EPA-650/2-73-026
REMOTE SENSING
OF
AIR POLLUTION
IN URBAN AREAS
by
M. L. Streiff and C. B. Ludwig
General Dynamics
Convair Aerospace Division
P.O. Box 1128
San Diego, California 92112
Contract Number 68-02-0020
Program Element No. 1A1010
EPA Project Officer: William F. Herget
Chemistry and Physics Laboratory
National Environmental Research Center
Research Triangle Park, N. C. 27711
Prepared for
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
Washington, D. C. 20460
August 1973
-------
This report has been reviewed by the Environmental Protection Agency and
approved for publication. Approval does not signify that the contents
necessarily reflect the views and policies of the Agency, nor does
mention of trade names or commercial products constitute endorsement
or recommendation for use.
11
-------
TABLE OF CONTENTS
Section Page
LIST OF FIGURES iv
LIST OF TABLES vi
SUMMARY vii
1 INTRODUCTION 1-1
2 ANALYSIS OF MOLECULAR SPECTRA 2-1
2. 1 THE CONCEPT OF EQUIVALENT WIDTH - 2-1
LABORATORY PROCEDURE
2.2 SPECTRA OF POLLUTANTS AND ABSORPTION 2-11
COEFFICIENTS
3 PERFORMANCE OF THE ROSE SYSTEM 3-1
3.1 GENERAL 3-1
3.2 SYSTEM NOISE 3-2
3.3 SCINTILLATION NOISE 3-3
3.4 SIGNAL STRENGTH 3-5
4 LONG-PATH TRANSMISSION MEASUREMENTS 4-1
4.1 ONE MILE MEASUREMENTS 4-1
4.2 TWO MILE MEASUREMENTS 4-15
5 EMISSION MEASUREMENTS 5-1
5.1 STEEL MILL 5-1
5.2 POWER STATION 5-5
5.3 HOSPITAL INCINERATOR 5-5
ill
-------
TABLE OF CONTENTS
Section Page
6 TRANSMISSION MEASUREMENTS OF EXTENDED 6-1
SOURCES
7 SUPPORTING LABORATORY DATA 7-1
7.1 DETECTOR, NOEE MEASUREMENTS 7-1
7.2 MONOCHROMATOR CALIBRATION - 7-6
TEMP. EFFECT
8 RECOMMENDATIONS 8-1
9 REFERENCES 9-1
APPENDIX
I REPORT TM 6-125 Ph-336 A-0
II MOLECULAR SPECTRA TABLES A-49
IV
-------
LIST OF FIGURES
Figure Page
2-1 True Spectral Line Shape 2-3
2-2 Observed Line Shape Assuming a Triangular Slit Function 2-3
2-3 Collision Broadened Curves of Growth for Various 2-7
Intensity Distributions
3-1 Relative Signals from Source and Reference Channels - 3-6
One Mile Path
3-2 Calculated System Transmission 3-7
3-3 Calculated Transmission Resulting from Aberrations 3-8
4-1 7 to 13 p, Scan with One Mile Path, Av = 5 cm"1 4-2
-1 -1
4-2 Noise at 1000 cm with One Mile Path, AV = 5 cm 4-3
4-3 7 to 13 M, Scan with One Mile Path, A\JI()= 1 cm~ 4-4
4-4 Noise at 1000 cm with One Mile Path, AM = 1cm 4-5
4-5 7tol3|j,I/I Scan with One Mile Path A v,= 5 cm" 4-6
o i
4-6 3 to 5u. Scan with One Mile Path, A\>.« 6 cm" 4-8
4-7 Noise at 2500 cm with One Mile Path, Lv^m 6 cm" 4-9
4-8 3 to 5 p, I/I Scan with One Mile Path, A v4"6 cm" 4-10
4-9 Volz Photometer Calibration and Typical Aerosol Optical 4-12
Thickness Variation in San Diego, California
4-10 Volz Photometer Measurements 4-13
4-11 Noise at 2500 cm with Two Mile Path, A\j = 3 cm" 4-15
-1
4-12 8 to 13 n Scan with Two Mile Path, Av «• 1 cm 4-16
4-13 Ozone Concentration with Time - July 28, 1972 4-18
4-14 3 to 5 n Scan with Two Mile Path, Ays 3 cm" 4-19
4 -i
4-} 5 Digitized Spectrum of CO Band, Av~ 1.8 cm 4-20
5-1 Sky Background 5-2
5-2 Coke Oven Plume Emission 5-2
5-3 Difference (Plume - Sky) 5-2
5-4 Stack Emission - Steel Mill, Av. = 9 cm" 5-4
-------
LIST OF FIGURES
Figure Page
5-5 Plume, Sky and Stack Scans Power Station Stack 5-6
#3, Av » 3 cm
5-6 Difference Spectrum (Plume - Sky) Power Plant 5-7
Stack f3
5-7 Plume Scan, 3 - 5p,, Power Plant Stack #3 5-8
5-8 Variations in Plume Emission - Power Plant Stack #3 5-9
i
5-9 Variations of Hospital Incinerator Emission, v=2390 cm 5-11
5-10 Comparison of Two Scans, 3 to 5n, Hospital Incinerator 5-12
6-1 Comparison of CO Spectra with A \>=* 4.4 cm and 6-2
1.3 cm"1
6-2 CO Spectrum at Refinery, Av« 1.3 cm 6-3
6-3 3 to 5 p, Scan at Refinery, A-j m 1.4 cm" 6-5
••I
6-4 Comparison of Spectra Near 3000 cm 6-6
6-5 7 - 13 n Scan at Refinery ANJ, Q = 0.65 cm~ 6-7
7-1 Waveanalyzer Measurements of Detector Noise 7-2
7-2 Analog Measurements of Noise 7-5
7-3 Divider Box Schematic 7-9
VI
-------
LIST OF TABLES
Table
Ai-1
Ar2
A. 3
A. 4
Af&
A. 6
A. 7
A. 8
A. 9
A. 10
A, 11
A. 12
A. 13
A. 14
A. 13
A. 16
A. 17
A. 18
Band Strengths and Band-Averaged Absorption
Coefficients and Fine Structure Parameters of
Pollutants
-1
Band Model Parameters for CO in 5 and 20 cm
Intervals
-1
Band Model Parameters for CO in 5 and 20 cm
Intervals
-1
Band Model Parameters for SOy in 5 and 20 cm
Intervals
Band Model Parameters for NOn in 5 and 20 cm
2
Intervals
-1
Band Model Parameters for NO in 5 and 20 cm
Intervals
Band Model Parameters for NO in 5 and 20 cm
Intervals
-1
Band Model Parameters for NH in 5 and 20 cm
3
Intervals
Band Model Parameters for HCHO in 5 and 20 cm
Intervals
Band Model Parameters for HO in 5 and 20 cm
Intervals
Band Strengths and Fine Structure Parameters of
Pollutants not Included in Table D-3
Absorption Coefficients for HC1
Absorption Coefficients for HF
Absorption Coefficients for CH
Absorption Coefficients for C2H
Absorption Coefficients for PAN
C..EL Benzene
D O
18-O0 - Ozone
Page
A-49
A-50
A-52
A-56
A-58
A-60
A-61
A-65
A-69
A-71
A-79
A-80
A-81
A-82
A-83
A-84
A-85
A-85
Vll
-------
SUMMARY
Remote sensing of air pollution was made using a scanning spectrometer (ROSE
system) and a Michelson interferometer. Both systems were used for transmission
measurements over nominal path lengths of 1 and 2 miles and for stack emission
measurements. A comparison shows that tiie spectra of both instruments are of
the same quality. The greater throughput and multiplex advantage of the inter-
ferometer system manifests itself in the shorter time required to take the spectra,
(typically, 10 sec versus 1200 sec). This time advantage is partly lost when the
effort is considered which is required to reduce the interferograms to spectra by
a computer. However, when an on-the-line computer is available, this effort would
be greatly reduced.
In addition to the above measurements, the ROSE system alone was used to
measure the pollution from an extended source (oil refinery) over a path length of
1/4 mile.
All of the spectra were analyzed and a number of pollutants were identified.
In the transmission measurements, the ozone and carbon monoxide concentrations
as a function of time were determined. The presence of other pollutants such as
nitrogen dioxide, sulfur dioxide and several hydrocarbons is indicated. A more
definite identification and determination of concentration can be made, when a
computer analysis procedure is operational. In the stack emission measurements,
pollutants such as carbon monoxide, sulfur dioxide and many hydrocarbons were
identified.
Transmission spectra taken during hot days were influenced by scintillation
noise, but much less than anticipated. Typically, a signal-to-peak-to-peak-noise
ratio of 26 observed during a test at night is reduced to about 9 during a test on a
hot day. Also, signal fluctuations during stack emission measurements were less
than anticipated. However, this holds true only when the field-of-view is smaller
Vlll
-------
than the plume.
The field ruggedness and operational performance of the ROSE system was
demonstrated by these tests to be satisfactory. A number of problems which did
arise during the course of the testing were corrected either during or after the
test period.
IX
-------
1
INTRODUCTION
Remote sensing of air pollution in the Los Angeles area with the "Remote Optical
Sensing of Emission (ROSE)" system is described. The work was conducted in
June, July and October 1971, under contract 68-02-0020 with the Environmental
Protection Agency. A portion of the study was supported by the General Dynamics
Electro Dynamic Division (Pomona); the results of this portion of the study are
reported in TM 6-125 Ph-336 which is reproduced in the Appendix of this report.
A complete description of the ROSE system is presented in report EPA-R2-
72-052 (GDC-DBE-72-001) along with a description of preliminary field tests made
in San Diego over a path length of about 1/4 mile on June 23 and 24, 1971.
