I2GMT0 s
FIGURE 14 MONTGOMERY STREAM FUNCTION ANALYSES (108cmV2), 12 AUGUST TO
14 AUGUST 1963 (See Figure 18 for meaning of symbols)
34
-------
o
m
co
\
o
CO
O
'55
- .
I- 5
£
§
•* .2
" 13
< 1
^ 0>
o ^
O c
O To
| e
ll
in
UJ
cc
D
O
(51)
35
-------
The leading edge of the stratospheric air intrusion, according to
these analyses, coincides with the edge of a tongue of dry air. The
dryness of the air intrusion is documented by the soundings from Omaha,
Nebraska, and Columbia, Missouri, on 13 August at 0000 GMT (Figure 16 ).
It appears that the frontal zone at these two stations was located at
slightly higher potential temperature than 304°K, as expected with the
diabatic processes of radiational cooling and entrainment that occur in
a rapidly subsiding air mass. Accordingly, the trajectory segments for
13 August were calculated for the 305°K isentropic surface. On 12 Au-
gust, 0000 GMT, the sounding at The Pas, Canada, intersected the frontal
zone under consideration (Figure 17). Higher potential temperatures
(the 308°K isentropic level) were chosen for the trajectory segment cal-
culations for that day.
The sparsity of radiosonde data over northern Canada and the Arctic
prevented the construction of reliable trajectories further back in
time. Resolute, Canada, had a dry layer above the 600-mb level on 11
August, 0000 GMT (Figure 17), which in all likelihood was connected with
the stratospheric air intrusion under consideration.
Figure 18 summarizes the trajectory analysis for this case. Within
three days, air presumed to contain relatively high ozone concentrations
descended from the vicinity of the 600-mb level to the 900-mb level and
traversed Canada and the United States. The high ozone concentrations
over Tallahassee on 14 August could therefore have been due to a stra-
tospheric air intrusion.
2.1.1.5. Seattle, Washington, 8 April 1964
The ozone sounding for this case is shown in Figure 19. In this
instance, the high ozone concentrations at low altitudes were not found
with any of the weather conditions that are usually associated with in-
trusions of stratospheric air. Figure 20 shows that the Seattle area
had had no passages of cold fronts, nor was their evidence of squall
line activity associated with an approaching front. Thus, it is un-
likely that the observed ozone concentrations at the lower altitudes
came from the stratosphere. Tropospheric origins are much more likely.
Thus, this case was eliminated from further consideration.
2.1.1.6. Seattle, Washington, 15 and 17 April 1964
The ozone soundings for these two days are shown in Figures 21 and
22. The prolonged period of relatively high concentrations at Seattle
was characterized by the passage of a cold front on 15 April and by sub-
sequent northwesterly flow aloft in the rear of a trough, as can be seen
in Figure 23 . Intrusion of stratospheric air under such conditions is
very likely. Unfortunately, detailed trajectory analyses were impossi-
ble because of the lack of adequate rawinsonde data over the Pacific.
Thus, these cases were not considered further.
36
-------
tc.
D
o
m
co
co
O
CO
CO
o
(N
co
CO
o
o
CO
o
o
-------
o
co
CO
o>
< g
CM T3
«- C
O
CM
O
CO
o
m
i
Q
<
LU
!D
D.
C
LU 0>
a e
^-* QJ
CO 3
P
< 3
,_ I
i- o
O co
U. CO
co «-
3 ^
Ifg
z < 2
ID en
Q CN 2
co — -a
oc
D
O
o
m
co
o
-------
u
FIGURE 18 COMPOSITE 6-HOUR TRAJECTORY SEGMENTS FROM
11 AUGUST 1963, 1800 GMT TO 14 AUGUST 1963, 1200 GMT
(Pressures in millibars are indicated next to the dots
corresponding to synoptic observaton times.)
39
-------
-25
1000
50 100 150 200
PARTIAL PRESSURE OF OZONE
250
300
FIGURE 19 OZONE SOUNDING AT SEATTLE,
WASHINGTON, 8 APRIL 1964
40
-------
5
a
tt
Q.
UJ
O
u.
tr
cc
Q.
tr.
Lii
u
O
(O
CL
u.
cc
(O
oc
UJ
cc
O
LLJ
OC
D
41
-------
(mb)
500
1000
-25
50 100 150 200 250
PARTIAL PRESSURE OF OZONE (>umb)
300
FIGURE 21 OZONE SOUNDING AT SEATTLE,
WASHINGTON, 15 APRIL 1964
42
-------
-25
1000
50 100 150 200
PARTIAL PRESSURE OF OZONE
250
300
FIGURE 22 OZONE SOUNDING AT SEATTLE,
WASHINGTON, 17 APRIL 1964
43
-------
SURFACE MAP, 13 APRIL 1964, 1800 GMT
SURFACE MAP, 14 APRIL 1964, 1800 GMT
500 MILLIBAR HEIGHT CONTOURS ATI 10.
7 00 P M , E S T , YESTERDAY
500 MB MAP, 14 APRIL 1964, 0000 GMT
500 MB MAP, 15 APRIL 1964, 0000 GMT
FIGURE 23 NORTH AMERICAN SURFACE AND 500MB CHARTS, 13 APRIL TO 17 APRIL 1964
44
-------
SURFACE MAP, 15 APRIL 1964, 1800 GMT
SURFACE MAP, 16 APRIL 1964, 1800 GMT
500 MB MAP, 16 APRIL 1964, 0000 GMT
500 MB MAP, 17 APRIL 1964, 0000 GMT
FIGURE 23 NORTH AMERICAN SURFACE AND 500MB CHARTS, 13 APRIL TO 17 APRIL 1964 (Concluded)
45
-------
2.1.2. Evidence of Ozone Transport from the Stratosphere as Deduced
from Studies of Radioactive Debris
2.1.2.1. General
Many radioactivity measurements have been made in the troposphere
and stratosphere (for references see Reiter, 1976a). Observations of
the geographic and seasonal distribution of ozone are also abundant
(Reiter, 1972; Reiter et al., 1975). The evidence from radioactivity
and ozone observations can be combined to estimate average and maximum
ozone concentrations to be expected at ground level as a result of the
transport of stratospheric air into the troposphere. Such an analysis
relies on a statistical treatment of the data, as opposed to the track-
ing of individual stratospheric air excursions by trajectory techniques
as was done in the preceding discussion. Radioactivity observations and
ozone measurements are not usually available at the same places and
times, so some license must be taken in comparing these data of dif-
ferent origins.
The premises involved in using ground-level radioactivity measure-
ments to estimate ground-level concentrations of ozone of stratospheric
origin are as follows:
1. The ratio of ozone concentration to radioactivity in the stra-
tospheric source area can be determined
2. All the measured radioactivity at ground level is of stratos-
pheric origin
3. The radioactivity and the ozone are transferred from the stra-
tosphere to the ground level by the same processes.
As is true of most methodologies, some difficulties are involved.
The ratio of ozone to radioactivity is somewhat uncertain because there
is an annual cycle of ozone concentration, in response to the cyclic
variations in incoming solar radiation and seasonal changes in stratos-
pheric circulation patterns. Also, the radioactivity levels changed
during the period of study in response to the decay times of the various
radioactive elements and due to atmospheric diffusion. Nevertheless,
with some reasonable assumptions, it is possible to estimate
ozone/radioactivity ratios. With regard to ozone concentration, it is
assumed that year-to-year variations of ozone are small, especially when
compared with season-to-season variations. With regard to radioactivi-
ty, the decay and removal rates of the radioactive materials can be used
to estimate concentrations at times other than when the measurements
were taken.
In essence, dry radioactive fallout is used as a surrogate for
stratospheric ozone. The behavior of stratospheric ozone is deduced by
analogy from the measured behavior of the radioactive fallout. There
are a number of short-comings to the approach and some of these will be
identified during subsequent discussions. One of these short-comings
arises from the fact that the radioactivity data are in the form of
24-hour averages and hence it is not possible to identify peak hour-
46
-------
average ozone concentration that can be compared directly with the
federal standard. Other short-comings arise from two sources: 1) uncer-
tainties in the radioactivity and ozone measurements, and 2) the failure
of ozone and radioactive matter to behave in a perfectly analogous
manner. Uncertainty in the low-stratospheric ozone-to-radioactivity ra-
tio is probably the most serious shortcoming. In spite of these diffi-
culties the technique provides a very useful tool for studying
stratospheric/tropospheric interchange.
2.1.2.2. Ratio of Radioactivity to Ozone in the Stratosphere
The mean meridional distributions of ozone (Dutsch, 1971), shown in
Figure 24 illustrate the seasonal differences in ozone concentration. *
The year-to-year variability may also be appreciable, although perhaps
not as great as that exhibited by the Arosa, Switzerland, data in Figure
25 (Wallace and Newell, 1966). The Arosa data may be more indicative of
shifts in stratospheric long-wave patterns than of the total mid-
latitude stratospheric reservoir of ozone. Lovill (1974) has found that
considerable short-term changes in total global ozone take place.
Reiter (1975b) suggests that these short-term changes are related to
fluctuations in the available potential energy of the atmosphere. Since
McGuirk et al. (1975) find a significant year-to-year variability in the
atmospheric energy cycle, it appears that the stratospheric ozone reser-
voir will display a similar variability. For purposes of the present
analysis, it is estimated that a 10-15% uncertainty in the assumed stra-
tospheric ozone concentrations will result from this interannual varia-
bility.
The production of the isotopes Sr-90 and Cs-137 is representative
of peak yields from U-235 and Pu-239. These isotopes are taken as a
measure of the total radioactivity production from nuclear devices.
Twenty years after the explosion of a nuclear device the combined ac-
tivity from Cs-137, Sr-90, and Y-90 amount to about 90% of the radioac-
tivity remaining in the environment. For "younger" debris this percen-
tage is considerably smaller.
Figure 26 shows the mean distribution of Sr-90 in the stratosphere
during the period May-August 1963, using measurements from Project Star-
dust (Seitz et al. , 1968). The units dpm/1000-scf are "disintegrations
per minute per 1000 standard cubic feet". The average ozone values for
the March-April period (from Figure 24) are also shown in the figure.
The region of interest,the lower stratosphere in the middle latitudes of
the northern hemisphere, is denoted by the shaded area in Figure 26. A
ratio of Sr-90 to ozone of about 500 units** appears to be appropriate
for much of the shaded region.
* It should be noted that the ozone concentrations are given in the
mixing-ratio units of ug/g (1 ug/m3 is approximately 0.6 ppm of 25 de-
grees C and standard pressure).
## The units of this ratio are derived from those of the figure and
are (ug/1000 SCF)/(dpm-g). These units are used throughout the discus-
sion.
47
-------
D
'A
IU
en
0.
1000
90°S 80 70 60 50 40 30 20 10 0 10 20 30 40 50 60 70 80 90°N
LATITUDE
1000
90°S 80 70 60 50 40 30 20 10 0 10 20 30 40 50 60 70 80 90°N
LATITUDE
FIGURE 24 MEAN MERIDIONAL DISTRIBUTION OF OZONE
(Mfl/8;
48
-------
400
350
3on
250
400
^350
n
2 300
z: 250
d
N
O
< "00
O 350
300
250
100
35(1
300
250
10)7 I'M!) 19-19 191)0 1951 1952 1153 1954
1932 19J3 19J4 I'll!) 1936 1937
1938 1939 19-10 1941 1942 1943 1944 1945 1946
1956 1957 1«L8 1959 1960 1961 1962 1963 1964
Source: Wallace and Newell, 1966
FIGURE 25 MONTHLY MEAN TOTAL OZONE AMOUNTS AT AROSA,
SWITZERLAND
As noted before, the ratio of Sr-90 to ozone will vary with time in
response to seasonal variations in ozone and the diffusion, removal, and
decay of the fission products. Therefore, corrections must be applied.
Figure 27 suggests that in the absence of nuclear tests, the northern
hemisphere stratospheric inventory of Sr-90 decreases by a factor of 1/e
every 14 months (where e is the base of natural logorithms). This de-
pletion of stratospheric strontium is mainly due to transport processes
into the troposphere of the northern hemisphere and into the strato-
sphere of the southern hemisphere. Correcting for the discrepancy of
about three months in the time periods to which the ozone data (March-
April) and the Sr-9U data (May-August) apply gives a value of 620 (rath-
er than 500) for the ratio of Sr-90 to ozone during the March-April
period of 1963. If the 10 to 15% interannual variability in stratos-
pheric ozone is ignored, then the 14-month 1/e decay can be assumed to
apply to the Sr-90 to ozone ratio from one year to the same season of
the next year.
In Figure 28a, an analysis by List and Telegadas (1969) of stratos-
pheric Sr-90 concentrations measured from September to November of 1964
has been plotted on the same graph with the mean ozone concentrations
for October through November (from Figure 24). An average Sr-90 to
49
-------
O
o •«
n 5
• «
.E 8
-I O
CD
il
Z CD
o E
£
03 Q
DC
3 <2
8 ft
M|f
S £ §
0 g s
N
O H
LU
CD
CM
LU
o:
LL
CO
Q
50
-------
104
o
103
55
O
^ 102
101
J^ei ' i'>€4' 1^6^ ' t«b(,
YEAR
Source: Krey et al, 1974
FIGURE 27 STRATOSPHERIC INVENTORY OF Sr90 IN THE
NORTHERN HEMISPHERE
ozone ratio of 250 to 300 appears to be valid for the shaded area.
In Figure 28b, the March-April, 1965, distribution of Sr-90 in the stra-
tosphere (List and Telegadas, 1969) is superimposed on the mean March-
April ozone distribution from Figure 24. In this case the ratios range
from 70 to 100 in the area of interest, with an average of about 90.
The ratios of Sr-90 to ozone calculated above are plotted as a
function of time in Figure 29, along with a line showing the changes ex-
pected with at a 1/e depletion rate of 14 months. It appears that the
1/e curve describes the decrease between (March-April 1963 and March-
April 1965) quite well. This says that fluctuations in the ratio due to
inter-annual variations in ozone concentrations are quite small. It
also appears from this diagram that the ratio computed for October-
November 1964 is too high. However, the ratio for autumn was strongly
influenced by seasonal changes in the ozone distribution. During au-
tumn, the ozone concentrations in the lower stratosphere of the northern
hemisphere tend to be lower than the spring concentration values (see
Figure 24) by a factor of about two. Accounting for this in Figure 29
brings the October-November, 1964, ratio approximately in line with the
expected ratio for an e-folding Sr-90 residence time of 14 months. This
is indicated in Figure 29 by the dashed box, which was based on a sea-
sonal reduction of a factor of two. Thus, the estimated autumn ratio is
quite consistent with the assumed 14-month e-folding time for Sr-90 and
the known seasonal variations in stratospheric ozone content.
51
-------
I
UJ
Ct
I
Ul
DC
o.
1000
•W ' • ••- •- ' - - '— '- ' • • •_ • - -*- f •-
90°S 80 70 60 SO 40 30 20 10 0 10 20 30 40 SO 60 70 80 90°N
LATITUDE
(a) OZONE FOR OCTOBER-NOVEMBER AND STRONTIUM-90
FOR SEPTEMBER-NOVEMBER 1964
E
Ul
QC
UJ
a:
a.
500 -
1000
90°S 80 70 60 50 40 30 20 10 0 10 20 30 40 50 60 70 80 90°N
LATITUDE
(b) OZONE FOR MARCH-APRIL AND STRONTIUM-90 FOR
MARCH-MAY 1965
FIGURE 28 ATMOSPHERIC OZONE (Solid Lines, M9/g) AND Sr90 (Dashed
Lines, dpm/1000SCF) DISTRIBUTIONS, FALL 1964 AND
SPRING 1965
52
-------
1
/
/
f
;
f
to
X!
O
•*
1
•H
P4
/
u
1
/
f
rH
a
•P
•ri
+J
Oi
e
0)
O
«
"p
4
/
C
•H
C
o
•H C
-P 0
0! -H
rl Oi
CS rl
> -P
F— I Q)
e e
0 0
Ul O
01 >
00 1 J
(DC X
0 N ^ ^
-p o X
r-—/
• /
A"
/
x
/
o o
o o
fx_ CD
o
o
m
o
o
m
u?
0)
c
o
•o
ro
CD
tD (D
en
ro
z
n
at
CM
OC
D
C3
ro
CD
0)
ro
<\J
O O O
O 0) 03
t_St_UOSOOOT)
OliVH
53
-------
2.1.2.3. Observations of Radioactivity at Ground Level
Surface fallout radioactivity observations conducted by the U.S.
Public Health Service Radiation Surveillance Network were used to esti-
mate possible ozone concentrations of stratospheric origin near the
ground. The radioactivity data are reported in units of pC m~3 (picocu-
rie per cubic meter of air passed through a filter), where 1 dpm/1000
scf = 0.0159 pC m-3.
It should be recalled here that these measurements are 24-hour
averages. The ozone estimates that are derived directly from them are
representative of the same averaging period. Such being the case, com-
parisons with the federal air quality standards for 1-hour average
ozone concentration are possible only indirectly.
