vvEPA
United States
Environmental Protection
Agency
Environmental Sciences Research
Laboratory
Research Triangle Park NC 27711
EPA-600 4-80-OOT~
January 1980
Research and Development
Synoptic
Meteorology and
Air Quality
Patterns in the
St. Louis RAPS
Program
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U S Environmental
Protection Agency, have been grouped into nine series These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are
1 Environmental Health Effects Research
2 Environmental Protection Technology
3 Ecological Research
4. Environmental Monitoring
5 Socioeconomic Environmental Studies
6 Scientific and Technical Assessment Reports (STAR)
7 Interagency Energy-Environment Research and Development
8. "Special" Reports
9 Miscellaneous Reports
This report has been assigned to the ENVIRONMENTAL MONITORING series
This series describes research conducted to develop new or improved methods
and instrumentation for the identification and quantification of environmental
pollutants at the lowest conceivably significant concentrations. It also includes
studies to determine the ambient concentrations of pollutants in the environment
and/or the variance of pollutants as a function of time or meteorological factors.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161
-------
EPA-600/4-80-001
January 1980
SYNOPTIC METEOROLOGY AND AIR QUALITY
PATTERNS IN THE ST. LOUIS RAPS PROGRAM
by
Elmer Robinson
Richard J. Boyle
Air Pollution Research Section
Chemical Engineering Department
Washington State University
Pullman, Washington 99164
EPA Grant No. 805142010
Project Officer
Thomas R. Karl
and
Gerard A. DeMarrais
Meteorology and Assessment Division
Environmental Sciences Research Laboratory
Research Triangle Park, North Carolina 27711
ENVIRONMENTAL SCIENCES RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NORTH CAROLINA 27711
-------
DISCLAIMER
This report has been reviewed by the Environmental Sciences Research
Laboratory, U.S. Environmental Protection Agency, and approved for publica-
tion. Approval does not signify that the contents necessarily reflect the
views and policies of the U.S. Environmental Protection Agency, nor does men-
tion of trade names or commercial products constitute endorsement or recommen-
dation for use.
-------
ABSTRACT
An objective, statistical synoptic weather map classification scheme de-
veloped by Lund to stratify map patterns for further study was used to type
regional weather patterns. The investigation extended over a 500-mile radius
of the greater St. Louis area and was intended for subsequent application to
air pollution studies. This analysis correlated sea level pressure data at
21 selected National Weather Service stations at a specified time on a given
day, with sea level pressures for these same stations on each of the other
days in the four-year period 1973 through 1976.
A total of 52 separate weather map types were identified with 12 map
types in the winter season, 13 map types in the spring season, 13 map types in
the summer season, and 14 map types in the autumn season.
To illustrate the potential usefulness of this method to air quality and
synoptic weather relationships, this paper also contains a preliminary compar-
ison of the weather typing system to air quality data. The data include total
suspended particles and carbon monoxide concentrations sampled in the Regional
Air Monitoring System of the Environmental Protection Agency, for a period of
approximately a year and a half during 1975-1976.
Results indicate that the typing program is able to show that synoptic-
scale weather does affect relative pollution levels in the St. Louis region,
if definitive synoptic model types are carefully chosen prior to application
to pollutant concentration levels.
Continued research is expected to indicate clearly not only how air pol-
lution levels in the St. Louis area depend on prevailing synoptic weather pat-
terns in general, but also how mesometeorological variables act as diffusion
parameters.
-------
CONTENTS
ABSTRACT iii
FIGURES iv
TABLES v
ACKNOWLEDGEMENTS vi
1. Introduction 1
Historical Background 1
Application of Ueather Types to Air Quality 2
Study Objectives 3
2. Research Techniques and Results 5
Synoptic Weather Map Typing Technique 5
Synoptic Map Typing Results 10
Air Quality Application 15
Air Quality Data Analysis 17
Total Suspended Particle Comparisons 17
Carbon Monoxide 22
3. Summary and Conclusions 26
References 28
Appendix - Synoptic Map Types Inventory 30
-------
FIGURES
Number Page
1 National Weather Service stations used for weather map typing
program 6
2 Correlation between network sea level pressures on June 1, 1973,
summer Type A base map and June 7, 1973, another summer Type A
map 9
3 St. Louis Regional Air Pollution Study air monitoring network,
1975-1976 16
VI
-------
TABLES
Number Page
1 Stations Used in this Study 7
2 Summary of Seasonal Synoptic Map Types St. Louis Regional Study,
1973-1976 12
3 Daily Map Type Assignments 13,14
4 Seasonal Map Types Ranked by Total Suspended Particle
Concentrations Averaged for Stations 103, 105, 106 and 108. . . 18
5 Results of Statistical Test for TSP Concentrations by Map
Type Pairs 21
6 Seasonal Map Types Ranked by Carbon Monoxide Concentrations
Averaged for Stations 103, 105, 106 and 108 23
7 Results of Statistical Test for CO Concentrations by Map Type
Pairs 25
-------
ACKNOWLEDGEMENTS
This report represents the efforts of a number of the Air Pollution
Research Section staff in addition to the named authors. Special recognition
should be given to Mr. F. Lee Bamesberger, Associate Chemist, who provided
programming and computer services as well as data handling advice. The assis-
tance of Mr. R. J. Valente and Mr. Mark Vuono in various data processing
chores is also appreciated.
In the WSU Statistical Services group the frequent and informative con-
sultations of Dr. R. B. Bendel were important aids in accomplishing the pro- .
gram.
The assistance of our EPA Project Officer, Thomas Karl and his replace-
ment Gerard A. DeMarrais were vital assistants to our program. They were
timely in response to our requests for assistance and patient with us as we
carried through with the program.
Finally, the senior author wishes to point out that this report is to a
significant degree the thesis research program carried out by Mr. Boyle during
his program of study toward the Master of Science degree in Engineering, and
a large part of Mr. Boyle's thesis is used directly in this final report.
vm
-------
SECTION 1
INTRODUCTION
Air quality, as characterized by air pollutant concentrations, is con-
trolled to a significant degree by the prevailing weather conditions from the
time of the initial pollutant emission to the time of sampling and measurement
of the atmospheric concentrations. Although the meteorological factors af-
fecting such pollutant dispersal exhibit a wide range of size and time scales,
the dispersion field is strongly linked to the synoptic scale of atmospheric
motion as a prime feature of the dispersion. The impact of synoptic weather
patterns, in this regard, is not new and has been treated widely in numerous
books and articles in both the fields of air pollution and meteorology.
Recently, synoptic influences have figured prominently in studies of urban
plumes, ozone transport via an anticyclonic model, and general "air mass
transport".
HISTORICAL BACKGROUND
Accounts dating back to about 300 B.C., such as De Ventis by Theophrastus
of ancient Greece, (Coutant and Eichenlaub, 1975), clearly demonstrate man's
early attempts to characterize the weather in two easily recognizable patterns.
Through the following centuries observant scientists were able to describe
various storm models and typical types of weather systems. Modern synoptic
meteorology moved ahead rapidly after the development of the polar front storm
models. As extended range forecasting techniques were developed in the 1930's
and 1940's there was considerable interest in methods to organize weather pat-
terns into sets of synoptic types. The results of these procedures could
serve as valuable forecast aids and as a surrogate for the memory of the fore-
caster.
Rigorous attempts to classify synoptic weather by types, especially in
the three decades prior to 1950, have been discussed at some length by Elliott
(1951). A six-day North'American weather type scheme was developed by Elliott
and Krick during the 1930's wherein each identifiable type covered a sequence
of days characterized by easily identifiable quasi permanent meteorological
factors. Further refinement of this technique resulted in the California
Institute of Technology catalogue of synoptic weather types of North America
which contained some 20 to 30 distinct types (Krick and Elliott, 1940).
As machine methods moved onto the scientific scene during the 1950's, the
arts and skills of forecasting were integrated with the computerized meteor-
ological statistical method of forecasting and analysis. Thus, it was through
-------
a natural progression of events that investigators tried computerized statis-
tical methods in developing weather types of generalized synoptic situations.
Typical pressure patterns lend themselves to numberical analysis which then
can be readily programmed into a computer system.
Application of simple linear correlation statistical procedures, augment-
ed by computer programming, resulted in a workable scheme by Lund (1963) of
synotic weather classification which is the major basis for the experimental
procedures described in this report. Basically, Lund's technique involved
correlating sea level pressure values at 22 northwestern United States weather
stations by statistical means to describe recurrent synoptic map patterns. He
developed a catalogue of map types which he believed would be most useful in
identifying reappearing map patterns and stratifying situations for further
study. Since publication of his technique in 1963, several additional invest-
igators have used modifications of his method. For example, Singh, et. al.,
(1978) have employed a qualification of Lund's style in analyzing synoptic
cl imatology.
