United States
Environmentai Protection
Agency
Environmental Sciences Research
Laboratory
Research Triangle Park NC 27711
EPA-600 2-79-139
August 1979
    ^^^rf4
    •no
Ambient  Air
Carbon  Monoxide
Measurements

-------
                 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-600/2-79-139
                                               August 1979
 AMBIENT AIR CARBON MONOXIDE MEASUREMENTS
                    by

             Lucian W. Chancy
        The University of Michigan
         Ann Arbor, Michigan 42109
            Grant No. R-803399
              Project Officer

            William A. McClenny
Atmospheric Chemistry and Physics Division
Environmental Sciences Research Laboratory
    Research Triangle Park, N.C. 17711
ENVIRONMENTAL SCIENCES RESEARCH LABORATORY
    OFFICE OF RESEARCH AND DEVELOPMENT
   U.S. ENVIRONMENTAL PROTECTION AGNECY
    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 publi-
cation.  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.
                                    11

-------
                               ABSTRACT

      This report discusses the application of a new type CO monitor to
special ambient field measurement problems.  The monitor, a gas filter
correlation (GFC) instrument was  designed specifically for use in the St.
Louis Regional Air Pollution Study (RAPS), but has been applied to several
other measurement requirements.   The monicor has an inherently fast
response of less than one second and has proved useful in documenting
extremely variable monitoring situations.

      The monitor was used in a total of ten separate studies and typical data
is presented from all tests  with the exception of the last which is still being
reduced.

      The most significant contribution from these studies  has been the
ability to document the extreme variability of carbon monoxide in our urban
environment.

      This report was submitted in fulfillment of Grant R-803399 by the
University of Michigan under the sponsorship of the U. S. Environmental
Protection Agency.  This report covers a period from November 1, 1974 to
October 31,  1977 and the work was completed as of October 31,  1977.
                                   iii

-------
                         CONTENTS
Abstract	iii
Figures	vi
Tables	vii
Abbreviations and Symbols	vili
Acknowledgment	-	ix

   1.  Introduction	   1
   2.  Conclusions	  2
   3.  Recommendations.	   3
   4.  Measurement Summary	4
   5.  Discussion	9

References	13
Appendices

   A.  Unique Ambient Carbon Monoxide Monitor Based
       on Gas Filter Correlation:  Performance and
       Application,	15
   B.  Carbon Monoxide Automobile Emissions Measured
       from the Interior of a  Traveling Automobile	21

-------
                            FIGURES

Number                                                         Page

    1.     Comparison between GFC monitor and Lincoln
           Laboratory's  Long Path Monitor .............  •  10

    2.     Comparison between GFC monitor and SAI Long
           Path Monitor  ........................  1 1
    3.     GFC San Diego Freeway "On" ramp measurements- ...  12
                               VI

-------
                                TABLES

Number                                                            Page

   1     Gas Filter Correlation (GFC) Carbon Monoxide
         Monitor Field Studies	   5
                                   vn

-------
                    ABBREVIATIONS AND SYMBOLS
ABBREVIATIONS
CO
DOT
EPA
GFC
MIT
ppm-v
RAMS
RAPS
SAI
             ---  carbon monoxide
             ---  Department of Transportation
             ---  Environmental Protection Agency
             ---  gas filter correlation
             ---  Massachusetts Institute of Technology
             ---  parts per million by volume
             ---  Regional Air Monitoring Station
             ---  Regional Air Polution Study
             ---  Science Applications Incorporated
             ---  sulfur hexafluoride
                                   Vlll

-------
                       A CKNOWLEDGEMENTS

    The studies reported here required the cooperation of numerous groups
in many locations.  The author is grateful to the  permanent Regional Air
Pollution Study staff which responded to the many requests for data and,
particularly, to Stan Kopczynski who arranged the instrument calibration
tests.   The author appreciated the cooperation of Dr.  D. Burch of Ford
Aeroneutronic who was responsible for the monitor design and Dr. F. F.
Marmo of the Department of Transportation who  arranged several of the
roadway tests.

    Thanks also go to the student staff for their whole hearted cooperation
in collecting most of the data:  Michael Travis, Jeff Kochelek, Carol
Chaney, and David McKinley.

    The author's special thanks are extended to Dr. W. A. McClenny for
his  many suggestions and review of this report prior to final preparation.

-------
                             SECTION 1

                           INTRODUCTION
      The monitor used in the studies to be discussed was designed and built
for the U.S. Environmental Protection Agency (EPA) by Ford Aeronutronic
and described by Burch. *  The expressed reason for obtaining the monitor
was for use in the St. Louis Regional Air Pollution Study (RAPS)2  to evaluate
a Long Path Laser CO monitor developed by MIT Lincoln Laboratory. ^ It
was also intended for use in a sub-grid characterization study.   However,
as a result of its proved usefulness in conducting the intended studies,  it
was also used in several ancillary studies.   The purpose of this report is
to discuss and document carbon monoxide measurements made with the GFC
monitor as a part of the EPA grant studies  conducted by the University of
Michigan.

-------
                             SECTION 2

                           CONCLUSIONS

    The GFC monitor proved to be extremely useful for micro scale
studies which involve rapid fluctuations in the ambient concentration of
carbon monoxide,  The monitor's rapid response of less than one second
permitted the documentation of the very rapid change in concentration as
well as the magnitude, which can exceed three orders of magnitude in a
few seconds.

    The original hope was that the monitor would be sufficiently portable
(gross weight about 35 pounds) so  that it could be back-packed to permit
path averaged measurementf  Although this goal was not realized, the
fact that it could be easily transported from one  location to another, set up
to operate in an automobile, or an airplane, made very interesting experi-
ments possible. The data obtained has added to  our knowledge of the nature
of the microscale variations in the concentration of carbon monoxide.

-------
                                 SECTION 3

                           RECOMMENDA TIONS

       The GFC monitor should be considered for any field program re-
quiring the measurement of the microscale variations in the concentration of
carbon monoxide.

       The technical problems involving the monitor which should be
examined are as follows:

       1) The precise reason for the zero drift with temperature
           should be  determined and corrections made in order to
           extend the temperature range for uncorrected operation.

       2) The battery supply should be re-designed so that the
          drain on all the batteries  is equal or approximately so
          during the  operation. It would be very desirable to
          extend the  battery operation to one complete working
          day.

-------
                                SECTION 4

                        MEASUREMENT SUMMARY

       Many of the measurements made with the Ford Aeronutronic Gas
Filter Correlation (GFC) carbon monoxide monitor for the Environmental
Protection Agency (EPA) have been previously reported.'*"'  A chronological
list of the individual studies and measurements is given in Table 1.

       The monitor itself has been adequately described by Burclr1 and is
also briefly described in Appendix A.  The  individual studies are briefly
described in this section following the order listed in Table 1.

RAPS POLLUTANT VARIABILITY STUDY -- July and August 1975 and
       July and August 1976 St. Louis, Mo.

       This study was performed over a  period of three years 1974-1976 in
St. Louis and has been described in a previous report.   The  GFC monitor
was the primary CO measurement tool during the final two years of the study.
Although the original plans  were to use the  monitor as a portable instrument
and the final weight made this impossible, it was  small enough to be easily
transportable which proved to be advantageous.

       The monitor was operated during  the St. Louis study inside an air
conditioned mobile van which was usually parked adjacent to a Regional Air
Monitoring Station (RAMS).

       Air samples were collected in the area at distances up to one kilo-
meter away from the station and brought  to the monitor for analysis.  The
data collected was used in the pollutant variability study.  In addition,  ambi-
ent measurements were  made for comparison with the Lincoln Laboratory
CO laser system housed in  the same mobile van.  3, 5  Spatially integrated
bag samples were collected by walking along the laser path for comparison
with the laser measurements.   An example of an ambient measurement
comparison between the  CO laser and the GFC is  shown in Figure 1.  Further
details of the laser comparison were reported by Lincoln Laboratory -- MIT.

-------
                          TABLE 1

         Gas Filter Correlation (GFC) Carbon Monoxide
                     Monitor Field Studies

 1.  Regional Air Pollution Study (RAPS) Pollutant Variability
    Study, July-August 1975 and July-August 1976 St. Louis, Mo.