The analysis of molecular spectra is set forth in Section 2 of this report.
The performance of the ROSE system under field conditions is discussed in
Section 3 of this report.
Transmission measurements are reported in Section 4 of this report. These
tests were made, with a nominal one-mile path length, from June 29 to July 2, 1971,
and, with a nominal two-mile path length, from July 27 to 29, 1971.
Emission tests were made of sources at three locations on successive days
from July 13 to 15, 1971 inclusive: a steel mill, a power plant and a hospital
incinerator. The results of these tests are reported in Section 5 of this report.
Extended source transmission measurements at an oil refinery location were
made on October 20 and 21, 1971 with a path length of approximately 1/4 mile.
The results of these tests are given in Section 6 of this report.
Supporting laboratory work is described in Section 7 of this report.
The program leader for these tests was Dr. C. B. Ludwig assisted by Dr.
M. Griggs; they were responsible for test planning (together with government project
officers), conduction of the field tests and report preparation until their departure
from the company in mid-1972. M. L. Streiff assisted by C. R. Claysmith was
1-1
-------
responsible for equipment operation and modifications during and after the field
tests. Mr. Streiff had principal responsibility for completion of the field test
report. Mr. G. W. Ashley was in charge of the Pomona interferometer activities.
The government project officers were Mr. John S. Nader and Dr. William
F. Herget. They were assisted by Dr. H. M. Barnes and Mr. R. Rollins during
the field tests.
1-2
-------
ANALYSIS OF MOLECULAR SPECTRA
2.1 THE CONCEPT OF EQUIVALENT WIDTH - LABORATORY PROCEDURE
The remote sensing of pollutants is presently done with scanning instruments
which have a spectra] resolution from 1 cm to several wave numbers. In the
case of diatomic and light polyatomic molecules, single lines can be observed
with a spectral resolution of 0.1 cm . In the case of heavier polyatomic molecules
and of lower spectral resolution, a number of lines will be observed simultaneously.
Thus, for the interpretation of measured spectra in terms of pollutant concentration,
spectral parameters have to be known botli For the single line and multiple line
cases.
2.1.1 Single Line with Lorentz Shape
The total absorption over a wavenumber interval A w containing the entire line is
given by the Ladenberg-Reiche function.
W = 2 na X e"X IQ (X) + ^ (X)) (2-1)
where W is the equivalent width (cm ), a is the half width at half height in cm ,
I , I are modified Bessel functions and
X = Su/2rrQ', (2-2)
where S is the line strength and u is the optical path (cm-atm). The line shape
for a collision broadened line is, according to Lorentz, given by
Q) + a
2-1
-------
The two parameters S and a are determined in the laboratory by making curve-of-
growth measurements. It is convenient to make measurements in the two limits
of small and large values of X:
W = Su for X < 0.02 p (2-3)
1/2
W=(4S*U) forx>12.5p-1 (2-4)
These approximations (called the 'linear" and "square root" approximations) are
valid within p percent. For instance, Equations (2-3) and (2-4) are accurate to
within 5 percent when x < 0.1 and x > 2.5, respectively. Experience shows that
for a reliable measurement of S three data points in the linear region should be
obtained and that for a reliable measurement of the product (Sa) three data points
should be obtained in the square-root region. Care must be exercised that the
assumptions of the validity of these approximations are fulfilled. This is done
by plotting the data points (W versus u) on logarithmic scales. In the linear region,
the plot is a straight line with a slope of 1, and in the square-root region, it is
also a straight line with a slope of 1/2. Between these two segments lies the
"transition" region.
Experimentally, the procedure is as follows: The spectroscopic instrument
with a slit function g is scanned over a single spectral line, whose true shape is
.shown in Figure 2-1. The true shape of the line is distorted by the slit function
g(uru ), and appears at the output of the instrument in the shape shown in Figure
2-2. One can show that the area under the spectral line (total absorption «*
equivalent width) is the same in Figure 2-1 and 2-2. That means that W is an
invariant and does not depend upon the slit function of the instrument. In other
words, the W determined in the laboratory with one instrument can be used to
interpret the field data taken with another instrument. In mathematical terms,
this fundamental fact is expressed as
*G. N. Plass JOSA 48,690 (1958)
2-2
-------
Fig. 2-1. True Spectral Line Shape
Fig. 2-2. Observed Line Shape Assuming
a Triangular Slit Function
2-3
-------
= J (l-IIN(u))A0(u)) ) d « (2-5)
W
Auo
P /• OUT N
= J (JL-I (u>)/lo(u>) ) du, (2-6)
Au)
IN OUT
where I (m) is the "true" intensity and I (tu) is the "apparent" intensity
as measured by the instrument, given by
OUT _ P IN / / /
Au/
When Equation (2-7) is introduced into Equation (2-6), one finds that the
normalized slit function can be integrated separately, giving unit value and
the equality shown in Equation (2-5) and Equation (2-6) holds.
In summary, the procedure for obtaining the laboratory data and the
subsequent interpretation of field data is as follows:
1. For each single pollutant line of interest, make three laboratory
measurements in the linear region of toe curve-of-growth by
changing u to obtain S = W/u, and three in the square-root region
2
to obtain a = W /4 Su.
2. For each single pollutant line of interest, calculate by computer
W as a function of u according to Equation (2-1) and electronically
plot these curves-of-growth. (The linear and square-root regions
are identical to the ones measured under Item 1, thus providing
a consistency check.)
3. From the atmospheric spectra measured in the field, obtain W of
the pollutant lines of interest according to Equation (2-6) and ;
obtain the desired quantity u from the curve-of-growth prepared
under Item 2.
2-4
-------
4. An alternate procedure would be to store tae values of S and at
for each single pollutant line in the computer. After W has been
measured in the field, input it together with the line identification
into the computer. The computer will iterate Equation (2-1) until
the proper u is found.
It is recommended that both procedures be followed. The operator in the
field has a ?et of curves-of-growth and can thus readily obtain u. The
computer analysis can be executed more accurately at a later date.
The term "pollutant lines of interest" refers to lines which can be
observed in the field without interference. This will eliminate making
laboratory measurements on a great number of pollutant lines. A decision of
what the "pollutant lines of interest" are can be made a) theoretically, based
upon existing spectra of the "clean" atmosphere and the present knowledge of
pollutant line position and b) experimentally, after initial field data have been
taken.
2.1.2 Multiple Lines with Lorentz Shape
If several lines contribute to the absorption in the spectral interval given by the
instrument slit function, band models must be used. The most useful band model
applicable to atmospheric pollutants is found to be tha statistical model. In that
case, the mean absorption is given by
1 -T = 1 - exp (-W/d) (2-8)
where Wis the mean equivalent width of the lines in this interval and d is
the mean line spacing. The following expressions exist for this ratio:
Equally intense lines
T=-^xe-X[lo(x)+Il(X)] (2-9)
2-5
-------
Exponential distribution of line intensities
W _S_ /-, Su
d d
1/S distribution of line intensities
(2-11)
All of these expressions have the same asymptotic expressions:
^ «4 u for ~ « 1 (2-12)
d d TTa '
In the transition region, the maximum difference in W/d is about 25%
between the equally intense lines and the 1/S distribution (seeFigure 2-3).
A standard notation for the parameters S/d and a/d is
S -1 -1
-r = k (cm atm )
•^ = a (non-dim.)
where k represents the average absorption coefficient and a is called
the fine-structure parameter.
2-6
-------
LADENBURG - REICH E.
(DELTA FUNCTION
DISTRIBUTION)
+ X - 1J (1/S DISTRIBUTION)
XA/1 + X/4: (EXPONENTIAL
DISTRIBUTION)
LOG X
Figure 2-3. Collision-Broadened Curves of Growth for Various Intensity Distributions
2-7
-------
The laboratory procedure is the same as previously outlined with one
exception. Instead of making measurements only in the linear and square-root
regions, measurements in the "transition region" are also required in order
to determine which of the three models (given by Equation 2-9), (2-10), or (2-11))
applies .
If an accuracy of ±12.5% in the transition region is deemed sufficient, the
expression given by Equation (2-10) is recommended. In that case
-1/2
-An t =ku fl+T1"")
V 4a /
or
2
Equation (2-14) is a linear relationship in u in that the observed quantity (-u/£ n t)
as a function of u is a straight line. The absorption coefficient k is determined by
the intercept of this line at u = 0 and the slope determines the product ak. These
relations were successfully employed in the determination of the curves-of -growth
*
of hot water vapor.
In summary, the procedure for obtaining the laboratory data and the sub-
sequent interpretation of field data is as follows:
1. For each wavenumber interval in which a pollutant has a number
of lines which are not interfered by lines from other molecules,
make laboratory measurements of the curves-of -growth.
2. From the atmospheric spectra measured in the field, obtain
t (<«) = I -I(u>)A0 ). DetermiD
the plots of the curves-of-growth.
t (
-------
3. Using a computer, determine which of the expressions for
the different line intensity distributions (Equations (2-9), (2-10),
or (2-11)) is the most appropriate one. Use that expression to
reduce the curves-of-growth in terms of k (uo) and a(uo) and store
in memory. The values for t (u)) measured in the field are then
readily reduced to u.
Again, as in the single line case, the decision of which of the spectral data
are of interest can be made both theoretically uid/or experimentally.
2.1.3 Laboratory Procedures
The maximum amount of data points which could be taken in the laboratory is
immense. One set of measurements could be taken at room temperature, using
air (mixtare of N and O ) as foreign gas broadener with 1 atm total pressure.