Although Lund considers this map typing scheme to be objective, in a 1971
study relating 850 mb data correlations with areal precipitation patterns for
California and the eastern United States, he admits that it is difficult to
arrange types in a meaningful order with regard to a particular weather event.
For example, a meteorological model to estimate the probability of precipita-
tion has not been developed as yet. On a smaller or mesometeorological scale,
Christensen and Bryson (1966) employed a variation of a weather typing mathe-
matical modeling procedure, which they called component analysis (factor ana-
lysis), that reduced a sample of data for Madison, Wisconsin, to the number
of variables needed to describe each day (type) meteorologically. They then
compared these variables with the synoptic situation for that day to determine
whether the models with certain meteorological factors formed a reasonable
synoptic pattern. The results were then grouped, after some manipulations,
into synoptic types by the Lund technique. Several papers written by foreign
investigators such as Zdravko (1968), Fliri and Schuepp (1968) and Bardin
(1971) also refer to Lund's method in their particular development of weather
types using additional meteorological parameters of temperature, precipita-
tion, duration of sunshine and cloudiness.
APPLICATION OF WEATHER TYPES TO AIR QUALITY
Although air pollution patterns are generally considered to be influenced
to a significant degree by local meteorological conditions, they are also
greatly affected by regional and large-scale synoptic weather systems that
move through the area. For example, a quasi-stationary antic/Ionic circula-
tion and in the summer season usually correlates positively with high or even
episodic air pollutant concentrations. Conversely, cyclonic circulation
is conducive typically to good dispersion and low concentrations of air pollu-
tants. There are many possible variations in the area of air pollution and
synoptic weather as shown by the fact that spring time trough situations,
quasi-stationary low pressure systems, and weak frontal systems were found to
correlate with significantly elevated 2,4-D concentrations by Peckenpaugh,
-------
et. al. (1974) in their investigation of long distance atmospheric transport
of the organic herbicide in central Washington State.
During the summer of 1974, Westberg, et. al. (1978) conducted a 24-day
experiment in Canton, Ohio, involving a vertical and horizontal profile study
of ozone concentrations. As a result of their observations of ozone behavior,
an ozone transport model was developed which logically explains long-range
ozone transport dispositions occurring in a quasi-stationary high pressure
situation through the eastern region of the United States. Subsequently, us-
ing a mesoscale approach, relationships between sea breeze effects and ambient
ozone concentrations were examined by Westberg, et. al. (1975) along the
New England coastline. It was found that this phenomenon could either moder-
ate or enhance pollution levels in the Boston area. However, in the Groton,
Connecticut region, high concentrations of ozone usually coincided with the
sea breeze effects, and aircraft measurements over the ocean indicated that
these high concentrations were part of a large urban plume from the New York
City - New Jersey area. Since synoptic weather patterns greatly influence
this type of atmospheric effect, a map typing classification method should
also have practical applications in these cases.
In trying to establish causal relationships between precipitation pat-
terns and the industrial effects of producing steel near Dunaujvaros in cen-
tral Hungary, Lowry and Probald (1978) concluded that there is a need for
both statistical testing of temporal differences, i.e., the pre-industrial
period (1934-43) versus the post-industrial period (1961-70), and a synoptic
classification method, in exploring definitive anthropogenic effects of pre-
cipitation using standard cl imatological data.
Traditionally, meteorological forecasters have incorporated, consciously
or subconsciously, synoptic weather types of significant atmospheric trends
into their analysis and prognosis procedures. It is necessary for meteorolo-
gists to scrutinize more closely the effects on air quality of definitive
weather patterns and for weather forecasters to appreciate the implications
involved in this process. As mentioned earlier, we believe that it is ex-
tremely important in conducting efficient air pollution control strategy pro-
grams that synoptic weather patterns be given consideration as a major influ-
ence in the dispersion region.
STUDY OBJECTIVES
One of the primary goals of this research was to identify synoptic weath-
er types by using Lund's objective technique of correlating sea level pres-
sures. Thus a set of seasonal weather map types was developed for application
to an experimental area around St. Louis for the time period covered by the
Regional Air Pollution Study (RAPS) program.
An additional aim of the study was to begin to understand the relation-
ships of synoptic weather patterns to ambient air pollutants in the St. Louis
area. For this purpose, total suspended particles (TSP) and carbon monoxide
(CO) data from the EPA St. Louis RAPS program were compiled and analyzed.
These particular pollutants were selected mainly because they are relatively
-------
inert and easily analyzed. The actual air monitoring data were collected by
the Regional Air Monitoring Study (RAMS) which was one part of the overall
RAPS program.
In our weather typing research, a tentative analysis of the Greater
St. Louis area air quality, as related to synoptic weather, was undertaken as
a first step in stratifying air pollution concentration levels into more homo-
geneous meteorological classifications. It is expected that this library of
weather types can be generally useful in assessing air pollution, mesoscale
precipitation, and other weather related factors in the St. Louis area.
The results of this research on regional weather systems should provide
other investigators working with RAPS data with an objective classification
of the prevailing weather systems. Emphasis of synoptic influences on RAPS
air pollution situations should increase the realization of the importance
of this phenomenon so that more reliable evaluations can be made of air pollu-
tion in the St. Louis area. The research program at WSU is continuing with
additional and more detailed analyses.
-------
SECTION 2
RESEARCH TECHNIQUES AND RESULTS
SYNOPTIC WEATHER MAP TYPING TECHNIQUE
Weather typing processes in the recent past have been limited mostly to
subjective applications and techniques by individual meteorologists.
The advent of computers has not only made possible much more objective
synoptic typing systems, but has enhanced the development of large scale dy-
namic modeling procedures that are more highly sophisticated and useful than
the older typing techniques. In this regard, statistical computerized map
typing is particularly useful for some specialized situations. Lund (1963)
has clearly demonstrated how a simple linear correlation technique can be ap-
plied on a mesoscale to develop such a system of synoptic map types. In his
study identifying maximum linear correlations between precipitation and the
850 mb surface level, Lund (1971) illustrated the usefulness of map typing in
understanding the occurrences of precipitation and cloudiness in the region
that was being analyzed. Additionally, regional and hemispheric circulation
patterns have been examined in a study by Arai (1964) in which he classified
ten years of daily winter 500 mb patterns into fundamental and subsidiary
weather types. Methods of weather typing were also used by Halter (1976) in
an assessment of weather conditions affecting the Pacific Northwest coast.
As stated previously, the synoptic map typing program described in this
study used the statistical or pressure correlation method developed by Lund
(1963). Specifically, the area included in the typing program was the region
of approximately 500 miles (800 km) radius centered on St. Louis, Missouri.
The weather data for the typing process were obtained from 21 first-order
National Weather Service stations distributed more or less uniformly over the
region. Figure 1 shows the network of stations and the region covered by the
sea level pressure analysis used in the program. Table 1 identifies the sta-
tions according to their National Weather Service call letter identifiers.
Data for the map classification process covered the four-year period, 1973-
1976. For this analysis, the sea level pressures at 0000 GMT (6 PM CST) were
used in the map typing program. This specific time was chosen to avoid early
morning radiation stability influences and to include to the maximum extent
possible the full impact of the daily synoptic weather situation on the pollu-
tants under investigation.
The total number of synoptic weather stations employed in this type
of appraisal is a compromise between the technical value of a very dense net-
work and the costs of accumulating, handling, and processing the data file.
Lund's (1963) initial study used a set of 22 stations over the area along the
-------
Figure 1. National Weather Service stations used for weather map
typing program.
-------
TABLE 1. STATIONS USED IN THIS STUDY^
Call Letters Name of Station
ALO Waterloo, Iowa
ATL Atlanta, Georgia
BNA Nashville, Tnnessee
DDC Dodge City, Kansas
DET Detroit, Michigan
(Wayne County)
DSM Des Moines, Iowa
FSD Sioux Falls, South Dakota
GRI Grand Isle, Nebraska
HTS Huntington, West Virginia
IND Indianapolis, Indiana
JAN Jackson, Mississippi
MEI Meridian, Mississippi
MEM Memphis, Tennessee
MKC Kansas City, Missouri
MSP Minneapolis, Minnesota
OKC Oklahoma City, Oklahoma
ORD Chicago, Illinois
(O'Hare Int'l.)
SGF Springfield, Missouri
SHV Shreveport, Louisanna
STL St. Louis, Missouri
TVC Traverse City, Michigan
*A11 stations are hourly-reporting National Weather Service Stations.