 2.  RAPS Quality Assurance Study, July-August 1975  St. Louis, Mo.

 3.  RAPS Department of Transportation (DOT) Roadway Study,
    July 1975  St. Louis, Mo.

 4.  General Motors DOT Roadway Study, October 1975  Milford,
    Mich.

 5.  RAPS Roadway Study, August 1976  St.  Louis, Mo.

 6.  RAPS Helicopter Installation,  July 1976 St.  Louis,  Mo.

 7.  Moving Vehicle  Emission Study, March 1977.

 8.  DOT, Silver Strand Roadway Study,  April 1977  San Diego,
    Calif.

 9.  Los Angeles Freeway "On" Ramp Measurements,  May 1977
    Los Angeles, Calif.

10.  Chicago Plume Study, July 1977 Muskegon, Mich.

-------
RAPS QUALITY ASSURANCE STUDY -- July and August 1975  St.  Louis,  Mo.

       The quality assurance study was conducted for a period of one week
prior to the RAPS 1975 summer intensive period and for one week following
the summer period.  The main purpose was to independently measure the
pollutant concentrations at the RAMS measurement locations and compare the
results.  The CO measurements were intended to be taken by collecting bag
samples at the RAMS stations and then measure the CO concentrations in the
bags with the  GFC.  This led to the surprising result that many of the stations
were reading  low, but in an inconsistent manner.  As a result, the GFC moni-
tor was compared directly with a gas chromatograph monitor in the RAPS
Laboratory.  The two instruments  reported identical results over a period of
24 hours.  The monitor was  subsequently placed inside a RAMS station and
attached to the air  sampling  manifold.  Several stations were sampled for
periods of at least  24 hours.  Two  samples of the data  collected are shown in
Appendix A Figures 4-A and 5-A.  The final conclusion was that the difference
in measurement was the result of the method of sampling the data rather than
the method of measurement.  The RAMS sampling procedure was to collect a
sample for a few seconds for analysis every five minutes.  Hence, variations
in concentration with periods faster than 10 minutes could not be accurately
reproduced.  This  fact has been pointed out'in previous reports.1*  The chief
advantage  of the GFC monitor in this study •was its portability and ease of
installation.

RAPS-DOT ROADWAY STUDY — July 1975 St. Louis,  Mo.
GENERAL MOTORS-DOT ROADWAY STUDY -- October 1975, Milford, Mich.

        These two studies are being reported together since they were a part
of the same cooperative effort with the Department of Transportation.  The
study in St. Louis was  essentially a trial run for the measurements carried
out later in the year at General Motors Proving Grounds under controlled
conditions. The basic  carbon monoxide measurements were made with a long
path GFC   monitor developed especially for this purpose.  During the St.
Louis study, sets of simultaneous measurements both upwind and downwind of
the roadway were made by collecting bag samples.  During several of the
tests, five simultaneous samples were collected and immediately taken to St.
Louis University about one and one-half miles  away for analysis on the GFC
monitor.  Several such sets  were collected and analyzed during one rush hour
time period.

       Based on the experience gathered during the St.  Louis  study, it was
decided that a real  time measurement of the background concentration would
be  the preferred measurement for the General Motors Study.  The monitor,
recorder,  and calibrating gases were  mounted in the rear seat of a passenger

                                     6

-------
car which was parked about 50 feet upwind of the roadway.   The sample was
collected through a tube extending about 5 feet above the roof of the automobile.
The roadway monitoring instrument used in the experiment was developed by
Science Applications Inc. (SAI) and described by Bartle. ^  The instrument
is an open path GFC capable of measurements up to 40 meters.  A description
of the experiment is the subject of a SAI report   and additional analysis of
the data was done by Dr.  Fred Marmo,    the  Department of Transportation
Project officer.

RAPS ROADWAY STUDY -- August 1976  St.  Louis,  Mo.

       The  study was conducted at two sites in the St.  Louis area for a period
of two weeks.  The main purpose in conducting the study was to measure the
magnitude of the real time variation in the CO concentrations near a roadway
and at specified distances away from the  roadway.  An example of the extreme
magnitude of the variation is contained in the data shown in Appendix A
Figure A-2a. Data taken at a distance of 100 meters from the roadway show-
ed virtually no rapid variation in concentration.  The data collected during
the study was also the subject of a student project. "

RAPS HELICOPTER INSTALLATION -- July 1976  St.  Louis,  Mo.

       The  RAPS helicopters were normally equipped with Andros CO moni-
tors.  These monitors were adversely affected by the aircraft environment.
Hence, as a result of the success  achieved in using the GFC, arrangements
were made to fly it on one of the RAPS helicopters.  Two flights were carried
out and some of the data collected on one trip is shown in Appendix A,
Figure A-6.  The instrument performed very well in this application.   The
data shown is for two passes through a power plant plume which is very
clearly defined as a result of the rapid  response.  Levels as low as  0. Ippm
were measured on both flights.

MOVING VEHICLE EMISSIONS STUDY -- March 1977

       This  study was designed to measure the CO concentration along inter-
state highways.** In the process of making the measurements,  it was  deter-
mined that the effect of emissions from individual vehicles could be measured.
A paper describing the  study is attached as Appendix B.

DOT-SILVER STRAND  ROADWAY STUDY -- April 1977 San Diego,  Calif.

       The  study, organized by DOT,  had as  its goal the measurement of the
roadway CO emission rates as a function of traffic density.  The basic plan
was to release measured amounts of SF& on the upwind edge of the roadway as
a tracer gas and measure both SF^ and CO on  the downwind edge of  the road-
way.  The background concentration of CO on the upwind side was subtracted
from downwind CO  concentration and ratioed with the SF^ measured downwind
to determine the effective CO emission from the roadway.  The wind speed,
wind direction,  temperature,  relative humidity,  and traffic  count were
measured throughout the test period.

-------
       The background measurements of the CO concentrations were made
with the GFC monitor operating on a battery pack.  The monitor could be
placed on either side of the road in any selected location.  Data was taken
with the optical cell completely open so that real time measurements were
made with a one second time constant.  During one of the tests when the wind
was blowing from an unfavorable direction, the Ford Aeronutronic GFC and
the long path GFC system   used as the primary monitor were compared on
the same  side of the roadway (Figure 2).   The data illustrates the rapid
variation  in the roadside concentration of CO as well as the excellent agree-
ment between the instruments.

LOS ANGELES FREEWAY "ON" RAMP MEASUREMENTS -- May 1977
       Los Angeles, Calif.

       The purpose for collecting this data was to determine the feasibility
of monitoring the CO emission from individual vehicles using the Los Angeles
Freeway  system.   The data (Figure 3) was collected with the monitor placed
on top of a parked automobile beside an "on" ramp.  It can be seen that the
peaks from individual vehicles are not quite resolved.  However, it is be-
lieved that with minimal instrument modification individual resolution could
be achieved.

LAKE MICHIGAN PLUME STUDY -- July 1977  Muskegon, Mich.

       The Lake Michigan Plume Study is a multi-year program funded by
the Department of Energy and conducted by Battelle Northwest.  The purpose
of the study is to analyze the urban plumes from Chicago and Milwaukee as
they move across Lake Michigan and observe the change in composition. In
order to carry out the study, Battelle has outfitted a DC-3 with a complete
complement of air pollution instruments.  The monitors available for
measuring the CO concentration were not completely satisfactory.   Hence, it
was arranged to have the Aeronutronic GFC flown on several missions.  The
flights were carried out during the late summer of 1977, but the data has not
yet been reduced.

-------
                                 SECTION 5

                                DISCUSSION

       The measurement of the carbon monoxide concentration as an indica-
tor of air pollution has been an established practice for many years and as a
consequence adequate monitoring techniques are available.  This is particu-
larly true if the major interest is to determine the  health hazard due to
ambient concentration.  Many of the present monitoring instruments set one
ppm as the minimum detectable signal which is reasonable from the stand-
point of possible health hazards.  However, the accepted background concen-
tration for the northern hemisphere is 0.15 ppm_v.  Hence, in order to study
the normal ambient background concentration, a more sensitive instrument
is required.

