&EFK
United States
Environmental Protection
Agency
Environmental Sciences Research EPA 600 2-79183
Laboratory September 1979
Research Triangle Park NC 27711
Research and Development
Microscale
Variations in
Ambient
Concentrations of
Pollutants in
St. Louis Air
-------
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 PROTECTION TECH-
NOLOGY series. This series describes research performed to develop and dem-
onstrate instrumentation, equipment, and methodology to repair or prevent en-
vironmental degradation from point and non-point sources of pollution. This work
provides the new or improved technology required for the control and treatment
of pollution sources to meet environmental quality standards.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/2-79-183
September 1979
MICROSCALE VARIATIONS IN AMBIENT CONCENTRATIONS
OF POLLUTIONS IN ST. LOUIS AIR
by
Lucian W. Chaney
The University of Michigan
Ann Arbor, Michigan 18109
Grant No. R803300
Project Officer
W.A. McClenny
Atmospheric Chemistry and Physics Division
Environmental Sciences Research Laboratory
Research Triangle Park, N.C. 27711
ENVIRONMENTAL SCIENCES RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, N.C. 27711
-------
DISCLAIMER
This report has been reviewed by the Environmental Sciences
Research Laboratory, U.S. Environmental Protection Agency, and
approved for publication. Approval does not signify that the
contents necessarily reflect the views and policies of the U.S.
Environmental Protection Agency, nor does mention of trade names
or commercial products constitute endorsement or recommendation
for use.
ii
-------
ABSTRACT
As part of the Regional Air Pollution Study (RAPS), a
series of studies were carried out in St. Louis during the
summers of 1974, 1975, and 1976 primarily to determine the sub-
grid concentrations of ambient air pollution. One primary
pollutant gas, CO, and one secondary pollutant gas, ozone, were
chosen to be representative. Methodology for determining sub-
grid concentration variations of these gases is discussed.
Portable monitors and the collection and analysis of bag
samples were used to determine pollutant concentrations. In some
cases the monitors were moved along selected paths while the
measurements were made; in other cases the monitors were placed
at selected sub-grid locations. The data were collected at six
sites during the first year, and at two sites during the final
two years. Both urban and rural sites were selected. All the
data were collected during daylight hours generally between 10:00
a.m. and 4-: 00 p.m.
This report was submitted in fulfillment of Grant R-803300
by the University of Michigan under the sponsorship of the U.S.
Environmental Protection Agency. It covers a period from
November 1, 1974- to March 31, 1979 and the work was completed
as of December 31, 1978.
111
-------
IV
-------
CONTENTS
Abstract ,., iii
Figures ,.,....,,.. , v
Tables, , vi
Abbreviations and Symbols........ ,. vii
Acknowledgements viii
1. Introduction 1
2. Conclusions 12
3. Recommendations ,. ., 14
4, Preparations
Monitor selection 15
Recorder selection 18
Calibration standards 18
Cooperative arrangements 20
Site selection 21
Personnel and facilities 22
5. Methodology
Monitoring paths 23
Helicopter supplemental data techniques 29
Field calibrations 29
Data collection procedures. 33
Data reduction procedure 33
6. Results
1974 ozone 35
1975 ozone and carbon monoxide 35
19 76 carbon monoxide 36
Data presentation 36
Probable sources of error. 41
7. Discussion 44
References 52
-------
FIGURES
Number
1 Regional Air Monitoring Systeni, Site Locations 3
2 Regional Air Monitoring Site 102 6
3 Regional Air Monitoring Site 103 7
4 Regional Air Monitoring Site 105 8
5 Regional Air Monitoring Site 106 9
6 Regional Air Monitoring Site 108 10
7 Regional Air Monitoring Site 113 11
8 Portable Ozone Monitor on Back-pack H>
9 Sub-grid Data Collection by Back-packing 23
Portable Monitors
10 Side by Side Comparison around St, Louis 25
University
11 Side by Side Comparison near Site 106 26
12 Area Plot around St. Louis University 27
13 Sub-grid Data Collection by Fixed Location 28
14 Ozone Sub-grid Data at Site 105 8/12/75 morning 30
15 Ozone Sub-grid Data at Site 108 8/12/75 midday 31
16 Ozone Sub-grid Data at Site 105 8/12/75 afternoon 32
17 Grand Ave. CO Concentration, August 20, 1976 37
18 Ozone Correlation Plot RAMS Station 113 39
19 Space and Time Scale Correlation 48
20 Pollutant Area Scale Comparison 49
21 Pollutant Variation vs. Time Comparison 49
vi
-------
TABLES
Number Page
1 RAMS Monitor Ozone Concentrations 38
2 RAMS Monitor CO Concentrations 40
3 Intensive Studies with Helicopter Support 43
RAMS Site 1Q8
vii
-------
ABBREVIATIONS AND SYMBOLS
AID
CO
cxr
EPA
GE
GFC
Hz
JAPCA
KI
MIT
NBS
NERC
NO
NO-
NO
0 X
ppb-v
ppm-v
RAMS
RAPS
RTP
SLU
SO.,
SRfl
Analytical Instrument Development Company
Carbon Dioxide
Carbon Monoxide
Environmental Protection Agency
General Electric Company
Gas Filter Correlation
Cycles per second
Journal of Air Pollution Control Association
Potassium Iodide
Massachusetts Institute of Technology
National Bureau of Standards
National Environmental Research Center
Nitric Oxide
Nitrogen Dioxide
Nitrogen Oxides
Ozone
parts per billion by volume
parts per million by volume
Regional Air Monitoring System
Regional Air Pollution Study
Research Triangle Park, N.C.
St. ILouis University
Sulfur Dioxide
Standard Reference Material
viii
-------
ACKNOWLEDGEMENTS
The cooperation of Rockwell International Air Monitoring
Center is gratefully acknowledged and, in particular, Mr, Al
Jones who supplied all the preliminary test site information
as well as the other Rockwell personnel who responded to the
numerous requests for RAMS data.
The Environmental Protection Agency Regional Air Pollution
Study permanent staff was very helpful in many ways, especially
Mr. Stanley Kopczynski who arranged instrument calibration
tests and Mr. William Best who helped make the arrangements for
the installation of the long path instruments.
Thanks are extended to the Anheuser-Bush and Monsanto com-
panies for permitting the placement of mirror reflectors on
their property.
The author is especially grateful to St. Louis University
and to Dr. John Forsberg who made laboratory space available
for daily calibrations.
Thanks also go to the student staff for their whole-hearted
cooperation in collecting most of the data; Jeff Kochelek,
Michael Travis, and Carol Chaney.
The reviews and helpful comments received from EPA Environ-
mental Research Center staff members Dr.. David Mage, Mr. Ray-
mond Rhodes and Mr. Arthur Coleman are sincerely appreciated.
The author thanks Dr. W.A. McClenny for his many suggestions
and review of this report prior to final preparation.
-------
SECTION 1
INTRODUCTION
The study evolved as a result of the EPA air pollution con-
trol objectives. One of the major objectives has been-,the de-
velopment of urban air quality simulation (AQS) models . Cur-
rently many models a,re being developed and several will be
applied to the RAPS data base, A set of AQS models will even-
tually be used to determine the environmental impact including
the health risk of all proposed new construction, such as, man-
ufacturing plants, shopping, centers, roadways, etc., on a given
total urban complex. Of course, impact studies are currently
required for all large projects and many are availabe for this
purpose. However, all the routinely usable models are relative-
ly simple and are usually applicable on the microscale rather
than on the meso or urban scale. As the scale increases in
magnitude the model complication increases rapidly in a very
non-linear fashion.
The nature and complexity of the modeling problems have
been discussed and outlined (Johnson, 1972)3. jje also identi-
fied the important research needs. The most important identi-
fied need was an adequate data base for validating both existing
and future models. The "RAPS program was designed primarily to
meet this need. The initial RAPS.program was described in a
companion paper (McCormick, 1972) . This paper identified the
need for area averaged measurements and outlined the original
plan to use a laser to make area wide measurements. The de-
tailed RAPS measurement plans were subsequently described
(Pooler 1974) and the measurement techniques and the data
collected has been reviewed (Schiermeier 1978) .
The original area measurement plan was to use a laser with
two turning mirrors to trace a triangular laser path around the
monitoring station. The height of the path and the precise form
would be determined in the field. It was suggested that the
laser path measurements be made at three typical sites such as;
downtown, urban, and suburban. It would have been desirable to
collect the data continuously over some extended period (2 weeks)
at each site.
The original objective of the study reported here was to
meet this challenge and collect data suitable for storage in
the RAPS data bank. The collection of such data was a goal
throughout the three year study period. However, it was known
from the start that the laser program might not develop rapidly
1
-------
enough and area averaged data would have to be collected using
conventional monitors as we have reported here.
It should be emphasized that the collection of data with
conventional point monitors at a cluster of points inside the
grid area is a significant departure from the proposed spatial-
ly integrated, measurements. The proposed measurements could
have been compared directly with the model calculations. The
information collected can only advise the modeler on how he
might interpret the point measurements.
