cxEPA
United States
Environmental Protection
Agency
Environmental Sciences Research
Laboratory
Research Triangle Park NC 27711
EPA-600/2-79-094
May 1979
Research and Development
Accuracy of
Remotely Sensed
SO2 Mass
Rates
PROPERTY OF
DIVISiOW
OF
METEOROLOGY
-------
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-600/2-79-094
May 1979
ACCURACY OF REMOTELY SENSED
S02 MASS EMISSION RATES
by
R. B Sperling
M. A. Peache
W. M. Vaughan
Environmental Measurements, Inc.
215 Leidesdorff St.
San Francisco, CA 94111
Contract No. 68-02-2711
Project Officer
W. F. Herget
Emissions Measurement and Characterization Division
Environmental Sciences Research Laboratory
Research Triangle Park, North Carolina 27711
ENVIRONMENTAL SCIENCES RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NORTH CAROLINA 27711
-------
DISCLAIMER
This report has been reviewed by the Environmental
Sciences Research Laboratory, U. S. Environmental Protec-
tion Agency, and approved for publication. Approval does
not signify that the contents necessarily reflect the views
or policies of the U. S. Environmental Protection Agency,
nor does mention of trade names or commercial products con-
stitute endorsement or recommendation for use.
11
-------
ABSTRACT
Remote sensing data of single-track power plant emissions and
local wind speed have been analyzed to determined SC>2 mass flux
for comparison with EPA referenced methods. Four days of S02
data were gathered from a moving platform by three upward-
viewing remote sensors -- two ultraviolet absorption spectro-
meters and an infrared gas filter spectrometer. Wind velocity
data were gathered by a laser-doppler velocimeter (LDV); sup-
plemental data were obtained from a tethered balloon (tele-
metered) and pilot balloons (optical theodolite). The data
matrix (S02, X-Y position, wind velocity for 120 traverses)
was computer processed; the end result was the S02 mass flux
derived from the remote sensing data. Comparisons were made
between these S02 fluxes (averages for 20 minutes and 60 min-
utes) and those derived from in-stack measurements. The results
of the comparisons show the relative accuracy of the remote
sensing technique for quantifying S02 mass emission rates. The
analysis shows that as averaging time increases from 20 minutes
to 12 hours the difference between the remotely measured SC>2
mass flux and the stack sampling S02 mass flux decreases from
about ±35% to ±10%. In general, no single wind measuring sys-
tem produced superior results over the other two. The LDV and
COSPEC, however, produced the best agreement with Method 6 (+6%)
when the plume was transported near the LDV instrument.
111
-------
CONTENTS
Abstract Hi
Figures vi
Tables vii
Acknowledgements viii
1. Introduction 1
Background 1
Purpose 1
2. Summary 3
Equipment 3
Measurements 4
Data Processing 5
Results 5
Comparison of Results 6
3. Conclusions and Recommendations 7
Conclusions 7
Recommendations 8
4. Data Processing 10
Wind Speed Data 10
Remote Sensor Data 14
5. Results 24
Remote Sensors 24
COSPEC III 24
COSPEC II 29
GFC 29
Reference Method 6 33
Analysis of Results 35
6. Comparison of Results 44
Time-Averaged Results 44
Means and Differences 48
Wind Measurements Accuracy 50
References 51
-------
FIGURES
Number Page
1 Field Activities 4
2 Traverse Route Map 5
3 Activity Summary 11
4 Wind Speed by Day 18
5 Typical Flux Calculation Printout 21
6 Typical Plume Profiles 22
7 COSPEC III S02 Flux Results 25
8 COSPEC II S02 Mass Flux Results 30
9 Reference Method 6 Mass Flux 33
10 Wind/Road Angle >±50° Plumes 36
11 Corner Plumes 37
12 Bifurcated Plumes 38
13 Double Plumes 39
-------
TABLES
Number Page
1 Wind Speed Summary 15
2 S02 Mass Flux Results - COSPEC III 26
3 S02 Mass Flux Results - COSPEC II 31
4 S02 Mass Flux Results - GFC 32
5 Stack Sampling Results - EPA Method 6 34
6 Analysis of Results - LDV Winds 41
7 Analysis of Results - TS Winds 42
8 Analysis of Results - PB Winds 43
9 Comparison of Results, 20-Minute Averages LDV . 45
10 Comparison of Results, 20-Minute Averages TS . . 46
11 Comparison of Results, 20-Minute Averages PB . . 47
12 Comparison of Results, 60-Minute Averages LDV . 47
13 Comparison of Results, 60-Minute Averages TS . . 48
14 Means and Differences of 20-Minute Averages . . 49
15 Means and Differences of 60-Minute Averages . . 49
VI1
-------
ACKNOWLEDGEMENTS
The data processing was performed by Michael Peache and
Jean-Jacques Garbarz; the analysis and report writing were done
by Mr. Peache, Dr. William Vaughan, and Roger Sperling, Project
Manager; Joan Geary, Suzanne Klimsza, Daisy Chan, and Evelyn More
prepared the report. All are EMI employees.