       The unique GFC monitor used in these studies has permitted the
measurement of ambient CO background concentrations in difficult to measure
locations.  These measurements have contributed to our knowledge of both
the spatial and temporal variations of CO in the ambient environment.  This
information should be useful to air pollution modelers who must use and
interpret existing data.to verify their models.  Hopefully, the data collected
on individual vehicles will help those responsible for implementing air pollu-
tion control strategies.  Furthermore, in any future microscale studies
which involve the measurement of carbon monoxide, serious consideration
should be given to using such a monitor because of  its ability to follow the
real time variations in ambient concentrations of carbon monoxide.

-------
-5-5680
Ul


UJ
pC o
o  5

-------
                                                           s?^fe*^?r*3
Figure 2.  Comparison with SAI Long Path Monitor -- Silver Strand
                 Emission Study -- San Diego  5-4-77.

-------
                                 PDT
CO
                                           or v/ 1 L. s i-n R. e.
                             LOS>
                                    , 1377
      Figure 3.  San Diego Freeway "On" Ramp Data.

-------
                              REFERENCES

1.   Burch, D. E. , F. J. Gates, J. D.  Pembrook,  "Ambient CO Monitor, "
    Final Rep. on EPA contract No. 68-02-2219, prepared at Aeronutronic
    Ford Corp. ,  Newport Beach, Calif. ,  issued as EPA Rep. 600/2-76-210,
    July 1976.

2.   Pooler, F. , Jr., J. Air Pollut. Control As soc. , 24, 228-31 (1974).

3.   Ku,  R. T. , E. D. Hinkley,  "Long Path Monitoring of Atmospheric Carbon
    Monoxide, " Lincoln Laboratory Progress Rep. to National Science
    Foundation, Contract No. NSF/RANN/IT/GI-37603.

4.   Chaney,  L. W. ,  "Studies of Microscale  Variations in Ambient Concentra-
    tions of Air Pollutants, " University of Michigan Report,  EPA Grant
    R-803399.

5.   Chaney,  L. W. ,  R. T. Ku,  and W.A.  McClenny, Paper 75-56. 6 at 68th
    Annual Meeting of APCA, Boston, June 1975.

6.   Ku,  R. T. , L. W". Chaney,  and W.A.  McClenny, Paper 75-56. 5 at 68th
    Annual Meeting of A PCA, Boston, June 1975.

7.   Chaney,  L. W. ,  W. A.  McClenny, Environmental Sci. and Tech.  Vol.  11,
    page 1186, Dec. 1977.

8.  Chaney,  L. W. ,  Science, Vol. 199, 17 Mar 1978, pp. 1203-1204.

9.   McClenny, W.A. and L. W. Chaney,  "Pollutant Variability in the Regional
    Air  Pollution Study, " accepted for publication in Jour, of Air Pollu.
    Control Association,  April 1978.

10. Bartle, E. Roy, "Measurements of Carbon Monoxide from Highway
    Traffic Projected  for 1985, " SAI-75-687-LJ, 26 Nov. 1975,  Contract
    DOT-TSC-113.

11. Bartle, E. Roy, "Field Calibration of SAI Model 300 Long Path GFC Gas
    Analyzer, " SAI-76-593-LJ, 22 April  1976.
                                    13

-------
12.  Marmo,  F. F. , and M. J.  Scotto, "SLAPS Field Program for the Valida-
    tion of Sub-Grid Scale Dispersion Models, " Technology for Environmental
    Analysis series,  June 18, 1974.

13.  Kochelek, J. , "The Definition of a  Traffic Mixing Cell by Carbon Monox-
    ide Measurements, " Student paper  presentation to American Chemical
    Society,  St.  Louis Chapter -- March 30, 1977.
                                    14

-------
                     APPENDIX A

           Ambient Carbon Monoxide Monitor Based
on Gas Filter Correlation:  Performance and Application
                          15

-------
Unique Ambient Carbon Monoxide Monitor Based on Gas Filter Correlation:
Performance and Application

Lucian W. Chaney
University of Michigan, Ann Arbor, Mich.
William A. McClenny*
Environmental Sciences Research Laboratory, U.S. Environmental Protection Agency, Research Triangle Park, N.C. 27711
• A new type of ambient air monitor for carbon monoxide
was developed. The monitor was based on an infrared ab-
sorption technique termed "gas filter correlation". Water
vapor was measured independently in the monitor, and a
correction for interference in the CO measurement was made
automatically. Initial applications of the monitor demon-
strated its low noise equivalent concentration of 20 ppb-v and
a linear range of response up to 100 ppm-v. In studies near
roadways, the temporal and spatial characteristics of auto-
motive emissions were established by placing the absorption
cell directly in the ambient air. Comparisons with a second
type of CO monitor in a regional air monitoring network re-
sulted in specific recommendations for CO sampling tech-
niques.
  During recent studies as part of the Regional Air Pollution
 Study (RAPS) in St. Louis (1), the need for specialized in-
 strumentation for air quality monitoring and atmospheric
 modeling studies led to several developmental efforts by the
 Environmental Sciences Research Laboratory, U.S. Envi-
 ronmental Protection Agency. The result of one of these ef-
 forts was a specific, sensitive, fast-response carbon monoxide
 monitor for use over the full range of likely CO concentrations,
 from background levels of 50 ppb-v to the part-per-million
 levels encountered in urban areas. This monitor is the first
 ambient CO monitor based on the  technique of gas  filter
 correlation (2). Burch et al. (3) have described the design and
 fabrication of the  monitor. This paper is an account of the
 results of field and performance tests with emphasis on special
 applications.
  The gas filter correlation (GFC) monitor is one of a class of
 instruments that measure the changes in transmission of ra-
 diation due to absorption by gaseous species. Some perspec-
 tive on the evolution of instruments similar to the  GFC
 monitor can be obtained by reference to Hanst (4). The uni-
 queness of the GFC technique is  the manner in which the
 target gas absorption is separated from the absorption due to
 other species. This specificity is accomplished in a signal
 processing procedure that involves the use of an optical filter
 cell containing a high concentration of the target gas. The gas
 filter cell provides an optimal filter for radiation at wave-
 lengths readily absorbed by  the target gas, i.e., the  filter
 transmission characteristics anticorrelate with the absorption
 spectrum of the target gas. The specific manner in which the
 gas filter correlation is applied to CO detection is discussed
 in the next section.
  Although the present results represent the first application
 of the GFC technique to ambient air monitoring of a criteria
 pollutant, the technique already plays a basic role in pollutant
 monitoring of various sources. Previous uses of the technique
 have included source  monitoring for a number of gases: in
 industrial effluents (5) for NO, CO, and SO2; in automotive
 exhaust (6) for CO, CH4, C02, and CH^O; and  in exhaust
 clouds from rocket launch sites (7) for HC1 and HF. The
 technique is generally applicable to simple molecules whose
 spectra exhibit vibrational, rotational fine structure at at-
Figure 1a. Optical layout of gas filter correlation (GFC) monitor for
CO
Arrow emanating from source indicates light path for CO detection


 mospheric pressure. Short-path, open-air instruments based
 on  GFC are finding  applications in situations where  the
 measurement region of interest is inaccessible, such as across
 highways (8) or airport  runways (9). Long-path, open-air
 measurements of the HDO molecular species are also being
 applied over path lengths of up to 10 km (10).