-------
northeast coast of the United States from Virginia to Maine and west to
Detroit and Sault Ste. Marie. Halter's (1976) typing program of the Pacific
Northwest also used a set of 22 stations covering the area bordered roughly
by Quillayute, Washington; Great Falls, Montana; Bryce Canyon, Utah; and
Oakland, California. Because about 20 stations have provided a useful typing
program, the present study around St. Louis used a total of 21 first-order
National Weather Service stations over an area borderd by the neighborhoods
of Minneapolis, Minnesota; Huntington, West Virginia; Atlanta, Georgia;
Shreveport, Louisiana; and Dodge City, Kansas.
The prior analysis by Lund (1963) was based on five years of winter sea-
son data incorporating the months of December through March. Halter (1976)
used four full years of data, 1970 through 1973, divided into a winter half,
November through April, and a summer half, May through October. In order to
insure that the selection of months for each season coincided with the natural
mean seasonal temperature pattern for the St. Louis area, an actual tempera-
ture plot covering the four-year time period of this study was made to deter-
mine if the lowest temperature actually occurred in December, January and
February and if the highest temperatures occurred in June, July and August.
This proved to be the case, and it was concluded that the season groups sel-
lected for this investigation were compatible with expected seasonal tempera-
ture cycles.
A map type is a group of weather maps with a sea level pressure pattern
distinctive enough to be classified separately from other weather patterns.
In terms of identified map types, Lund (1963) had ten types with 398 out of a
total of 445, or 89% being successfully typed. Halter (1976) was able to type
92% of his winter seasons into ten types and 94% of his summer season maps in-
to eight types. As will be discussed, the St. Louis typing study had compar-
able results and typed between 89% and 91% of the seasonal maps into 12 to 14
different types. Thus, the St. Louis program produced results that are gener-
ally equivalent to those of similar studies.
The simple linear correlation procedure of Lund (1963) is based on the
statistical interdependence between the sea level pressures at a given set of
stations on one day and the pressures at the same stations on a second day.
For example, in the St. Louis program, if the X-axis (abscissa) is used to
plot station pressures for day 1 and the Y-axis (ordinate) the pressures for
day 2, the pressure data from 21 stations will define a corresponding set of
21 points in the X-Y field. Standard statistical techniques can then be used
to determine the correlation between station pressures on day 1 versus those
on day 2. It is assumed, then, that a "good" correlation between station
pressures on the two days is indicative of similar synoptic map patterns on
the two days. In Figure 2, the correlation coefficient is 0.81 between June
1, 1973 and June 7, 1973 data.
Previous statistical typing by Lund (1963) and by Halter (1976) were
based on weather types being identified by correlations of 0.70 or greater be-
tween each map date within the type and the identified base map. The choice
of magnitude of the lowest acceptable correlation for determining a type
classification was examined in detail by Lund (1963). He determined that the
same key or base maps would be identified with the use of either 0.90 or 0.80
8
-------
SEA LEVEL PRESSURE(mb) JUNE 7,1973
o
o
o
Ol
o
INi
o
8-
Ol
CO
m
m
<
m
m
CO
co •
c
30
m _
co o.
C_
C
m .
-------
instead of 0.70. The major impact of the choice of a value for the lowest
acceptable correlation was in number of cases falling into each type and the
number of untyped maps. The higher the limiting correlation the fewer the
number of maps in each case, but not much difference in the identified types,
and the larger the number of untyped maps. After a careful consideration of
the limiting correlation value Lund (1963) chose 0.70 as his lowest acceptable
correlation. With this correlation limit he was able to type 89% of his map
file. Halter (1976) also used 0.70 as a lowest acceptable correlation and
typed more than 92% of his map file. In the present study a value of 0.70 was
also used for the lowest acceptable correlation coefficient for type identifi-
cation.
Following Lund's guidelines, the St. Louis regional synoptic map types
were determined by the following steps:
1. Correlate the 0000 GMT sea level pressures for each day of the sel-
ected season with the corresponding pressure on every other day of
the season over the four-year test period.
2. Identify the day which has the most other days related to it by a
correlation coefficient of 0.70 or greater. Designate this day as
the base map for Type A and the maps correlated with it (r _> 0.70)
as belonging to the Type A classification.
3. Remove all Type A maps from the map file and then identify the date
in the remaining maps which now has the most maps suitably correlated
(r _> 0.70) with it. This is the base map for Type B and the maps
correlated with it make up the Type B classification.
4. Remove all Type A and Type B maps from the file and proceed to ident-
ify the Type C base map and the maps in the Type C classification, as
was done for B.
5. Repeat the typing process until most (about 90% or more) of the maps
have been typed.
6. Since a given map may be well correlated with more than one type,
review all calculated correlations relating each map and the several
base maps. Determine that the map type assigned by the first cor-
relation check was the type that had the highest correlation. If
the given map is not assigned to the base map type having the high-
est correlation, move the test map to the map type which is most
highly correlated.
7. Continue the map typing process until each of the four seasons has
been completed.
SYNOPTIC MAP TYPING RESULTS
In this study, the WSU AMDAHL 470V/6-II computer system was used for the
correlation analysis of daily station pressures. The four years of data,
10
-------
1973-1976, were separated into the four conventional seasons: Winter:
December, January, February; Spring: March, April, May; Summer: June, July,
August; Autumn: September, October, November.
When this correlation-typing procedure was carried out on a seasonal
basis for the region around St. Louis, the result was the identification of
12 winter map types, A(W) through L(W); 13 spring map types, A(SP) through
M(SP); 13 summer map types, A(S) through M(S); and 14 autumn map types, A(A)
through N(A). Thus, there were a total of 52 map types, although similar map
types may occur in more than one season. Those maps in each season not sign-
ificantly correlated with any identified type were classified in a miscellan-
eous group designated 0.
Some of the specific results of this statistical correlation typing pro-
cess are shown in Table 2 which lists the assigned map type, the date of the
major synoptic pressure system which was the base map of highest correlation
factor for the type, and the number of other maps that were most highly corre-
lated with the base map. Typically, the number of maps in a given class de-
creases from a maximum for Type A, but there are some cases such as the spring
(SP) types after F where this does not occur. This is caused by the shifting
of maps to the type class with which they show the maximum correlation (see
Step 6 in the above typing process and Lund (1963), (Table 3).
The synoptic map types for each season on a daily basis are shown in
Table 3, Daily Map Type Assignments, for the four-year test period, 1973-1976.
The map types marked with an asterisk are the base or primary map types with
the highest correlation factor of a particular season. There is only one
date, July 31, 1975, for which there was no map type. This was due to missing
sea level pressure data. The file of 52 seasonal base or key map types re-
sulting from this study are reproduced in Appendix A.
The maps presented in Appendix A are the NOAA/NWS United States surface
analysis for 0000 Z. As such they cover a larger area and present more data
items than was used in the present computer typing system. This was done for
the convenience of the meteorologist using this map type file and who would
probably be interested in the relationship of the St. Louis regional weather
type to the synoptic situation over a larger area. This area around St. Louis
that was used in this typing study is indicated by the large circle centered
on St. Louis. The individual stations used in the typing process can be
identified by referring to Figure 1.
Persistence of flow patterns is another useful factor identified by this
map typing scheme. For example, in the autumn season, the most frequently
persistent weather situation over the four-year period involved map Type A,
which has 14 cases of a two-day persistence, four cases of a threeday persis-
tence, two cases of a four-day persistence, and one case of a seven-day per-
sistence. Two other map types, B and C, also have many combinations of per-
sistence in this same autumn season grouping. It is easy to use the synoptic
map typing classification to determine persistence and to predict potential
stagnation periods by the judicious application of Lund's synoptic weather
classification technique.
11
-------
1 —
CTi
r_
1
CO
^^
O1
^_
•\
>-
Q
1—
oo
z
o
t — 1
t n
UJ
CXL
OO
— )
o
t—
00
oo
1 1 1
LLJ
Q_
1—
a.
^f
^»
o
•— i
i—
Q-
0
>-
oo
<
o
oo
UJ
oo
U.
O
Qi
^r
2£
— s
oo
,
CM
UJ
i
ca
i—
c
E
3
•M
3
^
L-
CU
£
F
3
oo
C71
C
•r-
S-
a.
00
t-
cu
+J
c
•1—
3
O CO
CU
• CO
O fO
z o
Q.
4- (0
o 2:
CU CU
-I-J CO
13 U
Q CO
4-
O CO
CU
O 13
z o
Q.
4- 13
o 2:
CU CU
4-> CO
t3 13
Q CO
4-
O CO
CU
• CO
O fQ
z: o
o.
4- (O
o s:
CU CU
^ CO
a aa
O CO
O)
• CO
O fO
z o
D-
4- 13
o 2:
CU CU
4-> CO
fO 13
Q CQ
O)
a.