The guidance of Dr. William Herget, the EPA Project Officer,
throughout the project is gratefully acknowledged.
viii
-------
SECTION 1
INTRODUCTION
BACKGROUND
Environmental Measurements, Inc. (EMI) collected air quality
data with three remote sensing spectrometers at a southwestern U.S.
coal-fired power plant using a moving instrument platform and an
automated data acquisition system. The instruments were:
COSPEC III
COSPEC II
Gas-Filter Correlation Spectrometer
The measurements were made over a five-day period, 2-6 August,
1976, to gather upward-looking S02 data to be used to evaluate the
relative accuracy of the instruments for determining mass emissions
rates remotely.
Concurrent wind measurements were made with three systems:
Laser Doppler Velocimeter (LDV), Lockheed Missies
and Spacecraft Corporation1*
Tethersonde (TS), Intera, Inc.5
Pilot balloons (PB), EMI.1
Simultaneous in-stack reference method testing of S02 concen-
trations EPA (Method 6) and gas velocity EPA (Method 2) were made
by Entropy Environmentalists, Inc.3
The data collected from the moving laboratory have been re-
ported in tabular and plotted formats.1 These listings provided
the spatial S02 data needed to combine with the wind velocity pro-
file data for calculating S02 mass emission rates.
PURPOSE
All of the field data have been synthesized into a three-by-
four S02 mass flux matrix: mass emission rates from fhvee remote
sensing spectrometers for four sets of wind measurements. These
-------
remotely quantified S02 fluxes were compared with reference method
in-stack measurements to determine the:
Relative accuracy of the remote sensing method as
compared to the reference method, and
Improvements, if any, in remote sensing accuracy
using more accurate measuring equipment.
-------
SECTION 2
SUMMARY
EQUIPMENT
The data analyzed in this report were gathered by three remote
sensing spectrometers and three different wind measuring systems.
Spectrometers
Two of the spectrometers were Barringer Research Ltd. correla-
tion spectrometers: COSPEC III (serial number 6061) and COSPEC II
(serial number 5922) were provided by EMI and U.S. EPA/RTP, respec-
tively. The third instrument was a government-provided gas-filter
correlation spectrometer built by Science Applications, Inc. All
three instruments were installed in an EMI Air Quality Moving Labora-
tory in the upward-viewing mode.
Wind Measurements
Three different wind measuring systems were used to determine
wind velocities at the altitude of the stack emissions:
The van-mounted Laser Doppler Velocimeter (LDV) system
was located 800 meters northwest of the stack. From
this location it collected and analyzed horizontal and
vertical remote wind velocity data at altitudes from
30 meters to 800 meters above ground level (AGL). The
data were recorded on strip charts and on magnetic tape
for later analysis.
A tethered balloon system located 100 meters from the
LDV site was used to measure wind velocity from ground
level to a height of 600 meters. The data were recorded
on strip charts for later analysis.