In the spring and summer of 1963 the bulk of stratospheric radioac-
tive debris was less than 1 year old, stemming from the U.S. and USSR
test series of multi-megaton devices conducted during 1962 and ending in
December of that year. The relative disintegration rates of the indivi-
dual fission products (in the case of U-238, fission produced by a ther-
monuclear reaction spectrum—Freiling et al., 1965) and measurements in
rainfall at Westwood, New Jersey (Mahlman, 1965) indicate that radioac-
tivity from Sr-90 and Y-90 amounted to 0.9 to 4.0 percent of the total
radioactivity during spring and summer 1963. The following calculations
assume that the contribution from Sr-90 + Y-90 to the total stratospher-
ic radioactivity during spring and summer 1963 was 2%, which, from the
measurements quoted above, appears to be a reasonable average value, and
that the contribution from Y-90 can be neglected because it is a very
short-lived daughter product of Sr-90. These figures provide a basis
for estimating the stratospheric ozone concentrations that might have
been present when the radioactive fallout measurements were made. For
instance, 1 pC m-3 of total fallout would contain 0.02 pC m~3 of Sr-90
(2 percent of the total) which is equivalent to 1.26 dpm/1000 scf. Ac-
cording to Figure 29 , the spring and summer ratios of Sr-90 to ozone in
the stratosphere were about 500-650. During autumn of 1963, the ratio
should have been about 300 to 350. An average ratio of about 500 can
then be assumed for much of the year 1963. Hence, an average annual
fallout of 1 pC m~3 would correspond to average ozone concentrations, of
stratospheric origin, of about 2.5 x 10-3 p-g/g, or about 1.5 ppb.
The above procedure can be applied to the dry fallout (not includ-
ing precipitation washout) data as measured by the Public Health Service
Radiation Surveillance Network. The average fallout for 1963 is shown
in Figure 30. The highest value of 8.94 pC m~3 was observed near Las
Vegas. One must suspect that radioactive fallout in Nevada and adja-
cent regions included a significant fraction of tropospheric origin,
i.e., coming from the Nevada test site. A secondary region of relative-
ly large fallout values appears along the eastern slopes of the Rocky
Mountains. Downdrafts during chinook-wind episodes can carry upper-
tropospheric and stratospheric air down into the well-mixed planetary
boundary layer. Lovill (1969), for instance, finds relatively high
ozone concentrations at Boulder, Colorado, during strong chinook-winds.
The downward transport under such wind conditions occurs in conjunction
54
-------
with strong orographic lee-wave development. The fallout values of
about 6 pC m-3 in this region correspond to an average annual stratos-
pheric ozone contribution of nearly 9ppb.
According to Figure 30, a relative minimum in the radioactive fal-
lout distribution extends in a meridional direction over the midwestern
United States. To the east of this, values approaching 6.0 are seen in
several locations. According to the previous calculations the average
annual concentration of stratospheric ozone at these locations was about
7.5 ppb.
Most of the radioactive fallout is observed during spring (Reiter
1971; 1972; 1975a). Figure 31a shows the average dry fallout concentra-
tion during the first 6 months of 1963. The pattern of this chart is
similar to that of the preceding figure, but the values are approximate-
ly 50% higher. The equivalent semi-annual average ozone concentrations
are generally 15 ppb or less. Again, the anomalous value near Las Vegas
is discounted.
Figure 31b shows the number of days during which fallout values of
10 pC m~3 or greater (corresponding to about 15 ppb or about 19% of the
federal 1-hour standard) were encountered during 1963. A slight bias
may exist in these numbers because most stations reported 10 to 15% of
the daily data as missing. Again, the Nevada region is probably
anomalous because of local tropospheric contamination.
Figure 31c shows the maximum fallout reported during any day of
1963. If the Nevada area data are discounted because of possible local
contamination, we find that the maximum values were measured over the
eastern United States and in the Pacific Northwest. These maxima of
about 26 to 28 pC m~3 translate into equivalent ozone concentrations of
about 40 ppb, or approximately 50 percent of the federal 1-hour stan-
dard. It should be reemphasized that the values based on the radioac-
tivity data are the equivalent of 24-hour averages, and are not strictly
comparable to the federal 1-hour standard.
Returning to Figure 29, it can be seen that a Sr-90/ozone ratio of
about 220 is appropriate for the spring of 1964. The contribution of
Sr-90 to total fission product radioactivity would be about 4 percent by
this time (about two years after the major input of radioactive materi-
al). Thus, 1 pC m~3 of radioactive fallout during the spring of 1964
corresponds to stratospheric ozone contribution of about 6.9 ppb.
This correspondence can be used to interpret the pattern of maximum
fallout for 1964 that is shown in Figure 32 . Areas of maximum fallout
along the eastern slope of the Rocky Mountains and in a band stretching
from Texas into the northeastern United States bear some resemblance to
the 1963 pattern exhibited in Figure 31c. In drawing the isopleths, a
55
-------
O
Q.
a>
h-
§
<
ID
LLJ
QC
56
-------
cc
o
LL
K
I
o
UJ
o
ui
en
u.
<
u_
>
O
<
5
5
X
I
CO
LU
OC
57
-------
10
CO
O
O
I
8
x
<
CM
CO
58
-------
o
maximum value of 472.75 pC m~ , reported at Las Vegas on 14 March, and
another high value at Las Vegas (16 March) were disregarded as having
come from the Pike shot of 13 March at the Nevada test site. The max-
imum value outside of Nevada is about 8 pC m ^ or approximately
equivalent to a 55 ppb concentration of stratospheric ozone. This is
somewhat greater than the 40 ppb deduced for 1963, and is about two-
thirds of the 80 ppb 1-hour standard.
2.1.3. Ozone Observations at Zugspitze
To this point, the analyses of stratospheric influences have not
been strictly consistent for comparison to the federal one-hour-average
oxidant standard. The ozonesonde data represent near-instantaneous
concentrations in the lower troposphere. The concentrations inferred
from the radioactivity data are 24-hour averages. In this section,
ozone data from Zugspitze Mountain in Germany are examined in order to
estimate relationships between 24-hour and 1-hour averages at a ground-
level station. The elevation of the station (3000 m above sea level)
ensures that the ozone is mostly of stratospheric origin.
One has to assume that hourly ozone concentrations vary about the
daily mean value. A relatively high daily-mean concentration, there-
fore, produces a certain chance that hourly-mean concentrations for one
or more hours will exceed 80 ppb, although the daily average is well
below that value. The ozonsonde observations over North America re-
vealed that in 0.2% of the available cases, instantaneous ozone concen-
trations of stratospheric ozone exceeded 80 ppb in the lower tropo-
sphere. However, as noted before, ground-level observations were made
over one-hour averaging times. It is to be expected that the averaging
process, even over one-hour periods, would reduce excessive instantane-
ous concentrations. Also, it is expected that concentrations are
lowered during the mixing processes that bring the ozone to ground lev-
el. Destruction processes at ground level lower the concentrations even
more.
Hourly ozone observations from Zugspitze Observatory* were used
to compare daily one-hour-maxiraura and daily average ozone concentra-
tions. Figure 33 shows the distribution of daily one-hour maximum ozone
concentrations in relation to the daily mean values for a 529-day
period. To the right of the slanting shaded line the standard of 80 ppb
is exceeded. Figure 33 suggests that it is not exceeded under the "nor-
mal" behavior of maximum one-hour concentrations. However, two days,
January 8 and 9, 1975, departed significantly from this "normal"
behavior. The maximum one-hour ozone concentrations on these days were
in excess of 145 ppb. These days are indicated by the two points to the
right of the slanting shaded line in Figure 33. The hourly ozone values
for these two days are reproduced in Figure 34 . All evidence indicates
that these data are realistic. The following facts support this conclu-
sion:
* These data were generously provided by Drs. R. Reiter and Kantor
of the Institut fur Atmospharische Umweltforschung, Garmisch-
Partenkirchen, West Germany.
59
-------
ro u> g y - in
— — c\) ro 0> !
ft
p.
tn
O
LU
O
Ul
z
O
N
O
LU
QC
111
a:
D
O
V)
w
QC
QC
D
O
LU
a.
LL.
O
<
QC
O
z
LU m
8£
s&
LL IS
£3
5 t
O
qdd — SNOIXVHiN30MOO EO NV3W A1IVQ QNV HHOH HV3d N33MI,3H
60
-------
200
180
160
140
120
.o
- 100
n
O
80
60
40
20 -
" I T
ZUGSPITZE
Instrument off scale
10 14
JANUARY 8
18
22
10 K 18
JANUARY 9
22
FIGURE 34 OZONE CONCENTRATIONS AT ZUGSPITZE 8-9 JANUARY 1975
61
-------
* The hourly ozone concentrations on these two dates do not behave
spuriously but reveal an orderly increase from, and later an
orderly decrease to, the "normal" range of ozone concentrations
observed at Zugspitze.
• Two days constitute 0.4 percent of the available sample. Since
the excessive concentrations straddled the midnight hour they
could be considered a single case, i.e. 0.2 percent of the
available sample. Such a frequency of occurrence is in agree-
ment with the data from the North American ozonesonde network.
* Be-7 measurements suggested an influx of stratospheric air.
• Weather maps show the passage of a cold front with precipitation
and a strong jet stream, even at the 500 mb surface, early on
January 8. A short-wave trough which, on January 7 at 00 GMT,
was over the North Sea, passed the Alps by January 8, 00 GMT.
Such a weather situation is conducive to the import of stratos-
pheric air into the lower troposphere, as discussed earlier.
* An exceptionally strong trough was located in the stratosphere
over Europe, with sinking motions on its rear side capable of
moving large amounts of ozone from the middle to the lower stra-
tosphere over this region.
It appears, then, that the ozone reservoir in the lower
stratosphere—replenished by downward motions in the stratosphere—was
"tapped" by a typical intrusion event associated with the strong jet
stream and the advancing cold front. The anomalously high concentra-
tions observed at Zugspitz appear to be due to the unusual stratospheric
flow pattern which coincided with a rather typical intrusion event.
Such unusual transport processes in the lower stratosphere can easily
produce ozone concentrations just above the tropopause which are more
than twice as high as seasonally averaged values (see e.g. Reiter, 1971
and Reiter et al., 1975a). Over North America the climatological mean
position of the long—wave trough in the winter stratosphere makes these
occurrences somewhat more likely than over Germany.
It should be remembered that the Zugspitze observatory is at an
elevation of 3000m. Strong dilution to less than half of the concentra-
tions encountered in layers embedded in the middle troposphere (e.g.,
3000m) should be expected from mixing processes between there and ground
level. Even though short-term "spikes" of ozone concentrations at low
elevations can approach the undiluted values in the upper layers, con-
centrations averaged over one-hour intervals near ground level should be
less than half of the corresponding values encountered in the mid-
troposphere. If we assume that maximum concentrations of 2 to 3 times
the federal standard occur in such elevated layers (e.g., at Zugspitze),
it appears that the federal standard might be expected to be violated
occasionally at ground-level sites with elevations typical of populat-
ed areas. These instances would occur at cyclogenetically active loca-
tions of the middle latitudes in regions over which the stratospheric
long-wave trough pattern is able to establish a greater-than-normal
reservoir of ozone in the layers above the tropopause.
62
-------
2.2. Tropospheric Sources
Control strategies for ozone are necessarily tropospheric stra-
tegies because this is where controllable precursor emissions occur. In
this section, the problem of the origins of tropospheric ozone and the
meteorological factors that affect its formation are examined using two
different (but related) techniques. First, a Lagrangian approach is
used to define the history of the air and relate it to the observed
ozone concentrations. The second approach examines weather patterns,
large-scale ozone distributions, and the locations of major emissions
areas, in order to see what relationships there might be among these
different patterns at any given time.
Of course, the history of a given air parcel, with regard, to
meteorological factors and to anthropogenic emissions, is determined by
a succession of weather patterns and a certain distribution of emis-
sions. Thus, the results of the two approaches should be consistent.
However, the use of both approaches increases the probability of identi-
fying those factors that are of greatest importance. A dual approach
also provides opportunities for cross-verification that would not other-
wise be available.
2.2.1. Trajectory Analyses
2.2.1.1. The Trajectories
As noted in the Introduction, 120 air trajectories were constructed
and used as a basis for this part of the study. These trajectories are
shown in Figures 33 through 42. Positions at 3-hourly intervals are
marked. Trajectories that led to ozone concentrations in the upper 20
percentile for a station are plotted separately from the remaining tra-
jectories. The trajectories are identified by the date of their termi-
nation.
2.2.1.2. Oxidant Concentrations Related to Precursor Emissions and the
Meteorological History of the Air
A logical first step in this phase of the investigation was to ex-
amine the correlations between the ozone concentrations and the 12-hour
averaged indices of the different meteorological and emissions parame-
ters. The derivation of the meteorological and emission indices is dis-
cussed in detail in Volume 3 (Appendices A and B). Table 3 shows the
calculated linear correlation coefficients for those cases where the
significance is better than 0.05. It is important to note that 175
correlation coefficients were calculated in the preparation of this
table, so that many of the "correlations" that appear to be significant
at the 0.05 level are not indicative of true physical relationships. It
might be argued, that a better criterion for acceptance would probably
be the 0.005 significance level. However, the first entry in the table
would argue against the physical relevance of even this level of signi-
ficance. This entry suggests that there is a significant correlation
between the oxidant concentration at Queeny and the hydrocarbons
released into the air 48 to 60 hours before, whereas no such significant
relationship is noted for more recent emissions. This example suggests
63
-------
FIGURE 35 McHENRY TRAJECTORIES-OZONE CONCENTRATIONS IN TOP 20 PERCENTILE
64
-------
FIGURE 36 McHENRY TRAJECTORIES-OZONE CONCENTRATIONS NOT IN TOP 20 PERCENTILE
65
-------
S7-11
FIGURE 37 QUEENY TRAJECTORIES-OZONE CONCENTRATIONS IN TOP 20 PERCENTILE
66
-------
8-13 t
9-11
FIGURE 38 QUEENY TRAJECTORIES-OZONE CONCENTRATIONS NOT IN TOP 20 PERCENTILE
67
-------
FIGURE 39 WOOSTER TRAJECTORIES-OZONE CONCENTRATIONS IN TOP 20 PERCENT!LE
68
-------
FIGURE 40 WOOSTER TRAJECTORIES-OZONE CONCENTRATIONS NOT IN TOP 20 PERCENTILE
69
-------
7-12
FIGURE 41 YELLOWSTONE LAKE TRAJECTORIES-OZONE
CONCENTRATIONS IN TOP 20 PERCENTILE
70
-------
FIGURE 42 YELLOWSTONE LAKE TRAJECTORIES-OZONE CONCENTRATIONS
NOT IN TOP 20 PERCENTILE
.71
-------
CO
(U
CO
H
W
CO
W
o
M
H
§
o
O <£
N! C_J
O M
„ a
p [T"\
P" U
W P
CQ
Hydrocarbon
Emit • Ion a
a §
0 0
0 0
r*. CM
CM e*
irt *o
O O
5 **• o *n CM
o o o o o
o o o o o
01 N \r> in 9*
m tn *o «•> **>
o o o o o
1 I
8r>
0
o o
<••» r-
»n m
0 0
PJ -a- *a co O
i > i i i
«•* CM n *»
Insolation
§o
s
0 0
s s
? ?
M
0
o
m
n
o
i
§m
o
0 0
0 in
to m
0 0
m
§M n
O O
O O O
» CO m
o o o
1 1
(M O O ^
o o o o
o o o o
co r*. «n
o o o o
fSI *» xD CO O
i i 1 i 1
** M n -^
Relative
Humidity
o -- o «
SO O rJ O
O 0 O O
o o o o o
vr r^ PH o vo
>A fM •* CJ tN
o o o o o
o o o -« «n
o o o o o
o o o o o
o o o o o
o y
T«Aip«r*tur«
O *»
u
1 1
o *•
§«*> «•<
004
o o o
o o o o
T* m *e t*>
«n r^ *M r+
oooe>
*n
S«n
0
o e
CO O
«* ^
o o
1 I
§m
Q dl
5 o
o o o
o *o «o
o o o
1 1 1
rt
5 S
0 O
IH m
^ «n
o o
i
-------
that statistical tests of significance should be tempered with tests
based on logic and physical principles.
2.2.1.2.1. Emissions Indices
With the previously discussed exception of the correlation between
Queeny oxidant levels and hydrocarbon emissions 48 to 60 hours upstream,
there were no significant linear relationships noted between oxidant
concentrations and hydrocarbon emissions during any of the 120 trajec-
tories. The following explanations can be proposed:
1 . Hydrocarbons do not affect oxidant formation.
2. Trajectories with high hydrocarbon emissions did not occur when
meteorological conditions were conducive to oxidant formation.
3. Emissions and trajectory data are not adequate to detect correla-
tions.
4 • Enough hydrocarbons are generally present to allow oxidant forma-
tion.
The first of the possible explanations is known from laboratory ex-
periments to be untrue, and the second seems unlikely to be true for all
four of the sites. For example, the Wooster site has major urban areas
to the west-northwest (Chicago), to the southwest (Columbus and Cincin-
nati) and to the northeast (Cleveland). Yellowstone Lake is west
northwest of Chicago; Queeny is west of St. Louis. McHenry has urbanized
areas to the west (Columbus), north-northwest (Cleveland), northeast
(New York), east (Philadelphia) and southeast (Baltimore and Washing-
ton). The meteorological conditions associated with trajectories arriv-
ing from such a multitude of directions must be diverse enough to rule
out the possibility that high hydrocarbon emissions are only associated
with trajectories that occur during meteorological conditions that are
unsuitable for ozone formation. The third argument is refuted by the
fact that significant relationships were noted between oxidant levels
and NOx emissions. The fourth explanation suggests that hydrocarbons
are generally present in sufficient quantities for the formation of
ozone near these nonurban sites and, therefore, that ozone concentra-
tions will be governed by other factors.
73
-------
Table 3 thus indicates a stronger relation, at these four rural
sites, between NOx emissions and ozone formation than between hydrocar-
bons and ozone formation. This is somewhat surprising in that it was
anticipated that the occurrence of high hydrocarbon emissions and high
NOx emissions would be sufficiently in concert that their relationships
to ozone formation would be quite similar. There is some evidence,
presented later, to suggest that significant relationships between hy-
drocarbon emissions and ozone do indeed exist despite the fact that they
are not evident in the linear correlation analyses.