>^
h-
0*1 r*^*. ^f~ LO ^^ to *^ LO o^ i i r*^. vo *^ ^o
00 «?1- CVl CM OJ i— OJCMi— •—
LO i^O VO *O **O OO "^ ^t" LO
P""- r^s» p*^ f**«, {Q f«x» f*x. ^- (x^, j_^j fs^, ^^ |_^ i^.,,
r*-1*' P1*- t*»* P^». r^x
AAAA AA A A A
00 I-O CO ^«O A ^O OO •» p— A O^ M A ^3
OJ r— CMCMLT),— CMCO,— CMr— ^OOr—
> 4-* ^> -4-J Q^ > > f~i f\ j_% > c^ > >
oooocuooa>cucjocijoo
^o oo ^J* oo ^^ LO ^— co LO r*^> ^D oo o^
00^^"i— OJi— C\Jf— r— OJ
LOLO LOCOCO COOOLOLO
oo r**» P"*" oo P*"*- P**« P**« vo oo P^* f1^* P*** P"""*
P-*. P^* p**. p-^
AA AAA AAAA
,— 00,— i— CvJr-i— OOCTiCMCMOJOsJ
c c,— cn ,— ,— ^- c^-i— P— i— en
333^3133133313313
LO P*->. CO LO OO t*Q O*i O^J O*» CSJ LO O **O
OO LO OO OO C\J i— H t-H C\J t— 1 r—\
*& LO LO LO OO ^" LO
P*"* P*^. OO P^- OO P"*- LO P*^- LO OO P1** LO P*1^
r*1* P""* P"^ p*^ p*"* P^*.
A A A A A A A
LO OO ** r— ** ^— • ** ^— ** " CNJ ** CD
r— * r— ^f i~~ 00 CVJ LO CVJ CO CSJ C\J ^*11*' OO
^fekcx^ra^nskra^ask
O^ CO ^j1 p». C^ O"^ p"~ LO OO CD 00 P*1*
LO LO ^if ' CO CO CsJ C\J ^'~ ^~~ ^•—
^- vo r*^ r*^* LO vo P^» co ^~ ^^ uo f****
^^ r*>. ^^. ^^» r*** r^«
»\ «\ M A A *
(^ »\ CM CO •* •» ^O •* •* CM •» CT^
CMCOCMCMCOOOi — f-OOi — Lf>i —
f) C O ^ ^ *** (^ O C CJ O CI
curocurocucucucutacucuro
1 1 "^^ f^^> (~^ t 1 1 1 1 1 <*"^ >~5 l^^l f~} "~^
^C CO C_ 3 r~^ i i ) 1 1 CJ "r* »—< "~3 ^*S i 5^" ^i
^" o~> co *s
CO CM ^O i—
CO CO CT)
CT* CT* r^ ^^
CO CM IO CT>
co co oo
«3 CM CO fcft
co co to o
CO CO CTi
ir> vo i — »*
CO CM VD O
CO CO CT)
•o
CU
-o co a.
,
Q. CO I—
>> to
1— 0 +J
c:
• r— i— CU
O 13 13 O
CO +J ^J S-
•r- O O CU
2: 1— H- Q-
12
-------
TABLE 3. DAILY MAP TYPE ASSIGNMENTS
Year
Month
Day 1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Winter
D J F
H B D
A H B
0 C F
OBJ
C E I
B G G
H*G E
H E G
B B 0
F C H
A B H
J*A D
0 C I
D C B
E A B
GAG*
D A H
D C A
E L*C
B D*F
A J B
A C C
D C F
D F D
A 0 I
0 0 G
F E H
K B A
B F
E A
E D
1973
Spring
MAM
A E B
J*C C
L G C
D C*A
ABA
J B A
0 G E
D G E*
D C E
A E E
E B M
ADC
A A M
BAG
G B E
C E C
I A 0
A A B
DAG
D A A
H*B A
ABO
A 0 0
J D 0
G D A
0 G A
A C F
0 A C
G A C
DAE
F B
Summer
J J A
A*A*D*
A C D
A A H
0 D M
K A A
C A A
A A A
C A C
H I*C
ADC
C B F
J C 0
B E 0
ADO
A G*L
C B B
A M A
L A A
K F*0
K J*D
E J B
H B M
C M J
M A A
M K*A
C K A
E E A
E 0 A
F M A
A 0 M
D A
Fall
SON
A A I
A A B
ADD
0 B D
B J H
DAE
E A M
E A B
D A H
0 A A
BAA
D A C
D G C
H C I
D B B
DBA
B F A
J A 0
I 0 A
E E A
A A C
I A M
A A A
A F 0
A C A
M D A
D B B
0 B G*
0 0 C
0 C L
G
Winter
D J F
0 G I
BEE
GOG
H 0 A
A L D
D B E
B D G
B I*F
F G F
C E B
0 B A
A H C
0 A 0
J C D
C F 0
B C B
L D A
J E K
0 D F
0 C D
F L I
ABB
I 0 I
E 0 B
F A F
A C A*
F 0 A
AFC
B A
D A
B G
1974
Spring Summer
MAM J J A
B E J
BAB
B F D
E C D
B I C
B B H
A 0 A
A 0 0
G A H
D A A
0AM
D B I
H F A
A C G
MCA*
COB
ABA
K A A
D A A
G A A
E B 0
K*C 0
G L K
H H L
BAD
A A D
ABA
ABB
0 K F
I C B
A G
G A 0
M A K
M C E
A E 0
ABO
A A B
A A L
F A A
CCA
E A M
E B C
H M C
A H B
C E B
EGA
D B 0
C A 0
A C 0
C I A
C B M
I 0 L
GOO
G B G
G L B
G L*M
B 0 C
B 0 J
A E 0
C D E
C 0 C
F F
Fall
SON
D D A
H E 0
H*A B
E A 0
E I B
E B J
E J N
0 0 A
A J A
A A 0
1*1 G
B D G
H A B
F B C
0 J C
F F A
G H N
FDA
L D K*
BEG
H A 0
E I A
A A K
F C B
F B E
F N 0
I A C
K A B
GAD
B A 0
X
Type "0" is miscellaneous map type
*Primary map type
X no map type (data missing)
(continued)
13
-------
TABLE 3. (continued)
Year
Month
Day 1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Winter
D J F
A B G
LAG
A C D
A F I
K*J B
GAB
0 J C
E 0 E*
F D H
A C A
I B I
D B B
A B D
K A D
0 K E
E 0 I
BAB
F C B
C E F
E A A
H E 0
ONE
G A 0
D A C
0 C C
B L B
D I A
DAB
0 B
H D
0 E
1975
Spring
MAM
G D H
L C A
I I*C
B H A
B H A
AHA
C A J
L J D
ADD
D G D
J G 0
G 0 L
D A 0
I H M
A A G
HAH
ABO
0 E B
E E B
A H B
F*A B
A A A
0 B B
E A A
L H A
A A C
J A H
OBJ
C B F
B 0 C
F C
Summer
J J A
0 B K
A 0 0
A E 0
COO
K C G
E E B
GOB
B D M
BOA
0 D C
K D A
C D A
C 0 F
0 A J
C M 0
A A C
C A F
A C F
A C*B
A E*M*
M 0 A
A A M
A C A
B D A
B G 0
L 0 F
B E B
B*B A
B B F
B B 0
0 H
Fall
SON
0 H I
E J*I
L F M*
DAM
B C A
J E A
0 A C
E A I
E 0 0
AON*
B A F
HAG
JAG
A I C
ABC
EDA
AHA
0 H A
K C A
G C K
0 I G
HAG
H A 0
H K B
H J D
A A 0
A I I
ABA*
A H I
BAG
A
Winter
D J F
CAB
D B 0
0 B*K
OLE
DAE
E K B
G E C
A B F*
A A A
E J C
OFF
B A C
H B G
CCA
B C C
C B C
F G J
A A C
BCD
BCD
ABB
C*A B
F 0 A
ADA
B E C
F B F
C H K
B C A
A F 0
E G
B C
1976
Spring
MAM
A E E
F B C
L C I
FOB
COB
E H G*
L*D D
D D 0
0 A B
0 B E
A D*H
C A J
B A F
K B D
D B 0
I B C
BBC
B*B 0
BAB
E F B
L E H
0 0 D
BAD
BOD
A G H
OHO
A H J
A D J
F 0 A
C M*F
E F
Summer
J J A
G F G
G F G
B G B
B G A
B G J
0 H G
H C B
H*A B
H M M
ACM
A E A
M 0 C
A H F
A C F
K E B
A E B
M H B
E A B
DAB
H C M
A I M
A A B
A E B
KGB
A M M
A C A
A A 0
C C D
E A B
D I M
G 0
Fall
SON
D 0 A
AEG
I A G
L*M B
E*B C
A D B
A H B
B 0 F
B G G
FAB
A A H
A K B
A G N
L C J
D B*N
D H F*
L E C
A E C
B B B
B G B
B G G
A A G
D A C*
E B C
E H I
B D*0
H H B
0 F H
0 0 C
F G A
H
indicates the base map for this season and type
"0" indicates untyped maps
14
-------
In addition to determining persistence, an examination was made of the
transiting patterns of charts in a season into other types to determine if
there was any regular evolution from one type to another. To make this deter-
mination, a method used by Singh, et. al. (1978), in their investigation of
synoptic climatology during the summer monsoon season in India, was employed.