Pilot balloons were released from the tethersonde site;
the data were analyzed to determine wind velocity up to
1000 meters AGL.
All data have been reported separately by the individual con-
tractors . l >3' **' 5
-------
MEASUREMENTS
The equipment provided the following sets of measurements:
120 traverses with the moving laboratory
13 hours of vertical wind profile data from the LDV
20 hours of vertical wind profile data from the tethersonde
20 pilot balloon measurements of winds aloft
Figure 1 illustrates the field measurement activities, and
Figure 2 shows the traverse routes used by the moving laboratory
and the sites of the meteorlogical systems.
Figure 1. Field activities (clockwise from lower right)
Adding liquid nitrogen to GFC Spectrometer;
pair of upward-looking COSPEC remote sensors;-
observing flight of pibal toward tethered
balloon; checking MAP listing of van position
and spectrometer SC>2 data.
-------
BALLOON _
SITE A LUV
2UO 400
< 1
MtTEI2 optical depth data for the three spectrometers were
edited, and appropriate calibration factors were applied to convert
the millivolt reading to part-per-million-meters S02 (ppmM S02).
The X-Y coordinates of the moving laboratory were also edited to
establish the same coordinates system for all 120 traverses. The
final coordinates and SC>2 optical depth data were stored into a
computer for the computation of SC>2 mass emission rates.
Wind Data
The LDV system wind data were reviewed; discrepancies were
noted between the tabular listings and the plotted results. It
was determined that the data had been hand-processed and could
contain a systematic error of +10% (the difference between hand-
processed peak values and machine-processed average values). A
second processing by computer was requested. This computer pro-
cessing resulted in a new set of LDV wind velocity numbers.
Rather than to present both sets of numbers, it was decided to
present only the revised LDV values. The original data (LDV) were
computer-processed to form an LDV set, which was used as delivered,
RESULTS
The processed data were used to calculate SC>2 mass emission
rates yielding a three-by-four matrix: SC>2 mass emission rates
-------
from three remote sensing spectrometers for four sets of wind
measurements. The in-stack measurements of SC>2 mass flux were
converted to common units of metric tons per day SC>2 (MT/D S02)
for comparison with the remote sensing data.
COMPARISON of RESULTS
To compare the remote sensor results with the in-stack re-
sults all data were averaged over the same time periods. The
remote sensor S02 fluxes were averaged over 20-minute periods
during which the in-stack data were collected; 60-minute aver-
ages were also determined. The 20-minute averages were within
about ±35% of the reference data, whereas the 60-minute averages
were within ±20%. Extending the averaging time to 7-to-12 hours
further narrows the difference to ±101, showing clearly that the
relative accuracy of the remote sensing method is dependent on
the averaging time or, more precisely, the number of profiles
used in an average. There was no significantly superior wind
measuring system, considering the time-averaged comparison of
results. The pibal-derived S02 fluxes, however, were closer
to the reference method than were either the LDV' or the teth-
ersonde results for 20-minute averaged data.
-------
SECTION 3
CONCLUSIONS and RECOMMENDATIONS
CONCLUSIONS
The following conclusions can be drawn from the first-order
analysis presented in Section 6 of this report regarding the rela-
tive accuracy of the remote sensor flux calculations:
S02 Flux Accuracy
The relative accuracy of the S02 mass flux calculations using
remote sensor and wind speed data is dependent on the averaging
time or number of profiles. The approximate differences relative
to reference methods are:
±35% for 20-minute averages or 2-to-5 traverses
±20% for 60-minute averages or 4-to-13 traverses
±10% for 7-to-12 hour averages or 25-to-75 traverses
These results are consistent with previous studies. Single
measurements of a plume profile can have greater than ±50% error
because under most dispersion conditions the actual plume cross-
section is non-uniform, and several measurements (traverses are
required to provide a representative average profile. The greater
the number of traverses, the lower the expected error until the
minimum difference (±10% in the case of these data) is approached.*
Wind Measurement Accuracy
The relative accuracy of the three sets of wind speed data,
as shown in the individual sets of S02 mass flux calculations, can
be assessed as follows:
*Millan M. Millan, in his research for Atmospheric Environmental Service of
Canada, notes that their average for 18 to 22 profiles/hour at -1.5 km down-
wind is about ±15-17% different, which agrees reasonably well with this data.