It should be noted that two effects of possible importance have not
been included in these calculations. The first is the difference
between emissions on weekends, and those on weekdays. Also, the diur-
nal corrections that were used are appropriate to weekday conditions.
The introduction of weekend factors might change the results somewhat,
but it seems unlikely that they would be seriously altered. Perhaps
more serious is the fact that the emissions indices that have been used
reflect only the direct anthropogenic emissions. They do not include
any natural emissions or indirect anthropogenic emissions such as may
occur, for example, in the use of nitrogen compounds in agriculture.
It has not been possible to incorporate all the possible effects in
determining the emissions indices. No account has been taken of possi-
ble differences between stationary and mobile sources, and how the ratio
of emissions from the two types varies through the day or from season
to season. Neither has any accounting of temperature effects on eva-
porative emissions been included. Although such effects might cause the
magnitude of the results to change somewhat, it seems unlikely that the
results would differ substantively. Furthermore, the increased level of
detail in the treatment of the emissions would be inconsistent with the
countywide, annual-average nature of the emissions inventory that was
available as the basic data source or with the uncertainties in the tra-
jectories.
2.2.1.2.2. Meteorological Indices
The meteorological indices are very closely related in a physical
sense, so that high correlations between ozone concentrations and
several of the meteorological indices do not necessarily indicate
several independent relationships. For example, relative humidity tends
to be inversely related to temperature so that a correlation between
ozone and temperature can be expected to be accompanied by one of oppo-
site sign between ozone and relative humidity. Similarly, the relation-
ship between insolation and air temperature would lead to the expecta-
tion that the two indices would correlate in the same sense with ozone.
Table 3 shows that these expectations are generally met.
74
-------
Also, the value of an index for one time period will often be
closely related to its value for some other time period. In the case of
insolation, the values at 24-hour intervals should be similar. Thus, we
would expect to see similarities among insolation indices representing
alternate 12-hour periods. This is very evident in the combined data
results given in Table 3 . In the case of dew-point, there should be
little diurnal change within an air parcel because dew-point is a meas-
ure of absolute humidity and changes only with the addition or removal
of water vapor. It is a reasonably conservative property of the air, so
that results for one time interval should be similar to those for anoth-
er; Table 3 supports this.
The interrelationships among the meteorological indices mean that
there is considerable redundancy in the information they provide. If
one selects the best index and uses it to describe ozone behavior, there
is not likely to be much improvement in the description if one of the
other indices is incorporated in the regression equation. Furthermore,
as one adds variables to the regression expressions, these expressions
become more specific to the data from which they are derived and more
difficult to generalize to the overall population.
Thus, it is important to try to select the "best" index and use it
to represent the others. Table 3 provides a means for doing this. On-
ly two of the indices provide significant correlations at all the sta-
tions studied: insolation and temperature during the 12-hours preceding
the ozone observations. Of the two, temperature is consistently better
correlated with ozone concentration and appears to be the better choice
if only one meteorological index is used.
There is no a priori reason why the relationships between oxidant
concentrations and the various indices should necessarily be linear. In
fact, there is some reason to believe just the opposite. For instance,
in the case of precipitation there could be a scrubbing effect on either
ozone or its precursors. With no precipitation, oxidant concentrations
would be controlled by other factors; but in the presence of precipita-
tion there may be some upper limit to the concentration that can occur.
Whether that maximum is achieved will depend on other factors. Figure
43 presents some evidence that such may be the case. This figure is a
computer-generated scatter diagram of oxidant concentration versus the
composite precipitation index for the last 12-hours of the trajectory.
An asterisk is plotted for each case. Plotted numerals indicate the
number of cases at the indicated positions. Although there is a signi-
ficant negative linear correlation between these two variables, the
scatter diagram suggests that a more useful representation of the rela-
tionship between the two might be a curve describing the envelope of the
points. The dashed curve in Figure 43 is an approximation to this
curve. The practical significance of such a relationship is that it can
be used to define conditions for which high oxidant concentrations are
very unlikely, regardless of the emissions history of the air.
75
-------
! I i
i
1
-
•
.
.
•
1
•
•
i
•
1
-
•
•
1 ' !
o
1— 1
«3<
g i
fc
w
0
O
o o ,
I'l
Crl l_^
*
>-xX'
i * •
:
,
r •
'
1
t
I
.
#
'
* *
•
#
-
* * * * *
i
*
* '
.
•
* * * *
.
* * * * *
*
«
• • I****
*\
,-1
n)
4J (J
<1) z
r 00 DC
•H 3
1 Q
3 X
i— 1 LU
a
o 8 -
n "
§ I °
O <
w t
• a.
* C *M
r! < ff -
V) -I
5
«• LU
p:
g
LL
•O
»t**»***t«*
^ prj QI if) v* ^ 0 9 in •* ^
««K« — N 4 O N »
r)j«(\icM
-------
2.2.1.2.3. Rank Correlations
Before leaving the discussion of the relationships between ozone
concentrations and each of the individual indices, it is worthwhile to
examine the possibility of nonlinear relationships. One way of testing
for nonlinear, but raonotonic relationships is through rank correlation.
In this case, use has been made of Spearman's correlation of ranks (see
e.g. Langley, 1970). This test will show if there is a significant
monotonic relationship between two variables, even if it cannot be well
represented by a linear expression.
Table 4 presents the Spearman correlations between oxidant and each
of the individual indices. As in Table 3 , the value of the correlation
has been entered only when its significance is equal to, or better than,
0.05. The significance is shown in parentheses. The results in the
table are based on the combined data from all four sites.
The results presented in Table 4 suggest relationships that are
stronger than those evident in the linear correlations given in Table 3.
It is particularly interesting that there appears to be a relatively
strong relationship between ozone concentrations and emissions of both
NOx and hydrocarbons 24 to 36 hours earlier.
Both Table 3 and Table 4 exhibit evidence of diurnal effects on the
relationships. It should be remembered that daily maxiraum-one-hour con-
centrations of ozone were chosen for this analysis. Such maxima normal-
ly occur in the afternoon or early evening hours. Thus, the first,
third, and fifth 12-hour periods before the observation will most often
be the periods with the greatest insolation and the largest variability
in isolation from case to case. From this, it follows that the possi-
bility of significant correlation is greater for these periods. Simi-
larly, nighttime emissions are lower and have s-maller spatial gradients
than daytime emissions, so the sample correlations are likely to be less
for the even-numbered 12-hour periods than for the others. As is to be
expected, diurnal periodicity in correlation is also evident in Table 4
for temperature, relative humidity, and dew-point, but not for precipi-
tation.
2.2.1.3. Ozone Concentration Related to Combinations of Factors
2.2.1.3.1. Unstratified Data
In this subsection the ozone concentrations are related to dif-
ferent combinations of indices through linear regression. As noted be-
fore, it is desirable to limit the number of indices used in these ex-
pressions to as few as possible, otherwise the results will tend to be
specific to this particular data set. Some selection criteria must
therefore be used to decide which indices are most appropriate. One ap-
proach uses a step-by-step selection process to choose those variables
77
-------
Table 4
SPEAEMAN RANK CORRELATIONS BETWEEN OZONE, METEOROLOGICAL
AND CHEMICAL INDICES FOR THE COMBINED DATA FOR ALL SITES
Index
Hydrocarbon
Emissions
NO Emissions
X
Insolation
Temperature
Precipitation
Relative
Humidity
Dew-Point
Time Period Before Measurement (Hours)
0-12
0.18
(0.03)
0.19
(0.02)
0.49
(0.001)
0.68
(0.001)
-0.45
(0.001)
-0.34
(0.001)
0.40
(0.001)
12-24
0.31
(0.001)
-0.37
(0.001)
0.28
(0.001)
24-36
0.36
(0.001)
0.40
(0.001)
0.48
(0.001)
0.47
(0.001)
-0.31
(0.001)
-0.32
(0.001)
0.26
(0.003)
36-48
0.17
(0.04)
0.23
(0.007)
0.25
(0.004)
-0.31
(0.001)
0.30
(0.001)
48-60
0.27
(0.003)
0.26
(0.004)
0.33
(0.001)
0.29
(0.002)
0.30
(0.001)
that explain the greatest part of the variance in oxidant concentration
that remains unexplained by the variables selected in earlier steps.
This approach is used by the SPSS (Nie et al. , 1975) computer program.
When this step-by-step selection process was applied to the combined da-
ta, the four indices that explained the most variance were the follow-
ing:
1. Temperature during preceding 12 hours
2. NOx emissions during preceding 12 hours
3. Temperature 48 to 60 hours earlier
4. Hydrocarbon emissions during preceding 12 hours.
78
-------
These indices are listed in order of their decreasing ability to
describe the resulting ozone concentrations.
Practical considerations relating to the purposes of this project
can be invoked to provide a rationale for selection of indices to be in-
cluded in regression expressions. The purpose of this project is relat-
ed to ozone control strategies, hence the emphasis should be on those
indices that are related to controllable processes, i.e., emissions
rather than weather.
Another possible basis for selecting variables for inclusion in the
studies is the degree to which one variable can serve as a surrogate for
another. To a large extent, this ability is represented by the correla-
tion between the variable and its surrogate. The linear correlation
coefficient is a measure of the redundancy of information contained in
the data sets containing the two variables. Table 5 shows all the corre-
lations between meteorological indices that are 0.3 or greater. It is
evident from this table that the temperature during the last 12 hours
of the trajectory is highly correlated with most of the other meteoro-
logical indices—except the precipitation index. Temperature during the
last 12 hours of the trajectory would thus seem to be a good choice for
an "all purpose" indicator of the meteorology along the trajectory.
The emissions indices do not show the same degree of redundancy
with each other as temperature does with other meteorological indices.
Table 6 indicates that NOx emissions are significantly related to KG em-
issions for the same time period. This was expected. However, there is
viturally no relationship seen between different time periods. Also,
(though not shown in Table 6) there was no significant relationship not-
ed between emissions of either pollutant with emissions of the same pol-
lutant during a different period.
For the reasons presented above, most of the following discussions
are limited to multivariate relationships between ozone concentrations
and combinations of temperature, JSOx, and hydrocarbon indices. These
relationships take the form:
Ce = A + B T + B N + B H
where
Ce = estimated ozone concentration
A = constant determined by regression
B ,B ,B = coefficients of the temperature, NOx, and hy-
drocarbon indices, respectively, as determined by regres-
sion
T,1J,I1 = temperature, NOx, and hydrocarbon indices,
respectively *
* Units of temperature are degrees Fahrenheit ;units of the emissions
indices are 10~3 m~2 hr"1; units of concentration are ppb.
79
-------
CO
w
CJ
M
M
i-4
<
CJ
I— (
8
g
Qi ^-v
S T)
S S
pfl »r-|
m § ja
CO £q 0
i-H O U
rQ
CO CO M
H BS n w
m m PJ
i
CM r- to X C
—i oj n TT IT
O £*J •?* vt" JC
ni M ,-0 T
£
Q
I
£
s
i i
a
(
o
X
f) 01
r- co
•* oco
r- cr.t-
t^ 0) ClO
• • .^
tr —t
^ -v
i' i"
CN
n
i
r*- co
co n
''
<-< o — t t-* o>
CD ^- o r-ts-
^ ^ ^t ^t .
CO t^ h- XGO
CN m m ^- n
S^* -i
^ CO
o <-
«)
-------
Table 6
CORRELATIONS BETWEEN HYDROCARBON AND NO EMISSIONS
x
(COMBINED DATA)
(No values < 0.3 are shown)
Oxides of
Nitrogen
Emissions
0-12 Hrs.
12-24 Hrs.
24-36 Hrs.
36-48 Hrs.
48-60 Hrs.
0-12 Hrs.
0.39
12-24 Hrs
0.41
Hydrocarbon Emissions
24-36 Hrs.
0.38
36-48 Hrs.
0.66
48-60 Hrs
0.39
0.77
Linear regression equations of the above type were calculated for
each of the sites and for the combined data using both weighted compo-
site indices and indices referring to the last 12-hours of the trajecto-
ry. The weighted indices are an attempt to characterize conditions
throughout the 60-hour duration of the trajectory. The weighted average
indices were computed by giving five times as great a weight to the
values in the last 12-hours as to those in the 48 to 60 hour period;
while a weighting factor of four was given to values in the 12 to 24
hour period, and so forth. This subjective weighting scheme is meant to
reflect the likelihood of greater physical relationships with the more
recent events. It also minimizes the weight given to the earlier, more
uncertain, parts of trajectory. Table 7 gives the constants for a re-
gression equation that contains the weighted indices of NOx and hydro-
carbon emissions and the temperature index for the last 12 hours of the
trajectory. The multiple correlation coefficients are also given. The
missing entry for the coefficient of the NOx index indicates that the
addition of this variable did not improve the predictive capabilities of
the equation.
It is obvious that a zero coefficient for one of the terms in a re-
gression equation indicates no contribution from the corresponding in-
dex. Inasmuch as the values of the coefficients are derived from limit-
ed data sets, it is reasonable to wonder what the chances are that the
coefficients might have been zero if the total population had been used.
To some extent this can be estimated by establishing confidence inter-
vals for the values of the coefficients. Figure 44 shows the 95% confi-
dence intervals for each of coefficients shown in Table 7 . The fact
that these intervals all span zero for the weighted hydrocarbon index
tends to deprecate the importance of this index. The significantly
nonzero values of the coefficients of the temperature index support the
importance of this index as a predictor of ozone concentration.
81
-------
Table 7
REGRESSION CONSTANTS RELATING OZONE CONCENTRATIONS TO WEIGHTED
INDICES OF NOX AND HYDROCARBON EMISSIONS AND TO
TEMPERATURE DURING THE LAST 12 HOURS
Site
McHenry
Queen y
Wooster
Yellowstone
Lake
Combined
Constant
Term
-297.41
-146.35
-228.25
-61.07
-145.72
Coefficient
for Weighted
Hydrocarbon
Index
-7.8
3.4
-1.1
-0.8
-1.2
Coefficient
for Weighted
NOX
Index
15.0
6.7
— *
6.3
5.1
Coefficient
for Temper-
ature
Index
5.3
2.8
4.4
1.8
3.1
Multiple
Coeffi-
cient
Index
0.67
0.70
0.81
0.73
0.60
For this station, weighted NO emissions did not significantly
contribute to the explanation of any of the variance left unexplained
by the temperature and hydrocarbon emissions.
Table 8 summarizes the constants for the regression equations when
the indices for the final 12 hours of the trajectory are used instead of
the weighted emissions indices. The multiple correlations are also
shown. Comparison of these values with the corresponding correlations
in Table 7 shows that there is little difference between the predictive
capabilities of the weighted emission indices and the final 12-hour em-
ission indices. Figure 45 shows the 95% confidence intervals for the
coefficients given in Table 8. Again, the lack of a linear relationship
between ozone and hydrocarbon emissions is indicated.
Inasmuch as the various meteorological indices are highly cross-
correlated, the possibility of replacing the temperature index with a
linear composite of all the meteorological indices was investigated.
Factor analysis (Nie et al., 1975) was used to derive two uncorrelated
composite indices that explained virtually all of the variance in the
original set of 6 meteorological indices. However, when multiple re-
gression analyses were performed using these factors, no significant im-
provement was found over the results achieved by using temperature
alone. It appears that the dependence of temperature on other meteoro-
logical conditions is such that the relationship between ozone and tem-
perature is as strong as that which can be established between ozone and
any optimized combination of other meteorological parameters.
82
-------
00
•2
CM
CO
" S C
-
1J 1
_ 1 1
- x
H
~
*-r
S
tf
«rf
h 8
cfi
_ D
— Q
W
a
~ o
_ §
>>
-H
5 t
D
3
y :
>^
_(
c
V
c
o
a
c
ed K
-p
(S C
Q I
C
•o +
Q) C
C r
•H (.
,0 r*
E
o c
0 >
1
1!
5
u
-4
2
j
a
s
D
H
H
D
•4
X
g
r— f
O
Q
S
O
h-4
Is
^
>> a
C (D
CD W
®
3 0
y s
CD —
CS
(S iJ
-M ~"~
d Q
Q C
O —
•a -P >>
0 U) M SH
e s cu >> c
•H O -P G C) —
A 1-1 in CD R
S H 0 CD
O S 0" S "~
• . 'i 1:
00 ~
fyJ fe
S o
S 5
W >
r-< tf "
i pq
O CO
HO ""
S <*5
FH O — —
£ M
H C£
^ g
M H
H 03 —
*
Q
LU
H
o i
% UJ
o
«N i
CO
O
CM D
a
LU
Z
o
5 1
DC
O
LU
DC
00 QJ
I
Z
0 K
*-* r^
LU
O
u_
LL
00 LU
0
O
DC
O
? i
-1 CO
LU 0
* z Q
HI 2
O ~
? M
^p _«
IO 2
O) LU
§ 5
LU
QC
a
00 u_
t
83
-------
00
a
«
CO
00
o
00
•
CO
1
^fe
CN
'
CN
CO
o
f
00
-
_ H ^ O*
—
—
- 1
1 •• •
CO
o
g
—
IN
1
O
W !5
- S5 0
0 M
M H
co <
— co >
h- 1 p^
s w
CQ
S5 O
§ „
OS O
o w
~~ O OS
Q 2
H CQ
>>
H
— (
a
j
g
0) h
p
cd
Q
•o
V
d
•H
e -
o
u :
1
0)
ot
4
Q)
C
O
P
to ^
^ Q)
O P
-i in
H O
0) O
X & C
CO
O 'K
33 O
h-<
N H
I-H <;
i >
0 Cfi
w
w w
j2^ CQ
o o
I— 1
w «
CQ O
M 2
0
O W
SH CQ
^>
>> h
a a
CD 0)
>
c s a) >> ^
•HOP c a —
ft r-l 10 CU 0)
E r-l O O K
O (U O 3 O
~ ~
_
_
. -
• . ' 1 1
_
—
—
CQ
tf S
DO —
s s
^
03 J> "™"
r-l p£
1 pq
O BJ
CQ
w o
E-i o"
05 W _
W OS
E O
E-1 CQ
00
0 U
"* z
3
z
CM O
CO ^
p™
^
3
a
LU
3 Z
o
LU
DC
co a
*~ LU
CC
LU
X
1-
00 z
i
LU
O O
LL
li.