Li Lt
Another set of measurements could be taken at elevated temperatures, using a
mixture of O , CO and HO as foreign gas broadener. Considering that the
2i £ 2
diatomic and polyatomic pollutants of interest may amount to 20, that these pollutants
liave single or multiple infrared bands of about 200 cm , that at least five data
points are to be taken for each resolution element of 1 cm , and that at least six
points along the curve-of-growth curve are to be taken, it is easy to see that the
total number of data points could run into several hundred thousand. Thus, it is
Imperative to limit the measurements to spectral intervals which are most useful
in field work. As mentioned above, the "spectral intervals of interest" must be
determined before a program of laboratory studies is initiated. This determination
can be based on the theoretical knowledge of the relative location of pollutant lines
with respect to the "clean" atmospheric lines and/or based on the results of field
data taken by the ROSE system.
As outlined above, curve-of growth measurements are made by changing the
optical thickness u at constant total pressure p . Since u is given by the product
of cell length i and partial pressure p = C p , where C is the fractional concen-
tration and p is the total pressure, a change in either 4 and/or p is sufficient.
t p
2-9
-------
However, it must be remembered that a change in p without a simultaneous
change in C is not sufficient. In other words, a change in p alone does not result
^"^"" f
in independent values of transmission. This mistake has been made quite often
in the past, where researchers prepared a mixture of gases in a mixing tank and
added this mixture to the absorption cell at various total pressures, assuming
that they had thus measured different points along the curve of growth. The fallacy
of this procedure is evident if we rewrite Equation (2-10) in the following way,
remembering that a = a p :
1/2
W/d - in t .„./•, _ , _.
-L — = - = kCi (1 + - - ) (2-15)
pt Pt ^ 4ao }
It is clearly seen that if C and JL are kept constant and only p is changed, the
t
right-hand side of Equation (2-15) is a constant and neither k nor a can be
determined separately.
It is usually easier and more accurate to change the cell length JL rather
than the concentration C. This can be done in cells which have adjustable
multiple paths, such as in a White-cell arrangement.
The optical thickness to be measured in the laboratory must cover the range
of the optical thickness expected in the field. Thus, for an expected concentration
of 10 PPM of a given pollutant to be observed over a pathlength of 4 km, the
optical thickness is 4 cm-atm. If it would be possible to use a partial pressure of
I atm for the pollutant, a cell length of at least 4 cm must be used. However,
in this case, the measurements would be made under the condition of self -broadening.
The results would thus not be applicable to the atmospheric case, in which air-
broadening is dominant. This is especially true for polar molecules, which may
have a self -broadening coefficient (ratio of self -broadened half -width to nitrogen-
broadened half-width) of the order of 10. Thus to keep the actual half-width
within, say, 1% of the self-broadened value, the maximum concentration should
2-10
-------
not exceed 1000 PPM. At this concentration and 1 atm pressure, a pathlength of
40 m is required to produce an optical thickness of 4 atm-cm.
2.2 SPECTRA OF POLLUTANTS AND ABSORPTION COEFFICIENTS
The "pollutants of interest" will, to some extent, depend on the pollution monitoring
task at hand. However, there are a number of pollutants of sufficiently common
interest to be considered here along with the major infrared active normal atmos-
pheric constituents, CO and HO.
Li Li
Band strenghts, band averaged absorption coefficients and fine structure
*
parameters are given in Table A.I and band model parameters are given in
* _i _i
Tables A.2 through A. 10 at 5 cm and 20 cm intervals.
**
Band strengths and fine structure parameters are given in Table A. 11 and
**
absorption coefficients are given in Tables A. 12 through A. 18 for various
cm intervals.
For details of the generation of these tables the original references given
below should be consulted. Spectra of these and other pollutants are given in
the open literature which are too extensive for inclusion here.
* Air Pollution Measurements by Satellites, C. B. Ludwig, M. Griggs,
W. Malkmus and R. Bartle, NASA (in press).
** Study of Air Pollution Detection by Remote Sensors, C. B. Ludwig, R. Bartle,
and M. Griggs, NASA CR 1380 (July 1969).
2-11
-------
3
PERFORMANCE OF THE ROSE SYSTEM
3.1 GENERAL
The performance of the ROSE system has been evaluated in terms of the operation
of the mechanical, optical and electronic components during the field tests. In
view of the complexity of the system and, at times, unusually adverse field con-
ditions it is considered that the system performed well.
The performance of the mechanical components of the ROSE system was
excellent during the field tests which included set up at six locations and two moves
from Pomona to San Diego by truck. The system proved to be rugged and was
conveniently moved into position. One vendor-supplied mechanical component,
the reference chopper motor bearings, failed during the emission tests. These
bearings have been replaced and spares obtained for use if future trouble should
develop.
The optical components of the system also performed well during the field
tests. The mountings appeared to be stable and, even with severely hot ambient
temperatures, no alignment problems developed. The possibility of having to shield
the telescopes from direct sunlight proved to be unnecessary.
There was some difficulty with isolated electronic components believed to
have been caused, in part by ambient temperatures above the design values and,
in part, by non-standard voltages of portable power supplies. In particular, the
reference chopper drive has been modified to make its operation more reliable.
A number of wiring breaks were located and have been corrected. Subsequent to
the field tests the digital coupler was modified to provide more stable operation
of the digital system.
In the following sections the performance is described in terms of system
noise, scintillation noise and signal strength.
3-1
-------
3.2 SYSTEM NOISE
System noise is defined as the noise with no optical input to the system; i.e., with
the optical path blocked off. Ideally, the system noise equals the detector
noise but in practice other electrical components may contribute to the system
noise. However, it is useful to express the system noise in terms of the
detector noise.
The theoretical system noise is based on detector noise which may be
calculated from
fAA
Vn
theo. *
-2
where A = detector area - (cm )
2
= 0.004 cm nominally
Af = system bandwidth - (Hz)
= 1/4 (system time constant - sec)
R = detector responsivity - (volts rms/watt rms)
D = detector specific detectivity - (cm Hz 1/2/watt)
4
The value of R/D* at X , the wavelength of peak response, is 12 x 10 /
10 -5 m , 1/2 *
0.95 x 10 ~ 1.26 x 10 volts/cm Hz . R and D both vary with wavelength
to some extent but the ratio does not.
Using the above values the theoretical system noise at the detector is
f» . f* ,
Vn = 0.8 X 10* VAf = 0.8 X 10~ / V4r
theo.
which results in a noise voltage of about 0.4 p, volts rms for a bandwidth of
0.25 Hz (time constant = 1 sec.)
Vn//Af", = 0.8 X 10~6 rms volts / ^/Hz
theo
3-2
-------
The observed system noise is obtained by reading the peak to peak noise
voltage on the recorder trace and dividing this voltage by 5 (to obtain the approximate
rms value) and by the gain between the detector and the recorder and V Af to obtain
the equivalent rms noise voltage per VHz at the detector. This system noise may
then be compared to the detector noise from the above equation for the particular
bandwidth (or time constant) of the test.
Analysis of the system noise data gave the following results averaged for
each test.
Vn /V5z (at detector)
-6 . i—
One mile test 1.0x10 rms volts A/Hz
Emission tests 0.6-0.7 " " "
Two mile tests 1.1 " "
Refinery tests 0.5 " " "
Thus the blocked off system noise is within a factor of about 1.5 of the
detector noise. Considering the difficulty of noise measurement, especially
with small trace deflections, it is considered that the above deviations from
the detector noise value of 0.8 x 10 /VHz are not significant.
It should be noted that occasionally the noise exceeded the average values
given above by as much as a factor of three. It is believed that these transient
increases were mainly a result of electrical interference from outside the ROSE
system.
Previously reported laboratory values of system noise (or zero line noise)
corresponds to 0.5 to 0. 8 x 10 rms volts /'VHz. The system noise for the San
_/> .
Diego field tests corresponds to about 1.2 to 1. 6 X 10 rms volts /VHz.
3.3 SCINTILLATION NOISE
Scintillation noise results from variable atmospheric refraction between the source
and the receiver and is observed as a variation in the signal level. In the following
analysis the observed signal variation is divided by the total gain, G, to normalize
3-3
-------
different gains used on different runs and is multiplied by V 4 T, where T is the
system time constant, to normalize the different time constants used. This
process results in a signal noise which is referred to the detector and to unit
bandwidth of
V A/Hz =
-------
except for the early large values of signal noise, did give less noise than that
observed on the one mile tests.
In the one-quarter mile tests at the refinery the signal noise ranged from 2
to 30 M,volts /VHz with the average being about 11 jj, volts /VHz. Again no clear
cut variation with time of day was observed.
For the emission tests the signal noise varied from 0.4 to 1.6 p,volts /VHz
which is within a factor of about two of the system noise. In these tests, of course,
there is not the precise alignment requirement as there is for the transmissi on
mode of operation.
A factor contributing to the large scintillation noise during the one mile tests
was the use of a slit height equal to the theoretical image height, h , which was
2.3 mm for the one mile tests. Figure 3-1 shows the source and reference channel
signals as a function of slit height normalized to unity for the maximum mean value
of the source signal. The flattening of the source signal simply represents the
fact that the source image is finite. It is readily apparent that the variation of the
source signal for a given slit height depends on the value of the slit height and is
a maximum near h . This is interpreted to be a result of vertical movement of
o
the source image.
3.4 SIGNAL STRENGTH
The system transmission is fundamental to the calculation of the signal strength.
Calculations have been made for the transmission mode of operation and the results
are shown on Figure 3-2. Reflectivities for aluminum were estimated from measured
values for the two telescopes which were different; thus the two curves shown repre-
sent this difference. For the emission mode of operation the transmission would
2
be more than that shown by a factor (1/p.) .