 Description
  The GFC carbon monoxide monitor was designed and
 fabricated by Burch et al. (3). Details of the design features
 for optical, electronic, and mechanical components are given
 in ref. 3. The description  presented here is meant to provide
 a basis for understanding the main operational features of the
 monitor.
  A top view of the GFC monitor is presented schematically
 in Figure la, showing the components of the optical path for
 CO detection. During operation, sample air is continuously
 pushed through the sample cell. Radiation from the source is
 directed by optical transfer elements through the two main
 optical subsystems: the  rotating gas filter (designated as
 correlation cell  in Figure la) and the optical multipass (sam-
 ple) cell. The beam exits the sample cell through interference
 filter FC,  which limits the spectral passband to a few of the
 strongest CO absorption lines  in the 4.6-^m region. Detection
 of the transmitted radiation occurs at the PbSe thermoelec-
 trically cooled detector, C. The multipass system, originally
 described by White (11),  is ordinarily adjusted to 28 passes,
 giving a total optical pathlength in the cell of 11.2 m.
  Although the passband of filter FC is chosen to minimize
 interference from other gases, some residual H20 interference
 occurs. Therefore, HjO is also monitored, and the resulting
 signal is used in real-time to  automatically correct for H20
 interference in the CO reading. The optical path for H^O de
 tection consists of a double pass through  the sample cell. This
 is arranged by having a portion of the radiation that leaves
 mirror M4 overfill MS and fall on M7. The radiation reflected
 from M7 is focused on a window above FC and passes to a pair
 of filter-detector  combinations, one of which monitors ra
 diation in a spectral interval of weak HoO absorption and the
 other of which monitors  radiation in a spectral interval of
 strong H20 absorption. By comparing the amounts of radia-
 tion in the two  spectral intervals, H2O concentration can be
 determined. The filter-detector combinations for H20  de-
             Reprinted from ENVIRONMENTAL SCIENCE & TECHNOLOGY, Vol. 11, Page 1186. December 1977
               Copyright 1977 by the American Chemical Society and reprinted by permission of the copyright owner
                                                        16

-------
tection lie in a plane above the elements FC, LC, and Del C
of Figure la but are not shown.
  The gas correlation cell is constructed with two compart-
ments (Figure Ib): one compartment (gas cell 1) is filled with
one-half atmosphere of CO, and the other compartment (gas
cell 2) is filled with pure N2. Radiation transmitted through
cell 1 is completely attenuated at spectral positions where CO
absorbs strongly. The radiation transmitted by cell 2 is re-
duced by coating the exit window of the cell with a neutral
attenuator. In this way, the amounts of radiation transmitted
by the two cells are made approximately equal in the spectral
passband that reaches detector C through filter FC.
  In operation, radiation passes alternately through the two
cells as they are rotated by a synchronous motor drive. This
establishes a-signal modulation frequency of 33.3 Hz. Trans-
mission-to the deteetor is-constant if no absorption by the
ambient sample occurs. If CO is present in. the sample, the
radiation transmitted through cell  1 is not appreciably
changed, while that through cell 2 is changed. This imbalance
is linearly related'to CO^ concentration for small concentra-
tions: Other gas species absorb the radiatlon'-transmitted by
cells 1 and 2 in approximately equal amounts since their ab-
sorption structure does not correlate with that of CO (see ref.
2 for a more detailed treatment).
   Superimposed on the entrance window-of thejeell is a typical
 light chopper pattern; (Figure Ib) that creates a carrier-fre-
 quency 12 times the signal modulation frequency, i.e., a carrier
 frequency of 400 Hz. The detector output from  the CO
 channel is fed to two phase-sensitive amplifiers that separate
 the detector response at the signal frequency from the detector
 response at the reference (carrier) frequency. The signal due
 to CO is divided'by the reference signal to substantially reduce
 many of the cause&of sensitivity change, such as accumulation
 of material on optical components and variation in detector
 sensitivity.

 Zero and Multipoint Calibration
   During the performance tests and'applications to be dis-
 cussed, the instrument zero was determined by flushing the
 sample chamber with a gas containing negligible amounts of
 carbon monoxide. Four sources of "zero" air were used and
 found to agree within the noise limit of the instrument. These
 sources were: "ultrapure zero" air sold by Scott-Marrin, Inc.;
 prepurified argon; high^purified-grade helium sold by Air
 Products and Chemicals,  Inc.; and zero grade air sold  by
 Linde, Inc. The zero grade air (Linde) was further purified by
 passage through a "Hopcalite" CO scrubber sold by Mine
 Safety Appliance, Inc. A Standard Reference Material (SRM)
 purchased from the National Bureau of Standards, CO in  NS
 at 100 ppm-v, was diluted by dynamic mixing with one of the
 "zero" air sources and used to provide a multipoint calibration.
 The instrument  was shown to be linear over the range from
 20 ppb-v to 20 ppm-v for the 11.2-m pathlength. Adjustment
 of the number of passes in the multipass cell can be used to
 extend the upper limit of linearity to 100 ppm-v.
                                    GASCELlj
  The H2O channel was calibrated by establishing reference
signal levels at 0% relative humidity and 100% relative hu-
midity at 25 °C by bubbling zero air through water. The water
vapor interference for 100% relative humidity at 25 °C was 0.1
ppm-v of equivalent CO without any correction; with a cor-
rection the interference was reduced to less than the monitor
noise level of 20 ppb-v. Carbon dioxide interference was tested
by comparing C02 free air with air known to contain 330
ppm-v. The signal difference was undetectable.

Performance

  Convenience of operation is one of the main performance
features of the GFC monitor. No comsummables, pressurized
gases, mixer gases, or chemicals are required for operation.
The instrument response is independent of flow rate through
the sample cell and reasonably independent of temperature.
Chamber tests of the monitor, in which the ambient temper-
ature was varied to different controlled levels, showed small
variations in response to temperature changes over the range
16-32 °C. On a 2 ppm-v scale the sensitivity decreased 0.3%
of full scale per degree Celsius increase, indicating the change
in signal expected as the sample gas density decreases with
temperature. The zero reference level increased 0.3% of full
scale per degree increase; the exact cause of this effect is un-
known. Ambient pressure changes (usually ±3% of the mean
pressure) are also expected to alter instrument response due
to corresponding: gas density changes. Particular care must
be taken when using the monitor on an airborne platform since
t he pressure decrease-with altitude is roughly 0. 1% per 10m.
The lightweight (20 kg or 40 Ib) of the monitor and the small
size (62 by 27 by 12 cm) make it easily transportable, A dc-
to-ac converter along .with a set of six 6-V, 20-A-h batteries
permit operation in almost any location for at least  4 h.
  A unique feature of the monitor is that the-top and one side
of the sample section (Figure la) can be opened to the ambient
air. As a result of the open path, the response time can  be
optimized to permit accurate temporal characterization of the
CO from rapidly varying sources such as automobiles. For
reactive gases like HC1, which can be detected by the GFC
technique, integrity ofthe target gas can be maintained in  an
open path configuration  since no actual sampling occurs.
                 GASCEU)
  Figure 1b. View of components for gas filter cell; chopper and atten-
  uator attached with epoxy onto cell during assembly
  Three types of field tests have been selected to demonstrate
the application ofthe GFC monitor to monitoring problems:
roadside measurements, long-term comparisons with a gas
chromatograph-flame ionization detector for CO at selected
stationary sites in the RAPS, and helicopter measurements.
All measurements occurred during field experiments in the
RAPS during 1975 and 1976. Additional studies of compari-
sons with  CO monitors that1 incorporate  different types of
monitoring techniques, e.g., standard nondispersive infrared
and electrochemical monitors, have been made with excellent
results. Two of the initial applications of the GFC monitor
have been  reported elsewhere, i.e., laser long-path evaluations
(12) and comparisons of RAPS  station readings with area
averages (13).
  Roadside Measurements. For routine monitoring, ambi-
ent air is pushed through the-enclosed sample cell at a rate
determined by the pumping speed. The time required to ex-
change a factor e "' of the gas volume (2 L) of the sample cell,
assuming a well-mixed sample, was 40 s for the usual flow rate
of 3 L/min.
   With the top and side panels of the monitor's sampling cell
removed,  the measurement time constant depends on the
electronics (1 s minimum) rather than the gas exchange rate.
Hence, an in-situ measurement can be made  in which the
measurement path is essentially immersed in the ambient air.
                                                                           Volume 11, Number 13, December 1977   1187
                                                      17