-------
LDV1 --
The reprocessed LDV data corrected the positive bias of the
original LDV data. Overall, the LDV average mass flow (78.9MT/D)
was 17% greater than the average Method 6 determination (67.4MT/D).
However, selecting data associated with SE winds that brought the
plume over the LDV site and using only the Method 6 results for the
same time frame, the LDV results with COSPEC III are only 6% higher
than the average for Method 6 (70.1 vs 66.2).
TS --
The tethersonde data had a negative bias producing fluxes over
the long term within -101 of the reference method.
PB --
The pibal data produced the best results (<±5% in the long term),
but the number of 20-minute averages was smaller than the other two
methods.
It appears that the LDV system may be the most accurate of the
systems tested, provided that it is used near the plume to measure
the wind field near or in the plume. This finding from the subset
of SE winds implies that a mobile remote wind monitoring system
would be desirable for remote sensor plume studies.
RECOMMENDATIONS
The following recommendations are offered to assist in advancing
the state-of-the-art of remote sensing emissions monitoring:
Further Analysis
Further analysis of the body of data in this report could lead
to:
Interpretation of the relative accuracy of remote
sensor S02 mass fluxes expressed in terms of error
intervals and confidence limits.
Comparison of these data (1976) with previous simi-
lar data (1975) to determine how to optimize remote
S02 mass flux measurements.
Identification of measurement protocols that should
be followed to minimize the error in the flux calcu-
lation and conditions that should be avoided that
degrade the measurement technique.
Selection of the most suitable wind velocity measure-
ment system to accompany remote sensor measurements
used for emission rate calculations.
-------
Further Testing
With or without further analysis, further field testing could
lead to:
Confirmation of measurement conditions that tend to
optimize the flux calculation.
Verification of the most suitable wind velocity mea-
surement system for remote sensor field work, such
as a mobile remote wind sensing system to operate in
conjunction with a remote sensor team.
-------
SECTION 4
DATA PROCESSING
The first step in processing the remote sensor and supporting
data was to prepare an activity summary relating all sets of mea-
surements in time. Figure 3 shows the times for every field mea-
surement.
Each set of data was reviewed for the four measurement days
to determine its suitability for further analysis. The three sets
of wind data were reviewed independently; the three spectrometer
data sets were individually prepared for merging with the wind data
to calculate S02 mass flux. For each of the 414 flux calculations
a plume profile was drawn to aid in the analysis.
WIND SPEED DATA
Before wind speed figures could be selected from the various
data sets for the individual times of the remote sensor traverses,
the individual altitudes first had to be selected. This was done
by calculating plume rise and vertical dispersion; the appropriate
wind speeds within the plume were then taken from the vertical wind
profiles.
Plume Height
The extensive wind information could not increase the accur-
acy of the flux calculations unless the height of the plume was
known. If the wind speed varied considerably with altitude, it
was imperative to know the height of the plume because the flux
calculation results are directly proportional to the wind speed.
A 20% error in wind speed causes a 20% error in the flux calcu-
lation.
To obtain an approximate plume height (since no direct mea-
surements were made) the existing wind information was used in
conjunction with the Briggs plume rise formula. (Though its agree-
ment with a wide range of plumes is established6, its use during
unstable conditions such as those of these tests is not so well
established.)
10
-------
3 AUG 76
I'lliAL
CROSS - S r \CK
MI.TI101) 6
COSl'l.C II, III
A A
8 1J 10 11 12 13 14 15 10 17 If.