A
Figure 3-2 shows the transmission calculated from all known causes except
aberration. The transmission from aberrations for the transmission mode of opera-
tion is shown on Figure 3-3 for a range of 0.4 km. For the emission mode of
3-5
-------
101 L/mm 7 y, Filter
3.0
2.0
1.0
= 726; v ~ 1000 cm
S H 300 n xh
-1
Reference
Relative
Voltage
h =2.3 mm
o
4 6
Slit height, h, mm
10
Figure 3-1. Relative Signals from Source and Reference
Channels - One Mile Path
3-6
-------
Transmission Mode of Operation
4
T = T X p
t c a
T 83 I3 X T. X
C I f
(r & Common Path Transmission)
gr
T s IRTRAN II (Entrance slit, field lens, det. cass. field lens, dewar window)
T. a Filter transmission
p »Aluminum reflectivity
3
e • Grating efficiency
5000 3000
I I I
.20
.15
4
.10
,05 -
cm
-1
1000 900
I I
800
I
TRANSMISSION LOSS
FROM ABERRATIONS
NEGLECTED.
8 10
Wavelength-Microns
12
14
Figure 3-2. Calculated System Transmission
3-7
-------
l.Q
0.9
O.i -
Transmission
resulting from
aberrations
0.7
o.d
Geometrical
image size
"exceeds detector
size
Slit Widii - Microns
Transmission Mode of Operation Range = 0.4 km
Loss from Geometrical Image Size Exceeding Detection
Size Included.
2000
Figure 3-3. Calculated Transmission Resulting from Aberrations
3-8
-------
operation there should be little or no transmission loss from aberrations for a
uniform source field.
The signal strength in the laboratory were found to be 0.55 to 0.67 of that
calculated using the above calculated transmissions (see EPA-R2-72-052 or
GDCA-DBE-72-001 page 6-3). The difference between these ratios and unity
represent unknown losses which may come from deviations from the calculated
values of reflectivity or tranmission from aberrations or from misalignment,
scattering or diffraction effects.
Signal strengths in the San Diego field tests agreed with the laboratory data
(ibid page 6-7).
For the one mile tests in Pomona the signal levels were considerably below
the theoretical values as a consequence of the receiver being at the edge of the
beam. The signal levels as well as the signal noise showed wide variations.
The signal strength for the two mile tests at v = 2500 cm showed a range
from 0.48 to 0.59 of the theoretical value with an average of 0.54. At M = 1000 cm
the signal strength ranged from 0.37 to 0.53 of the theoretical value the average
being about 0.45. These values probably are lower than those observed in San
Diego at one-quarter mile because of the lower transmission of the two mile path.
For the refinery tests the signal strength was from 0.30 to 0.40 of the
theoretical value. Since the path length was about one quarter mile it is suspected
that the detector toroid was displaced from the optimum position during transport
of the system. It is recommended that the detector toroid adjustment be checked
after each move before testing (at least until experience indicates otherwise).
3-9
-------
LONG PATH TRANSMISSION MEASUREMENTS
4.1 ONE-MILE MEASUREMENT
A description of the location for these tests is given in Appendix I.
As previously mentioned, these tests were influenced by scintillation noise.
This noise was greater than expected. It was subsequently traced to the unfavorable
position of the receiver telescope near the edge of the beam. In addition, a con-
siderable portion of the line-of-sight was over a hot blacktop surface and one
edge of the beam was close to metal lamp posts.
A typical spectrum of transmitted intensity (I) from 7.5 to 12.5p, is shown
in Figure 4-1. It was recorded with a slit width of 1500 p,, which corresponds to
about 5 cm at 1000 cm , and a time constant of 0.1 sec. The main features
_^
are the strong absorption lines due to HO between 1100 and 1330 cm , the band
-1
absorption of O between 1000 and 1100 cm , and the weak absorption lines due to
3 -i
HO between 800 and 1000 cm . The instrument noise in this trace is very small,
-1
as can be seen at 1320 cm , where the atmospheric transmission is reduced to
zero due to strong water absorption and thus acts as a shutter. The fluctuating
signal between the identified water lines notably for DJ< 1000 cm , arises from
the atmospheric scintillation, as can be seen from the record in Figure 4-2, which
was taken at 1000 cm for a two-minute time interval.
By increasing the spectral resolution to 1 cm , many of the strong center
lines between 1100 and 1330 cm become better defined (see Figure 4-3). Line
identification numbers were taken from Air Force Surveys in Geophysics No. 142.
The increase in amplification by a factor of 10 also increases the noise in the
signal, as can be seen in the record of Figure 4-4.
A typical spectrum, in which the atmospheric transmission ([/I ) is recorded,
is shown in Figure 4-5. The spectral slit width is the same as used in the recording
of Figure 1, but the amplification is reduced by a factor of 2.5. The spike shown at
4-1
-------
r~
.. .......
-— ...... ---.,-^-i
o
_o
CO
o
_o
o
-o
o
.0
o
o
-o
o
-o
00
s
o
in
II
-------
T —
•'•--' J 7 I
fi
o
s
o
o
o
o
r-l
o
4
2
S,
4-3
-------
II
o
T-l
>
oj
A
o
fl
&
d.
co
i-i
3
t-
co
-------
-,. .-
CO
CO
o
CO
i!
I • .
: ' :
1. ' ' '
1
1 .
1 '
, f ' .
-H-4
' , > 1
i , •
. :.! , i '
1 • r
J.^
t i •
1 ' '
.In:
Hi:
III:
J.i . j.
-i J V. *
r
•*-»
8-
aS
i §
rt co
CO 0
! ; : 1
— - —
1 : i
.
.-;-
j . :
1
-cl-i.
' 1
1' , t
- -':-
-.1:1-
• .
i > : .
iiji
' ; * •
:.r.ti
H
'§
O
o c
S i
II *
i
• • 1
' : i
'—
'
, . '
J- U-U
,
,
T' ~"7
-•+— 1-
-;-- rr
'• \ '
1 ' ;
1 ' ' I
j J..LL
1
1
5
!
1 " "
— •
"
_...:.
. — -y—
• i . :
1
•-r""""-T
' i T'
.' 1
"''T'TT'
1
, . i
; l ' i
.//.
I.CT
• -, ,
! .
o g
™ w
n cq
1 *'(
: ..L-
_„-
1
_.__
' ' i
. -- "|
1 j •
! ! '
' i , :
-:!ii
j. ;.
-- -'
•
-i : . -
_. -^
'
I ' '
• i ' i
4_, ..
! ! ;
-1.^.
U. - J A .
i
.... -
/ .
=
• ' •
T°~ "
1
. t
'
• ' • 1
, , ]
i ~ r
: : ; : j
F.L : ,1
1':
1
'\ir '
1 ' ' i
•„ '-
«^J]
|-
^T
"^
^ i
fc
P
J^" 1
r~3t_ — —'
>
•••n —
i i ,
ill!
:.j i i.
^ . - .
_j
! _j
_ ,-^ -'
i
i
__
' "*
-_
,
-r1-.
i
<
Ll!'
i
-i L.L.I
1
t
i
L~
f
r; --
; , ,
£ ;
^ '
£
ct
i;:'!
" ' "j~
t
! ' ' ,
_j !_
! '. '' .
'
i ' ;
, ' : i
i • ' ;
, i
~r— ~TT~
» , [
i.;L
i
1
-
-
i
i
rH
rH
II
O
i
i "
i *
J
^
'%
rH
'a
o
o
o
o
•3
0
M
I t
: 2
i PH
•
i
.
4-5
-------
U5
II
I
Pi
J
U
co
5
UJ
-
g.
-------
1010 cm is identified as VIB ('Vehicle in the beam"), an all too frequent occurrence
in the measurements conducted here.
The only pollutant identified unambiguously is ozone. The quantitative analysis
proceeds as follows. As has been shown in Section 2.2, the band absorption can be
represented by the statistical band model. Referring to Figure 4-3, the I level is
-1 ° -1
assumed to be the connecting line between the levels at ~ 1000 cm and 1080 cm .
At 1055 cm I = 14.5 and I = 12, thus IA =0.83. Using the absorption coefficient
and fine structure parameter from Section 2.2, the partial pressure of ozone is
-6
calculated to be 0.12 x 10 atm, which corresponds to 12 pphm. This value agrees
well with the hourly average value of 11 pphm recorded by the Pomona APCD station
at 3 ppm (see Table I of the Appendix).
A typical spectrum of transmitted intensity from about 3 to 5 p. is shown in
Figure 4-6, which was recorded with a slit width of 700^ (A v at 4 p, or 2500 cm
-1
equals 6 cm ) and a time constant of 0.1 sec. The main features are the strong
water absorption between 3400 and 2900 cm , the absorption due to HDO near
2700 cm , the absorption due to CO at 4.3|i, due to NO at 2230 cm and due
-I2 2
to HO between 2200 and 2000 cm . The amplification is the same as used in
2 -i
Figure 4-1. The signal at 2500 cm shown in Figure 4-7 indicates again large
fluctuations due to scintillation.
The corresponding record of atmospheric transmission (I/I ) is shown in
Figure 4-8. The unusual feature in the CO absorption band between 2300 and 2400
-1 2
cm originates from the absorption in the I trace, which cannot be eliminated since
the instrument is not flushed.
4-7
-------
! 3
o —•—
;T _ ___
I
CO
II
I
PH
I
I
2
CO
-------
._T
70206
Same as
_. . _
""
S/N-3
PP
Figure 4-7. Noise at 250C cm with One Mile Path, Ay, = 6 cm"
4-9
-------
' '
1 <
1
— • .-
• ' '
, i . .