During the RAPS program a large share of the total project
effort was directed towards the development of the laser sy-
stem. Due to problems with frequency drift and calibration
neither laser system would function unattended. However, a
limited amount of CO data was collected by MIT Lincoln Labora-
tories for storage in the RAPS data bank.
Following the completion of the RAPS program considerable
effort has been devoted to the continuing development of both
laser systems. At the present time (Jan. 1979) it is felt
that the fundamental problems of frequency stability and multi-
moding of the diode laser have been solved. A continuing effort
is being directed towards solving the non-uniform energy distri-
bution in the CO, laser. In the event of a future RAPS program
the collection of valid area averaged data seems almost cer-
tain.
The studies being reported here were all conducted inside
a basic 2.8 km grid area and have been termed sub-grid studies.
However, they belong to the general class of air pollution
studies termed microscale. Microscale studies are usually re-
stricted to distances of no more than few feet to approxi-
mately one kilometer. The purpose of the studies has been to
determine the pollutant variability at RAMS sites (Fig. 1)
utilized by the RAPS program.
Pooler , in an overall description of the program, recog-
nized that the RAPS fixed point monitors located at 25 stations
throughout the St. Louis metropolitan area could measure the
pollutant concentration only at given points. However, atmos-
pheric modeling can only predict the averagejconcentration over
grid volumes with bases of no less than 4 km extending to the
inversion height. Hence, there was a need to either determine
the monitoring station bias or directly measure the volume
average.
Specifically, the studies have addressed the following pro-
position: Given an air quality simulation model which will pre-
dict the average concentration in a defined volume:
(1) What are the spatial and temporal characteristics of
2
-------
HJ&-GKIB srooy SITES
Figure 1. Regional Air Monitoring System Site Locations.
3
-------
pollutant concentrations inside the grid volume?
(2) How accurately can a point measurement made at a
specific location, for the purpose of verifying the
model, represent the volume wide concentration?
(3) Can a calibration factor be applied to a point
measurement to estimate more accurately the volume
average?
The objective of the studies described in this report has
been to answer the posed questions. The methodology and tech-
niques employed during the study depended on the instrumenta-
tion both available and under development. Concurrent EPA
sponsored programs were directed toward the development of
new air monitoring instrumentation. The EPA sponsored deve-
lopments which were evaluated for use in meeting the study
objectives were a tuneable diode laser system developed by
MIT Lincoln Laboratory,7 a CO2 laser system developed by the
General Electric Company,8 ana gas filter orrelation spectro-
meters developed by both Ford Aeronutronic and Science Appli-
cations, Inc. In addition, several commercially developed
portable monitors were evaluated and used for the study.
The studies were performed in St. Louis concurrently with
the RAPSsumroer intensive studies of 1974-1976. These time
periods were chosen since many of the RAPS ancillary services
such as meteorological information and helicopter data were
not available at other times. The summer intensive periods
also presented an opportunity to compare results with other
investigators.
The 1974 studies were primarily ozone measurements made at
six selected RAMS sites (Figs. 2-7). These measurements indi-
cated good correlation between the RAMS measurements and the
area, averages determined with the portable monitors.
During the second year, an effort was made to include mea-
surements of more pollutants and, as a consequence, the mea-
surements were confined to two stations, 105 an inner city site
(Fig. 4) and 108 a rural site (Fig. 6). The ozone area measure-
ments again showed good correlation with the RAMS measurements,
but the carbon monoxide data correlated poorly. The efforts
to measure SO- and NO/NO on an area wide basis were unsuc-
cessful and were dropped from further consideration.
The measurements made during the final year 1976 were limi-
ted to carbon monoxide. The measurements made the previous
year at RAMS sites 105 and 108 were repeated. The measurements
confirmed the previous year's results. Additional studies were
made at two roadway sites and some helicopter data was obtained.
-------
The conclusions and recommendations are presented in the
following two sections. A description of the overall study
is given in Sections 4 and 5. The data which has been tabu-
lated according to the pollutant measured and the RAMS site,
has been summarized by best fit linear equations given in the
results. The discussion represents an attempt to relate this
study to the overall air pollution measurement problem.
-------
Figure 2. Regional Air Monitoring Site 102.
6
-------
Figure 3. Regional Air Monitoring Site 103.
7
-------
00
ST. LOUIS RAMS SITE 105
2 km RADIUS
-*•"•*- LONG PATH MONITOR
—— POlNflONiTOR PATHS
ANHEUSER-BUSCH
Figure 4. Regional Air Monitoring Site 105.
-------
B) MAGNOLIA
TOWER GROVE PARK
ARSENAL
Figure 5' Regional Air Monitoring Site 106;
-------
LHELICOPTER FLIGHT
1 PATH
Figure 6. Regional Air Monitoring Site 108.
10
-------
X (A)
Figure 7. Regional Air Monitoring Site 113.
11
-------
SECTION 2
CONCLUSIONS
MICROSCALE VARIATIONS
The microscale variations in the ambient concentration of
air pollutants are dominated by local sources, usually automo-
bile traffic. The temporal and spatial variations can be large.
The temporal fluctuation in the CO concentration was measured to
be as much as two orders of magnitude in 30 seconds and the spa-
tial variation was measured to be as large as one order of
magnitude over a distance of 100 meters. On the other hand,
the ozone concentrations at rural stations have been measured
to be constant within 2% over a distance of 50 km. The magni-
tude of the temporal and spatial variation inside the grid area
depends on the pollutant, the sources Inside the grid area, the
location of the monitor with respect to the sources, and the
meteorological conditions at the time the measurements are
made.
AREA AVERAGES
The measurements of ozone concentrations at six RAMS sites
established that whenever the ozone concentration reached sum-
mer values of 50 ppb to 200 ppb, the station measurements re-
presented the area averages within 10%. Correction factors
within the 10% range were determined for each measured site.
The CO measurements were made at two sites. One, an inner
city site, RAMS 105, and the other a close rural site RAMS 108.
These measurements established two facts: (1) The RAMS reported
CO concentrations have a high probability of being low due to
undersampling the data. The values measured during the study
were on an average low by 50%. (2) The coefficient of varia-
tions of the station measurements from the measured area aver-
ages were 15% for the rural station and 60% for the inner city
station.
LONG PATH MONITORS
Considerable effort was devoted to the evaluation of both
the long path ozone and long path carbon monoxide monitors.
Data taken in test runs of the two systems showed good agree-
ment between point monitor readings and long path monitor rea-
12
-------
ding, demonstrating the feasibility of making long path
measurements. Monitoring of CO at the RAMS sites 105 and 108
with the diode laser system was successfully accomplished du-
ring the 1975 summer intensive period10 with less extensive
monitoring periods in summers 1974 and 1976. Monitoring of
0-, at the RAMS site 103 with the C02 laser based system occurred
during the 1974 summer intensive. However, the system "zero"
prevented any systematic data collection.
For both systems, long path monitoring was limited to single
path averages instead of the area monitoring originally envi-
sioned.
13
-------
SECTION 3
RECOMMENDATIONS
DATA SAMPLING
This study has shown that the RAMS often reported, as a
result of under-sampling the data, low CO values. It is im-
perative that whenever an air monitoring station is to be esta-
blished, a determination should be made that the data sampling
rate is twice the highest frequency appearing in the data. This
can be done by either selecting a sufficiently high sampling
rate or integrating the original data to eliminate the high
frequency components.
STATION SITE SELECTION
This s£udy has shown that the concentrations measured by
an air monitoring station are strongly affected by its location
relative to local sources. If the data collected is to be cha-
racteristic of the urban area, then the station should be sited
100 meters and preferably 200 meters away from any local source.
LONG PATH CO LASER MONITOR
The long path CO monitor development should be continued
in order to demonstrate the capability of making routine field
measurements. This will require further development of the
monitor itself and the design and implementation of a demon-
stration study. The demonstration study could be one of the
following: a roadside emission survey to verify existing mo-
dels, an area emission study to verify existing emission in-
ventories, a shopping center emission study, or an airport
emission survey.
LONG PATH OZONE LASER MONITOR
The monitor which is currently under development should be
evaluated to determine its effectiveness, and the feasibility
of applying the same technique to other pollutant gases.
14
-------
SECTION 4
PREPARATIONS
MONITOR SELECTION
The pollutant variability study was designed to answer
basic questions regarding the variability inside a given grid
area. In the initial consideration of the methodology to be
employed in the conduct of the study, monitoring consideration
was given to all of the criteria pollutants: that is, all pol-
lutants for which maximum dosage levels have been established.
The pollutants that fall into this classification are ozone,
carbon monoxide, sulphur dioxide, nitric oxide, and non-methane
hydrocarbons. The criteria pollutants are also the pollutants
for which various monitors have been developed and are contin-
uing to be developed. It was, of course, desirable to use the
most modern techniques available and to measure as many of the
pollutants as possible. Carbon monoxide, non-methane hydro-
carbons, and nitrogen oxides are primary pollutants contained
in auto emission. Carbon monoxide is considered to be a non-
reactive compared to the other pollutants and is certainly the
most abundant. Hence, its measurement was considered mandatory.
However, at the beginning of the program no satisfactory porta-
ble instrument was available.