-------
Figure 2a. CO concentration vs. time on sidewalk near inner-city
roadway
 Actual temporal variations of CO concentrations can be ob-
 tained in this way.
   A set of roadside measurements was planned in which the
 in-situ monitoring feature was required. Two questions were
 to be addressed: What are the temporal characteristics of the
 CO concentrations near a roadway? How far from the roadway
 can the characteristic temporal fluctuations be distinguished?
 These are basic questions that have implications to roadway
 modeling and to placement of stationary monitoring sites with
 respect to street and highway traffic.
   Although the combination of the monitor, battery pack, and
 chart recorder could not be hand-carried, they were easily
 mounted on a small laboratory cart that could be rolled along
 the sidewalk or across the surrounding area. The monitor
 sampling space was approximately 1 m off the ground. Mea-
 surements were made on the downwind side of the roadways,
 at the roadway edge, and then at intervals of 25 m to a distance
 of 100 m. Data were recorded at two locations. One was an
 inner-city location near  Grand Avenue on the campus of St.
 Louis University, and the other was near Page Road, a four-
 lane suburban highway at a location  1 mile east of interstate
 highway 270.
   Figure 2 gives two in a set of measurements taken at Grand
 Avenue. Figure 2a, showing data taken at roadside, indicates
 a minimum measurement of approximately 0.2 ppm-v and a
 maximum of more than 17 ppm-v, at which point the recorder
 saturates. The peaks, which occur every 1.5 min,  are syn-
 chronized with a nearby traffic light. The concentration
 minima appear to represent the urban background, while the
 peaks are due to the local automotive traffic.  This separation
 is exactly what is  required for microscale measurements.
   In Figure 2a, note that the concentration changes from 0.2
 ppm-v to more than 17  ppm-v in only 30 s. The maximum
 frequency component in the data is about 10 cycles/min or 600
 cycles/h. This characterization of the temporal variation of
Figure 2b. CO concentration vs. time 50 m downwind of roadway

1188  Environmental Science & Technology
the CO concentration will be used in the next subsection to
explain the results of comparison studies between the GFC
monitor and a monitor that takes discrete samples. The
electronic time constant during this measurement was 3 s.
Thus, with a 1 m/s wind speed, a spatial variation of 3 m could
be resolved. This is more detailed spatial resolution than is
normally required, but illustrates a capability that has been
lacking in previous microscale studies.
  Figure 2b illustrates the decrease in CO variability as the
distance  between roadway and monitoring location is in-
creased. Additional measurements at Page Road (14) and at
Grand Avenue provided the basis for a more generalized ob-
servation. At no time in this study was the effect of roadway
traffic measurable 100 m from the roadway. Although this set
of observations is very limited and is not considered a suffi-
cient study of CO variability near roadways, the main con-
clusion is essentially in agreement with the results of urban
survey measurements by Ott (IS). Whenever Ott measured
CO sufficiently far away (200 m in his results) from a road or
local source, he measured the urban background. Since inte-
grated samples were used in his studies, the nature of varia-
tions in CO concentrations as a function of distance was not
apparent as with a real-time monitor.
  A few examples of previous studies  that could have bene-
fited from the use of the GFC monitor are the San Jose street
canyon study (16), the St. Louis street canyon study (77), a
San Jose urban survej (15), the Oakbrook shopping center
study (18), and the New York City Roadway Configuration
Study (/9).  In addition, the following proposed studies are
now possible: the measurement of peak to average concen-
trations (20) and the measurement of the  distribution of
pollutants on the near-freeway microscale within the freeway
corridor (17).
  Long-Term  Comparison  in Selected  Regional  Air
Monitoring Stations. Prior to and following the 1975 summer
"intensive" period of research studies in the RAPS, a quality
assurance study was made of the entire Regional Air Moni-
toring System (RAMS) network. During the study, bag sam-
ples of ambient air were collected at the manifold inlet to in-
dividual stations for periods of 20 min.  Subsequently, the bag
samples were analyzed for CO with the  GFC monitor, and the
resulting readings were compared with the reported station
average for the same time period. Each of the RAMS stations
was equipped with a Beckman 6800 gas chromatograph with
a flame ionization detector. For this instrument, ambient CO
measurements were continually obtained, with updates at the
end of 5-min intervals. Readings were digitized to the nearest
0.1  ppm-v (this digitization is especially evident in Figure 4,
which is presented later). During the  20-min period of bag
sampling, four gas chromatographic cycles were completed.
A total of 19 preintensive and 16 postintensive data sets were
collected. The corresponding 35 data comparisons are shown
in Figure 3. Also shown in Figure 3 are 20 data sets of the same
type taken at site 105 (an urban site near the center of St.
Louis) in the RAMS network. Assuming that the GFC read-
ings were correct, these  measurements  indicate that the
RAMS network was reporting low values for the majority of
the comparisons, with a few exceptionally high station read-
ings. Both the set of network monitors taken as a whole and
the monitor at station 105 showed roughly equivalent scatter
diagrams.
  As a result of the comparison measurements, the following
question was addressed:  Do the  RAMS stations report low
values and if so, why? In an attempt to answer this question,
the GFC monitor was installed in several of the RAMS sites,
and a direct comparison was made over extended periods of
time.
  There are a total of 25 RAMS sites located on approxi-
mately concentric rings centered on  downtown  St. Louis.
                                                       18

-------
                                      X,
       II
       It
                                f
Figure 3. Scatter diagram showing comparison between measurements
of 20-min bag samples and simultaneous RAMS monitor measure-
ments
Figure 4. Comparison of CO concentration as measured by GFC and
RAMS monitors at site 122
                    Sill 1W
                    list" '5
                    niB»IIW:C5I
                    si LOWS n«reituot

                   — "HltCO 'DSC CAS fllTIBCnUBtt illOH
                   •BlCMUtH^im CKMnovajosnam
                           am   mst
Figure 5. Comparison of CO concentration as measured by GFC and
RAMS monitors at site 106
 There is one station near the center, six inner-city sites located
 on a 4-km radius, six near-suburban locations on a 9-km ra-
 dius, eight far-suburban locations on a 15-km radius, and four
 rural sites on a 40-km radius. All the stations used a Beckman
 6800 gas chromatograph to measure CO concentrations, and
 there was an identical unit in a central laboratory maintained
 by the RAPS staff. It was decided to compare the monitors
 in the laboratory and in at least one station on each concentric
 ring.
   The comparison between the GFC monitor and the Beck-
 man 6800 in the central RAPS laboratory was performed over
 a period of approximately one day. Both instruments were
 calibrated with the same gas, a Standard Reference Material
 (SRM) obtained from the National Bureau of Standards. The
 data proved to be virtually identical, having a correlation
 coefficient of 0.99.
  Examples of data collected at one of the RAMS stations are
shown in Figure 4. This station, site 122, was in a rural area;
 lack of local sources evidently accounts for the correlation
coefficient value of 0.96 and the close agreement with respect
 to absolute concentration. Monitoring sequences at sites 119
 (far-surburban site), 106  (near-surburban site),  and 105
 (inner-city site) were also  taken. The data were similar  to
 Figure 4, although the range of concentration variation was
 larger, e.g., from a maximum of 3.5 to 0.1 ppm-v at site 119.
 The correlation coefficients for sites 119 and 106 during 10-h
 monitoring sequences were 0.96 and 0.37, respectively, while
 the correlation coefficient for a 4-h monitoring sequence  at
 site 105 was 0.72. The data  from site 106 are shown in Figure
 5. The most significant difference in response between the two
 monitors at site 106 was during the initial 2-h period. Exclu-
 sive of this period the correlation coefficient was 0.95. How-
 ever, the RAMS data are consistently lower than the GFC data
 during the remainder of the 10-h sequence.
  Agreement between the two monitors at sites 122 and 119
 was excellent with a high correlation. Data from sites 105 and
 106  indicated significant  differences between the 5-min
 readings provided  by  the  Beckman 6800 and the corre-
 sponding GFC monitor readings. In summary, the laboratory
 data and1 data from the more remote sites demonstrated basic
 agreement between the two types of instruments, but the data
 from urban sites and the quality assurance data indicated  a
 lack of agreement for short-term comparisons. This set of
 observations could be explained by considering the gas chro-
 matograph sampling procedure and the data on temporal
 fluctuations in CO concentrations obtained  in  the near-
 roadway study.
  The RAMS manifold pump moves at least 400 L of aii/min
 through the station manifold. The gas ehromatograph extracts
 10 mL of air during injection and repeats this sampling pro-
 cedure every 5 min. The number of samples measured per
 hour is then 12. Since the data in Figure 2 indicate the likely
 existence of frequency components in the ambient CO con-
 centration of up to 600 cycles/h, it is obvious that the average
 of a limited number of gas chromatographic readings can be
 in error. The use of an integrating flask between the RAMS
 manifold and the gas chromatograph would serve to reduce
 the temporal variations in the sampled air and hence reduce
 the error. The best solution would be to use a continuously
 sampling monitor.
  A straightforward "yes" or "no" answer cannot be given to
 the original question, "Does the RAMS network report low
 CO concentration?" However, it has been established that the
 reported concentration  average for 20-min readings can
 misrepresent the true time-averaged concentration and that
an explanation can be formulated in terms of the sampling
 technique.  In other words, due to the nature of temporal
 variations in ambient CO concentrations, the probability that
discrete sampling intervals  occur during a period of low CO
concentration can be significantly greater than during a period
of high CO concentration. Since most of the RAMS sites are
located near lightly traveled streets where high concentrations
occur infrequently and in an erratic manner, the probability
that any sample measurement will be lower than the  true
average concentration is increased, that is, a predominance
of low readings is recorded by the Beckman 6800 with a few
exceptionally high readings. The data represented in Figure
3 are consistent with this explanation. It also seems obvious
from the Figure 3 data that, as averaging time is increased, the
RAMS average for CO approaches the average for the GFC
monitor.
  Helicopter  Measurements. In addition to the 25 sta-
tionary monitoring stations that form the  backbone of the
RAPS data collection program, data were also collected during
the summer intensive periods by two instrumented helicopters
which provided vertical profiles of pollutant concentrations
by monitoring during downward spirals over selected RAMS
sites. As a result of the need for a better CO monitor for in-
flight measurements, two test flights for the GFC monitor
                                                                          Volume 11, Number 13. December 1977   1189
                                                       19