.--i-|»;T
;l; _
1
: •
'>''.'
*-rf-
L _ '. ...:_
'; • i
i. _'.:
[_ 1 i •
Q._;ii
1 • i
-T- 1-T7-
r_
.
r-
1."
s
o
i
o
o
o
o
o
• in
. 01
o
o
o
s
o
to
II
2
co
00
8
J
-------
During the week in which the one-mile tests were performed, atmos-
pheric turbidity measurements were made with the Volz sun photometer. *
This instrument consists of a small aperture which was aligned so that
energy from the sun fell on a filter-detector combination; the detector output
was read from a microammeter. The secant of the sun zenith angle associated
with the measurement was determined by means of an auxiliary scale and a level. The
air mass of the atmosphere was assumed to be equal to the secant of the sun zenith angle
(valid for 0° < M < 80°).
The aerosol optical thickness, T , was determined from the fol-
aer
lowing relation.
I = I exp (-(T + r . )M)/f
o aer atm
where
I = meter reading of the measurement
I = meter reading with zero air mass ( = 51 with the present
instrument, S/N G120).
M = sec 9 (9 = sun zenith angle)
T - Rayleigh scattering and ozone optical thickness (estimated)
atm
T = aerosol optical thickness
aer
f a a distance correction factor
A series of measurements were made in San Diego which show that
the zero air mass deflection (extrapolated to zero air mass) was still equal to
the original calibration value of 51 and also show typical variations of the
optical thickness during the day, see Figure 4-9.
The results of the photometer measurements made during these field
tests are shown on Figure 4-10. For comparison, Elterman's data for clean
standard continental air and data taken in San Diego are also shown. Most
of the data indicate an optical thickness about twice to three times Elterman's
value.
Attempts to correlate T with signal strength were unsuccessful;
*F. Volz, Arch.Met. Geophys. Biokl. B, 10 p 100 (1959).
4-11
-------
51
10
3 4
AIR MASS, M
0.3
aer
0.2
o.i
i
I
O 4/1/71
0 4/2/71
A 4/3/71
O 4/4/71
10 11 12 13 14 15
TIME(PST)
16 17
IS
19
20
21
Figure 4-9. Volz Photometer Calibration and Typical Aerosol Optical
Thickness Variation in San Diego* California
4-12
-------
01
-------
perhaps the optical thickness value measured with the Volz photometer
(which is an altitude integral) does not necessarily represent the horizontal
optical thickness required for transmission calculations.
4., 2 TWO-MILE MEASUREMENT
A description of the location for these tests is given in Section 1. 2 of
the Appendix.
In these tests, the receiver telescope was placed in the center of the
source beam. A reduction in scintillation noise was expected. However, the
fluctuations are again substantial, as shown in Fig. 4-11. The record taken
at 10 a. m. , shows a signal-to-pp noise ratio of about 9. The time constant
was 1 sec, i.e. , ten times longer than used in the record of Fig. 4-7, in
which a signal-to-pp noise ratio of about 3 is observed. Thus, the apparent
"improvement" results mainly from the longer time constant. On the other
hand, the atmospheric conditions for the two mile tests were less favorable.
A comparison of the relevant parameters is given in Table I.
Table I
1 Mi 2 Mi
Date 2 July 1971 29 July 1971
Time 9 a. m. 10 a. m.
Air Temperature 72° F 90° F
Relative Humidity 59% 45%
Evidence of the greater stability of the night time atmosphere is clearly
visible in the record taken at 9:15 p. m. (Fig. 4-11) in which the signal-to-
pp noise ratio is about 26.
A typical spectrum of the transmitted intensity in the spectral range
from 8 to 13|j, is shown in Fig. 4-12. A time constant of 1 sec was used,
which was subsequently determined to be too long for the scan speed employed.
Thus, some of the fine structure in the spectrum was losl. Nevertheless, the
4-14
-------
2 Minutes
72904
End
s
o
o
o
-.5V
Day
10:30 AM
S/N~9
PP
• r, j ~ ^ zi—
?• • ' T , ' .
72816
; End
Night
9:15 PM
S/N ~26
PP
Figure 4-11. Ndse at 2500 ctn"1 with Two Mile Path, A v. =3 cm"1
4
4-15
-------
3
a
•»
a
•*
a
-------
strong water lines throughout the entire spectral range are very well
resolved. This becomes especially important in the ozone region, in which
the many water lines, if not resolved, would introduce large errors causing
an apparent reduction of the ozone concentration. Based upon our present
knowledge, the band contour of ozone has been sketched in. Reducing the data
obtained during the day of July 28, results in a plot of ozone concentration
versus time (see Fig. 4-13). These data are compared with the hourly
averaged values taken by the Pomona Station of the Los Angeles APCS (see
Table V of the Appendix). The agreement is good. The time shift of about
1 hour between the two traces may be significant, indicating an 63 build-up
near the ground to occur faster than in elevated layers.
A typical spectrum of the transmitted" intensity in the spectral range from
3 to 5 p, is shown in Fig. 4-14, recorded with a time constant of 1 sec. For
the spectral region from 1900 to 2300 cm the amplification was higher by a
factor of 2 than for the rest of the spectrum. A spectrum produced by the
digital system for the CO absorption band is shown on Figure 4-15. The
magnetic tape was reformatted and the results plotted on the SC4020 with the
frames placed side by side.
4-17
-------
W
s
H
t
>-3
00
CM
I
0)
S
•r-t
H
ca
§
-------
t
I
s
I
(M
-------
a
o
(O
r-I
I
(
V
o
a
h-
o
o
10
i-Z(-UZ •)-.»
^r
n
o
a
g
m
a
o
s
o
O
o
I
14
-U
8
•8
Sf
4-20
-------
8
8
•a
§
.3
6 g
s a
a
i
a
a
r.
o
o
10
o
o
n
a
o
M
4-21
-------
o
o
8
S
4-22
-------
5
a
•#
2
c«
I
>
4-23
-------
5
EMISSION MEASUREMENTS
For each stack, at least two measurements are necessary. One spectrum is
taken from the plume, the other one adjacent to the plume to obtain the sky
background. The two spectra are then differenced. This is necessary even
in the case where the plume fills the field of view of the instrument. For
the quantitative determination of the pollutant concentration in the stack
plume, the absolute temperature of the plume must be known.
5. 1 STEEL MILL
The location and absorption of the stacks of the steel mill selected for these
tests are described in Section 2. 1 of Appendix L A typical spectrum of
the sky background from 7 - 13|i is shown in Fig. 5-1 and that of the plume
emission in Fig. 5-2. In both cases, the spectral resolution was 0. 5 to 4. 5
cm . The reference source was at 106°F. The zero line (no energy differ-
ence between target and reference source) is in the center of the recording
paper. When the target is colder than the reference source, the deflection
is upward; if it is warmer the deflection is downward. In comparison, the
two spectra appear to be very similar, except for one obvious region around
7. 6^, where the plume emission shows a rather strong emission. However,
even in the regions between 8 and 10. 5y,, there are differences which become
apparent when the differences between the two spectra are plotted; in Fig.
5-3 the differences are shown. An attempt has been made to identify the
species responsible for these bands. Normally, this differencing process
is to be done by a computer. However, we experienced difficulties with the
electronics and no digital records were taken. The failure of the electronics
was attributed to the heat in the equipment, which was located in an unvented
van, whose temperature at times reached 112°F.
The most prominent spectral features of the sky emission curve
5-1
-------
u i
1 1
a 5
o o
I! 1
a o s
"
J, a 2
s a a
5-2
-------
shown in Figure 5.1 (71302) are the emission by stratospheric and low-level ozone near
1000 cm and the HO spectrum.
An estimate of the effective sky radiance may be made as shown below
using the final equation from Section 3.3.1 - Emission Mode of Operation,
of the final report. (EPA-R2-72-052 or GDC-DBE-72-001 page 3-22).
Using N ° for the van temperature of 106°F (314° K) and other values
appropriate to the test at X = lOp.,
—
Ns
(—^) (12x104)(. 25)(1. 16x10 3)(0. 021){10 5)(0. 58)
L ]
12.42xlO-4(. 81)
-4 2
= 10.5x10 u)/cm sterM*
for which the equivalent blackbody temperature is
T = 302°K.
s
The effective temperature difference is thus about 12° C between the van and
the atmosphere at an elevation angle of about 5 . The value of sky radiance
is 10 to 20% higher than the values cited in the Handbook of Military Infrared
Technology (Fig. 5-5 for the clear nighttime sky at Cocoa Beach, Fla. , and
Fig. 5-8 for a sky with cirrus clouds).
The difference is perhaps due in part to the higher ambient tempera-
tures in Los Angeles and, in part, to the greater pollution particularly at
lower levels.
The plume emission shown on Figure 5-2 at first glance looks much the
same as the sky background but taking differences as shown in Figure 5-3
reveals important features of the spectrum. A number of candidate pollu-
tants are shown at their respective spectral locations.
A typical example of the plume emission from a steel mill stack from
3 to 5)i is shown in Figure 5-4. The sky background scan and stack wall
5-3
-------
8 §•.
us
£
5-4
-------
scan were similar to Figure 5-4 except there was no spectral features in
the CO and CO2 regions. The spectral slit width at 4p, was about A v =
9 cm . The strong emission from hot CO and CO is apparent in Figure
5-4; however, the greater portion of the hot CO band is absorbed by the
CO in the atmosphere.