Ozone
Satisfactory portable monitors of ozone have been developed.
An AID Model 560 monitor was used satisfactorily in a previous
study and for that reason it was selected for this study. Two
monitors were obtained. The total weight of the monitor and
an Esterline-Angus Model T171B recorder mounted on a backpack
was less than 35 pounds (Fig. 8). The monitor had ample sensi-
tivity and fast response. The fundamental problem, which was
common to all the portable monitors, was sensitivity to temper-
ature changes. The monitor contains an uncooled photomultiplier
tube and the electronics and gas flow rates are temperature
dependent. A first order correction for temperature change was
made by calibrating before and after each daily data collection
period.
A long path laser monitor for ozone was under development
at the beginning of the study and the eventual use of the moni-
tor for area averaged measurements was a possibility. The mo-
nitor was tested in St. Louis during the first year of the study,
15
-------
Figure 8. Portable Ozone Monitor on Back-pack.
16
-------
Carbon Monoxide
The method selected for measurement was to collect air
samples in 90 liter Tedlar bags attached to backpacks. The
bags were returned to the RAMS laboratory for chromatographic
analysis. The bags selected were 5 mils thick, which were
satisfactory compared to 2 mil bags used by other investigators.
However, only a limited amount of data could be collected using
this technique. If the bags had to be stored for more than
four hours before analysis the data was subject to question
either due to diffusion into or out of the bag. Fortunately,
as a result of an EPA contract to Ford Aeronutronic, a gas
filter correlation monitor11 became available during the second
year of the study. This monitor was used in conjunction with
the bag samples to obtain data.
A long range possibility for measuring CO was a long path
laser monitor being developed by MIT Lincoln Laboratory for
EPA, It was not considered that this development would be
ready for use during the first year of the study, but it would
be available for on-site testing in St. Louis and would perhaps
be available during the second or third year to make area ave-
raged measurements. A limited amount of long path data collec-
ted during the second year was incorporated into the RAPS data
bank.
Sulphur Dioxide
The monitor selected to measure sulfur dioxide was a Meloy
Model 165A. The unit weighs 50 pounds and operates satisfac-
torily in the laboratory. However, as a portable unit there
are several problems: temperature sensitivity, limited battery
life, limited hydrogen supply, and long warmup time. The
combination of limited battery life and hydrogen supply coupled
with the long warmup time made the monitor difficult to use,
but the temperature sensitivity made it impossible on hot days.
Nitric Oxide
Prior to the beginning of the second year a portable che-
miluminescent monitor for NO, NO became available. The moni-
tor, a McMillan Model 2200, was Initially tested in the labor-
atory and found to be operable, but it never performed satis-
factorily in the.field due to multiple failures of the acid
storage battery selected to meet the power requirements.
Non-methane Hydrocarbons
As a result of the leakage problem experienced when collec-
ting CO in bags and transporting them to a laboratory for ana-
lysis, no consideration was given to measuring non-methane hy-
17
-------
drocarbons.
In summary, the monitors selected for the study were:
Ozone: AID Model 560
Carbon Monoxide: First year bag sample and labora-
tory analysis
Second and third year, Ford Aero-
nut ronic GFC
Sulfur Dioxide: Meloy Model 165A.
RECORDER SELECTION
An Esterline-Angus Model T171B portable chart recorder
was used in a 1973 long path evaluation test and found to be
satisfactory. A second identical unit was purchased prior
to the 1974 study for use with the second ozone monitor. Two
Hewlett-Packard Model 680 strip chart recorders were obtained.
One was attached to the RAMS ozone monitors and one was at-
tached to the CO monitor.
Prior to the final selection of the portable strip chart
recorder for data collection, some thought was given to the
use of a magnetic tape recorder. This would have several ob-
vious advantages: elimination of jammed paper drives, torn
paper, and the irregularity in the flow of the ink. The dis-
advantage at that time was that a separate, bulky instrument
was required for play back and no convenient space was avai-
lable. Hence, the decision was made to use the strip chart
recorders.
CALIBRATION STANDARDS
The calibration procedures and the selection of standards
are probably the most important aspects of any air pollution
study. Specifically, the calibrations should be traceable
to NBS and the procedures should follow FPA reference methods
or their equivalent as outlined in the Federal Register Vol.
36, No. 228, Nov. 25, 1971. For most of the pollutants, it
was necessary to provide both a primary laboratory standard
as well as a secondary field standard.
Ozone
The primary ozone calibration selected for the first
year's study was the KI technique which is the current EPA
reference method. An ultraviolet ozone generator, manufac-
tured by South States Industries, was used as the secondary
standard. The generator was originally calibrated at EPA-
NERC-RTP in June 1974 and re-calibrated upon arriving at St.
Louis in July 1974. These two calibrations agreed to within
18
-------
1%. The generator output was calibrated in St. Louis on two
later occasions.. These last two calibrations were found to
be significantly lower than the first two calibrations. How-
ever, if a decrease in ozone generator output of 1% for each
100 hours of operation was assumed, the agreement between all
the calibrations was within 3%.
12During the second year, the ozone generator was calibra-
ted by a dynamic gas phase titration of O3 with NO. The
NO used in this titration was a standard reference material
(SRM) obtained from the NBS. This method is listed as an
EPA equivalent to the KI technique. The advantage for field
operations was that no extra equipment was required and the
results seemed to be more repeatable and were directly trace-
able to NBS.
Carbon Monoxide
The primary standard selected for the first year was a
set of three span gas concentrations of CO in nitrogen ob-
tained from Scott Research Company. The gas concentrations
were checked by infrared absorption spectroscopy at EPA-RTP
prior to the summer study and again in St. Louis by gas chro-
ma tography.
The procedure was unsatisfactory due to discrepancies
between the various analyses. The conclusion was that the cy-
linder material was reacting with the CO to change the con-
centrations. This was based on the fact that all the concen-
trations were decreasing and the smaller the concentration the
higher the percentage change. During the following years,
NBS made available an'SRM, CO in nitrogen, which was used.
The SRM CO in N2 concentration (94 ppm-v) was dynamically
diluted with Scott Ultrapure air to obtain a seried of cali-
bration points. The flow rates were measured with calibrated
bubble meters.
Sulfur Dioxide
The primary, as well as the secondary, method selected
for SO- calibration was to use calibrated permeation tubes in
a dynamic calibration system. The tubes used during the first
year were calibrated on Cahn/Ventron Recording Electro Balance
System Model R-100 at EPA-NERC-RTP. These were installed in
a Bendix Dynamic Calibration System Model 551446.
At the beginning of the second year SRM permeation tubes
became available for use as primary sources and commercial
tubes were obtained as secondary standards.
Nitric Oxide
19
-------
The SRM, NO in nitrogen, obtained for the ozone calibra-
tion was also used for the nitric oxide calibration. The SRM
was dynamically diluted with Scott Ultrapure grade air to ob-
tain the calibrating mixture. The calibrating procedure was
to measure the flow with a bubble meter and rotometer in ser-
ies, then to remove the bubble meter and adjust the rotometer
to the original reading.
Nitrogen Dioxide
The calibrating method selected for N0_ was an SRM per-
meation tube furnished by NBS.
COOPERATIVE ARRANGEMENTS
RAPS Staff
This study was one of many carried out in connection with
the overall RAPS program. During the five year period cover-
ing the entire study, a permanent staff was assigned to St.
Louis to oversee the collection of data from the 25 RAMS sites
as well as to assist the field investigators. Prior arrange-
ments were made with the staff for the following: (1) assis-
tance in selecting sites, (2) obtaining permission to locate
laser reflectors on private property, (3) granting..permission
to use space inside the RAMS to locate monitors and recorders,
(4) arranging with the prime contractor Rockwell International
to increase the available power, phone service, and parking
facilities at the laser monitoring sites, (5) providing a
laboratory service of making gas chromatographic measurements
of collected air samples, and (6) providing requested output
data from the RAMS.
RAPS Helicopter
The RAPS Helicopter support consisted of two helicopters
outfitted with a complete set of air pol3ution monitoring in-
struments. These helicopters were normally based in Las Vegas,
Nevada and were sent to St. Louis for the summer intensive
studies of 1974-1976. The purpose of the helicopter measure-
ments was to add a third dimension to the data being collected
by the 25 ground-base stations. Hence, the usual routine was
to collect data on a regular schedule over many of the ground
stations. However, a limited amount of time was made availa^
ble for cooperative measurements with individual investigators.
Prior to each year's study, we made a request for simultane-
ous data collection over the sites we were investigating. Two
or three data collection periods were arranged during each
summer period.
MIT Lincoln Laboratory
20
-------
The MIT Lincoln Laboratory was responsible for the deve-
lopment of diode laser CO monitor which was originally tested
during the RAPS program. We undertook to evaluate the perfor-
mance of this system with the expectation that the data could
eventually be used to make area averaged measurements around
the monitoring sites. Numerous sets of simultaneous measure-
ments were made using the diode laser the GFC monitor located
in the laser van, and bag samples collected in the area. The
van was a convenient field location for the GFC monitor through-
out the study period.