-------
Figure 6. Measurements of CO concentration in industrial plume with
GFC monitor mounted in RAPS helicopter


were arranged, one on July 21 and the second on August 12,
1976.
  For the first test flight, during which routine monitoring
spirals were flown, the ambient concentration never exceeded
0.2 ppm-v. During the second flight, the helicopter made two
passes through a power plant plume. Variations in the monitor
response as a function of time are shown in Figure 6. Data were
only available as 1 -min averages. The flight path carried the
monitor through the plume, and after turning, back through
the plume again. While in the plume the monitor saturated
at 8.5 ppm-v. During  the same flight the concentration
dropped to less than 0.1 ppm-v. However, during most of the
flight the concentration exceeded 1 ppm-v.
  Calibrations of the monitor just prior to and immediately
after the second flight agreed within ±0.05 ppm-v. However,
because no in-flight calibrations were made, the data in Figure
6 must be considered qualitative. There are no known prob-
lems associated with flying the GFC monitor,  and the data
appear reasonable.

Conclusions

  An ambient real-time CO monitor based on gas filter cor-
relation has been developed. The monitor is sensitive enough,
with a noise equivalent concentration of 20 ppb-v, to monitor
background CO levels, or it can be used to monitor compliance
with Federal air quality criteria levels for CO (9 ppm-v for an
8-h average and 35 ppm-v for a 1-h average). Operation is
simple and flow independent; no consummables are required.
The monitor incorporates a unique feature that reduces re-
sponse time to a minimum.
  The monitor is particularly useful for roadway or microscale
studies  where short time variations in concentrations are
important. Measurements to demonstrate the monitor's ca-
pability to characterize the temporal and spatial  variations
in CO concentrations near roadways suggest its use in future
microscale studies. Successful use in a helicopter has also been
demonstrated.
  The monitor has been used to determine the reason for
apparently low CO concentrations reported by the RAMS
stations. Comparisons  between the GFC monitor and the
discretely sampling monitor being used in the RAMS stations
were made at selected stations. The nature of CO variability,
especially in  urban areas,  can bias a discretely sampling
monitor unless an appropriate means for sample integration
is provided. Information and insight gained tn this study
should prove valuable in planning for future studies.

Acknowledgment

   The authors acknowledge the help of the following indi-
viduals in providing information and equipment as part of
comparison studies with the RAMS: J. A. Reagan (EPA), F.
Pooler (EPA), and R. F. Mindrup, Jr.  (Rockwell Interna-
tional).

Literature Cited

(1) Pooler, F., Jr., J. Air Pollut. Control Assoc., 24,228-31 (1974).
(2) Burch, D. E., Gryvnak, D. A., "Analytical Methods Applied to Air
   Pollution Measurements", R. K. Stevens and W. F. Herget, Eds.,
   Chap. 10, Ann Arbor Science, Ann Arbor, Mich., 1974.
(3) Burch, D. E., Gates, F. J., Pembrook, J. D., "Ambient CO Moni-
   tor", Final Rep. on EPA Contract No. 68-02-2219, prepared at
   Aeronutronic Ford Corp., Newport Beach, Calif., issued as EPA
   Rep. 600/2-76-210, July 1976.
(4) Hanst, P. L., "Advances in Environmental Science and Tech-
   nology", J. N. Pitts and R. L. Metcalf, Eds., pp 91-213, Wiley, New
   York, N.Y., 1971.
(5) Herget, W. F., Jahnke, J. A., Burch, D. E., Gryvnak, D. A., Appl.
   Opt.. 15,1222-8(1976).
(6) Burch, D. E., Pembrook, J. D., "Instrument to Monitor CH4, CO,
  and C02 in Auto Exhaust", Final Rep. on EPA Contract No. 68-
  02-0587, prepared at Aeronutronic Ford Corp., Newport Beach
  Calif., issued as EPA Rep. 650/2-75-030, Oct. 1973; Burch, D. E.
   Gates, F. J., Gryvnak, D. A. Pembrook, J. D., "Versatile Gas Filer
   Correlation Spectrometer", Final Rep. on EPA Contract No. 68-
  02-1227, prepared at Aeronutronic Ford Corp., Newport Beach,
  Calif., issued as EPA Rep. 600/2-75-024, Aug. 1975.
(7) Bartle, E. R., Kaye, S., Meckstroth, E. A., J. Spacecr Rockets,
  9,836-41 (1972).
(8) Marmo, F., U.S. Dept. of Transportation,  Cambridge, Mass..
   private communication, 1976.
(9) Herget, W. F., U.S. Environmental Protection Agency, Research
  Triangle Park, N.C.. private communication, 1976.
(10) Meredith, R. E., Science Applications, Ann Arbor, Mich., private
  communicaton, 1976.
.(11) White, J. U., J. Opt. Sac. Am., 32,258-88 (1942).
(12) Ku,  R. T., Hinkley, E. D.. "Long-Path  Monitoring of Atmo-
  spheric Carbon Monoxide", Lincoln Laboratory Progress Rep. to
   National Science Foundation, Contract No. NSF/RANN/IT/
  GJ-37603.
(13) Preliminary  Rep. to EPA  by the University of Michigan,
  "Sub-Grid  Scale Characterization in the St. Louis Regional Air
   Pollution Study", EPA Grant No. R-803399, Grants Office at Re-
  search Triangle Park, N.C.
(14) "Sub-Grid Scale Characterization Studies in the Regional Air
   Pollution Study in St. Louis", L.  W. Chaney. Univ. of Michigan,
  Ann Arbor, Mich., in preparation.
(15) Ott, W., PhD dissertation. Stanford University, Palo Alto, Calif..
   Oct. 1971.
(16) Johnson, W. B.. Dabberdt, W. P., Ludwig, F. L.. Allen, R. J.,
  "Field Study for Initial Evaluation of an Urban Diffusion Model
   for Carbon Monoxide", Contract CAPA 3-68( 1-69), Stanford Re
  search Institute, June 1971.
(17) Ludwig, F.  L., Dabbardt, W. F., "Evaluation of APRAC-1A
   Urban Diffusion Model for Carbon Monoxide", Contract CAPA
   3-68(1-69), Stanford Research Institute, Feb. 1972.
(18) Patterson, R. M., Record. F. A., "Monitoring and Analysis of
   Carbon Monoxide and Traffic Characteristics at Oakbrook",
   EPA-450/3-74-058, Nov. 1974.
(19) General Electric Co., "Study of Air Pollution Aspects of Various
   Roadway Configurations", PB 211235,  New York City Contract
   No. 209624.
(20) Slade, D. H., "Meteorology and Atomic Energy". TID-24190,
   July 1968.
Received for review March 16,1977. Accepted July 11,1977. Mention
of commercial products or company names does not constitute en-
dorsement by the U.K. Environmental Protection Agency.
1190   Environmental Science & Technology
                                                          20