LU
O „
o ">
00 CC D
0 0
IL *T"
$2 CN
t *~
co S H
1
UJ — i
^» cc
Tf O
CN — CO
' •» ?
O
O ^
^^ CO
JO
°? a? LU
U)
LU
?OC
3
iL
00
t
84
-------
Table 8
REGRESSION CONSTANTS RELATING OZONE CONCENTRATIONS TO
TEMPERATURE AND EMISSIONS INDICES FOR THE LAST
12 HOURS OF THE TRAJECTORY
Site
McHenry
Queen y
Wooster
Yellowstone
Lake
Combined
Constant
Term
-219.01
-138.98
-237.43
-68.02
-141.38
Coefficient
for Hydro-
carbon Index
2.0
4.9
0.3
-0.7
-0.8
Coefficient
for N0x
Index
11.4
-7.0
-0.5
2.7
3.0
Coefficient
for Temper-
ature Index
3.9
2.9
4,5
1.9
3.1
Multiple
Corre-
lation
0.71
0.69
0.81
0.72
0.59
Another measure of the effectiveness of a regression equation, be-
sides the multiple correlation coefficient, is a comparison between the
values predicted by the equation and the values observed. Table 9
shows the standard errors for the two classes of equations— those using
the weighted emissions indices and those using the emissions during the
last 12 hours of the trajectory. By this measure there is little
difference between the two kinds of index. Figures 46 and 47 are scat-
tergrams that illustrate how well the observed concentrations are fit by
the regression equations derived from the total data set. There is one
anomalous point at the far right of each diagram. The data for that
point were reexamined and appear to be correct; but it should be recog-
nized that only slight miscalculations in the trajectory analysis could
have caused the assignment of unrealistically high values for the emis-
sions indices. In this case, the air appeared to have passed over a ma-
jor urban area, when in fact the air may have passed over an area of
much lower emissions only a few tens of kilometers away. In general,
the points in Figure 46 and 47 form tight, consistent patterns. There
appears to be a change of slope at around 80 to 100 ppb.
Table 9
STANDARD ERROR OF ESTIMATE OF THE REGRESSION EQUATIONS
Site
McHenry
Queeny
Wooster
Yellowstone Lake
Combined Data
Equation
Weighted
Emissions
65 ppb
35
22
24
47
Emissions 0 to
12 hours prior
60 ppb
36
22
23
47
85
-------
*
*
*
*
•
I
I
\
\ ""
\
V
\
\
"T
A
* •
« * *
#*
* *
«
t
*
* *
-H
4
4
•
* *
*
* *
* *
•
\"
*\ N
x * *
1 •
o
n
.
in
o
.
,\
to
~
S
c
' ' b
* s
"" C
fe
1
• a
N a
fO ft
r^
?"
g
• E-
e £
•ri
C
ft
<
t—
• r
3 a
pt
j2
O
h-
E-
• ..,
00 S
10 g
PL
1
• ft
ro c
N
c
L
» •
• to
LJJ
•^ w
03 W
^UU
ZUJ "g
O U c
^ o
LU r- W
l|fe
° UJ _l
Iss
DC C/J •"
HI CO (TJ
EQ OC ^
O C3 CC
O oc <
5oS
LU ;=
DC (_ S
LU < LU
> — Q
C/) LU <
(O
U
CC
N
«
n
10 -
* •"
CM ni
qdd NOIXVH1N30NOO 3NOZO QaAHaSHO '°0
86
-------
+
•
•'
'
'
1
*
#
'
* ' *
i
,
_
>• i
\m
o
\ ' "
\ O
\ °
'
\
\
A
' \
.\
*
(NJ
#» * *
*
*
*
,
- « - j
1 0
1 -
J OJ
1
i.
i in
1 9
I -
*
I
1
JJ i
IN C
1 **
l +
f
1 !z
i. £
!* £
12 a
-..—..„_.,....,-.- ««-^ £
« i f*
(M 1
* ' Z
* I C
.« « * • H
V # * 1 m CA
\ *« IN c/
* - [i
\ * * g
*\.% ! |
#\» *(\i » i s
*** i ^
\» * • F
» 1 o E-
* 101-
* # 1 - 5
N 1
ww . * + C
»\ * N # * » 1 £
* \ * * 1 p
* N Y^* : i. I
\* : Is ^
* »\ * #1 V
V » | [!
\l C
* » 1 h-
* 1 f-
* • <
I
UJ Z
W J^
o -3
— ^»
i o 2
f <
< D
DC O
O UJ
< z
—
cc w
H UJ
0 P OC
«f O
O UJ
CO OC
^
^
UJ
cc
C3
qdd
NOIXVULN30NOO 3NOZO
87
-------
Other measures of the effectiveness of the regression equations are
available. The degree to which a regression equation that is based on
one set of data fits other data sets provides an assessment of the ap-
plicability of the equation. Such a comparison was made using the five
data bases (4 sites and the combined data set) and the ten equations
(one per site using weighted-average emissions, and one per site using
emissions for the last 12 hours). None of the equations based on data
from other locations fit the Wooster data very well; correlation coeffi-
cients range from 0.35 to 0.37. However, the Wooster equation provides
reasonably good estimates for the concentrations, at other stations with
correlations ranging from 0.60 to 0.71. The Wooster equation that is
based on emissions during the last 12 hours of the trajectory also pro-
vides a good fit to data from the other stations, with multiple correla-
tions from 0.55 to 0.70.
When the equations based on the combined data set are applied to
the subsets for the different stations, the multiple correlations range
from 0.55 to 0.72 for the equation using the emissions during the final
12 hours, and from 0.58 to 0.72 for the equation using the weighted em-
issions. The standard error of estimate for the weighted-index equation
lies between 24 and 67 ppb for the different sites; using the equation
based on the final 12-hour indices, the extremes are 23 and 64 ppb.
Multiple-regression equations were also formed using combinations
of indices other than those discussed above. For example, the rank
correlations suggest that emissions during the 24-36 hour time interval
are important. A regression equation was formed using the temperature
and NOx indices for the 0 to 12 hour period and the NOx index for the 24
to 36 hour period. The resulting equation is as follows:
Ce - 2.89 T(0) + 2.45 N(0) + 2.55 N(24)-133.4
where
Ce « estimated ozone concentration
T(0),N(0) - average temperature (°F) and NOx (10"3 ton mi'2
hr~l) emissions 0 to 12 hours before observation.
N(24) » average NOx emissions, 24 to 36 hours before
observation
The correlation achieved with this combination of indices was 0.62, very
slightly better than with the combinations discussed above. Figure 48
shows the fit achieved with this equation.
88
-------
*
*
*
*
• • • *
\
\
\
""A
w
O ^L
o \
\
:~~~\
*
* *
*
*
* *
*
*
•
*
i
• • *
*
*
*
*
* *
\ * t\l IV •
\*« *
A , .
*V
» • *
\" ' . *
« *
* • ni
•
n> *
• •
\ni *
» «V * * * •
• \ * * *
* * \ •
* \* *
• » • *
•
4 0
• v'
• \ * *
\
•
*
•
,•
» •
o
.Q
a
a
X
1-
X
• ill
3 H
Z
Z
. o
O i
* rf
2
• o
£ z
D
Z
. o
£ H
QC
Z
01
O
O
LU
z
3 N
0
Q
HI
1-
H
co
LU
1^ O
• -c
t
LJ
UJ Q
||
O M H
X "~ <
| ^ m
> h- UJ
O \r LU
< UJ o
^ DC U-
UJ < ,«
0 OC X?
Z UJ D
O °- o
°s°
z *~ S
N ^ '
^* o 2
UJ l~ in
CO < £
03 D
O O uj
O OC
lie
S S M
Q "J —
c 52
LU ^
1 — «^_ (i i
< O x
o oc o
CO U. Z
00
LJJ
qdd
SIMOIlVb±N30NOD 3NOZO Q3AH3S8O '°0
D
a
89
-------
Another equation that was checked used emissions indices for the 24
to 36 hour period for both hydrocarbons and oxides of nitrogen, along
with the temperature index for the final 12 hours before observation.
It was observed that the addition of the hydrocarbon index did not con-
tribute to the explanation of the ozone concentrations.
2.2.1.3.2. Stratified Data
Figures 46, 47, and 48 suggest that a better fit might be achieved
if the data were stratified according to the observed ozone concentra-
tion. Two equations were derived using the temperature, NOx, and hydro-
carbon indices for the last 12 hours. One equation was derived to fit
the cases where the observed ozone concentration exceeded the federal
standard of 80 ppb; the second equation fit those cases with lower ozone
concentrations. The resulting equations were as follows:
Ce « 1.14 T(0) + 1.52 N(0) + O.li HC(0) - 38.8 (03 > 80 ppb)
and
Ce = 0.93 T(0) + 2.22 N(0) + 0.31 HC(0) + 48.9 (03 £ 80 ppb)
These equations suggest that the dependence of ozone concentrations
on temperature is greater for the concentrations above 80 ppb than for
concentrations under the standard.
Figure 49a shows the scattergram of observed concentrations versus
those calculated from the above equations (all sites combined). It is
apparent that the two expressions achieve better agreement than the
corresponding single equation illustrated in Figure 46 . This piecewise
linear regression results in a correlation between observed and estimat-
ed ozone concentrations of 0.76, as compared to the 0.59 obtained with
the single expression (Table 8). The standard error was reduced from 47
ppb (Table 9) to 38 ppb.
During the derivation of the equations just discussed, it was
discovered that the insolation indices for the last two 12-hour periods
of the trajectory S(0) and S(12) were better predictors of the higher
ozone values than the temperature index. With these parameters substi-
tuted for temperature, the equation for the cases where observed ozone
concentrations were over 80 ppb is:
Ce = 298 S(12) + 133 S(0) + 2.40 N(0) - 1.79 HC(0) + 35.5
Using this equation, in combination with the one for low ozone con-
centrations given earlier, results in a correlation of 0.81 between ob-
served and estimated concentrations. The standard error of the estimate
is 34 ppb. The scattergram is shown in Figure 49b. It is thus seen
that ozone concentrations can be reasonably well estimated on the basis
90
-------
•a
P.
o
H-t
EH
W
O
o
o
§
w
w
§
. I
I I
I I
} (a) ESTIMATED FROM TEMPERATURE AW
i
i
i
i
i
i
i
1
• *
* *
I
T
D EMISSIONS !
i
i
i
i
i
i
i
i
i
Q mo.
W
CQ
ffl
O
(b)
ESTIMATED FROM TEMPERATURE ,
AND INSOLATION
'3. 31. Sn. 68. 86. 10». U2. UO. 159. 17T. 195.
Ce, OZONE CONCENTRATION ESTIMATED WITH REGRESSION EQUATION- ppb
FIGURE 49 SCATTER DIAGRAM OF ESTIMATED VERSUS OBSERVED OZONE FOR
TWO PIECEWISE LINEAR REGRESSION EXPRESSIONS
91
-------
of the recent exposure of the air to sunlight. It is worth inquiring
whether the same degree of predictability can be achieved without using
any emissions indices as predictors. Piecewise regression was used in
the same manner as discussed above, but with only temperature and inso-
lation indices used. The equations are:
Ce = 1.15 1(0) - 31.1 03 < SOppb
Ce = 299 S(12) + 136 S(0) + 38.9 03<80ppb
The concentrations are estimated with these equations nearly as
well as with those that include emissions. The scattergram of observed
ozone versus estimated ozone concentration is shown in Figure 50. The
correlation is 0.79 and the standard error is 36 ppb. These results
show that the addition of the emissions indices to the piecewise regres-
sion equations explains only about three or four percent more variance
in the ozone data. The errors in the estimates are reduced by only one
or two ppb.
The strong dependence of ozone concentration on meteorological con-
ditions suggests that stronger correlations might be seen between ozone
and emissions if the data were stratified according to whether meteoro-
logical conditions are unfavorable or favorable to ozone formation. Ac-
cordingly, correlations between ozone concentration and emission indices
were calculated from two subsets, as follows: 1) those for which the
average temperature during the last 12 hours was greater than 70°F, and
2) those for which there was no precipitation during the last 12 hours.
There was only one case where the temperature index for the last 72
hours was less than 70°F and the federal standard of 80 ppb was exceed-
ed. Data from all stations were used. However, for the subset of data
for which the temperature index for the last 12 hours was greater than
70°F, no significant correlation was found between ozone concentration
and any of the hydrocarbon emissions indices. In this higher tempera-
ture data set, ozone was significantly correlated with NOx emissions 0
to 12 hours and 24 to 36 hours before the observation. However, the
correlations were about the same as those shown in Table 3 for the un-
stratified data. Because of the reduced sample size, the correlations
were slightly less significant.
For those cases where no precipitation occurred along the air tra-
jectory during the last 12 hours, there were no significant correlations
between ozone and the HC emissions indices. Significant correlations
were found with NOx emissions during the preceding 12 hours (0.32 corre-
lation) and the 24-36 hour period (0.29). Again, however these values
do not differ much from those found for the unstratified data (Table 3)
92
-------
•a
P.
o
I—I
w
o
w
03
m
a
13
oa
w
o
M
EH
S5
I
O
O
§
o
CO
o
LU
CC
C5
LU
or
HI
Z
LU
CO
LU
o
LU
ac
O
a
LU
CC
LU
CO
CQ
O
CO
CO
CC
LU
>
a
Is
5 O
.— w
a
UL 2:
o <
S LU
< DC
DC =>
0 t?
01 5
LU OC
t ^
< 1
O LU
CO |-
g
H
s
LU
CC
a
qdd SNOIXVULN30NOD 3NOZO
o
' 0
93
-------
2.2.1.4. Characteristics of the Trajectories
2.2.1.4.1. Background
The practical objective of this project was to determine relation-
ships between ozone concentrations and anthropogenic emissions in the
hope that such relationships will be useful for the formulation of con-
trol strategies. In the case of a secondary pollutant like ozone, this
problem has two aspects. The first is identifying the controllable fac-
tors causing high ozone concentrations. The second is identifying the
temporal and spatial relationships between the causative factors and the
resulting ozone concentrations. Any control strategy will have to deal
with the likely spatial relationships between pollutant sources and the
areas impacted by the resulting ozone.
Thus far, only the relationships between ozone concentrations and
the various indices have been discussed. The strongest ties have been
observed to be with solar radiation and with temperature. These obvi-
ously are not cntrollable items. Based on the linear correlations, NOx
emissions during the last 12 hours of the trajectory and those between
24 and 36 hours before the end of the trajectory appear to be the most
important factors. This section examines the distances and directions
traveled by the air during the last 12 hours and during the last 36
hours of the trajectories. In some instances, the data are stratified
into two categories: "high" ozone cases and "not high" ozone cases. As
discussed earlier, and in Appendix A (Volume III), half the cases were
selected from those days which had maximum-hour ozone concentrations
among the highest 20% of the available data sample. These are the
"high" cases. The remaining half of the cases, divided about equally
between values in the lowest 20% and those within 10 percentile of the
median of the maximum-hour concentrations, are the "not high" cases. To
some extent the following statistical analyses simply quantify the
distance/direction relationships that are qualitatively evident in Fig-
ures 33 through 42.
2.2.1.4.2. Travel Distances
Figure 51 is a scatter diagram of ozone concentration at the four
rural sites (combined) versus the straight-line distance between the
monitoring site and the location of the air 12 hours before the observa-
tion. Figure 52 is similar, but uses the 36-hour separation between
the air parcel and the site. The figures indicate that the higher ozone
concentrations occur more often with lower wind speeds and correspond-
ingly shorter net travel distances. The dashed lines in Figures 51 and
52 are envelopes that might be used to estimate the maximum ozone con-
centrations at a rural location in an air mass that has come from the
distance shown on the abscissa during the preceding 12 hours (Figure 51)
or 36 hours (Figure 52). In about one half of the instances where the
federal standard of 80 ppb was equaled or exceeded, the air was ap-
94
-------
*
.
d
.$r ,
/v y
4^?' f
>§y f
•^ »
7^
XV *
• «f ******** + N-M — i^i + fcaawi.!*^***.
-
f
/
f
/ *
^ y
o* *^
*Y ^^
^y ^^
,£f *
/f
^/
'
*
*
* «
Ijt
*
*#
*
*
*
* 1 N
1 «
# *
•i •
* * 1 *
» i n
* -f
* 1
* 1
* N # *# 1
» *# * •
i m
* i e
« i n
» » * i
* » +
* « i
** I
» » i
* CM |
* * * T«
« 1 (VI
M * * 1
* « * +
« 1
» * !
* »
* 1 N
1 CD
* * 1 -
* 1
# » +
* * 1
« N
I •*
* |
+
1
* 1
1
« * * •
* | O
* * is
* » 1
» 1
* 1
1
» i
* i
-* * *
I
W
o
PK
W
PQ
Pi
Q
(N
-------
*
«-«»«««-^Mfc.«».««M-
^
^
^^
O 4
A? f
* /
/
/
/
/
f
/
t
/ * #
J
/
J
f
f
/
/
f
^
iff
'
# • • •
r- r> 0v to
* - N •*
rt n N w
nrfd
, >
^^> f
£3 j
^
j^v' X
Cy ^^ ^
^^"^
..cy y *
/
ly /
N
T /
/
* *»
*
*
*«
»
*
*
*
*
*
'
— N n
-« s •*
CM - —
KT/-ITTWTCI TW KTCITTWU
f
*
X
f
f
* *
/ »
/ #
,
*
*
4
* *
**
*
4t
* » •
*
* *
\
* » »
« M *
*«
« *
*
*
«
•
* *
* # #
*»
*
# » ft
*
» *
*
* * *
•
*
*
* * **
* *
*»
*
*
* *
» * *
*
* .*
•
*
* * *
* *
*
* *
* *
'
•
»
• • « *
O» « - N
ON*
n
T kTCf^kTr*^ &r\
t *
tt
C
CO
•
z
« 0
« —
s \-
* CO
O
Q.