5. 2 POWER STATION
The location and description of the studies of a power generating station
are described in Section 2. 2 of Appendix I. Spectra of both Stack #3
and #1 were taken. A typical spectrum of the emission of Stack #3 in the 7
to 13|j, region is shown in Fig. 5-5 along with the sky background and the
emission from the stack wall. In these traces, the colder temperatures are
recorded downward. The difference between plume and sky emission is
shown in Fig. 5-6. Besides SO , a number of hydrocarbons are possible
Lt
contributors to this spectrum as indicated in Fig. 5-6.
A typical spectrum iti the 3 - 5|j, region is shown in Fig. 5-7 whose
prominent feature is the emission from CO and the total lack of CO. An
indication of the small fluctuation in the signal is given in Fig. 5-8, which
was taken at the peak CO emission at 2475 cm . We believe that the
£»
absence of large-scale fluctuation is due to the fact that the field-of-view was
smaller than the plume but larger than the eddies in the turbulent plume
so that a spatial averaging took place.
The emission in the 8 - 13|j, region from Stack #1 is very similar to
the one from Stack #3, except for the absence of SO emission, which is in
&
agreement with the results obtained with the interferometer.
5. 3 HOSPITAL INCINERATOR
A brief description of hospital incinerator stack is given in Section 2. 3
of Appendix I. The spectra we observed show great variations both in
characteristics and strength. Records taken at one wavelength indicate
5-5
-------
Plume
71414
101 L/MM
1 n Filter
S=900(j, X 10 mm
3/2 RPM
.5 mm/sec
G=50 x 10 X 100
Q = 10 T • 1.0
Sky
71413
Stack Wall--
71416
1 1
600
_ ,.
1
800
1 . 1
1000 1200
-1
v - cm
I 1
1400
Figure 5-5. Plume, Sky and Stack Scans - Power Station Stack #3, A\>10 a 3 cm
5-6
-------
w
o
.3
s
f-l
-------
ORS
-IV
71409
" — +iv ^ ,
0
T
1800
. — j - <\l ; -
^
2000
' i I" ' i
£ . . • .
r
2200
- I <3N 1 - ',^ '
.-^ r^ ^
T ^ T
2400 0600 2800
-1
- cm
Figure 5-7. Plume Scan, 3 to 5 p, Power Plant Stack #3
5-8
-------
-.r.
71408
***&Sv^
i .jj; \y-\ ;-•;•-• -t-
i-^FT*''"' '
±.Q.-^
:m-
"_ L_1 I
_" ^___^ " ^ 2475.5 CM"1
_ VARIATION IN PLUME EMISSION
Figure 5-8. Variations in Plume Emission - Power Plant Stack
5-9
-------
that the variations originate in the plume, an apparent result from the
different loadings of the incinerator. A typical trace taken at 2390 cm
over a period of several minutes is shown in Fig. 5-9.
At another time, the plume appeared visibly rather steady. A comparison
of two spectra recorded between 3 and 5p in a time interval of 2 minutes is
shown in Fig. 5-10. The emission between 2600 and 2900 cm has changed
drastically. The upper curve is indicative of a black body emission from
soot particles, while the lower one is indicative of gaseous emission. This
is the wavelength region, where H S and HGt are emitting.
5-10
-------
I Blockxl 1
'— with *t
Ey«pl»c«
< :
•i •.
•Udltooto
».MV/DlT
L—I
Figure 5-9. Variations of Hospital Incinerator
Emission v = 2390 cm~
5-11
-------
JO
71513 And
Repeat After
HC1
i i I
2000
I
2500
v - cm
-1
3000
3500
Figure 5-10. Comparison of Two Scans, 3 to 5 n, Hospital Incinerator
5-12
-------
TRANSMISSION MEASUREMENTS OF EXTENDED SOURCES
Pollutant concentrations in the vicinity of a refinery in Southern California were
measured over a pathlength of approximately one quarter mile. Because of prevailing
winds from the west, a north-south path just inside the eastern boundary of the
refinery was chosen. Most of the expected emission was located at the north end
of the path near the source; the receiver was located at the south end of the path
in a relatively clear area. Although the prevailing wind was from the west, the
wind direction varied and, at times, came from the southeast.
At two such times (i. e., with the wind from the southeast) runs were made
within about 40 minutes at two different slit widths. These runs clearly show
the effect of using a spectral slit width equal to or less than half the spectral line
* -1
spacing. The lines of interest here are the CO lines from about 2100 cm to
2190 cm which have a spacing which varies from about 4.2 to 3.3 cm . One run
with a spectral slit width at 2150 cm of 4.4 cm is shown at the top of Figure 6-1
and shows very little, if any, of theCO spectrum. The other run at the bottom of
Figure 6-1 with a spectral slit width of 1. 3 cm shows the CO spectrum quite
clearly. Since the wind was blowing toward flie refinery the difference was pre-
sumably not a result of a CO level change but was a result of setting the spectral
slit width less than half the spectral line spacing. An analysis indicates that the
CO concentration was about 2 ppm which is about what would be expected for the
ambient level.
Later in the afternoon there was a strong west wind. A CO spectrum
taken at that time is shown on Figure 6-2 with instrument settings identical
*
GDCA-DBE-72-001 page 5-16.
6-1
-------
— i .- -j^- -4 ,_
— ~-\ ~: -: hz"::~. ~ ~L:~r-r r
^OEEFTF^
. 1 .:.i._~_T
•-•'- 02102
1-^240 L/MM
~r_ 3 p, Filter
S= 700p, x 6.6 mm
;:-_-_-._-. 6 RPM 2.5mm/sec
~ ~ G = 100 x 1 x 10
-
4.4 cm —
0.4 VA>iv
I
_i _j
2000 cm !.: r.iT
r^g.-r , i^El
-—:-zrrr;r-:rmr 2500
SANBORN
EEEt"— 3 n Filter
— Sc2QQ(t x 6.6 mm
r.z:4r-Z:i^/4 RPM
0.5 mm/sec
100 x 10 x 10
— A v a 1.3 cm
rz;; _"~.o. 1G V/Div
Figure 6-1. Comparison of CO Spectra with A v » 4.4 cm" and 1.3 cm"
6-2
-------
s
o
co
i-T
D
O
I
•a
s
|
o
a
co
O
O
oj
CO
-------
to that at the bottom of Figure 6-1 (except . OSf/Olj/). Comparing the
bottom of Figure 6-1 with a southeast wind blowing over a freeway and
Figure 6-2 with a west wind blowing across the refinery, the CO level from
the refinery is no more, perhaps less, than that from the freeway.
A complete scan with a strong west wind (taken about 15 minutes
before Figure 6-2) is shown in Figure 6-3. Although the slit is narrower a
longer time constant was used which suppresses line spectra somewhat.
Even so, there does not appear to be much CO present. A smell of
hydrogen sulfide, H S, was noted at the source end of the path at the time the
o
run shown in Figure 6™3 was taken. H S absorption would be expected at
1100-1400 cm , 3800-4000 cm" and at 2700-2800 cm" . No significant
absorption was observed probably as a result of only a small portion of the
total path containing H S.
Figure 6-4 shows a comparison of portions of two runs one with a south-
east wind and one with a west wind. The significant difference is two sharp
absorptions at 2960 and 2970 cm with the west wind from the refinery.
There are a large number of hydrocarbons which have absorption at about
3000 cm so that positive identification could not be made with the comparison
spectra available (R. H. Pierson et al Analytical Chemistry, 28, 1218
(1956)). The spectra were sketched because the absorption features of in-
terest were faint on the original record.
A scan in the 7 to 13^ region is shown on Figure 6-5; a strong west
wind was blowing (from the refinery) when this run was made. No evidence
of SO absorption in the 1100-1200 cm region or other pollutants was
observed on this spectrum.
6-4
-------
:'^™ n '-M
^ffljVP ' _J
-f — 2000 ! [ Jr.
r -..' * =
. i _ "_-._\ - r ^,. _-^__
•~.—.--". -.IT \-----r.. ._:•._
~- :\- — ~r T"-:^ , — -~ -
; : — : — ; . : j
^ir:^^_ ^_^^::-:::
J : r f ,-*, ^.--, ', ..-. ' — i_l — . — ; i
Figure 6-3. 3 to 5 p, Scan at Refinery, A \> = 1.4 cm
-1
6-5
-------
'••' vt i *<&'-%
240 L/MM fyH/Wl
3^ Filter 1 I/I
o //I T> DA T I
3/4 K-PJYL ^
0.5 mm/sec V
T a 0. 1 sec F
AvQ ,= 2.4 to
o • o _ _ "J.
2.7 cm
02108 S. E. Wind S= 175 p,
02113 W. Wind &= 200 n
2800
/ ' ' -
* A
1 ! -^
M \ y
ml *• :? « 5 «
'1 i
; jj
I S 2*
* *k ^ ^
iff f X ^
< * ZERO
32
f /
12
Figure 6-4. Comparison of Spectra near 3000 cm
-1
6-6
-------
6-7
o
•o
o
I
o
i §
S
o
in
<£>
•
o
tl
o
-------
7
SUPPORTING LABORATORY DATA
7.1 DETECTOR NOISE MEASUREMENTS
During and after the period of the field tests, measurements were made in
the laboratory to define more precisely the detector noise and detector plus
system noise. Measurements of detector noise were made directly at the
preamplifier output with a waveanalyzer. Measurements of the indicated
noise of the detector plus system for both the analog and digital systems.
Finally, the results of the various measurements are compared.
7. 1. 1 Wave Analyzer Measurements
The noise of the cold detector with the beam blocked at the entrance slit
was measured at the preamplifier output by means of a Hewlitt-Packard
302A waveanalyzer (Af = 6 Hz) with an integrator attached to the output.