Within each season, a simple Chi Square statistic was computed to determine
if observed frequencies of occurrence of one map type were greater than the
frequency of occurrence expected by chance for that type. Results of the
test in this study indicated that there were significant differences between
observed frequencies and expected frequencies for some pairs of map types.
Accordingly, the map type file could be used to determine the more probable
transitions by map types based, and the result might have obvious synoptic
forecasting applications.
One would normally expect more intra-seasonal transitions between types
in spring and autumn. Although the Chi Square test for all seasons is not
included in this study, all seasons were tested and more predictable or non-
random transitions between map types occurred in autumn and spring than in the
summer and winter seasons. This might be expected on the basis of the general
persistence of major weather systems in summer and winter and the more rapid
system developments in spring and autumn.
AIR QUALITY APPLICATION
Direct and indirect effects of synoptic weather systems on air pollution
intensity have been reported for many years by air quality meteorologists,
scientists and engineers. Continuing efforts are being directed to improve
both simple and sophisticated mathematical air quality models for use in
nationwide pollution control strategy programs. As part of this general
attempt to better understand local and regional air pollution conditions, an
additional objective in this study was to find relationships between synoptic
map types and specific air pollution concentrations over a given time inter-
val. In order to test the validity of this objective in the St. Louis area,
data from the St. Louis RAPS/RAMS network for total suspended particles (TSP)
and carbon monoxide (CO) were analyzed by synoptic map type. Pooler (1974)
EPA, described the criteria, modeling, and objectives of the RAPS program in
detail in his paper on the RAPS network siting requirements. The RAPS network
of 25 stations was arranged in an expanding pattern of four concentric circles
originating from the downtown St. Louis Arch at radial intervals of 4, 10, 20
and 40 kilometers, as shown in Figure 3.
The selection of TSP and CO concentration data to determine possible re-
lationships with weather types was made for several reasons. First, essen-
tially non-reactive pollutants were assumed to provide a situation uncompli-
cated by photochemistry and other rapid reaction systems. Second, the pollu-
tant sources for both pollutants should be widely distributed but with differ-
ent source patterns. Third, and obviously, the data should have been reported
by the RAPS monitoring network. In addition both TSP and CO are important
urban area and regional pollutants and a correlation between these measure-
ments and weather types could be useful, directly, in increasing the under-
standing of air pollution problems in the St. Louis area.
15
-------
• RAPS CENTRAL
• RAMS STATIONS
Figure 3. St. Louis Regional Air Pollution Study air monitoring network,
1975-1976.
16
-------
In our study, we concluded that from the network the four centrally lo-
cated sites—numbers 103, 105, 106 and 108--were the most likely to show how
synoptic weather might influence TSP and CO pollution concentrations. These
four RAMS stations provide an indication of the average levels of TSP and CO
across the central part of the St. Louis commercial and industrial area. It
was expected that changes in synoptic weather, as indicated by weather types,
would give the optimum relationship between the synoptic influence in this
part of the city and the levels of concentration of these pollutants.
Air Quality Data Analysis
In the RAPS program, air quality data were collected on both TSP and CO
for 22 months during the period from January 1975 through December 1976.
Sampling for TSP was done every third day and was the usual 24-hour average
Hi Vol sample collection. Sampling for CO was the usual continuous non-
dispersive IR recording instrumental observation. This sampling pattern
resulted in an approximate three-fold difference in the total number of daily
average values obtained between TSP and CO. One factor that influenced the
selection of station data for analysis was the fact that not all air quality
measurements were taken at all of the RAMS/RAPS stations. Thus some type of
preliminary data screening and station selection was necessary to determine
a suitable procedure for correlating St. Louis air quality with the synoptic
weather types. Several combinations of stations were tried, including net-
work averages, and it was decided to use an average of the four central RAPS/
RAMS network stations previously mentioned, numbers 103, 105, 106 and 108, see
Figure 3.
The next step was to relate air quality concentrations for TSP and CO by
date with the seasonal map type for the same date. After seasonal geometric
means, variances, standard deviations, and coefficients of variance had been
computed for averages of the four selected RAMS stations 103, 105, 106 and
108, the data were analyzed statistically. The geometric mean was chosen to
insure greater sensitivity throughout the statistical population. Indications
of significance between map types was sought. The first test performed on the
data was an analysis of variance and an F test. Uhen the results of this pro-
cedure indicated that there were differences between map types within each
season, a more powerful statistical procedure, the Duncan's new multiple range
test (Steel and Torrie, 1960) was employed to determine more conclusively air
quality differences between pairs of synoptic types. The WSU computer sys-
tems, which incorporated the Statistical Analysis System (SAS) algorithm, was
applied to these data.
Total Suspended Particle Comparisons
The tabular ranking of seasonal maps by high to low TSP concentrations,
map types, number of cases, and average concentration values measured in
yg/m3, for RAMS stations 103, 105, 106 and 108, is given in Table 4. It
should be noted that the National Ambient Air Quality Standard for TSP is
260 yg/m3 averaged over 24 hours. None of the averaged St. Louis data
17
-------
to
o
i—
to
o
u_
Q
LjJ
CD
C£
UJ
00
z
o
1— 1
1—
C£.
Z
UJ
o
o
UJ
1
o
£
o;
CL.
Q
UJ
Z
UJ
Q-
to
to
o
1—
^_
co
o
UJ
^
^^
a;
to oo
1 1 1 <~*i
Q- r—
1 — *
Ct. O
S "~
n
I LO
< o
Z i—
o
to •>
*£. CO
uj o
to i—
,
^^
Ul
i
CQ
1—
CO
£
3.
Q-
to
^~
"00 z
u.
-K
*
a)
a.
I—
co
CD
to
1- 1—
cu
E
^3
to
*
*
cu
Q.
>>
CO
^
^^^
0)
a.
CO
CT>I—
C
•^
s_ z
Q.
to
*
Qi
a.
^
1—
co
E
en
—i
D-
J- 00
C +-
•r- "Z.
3
(U
Q.
>>
I—
V
C
(O
a:
cncniocnLOCM! — ^- co **• «*
«*COCOCMLOCMi — CncMCncM
i — OOOCOCOCOr-»lv--LOLO
^-* ?*— r"- r—
CMiDCn,-(^j-cOLOcnLOcnLO
f— CO CM r— CM
s:<:u.oujQCQ3:i— icocn^cMLOOus^
r— CM r— i— CM
CJ3i^Oi— lU.cC'OUJCOQ — 1
coooOLOLOcor^co
(U
Ol
H3
s_
a>
CO
c
o
•1^
•f-3
(O
(O +->
+3 l/l
re
•^'s
o
C S-
cu
re •» ^)
(O +-> >-3 E
-i-> re 3
re -o -o c
"O C
o re (/>
o c ai
C >— I 4->
•* re
•> ^ l/l U
n: cu T-
cu a. -a
co a. >, c
Q. >>l— T-
>>t— *
1 — * * Z
***•>-
18
-------
approached this level. As shown by the seasonal columns in Table 4, each
classification has, at least qualitatively, three concentration categories of
map types. There is a group at the upper end of the distribution, one at the
low concentration end, and a group that is intermediate between these two.
For example, there are four autumn map types with concentrations >100 yg/m3,
five map types with concentrations between 60 and 100 yg/m3, and two map types
with concentrations <60 yg/m3. In order to examine this phenomenon more
clearly, each season was reviewed separately by map type and TSP concentration
and with reference to the data in Table 4.
(A) Winter - an examination of the five winter weather types L, E, F, A,
and B with TSP concentration levels which exceed 85 yg/m3 indicated simi-
lar wind flow patterns over the St. Louis area, predominantly from the
western quadrants. Additionally, map types A, F, and E, with the highest
concentration levels, that is TSP concentrations of 85 yg/m3 or higher,
were significantly different from the two map types H and G with the low-
est concentration levels, i.e., TSP concentrations of 65 yg/m3 and below.
In the winter group of map types with concentrations between 75 yg/m3 and
85 yg/m3, there was no pattern or much variation of actual TSP concentra-
tions. This was substantiated by the statistical non-significant results
of the Duncan's test comparing map types K, C, D, I, and J.
From a meteorological synoptic standpoint winter weather can include
both stable, stagnant weather situations that favor pollutant accumula-
tion and storm situations that provide excellent pollutant dilution.
With regard to TSP concentrations precipitation and wet ground would tend
to reduce ambient concentrations and surface sources of fugitive dust.