« *
Q> |~j Jri
4f ^ ^
" 5 ^
OH OC
<
' W IN
* PS T
"" O r~
n ^ I-
^
w ai
esi O
* 3 2
o O
^* V— <<
2 PS
g 52
^ 2 m
J7> O rrt
*** 12 O rr"
1 zo
H 0 ^
W N (o
CQ O CO
•
(f) [T]
-------
parently within about 180 km of the receptor, 12 hours earlier. Simi-
larly, one half of the cases where the standard was violated involved
air that had been within 500 km of the site 36 hours earlier.
The information shown in Figures 51 and 52 can be displayed in
still another way. One can examine the frequency distributions of trav-
el distance for the two different classes of data: "high" and "not high"
ozone concentrations. Figure 53 shows the results of such analysis. As
expected, the travel distances for the higher concentration cases tend
to be less than those for the lower.
Analysis of the travel distances associated with the higher concen-
tration can be useful in guiding the formulation of control strategies.
The frequency distributions of travel distance, combined with the time
span over which emissions are influential, roughly define the dimensions
of the area within which controls should be applied.
2.2.1.4.3. Directional Effects
It is not sufficient to define only the travel distance between the
ozone observation and the related emissions; the direction is also im-
portant. Figure 54 is a scatter diagram of ozone concentrations versus
the direction from the observation site to the position of the air 12
hours before the observation. Figure 55 shows the scatter diagram for
the air position 36 hours before the observation. The dashed lines in
Figures 54 and 55 are estimated envelopes of the maximum ozone concen-
tration that might be expected at a rural location in an air parcel that
has traveled from the direction shown on the abscissa during the
preceding 12 hours (Figure 54) or 36 hours (Figure 55). Note that for
both time periods the highest ozone concentrations occur when the air
arrives from directions southwest through northwest, and the lowest
ozone values are brought in by air coming from the east through south.
Although Figures 54 and 55 show the extremely high ozone concentra-
tions to be associated with arrivals from the northwest, more "high"
ozone cases arrive from the southwest. When the flow directions are di-
vided into four quadrants—0° to 89°, 90° to 179°, and so forth—the
largest number of "high" ozone concentrations occurred when the air came
from directions between 180° and 269°. For the "not high" cases, the
most frequent directions were in the 270° to 359° quadrant. Figure 56
summarizes the findings of this analysis.
97
-------
2000
,-q
W
u
1000
700
£ 500
Q
g 300
I
2;
W
03
w
200
100
70
50
12-HOUR
I
Top 20-Percentile
Other Cases
I
I
I
I
10 20 30 40 50 60 70 80
PERCENTAGE < ORDINATE VALUE
90
FIGURE 53 FREQUENCY DISTRIBUTIONS OF 12 AND 36 HOUR
TRAVEL DISTANCES
98
-------
\
• * *
•
)
s «
V
^/
?/
o
d
LLJ
OC
O
HI
UJ
01
(NO
E-i
00
HI "2
>ai
22
I- in
< OQ
°ECO
^^ fin
N !I
u- n.
«j
S ^
o: °-
~ oc
Si
111
g
LL
IM 1M
qdd NOIXVXS XV NOIXVHXNaONOO 0
99
-------
k
H
V
« 02
00 QH
"§
CO
~ w
tn ,
(M 2*
O
H
M
• co
S8
05
O
M
H
OT
O
en
o>
0)
^
fac
OJ
•o
H W
o ra
- w
o
LU
CC
Q
LU
I
l-
co
CO
LU
LU
OC
I
o
<
OC
I-
LU
O
z
o
o
LU
O
N
M O
tf _
P5 S
< <
DC
§ «
o "t
LU
CD
CO
DC
O
I
co
co
Q E
UJ LU
O
CO
in
ID
LU '
OC
=3
qdd
NOIiVXS XV NOIiVHXNaONOO O
100
-------
60
50
W
w
o
o
40
20
sr io -
A
O ~ High 03 values,
12 hour travel direction
D ~ High 03 values,
36 hour travel direction
• - Not high 03 values,
12 hour travel direction
A - Not high O3 values,
36 hour travel direction
I II
000- 090-
089° °
HI IV Quadrant
180- 270-
179° 269° 359C
FIGURE 56 FREQUENCY DISTRIBUTION OF 'HIGH' OZONE AND 'NOT HIGH' OZONE
TRAVEL DIRECTIONS (Combined data)
2.2.2. Studies of Synoptic-Scale Ozone Distributions and Weather Pat-
terns
2.2.2.1. Large-scale Spatial Distributions of Ozone Concentration
2.2.2.1.1. Background
The daily peak-hour ozone concentration observed at SAROAD stations
during 1974 were used as the basis for a set of isopleth maps of the
peak-hour ozone concentrations over the eastern* United States for
each day of the year. The locations of SAROAD sites measuring ozone
during 1974 are shown in Figure 57.
* The states east of, or traversed by, 100 degrees W. meridian.
101
-------
•f --.-
FIGURE 57 LOCATIONS OF SAROAD SITES IN THE EASTERN UNITED STATES
MEASURING OZONE DURING 1974
102
-------
Some subjectivity was involved in the ozone isopleth analysis. In
order to eliminate the worst of those cases where keypunch errors in the
data cards had occurred, or where the decimal place indicator or the un-
its were misreported, no values were used which were greater than 500
ppb during the summer or late spring months or greater than 300 ppb at
other times. Occasionally some observations still appeared to be
anomalously high relative to neighboring observations. In such cases a
subjective decision was made whether to retain or discard the data, usu-
ally based on the hourly observations at the site before and after the
observation in question. In one instance, data from a given site ap-
peared anomalously high for a full 3-month period, and was discarded
when the agency confirmed that they had experienced trouble with a newly
installed data acquisition system during that time. Other than this in-
stance, very few data were discarded.
Volume III of this report contains the isopleth anaysis of
maximum-hour ozone for each day of 1974. The ozone analyses are paired
with the morning weather maps for each day. The objective in comparing
these maps was to investigate geographical relationships between weather
patterns and ozone concentration in the context of a defined geographi-
cal pattern of precursor emissions (NEDS data, discussed earlier).
Three approaches have been used to establish those relationships.
The first of these appoaches was the simplest. The ozone maps for each
day were examined to see where and when violations of the federal oxi-
dant standard occurred. Eight geographical regions of frequent ozone
violations were identified. The number of incidents per month in each
of these regions were counted. This provided a picture of seasonal and
geographic variability. Those parts of weather systems most subject to
ozone violations were also identified. The frequency of occurrence of
the higher ozone concentrations in the different parts of the weather
systems were determined. As might be expected, some meteorological con-
ditions are much more conducive to widespread ozone violations than oth-
ers.
The second approach to the analysis of the ozone and weather pat-
terns was more formal (and less successful). Pressure patterns and
ozone patterns were classified objectively using the technique proposed
by Lund (1963). Briefly, each of 134 days during 1974 on which ozone
violations were widespread in the eastern United States was character-
ized by a set of 20 pressures read from the daily weather map at regu-
larly spaced points in the eastern United States. The correlations
between the twenty pressures for each day and those at the same twenty
points on each of the other days were calculated. The daily pressure
pattern that correlated with the most other daily patterns at the 0.7
significance level or greater was considered to be the prototype for the
most important group of pressure patterns. In theory, a second proto-
type can then be similarly chosen from among those daily pressure pat-
terns that are not in the first group. The classification can continue
so long as there are a significant number of cases in a class. The
ozone patterns can be similarly classified. The coincidences of weather
and ozone patterns was to be studied but, as is discussed later, this
was not possible because these were not enough distinct types of ozone
or pressure pattern found in the data set.
103
-------
Table 10
FREQUENCY OF OCCURRENCE OF DAYS WHEN THE FEDERAL OXIDANT
STANDARDS WERE VIOLATED IN THE EASTERN UNITED STATES
Month
January
February
March
April
May
June
July
August
September
October
November
December
No. of Days with
03 S 80 ppb
in the Eastern U.S.
(1974)
2
4
10
21
30
30
31
31
21
25
7
3
In the third approach, some of the trajectory analyses discussed in
the preceding section were used in the interpretation of the ozone iso-
pleth anayses. The trajectories were used in a qualitative way to il-
lustrate the recent history of the air in the high ozone areas.
2.2.2.1.2. Frequency of Occurrence of Ozone Concentrations Above the
Federal Standard
The first step in the analysis of synoptic ozone patterns was to
determine how frequently ozone levels exceeded the federal standards in
one of more parts of the eastern United States. The maps were examined
and the days with values of 80 ppb or greater were counted. Table 10
shows the results of this analysis. As expected, the table shows that
violations of the standard were an everyday occurrence somewhere in the
eastern United States during the warm summer months and were quite in-
frequent during the winter.
104
-------
2.2.2.1.3. Areas of Most Frequent Ozone Standard Violations During 1974
A preliminary comparison of the daily map series suggested that
areas of high ozone concentrations were most frequently found in the
following regions:
• Florida peninsula
• Texas-Louisiana Gulf coast
• New York-New England
• Western portions of Oklahoma, Kansas, and Nebraska
• Southeast of Lakes Erie and Ontario
• Washington-Philadelphia
• South or southwest of Lake Michigan
• St. Louis and Ohio River Valley
Figures 58 through 62 show examples of high ozone concentrations
occurring in the listed areas. Figure 63 shows the counties with the
highest NOx emissions densities in the United States. The average NOx
emissions in the blackened counties exceeds 75 tons mi~2 yr~l« There
are only 134 such counties in the entire U.S. Of course, many of these
same counties are also among the highest in hydrocarbon emissions. Fig-
ure 63 shows that, with the exception of the western Oklahoma-Kansas-
Nebraska area and perhaps the Florida peninsula, the areas of the most
frequent high ozone concentration also contain regions of major emis-
sions. This simple comparison shows that the expected relationship
between anthropogenic emissions and high ozone concentrations is subjec-
tively identifiable in the data.
Table 11 summarizes the number of days per month that the federal
oxidant standard was violated in each of the listed areas during 1974.
In most of the various areas the annual trends parallel those of the
country as a whole. As might be expected, the warmer, more southerly
locales have occurrences of ozone concentration greater than 80 ppb in
the early spring and late fall more often than do most of the more
northerly areas. Surprisingly, New England also has rather frequent
violations of the standard in these seasons, as do the western parts of
Oklahoma, Kansas, and Nebraska. The violations in the remaining areas
are limited almost completely to July and August.
105
-------
o o o o
ID CO CO CD
UJ
cc
HI
(A
111
UJ >
to oc
Z UJ
g >
-------
/^;%fiw^4/ \-
^w^fS^^t^
Ct ^fe^^f /s%s4 ^/•>v
\ ^ r^W^^^^^ ?&#-' "' "' 4
?-VOT ^?/>
-------
a
i
o
o
I
I-
§
g
QC
Z
ui
O
Ul Z
Z <
O -1
N a
o z
Ul
— ui
UJ <
oc
D
108
-------
,"'6^' .„•>„ •fo£!tf~9«H ^,r,V, <
r*P:1%FW-/^ I?
a
o
o
CNJ
I
O
O
QC
Q
<
UJ
E
UJ
Ul
LU
O
b
S8
O u.
z 5
"J $
O <
Z w
§8
Ul -I
N X
O 01
O
Ul
<0
UJ
oc
D
a
UL
109
-------
gs88 88
QC
o
LU
I
(O
1 *
* o
z 9
DC en
UJ K
fe o
UU CJ
5 <
Z X
LU O
O H
Z O
81
I H
« Q
O
UJ
s!
X LU
LU Q-
(O
LU
tr
o
110
-------
FIGURE 63 COUNTIES WITH AVERAGE ANNUAL NOX EMISSIONS
GREATER THAN 75 t
It should be noted that, although the data in the Kansas area some-
times appear out of place, they are not consistently higher or lower
than other surrounding values. Discussions with personnel of the
Wichita-Sedgwick County Department of Community Health, the agency
responsible for the measurements, provided no reason to discard the da-
ta. It is interesting that the high concentrations are found rather
frequently during seasons other than the photochemically active summer
season. Also, the area is relatively free of major anthropogenic
sources of precursor emissions. These facts suggest that an important
natural source mechanism may be operative. Earlier in the study, this
area was seen to have relatively high radioactive fallout, and hence
stratospheric ozone at ground level (Figures 30-32). Of course, the
concentrations discussed presently are much higher than those to be in-
ferred from the fallout data or to be otherwise derived from the stra-
tospheric portion of this study. Thus, no explanation can be offered at
this time for the cause of these high ozone values.
Ill
-------
Table 11
NUMBER OF CASES FOR EACH MONTH WITH DAILY MAXIMUM OZONE > 80 ppb
IN SPECIFIED REGIONS OF THE EASTERN UNITED STATES
Florida
Peninsula
Texas -Louisiana
Gulf Coast
New England
Western Oklahoma,
Kansas, Nebraska
SE of Lakes Erie
& Ontario
Washington- Phila-
delphia Corridor
S or SW Shores of
Lake Michigan
St. Louis and Ohio
River Valley
Other Areas
January
0
0
2
0
0
0
0
0
0
February
0
2
0
0
0
0
0
0
2
•g
M
1
5
5
0
3
0
0
0
0
1
r— 1
•H
H
o,
<
3
5
6
8
0
2
0
0
5
>!
S
8
8
10
22
2
3
1
3
0
>
rH
3
>n
9
15
9
22
20
15
10
22
6
August
3
11
18
20
21
7
9
13
6
September
1
4
13
12
2
3
2
5
5
October
2
10
1
15
0
5
0
0
5
November
0
2
0
1
0
2
0
2
0
December
0
1
0
0
0
0
0
0
2
2.2.2.2. Relation Between Ozone Distributions and Weather Features
2.2.2.2.1. Winds Associated with High Ozone Concentrations
Table 12 was prepared in order to determine whether there were
differences from region to region in the relation of wind speed and wind
direction to the frequency o£ high-ozone events. The table shows that
very light winds are important to the occurrence of high ozone concen-
trations in several regions. This result is consistent with the findings
of the trajectory study, which showed shorter travel distances
(corresponding to lower wind speeds) to be associated with higher ozone
concentrations.
112
-------
Table 12
WINDS REPORTED ON MORNING WEATHER MAP IN AREAS
WHERE PEAK-HOUR OZONE EXCEEDED 80 ppb DURING THE DAY
(Number of days from June through August)
Region
Florida Peninsula
Texas -Louis iana
Gulf Coast
New York-New
England
Western Oklahoma,
Kansas, Nebraska
SE of Lakes Erie
and Ontario
Washington -Phil -
adelphia Corridor
S or SW shore of
Lake Michigan
Ohio River Valley
& Surroundings
Calm
11
17
3
1
20
9
7
21
Surface Winds
> 2 m/s
N to E
5
10
4
7
1
5
0
2
E to S
4
0
7
13
10
7
5
1
S to W
2
7
26
36
6
7
6
7
W to N
0
1
2
11
1
4
1
0
Two regimes are evident from the table, that is, those for which
the high ozone appears to have resulted from nearby precursor emissions
and those for which the ozone (and/or its precursors) appears to have
been transported by the wind. Most of the situations in the New York-
New England area and in the western Oklahoma-Kansas-Nebraska area appear
to fall into the latter regime. Most of the situations in the St.
Louis-Ohio River Valley area fall into the former regime. Both regimes
are evident at the other locales. These observations are qualitatively
consistent with the geographic distribution of precursor emissions. Al-
so in agreement with the distribution of emissions — for the transport
113
-------
cases — is the relation between the occurrence of high ozone concentra-
tions and the wind direction. This is especially evident (Table 12) for
the New York-New England area, where the S-W direction indicates that
the air tends to come from a direction along the east coast urban corri-
dor on high ozone days. This deduction has been confirmed by Cleveland
et al. (1975) and Ludwig and Shelar (1977) in more detailed studies of
this part of the country.
2.2.2.2.2. Weather Patterns Associated with High Ozone Concentrations
Earlier sections addressed the meteorological conditions that might
be expected most often in conjunction with high ozone concentrations.
According to the trajectory analyses and the stratospheric intrusion
studies, the following types of meteorological situations are likely to
be significantly associated with the occurrence of ozone concentrations
in excess of the federal standard:
1. Warm air ahead of a cold front
2. Warm sector of a frontal wave
3. Western side of an anticyclone (high pressure area)
4. Other parts of an anticyclone
5. Region behind a vigorous cold front, especially during cyclo-
genesis
6. Squall lines
The first three of these meteorological situations are related to
the observed tendency for high ozone concentration to occur with warm
temperatures and southwesterly winds. Item 4 is a meteorological situa-
tion likely to be characterized by light winds and abundant sunshine.
The final two situations are associated with stratospheric intrusion.
Table 13 summarizes the number of times per month during 1974 that ozone
concentrations in excess of 80 ppb were associated with the these
meteorological situations. The table shows that the six specified si-
tuations account for a substantial majority of the incidents of high
ozone. It should be noted that the meteorological situations thought to
be associated with ozone transport downward from the stratosphere are
involved in many fewer cases than are those associated with the horizon-
tal transport of ozone and/or its precursors.