Tl Tilll: _ ______ _____
(i AI If. 7C
1'IHAL M HU" /D
LUV ___ _ ^_^_____
CROSS -ST\L'K
Ml 11IOD 0
A A A A A A A
COSI'KC 11,111
I I I
8 y 10 11 12 13 14 15 10 17 IS
TL11ILR __________
R Allfi 7R
PTRAI J ^"^ '°
PiBAL »
LUV ^^^_^^^^ _____
CROSS - STACK »...~-«^^»..«^.^»......,
Ml TIIOD I) AAAAAAA
COSl'lC 11, 111
3 9 10 11 12 13 14 15 16 17 18
PIB<\L 6 AlK 76
LUV _ ^^_ ^
CROSS-STACK »«.«...«.«.«.«.«.
ML'11101) 0
A A A A A
COSl'LC II, III
3 10 11 12 13 14 15 16 17 18
Figure 3. Activity Summary.
11
-------
Ah = 2.0 Fa/3 X 2/3
where
F = g AT Vs d2
4 T
s
Ah = height of plume axis above the top of the stack at
given distance downwind (plume rise)
g = acceleration due to gravity
AT = T -T
T = absolute temperature of ambient atmosphere
T's = absolute temperature of stack gas
Vs = stack gas velocity at stack top
d = diameter of stack opening
X = downwind distance
u = wind speed
The average stack gas veloci'ty and temperature as measured
by Entropy Environmentalists Inc. were used because of a rela-
tively small variation in these values during the test period.
Likewise, the average ambient temperature was used because the
14°K variation would have caused only a + II variation in the
calculated plume height. To determine the downwind distance of
each traverse, a computer program was used to find the point of
maximum concentration and then calculate the distance of that
point from the power plant stack. Finally, as an initial wind
speed input, the wind speed at stack height, as determined by
the Laser Doppler Velocimeter or the Tethersonde, was input and
the plume rise was calculated.
The plume rise when added to stack height gave plume height.
The wind speed at this height was then compared with the speed at
stack height. If the two speeds varied considerably, the plume
rise calculation was repeated using the wind speed at the calcu-
lated plume height and the wind speeds were again compared. It
must be remembered that the accuracy of the plume rise calculations
is probably within a factor of two, so that the results were not
expected to converge on an exact figure for plume rise, but were
used to determine the general region in which the plume was likely
to be moving at the time it was sensed by the COSPEC and GFC.
These height determinations were then used to choose the approp-
riate wind speeds from the wind profiles.
A second set of calculations was made to determine the prob-
able vertical dispersion of the measured plume. To generalize
the process Stability Class B was assumed for all four measure-
ment days; this is based on the conditions of strong insolation
and wind speeds greater than 3 meters per second. According to
Gifford (Ref.6,p. 259) qz is on the order of 20-to-30 meters at
200-to-300 meters downwind of the stack. These assumptions were
all reasonable for the typical plume measurement made with the
12
-------
moving laboratory. Hence a 6a vertical plume dispersion of +75
meters, centered on the plume height calculated previously, was
used for selecting wind speeds. For example, if the calculated
plume height was 200 meters, wind speed data from 125 to 275
meters altitude above ground level would be selected for that
traverse.
Each set of wind data was studied in turn, starting with
the LDV data.
LDV
The laser Doppler Velocimeter wind speed data were originally
presented tabularly for vertical sweeps to 80 meters altitude and
for temporal measurements made at fixed altitudes for periods of
30 seconds. Selected data were also plotted with time. Compari-
son of the tabular and plotted velocity data revealed some discre-
pancies -- differences of up to 30%.
Rather than completely discarding the original data, both sets
were retained for the purpose of Table 1 and identified as follows:
LDV - original hand-processed results decreased by 101
to approximate the averages produced by computer
processing.
LDV - new computer-reprocessed results. (For 3 August
LDV could not be computed because the full set
of necessary data was either not available or not
adequate for this calculation.)
From the LDV data wind speeds within the plume (plume height
±75 meters) were selected that were coincident with the traverse
time recorded by the moving laboratory. Where two or three values
were available, they were averaged; in many cases only one value
existed.