(B) Spring - This season is typically unstable with rapidly moving,
active weather systems. The map types, with their overall lower TSP con-
centrations when compared generally to the other seasons, support this
contention. Despite the seasonal tendency for dynamic instability, map
types G, K, and C had concentrations of 90 yg/m3 or higher and two in
this group with the largest number of cases, C and K, were significantly
different from types A, B, D, and E, whose concentrations were below
65 yg/m3. The weather patterns of the map types with the higher concen-
trations, e.g., 90 yg/m3 or higher, had wind flow from the west, south-
west and northwest sectors. The remaining weather map types had active
synoptic situations where frontal activity was either just approaching
or just receding from the St. Louis area, within the previous six-hour
period.
(C) Summer - Traditionally, air pollution episodes are associated with
strongly modified or quasi-stationary systems of high pressure. When
this situation becomes widespread, air pollution episodes similar to the
ones described by Lynn, et al. (1964) and a number of other investigators
can occur. Accumulation of particulate pollutants leading to wide spread
visible haze has been observed to be highest generally in the western or
southwestern sector of anticyclones moving slowly across the US midwest
and east coast from northwest to southeast, particularly if there is a
frontal system or trough behind the high that brings in upper air clouds
19
-------
which tend to limit mixing and dispersion of pollutants. This western
sector of these modified continental polar anticyclones is frequently
identified as the "backside" of the High. Map type M (summer) is an ex-
ample of this type of phenomenon where St. Louis is on the westside of an
anticyclone centered in West Virginia and shows a high cloud layer, at a
reported 25,000 feet on the August 20, 1975 base map, due to a trailing
front to the northwest. The average TSP concentration of 161 yg/m3 asso-
ciated with this map type exceeds the 150 yg/m3 secondary National Air
Quality Standard for a 24-hour maximum concentration.
Although only two map types, 0 and M, corresponded to daily TSP con-
centrations that were >115 yg/m3, they both were significantly different
than map types F, L, and I which had the lowest concentrations in this
season, i.e., <65 yg/m3.
(D) Autumn - Although this season is normally transitional with frequent
unstable weather conditions, stable conditions are also frequent and
average pollution concentrations were generally higher than in the spring
season.
Four map types, M, A, F, and C, had TSP concentrations of over
100 yg/m3, and they were significantly different than map types G and L
associated with TSP concentrations of <60 yg/m3, the lowest concentra
tions in this seasonal grouping. Wind flow for these four high concen-
tration map types, which were dominated by strong high pressure cells,
ranged from 4 to 10 kts, while those with the lowest concentrations had
velocities from 12 to 16 kts, generally characteristic of good dispersion
characteristics.
As a result of these analyses, it is has been concluded that the various
map classifications by the Lund (1963) statistical technique, used in conjunc-
tion with related TSP concentrations, generally identify synoptic weather pat-
terns with specific pollution concentrations that may be classed as high, med-
ium and low level s.
Table 5 lists the results of comparing the average concentrations by map
type and indicates whether or not the differences between TSP concentrations
for pairs of map types were statistically significant (S) or non-significant
(N). Map types in Table 5 having statistical significance for TSP concentra-
tions, designated S, are generally similar to those map types identified qual-
itatively by ranking in Table 4. Subsequent discussion examines the situation
in more detail. In the winter group of Table 4 map types A, E, and F showed a
significant difference from map types G and H, which correlated with the win-
ter group in Table 4. However, map type L, which has only three cases, did
not show statistical significance, although its average concentration level
was high. This is an example of a common statistical problem in dealing with
small sample numbers and significance testing. In the spring season, map
types C and K were significantly different from map types A, B, D, and E which
correlated with the ranking by concentration of these map types. Map type G,
with the highest concentration and lowest number of cases, again did not mani-
fest statistical significance in this group. In the summer group map types J
and M were significantly different from map types F, L, and I, which agreed
20
-------
TABLE 5. RESULTS OF STATISTICAL SIGNIFICANCE TESTS FOR TSP CONCENTRATION
DIFFERENCES BETWEEN MAP TYPE PAIRS
Statistical Significance Noted by S, Non-significance by N
WINTER SPRING*
ABCDEFGHIJKL ABCDEFGIHKL
ANNNNNSSNNNN ANSNNNNNNNN
B NNNNNNNNNN B SNNNNNNSN
C NNNNNNNNN C SSNNNNNS
D NNNNNNNN D NNNNNSN
E NSSNNNN E NNNNSN
F SSNNNN F NNNNN
G NNNNN G NNNN
H NNNN I NNN
I NNN J N N
J N N K S
SUMMER** AUTUMN***
A B
A N
B
C
D
E
F
G
H
I
J
L
C D
N N
N N
N
E F G
NNN
NNN
NNN
NNN
N N
N
H I J
N S N
N S N
N S N
N S N
N S N
N N S
N S N
S N
S
L
N
N
N
N
N
N
N
N
N
S
M
S
S
N
S
S
S
N
S
S
S
S
A B
A S
B
C
D
E
F
G
H
K
L
C D
N N
N N
N
E
N
N
N
N
F G H
N S N
NNN
N S N
NNN
NNN
S N
N
K
N
N
N
N
N
N
N
N
L M
S S
N S
S N
N N
N N
S N
N S
N N
N N
S
* Map type H missing, no data
** Map types K and M missing, no data
***Map types I, J, M, and N missing, no data
21
-------
with the ranking by concentration of these map types. However, in this in-
stance, map type M, with the lowest number of cases, did indicate statistical
significance. Finally, in the autumn season, map types M, A, F, and C were
significantly different from map types G and L which also corresponded to the
ranking by concentration for this group of maps. The number of cases in this
autumn group was fairly high, i.e., nine or more.
In this study, Duncan's test proved to be a valid procedure for determin-
ing the statistical significance between map type pairs and their correspond-
ing levels of TSP. Not unexpectedly, it appears that when the number of cases
involved in the statistical computations was much less than 10, it was usually
not possible to demonstrate significance. Although the one case in the summer
group involving map type M was an exception to this pattern, the especially
high concentration average of 161.0 yg/nr was probably a contributing factor.
Carbon Monoxide
Carbon monoxide concentrations and weather type relationships were also
examined qualitatively by concentration ranking and statistically using the
Duncan's multi pi e-range test. Two methods of data averaging, a total daily
basis and by peak traffic periods, were used in this phase of the study.
First, seasonal tabulations of simple daily averages from the four cen-
tral area RAMS stations (103, 105, 106 and 108) were sorted according to map
seasonal type. The map types were then ranked according to average CO con-
centrations. The results are shown in Table 6. The indicated number of ob-
servations, N, is the number of daily average station values for CO. At
least 13 hourly averages from a given station were considered necessary to
produce a daily average value from that station. In a second data file the
CO data were averaged to include observations made during the morning (0600 to
1000) and evening (1600 to 2000) traffic periods. This would tend to identify
maximum local concentration situations. Neither of these two groupings of the
CO data, the 24-hour daily average or the morning plus evening peak traffic
average, correspond to the specifications of the National Ambient Air Quality
Standards for CO. These are 9 ppm for a 8-hour average and 35 ppm for a 1-
hour average, neither to be exceeded more than once per year.
The daily average CO concentrations shown in Table 6 indicate variations
by factors of two to three between maximum and minimum 24-hour average values
by map types, intra-seasonally. As with the TSP values, the differences be-
tween the seasonal map types having the highest and lowest concentrations
tested statistically significant at the 5% level. A comparison of the rank-
ings of map type by pollutant concentration shown for TSP in Table 4 and for
CO in Table 6 indicates that there is generally a different order of map types
for each of these two pollutants. Except for the map types discussed below,
there is little similarity between the order of intraseasonal ranking between
map types for TSP and CO when the higher and lower concentration categories
are considered. For example, in the autumn when TSP values were generally
higher than in the other seasons, map types M, A, and F had the highest aver-
age TSP concentrations. For CO concentrations, these same map types ranked
fifth, seventh, and second, with only type F in both lists. The three highest
22
-------
TABLE 6. SEASONAL MAP TYPES RANKED BY CARBON MONOXIDE CONCENTRATIONS AVERAGED
FOR STATIONS 103, 105, 106, and 108
Winter Spring Summer Autumn
Rank Type N Cone (ppm) Type N Cone (ppm) Type N Cone (ppm) Type N Cone (ppm)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
N
F
J
C
E
L
A
K
D
M
B
G
H
I
8
25
18
29
37
14
139
8
28
10
86
53
52
35
1.081
1.050
.823
.674
.636
.617
.556
.505
.490
.472
.449
.419
.324
.300
I
J
F
L
E
H
D
A
G
C
B
K
14
13
33
9
51
20
42
98
25
79
99
19
2.260
1.277
1.045
.962
.960
.934
.893
.838
.772
.736
.718
.705
M
K
A
J
F
I
E
D
B
H
L
C
G
7
4
107
26
30
19
39
68
106
53
24
49
20
.768
.542
.471
.447
.416
.408
.397
.391
.382
.370
.341
.320
.309
H
F
L
B
M
C
A
K
J
G
D
E
I
29
30
8
105
49
66
136
11
4
45
30
36
7
.572
.509
.477
.459
.436
.388
.383
.376
.360
.357
.339
.320
.213
*n = number of daily station averages where a minimum of 13 hourly averages
are required to calculate a daily average.