2.2.2.2.3. Attempts at Objective Comparisons between Weather and Ozone
Patterns
A scheme for the objective comparison of ozone patterns and pres-
sure patterns was described earlier (Section 2.2.2.1.1). Unfortunately,
114
-------
only 42 of the 134 daily pressure patterns could be classified into the
same category, and only eighteen of the ozone patterns fell in a single
category. There was virtually no correspondence between the two groups.
Thus, it was not possible to associate certain ozone patterns with cer-
tain pressure patterns.
The prototype day for the one pressure pattern category was June 4,
1974. It is shown in Figure 64 along with the ozone pattern for the
same day. The prototype ozone pattern occurred on 2 October 1974. It is
shown in Figure 65, with the weather map for the same day. The proto-
type pressure pattern shows that the frequently recurring weather pat-
tern in this sample involved a high pressure cell near the northeast
coast, with air flow from the Gulf of Mexico up into the Midwest. The
most frequently recurring ozone pattern (Figure 65) had high concentra-
tions in the southwest and the eastern parts of the Gulf, with flat gra-
dients and lower concentrations prevailing over much of the rest of the
eastern United States.
Table 13
METEOROLOGICAL FEATURES ASSOCIATED WITH
HIGH OZONE CONCENTRATIONS
(Number of Cases per Month, 1974)
Warm Air Mass
near Front
Warm Sector of
Frontal Wave
West Side of
Anticyclone
Center or East
of Anticyclone
Squall Line
Behind Strong
Cold Front
Other
January
1
0
0
0
0
1
0
February
0
1
1
2
0
0
0
March
6
1
2
2
0
2
1
i-i
•1-1
&
<
5
0
12
4
0
2
6
>%
15
4
11
14
1
1
10
0)
c
•->
8
4
22
15
1
6
15
>,
rH
>-)
19
8
35
44
0
7
15
•U
CO
3
00
3
23
4
30
20
0
10
21
September
9
2
18
11
0
0
7
October
8
2
14
10
0
2
2
November
2
1
0
2
0
2
0
December
0
0
2
1
0
0
0
115
-------
\ W'S'> *^«» ^ V,«^?V--'5r >»-.! 3- 7 J^ 1 >>!
)\"9 - tiv,% * .-^^.^tfvs i -t^—*^ -iP*
<
o
lil
1
LU
t-
QC
O
CC
LU
<
o.
UJ
O
N
O
LU
X
o
I
£
QC
I
LU
QC
0.
O
oc
s
DC
O
116
-------
r~U4 WSW£\^V - "
ft ^tf*v*MBra
*» /r* ».!»-'* 4^.--&f%
<
0
111
<
w
X
cc
o
U.
OC
01
UJ
I
Q
<
Z
CC
LU
LU
O
N
O
LU
DC
0.
LD
(O
LU
DC
a
117
-------
Table 14 Lists the days classified as having pressure and ozone
patterns similar to the prototypes. The only day common to both lists
is May 28. It should be noted that because of the nature of the corre-
lation coefficient, the features within each group show considerable
variation in intensity. The classification system tends to emphasize
the relative positions of high and low features rather than their rela-
tive intensities.
The lack of repetitious patterns does not necessarily mean that
ozone is uncorrelated with atmospheric pressure. The correlations
between the 20 pairs of pressure readings and ozone concentrations were
determined for each of the 134 days. Table 15 shows the results of this
analyses. (The 20 points are shown in Figure 66.)
Table 14
DATES CLASSIFIED AS HAVING PRESSURE
AND OZONE PATTERNS SIMILAR TO THE PROTOTYPES
Dates with
Similar to
1974
April 27
April 28
May 8
May 11
May 14
May 16
May 21
May 28
June 3
June 5
June 6
June 7
June 13
June 14
June 18
June 19
June 28
June 29
July 2
July 6
July 7
July 13
Pressure patterns
that of June 4,
July 17
July 18
July 24
July 25
July 26
August 7
August 12
August 15
August 16
August 19
August 20
August 21
August 25
August 26
September 10
September 11
September 26
September 27
October 11
Dates with Ozone Patterns
Similar to that of October
2, 1974
January 30
May 6
May 7
May 12
May 26
May 27
May 28
June 16
June 27
July 22
July 23
September 20
October 3
October 9
October 19
October 20
October 26
118
-------
FIGURE 66 LOCATIONS OF GRID POINTS USED FOR CLASSIFYING OZONE AND
PRESSURE PATTERNS AND FOR PRESSURE-OZONE CORRELATIONS
119
-------
Table 15
FREQUENCY OF CORRELATION VALUES BETWEEN PRESSURE AND OZONE
AT 20 POINTS IN THE EASTERN UNITED STATES
Correlation
No. of Cases
-0.999
to -0.3
16
-0.299
to -0.1
31
-0.099
to 0.099
44
0.01 to
0.299
26
0.3 to
0.599
16
£0.6
1
At first it may seem surprising that the correlations are not
greater since it is generally noted that high ozone concentrations are
frequently associated with high pressure cells. Indeed, if high-
pressure cells were the only favorable location for high ozone concen-
trations, then the correlations should generally be high. However, as
has been noted already, there are several other types of weather situa-
tion that are also associated with high ozone concentrations. One of
these is the warm air just ahead of a front or in a frontal wave. In
general, these are regions of relatively low pressures, clearly such in-
stances contribute to negative correlations between presuure and ozone
concentration. There are even frequent instances when high ozone con-
centrations are found in association with both high and low pressure
areas on the same day. Figure 67 shows an example of this that oc-
curred on 21 July. One area of high ozone is in the warm sector of the
frontal wave, near the low pressure center that is situated over the
Minnesota-South Dakota border. Another high ozone area is in the high-
pressure ridge over the Texas Gulf coast. It is not surprising that the
correlation between pressures and ozone concentrations at the twenty
grid points was -0.2 on this day.
The highest correlation in the set of 134 cases occurred on 28
April (Figure 68). Surprisingly, this is a case dominated by a high
ozone area ahead of a cold front. The high correlation of 0.64 is the
result of the fact that over the area, (eastern U.S. as whole, both
pressure and ozone generally increase as one moves from the northwest
toward the southeast (except over the southeastern United States itself,
where both gradients are rather flat).
The highest negative correlation in the sample, -0.5, was for 20
September. The ozone and weather maps for this date are shown in Figure
69• A front is seen stretching diagonally from southwest to northeast
across the eastern United States. This front is situated in a low-
pressure trough between two high pressure cells. A ridge of high ozone
concentrations is seen to be just ahead of the frontal trough. This
coincidence of high ozone values and lower pressures accounts for much
of the negative correlation.
120
-------
- •* OVW*\ S \ ^-x\-> °^-» ' ^
-*r s<- PCiJfAi-/ \ ^- V Vw5*>,y.AV ,i
^^i-p4. V Cv-A^^-jr^K
;^s?:f vkv\ v"rK^7-;"'Xl:,xi^ IK *\'
O)
CN
<
a
z
in
IT
O
LL
oo
Q
111
Z
O
N
O
EC
LU
LU
to
LU
DC
D
O
LL
121
-------
^%^
*__/;<,;
- •/N^r^^-SS'iXs*"C-J^*1 \?s i*@* V-^.-j /
, 'v> v$$H>$* vx--v^. ^ * ^O--
Vv1*. lW^£^ ^'\> v< *% ^ v* r^
\;^^J;^^v ^ ^;# • t -v
It^^^^SS^l^ ;p
,. ^«. ,.\^*,»^,, -p^7
^^/
.
^<-! j^T ,' -X . m^_^ , ->x. ^v 'CjC-f.,.^?,, . •««-.«- .
^i^^ ' , \.S&
: a-3 £r-« Ki F-V'
/ ) ~tv ' xriW c ' ^'--t- j«' ,,
Aft^H^f
. ' - f _ ' dPflf -T>« ^^^il^a& ;-*<: fr
^ ; C 'C? jTfTrff!
"
QC
UJ
LU
O
O
CM
00
csl
J
Q
Z
O)
_r
cc.
8
a
z
(A
OC
O
LL
O
H
m
cc
te
UJ
O
N
O
oc
LLJ
UJ
§
oo
CO
HI
DC
2
a
122
-------
S
O)
CC
HI
ffi
LU
(A
CC.
a.
DC
O
a
in
I
O
o
oc
LU
I
O)
ai
cc
123
-------
Singh, Ludwig and Johnson (1977) have examined the day-to-day vari-
ations in pressure and ozone concentration at several remote monitoring
sites in the west. They found that the temporal variations (at fixed
sites) of these two parameters, like their spatial variations discussed
above, are not closely related. Figure 70 shows an example of this for
the site at McCrae, Montana. In this figure, daily pressure values, ob-
tained by interpolation from the daily weather maps are plotted on the
same graphs as peak-hour and daily average ozone values. As can be
seen, the fluctuations in ozone values from day-to-day do not follow
the pressure changes with any consistency.
2.2.3. Combining the Trajectory Approach with Synoptic-Scale Comparison
The trajectory studies indicated that there are preferred air mo-
tions associated with ozone concentrations in the highest 20 percentile
at the various stations. While some directions may be favored over oth-
ers, it is obvious from the trajectories (Figures 35 through 42) that
there is virtually no uniformity. However, some general observations
about the air motions can be made from an examination of the figures.
Probably the most generally applicable of these is that the higher ozone
concentrations are almost always associated with trajectories of clock-
wise curvature. Such motions are often associated with air flow in the
vicinity of high pressure centers. Subsidence and clear skies (hence,
abundant sunshine) are generally associated with such flow.
Cases were chosen for study from among those days for which high
ozone (top 20 percentile) trajectories were constructed for two or more
of the four sites. The trajectories were superimposed on the weather
and ozone maps so that their relationships with both the weather pat-
terns and the ozone distributions could be seen more clearly (Figures 71
through 81). The trajectories represent 60 hours of travel, except
where they extend outside the area covered by the map.
Such depictions for 7 and 8 July, 1974 are shown in Figures 71 and
72. A large high pressure cell dominates the eastern U.S. on both of
these days. The trajectories ending at Wooster, Ohio, and McHenry,
Maryland, show that the winds in this anticyclone had been quite light.
On 7 July, the air arriving at Wooster had spent 2 1/2 days meandering
across central Ohio from Cincinnati, accumulating emissions along the
way. The air arriving at McHenry traveled somewhat farther. On this
same day (7 July), the air arriving at Yellowstone Lake, Wisconsin, had
traveled a clockwise loop from central Illinois. A center of high ozone
is seen northeast of Yellowstone Lake. If the air arriving at this
center had traveled a path parallel to the Yellowstone Lake trajectory.
It would have left the Chicago area 2-1/2 days before.
On 8 July the trajectories ending at Mcilenry and Wooster were simi-
lar to those for the day before, indicating continued stagnation in the
region. The results of this stagnation, and the concommitant accumula-
tion of pollutants, is evident in the widespread high ozone concentra-
tions. In general, the additional day of accumulation caused even
higher ozone concentrations throughout most of the midwest.
124
-------
120
100
o. 80
3
r>
0
60
40
20
0
120
100
80
a
£ 60
n
O
40
20
0
120
100
80
a. 60
JUNE 1974
-A
n
O
40
20
0
i ' i • i " r ' i
DEC. 1974
i ' l ' I ' i
APRIL 1975
NOV. 1974
P
**W
PRESSURE (mb) = P + 930
DAILY 1-HR 03 MAX
FEB. 1975
i • i
i i
MAY 1975
10 15 20 25 30 0 5 10 15 20 25 30
TIME — days
TIME — days
Source: Singh, Ludwig and Johnson, 1977
FIGURE 70 OZONE-PRESSURE RELATIONSHIPS AT McRAE, MONTANA
125
-------
gy ^T-^m-^ry4/%^^v% • '.-if'^I/ "fi
if-4f;lf% I^WxI^^W' V^W V
tVs^/v&C NCfrfV ^ ^ W^ yr /
w^^te^A il •* \ 1 y /-:
^^
'^vj^'v 11^ f*!
i-^^^rf^^^i!
^^x^^P^f ^^
^ W^. ,^ 1-, SI i\ £ .^' - *
i-
(3
§
S3
1-5
JH"
P
z
D
DC
O
U.
CO
UJ
DC
O
O
UJ
3
cc
t-
o
z
<
±>
m
Q
HI
O
N
O
cc
UJ
I
<
111
UJ
DC
126
-------
UJ
cc
M
i z> ec
2 ^ UJ
*il
<(*—**-
lie;
Y>>/4T
^pJ-^iV*
/fe-4 J
"M.-5^' ?»o .. «r.
r? ##* irtfn **f #*\
Utr^M V tf** V
;* f?>*VP
^
\m%4'V v ^< A* ) '^.^5
- /f %&* •.Sijfc '^ I ' * ' ti\ -
,^^^^%,%v\
'^.-VteP&v^" V -\
• /•?*• •» v • •«. ••. *. ^ •
^fe^d
'"A ^\ tr ^ / ^ ^\ -4 &- ^/
' \ = \i? t> ^;'fe^r-^o^
rX
\f*^;! n^^1' ?'
$$$**%$$
li0&yMJt^;,
4Af
cc
UJ
X
111
3
O
S
at
oo
-------
Figures 73 through 77 depict the period 10 July through 14 July
1974. The high pressure area that led to the stagnation seen in the
July 7-8 cases was no longer present and the three trajectories shown
arriving at Queeny, Wooster, and Mcllenry on 10 July (Figure 73) show
much longer travel, generally arriving from the southwest ahead of an
advancing cold front. Although ozone levels exceeded the federal stan-
dard in many areas, they were well below those observed during the stag-
nation episode two days earlier. The air arriving at McHenry on July 10
had moved generally up the Ohio River Valley and had apparently accumu-
lated enough precursor emissions to cause ozone concentrations in excess
of 100 ppb. The other two locations, Wooster and Queeny, did not have
concentrations quite so high.
By 11 July (Figure 74), the cold front had passed McHenry. The
cleaner air moving in from the north caused concentrations in the
Maryland-West Virginia-Pennsylvania area to be lower. Queeny was still
under the influence of warm air arriving from the south and appears on
Figure 74 to have been near a center of quite high ozone concentrations.
However, the 173 ppb observation at this center seems to be unrepresen-
tative; other values nearby are nearer 75 ppb.
By 12 July, a new high pressure area had been established in the
upper midwest, as shown in Figure 75. The air arriving at Wooster and
McHenry had come from Canada. The trajectory arriving at Yellowstone
Lake had come from far to the south during the preceding 2-1/2 days. It
would have been interesting to investigate the source of the air with
the high ozone concentrations around St. Louis, but no trajectory had
been costructed for that area for this day.
The air arriving at Wooster and McHenry on 13 July (Figure 76) had
covered much shorter distances during the preceding 60 hours than had
the air that arrived at the same locations a day earlier. As might be
expected, the ozone concentrations increased. Concentrations exceeded
140 ppb just south of Toronto. On 14 July, the high pressure area per-
sisted and thus the air movement remained quite slow. As the trajec-
tories in Figure 77 show the air at Wooster and McHenry had come from
the west rather than from the north as on the preceding day. Judging
from the Wooster and McHenry trajectories, the air in the region of very
high ozone concentrations (around 200 ppb) probably had come from De-
troit and southern Michigan.
As shown in Figure 78, 18 July was a day on which the eastern Un-
ited States was dominated by a large high pressure center off the east
coast. The air arriving at McHenry had spent the preceding 2-1/2 days
traveling in an almost complete circle over West Virginia. The air in
Wooster had traversed a path of similar shape over Ohio. The air at the
high ozone area north of McHenry had probably circled over Pennsylvania.
By contrast, the air that arrived at Queeny and Yellowstone Lake had
moved much greater distances, traveling northward ahead of an appproach-
ing cold front.
128
-------
o>
DC
O
LL
cn
LU
cc
o
o
o
z
<
m
Q
LU
•z.
o
N
O
cc
LU
CO
LU
CC
D
O
129
-------
Sr^S1-* *
IV &.'$*• \tVA'
,CI/f^,^. i ^
TV«V^
*V-^9< ;*r Sv ^
^1'. ^-OS^ 4l
S
o>
>-
D
tr
O
u.
(O
UJ
O
LU
CC
O
CO
Q
LU
O
N
O
o_~
S
CC
UJ
5
UJ
CC
D
(9
130
-------
- - ^,v&, 3
\^*V4:
w
_J^*? X. I.J9V. "^-^ '• yfr |j- ' rji- s,gS»$
^%4^ ^5^1
ra^v^^iN^
M" k%M^Aifc; ,r
Q.
<
cc
Ul
X
<
UJ
g
I-
o
8
CM
OC
O
u,
to
UJ
E
I
UJ
cc
I-
o
<
o
F
CO
UJ
I
O
cc
Ul
in
r«.
UJ
c
o
K
u,
131
-------
00
<0,^& °^«, ^V ?' 1 / ) '
I \ ^^ V*S« T^*'1* ' ' * »V* 1$! • / / -
$^wfc*
n*^^^*^
\ ^V v> "'* ii^P^^i ?^ \ »T
#> V^\|?%>%
'• V\'.P» (V«\ nft->4>- 'tor', Si
\^
,0V^
^^
f" 'f-S
\X n\
j^\ H+ . *:
tfitt,?.^v$
OL
<
CC
111
X.
<
ID
C3
O
O
CM
m
oc
g
O
LU
•z.
o
N
O
oc
UJ
I
UJ
CO
ID
OC
o
132
-------
KWfc-S'-.i
V * »*& .1 ?•» v
en
cc
O
CO
LU
DC
O
o
LU
O
g
h-
m
LU
O
a."
<
oc
LU
LU
LU
CC
133
-------
l,^^ft^^^
^fe*MvA *%
y^~^i^\sA, W?,^,. '^-
s
a>
CO
cc
o
LU
c/>
LU
E
o
d
LU
3
cc
I-
o
<
z
o
H
D
00
cc
O
LU
1
O
cc
LU
LU
00
r*
LU
CC.