General Electric Company
The GE Company under contract to EPA had built a CO- laser
to measure ozone. We undertook to evaluate the performance of
this laser system during the 1974 summer study. We selected
the site for the evaluation study and also used that as one
of the pollutant variability sites. Arrangements were made to
carry out a series of comparison measurements on many differ-
ent occasions during the summer period.
Aerospace Corporation
Under contract to EPA, the Aerospace Corporation developed
a pulse fluorescence NO2 monitor which was tested during the
RAPS program. We arranged for the laboratory space required
to carry out their evaluations and provided air sampling bags
and a backpack sampler for the collection of air samples.
Ford Aeronutronic
A contract was awarded to Ford Aeronutronic to design and
build a GFC CO monitor. This monitor was brought to St. Louis
for testing at the end of the 1974 period. Data was collected
for about one week and comparisons were made with the Lincoln
Laboratory monitor as well as with bag samples measured by gas
chromatography at the RAPS center.
SITE SELECTION
In order to implement the study it was necessary to se-
lect the monitoring sites, determine the monitoring paths,
and field calibrate the monitors.
The choice of sites was limited to the Regional Air Moni-
toring System (RAMS) sites in the St. Louis area. The area
map (Fig. 1) indicates the locations of 21 of the 25 RAMS
sites. The four remote stations are beyond the borders of
the map. The stations are located on three concentric rings
centered on the downtown area plus one station near the center
and four remote stations 40 km away in the general direction
21
-------
of the cardinal compass points. The six stations on the inner
ring, with the exception of 103 in the stockyard area, are
inner city locations. The six stations on the center ring, are
near-suburban locations while the eight stations on the outer
ring are far-suburban locations.
As soon as the site locations were finalized, a survey
was made to select three sites suitable for testing the long
path laser systems being developed by EPA and also suitable
for the pollutant variability study. The basic requirement
for the long path testing was that an unobstructed one kilo-
meter path at ground level be available, and for the pollutant
variability study a variety of sites was required. In addi-
tion to the three sites selected to meet the long path laser
system requirements, an additional three sites were selected
for the pollutant variability study. It was expected that the
inner city stations would have the greatest variability. Two
were found, 103 and 105, from which one kilometer line of
sight paths were possible. The third long path site was lo-
cated on the second ring. This site, 108, was ideal for tes-
ting purposes because a levee ran across one edge of the site.
A retrorefleetor could be placed on the levee at varying dis-
tances from the station. Another advantage of the sites was
the proximity to industrial plumes which cotild produce a large
variability. The three additional sites selected for the pol-
lutant variability study consisted of 102 and 106, in the in-
ner ring and 113 in the center ring. The remaining inner city
sites were eliminated for security reasons.
PERSONNEL AND.FACILITIES
The personnel required to calibrate and move the monitors
in 1974 were two college students supplied under a task order
to Rockwell. The facilities needed to set up and store the
equipment were furnished by St. Louis University, Department
of Chemistry. St. Louis University provided about 250 square
feet of laboratory space and the spectrometer used for the
KI determination of ozone.
During the last two years, three college students, em-
ployed by the University of Michigan, performed the tasks of
calibrating and moving the monitors as well as carrying out
some of the data reduction. St. Louis University again pro-
vided the required laboratory space.
22
-------
SECTION 5
METHODOLOGY
MONITORING PATHS
Many details of the methodology employed were developed
in the field as the project developed. The original concept
for conducting the study was to hand carry portable monitors
and recorders over previously selected, orthogonal paths ap-
proximately one kilometer long, leading away from the fixed
monitoring site.
gAS/C GKIO-Z.8KM
Figure 9. Sub-grid Data Collection by Back-packing Portable
Monitors.
The data collected by the portable monitors were averaged
and the average value was compared with the average measure-
ment reported by the fixed station monitor during the same
time interval. This was the procedure which we expected to
follow throughout the study. However, the procedure was al-
tered during the study due partly to the sites selected.
Geodetic survey maps were obtained for each station and
paths were initially laid out on the maps (Figs. 2-7). The
selected paths were inspected to determine if the monitors
23
-------
could be safely carried along the paths. Occasionally there
were obstructions or blocked streets not noted on the maps.
One path was eliminated because the neighborhood was badly
deteriorated. Before attempting to take data, trial runs were
made to establish satisfactory walking speeds. The two people
carrying the monitors used watches and noted their times at
major street crossings and at the turnr-around point.
Prior to the collection of data at any of the selected
sites, preliminary testing was carried out in the vicinity
of St. Louis University.
During the previous evaluation of a long path system ,
the monitor and recorder were hand carried. However, it was
determined that it would be more satisfactory to strap the
monitor and recorder to a backpack (Fig. 8). It was also
determined from preliminary tests that more consistent results
could be obtained if sample inlets were placed at a height of
10 feet, which corresponds closely to the position of the gas
inlet on the RAMS stations.
As soon as a reasonable configuration of the instrument
package was complete, data were recorded while both instru-
ments were moved along the same path (Fig. ID). The average
difference between the monitors was 5%, and the standard de-
viation of the individual instrument signal from the average
value was 30%. A similar test was performed near site 106
after the completion of a data set (see Fig. 11). In this case
the difference between monitors was 9%, and the standard de-
viation was 15%. These differences were probably due to the
drifting of the zero and the sensitivity of the monitors.
Another part of the preliminary testing involved deter-
mining the length of the recordings, synchronizing the re-
cordings and walking speeds, and selecting the paths to be
traveled. The area around St. Louis University, shown in
Fig. 12, was treated as if it were a RAMS site to be investi-
gated. The paths traveled are noted by numbers on Fig. 12.
After taking data for a few days, it became obvious that by ^
walking along the streets the data was biased by the traffic
Hence, in order to obtain a better area average, some of the
measurements had to be taken at points away from the streets.
An attempt was made to assess the average land use of the
area by the use of maps, observations from the top of a 15-
story building, and driving around the area. Finally, an
experiment was performed to determine the changes in the
readings as the monitors were moved away from the street.
Subject to the prevalent wind conditions, the measurements
which were made at points adjacent to the street were as low
as 30% of the ambient level. By moving the monitors 100 feet
from the street, the readings were at least 90% of the ambient
24
-------
DZDNE:
3 UUL- 7H
RBPS &HTR
N)
/ooo
Figure 10. Side by Side Comparison Around St. Louis University.
-------
GI
PZDNE:
S3 t-lUL. VH
Instrument #1: Digitized at 10 points per minute
Mean = 81.41 Std. Dev. = 13.70
Digitized at 5 points per minute
Mean = 81.58 Std. Dev. = 12.22
Diff. Between 5 point £ 10 Point
Mean = -.2% Std. Dev. = 10.8%
Instrument #2:
Digitized at 10,.points per minute
Mean = 89.43 "Std. Dev. = 13.67
Digitized at 5 points per minute
Mean = 89.52 Std. Dev. = 13.04
Diff. Between 5 Point & 10 Point
Mean = -.1% Std. Dev. = 4.6%
K)
en
MErexs
BOO
/ooo
Figure 11. Side by Side Comparison Near Site 106.
-------
NJ
Figure 12. Area plot Around St. Louis University.
-------
level. At points 200 feet from the street, the street effects
were not noticed. The same experiment was repeated each year
and in no case were the effects from roadway observed at dis-
tances greater than 100 meters.
As a result of these tests an alternate data collection
technique was developed.
FIXED
ISA MS
Figure 13. Sub-grid Data Collection by Fixed Location.
After surveying the area, four fixed points, all located
approximately 1 km from the RAMS station or central reference
point were selected. These points are designated on Fig. 13
and all the site maps by the letters A,B,C and D. The mea-
surement procedure was to take data simultaneously at A, B
and the RAMS station, and then to move the monitors as rapidly
as possible to positions C&D and record a second simultaneous
set of data.
The operating procedure in the case of using a monitoring
path (Fig. 9) was to have the two monitor backpackers start
at RAMS station, walk in opposite directions to the edge of
the grid, turn around, and return. With a little practice it
was possible for two people to walk at approximately the same
rate. A second traverse would then be taken in a direction
orthogonal to the original traverse. The total averaged data
was then compared with the data collected at the station during
the same interval.
28
-------
At each site at least one set of data was collected while
walking the monitor or bag sampler along selected paths. How-
ever, the bulk of the data was collected at the sub-grid mo-
nitoring locations. Data was recorded for fifteen to twenty
minutes at each location. One or two complete data sets were
taken during the morning, and one complete set was taken dur-
ing the afternoon. Occasionally, rain or other unforeseen
factors prevented a data set from being taken.
HELICOPTER SUPPLEMENTAL DATA TECHNIQUES
A limited amount of helicopter data was collected every
year. The measurements were made at 103 in 1974 and at 108
in 1975 and 1976. The collection procedures were the same in
both cases with two exceptions. During 1974 only ozone was
measured and the helicopter performed a mission for another
project while the monitors were being moved. This resulted
in an excessive waiting time for the portable monitors.