-------
                  APPENDIX B

Carbon Monoxide Automobile  Emissions Measured
    from the Interior of a Traveling Automobile
                       21

-------
Reprint Series
17 March 1978, Volume 199, pp. 1203-1204
              Carbon Monoxide Automobile Emissions
       Measured from the Interior of a Traveling Automobile
                               Lucian W. Chaney
            Copyright © 1978 by the American Association for the Advancement of Science
                                     22

-------
 Carbon Monoxide Automobile Emissions

 Measured from the Interior of a Traveling Automobile

   Abstract. During a procedure to monitor carbon monoxide (CO) concentrations
 inside a traveling car, it was discovered that CO emissions from individual passing
 vehicles produced accurately measurable increases in the CO concentration. The CO
 produced by individual vehicles varied by three orders of magnitude; this finding
 demonstrates that a relatively small number of cars can be responsible for  a high
 percentage of total vehicle CO emissions.
   The purpose of this study was to deter-
 mine the CO concentrations to which the
 driver of a typical passenger car was ex-
 posed while traveling under a variety of
 traffic conditions. The maximum concen-
 trations of CO  permitted by the Fed-
 eral  Ambient  Air Quality  Standards,
 (FAAQS) (/) set in accordance with the
 Clean Air Act,  are 9  parts per miUion
 (ppm) for 8 hours and 35 ppm for 1 hour.
 While recording CO concentrations in-
 side automobiles on lightly traveled sec-
 tions of  an interstate highway, I ob-
 served  that CO  emissions from individ-
 ual passing cars produced clearly defined
 peaks on  the strip chart recorder. Sub-
 sequent measurements on 760 vehicles
 showed a surprisingly  wide variation in
 CO concentrations from 0.05 ppm to 45
 ppm. Analysis of the data showed that a
 large proportion of the vehicles  mon-
 itored produced only a small amount of
 the total CO, whereas a relatively small
 number of the total vehicles contributed
 almost  half of the total monitored CO.
   The gas filter correlation monitor used
 for this study was especially designed for
 the Environmental  Protection Agency
 for use in  modeling studies  (2) carried
 out during the  St. Louis Regional Air
 Pollution  Study  (.J). The monitor is port-
able and can be battery-operated; hence,
the monitor can  readily be operated on
the front passenger seat of an automo-
bile. I determined the CO concentrations
by  measuring the  attenuation of a fo-
cused infrared beam which makes 28 tra-
verses of a cell 0.6 m long. This is ac-
complished with  mirrors placed at each
end of the cell. The top  of the cell com-
partment was removed  to permit an in
situ measurement of the CO concentra-
tion. The time constant was set at 3 sec-
onds.  The  windows of the  automobile
were  closed, and  the  ventilation  was
forced by  operating  the  automobile
blower fan at the maximum speed. The
data were recorded on  a portable strip
chart  recorder mounted directly under
the monitor so that the  record could be
observed and marked by the driver. Two
small  (15 cm in outside diameter and 51
cm long)  high-pressure cylinders, one
containing argon and the other contain-
ing 3.0 ppm of CO  in nitrogen were
placed on the floor behind the front seat.
The gases in these two cylinders were
used in a daily, zero and span, two-point
calibration, usually carried out at a rest
stop. Before and after the trip, I carried
out a  five-point calibration, using a gas
mixture calibrated  by the  National Bu-
reau  of Standards.  The average  dif-
ference between the calibrations was 3
percent, and I estimate  that the  maxi-
mum measurement error  is 10 percent.
  The data were collected during a two-
part, cross-country trip. Part 1 was from
Chicago to New Orleans,  18 to 20 March
1977, and part 2 was from New Orleans
to San Diego, 27 March to 2 April 1977.
The expectation was that the average CO
concentration  would gradually increase
as cities were approached, reach a maxi-
mum near the center of a  city, and fall to
a minimum in the rural areas. In general,
this was found to be true; however, very
large fluctuations  occurred depending
mostly  on  traffic density  and  traffic
speed.
  Examples of data collected on heavily
traveled interstate highways are  those
from the Dan Ryan Expressway (1-94) in
Chicago (Fig.  la) and the  San  Diego
Freeway (1-405) in Los  Angeles  (Fig.
Ib). The peaks in the concentration are
primarily  the  result of  traffic  slowing
down as a result of congestion. When the
traffic slowed to 10 miles per hour (mph)
or less (1 mile  =• 1.6 km), the CO con-
centration  usually exceeded  15  ppm:
when it halted completely, the CO con-
centration was about 45 pp'rii.
  In order to compare the CO concentra-
tion on  interstate highways with that in
maximum downtown traffic, I made a
trip through  downtown New Orleans be-
tween 1200 and 1230 C.S.T. on 23 March
1977. The traffic  moved  at only a few
miles per hour and stopped at every traf-
fic light. The CO concentration varied
from 2 to 50 ppm, depending primarily
on the number  of stationary  or acceler-
ating  vehicles in close vicinity  to the
       _ a /Traffic stoppid
                                                            1430                    150C-                    1530
                                                                                    Time (M.S.T.I
                                                            Fig. 1. The CO concentration as a function of time inside an automo-
                                                            bile traveling on an interstate highway, (a) A 12-minute period in the
                                                            late morning on 17 May 1977 while traveling south on 1-94 through
                                                            Chicago; the observed peaks occurred when the traffic slowed as a
                        '°'°       n>2»       "so       io«o     result of congestion, (b) A 50-minute period in mid-morning on 10
                           Time (POT.)                          May J977 whiie traveling north on f-405 from Santa Monica to Ven-
tura, California; the peaks occurred as a result of congestion at the freeway interchanges, (c) A 75-minute period on 1 April 1977 along a rtirai
section of 1-8 (starting point, 150 miles west of Tucson, Arizona); the peaks  are due to the passage of individual vehicles.
SCIENCE. VOL. 199. 17 MARCH 1978                   0036-S075/78/0317-l203$00.50/0 Copyright ©  1978 AAAS                   1203
                                                23

-------
.a
~
.Percent of t
	
	