The measured RMS noise voltage, V , was divided by the nominal preamplifier
K
gain to obtain the RMS noise at the detector, V . The noise at the detector
d
was divided by \j6 to obtain V / yHz, the RMS voltage at the detector per
root-Hertz. Data were taken for a period of about 12 integrator time
constants and averaged to obtain the results shown on Figure 7-1. During
these tests the source chopper was on and the reference chopper was off.
The wave analyzer output is a current source of 1 ma full scale arid use
of a 1 K load resistor converts this to a 1 volt full scale voltage source.
An RC filter of 1 megohm and 5 ^fd was used as an integrator (r - 5 seconds).
The output was read on a CIMRON digital voltmeter (Z. = 1000M on 1 V
in
range) which negligibly loads the filter.
Figure 7-1 shows clearly the lower noise resulting from use of the
highest preamplifier gain. The lowest noise of about 0.39 p,v/^Hz is some
what less than that measured by the detector manufacturer of 0. 44 (j,v/ /Hz).
The highest noise for the preamplifier plus detector is 0. 72 y,v/ yHz which
7-1
-------
H-P
302A
1 v^">-
11K
n v»nrt I
li DVM
fd L? JQOpJL
O.i
0.4
Microvolts
0.2
2/5/72
Source Chopper on
f -~ 550 hz
Beam blocked at entr. allt
10
100
Prearap. Gain - Gr
lOOO
Figure 7-1. Wave Analyzer Measurements of Detector Noise
7-2
-------
is slightly less than that measured previously (0. 80 M,V/ /Hz). The later
measurements are considered more precise than the earlier value because of
an improved integrator.
Wave analyzer noise measurements were also made with the preamplifier
input open (internal 10K only) and with a 12. 7K 1% precision resistor at the
preamplifier input to simulate the detector. Data were taken and reduced in
a manner similar to that used for the cold detector. With the preamp. input
open the noise referred to the input with G = 1000 and Af = 6 was (35p,v/
1000) = 35 nv. Thermal noise for a 10K resistance and Af = 6 is 31. 6 nv
which gives a noise figure of 0. 6 db. With a 12. 7K resistor on the preamp.
input the noise was 30 ^iv/1000 = 30 nv at G = 1000 and 3. 0 p,v/100 = 30 nv
P
at G = 100. (At f = 330 Hz the noise was 32 uv/1000 = 32 nv at G = 1000).
P P
Thus no drop in noise at high preamp gain was observed. The noise per unit
bandwidth is thus about 12 nv/ KHz = 0. 012 p,v/ V Hz which is small compared
to the noise with the cold detector.
7. 1. 2 Analog System Measurements
During and after the waveanalyzer measurements, records were made with
the source channel analog system (f = 560 Hz approx. ) consisting of the cold
detector, (beam blocked at entrance slit), preamp. , selective amplifier,
lock-in amplifier compensation amplifier and analog recorder. The source
chopper was on and the reference chopper was off. The system was tuned
up and calibrated just before the records were made. Record lengths
were sufficient to assure that representative values were obtained (At/j
ranges from 40 to 170, 000 with most of the data between 100 and 10, 000.
Because the RMS value of a random signal is difficult to access with
precision, the peak to peak values of each record were read. Voltage
at the detector (peak to peak) is obtained by dividing the measured value
by the overall gain G = (G )(G, )(G ). Peak to Peak voltage at the
P fs xs __
detector per unit bandwidth is calculated by division by |/Af = \TT/4T.'
7-3
-------
The peak to peak results are shown on Figure 7-2 as a function of preampli-
fier gain. The highest values (between 6 and 7) are occasional peaks and the
lowest values (between 1 and 2) are for very small recorder deflections.
Except for these, most of the analog data lie between 0. 67 and 1. 5 of
the equivalent peak to peak noise from the wave analyzer measurements
indicated by the dashed lines ( = 5 Vd(RMS)/ ¥~Hz). The scatter is an
indication of the difficulty of measuring noise by the peak to peak method.
Even considering this scatter the average of the analog data is in resonable
agreement with that from the wave analyzer. Note that the lock-in gain,
G = (DC volts out/RMS volts in), is used for noise calculations.
xs
Also shown on Figure 7-2 is the digital result described in< the next
section.
7.1.3 Digital System Measurements
Data were taken after the field tests to determine the system noise as
indicated by the digital system. Both DVM's were connected to the
MONITOR terminal of the reference lock-in amplifier. A signal of about 8
volts was simulated with the lock-in zero suppress circuit. Digital data
were taken on paper tape at about 3. 3 data points per second and 50 data
points were taken. The -010 DVM was on DC so that it measured voltage
not ratio.
First, data were taken with the DVMs in the condition in which the
field tests were made; i. e. , with a filter on the input of the -010 DVM as
installed by the manufacturer. It was decided that the system would
indicate I/Io more properly if this filter were disconnected. After dis-
connecting the filter, data were taken again.
Each set of data consisting of 50 data points was analyzed; the
(maximum-minimum) gives the peak-to-peak voltage and the RMS value of
the deviation from the mean was calculated using the H-P 9100A computer.
The results for G = IK x 1 x 100 with the cold detector (beam blocked at
7-4
-------
(p-p)
T
O 1. 0 sec
A 0.1
D 0.01
r ' - pp Detector Noise from
r, f Wave Analvzer Data
1
6
5
4
3
2
1
0
" x2 ' --~ -^ '
O ""^
o ^
XI. 5 A >
"""-o — ^ \ -
* * N x
li1 ^ * \^
^ o a !?*^fl 1
xO.67 ^ n A » *
^JCO.5 J ~^"^^>A *
S-— c;
i i
Digital
Result
"^ (Without
Filter)
10 100
Preamp Gain - Gp
1000
Figure 7-2. Analog Measurements of Noise
7-5
-------
the entrance slit) are tabulated below and shown on Figure 7-2. The refer-
ence chopper was on during these tests. The voltage at the detector is
obtained by dividing the measured lock-in output voltages by the total gain,
G. which includes G , the normal lock-in gain.
xs
With Filter - nv/ylrf1 Without Filter - p,v/
T -015 -010 -015 -010
0. 01 sec
0.03
0. 1
0.3
1.0
Avg
Avg (pp/
RMS)
4.
3.
2.
2.
3.
3.
07
47
66
24
46
18
4. 5
0.92
0.76
0.71
0.68
0.84
0.78
1.34
1.93
2.20
2.50
3. 16
2.23
4.
0.28
0.45
0. 53
0. 59
0.85
0. 52
5
6.
3.
3.
2.
2.
3.
32
27
07
44
44
51
4.6
1.09
0.73
0.63
0. 54
0.70
0.74
4.
3.
2.
2.
2.
3.
63
97
86
22
42
22
4.6
0.87
0.82
0.70
0.52
0.65
0. 71
Note that, before removal of the filter on the -010 DVM, the noise (and
also the response) is substantially reduced compared to the 015 DVM
particularly at the shorter time constants. After removal of the filter,
the noise data (and therefore the response) for the two DVM's are much
more similar. Note also that the ratio of the (pp/RMS) is very nearly
equal to the previously assumed value of 5.
The filter of the -010 DVM was left disconnected to make the
response of the two DVM's more nearly equal.
7.2 MONOCHROMATOR CALIBR. - TEMP EFFECTS
Because of the high ambient temperature during some of the field tests (particularly
the emission tests) it is appropriate to record the results of a prior check
of temperature effects on the wavenumber calibration of the monochromator.
An examination of the approximate effect of temperature on the mono-
chromator wavenumber calibration was made by placing a heat lamp above
the monochromator. Time did not permit the elimination of spatial
7-6
-------
temperature gradients; rather, the monochromator was heated to about
40° C and allowed to cool slowly. The temperature and the wavenumber
drum readings for the 5th, 7th and 13th order of a mercury lamp were
observed as a function of time. From the drum readings at 40, 35 and 30° C
the wavenumber calibration constants were determined:
v = A + B (WD).
A is the wavenumber corresponding the zero on the drum and B is the wave-
number interval corresponding to one drum turn (WD is the drum reading in
turns). A prior calibration at 25° C is also included for comparison. The
total cooling time (to about 29° C) was 5 hours.
Temperatures measured at the center of the top of the cover and
at a hole in the base of the monochromator near the Photomultiplier tube
agreed within about 1"C, the cover temperature being slightly cooler.
The results of the test are given below 5
40° C Order i> WD
5 3662.3 18.391
7 2615.9 10.069
13 1408.9 0.454
35° C
5 18.376
7 10.056
13 0.455
B cm /turn
125.650
125.763
-1
A cm
1351.4
1351.3
5
7
13
25°C*
5
7
13
18.351
10.042
0.455
18.334
10.042
0.471
125.938
126.171
1351.2
1349. 1
*Calibration the previous day (5/26/70); the monochrometer has since been
realigned so that only relative values are important.
7-7
-------
B. "'"''"
WD - WD
I/_ - 0. 454
7. 3 DIVIDER BOX
In order to calibrate the system completely from the detector to the DVM
it is necessary to insert a divider box (Figure 7-3) ahead of the preamplifier
with the divider box having a nominal division ration equal to the nominal
preamplifier gain. The lock-in calibration signal is applied to the divider
box.
The measured output impedance of the lock-in calibration signal
is 52 ohms. The measured input impedance of the preamplifier is 10. OIK
ohms. These values were used in the calculation of the division ratio.
Resistance values were measured on a Cimron 6753 digital multimeter to
an accuracy of . 005% FS on 0. 01% RDG (whichever is greater) on the 1,
10 and 100K ranges and 0. 005% FS or 0. 02% RDG (whichever is greater)
on the 1000K range.
The calculated values of the division ratio, Ei/Eo, agree with the
nominal ratio within 1% of the nominal value.