The collection procedure was to locate the portable ozone
monitors and the CO bag collectors at positions A and B while
the helicopter made three passes overhead at heights of 60,
215, and 460 meters. The monitors were then moved to C and
D and three more passes were made. The purpose in taking the
data was to investigate the ability of RAMS to represent pol-
lution in a three dimensional area (see Table 1).
On August 12, 1975 a special effort was made to collect
data from two sites. Morning and afternoon sets were collec-
ted at urban site 105 and a midday set was collected at site
108 in conjunction with a helicopter run. The three data
sets are reasonably representative of the data collected dur-
ing the study and demonstrate the difference in variability
between a rural and an urban site.
FIELD CALIBRATIONS
Field calibrations must be carried out in a minimum of
time and often under adverse conditions. Hence, the tech-
niques selected emphasized convenience rather than accuracy.
Ozone
The calibration instrument was an ultraviolet ozone gen~
eratbr which could be adjusted with a simple sleeve setting.
This required a clean air supply for operation and had to be
left set up in our laboratory space at St. Louis University
(SLU). The daily procedure was to calibrate the portable
monitors at SLU. The monitors were then taken to the RAT-IS
site and simultaneous recordings of the RAMS monitor and the
portable monitors were made at the station inlet. The data
29
-------
I SB
SITE I 0S NW
12 RUG 7S
092S-0SH0 CDT
ST.LDUIS RHPS STUDY
I SB
Cfl
Q_
LJ
z
n
M
a
SITE IBS NE
12 RUB 7S
03SB-I00S CDT
ST.LDUIS RHPS STUDY
S E 7 S 9 IB II
TIME C M I MUTES )
12 13 IH
0 I 2 3 M
6 7
TIME
8 3 IB II 12
< MINUTES)
13 IH IS
SITE 105 SW
12 HUE 7S
0SS0-I00S COT
ST. LOU IS RFIPS STUDY
I SB
I2S
SITE: IBS SET
12 RUG 7S
0a2S-03H0 COT
ST. LOUIS RHPS STUDY
E 7 B 3 IB || 12 13 IH IS
T I ME < M INUTE5 )
S E 7 8 9 IB II 12 13 IH IS
TIME C MINUTES >
Figure 14. Ozone Sub-grid Data at Site 105 8-12-75 a.m.
-------
ISB T
125-
CO
O.
0.
Ul
a
M
a
75
SITE IBS NH
12 HUB 75
I2HS-I2M9 COT
ST. LOU I 5 RftfS STUDY
150
125
ta IBB
8:
75
-SB
25 • •
U)
H
5 E 7 B 3 IB II
TIME < MINUTES)
SITE IBB NE
12 'RUE 75
I 307-13 IB CI>T
ST.LOU IB RHPS STUDY
12 13 IH 15
5 E 7 B 9 IB II
TIME CMINUTES)
12 13 IH 15
150
125
ta IBB
a.
a.
75 •
25
SITE IBB 5W
12 HUE 75
1309-1319 COT
ST. LOUIS RHP5 STUDY
I5B
125
D_
a.
ui
1=3
M
a
75 •
25 •
S E 7 B 9 IB II 12 13 IH IS
TIME C MINUTE5 >
SITE IBB 5E
12 RUB 75
1253-1300 CDT
ST.LOU IS RRPS STUDY
5 E 7 B 9 1011
T I ME C MINUTE5 )
12 13 IH 15
Figure 15. Ozone Sub-grid Data at Site 108 8-12-75 noon.
-------
I SB
125
CD IBB
O.
O.
75
25
SITE I0S NW
12 HUE 7S
I4H0-IHS5 CDT
ST LOUIS RRPS STUDY
I SB i
125 •
CQ
Q-
O.
UJ
z
C3
7S
2S •
U)
S B 7 8 3 IB II
T I ME < MINUTE5 )
12 13 m
SITE IBS NE
I 2 RUE 7S
I5I0-IS2S CDT
ST.LDUIS f!RPS STUDY
5 E 7 B 3 IB 1 I
TIME < MINUTES >
12 13 IM 15
150 -r
125
m
a.
n_
7S
D
M
a
BITE IBS SM
12 RUB 7S
ISI0-I52S CI>T
ST.LDUIS RRPS STUOY
I SB i
I2S •
tD
D_
CL
UJ
SITE IBS SE
12 RUE 7S
IHH0-IHS5 CDT
ST.LOU I 5 RHPS STUDY
B I
5 E 7 B 9 10 II 12 13 IH 15
T I ME C M I MUTES )
5 E 7 B 9 IB I I 12 13 14 15
T I ME C M INUTES >
Figure 16. Ozone Sub-grid Data at Site 105 8-12-75 p.m.
-------
set for the day was collected after which another simulta-
neous set was recorded. The portable monitors were then re-
turned to SLU and a final claibration using the ozone genera-
tor was performed. This procedure was required in order to
account for the drift in the portable monitors during the day.
Carbon Monoxide
s.
The procedure developed after the first year was to peri-
odically (every 4 weeks) perform a 5 point calibration of the
monitor by the dynamic dilution of the SRM obtained from NBS.
Then, immediately following the calibration, the monitor was
used to measure the concentrations of newly mixed field cali-
bration gases. The field calibration gas was stored in small
cylinders about 17 inches high, which were easily handled by
one person. The normal daily procedure was to make a twice
daily check of the monitor with one span gas and one zero gas.
Several gases were available for zeroing the monitor? Scott
Ultrapure Air, Linde Prepurified Argon, and Linde High Purity
Helium. All of these produced zero values within the noise
figure of the GFC monitor (20 ppb-v).
DATA COLLECTION PROCEDURES
The data collection procedure followed throughout the
study was to collect at least 10 complete data sets at each
site over a period of one week. The daily procedure was to
perform two zero and span calibrations of the ozone monitors
at SLU and one or two zero and span calibrations of the GFC
monitor located in the Lincoln Laboratory van parked adja-
cent to one of the RAMS stations. While the monitors were
being calibrated, the Tedlar bags used to collect the air sam-
ples were flushed with helium and evacuated.
The equipment taken to the site to be studied consisted
of four backpacks: two each had an ozone monitor and recor-
der and two each had an air sampling pump and Tedlar bag.
Data was collected by both ozone monitors and an air sample
was collected at the RAMS inlet monitor. The monitors and
the air sampling systems were then taken to the sub-grid lo-
cations, or were backpacked along the selected paths as pre-
viously outlined. The monitors were returned to the station
and the six air samples collected were analyzed. This com-
pleted one data set. Usually two, and sometimes three, data
sets were collected at each site per day. At the end of the
day, additional ozone data were collected at the station mani-
fold. The laboratory calibrations of the ozone generator and
the CO monitor were performed once a month.
DATA REDUCTION PROCEDURE
33
-------
The analog data recorded on strip charts was digitized
at 10 points per minute, calibration corrections based on the
difference between the morning and afternoon calibrations were
applied to each data set, and the mean and standard deviation
from the average of the data set was calculated. Further
analysis of the data required that the RAMS data be correlated
with the sub-grid or portable monitor data. This was done by
fitting straight lines to data pairs consisting of the RAMS
data and the sub-grid data taken at corresponding times. A
straight line was fitted to the data set from which were ob-
tained values for the slope, intercepts, and their standard
deviations. The correlation between the RAMS and the sub-grid
measurements was also determined, then the relationship between
the RAMS and the average of the four sub-grid measurements was
established. The standard deviation calculated for each data
set was adjusted by a ratio derived from Student's t distri-
bution based on the number of data points used and an 80 per
cent confidence limit. Thus, each data set was summarized by
a best fit linear equation with 80% confidence limits.
34
-------
SECTION 6
RESULTS
1974 OZONE
During 1974 the pollutant variability study was confined
to ozone measurements although it was originally intended that
S02 and CO measurements should also be made. The sample pump
failed in one of the two S02 monitors. Because of the delays
and problems in making the measurements only a few exploratory
measurements were taken. As previously explained the technique
of collecting bag samples and returning them to the laboratory
for analysis was not satisfactory because of the limited num-
ber of samples and the possible contamination.
1975 OZONE AND CARBON MONOXIDE
During this study a maximum effort was made to increase
the number of pollutants and to take more data at a given site.
This required a compromise on the number of sites. Data was
taken at two sites only, 105 and 108. These were also the
sites selected for testing the long path laser system. A few
ozone data sets were collected at site 108 by walking along
the orthogonal paths, but most were collected at the sub-grid
locations.
The CO data was obtained by collecting a bag sample while
ozone was being measured, and usually within a half hour the
sample was measured with the GFC monitor which was located in
the Lincoln Laboratory van parked adjacent to the RAMS.
Two calibrated S0» monitors were taken into the field for
data collection. However, during £he first day.of the study
the outside temperature reached 98 F and it was immediately
apparent that the monitors would never operate in this envir-
onment. Hence, the SO, measurements were dropped from further
consideration. Enough data were collected in 1974 to indicate
that most of the time the concentration was below the monitor
noise level (lOppb) but whenever a plume appeared it was quite
high, hundreds of ppb.