D
g
LT
134
-------
A high pressure area was over the Great Lakes on 21 July 1974 (Fig-
ure 79). This day was discussed earlier (Section 2.2.2.2.3) in connec-
tion with Figure 67 and the correlation between ozone concentration and
pressure. Weak pressure gradients and light winds prevailed over Mis-
souri, Kansas, Arkansas, and Oklahoma. The air arriving at Queeny had
meandered from the southwest rather slowly with these light winds. The
air at Wooster and Yellowstone Lake had traveled greater distances from
the north; ozone concentrations at these locations were somewhat higher
than the federal standard.
The weather pattern for 22 July, 1974 (Figure 80) shows weak pres-
sure gradients and light winds over much of the eastern United States.
Uzone concentrations had increased in much of the area in response to
the accumulations of pollutants in this slowly moving air. Figure 80
shows the trajectories of the air arriving at Yellowstone Lake, McHenry
and Queany. Judging from the trajectory arriving at McHenry, the air in
the high ozone area south of Lake Erie had passed near Toronto, Buffalo,
Cleveland and other Ohio cities, before arriving in southern Ohio.
The maps for 26 July 1974 are shown in Figure 81. This was another
day of weak pressure gradients and light winds. The Wooster and Queeny
trajectories both exhibit the rather short circular character seen in
the other weak gradient cases. The Uooster trajectory spent 2-1/2 days
over Ohio, terminating near the center of a high ozone area with concen-
trations in excess of 140 ppb.
135
-------
mX^X'
*-y_Un\••,
~i£? *• *>, v,- >.
W*^m'*fc/t>
QC
o
u.
CO
LU
OC
I
LL1
OC
I-
o
<
g
i-
m
oc
»
Q
LU
z
o
N
O
oc
LU
r
LU
O)
r^
LU
OC
D
O
136
-------
•"x&Mt.k ^
K
,>^m¥Witm
«e*gfe£^-~#.:J$
*3V» ^^
^
Mt- -4
!*iV
••w
^n
a.
<
oc
ai
I
<
uj
g
o
o
CM
^?«fi i -i fj£«iJ ' V H i ' "'• ' iK -5? f ? «-^
^w* l_ | l*5Ts i .Vi\ of, i s ; €V ;<^! !^
V^gfe: € 4w9 ^r>A. FV '\^ ..'
Q.
<
2
oc
LU
I
<
LU
? §
2 LU
« OC
<
Q
I
137
-------
8
cc.
o
LL.
(O
UJ
E
o
UJ
3
QC
I-
o
OQ
cc
«
o
o
N
O
cc
LU
UJ
CO
UJ
cc
D
o
138
-------
3. DISCUSSION
This section is an attempt to draw together the results presented
in the preceding section and to compare them with the results of other
studies. Many of the more important studies of tropospheric ozone have
been reviewed in Volume II of this report. That review was prepared and
published during the early phases of this project. In the meantime,
several other important studies have been completed and will be referred
to here. An important feature of much of the recently published infor-
mation is its consistency. There is considerable agreement concerning
the importance of the stratospheric contribution to tropospheric ozone,
the extent of long range transport within the troposphere, and the
meteorological factors that have the greatest influence on ozone produc-
tion and accumulation.
Any review of results requires a vantage point. In this case the
vantage point is that of the control strategist assessing the difficulty
of his task. That strategist is attempting to answer the same questions
that this study has attempted to answer. Those questions are:
1. How much does ozone of stratospheric origin contribute to
ground-level oxidant concentrations?
2. What are the anthropogenic causes of high oxidant values; that
is, how do HC and NOx emissions interact with weather features
to cause high oxidant concentrations?
3. What are the implications of these answers to control stra-
tegies -- is it possible to define geographic regions where
unified oxidant control strategies are feasible? If so, how
can these regions be defined?
4. What new research is needed?
Another question, that of the possible role of geological and biological
precursor emissions, was not addressed in this study. The analysis
described in the preceding section have provided partial answers to the
other questions. The results of other recent studies serve to make some
of these answers more nearly complete.
3.1. How Much Does Stratospheric Ozone Contribute to Ground-Level Ozone
Concentrations?
3.1.1. General
There are two answers to the question of stratospheric impact and
both are of importance to the strategist. First, there is the question
of the long-term average contribution of stratospheric ozone to ground-
level concentrations. This is the baseline, the foundation that tells
the strategist what would be left if all anthropogenic, geologic, and
biological sources of precursors were removed. It is a generally ir-
reducible reservoir that the strategist must recognize.
139
-------
The other answer to the stratospheric question is the one that ad-
dresses short-term effects. Does ozone ever travel from the strato-
sphere to ground-level while undergoing so little dilution that it ar-
rives at concentrations greater than the federal standard? The answer
to this question will determine whether there are some "violations" of
the standard that are completely beyond the control of the strategist.
3.1.2. Long-Term Average Stratospheric Contributions
The analyses of radioactive fallout data suggest that the seasonal-
ly average stratospheric contribution to ozone near ground level is on
the order of 15 ppb during springtime. This is in qualitative agreement
with the values presented by Fabian and Pruchniewicz (1976). They found
seasonally averaged concentrations in the free troposphere of about 20
ppb in mid-latitudes during the spring and early summer. They attribute
most of this to stratospheric origins, but concede the possibility of
other contributions. Singh, Ludwig and Johnson (1977) indicate even
greater natural, probably stratospheric, contributions. Thus, the evi-
dence points to a significant stratospheric contribution to seasonally
averaged, ground-level ozone contrations in the middle latitudes of the
northern hemisphere. This contribution may be 20 ppb or more in the
springtime. The evidence provided by radioactive fallout suggests that
the effects are most pronounced near the typical locations of the polar
and arctic jet streams. Fabian and Pruchniewicz (1976) reached the same
conclusion from observations of ozone in the eastern hemisphere. They
also suggested that the subtropical subsidence areas were also regions
of important stratospheric ozone contributions to tropospheric ozone
concentrations.
3.1.3. Short-term Stratospheric Contributions
One thing is certain about short-term stratospheric contributions
to ground-level ozone concentrations—they do not cause the federal oxi-
dant standard to be exceeded frequently. The analyses presented here
suggest that stratospheric contributions exceeded the federal standard
only 0.2 percent of the time in the lowest few hundred meters of the
troposphere.
Nevertheless, there have been incidents in which stratospheric
ozone reached ground-level very directly and with much less than the
usual dilution. The Zugspitze results show clear evidence of such an
event at a high altitude (3000 m) ground-level site. Lamb (1976) makes
a very strong case for stratospheric intrusion as the cause of hour-
average ozone concentrations as high as 230 ppb in Santa Rosa, Califor-
nia, during the predawn hours of 19 November, 1972. The incident re-
ported by Lamb lasted less than 5 hours and was restricted to an area
with dimensions of only a few tens of kilometers.
As with the Zugspitze case, the Santa Rosa incident was associated
with a a rather special set of meteorological circumstances. In Santa
Rosa, a classical stratospheric intrusion event brought ozone in high
concentrations to the middle and upper troposphere where it was were
then "tapped" by the downdrafts associated with a convective shower.
This is similar to the mechanism hypothesized in Section 2.1.1.4 for the
140
-------
high ozone concentrations observed over Tallahassee, Florida on 14 Au-
gust, 1963.
Danielsen and Mohnen's (1976) analyses of stratospheric penetration
behind the jet stream over the United States during April 1975 show con-
centrations of 200-300 ppb at altitudes of about 6-8 km. While these
data confirm the importance of stratospheric intrusions in transferring
ozone from the stratosphere to the troposphere, they are not examples of
ground-level effects. Such intrusions can seriously affect ground-level
concentrations only in special circumstances when convective circulations
bring these high concentrations to the ground via a rather direct route.
It must be concluded that direct penetrations of stratospheric air
to ground-level and the consequent high ozone concentrations are rare
events. As is true of all rare events, it is difficult to accumulate
reliable statistics concerning their frequency. Nevertheless, the real-
ity of such incidents must be considered in the interpretation of ozone
records and in the formulation of control strategies.
3.2. What are the Tropospheric Causes of High Ozone Concentrations?
3.2.1. Meteorological Factors
It is clear from the tropospheric analyses that meteorological fac-
tors play a very important role in determining the ozone concentrations
at locations that are removed from urban areas. The results show that
those atmospheric conditions that are associated with warm temperatures
are also conducive to the formation of ozone. Meyer et al. (1976) have
obtained very similar results from studies that used techniques very
much like those used here. They found that recent air temperature was
the parameter most highly correlated with ozone concentrations.
The correlation between ozone concentrations and the temperature of
the air parcel has been shown to be a reflection of a complete set of
meteorological factors that combine to provide the conditions conducive
to the photochemical production of ozone from its precursors. As a
result, it is possible to identify those meteorological patterns where
ozone buildups are likely.
The results presented here are in good agreement with the findings
of others. It is generally agreed that high pressure areas are favored
locations for the accumulation of high ozone concentrations in the
eastern United States (e.g. EPA, 1975; Bach, King and Vukovich, 1976;
Wolff et al., 1976). The low wind speeds and the vertical mixing in the
vicinity of high-pressure centers allow precursors to accumulate while
the prevailing fair skies and abundant sunshine promote photochemical
activity. If temperatures are also high, this appears to further
enhance the formation of ozone.
Under some circumstances, the same combination of conditions is
found in areas other than high-pressure areas. For instance, fair skies
and warm temperatures are found in prefrontal warm air, but light winds
and stable stratifications (which limit the vertical mixing) are not as
common as in the high pressure areas. It appears that a build-up of
-------
ozone can nevertheless take place in those instances in which the air
spends considerable time over high-emissions areas. This is exemplified
by prefrental air flow from the southwest along the east coast urban cor-
ridor. Ludwig and Shelar (1977) have presented some rather dramatic ex-
amples of ozone buildups in prefrontal warm air over New England with
subsequent abrupt drops in concentration as cleaner, cooler air moves in
with the windshift behind the front.
3.2.2. Emissions Factors
If one carefully reviews the meteorological factors found to be im-
portant to the photochemical formation of ozone in the troposphere, it
becomes apparent that these same conditions are conducive to the accumu-
lation of high concentrations once it is formed. This set of conditions
is optimized near high pressure centers. Furthermore, the high pressure
centers with their light and variable winds are areas in which the cal-
culated trajectories are least reliable. Thus, the path of the air is
most difficult to trace during precisely those situations when any pre-
cursors that are introduced are most likely to produce high ozone con-
centrations. This may explain why the contributions of precursors were
not better defined by the statistical analyses of conditions along the
trajectories.
Meyer et al. (1976) found in their trajectory analyses that for
those stations that were near major emissions areas, ozone concentra-
tions were significantly correlated with the hydrocarbons introduced
into the air during the last several hours of the trajectory. They con-
clude that local hydrocarbon emissions are important to ozone formation
near cities.
Other studies have established the influence of urban sources on
ozone concentrations at fairly great distances from the cities. Cleve-
land et al (1976) suggested that the influence of the New York area em-
issions may extend for as much as 300 km downwind. Analyses by Ludwig
and Shelar (1977) identify ozone "plumes" from the New York area 100 km
downwind. The Boston ozone plume was observed by Zeller et al. (1976)
at distances of 50 or more kilometers from the city. These are by no
means isolated observations. It can be assumed that identifiable urban
effects may extend for distances of 100 km or more beyond major source
areas. However, under optimum ozone formation/accumulation conditions,
these distances are probably reduced by the low speed of the winds.
Furthermore, the more erratic air motions during the optimum meteorolog-
ical conditions reduce the likelihood of definitively following the path
of the pollutants.
Before leaving the subject of emissions, it should be noted that
the methods by which the emissions were estimated suffers some shortcom-
ings that the measures of meteorological conditions do not. The first
of these is the use of total hydrocarbon (THC) emissions data from the
NEDS countywide inventories. Total hydrocarbon emissions may be poor
representatives of those hydrocarbons which serve as precursors in ozone
formation processes. Also, the NEDS inventories do not include natural
sources. Although control strategies must treat anthropogenic emis-
sions, better understanding of the mechanisms operating to form ozone
would be achieved if estimates of natural emissions were available.
142
-------
A definite drawback in the statistical analysis of the trajectories
was the poor spatial resolution of the emissions data (county-wide aver-
ages) coupled with the uncertainty as to the actual paths that the air
had taken. A change in the calculated position of an air trajectory of
a few tens of kilometers could cause the emissions from a major source
county to be included or excluded depending on where the trajectory is
placed. Regression and correlation analyses are quite sensitive to such
changes. By contrast, the meteorological observations tend to be
representative of much larger areas and are therefore much less sensi-
tive to uncertainties in the trajectory.
3.3. Implications for Control Strategies
3.3.1. Interactions Between Ozone of Stratospheric and Tropospheric
Origin
The analyses show that stratospheric effects are not negligible.
There are long-term tropospheric levels of stratospherically originated
ozone of 25 to 30 percent of the federal standard. Ozone concentrations
in the lowest two kilometers of the atmosphere may be in excess of
federal standards perhaps 0.2 percent of the time because of individual
intrusion events. But concentrations above the standard at ground-level
are apt to be less frequent, except on mountain peaks, because of mix-
ing and surface destruction processes. Undoubtedly, rare circumstances
do result in concentrations of stratospheric ozone in excess of the
federal standard for short periods at elevations more typical of popu-
lated areas.
All the evidence indicates that spring is the period for the most
pronounced stratospheric interaction with the troposphere over most of
the United States. From the strategic standpoint, this is fortunate be-
cause it means that the periods of maximum stratospheric effect and max-
imum trophospheric photochemical ozone production do not coincide. Ana-
lyses of ozone data from remote locations in the western United States
(Singh, Ludwig and Johnson, 1977) indicate that background ozone concen-
trations reach their highest levels (40 to 80 ppb) in late winter and
spring at the most remote sites. Stratospheric effects are postulated
as an important factor in this behavior. At some of the less remote
stations the high concentrations continue into summer and early fall.
It is presumed that local photochemical ozone production at these sites
more than offsets the declining stratospheric contribution. Observed
increases in NOx concentrations (of unknown origin) in late spring and
summer at these same sites tend to support this view.
The likely locations for stratospheric intrusions generally fall
outside the areas where low altitude accumulations of photochemical oxi-
dant are most likely. The tropospheric analyses have shown that the
warm, clear conditions associated with high pressure cells or the warm
conditions just ahead of cold fronts are most conducive to tropospheric
ozone formation. On the other hand, stratospheric studies suggest that
the area just behind a fast moving cold front, accompanying an outbreak
of polar or arctic air, is a more likely area for a significant intru-
sion of stratospheric ozone. The meteorological conditions suitable for
the tropospheric buildup of ozone are quite different than those accom-
143
-------
panying stratospheric intrusions, so the two processes are unlikely to
be operative at the same place and time.
Overall, it does not appear that stratospheric ozone and tropos-
pheric ozone are apt to occur at high concentrations in combination.
Nevertheless, the possibility of high ozone concentrations from the
stratosphere does have important consequences for the formulation of
control strategies. Strategies are often formulated so that if they had
been in effect the standards would not have been violated during some
specific historical incident, e.g. for the day of the second highest
concentration for some past year. If the specific incident for which an
oxidant strategy is designed involved important ozone contributions from
the stratosphere, then the strategy would be faulty in assuming that
controls on anthropogenic emissions are suitable for attaining standard.
Obviously, strategy formulation should involve the examination of indi-
vidual cases to assess the importance of stratospheric contributions.
3.3.2. Strategies for Control of Ozone of Anthropogenic Origin
The studies reported here have emphasized regions that are well
removed from urban areas. The rationale has been that such locations
could be used to determine the longer-term processes that govern the
larger scale features of ozone patterns. These larger-scale features,
while important, cannot be considered to the exclusion of the smaller
scale processes that affect ozone concentrations near urban areas. As
noted before, urban emissions are known to influence concentrations for
distances of a hundred kilometers or more from their centers (see e.g.
Martinez and Meyer, 1976).
EPA (1977) has pointed out that the present strategy of controlling
hydrocarbon emissions in order to reduce ozone concentration is effec-
tive only where the ratio of the concentration of non-methane hydrocar-
bons (NMHC) to that of oxides of nitrogen (NOx) is less than about 30:1.
Low ratios are found in and near cities or large NOx point-sources. As
a general rule the ratios are quite high in rural areas, primarily be-
cause, as Meyer (1977) points out, the concentrations of NOx are very
low.
The above facts could explain why Meyer et al. (1976) found close
relationships between ozone and hydrocarbon emissions for stations near
cities, but not at more remote sites. Similarly, the results of the
present study can be explained on this basis. Using the fact that ozone
concentrations are closely related to hydrocarbon emissions where the
NMHC/NOx ratios are high, Meyer (1977) proceeded to delimit the dis-
tances from urban regions where such controls might be effective. Using
average per capita emissions data, he proposed that the effectiveness of
controlling hydrocarbon emissions would extend about 20 miles outward
from the center of a city of 200,000 people and about 85 miles out from
a city of 4 million.
Figure 82, taken from Meyer's paper (1977), shows the areas that
are within specified distances (related to population) of major urban
144
-------
centers. According to Meyer's proposal, these are the areas in which
oxidant concentrations are largely controllable by controlling hydrocar-
bon emissions. It is interesting to note the close correspondence of
the shaded areas in Figure 82 to many of those areas where widespread
violations of ozone standards are found to occur. Another important
feature of Figure 82 is the generally southwest to northeast alignment
of the overlapping areas along the east coast, in the Ohio River Valley
and along the shores of the Great Lakes. This alignment almost certain-
ly affects the frequency with which southwest winds and high ozone con-
centrations are seen to be concurrent in these regions.