7-8
-------
Lock-In
Cal. Output
Ei
n /VWL f~\
t
r
-pi
^
Divider Box
<
'
t
.
;99.53
: K
Nominal Divisior
Ratio E./E '
1 0
1
4
<
i
J49.7
' K
T *-^" S
! >10.035K
< 1 o
' K ' Q
100. OK
20
299. 9K
50
500. 7K
o
100
999. 8K
Q
2Tto
500
«
iiooo
101.7
_J
[ Preamp
Direct
Input ,.
— nE? InV^ f
t'j '
|
M
•
*
; 10. OIK
— O *23.6V
C 50mfd
i
Nominal Div. Ratio Actual Div. Ratio (Ei/Eo) Actual/Nominal
1
2
5
10
20
50
100
200
500
1000
1.0068
2. 0108
4.989
10.002
2 0. 008
50.017
100. 13
200. 16
496.5
1009. 1
Figure 7-3. Divider Box Schematic
1.0068
1.0054
0.9978
1.0002
1.0004
1.0004
1.0013
1.0008
0. 9930
1.0091
7-9
-------
RECOMMENDA TIONS
1. For emission tests a standard source should be incorporated between the
receiver chopper and the receiver telescope to check signal levels.
2. Tweak the detection toroid after each move before testing.
3. In the transmission mode of operation a slit height of about twice h would
be used to reduce signal noise (from that with height a h ). Because the
relative source to reference signal is reduced, the rsig control will have
to be increased by roughly the same factor of two.
4. The System Operating Log should be expanded (perhaps to 11 x 17) to
include the source and reference chopper frequencies (or, at least, if each
is on or off) and also the value of T sig. A section for notes should be
included; more copious notes in the field tests would have greatly aided
the data analysis.
8-1
-------
9
REFERENCES
1.. Design and Construction of a System for Remote Optical Sensing of Emissions
(ROSE System) M. L. Streiff and C. R. Claysmith, EPA-R2-72-052 or GDC-
DBE-72-001, January 1972.
2. C. B. Ludwig, Applied Optics 10, 1057(1971).
3. C. B. Ludwig, M. Griggs, W. Malkmus and R. Battle, Air Pollution
Measurements by Satellite, NASA (in press).
4. C. B. Ludwig, R. Bartle and M. Griggs, NASA CR1380 (July 1969).
5. F. Volz Arch. Met. Geophys. Biokl. B, 10 p. 100 (1959).
6. R. H. Pierson, et al, Analytical Chemistry 28 , 1218 (1956).
9-1
-------
TM 6-125PH-336
CONTRACT NO. WA-4800
APPENDIX I
AIR POLLUTION MEASUREMENTS
IN THE INFRARED
SEPTEMBER 1971
SUBMITTED TO:
GENERAL DYNAMICS
Convair Aerospace Division
BY
GENERAL DYNAMICS
Electro Dynamic Division
P.O. Box 2507, Pomona, California 91766
A-0
-------
AIR POLLUTION MEASUREMENTS IN THE INFRARED
FOREWORD
The General Dynamics Electro Dynamics Division, Pomona operation was
funded under a subcontract (WA 4800) to General Dynamics, Convair Aerospace
Division to assist in conducting a series of infrared spectral measurements
for the Environmental Protection Agency, Research Triangle Park, North Carolina.
The primary objectives of these measurements were to demonstrate the
feasibility of detecting and identifying air pollutants and to compare
measurement data taken with three different instruments. The instruments
used in the measurements were:
(1) Convair Grating Spectrometer
(2) Bendix Filter Wheel Spectrometer and
(3) Pomona Operation Interferometer Spectrometers
The Pomona operation acknowledges the cooperation of Mr. G. Rounds of
Kaiser Steel Corporation, Fontana, California, Mr. W. Faulkner of the Air
Pollution Control District, Los Angeles, California, and Mr. T. Banks of the
Pomona Valley Community Hospital, Pomona, California, for providing the
Pomona operation with data regarding the concentrations of stack effluents
and readings of the APCD Pomona Valley monitoring station.
A-l
-------
SUMMARY
The Pomona operation of General Dynamics, in support of the Environmental
Protection Agency, provided:
(1) liaison service with various industrial personnel to obtain access
to industrial plant property, where required, and analytical
data, whenever possible, on concentrations of stack effluents
(2) selection and surveying of sites for performing atmospheric trans-
mission measurements
(3) transmission measurements of an 1800°K blackbody source over 1
and 2 mile paths, and
(4) emission measurements of selected industrial stacks.
For both measurement functions (items 3 and 4), Michelson type inter-
ferometer spectrometers covering the spectral region from 716 cm"'- to 5000
cm~l were utilized. Both instruments were developed by the Pomona Operation
of General Dynamics. A description of the instruments as well as the measure-
ment and data reduction procedures are provided in the Appendices.
The one mile transmission measurements were performed on June 30, July
1 and July 2, while 2 mile measurements were made on July 27 and 28. The
spectral data along with concentrations of O-^, NO, N02, and CO measured at
the Pomona APCD monitoring station obtained through the courtesy of the
Public Information Service Department of the Los Angeles County Air Pollution
Control District, and meteorological data obtained during the measurements
are presented in Section 1.
Spectral measurements of thermally excited gaseous stack effluents at
Kaiser Steel Corporation, Fontana, California, Southern California Edison
power generating station, Etiwanda, California, and Pomona Valley Community
Hospital were made during the week of 12 July. Spectral data, corrected for
instrument response, along with available data on concentrations of major
pollutants are presented in Section 2.
The transmission and emission measurements demonstrate the efficacy of
interferometer type spectrotnetry in detecting gaseous pollutants in the
atmosphere. Transmission data can, for example, readily detect the presence
of small quanties of ozone while emission measurements can remotely identify
CH4, S02 and CO from stack effluents. The rapid data collection (on the order
of 1 sec per spectrum) coupled with the high resolution of the interferometer
provides obvious advantages particularly if the source emission shows rapid
variations or the source itself is non-stationary (automobiles, airplanes, etc.)
A-2
-------
While the present system requires computer reduction for highest resolution,
instantaneous data (on the order of 1 sec) can also be obtained at reasonable
resolution (8 cm~^), see Figure 20. Moreover, simple state-of-the-art attach-
ments (comnierciallv available) can be utilized to provide highest resolution
data (~lcm"l) for any portion of the spectrum.
A-3
-------
Section 1.0
ATMOSPHERIC TRANSMISSION DATA
1.1 ONE MILE MEASUREMENT
One mile transmission measurements with the Pomona operation inter-
ferometer spectrometer and the 1800°K blackbody source provided by General
Dynamics, Convair were made during the week of June 28, 1971. The location
of source and spectrometers are indicated on the map shown in Figure 1.
The source was placed at an elevation of approximately 50 feet above the
spectrometer location, and the instrumentation was arranged as shown in
the photograph of Figure 20
Hourly average concentrations of NO, N0£, 03, and CO (as provided by
the local air pollution monitoring station) are presented in Tables I through
IV and also in Figures 3 through 6 to indicate the diurnal variations of the
concentrations of the pollutants during the week of June 28, 1971. The
hourly average concentrations show a maximum ozone concentration during the
mid-afternoon hours, while the NO and N0£ concentrations reach a maximum
level during the early morning and late afternoon and evening hours. Carbon
monoxide concentrations are relatively constant throughout the day.
The spectral data in Figures 7 through 9 are presented as relative
intensitites corrected for instrument response only. The mid-ir region
measured with an InSb (pv) detector is shown from 2000 cm"l to 5000 cm~l
while the long wavelength information measured with a HgCdTe detector is
shown from 720 cm~l to 2000 cm'l. In order to present the spectra in
sufficient detail, the entire spectrum was divided into three separate
plots. Figure 7 shows the region from 720 to 2000 cm'1, Figure 8 from 2000 to
3500 cm-1, and Figure 9 from 3500 to 5000 cm-1, The relative intensity scales
of all three figures are approximately identical, i.e., the actual intensities
in Figure 7 (scale values 0 to 8) are considerably lower than those in Fig-
ures 8 and 9 (scale values from 0 to 60).
The bpoctral data clearly indicate the sensitivity of the ozone absorption
to changes in concentration. The ozone absorption at 1042 cm"*- (^3) and
1103 cni'l ( Vi) is of particular interest since it alone of all the gases
of interest is relatively unaffected by interference from water vapor, C02
and N'fl absorption.
The applicability of transmission data in the determination of ozone
concentration in the atmosphere is fully demonstrated in the curves of Figure 10
in which spectra with 3 different ozone concentrations are shown superimposed
with the zero levels of relative intensity displaced slightly. The hourly
average concentrations of ozone at the time of the measurements were
approximately 6, 9 and 13 parts per hundred million and the increased absorp-
tion as the concentration increases is readily discernible.
Figure 11 displays the same information but confined to the spectral
region primarily affected by ozone. The relative magnitude of the 03
absorption becomes more apparent in this figure.
A-4
-------
*LT,oN TT
, IV -' i '-•',s.
1800*K Source Locatton
1 Mile Measurement
Elevation: 860 Feet
Spectrometer Location
10,000 Foot Measurement
Elevation: 790 Feet
Figure 1. Map of the Pomona Area Showing Locations of the 1800°K
Blackbody Source and the Instrumentation for the One and
Two-Mile Transmission Measurements.
A-5
-------
A-6
-------
Table I
HOURLY AVERAGE VARIATION OF OZONE CONCENTRATION (pphm) DURING THE
WEEK OF JUNE 28, 1971
DX^
June 28
29
30
July 1
2
HOI
^ |