The ozone and CO data were collected for a period of
two weeks at each station site. The ozone data were essen-
tially in agreement with that collected in 1974, Since there
was no statistical difference between corresponding sets, it
35
-------
was considered that the study objective had been reached. The
CO data indicated that the station was reading low and for this
reason more data was collected,
A further indication that there was a discrepancy in the
CO dafca was the conclusion of a separate quality assurance
study which showed low CO measurements. At the end of the
1975 study, the GFC monitor was placed in several RAMS station
manifolds where the GFC and the station Beckman 6800 gas
chromatograph-jwere compared. The data is contained in a se-
parate report which describes the GFC CO monitor application.
1976 CARBON MONOXIDE
The primary objective of the 1976 study was to repeat
the CO variability measurements at stations 105 and 108 in
order to determine the cause of the low CO values reported
by RAMS. The procedure in making these measurements was al-
tered in that a bag sample was collected on the roof of the
RAMS simultaneously with the collection at the sub-grid lo-
cations. Thus, for each data set there were five bags-one
colleced for 1/2 hour on the station roof, and four 15-minute
collections at each sub-grid location.
This data demonstrated that there was a difference be-
tween the bag sample collected, on the roof and the RAMS mea-
surement. It was suspected from the Quality Assurance Study
that a sampling problem existed in the RAMS CO measurements.
The RAMS CO measurements are made on samples collected during
a 2 second interval every five minutes. Hence, signal fluc-
tuation periods shorter than five minutes are not observed.
This can lead to an appreciable error if the data is not aver-
aged over a sufficiently long period.
In order to determine the possible signal variation, it
was decided to conduct a small roadside study. This was done
at two locations in the St. Louis area. One study was con-
ducted on Grand Avenue near St. Louis University, and all
the others on Page Road near the EPA-RAPS laboratory. The
GFC monitor was set up with the sample cell open and data were
collected at intervals of 25 meters away from the roadway.
The ^§ta set collected has been included in the separate re-
port which describes the GFC monitor applications. An ex-
ample of the roadside data is shown in Fig. 17.
DATA PRESENTATION
The results of the study are summarized by a set of best
fit linear equations fitted to all the data collected at each
test site. This data and further analyses of the data are
discussed in published paper
36
-------
18 -
16
14
i
£ 12
c
o
-H
.U
£ 10
4-*
c
*"•
y
5 s
o
u
Recorder -
Saturation
Grand Ave. one block ti. of US 40, St.Louis, MO.
August 20, 1976
On Sidewalk
Stability Class A
Wind Speed 2±.6 m/s
Wind Direction 153°±3b
Traffic Flow 400
1221
1225
1230
TIME CST
Figure 17. Grand Avenue CO Concentrations August 20, 1976.
-------
Ozone
The results of the ozone data collection are given by
six equations, one from each site studied (see Table 1). The
number of data points for each set is indicated beside each
equation. Each point represents the average of 15-20 minutes
of concurrent data. The data from four of the stations was
collected in 1974 and the data from the other two was collected
during 1974 and 1975. A single example of how well a given
equation fits the data is given in Fig. 18 for RAMS 113, a
suburban site. The average slope for all the data sets is
0.99 and none of the sets are statistically different from
1.0.
TABLE 1. RAMS MONITOR OZONE CONCENTRATION-.
Related to Area Concentration.
Variations represent 80% confidence limits
Station No. of Points
RAMS 102 = 1.14 (+.17) area av. -K+11) 14
RAMS 103 = 0.86(+ .06) area av. +4 (+4) 13
RAMS 105 = 0.88(+ .10) area av. +4 (+6) 52
RAMS 106 = 1.07(+ .08) area av. +2 (+8) 32
RAMS 108 = 0.98(+ .07) area av. +2 (+4) 58
RAMS 113 = 1.00(+ .26) area av. +K+27) 20
AVERAGE =0.99
Carbon Monoxide
The CO data was analyzed by the same technique as the ozone
data, linear regression by the method of least squares, and de-
termining the correlation coefficient. The basic objective was
to determine the relationship between the RAMS station monitor
and the area average. This relationship is expresses by equa-
tions 1 and 2 listed in Table 2. The very poor correlation
38
-------
sea. T
I
CO
w
o
w
M
Q
I SB.
u> a
« s
o
SB.
o
PH
ST. DEV. OF SINGLfi POINT
I-— H
T
ST. DEV. OF SLOPE
1:1-
LINEAR REGRESSION FIT
v= «x
5- 7.004
RAMS STATION 113 OZONE MONITOR IN ppb
Figure 18. Ozone Correlation Plot RAMS Station 113.
-------
TABLE 2. RAMS MONITOR CO CONCENTRATION
RELATED TO AREA CONCENTRATION.
Variations represent 80% confidence limits.
Equation No.
(1) RAMS 105
(2) RAMS 108
(3) BAG 105
(4) BAG 108
(5) RAMS 105
(6) RAMS 108
,07 (±.07) AREA + .5 ( + .12)
,8 AREA - .11
,35 AREA +.66
,94(+.06) AREA + .004(+.03)
,29 (BAG 105) + .28
86 (BAG 108) - .11
Correlation
Coefficient (r)
.10
.82
.75
.94
.20
.88
%D
36
38
1.5
5.0
•35
•35
S%D
34
21
17
11
30
21
No. of Pairs
70/40
27
40
33
10
9
-------
shown in equation 1 was one of the reasons for collecting bag
samples at the station while the area average measurements were
being made.
The relation between the bag measurements and the area
averages are expressed by equations 3 and 4. The correlation is
much better, particularly at site 108. These equations express
the basic site variability during the measurement period.
The variability between the RAMS monitor measurements and
the bag measurement technique are expressed by equations 5 and
6 which relate the monitor measurements to the measurement of
the bag samples collected on the station roof.
It is apparent from an examination of equation 1 that this
technique of analyzing the data is not completely satisfactory.
Hence, it was suggested that the data pairs be examined by de-
termining the_present_difference (%D = Y-X/Y+X/2 • 100) and
calculating %D and S%D. These values are listed with the equa-
tions in Table 2.
It can be seen that the correlation between the RAMS moni-
tor and the area average is much better at site 108, a rural
station, than at site 105 an urban station. However, the per-
cent difference between the fixed monitor and the area average
is nearly the same at both sites. We have attributed this to
the data sampling technique used in connection with the station
monitors.
Helicopter Measurements
The only helicopter data significant to this study were
collected in 1975. These data are summarized in Table 3. This
indicates that the ozone concentration is very uniform over the
entire grid area and that it reaches a maximum at about 60 me-
ters and drops at higher altitudes. The values on the ground
are slightly lower and more variable, as might be expected.
The CO concentraitons show much more variation and some of
the concentrations are much higher than the ground level values.
This might be due to a plume from a factory in Granite City or
it might be temperature drift of the CO analyzer. Because of
the limited amount of data no analysis was attempted.
PROBABLE SOURCES OF ERROR
One of the major sources of error is uncertainty in the ca-
libration of the portable monitors and the station monitors.
Much of the uncertainty in these calibrations was reduced during
the 1975 and 1976 study by using SRM's as calibration standards
for all monitors.
41
-------
There is a measureable drift in the zero and sensitivity
of the portable ozone monitors with temperature. Much of this
error was corrected by measuring the sensitivity of the monitors
before and after the data set each day. The temperature drift
was measured in an environmental test chamber and over the tem-
perature range from O'G to 35 C, one monitor changed 10% and the
other 20%. However, on some days during the 1974 study larger
drifts were noted. This was probably due to absorbed solar
radiation in addition to the high air temperature.
The drift in sensitivity of the CO monitor was largely
corrected by operating the monitor in a temperature stable envi-
ronment, namely, the Lincoln Laboratory van. The CO monitor was
tested in an environmental chamber and a sensitivity drift of
0.3% per degree C was measured.
The RAMS CO monitor samples CO once every five minutes.
Data collected by the Philco-Ford GFC has shown that the con-
centration of CO can fluctuate with a period of less than 5
seconds (Fig. 17). When this happens the RAMS values will not
accurately represent the CO concentration. The data collected
in that study and subsequent studies shows that the typical
RAMS measurement is low.
The typical low average reading is due to the fact that
the average does not cover a sufficiently long sampling period.
The minimum averaging period can be determined from the sampling
theorem which states that two samples are required per period of
the highest frequency to be represented. If the highest fre-
quency of the concentration fluctuation is 5 seconds then a
sample is required every 2.5 seconds. During the 5 minute. (300
second) interval between RAMS measurements 300/2.5 = 120 samples
are required. In order to accumulate 120 samples the data must
be averaged for 120 * 5 minutes = 600 minutes or 10 hours.
The concentration is normally not constant nor is it averaged
over such a long period. The user should be aware that the ty-
pical measurement will be low, but an occasional measurement
will be very high.