otal vehicles
Percent of
total increase in
CO concentration
i — J.— ,

, 	 ,
1
1
1
     .05 ppm <.5ppm «5ppm "30 ppm > 30 ppm
      Class 1 Class 2  Class 3 Class 4 Class 5
                                                 I
                                                              Q New Orleans to
                                                                   Brookshire, Tex.
                                                              D Brookshire, Tex., to
                                                                     Wilcox,  Ariz.
                                                              • Wilcox, Ariz., to
                                                                  San Diego, Calif.
                     I
     .05 ppm < .5 ppm <5ppm <30 ppm >30 ppm
     Class 1 Class 2 Class 3 Class 4 Class 5
 Fig. 1. Data summary of 760 vehicles monitored on 1-8 from New Orleans to San Diego (28
 March to 2 April 1977): (a) distribution of vehicles and CO emission by class; (b) distribution of
 vehicles by class for three subgroups of the total sample.
monitoring vehicle. The results of these
experiments  correlated  with  the  ex-
pectations.
  The surprising results occurred along
the rural, rather than the u.-ban. portions
of the route. The strip chart used for re-
cording the rural data required that the
sensitivity and chart speed be adjusted to
accommodate the maximum CO concen-
tration and the traffic density. As a result
of the attention required to make these
adjustments, the speed of the monitoring
vehicle was somewhat  slower than nor-
mal. Thus, many of the other vehicles on
the highway were passing at their normal
cruising speed of about 60 mph. I noticed
that  the CO  concentration in the in-
strumented   vehicle   would  increase
abruptly a few seconds after the passage
of some but not all vehicles. I decided to
monitor the increase in CO concentra-
tion due to the passage of single vehicles
in a  uniformly  controlled manner. The
procedure that 1 adopted was to drive the
monitoring vehicle at  about 52 mph so
that most of the other traffic passed with
a differential speed of from 3 to 10 mph.
The usual pattern of the passing vehicles
after  passing was to return to the travel-
ing lane about 50 feet ahead of the mon-
itoring  vehicle.  If for some reason the
passing vehicle  remained in the passing
lane and there was no other vehicle ap-
proaching, then I also moved the mon-
itoring vehicle  into the passing lane. A
sample recording of some of the data col-
lected by this technique is shown  in Fig.
Ic. A few of the peaks  that were posi-
tively identified  with a given vehicle are
noted.
  The monitoring procedure  described
above was  developed  during the trip
from  Chicago to New Orleans, and was
used during the trip from New Orleans to
San Diego to collect  the data summa-
rized  in Fig. 2. The only vehicles includ-
ed  in the  tabulation  were  those that
passed singly with sufficient distance be-
tween the other vehicles to permit a clear
identification  of the peak in  the  mon-
itored CO concentration with the passing
    1204
vehicle.  In order for this condition to
hold, the traffic density had to be suffi-
ciently low. Hence virtually all the data
were collected in rural areas, usually 20
miles 01 more from a metropolitan area.
  While  recording the data, 1 observed
that heavily loaded  vehicles  produced
high CO  concentrations. Also, whenever
a grade was being ascended, the CO con-
centrations increased. In order to  deter-
mine the effect of a steep grade, 1 fol-
lowed a  truck over the Sierra Nevada
mountains. The CO concentrations  in-
creased and decreased by more than one
order of magnitude, depending on the
grade. The CO concentration inside the
monitoring vehicle reached  50 ppm  or
the most severe part of the grade and re-
mained over 25 ppm  for Vi hour. Clearly.
under some circumstances the potential
to exceed the FAAQS exists.
  I monitored a total of 760 vehicles by
this technique on part 2 of the trip.  These
were divided into five classes based on
the maximum increase m CO concentra-
tion measured inside the monitoring ve-
hicle. The classes were as follows: class
I. maximum increase less than 0.05 ppm;
these vehicles were  all 1975. 1976, or
1977 models lightly loaded (less than half
of the designed load):  class 2, maximum
increase   between 0.05 and  0.5  ppm;
these were mostly  1970 through 1974
models and some newer cars carrying
heavy loads (100 percent or more  of the
designed  load); class 3, maximum  in-
crease between 0.5 and 5 ppm; this class
includes  largely older cars but also some
heavily loaded newer vehicles such as
motor homes and pickup trucks: class 4.
maximum  increase  between 5 and  30
ppm: these were all heavily loaded vehi-
cles, most of them older than 1970; and
class 5. maximum increase greater than
30 ppm;  there were  two vehicles in this
class: one was a pickup truck pulling a
four-horse trailer: the other (CO concen-
tration. 45 ppm) was a no-brand fuel oil
delivery  truck
  The peak  measured increase in  CO
concentrations for  all  the vehicles  in
 each class were added, and the percent-
 age contribution of each class to the total
 concentration  was  computed. The re-
 sults are shown in Fig. 2a. If the vehicles
 falling in class 1 and class 2 are grouped
 together, they represent 76 percent of all
 the vehicles observed and contribute 12
 percent of the total CO; class  4 and class
 5 combined represent 3.3 percent of the
 total vehicles and contribute  45 percent
 of the total CO. The finding that 30 per-
 cent  of the vehicles monitored appeared
 to be equipped with catalytic  converters
 is in agreement with a recent study done
 in  California (4).
   In  order to develop some  feeling for
 the variability of the data, I divided the
 total  data set into three nearly equal
 pans. The first third contained the data
 collected from New Orleans  to  Brook-
 shire, Texas;  the  second third  from
 Brookshire, Texas, to Wilcox, Arizona;
 and the last third from Wilcox, Arizona,
 to San Diego, California. The percentage
 of vehicles in each class was determined
 for each third of the total sample (Fig.
 2b). In  the second third of the sample.
 there was a smaller percentage of the ve-
 hicles in class 2 and a greater percentage
 in class 3 than for the sample as a whole.
 The reason is probably due to more up-
 hill grades in this section of the route as
 well  as differences  in wind speed  and
 wind direction. However, the overall
 pattern  is  much the  same for all three
 sections of the route, which reaffirms the
 character of the total distribution
   As a result of the experiments con-
 ducted in  the urban areas. I  conclude
 that most drivers are not exposed to tox-
 ic doses of CO but that under certain cir-
 cumstances the potential exists  to ex-
 ceed  the  FAAQS.  The major observa-
 tions  resulting from the data are that the
 CO automobile emissions can vary  by
 three orders of magnitude depending on
 a number of factors and that more atten-
 tion should be directed toward that small
 minority of vehicles which are the  major
 polluters.
                  LUCIAN  W. CHANEY
 107 Research Activities Building,
 University of Michigan.
Ann Arbor 48109

           References and Notes
 I.  Fed. Kegist. 36 (No. 84), (30 April 1971).
 2.  D. E. Burch. F. J. Gales, J. D. Pembrook./lm-
    bient CO Monitor (final report on EPA contract
    68-02-2219. prepared at Acronulronic Ford Cor-
    poration, Aeronutronic  Division. Newport
    Beach. CaJif.. and issued as EPA Report 600/2-
    76-210. July 1976).
 3.  L.  W. Chancy and W  A. McClenny. Environ.
    So. Technol., in press.
 4.  "Los Angeles Field Modeling and Measurement
    Study" (Report AMC 8305.41R. Rockwell Inter-
    national. Air Monitoring Center, Newbury Park
    Calif..  EPA contract 68-02-2463.  18 February
    1977).
 5  Supported by Environmental Protection Agency
   grant R-803399.
6 October 1977: revised 5 December 1977

                     SCIENCE. VOL. 199
                                                          24

-------
                                   TECHNICAL REPORT DATA
                            (Please read Instructions on the reverse before completing)
 REPORT NO.
 EPA-600/2-79-139
             3. RECIPIENT'S ACCESSION-NO.
 TITLE AND SUBTITLE
 AMBIENT AIR CARBON MONOXIDE  MEASUREMENTS
                                                            5. REPORT DATE

                                                              August  1979
             6. PERFORMING ORGANIZATION CODE
 . AUTHOR(S)

 Lucian W. Chaney
             8. PERFORMING ORGANIZATION REPORT NO,
9. PERFORMING ORGANIZATION NAME AND ADDRESS
                                                            10. PROGRAM ELEMENT NO.
 University of  Michigan
 Ann Arbor, Michigan  48109
               1AD712B   BB-32 (FY-78)
             11. CONTRACT/GRANT NO.

                R-803399
 12. SP9NSORING AGENCY NAME AND ADDRESS
 Environmental  Sciences Research  Laboratory - RTF,  NC
 Office of  Research and Development
 U. S. Environmental Protection Agency
 Research Triangle Park, North Carolina  27711
             13. TYPE OF REPORT AND PEfllOD COVERED
                Final
             14. SPONSORING AGENCY CODE

               EPA/600/09
 15. SUPPLEMENTARY NOTES
 16. ABSTRACT
 A new type  CO monitor was  applied to special  ambient air measurement problems.  The
 monitor, a  gas filter correlation (GFC)  instrument, was designed  specifically for
 use in the  St.  Louis Regional  Air Pollution  Study (RAPS), but has  been applied to
 several other measurement  requirements.  The monitor has an inherently fast response
 of less than  one second and  has proved useful  in documenting extremely variable
 monitoring  situations.  The  monitor was  used in  nine separate studies and typical
 data are presented.  The most  significant contribution from these  studies has been
 the ability to document the  extreme variability  of carbon monoxide in urban
 environments.
17.
                                KEY WORDS AND DOCUMENT ANALYSIS
a.
                  DESCRIPTORS
                                               b.lDENTIFIERS/OPEN ENDED TERMS
                           c. COSATI Field/Group
 *Air pollution
 *Carbon monoxide
 *Monitors
  Field tests
                                 13B
                                 Q7B
                                 14B
18. DISTRIBUTION STATEMENT


  RELEASE TO PUBLIC
19. SECURITY CLASS (ThisReport)
  UNCLASSIFIED
                                                                          21. NO. OF PAGES
20. SECURITY CLASS (Thispage)

  UNCLASSIFIED
                           22. PRICE
EPA Form 2220-1 (9-73)
                                             25

-------