23
-------
concentration CO maps in the autumn were H, F, and L; for TSP these types
ranked eighth, third, and eleventh. The other seasons show a similar lack of
close correlation between the concentration ranking of these two pollutants
and given map types. An explanation for this situation is not readily appar-
ent, and this is an area for research in future investigations.
In the compilation of CO concentration on the basis of averages over the
two, four-hour peak traffic periods for our 4-station data set, there seemed
to be a strong bias toward the winter season and toward some relatively high
individual concentration levels within some of the wintertime synoptic weather
situations. There were, in all, seven map types with relatively high traffic
period CO. Six of these were winter map types, A, C, I, J, K, and L, with
concentrations ranging from a low of 6.4 ppm to a high of 35.2 ppm. There
was one summer map type, M, which had a relatively high concentration of 11.3
ppm. These seven were the only types during the RAPS period of 1975 and 1976
which could be classed as potentially significant in terms of CO air quality
standard, 9 ppb for a continuous eight-hour average. The winter season ap-
pears to have the most frequently occurring weather that is conducive to
occasional local carbon monoxide buildup. There was one characteristic that
was common to five of the six winter map types: a southerly wind flow with a
moderate speed of 10 to 12 mph. The one winter type that differed from this
pattern, L (W), had a northwesterly flow at about 10 mph. During the summer
season, the one significant map type, M(S), had a southerly flow. Generally,
wind flows from the south resulted in higher concentration levels of pollu-
tants in most instances.
The application of the weather typing concept to include the local, peak
traffic situations does not appear to be productive. It is likely that local-
or short-term weather factors, such as wintertime urban stability conditions,
might be difficult to categorize on the basis of larger scale weather systems.
Additionally, local effects in the vicinity of the sampling sites due to traf-
fic peaks, day of the week, traffic changes, and industrial operations might
be creating enough interference to mask detectable synoptic weather influences
Some station location and sampling problems that were encountered in the RAPS
CO monitoring program have been examined by McClenny and Chaney (1978).
An analysis for statistical significance by the Duncan's test similar to
that discussed above for TSP was performed for the daily average carbon mon-
oxide concentrations. The results are tabulated in Table 7 where "S" indi-
cates a statistically significant difference between the map type pair and,
"N" indicates non-significance. In the winter season, the two highest CO con-
centration map types, N and F, were significantly different from the two low-
est CO concentration map types H and I. Both the high and low concentration
categories were significantly different from the intermediate map types B and
D. The third highest winter type, J, showed a significant difference from map
types B, G, H and I, but not from D. The number of cases that were available
in each category was important in the statistical outcome of the Duncan's test
in the CO comparison as it was in the TSP analysis. This is particularly ap-
parent in the summer map types M and K with 7 and 4 cases, respectively, where
only minimum significance could be established.
24
-------
TABLE 7. RESULTS OF STATISTICAL SIGNIFICANCE TESTS FOR
DIFFERENCES BETWEEN MAP TYPE PAIRS
:0 CONCENTRATION
Statistical Significance Noted by S, Non-significance by N
WINTER
ABC
ANN
B S
C
D
E
F
G
H
I
K
L
M
ABC
A N S
B N
C
D
E
F
G
H
I
J
K
L
DEFGHIJKLMN
NNSNSSNNNNN
NNSNSSSNNNS
NNNSSSNNNNN
NSNNSNNNNS
SSSSNNNNN
SSSNNNSN
N N S N N N S
N S N S N S
S N S N S
N N N N
N N N
N
SUMMER
DEFGHIJKLM
NNNNNNNNNN
NNNNNNNNNN
NNNNNNNNNS
NNNNNNNNN
NNNNNNNN
N N N N N N N
N N N N N S
N N N N N
N N N N
N N N
N N
N
SPRING
ABCDEFGHIJKL
ANNNNNNNSNNN
B
C
D
E
F
G
H
I
J
K
N N N
N N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
S
S
S
S
S
S
S
S
S
N
N
N
N
N
S
N
N
N
N
N
N
N
S
N
N
N
N
N
N
N
N
S
N
N
AUTUMN
ABCDEFGHIJKL
A
B
C
D
E
F
G
H
I
J
K
NNNNNNSNNNN
NNSNNNSNNN
NNNNNNNNN
N N N
N N
N N N N
N N N N
N N S N N N
S N N N N
S N N N
N N N
N N
N
25
-------
SECTION 3
SUMMARY AND CONCLUSIONS
The statistical map classification process developed by Lund (1963) was
used to compile a comprehensive set of synoptic weather map types for the
St. Louis, Missouri, region on a seasonal basis for the four-year period 1973-
1976. This covered the 1975-76 operational period of the EPA Regional Air
Pollution Study centered in St. Louis. The map types were used along with av-
erage TSP and CO concentration data from the central area of the St. Louis
RAPS network to assess average TSP and CO pollutant concentrations on the bas-
is of seasonal synoptic weather types.
Statistical methods including analysis of variance and Duncan's multiple-
range test were used to test the differences between the pollutant means cal-
culated for the various seasonal synoptic weather types. The analysis concen-
trated on conditions measured at four central St. Louis RAPS stations, numbers
103, 105, 106, and 108.
As would be expected from considerable prior information synoptic scale
weather, when expressed as weather types, influences the pollutant concentra-
tions of both TSP and CO, sometimes to a significant degree. For example, the
TSP map types may be easily categorized into three pollutant concentration
categories of high, medium, and low relative to the seasonal average. One
summer map classification, type M, with an average 4-station concentration of
161 yg/m^ can be compared to the "classical" summer anticyclonic model which
is noted for its high levels of pollutant concentrations. The general wind
directions in the St. Louis area as related to the synoptic situations, par-
ticularly in the spring season, also appear to play an important role in the
observed concentrations of pollution.
The differences between carbon monoxide concentrations for selected map
types were also significant. However, there was a different set of map types
associated with high concentrations of carbon monoxide than with high TSP lev-
els. Future research should investigate this aspect.
Although there have been difficulties mentioned in this report with re-
gard to the statistical treatment of the data file, the results of this inves-
tigation comparing selected weather classifications with air quality should
provide a reasonable basis for evaluating the RAPS data and for further re-
search. The tabulation of daily weather by an objective weather typing tech-
nique should have applications to other research in the modeling of air pollu-
tion conditions in the St. Louis area. This research program is continuing
with the particular goal of developing, if possible, objective weather typing
26
-------
systems that are more closely related to air pollutant concentration patterns
than the 1963 sea level pressure pattern correlation system of Lund used in
this report.
27
-------
REFERENCES
1. Air Pollution Research Section, Department of Chemical Engineering,
College of Engineering, Washington State University, Pullman, Washington.
Atmospheric Drift of 2,4-D in the Lower Yakima Valley 1974, 1975, 1976,
1977, and 1978.
2. Arai, Y. Classification of 500 mb Patterns Related to the Abnormal,
January, 1963. Papers in Meteorology and Geophysics 15:93-114, 1964.
3. Bardin, G. I. Typification of Pressure Fields According to Their Nat-
ural Orthogonal Functions. Problemz Arktiki i Antarkitiki, Leningrad
(32):52-61, 1969. (See: Meteorological and Geoastrophysical Abstracts
22(4)231, April 1971.)
4. Christensen, W. I., Jr., and R. A. Bryson. An Investigation of the Po-
tential of Component Analysis for Weather Classification. Monthly
Weather Review 94(12):697-709, December 1966.
5. Theophrastus. DeVentis, edited by V. Coutant and V. L. Eichenlaub,
University of Notre Dame Press, 1975.
6. Elliott, R. D. Extended-range Forecasting by Weather Types, edited by
T. F. Mai one, Compendium Meteorology, American Meteorological Society,
Boston, MA, 1951. pp. 834-840.
7. Fliri, F and M. Schuepp. International Congress on Alpine Meteorology,
9th Brig, Switzerland, September 14-17, 1966, Wissenschaftliche
Abhandlungen, Zurich, 1967. Meteorological Geoastrophysical Abstracts
19(15):215-229, May 1968.
8. Halter, B. Site Evaluation Research for NOAA/GMCC Program. Thesis,
Washington State University, Environmental Science Program, February,
1976.