42
-------
TABLE 3. INTENSIVE STUDIES WITH HELICOPTER
SUPPORT RAMS SITE 108
Date Time (CST) Position
(03>PPb
S=10ppb
(CO)
S=.15ppm
Measurement
system
8-7-75
.11:45-12:05
NW
SE
RAMS
Helicopter at 60M
Helicopter at 215M
Helicopter at 460M
8-7-75
12:10-12:25 NE
SW
RAMS
Helicopter at 60M
Helicopter at 215M
Helicopter at 460M
8-12-75
11:52-12:05 NW
SE
RAMS
11:52-53 Helicopter at 60M
11:54-56 Helicopter at 215M
11:57-58 Helicopter at 460M
8-12-75
12:08-12:20 NE
SW
RAMS
12:12-13 Helicopter at 60M
12:14-15 Helicopter at 215M
12:16-20 Helicopter at 460M
54
46
59
60
60
60
63
54
60
60
60
60
89
86
89
99
96
93
79
89
87
95
93
92
0.38 I Bag Samples
0.87 (Measured by GFC
0.29 -Beckman 6800
1.0 lAndros measurement
1.5 fin helicopter
2.0 )Error not known
0.47 )Bag Samples
0.64 /Measured by GPC
0.30 -Beckman 6800
0.41 )Bag Samples
1.03 /Measured by GFC
0.3 -Beckman 6800
1.2 Bag Samples
1.06 Measured by
1.15 Beckman 6800
0.64 \Bag Samples
0.42 /Measured by GFC
0.3 -Beckman 6800
2.45 )Bag Samples
0.89 [Measured by
0.91 )Beckman 6800
43
-------
SECTION 7
DISCUSSION
In general, the correlation between area averages and RAMS
values is greater at rural than it is at urban sites. This is
not surprising; rural sites contain considerable fewer sources
of pollution than do urban sites. Consequently, the pollutant
in the rural sites is more evenly distributed than it is in an
urban area.
The correlation of 0.. RAMS data with the area average mea-
surement is greater than the correlation of CO RAMS data with
the area averages. This is probably due to the differences in
the method by which the pollutants are produced.
Ozone occurs naturally and is also a widely dispersed se-
condary pollutant. It seems that a relatively uniform concen-
tration could be spread across an entire air mass. In any event
it is dispersed on. a macroscale. These variations within the
urban area are due to titration with NO which is dispersed on
a microscale. The higher the total ozone concentration, the less
is the effect of local sources on the concentration. Carbon mo-
noxide is a primary pollutant and is not so evenly dispersed,
especially, in urban areas. Under conditions when the back-
ground CO is low, for example, during the passage of a cold
front, most of the measured pollution is from local sources.
If we define local sources to be those no more than 100 meters
away, then there will be little correlation between sub-grid
locations 1 kilometer apart.
During the course of this study, a large number of measure-
ments were made of the ambient concentrations of both ozone and
carbon monoxide. In addition, a significant portion of the
RAPS data base collected during the same time periods was ex>-
amined. The primary observations to be noted is that the vari-
ability in pollutant concentration id difficult to correlate
with any single factor. In fact, the variation is dependent on
many factors making it nearly impossible to develop simple re-
lationships between any single factor such as wind speed, wind
direction, or source strength and the pollutant concentration.
RELATION TO SIMILAR STUDIES
The variability data which was collected during this study
was averaged over relatively short time intervals (10 min. to
30 min.). These times are short compared to any model which
44
-------
will be applied to the RAPS data base. Also the number of sam-
ples were limited and all were collected during the summer day-
light hours. Hence, the measured correlation between the moni-
toring station and the area wide averages cannot be applied as
corrections, but noted as a probable error. However, such data
has proved useful in other similar studies. Examples of such
studies are "An Urban Survey Technique for Measuring the Spa-
tial Variation of Carbon Monoxide Concentrations in Cities" by
Wayne Ott , "The Areal Representiveness of Air Monitoring Sta-
tions—Fresno-jjtudy Phase I" by State of California Air Re-
sources Board , and "Selecting Sites for,Carbon Monoxide Moni-
toring" by F.L. Ludwig and J.H.S. Kealoha .
CO STUDIES
Ott performed a CO survey of the San Jose urban area (13
square mile grid) over a six month period by collecting bag sam-
ples which were analyzed for CO and compared with a. permanent sta-
tionary monitor. The primary question which he sought to an-
swer was, "How representative was the station monitor?" This
was essentially the same question we asked regarding the indi-
vidual stations in the St. Louis RAPS network. Ott's survey
extended over a six-month period from October through March com-
pared to our summer investigation of St. Louis. The San Jose
area was divided into 9 squares whereas we used 4 squares. In
each case the sample collections were made at fixed sites inside
the squares. The basic procedure in both cases was to collect
two or more simultaneous samples including the monitoring sta-
tion at the selected locations over an extended time period.
We each performed an almost identical supplementary exper-
iment in order to try to determine the "street effect" (Ott's
terminology) . Data was collected on a sidewalk near a very bu-
sy street and the sampling point was moved back 25 feet at a
time until the "street effects" were no longer observed. Ott
reported that the "street effects" disappeared after 200 feet
and we reported that they disappeared after 100 meters. We
were able to make the measurements with a portable monitor,
whereas Ott had to have the bag sample analyzed. Hence, our
sensitivity must have been greater which makes the agreement
very impressive. These results are in agreement with an analy-
sis of Los Angeles monitoring data which showed a correlation
between the monitored value and the distance of the monitor
from the street.
Our measurement of the "street effect" was done by placing
the monitor on the sidewalk along a busy street as compared to
Ott's technique of walking along the street. A sample of our
results is shown in Fig. 17. We believe that this confirms
Ott's conclusion that the measured CO consists of two compo-
nents, "urban background," the area between the spikes, and
45
-------
"street effects," the area under the spikes.
Our conclusion is that the "urban background" is CO which
has dispersed until it reached the inversion and dispersed back
to the surface, or else dispersed downwards after an initial
plume rise. The contributions from individual sources are no
longer identifiable. The "street effects" or spikes on our
records are due to intercepting the initial plume as it is ex-
hausted from the individual automobile or a cluster of auto-
mobiles. It appears that there is an initial plume rise or
buoyancy which normally prevents the initial plume from being
detected at ground level more than a few hundred feet from the
source. This effect has been identified as a reason for the
disagreement between some gaussian plume models and field mea-
surements (Chock) . It has also been suggested that near mid-
day, particularly in the summer, a street temperature higher
than the surroundings may contribute to the effect.
A typical inversion height at mid-day during the summer
will be several thousand meters and the urban background may
be very low (less than 0.5 ppm) so that most of the total mo-
nitored CO is due to "street effects" Under these circumstances
two monitors only 500 meters apart may monitor CO from complete-
ly different sources. We believe this explains the complete lack
of correlation of much of the CO data collected at RAM site 105.
There have been numerous microscale studies directed towards
measur^g the distribution of CO such^as, the St. Louis street
canyon^o, the San Jose street canyon , the Q§k Park shopping
center , and the New York submerged roadway . These studies
have a common characteristic in that they tried to establish an
average CO distribution in a limited area over a limited time
period. This is difficult task in that a dynamic situation is
being described by static measurements. Rather than stable gra-
dients, many individual plumes move in a turbulent fashion through
the area with order of magnitude concentration variations. In
all of these cases the total concentrations are high and the
"urban background" component is relatively low which is a dif-
ferent condition than the measurements that we have reported.
OZONE STUDY
The results of our ozone study show that there is no sta-
tistical difference between the RAM point measurements and the
portable monitor area-wide measurements. These results can be
compared with a Fresno study done by the California Air Resour-
ces Board to determine the represent!veness of a single moni-
toring station located near the city center for a surrounding
64 square mile area. Ozone was measured from June 22, 1976 to
Nov. 30. 1976 at 20 locations throughout the area. A pattern
of mean regression slopes for the maximum hour concentrations
46
-------
were drawn. This analysis showed the maximum concentration gra-
dient to be 5% over 1/2 mile. This is less than a statistically
significant difference for our experiment. This agreement regar-
ding the microscale variation of ozone adds weight to our conclu-
sion that RAM ozone measurements should be representative of
the surrounding area.
RELATIONSHIP TO SITE SELECTION
• A result of this study is a recommendation that monitoring
stations be sited 100 meters off a roadway in order to minimize
"streets effects.11 Ott made a similar recommendation and fol-
lowed with a paper suggesting several types of locations depen-
ding on the data desired. Ott's suggestions were used by Ludwig
and Kealoha (L and K) in the development of criteria for the
selection of CO monitoring sites.
It was suggested by L. and K. that the measurement scales
which may span 7 orders of magnitude from less than one meter
to 40,00km (the earth's circumference) be divided into seven
categories each spanning approximately one order of magnitude.
The proposed categories and the approximate scales as well
as the standard meteorological turbulence scales are shown in
Fig. (19).
The discussion by L. and K. of the seven proposed scales
concluded that only the tjhree shown by the solid hatching should
normally be, considered as monitoring sites. .It can be seen that
each of these categories is included in each of the three mete-
orological scales. Hence, for the purpose of this discussion
the meteorological nomenclature will be used.
As can be seen in Fig. (19) there is a close relationship
between the spatial and time variations of either atmospheric
turbulence or pollutant concentration. We have tried to illu-
strate the scale of the spatial variation in Fig. (20) and the
temporal variation in Fig (21).