Dodge (1976) has shown that NOx concentrations control ozone con-
centrations when NMHC/NOx ratios are high. As already mentioned, Meyer
(1977) showed that high ratios prevail in rural areas. Thus, the find-
ing of the present study—that rural ozone correlates better with NOx
emissions than with hydrocarbon emissions—and the the similar results
obtained by Meyer et al (1976), is not surprising. Singh, Ludwig, and
Johnson (1977) observed that, in the summer months, concentrations of
ozone in remote locations appear to be controlled by the presence of ox-
ides of nitrogen. All these results suggest that the control of oxidant
in rural areas may require a quite different strategy than is applicable
to urban areas and their environs.
For the rural areas, control of the contributing NOx emissions is
clearly the strategy of choice. The correlations of ozone with NOx em-
issions are significant over long periods of time that correspond to
travel distances of four or five hundred kilometers for the very high
ozone concentrations. So it appears that the control of rural ozone
will require that much larger geographic regions be considered than for
the urban situation. The implications are that the regions where con-
trol of NOx could be required might extend hundreds of kilometers upwind
of the areas where the violations occur.
Great care would have to be taken in the design of such strategies
however, because the reduction of NO emissions over areas with dimen-
sions of several hundred kilometers might well lead to higher ozone con-
centrations within the urban environs. Probably, greater emphasis
should be given at present to hydrocarbon control in urban areas than to
NOx control in rural areas because of population-exposure considera-
tions. However, it does seem clear that the continued siting of large
NOx sources in rural areas would tend to aggravate the present oxidant
problem in those areas. Meyer (1977) has pointed out that large por-
tions of the eastern U.S. are subject to prolonged summertime periods
of stagnation during which large concentrations of hycrocarbons can ac-
cumulate from the multitude of sources in the area. Introduction of
large amounts of NOx into such an air mass would probably be undesirable
in the rural areas.
145
-------
3.4. Recommendations for Future Research
It is important that better statistics be developed to define the
importance of stratospheric intrusion. Initially, such investigations
should use existing data from the SAROAD system and from past special
studies. Those geographic areas that have been identified as undergo-
ing the highest impact should be emphasized. For instance, cases of
anomolously high concentrations in the western portions of Kansas, Ne-
braska and Okalahoma should be studied individually, using isentropic
trajectory analyses to identify instances of stratospheric influence.
Other regions having episodes with high probability of stratospheric in-
fluence should be treated similarly.
Since stratospheric ozone penetrates to within a few kilometers of
the surface much more often than it does to near-ground-level, the mix-
ing and ozone destruction processes in the lowest layers are quite in-
fluential in determining stratospheric influence at ground-level. Past
aircraft studies of ozone aloft could be used to investigate these im-
portant processes. In such investigations, the ozone need not be of
stratospheric origin. Often there are layers of ozone aloft that ori-
ginated at the surface (see e.g., Miller and Ahrens, 1970; Ludwig and
Shelar, 1977). The behavior of these layers may, by analogy, provide
information on the extent to which ozone concentrations in elevated
layers introduced by stratospheric intrusion are reduced during mixing
to ground-level.
Meyer's radii (Figure 82) can be refined by a consideration of
weather effects. That is, the circular areas might be better defined as
asymetric areas, where the asymetry is governed by the nature of those
weather conditions known to be most likely to accompany high ozone con-
centrations. The 1976 data from the Regional Air Pollution Study (RAPS)
in St. Louis is now available and could be used to test this hypothesis.
Data from other relatively dense monitoring networks like those in Los
Angeles, San Francisco, Chicago and Southern New England might also be
used in such a study.
Where possible, ozone records from rural areas should be examined
in combination with HOx records at the same locations. If ozone concen-
trations in nonurban areas are in reality related to NOx emissions, then
there may be evidence of this in the monitoring data. Unfortunately
routine NOx measurements generally lack the threshold sensitivity neces-
sary for such a determination. Data from past special studies may be
found to be better in this respect than routine monitoring data.
There are theoretical investigations that would be useful to the
formulation of control strategies for reducing ozone levels in nonurban
areas. Existing computer simulation models might be used to evaluate
the effect of adding nitric oxide to the well aged remnants of urban em-
issions. This would help to define the impact (both initial and net)
of some major NOx sources such as power plants in rural areas. Data
from field studies, in combination with the model results, would also
be useful to the evaluation of such impacts.
146
-------
147
-------
In summary, it is recommended that existing routine data and data
from past special studies be used to 1) investigate the importance of
stratospheric intrusions, 2) determine the extent oi the direct urban
influence on ozone concentrations in the urban environs, and 3) investi-
gate the more widespread nonurban ozone problem. It is expected that
considerable information can be extracted from existing data. Further-
more, any studies using existing data will provide a strong basis for
the design of future data collection projects so that they will fill the
gaps in the existing data base most effectively.
Finally, in the present study, the selection of cases for the con-
struction of trajectories was made before the daily ozone maps were
drawn. In retrospect, it would have been better to have performed the
studies (i.e., the trajectories and the map-comparisons) in the reverse
order. This would have made possible a more judicious selection of the
cases so that trajectories could have been constructed for those cases
that appeared from the map comparison to have the greatest likelihood of
yielding useful results. The cases, discussed earlier, for which it was
possible to compare trajectories, weather maps, and ozone patterns, sug-
gested that more such analyses would be quite productive. This is
therefore recommended.
148
-------
REFERENCES
Bach, W.D., W.J. King and P.M. Vukovich, 1976: Nonurban ozone concentra-
tions in transient high pressure systems. Presented at the Nonur-
ban Troposphere Symposium, Sponsored by Amer. Geophys. Un. and
Amer. Met. Soc., Hollywood, Fla., 10-12 November.
Cleveland, W.S., B. Kleiner, J.E. McRae, and J.L. Warner, 1975: The
analysis of ground-level ozone data from New Jersey, New York, Con-
necticut, and Massachusetts; Transport from the New York City
Metropolitan Area, Mimeo Report, Bell Laboratories, Murray Hill,
New Jersey, 65 pp.
Cleveland, U.S., B. Kleiner, J.E. McRae, and J.L. Warner, 1976: Photo-
chemical air pollution: transport from the New York City area into
Connecticut and Massachusetts. Science, 191, 179-181.
Danielsen, E.F. and V.A. Mohnen, 1976: Project Dustorm R.eport: Ozone
measurements and meteorological analyses of tropopause folding.
Proceedings Joint Symposium on Nonurban Tropospheric Composition.
Amer. Geophys. Un. and Amer. Met. Soc., Hollywood, Fla., 10-12 No-
vember 1976, pp. 17-1 to 17-13.
Dodge, M.C., 1976: Combined use of modeling techniques and smog chamber
data to derive ozone precursor relationships. Presented at Inter-
national Conference on Photochemical Oxidant Pollution and Its Con-
trol; Raleigh, N.C., Sept. 13-17.
Dutsch, H.U., 1971: Photochemistry of atmospheric ozone. Adv. Geophys.,
15, 219-322.
Environmental Protection Agency, 1975: Control of photochemical
oxidants—technical basis and implications of recent findings. Re-
port No. 450/2-75-005, 35 pp.
Fabian, P. and P.G. Pruchniewicz, 1976: Final Report on Project "Tropos-
pharischea Ozon", Max-Planck-Institut fur Aeronomie, Report MPAE-
W-100-76-21, Kutlenburg-Lindau, Germany, 28 pp.
Freiling, E.G., G.R. Crocker and C.E. Adams, 1965: Nuclear-debris forma-
tion. In: Radioactive Fallout from Nuclear Weapons Tests (A.W. Kle-
ment, Jr., Ed.), U.S. Atomic Energy Commission, CONF-765, pp. 1-43.
Heffter, J.L. and A.U. Taylor, 1975: A regional-continental scale tran-
sport, diffusion and deposition model, Part I: Trajectory model.,
National Oceanic and Atmospheric Administration Technical Memoran-
dum, ERC ARL-50, pp. 1-16.
Bering, U.S., 1964: Ozonesonde observations over North America, Vol. 1,
Research Report, Report AFCRL-64-30(l), pp. 1-512, Air Force Cam-
bridge Research Laboratries.
149
-------
Herlng, W.S. and T.R. Borden, Jr., 1964: Ozonesonde observations over
North America, Vol. 2, Report AFCRL-64-30(II), Air Force Cambridge
Research Laboratories.
Hering, W.S., and T.R. Borden, Jr., 1965a: Ozonesonde observations over
North America, Vol. 3, Report AFCRL-64-30(III), Air Force Cambridge
Research Laboratries.
Hering, W.S., and T.R. Borden, Jr., 1965b: Mean distribution of ozone
density over North America, 1963-1964. Report AFCRL-65-913, Air
Force Cambridge Research Laboratory.
Hering, W.S., and T.R. Borden, Jr., 1967: Ozonesonde observations over
North America, Vol. 4, Report AFCRL-64-30(IV), Air Force Cambridge
Research Laboratories.
Holzworth, G.C., 1972: Mixing Heights, Wind Speeds, and Potential for
urban air pollution throughout the contiguous United States. En-
vironmental Protection Agency, Office of Air Programs, Pub. No.
AP-101, 118 pp.
Lamb, R.G., 1976: A Case Study of Stratospheric Ozone Affecting
Ground-Level Oxidant Concentrations. Systems Applications, Inc.,
San Rafael, California, 18 pp.
Langley, R., 1970: Practical Statistics Simply Explained. Dover Publi-
cations, Inc., New York. 399 pp.
List, R.J. and K. Telegadas, 1969: Using radioactive tracers to develop
a model of the circulation of the stratosphere. J. Atmos. Sci.,
26, 1128-1136.
Lovill, J.E., 1969: Transport processes in orographically induced gravi-
ty waves as indicated by atmospheric ozone. Colorado State Univer-
sity Atmospheric Sciences Paper No. 135, 78 pp.
Lovill, J.E., 1974: Contribution to Chapter 2, in Reiter et al., 1975.
Ludwig, F.L. and E. Shelar, 1977: Ozone in the Northeastern United
States. Final Report, EPA Contract 68-02-2352, prepared by Stan-
ford Research Institute, Menlo Park, California. EPA Report No.
90/9-76-005, 286 pp.
Lund, I., 1963: Map classification by statistical methods. J. Appl.
Meteorol., 2, 56-65.
Mahlman, J. D., 1965: Relation of Tropopause-level Index Changes to Ra-
dioactive Fallout Fluctuations. Colorado State University, Atmos-
pheric Science Technical Paper No. 70, pp. 84-109.
Martinez, E.L. an E.L. Meyer, 1976: Urban-nonurban ozone gradients and
their significance. Presented at Air Pollution Control Association
Technical specialty Conference on Ozone/Oxidants: Interactions with
150
-------
the Total Environment; Dallas, Texas, March 12.
McQuirk, J.P., E.R. Reiter and A.M. Barbieri, 1975: On the variability
of hemispheric scale energy parameters. Colorado State University
Environmental Research Papers, No. 1, 15 pp.
Meyer, E.L., W.P. Freas, J.E. Summerhays and P.L. Youngblood, 1976: The
use of trajectory analysis for determining empirical relationships
among ambient ozone levels and meteorological and emissions vari-
ables. Paper presented at International Conference on Photochemi-
cal Uxidant Pollution and Its Control, Raleigh, N.C., 12-17 Sep-
tember.
Meyer, E.L., 1977: Establishing organic emission control strategies as a
function of geographic location. EPA Office of Air Quality Planning
and Standards, 36 pp.
Reiter, E.R., 1971: Atmospheric transport processes; Part 2: Chemical
tracers. U.S. Atomic Energy Commission, Critical Review Series,
TID-25314, 382 pp.
Reiter, E.R., 1972: Atmospheric transport processes; Part 3: Hydrodynam-
ic tracers. U.S. Atomic Energy Commission, Critical Review Series,
TID-25731, 212 pp.
Reiter, E.R., 1975a: Atmospheric transport processes; Part 4: Radioac-
tive tracers. U.S. Energy Research and Development Administration,
in press.
Reiter, E.R., 1975b: The transport of tracers by global circulation sys-
tems of the atmosphere. U.S. Energy Research and Development Ad-
ministration Report COO-1340-46, 12 pp.
Reiter, E.R., 1975c: Stratospheric-tropospheric transport processes and
their relation to the influx of ozone into the troposphere. Report
to Stanford Research Institute, 12 pp.
Reiter, E.R., 1975d: Stratospheric-Tropospheric exchange processes.
Rev. Geophys. Space Phys., 13, 459-474.
Reiter, E.R., 1976a: Atmospheric transport processes, Part 4: Radioac-
tive Tracers. U.S. Energy Research and Development Administration,
in preparation.
Reiter, E.R., 1976b: Ozone concentrations in the lower troposphere as
revealed by ozonesonde observations. Report to Stanford Research
Institute, 26 March, 1976, 106 pp.
Reiter, E.R., E. Bauer and S. Coroniti, 1975: The natural stratospheric
1974. CIAP Monograph No. 1, U.S. Department of Transportation.
Seitz, H., B. Davidson, J.P. Friend and H.W. Feely, 1968: Final Report
on Project Streak. Numerical models of transport diffusion and
fallout of stratospheric radioactive material. Isotopes, Inc.,
151'
-------
Westwood, N.J., Report No. NYO-3654-4, 97 pp.
Singh, H.B., F.L. Ludwig and W.B. Johnson, 1977: Ozone in remote loca-
tions. Final Report, prepared for Coordinating Research Council,
Stanford Research Institute, Menlo Park, California, 157 pp.
Volchok, H.L., and L. Toonkel, 1974: Worldwide deposition of Sr90
through 1973, U.S. Atomic Energy Commission, Report HASL-286, pp.
1-17 to 1-36.
Wallace, J.M., R.E. Newell, 1966: Eddy fluxes and the biennial stratos-
pheric oscillation. Quart. J. Roy. Meteorol. Soc. , 92,(394):
481-489.
Wolff, G.T., P.J. Lioy, G.D. Wight, R.E. Meyers and R.T. Cederwall,
1976: An investigation of long-range transport of ozone across the
midwestern and eastern United States. Presented at International
Conference on Photochemical Oxidant Pollution and Its Control;
Raleigh, N.C., 13-17 September.
Zeller, K., R.B. Evans, C.K. Fitzsimmons and G.W. Siple, 1976: Mesoscale
analysis of ozone measurements in the Boston environs. Presented
at Symposium of Nonurban Tropospheric Components, Hollywood, Fla.,
10-12 November.
152
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-450/3-77-022a
3. RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE
The Relation of Oxidant Levels to Precursor Emissions
and Meteorological Features. Volume I: Analysis and
Findings
S. REPORT DATE
September 1977
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
F.L. Ludwig, E. Reiter, E. Shelar and W.B. Johnson
8. PERFORMING ORGANIZATION REPORT NO
9. PERFORMING ORGANIZATION NAME AND ADDRESS
SRI International
Menlo Park, ,CA 94025
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
68-02-2084
12. SPONSORING AGENCY NAME AND ADDRESS
Environmental Protection Agency
Office of Air Quality Planning and Standards
Research Triangle Park, N.C. 27711
13. TYPE OF REPORT AND PERIOD COVERED
Final
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
16. ABSTRACT
Published ozonesonde data, radioactive fallout measurements and alpine ozone observations have been
used to estimate the stratospheric contribution to observed ozone concentrations at ground level. Long term
average effects from the stratosphere over the U.S. are on the order of 10 ppb, with a springtime maximum
around 20 to 25 ppb. Short term stratospheric intrusion events resulting in one-hour-average concentrations
of stratospheric ozone in excess of 80 ppb in the lower troposphere have a frequency of only about 0.2 per-
cent. Still fewer (but some) of these events lead to ground-level impacts of such a magnitude.
Tropospheric causes of high ozone concentrations away from cities have been investigated by statistical
analysis of meteorological conditions and the precursor emissions occurring along air trajectories and by
comparisons of weather maps and large-scale 03 distributions. Meteorological factors are statistically more
strongly correlated with ozone concentration than are emissions, with air temperature being the most highly
correlated. At sites well removed from cities, the upwind emissions of oxides of nitrogen are more strongly
related to ozone concentrations than are the emissions of hydrocarbons. Widespread violations of the federal
oxidant standard are most likely to be found in association with a stagnant high-pressure system or in the
warm southwesterly flow in the western portion of a high pressure area, often ahead of an approaching cold
front.
The results of this and other studies suggest that not all violations of the federal oxidant standard
are controllable and this fact must be considered in the design of control strategies. Also, for areas
within about 125 ton of large cities, control might be achieved through the reduction of HC emissions. In more
remote areas, control strategies involving NOX control throughout large regions must be considered.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS C. COSATI Field/Group
Tropospheric Ozone
Stratospheric Intrusion
Oxidant Control Strategies
Meteorological Factors Affecting Ozone
18. DISTRIBUTION STATEMENT
Unlimited
19. SECURITY CLASS (ThisReport)
Unclassified
21. NO. OF PAGES
170
20. SECURITY CLASS (Thispage)
Unclassified
22. PRICE
EPA Form 2220-1 (Rev. 4-77) PREVIOUS EDITION is OBSOLETE
153
-------
3D
o>
m
D
S
3?
II
O
-c
3J
tif
o 3 :± H
2s;' o m
H
ff!
n
03
r
HH
n
>
H
NH
O
^
^!
O
•
w
NO
S
2
tD ">
a I
cu o>
Q. ~
Q.
o
_. c
fD ^
S-o
Q- H
o
\
CC
6
CD cn
5 3
.
O QJ
ID
O
-<
c/>
TJ
m
O
-n
O
°°£
O^
0=<
rn
2
O
25
7; •
C
co H -n
in m m
-I
m
O
m
z
n
------- |