9. Krick, I. P. and R. D. Elliott. Synoptic Weather Types of North
America, Meteorology Department, California Institute of Technology,
Pasadena, CA, December 1943.
10. Lowry, W. P. and F. Probald. An Attempt to Detect the Effects of a
Steelworks on Precipitation Amounts in Central Hungary. Journal of
Applied Meteorology 17(7):964-975, July 1978.
28
-------
11. Lund, I. A. Map-pattern Classification by Statistical Methods. Journal
of Applied Meteorology 2:56-65, February 1963.
12. Lund, I. A. Correlations Between Area! Precipitation and 850-millibar
Geopotential Height. Monthly Weather Review 99:691-697, September 1971.
13. Lynn, D. A., B. J. Steigerwald, and J. H. Ludwig. The November-December
1962 Air Pollution Episode in the Eastern United States. U.S. Depart-
ment of Health, Education, and Welfare, Division of Air Pollution,
Cincinnati, OH, September 1964.
14. McClenny, W. A. and L. W. Chaney. Pollutant Variability in the Regional
Air Pollution Study. Journal of Air Pollution Control Association, 28
(7):693-696, July 1978.
15. Peckenpaugh, J. A., L. M. Reisinger, and E. Robinson. Long-Distant
Atmospheric Transport of 2,4-D. Paper No. 1974-12 presented at the
Pacific Northwest International Sec'tion, Air Pollution Control Associa-
tion, Boise, Idaho, November 17, 1974.
16. Pooler, F., Jr. Network Requirements for the St. Louis Regional Air
Pollution Study. Journal of Air Pollution Control Association, 24(3):
228-231, March 1974.
17. Singh, S. V., D. A. Mooley, and R. H. Kripalani. Synoptic Climatology
of the Daily 700 mb Summer Monsoon Flow Patterns Over India. Monthly
Weather Review, (4):510-525, September 1971.
18. Steel, R. G. D. and J. H. Torri. Principles and Procedures of Statis-
tics. Chapter 7, McGraw-Hill Book Company, Inc., 1960. pp. 99-131.
19. Westberg, H., E. Robinson, D. Elias, and K. All wine. Studies of Oxidant
Transport Beyond Urban Areas, New England and Sea Breeze - 1975. EPA-
600/3-77-055, June 1977.
20.. Westberg, H., K. All wine, E. Robinson, and P. Zimmerman. Light Hydro-
carbon and Transport Studies in Ohio - 1974. EPA-600/3-78-007, January
1978.
21. P. Zdravko. Weather in Slovenia and its Development by Regional Weather
Types for the Five-year Period 1955-59. Drustvo Meteorology Slovenije,
Ljubljana, Raszprave, 8:31-60, 1967.
29
-------
Appendix
Base Maps for Synoptic
Weather Map Types
Section 1 Winter
Section 2 Spring
Section 3 Summer
Section 4 Autumn
30
-------
Appendix A - Section 1
Winter Map Types
31
-------
-------
33
-------
\ "I^K..
V tn K- ' «^-t ,/v, ' —'• •-
*?- 5TS-S =5 % »
•"•>*. «u AS
34
-------
;\
-------
-------
-------
38
-------
-------
-------
*YA** ^\
^^iM**:«Mj^*C^*,i \L/ '
&V*to
>\-' A"Jl rr'^
^ ^. .1 rt A_J-i *-*¥A 1\ ' l
s$^^Qt<$,/_
-------
-------
^ppT^ ' s J°; ^
i—T—^}
-------
Appendix A - Section 2
Spring Map Types
44
-------
-------
\
-------
\
-------
£-;'/*• ES* -i. '«
• .*-«—
.
43
-------
\ \ \
49
-------
-------
-------
-------
-------
\
54
-------
-------
-------
?.
"S^V*
v"5 -^
3* a.
.£< 19
~ •-«£
r-S?
9- i- ~° • .,,£¥ '-;
•,-"•-° ^.-L ,:;-'l. ^-"
»«• ?r" '?»-, '" "'J- Sr"' *'"'•• '"'• ' - -~ "6
.-•^-j,-is „.-:«?:-. :.^ -t; - •>, .^ • :•.,- 0 - --
'>- ~ r^' .," e- ^ ' ?T^vS - ^^ ^ *-?'-"-
s: vo
r^
Oi c-v
0.,-H
>1
^i- -i. - ' ' Q
«, -5 & "s. =" ^!
T_ T-. ao.^, ^ O n->
-------
Appendix A - Section 3
Summer Map Types
58
-------
-------
00
60
-------
*,*-£**&
61
-------
-------
-------
64
-------
O CO (/>
o =
-------
66
-------
o-
67
-------
68
-------
•"V
, , CM "? i ; —j -= i -
i Vv' < °
u r^r-J i -*
1/1 LPui I Z O
./- O K, =-
-------
70
-------
-------
Appendix A - Section 4
Autumn Map Types
72
-------
•^4^ ^ ^jps% * r %'& "V^ a ™
:;7l\. j ,°; ""%'. %' 'i~Ss- '-'i '•£' ^_fj "~' "' ^i' ::,4 "§
Xiil H:i ^~i it®
• J '- ie ~'«
~' 1
i^ I.T , f,
T\\ ^ ^ / ~" "^ r.-^.
•V -i, yi' ' S- -v-'
•" ^§ . •«' i
'e
-------
^Ma^irii i
s^s
'= \x^U^°
- ~a '~vo Ji ^T 2?
& & '*'•:&„<&.
£
". ^|^\5 '*«
»g .^'-
TT1^"^ IOT--
74
-------
c-X^V /WSo- r^\ £>A V " ' ^ & /
&£&!** &. ^ - s"fe / ^
F-^-^iBi^r »^'1 rs -/Si *S
©3ri-jfii^' ^( a- aSi / ^ "
VS,k J,*s~«i; s- ?i fii ^ r /, gi .£ j
^.fr"^^ ~X ?a.^ ?S5 / S'^ , To%
{~5i\ ; j* i £,* P:iai / i e • / -s. n/.
-------
-------
77
-------
\
-------
\
Jt/s, In
-------
s»
------- OCR error (C:\Conversion\JobRoot\00000CUU\tiff\20015O6Z.tif): Unspecified error
-------
<5
v
\
< O Z =
UJ O ,* «/)
^ ^o"
< "
s - •
Q Sz^
K-. O • T*
^ O CO O
2 03«
N
-------
-------
-------
\
-------
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1 REPORT NO.
EPA-600/4-80-001
3. RECIPIENT'S ACCESSION-NO.
4. TITLE AND SUBTITLE
SYNOPTIC METEOROLOGY AND AIR QUALITY' PATTERNS IN THE
ST. LOUIS RAPS PROGRAM
5. REPORT DATE
January 1980
6. PERFORMING ORGANIZATION CODE
7 AUTHOR(S)
Elmer Robinson and Richard J. Boyle
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Washington State University
Pullman, Washington 99164
10. PROGRAM ELEMENT NO.
1AA603 AE-006 (FY-79)
11. CONTRACT/GRANT NO.
805142
12. SPONSORING AGENCY NAME AND ADDRESS
13. TYPE OF REPORT AND PERIOD COVERED
Environmental Sciences Research Laboratory - RTP, NC
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
Final 5/77 - 10/79
14. SPONSORING AGENCY CODE
EPA/600/09
15. SUPPLEMENTARY NOTES
16. ABSTRACT
An objective, statistical synoptic weather map classification scheme developed
by Lund to stratify map patterns for further study was used to type regional
weather patterns. The investigation extended over a 500-mile radius of the greater
St. Louis area and was intended for subsequent application to air pollution studies.
This analysis correlated sea level pressure data at 21 selected National Weather
Service stations at a specified time on a given day, with sea level pressures for
these same stations on each of the other days in the four-year period 1973 through
1976. A total of 52 separate weather map types were identified with 12 map types in
the winter season, 13 map types in the spring season, 13 map types in the summer
season, and 14 map types in the autumn season.
To illustrate the potential usefulness of this method to air quality and
synoptic weather relationships, a preliminary comparison of the weather typing
system to suspended particles and carbon monoxide concentrations was also made.
Results indicate that the typing program is able to show that synoptic-scale
weather does affect relative pollution levels in the St. Louis region, if definitive
synoptic model types are carefully chosen prior to application to pollutatn con-
centration levels.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
c. COS AT I Field/Group
Air pollution
* Synoptic meteorology
* Meteorological charts
* Atmospheric pressure
St. Louis, MO
13B
04B
08B
13. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19. SECURITY CLASS (ThisReport)
UNCLASSIFIED
21. NO. OF PAGES
95
20 SECURITY CLASS (Thispage)
UNCLASSIFIED
22. PRICE
EPA Form 2220-1 (9-73)
37
------- |