Referring to Fig. (20) which illustrates the relative size
of the three scales it should be noted that any station located
inside the urban plume will monitor pollutant concentrations on
all three scales. The magnitude of the microscale CO contribu-
tion (0.5 ppm to 50 ppm) will depend on the roonitor location
relative to the local sources (less than 1 km away). The meso-
scale or urban CO contribution (0.1 ppm to 20 ppm) will depend
on the inversion height, wind speed, and wind direction. The
macroscale or synoptic CO contribution (0.05 ppm to 0.5 ppm)
will depend on the origin of the air mass. An artic high may
have traveled 1000's of kilometers over uncontaminated areas
and have contributions below the global CO background level
47
-------
£A*TH 40,000 KM
T£0*01-0&I CAL
SF>ACE SCALE:
Fig. 19 Space and Time Scale Correlation
-------
Figure. 20. Pollutant Area Scale Comparison.
*
o
<^
5
0
V*
,MESO SCALE.
,V\*CSO SCALE
12.
IB
SO
Figure 21. Pollutant Variation vs. Time Comparison.
49
-------
(0.16 ppm). Of course, it is possible to be located on the
backside of a stagnant high and measure fairly high CO con-
centrations (0 . 5 ppm).
An illustration of the time variation in the pollutant
concentration is shown in Fig. (21) . If we assume that the
monitors used will follow the time variations in the concen-
trations and that these variations are faithfully recorded,
then it would be possible to apply filtering to the data in
order to determine each of the three components separately.
The ability to separate the measurements into components
would increase the usefulness of the data. Models designed
to measure either microscale or mesosclae effects could be
verified directly without spurious interferences. Also, the
problem of station siting would be simplified and probably the
number of stations required to obtain satisfactory data would
be reduced.
If other studies such as the one reported here were to be
made, it would be appropriate to compare and examine only the
highest frequency components. On the other hand, if compari-
sons were to be made between stations on a regional basis as
in the RAPS, then the mesoscale or midfrequency variations
should be examined. In the case of ozone transport studies,
the macroscale or low frequency portions of the signal should
be examined.
SUGGESTIONS
RAPS DATA BANK
Most of the monitoring instruments used in the RAPS net-
work were capable of responding to microscale variations.
However, the data sampling, averaging, and readout process may
have obscured the microscale components. A careful examination
of the data should be made to determine if filtering can be
applied. If filtering is possible, a selected set of filtered
data should be tested for model verification.
FIELD TESTING
The most effective means of testing the filtering technique
would be by application to an existing data collection network.
The monitors to which the technique could be readily applied
are the continuous reading fast response instruments such as
the ozone and NO chemiluminescence monitors. If additional
network data collection channels are available, the data
could be passed through filters and collected on separate
channels. Otherwise, the data can be collected in digital form
and mathematical filtering applied to obtain the desired out-
50
-------
put signals.
It is felt that the possible benefits to be gained by
using filtered data requires the testing of the technique.
51
-------
REFERENCES
1. RAM User's Guide -EPA-600/8-78-016 a&b, November 1978.
2. Third Synposium on Atmospheric Turbulence, Diffusion, and
Air Quality - Oct. 19^22, 1976 Raleigh, N.C. - American
Meteorological Society.
.*'
.-''
3. Johnson, W.B. - Validation of Air Quality Simulation Models,
Proceedings' of the Third Meeting of the Expert Panel on
Air Pollution Modeling. Paris, France,Oct. 2-3, 1972.
4. McCormick, R.A. - The U.S. Regional Air Pollution Study,
Proceedings of the Third Meeting of the Expert Panel on
Air Pollution Modeling. Paris, France, Oct. 2-3, 1972.
5. Pooler, F. Jr., JAPCA 24:3, pp. 228-231 (1974).
6. Schiermeier, F.A. - Air Monitoring Milestones. Environmen-
tal Science &:. Technology Vol. 12, No. 6, p. 644 June 1978.
7. Ku, R.T., and E.D. Hinkley, "Long Path Monitoring of Carbon
Monoxide," Lincoln Laboratory, NSF/RANN/IT/GI-37603, Tech.
Report, April 1976.
8. Craig, S.E., et al, "Development of a Gas Laser System to
Measure Trace Gases by Long Path Absorption Techniques,"
Contract No.. 68-02-0757, Report No. EPA-650/2-74-046-A,
June 1974.
9. Burch, D.E., F.J. Gates, and J.D. Pembrook, "Ambient CO
Monitor," Final Report on EPA Contract No. 68-02-2219
issued as EPA Report 600/2-76-210, July 1976.
10. Ku, R.T., E.D. Hinkley, and J.O. Sample, Appl. Opt. 14.:4,
p.854 April 1975.
11. Chaney, L.W. and W.A. McClenny, "Unique Ambient CO Monitor
Based on Gas Filter Correlation - Performance and Applica-
tion, " Environmental Science and Technology, Vol. 11, p.
1186, Dec. 1977.
12. Rhmem, K.A., et al, Environmental Protection Technology
Series, EPA-R2-73-246.
13. McClenny, W.A., et al, JAPCA, Vol. 24, No. 11, p. 1044, Nov.
1975.
52
-------
15. Chaney, L.W., Summary of Summer 1975 Field Exercises,
presented at RAPS Meeting, October 29-30, 1975, EPA-
650/3-75-009.
16. Chaney, L.W., "Carbon Monoxide Ambient Field Measurements,"
EPA-Grant R-803300, July 1978.
17. McClenny, W.A. and L.W. Chaney, JAPCA Vol. 28, No. 7,
p. 693, July, 1978.
18. The Areal Representativeness of Air Monitoring Stations—
Fresno Study Phase I (Oxidant), State of California Tech-
nical Services Division, March 1977.
19. Ludwig, F.L. and J.H.S. Kealoha, Selecting Sites for Carbon
Monoxide Monitoring -EPA-450/3-75-077 Sept. 1975.
20. Chock, D.P., "General Motors Sulfate Dispersion Experiment:
Assessment of EPA HIWAY model," JAPCA, Vol. 27, p. 39.
Jan. 1977.
21. Johnson, W.B., W.F. Dabberdt, F.L. Ludwig, R.J. Allen,
"Field Study for Initial Evaluation of an Urban Diffusion
Model for Carbon Monoxide," Contract CAPA 3-68, (1-69),
Stanford Research Institute, June 1971.
22. Ludwig, F.L. and W.F. Dabberdt, "Evaluation of APRAC-1A
Urban Diffusion Model for Carbon Monoxide, Contract CAPA
3-68 (1-69), Stanford Research Institute, Feb. 1972.
23. Patterson, R.M. and F.A. Record, Monitoring and Analysis
of Carbon Monoxide and Traffic Characteristics at Oakbrook"
EPA-450/3-74-058, Nov. 1974.
24. General Electric Co. "Study of Air Pollution Aspects of
Various Roadway Configurations," PB 211235, New York City
Contract No. 209624.
53
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
E^-6T0N0°/2-79-183
2.
3. RECIPIENT'S ACCESSION-NO.
4. TITLE AND SUBTITLE
MICROSCALE VARIATIONS IN AMBIENT CONCENTRATIONS
OF POLLUTANTS IN ST. LOUIS AIR
5. REPORT DATE
September 1979
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
Lucian W. Chaney
8. PERFORMING ORGANIZATION REPORT NO,
9. PERFORMING ORGANIZATION NAME AND ADDRESS
The University of Michigan
Ann Arbor, Michigan 48109
10. PROGRAM ELEMENT NO.
1AA603 AA-12 (FY-77)
11. CONTRACT/GRANT NO.
R-803399
12. SPONSORING AGENCY NAME AND ADDRESS
Environmental Sciences Research Laboratory - RTF, NC
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park. N.C. 27711
13. TYPE OF REPORT AND PERIOD COVERED
1Q7A-77
14. SPONSORING AGENCY CODE
EPA/600709
15. SUPPLEMENTARY NOTES
16. ABSTRACT
As part of the Regional Air Pollution Study (RAPS), a series of
studies were carried out in St. Louis during the summers of 1974, 1975,
and 1976 primarily to determine the sub-grid concentrations of ambient
air pollution. One primary pollutant gas, CO, and one secondary pol-
lutant gas, ozone, were chosen to be representative. Methodology for
determining sub-grid concentration variations of these gases are dis-
cussed.
Portable monitors and the collection and analysis of bag samples
were used to determine pollutant concentrations. In some cases the
monitors were moved along selected paths while the measurements were
made; in other cases the monitors were placed at selected sub-grid
locations. The data were collected at six sites during the first year,
and at two sites during the final two years. Both urban and rural
sites were selected. All the data were collected during daylight hours
generally between 10:00 a.m. and 4:00 p.m.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
c. COS AT I Field/Group
*Air pollution
*Carbon monoxide
*0zone
*Measurement
*Concentration (composition)
Variations
St. Louis, MO
13B
07B
07D
8. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19. SECURITY CLASS (ThisReport)
UNCLASSIFIED
21. NO. OF PAGES
64
20. SECURITY CLASS (Thispage]
UNCLASSIFIED
22. PRICE
EPA Form 2220-1 (9-73)
54
------- |