All wind speeds are summarized in Table 1. If there were
no data at the proper altitude the nearest-altitude wind speed
was selected and coded "a". Also, if there were no data within
the time limits the nearest-time wind speed was selected and
coded "£". (See Section 5 for further discussion of these Error
Codes.)
Tethersonde
The same criteria were applied to the tethersonde data, and
the best wind speed values were chosen and tabulated. (See Table
1.) Over half of the tethersonde measurements were made at alti-
tudes lower than the caculated plume; they are coded "a". None
were out of tolerance with respect to time.
13
-------
Pibal
The twenty pilot balloon measurements were treated similarly.
However, the selected pibal wind (all of which were in tolerance
for altitude) were purposely applied to adjacent time intervals
to simulate the situation often necessitated by extrapolating in-
frequent pibal measurements. They are coded "2>" in Table 1.
Wind Summary
The selected wind data presented in Table 1 are also plotted
in Figure 4. These daily plots show the differences between the
four sets of velocities. It is important to note:
The differences between the LDV and LDV wind data
are significant; the LDV results tend to be higher
by as much as 30%, so they were not used for final
computation.
The tethersonde data tend to be low, principally
because the balloon was often tethered at altitudes
below plume heights determined after the field pro-
ject.
The pibal data show general agreement with other
results.
REMOTE SENSOR DATA
The remote sensor S02 optical depth data required further
processing prior to merging with the wind data (in the flux cal-
culations). The COSPEC and GFC were treated in a consistent fash-
ion.
COSPEC
When making mass flux calculations using COSPEC (or GFC) mea-
surements, it is imperative to accurately determine a zero refer-
ence (background) level, which is subtracted from the COSPEC (or
GFC) values, thereby leaving only a signal due to the S02 of the
measured plume. In working with digital results (which are aver-
ages over 20 meters, as provided by the MAP System), it is diffi-
cult to spot a background value such as might be done by drawing
a baseline on a chart record output; therefore, a different tech-
nique was used.
Most traverses under the plume were made so that there were
five to ten 20-meter averages on either side of the plume that
were measurements of background levels. Each traverse was evalu-
ated, and an average of five readings in the background region on
either side of the plume was calculated to provide an average back-
ground. If the average background on one side was more than six
14
-------
3
C
>croco ooor- o m
fi
O
O
in o r-j (
^
CMf
>"
rt
s!
3
CO
Q
n
w
o.
CO
Q
^^
i i
IS
, j
. T
M
<
E-H
- §
n "*
UJ --,
h-
X
O
Q X
H ~
UJ CO
;Q- :>:
un *
1L ^
-J X
O
i
LU X
: ''i .
O- X
r-vO^rr-orotn.ocN^in^o-^
O c^cor-cor-r-r-vocococo
O o o
O 00
i corMCNCNrMtNrNrNO i i o o
m o m
iu-\ rf (J- O^CNCOCN ^O
K1M"CNCN rJOO ^T
o--o^oo o *
o o o in cj\ ^j -7
1 1 1 IN"|I 1 I |T^I^OO
(N^ntNCN rrmoo ^J ^1
oo--
^*- r~-nOCT^corocoo'y^coK>
IOTCO OO^ r~~ vDCNiCT^i'XDr-r^ir'rr c>ccr---CM
COOO OOOOOOOOOOOO CM
o c o o
^co^-coor-r-i- c.^^^0
m^vO^OinvDinmin my-,^^
OOOOOOOOO 0 ^
ooooooooo e o ^ ,(
I I I i t I I I I i i -t -r |
m in m m in in m in ir> .." j".
j
OCO(N K> ^^^-TT
oo K> r-J'xlir'Oi
???^ss?^s ~"^TT ;
\D^rcooio^cricNrr,fT, "^ r- co i
s§§oooss§ --^.^
ooo o
o\D-«-CD'^or- cOCNOocNirNrMmcnor-omc"-m TJJ-\O
ooo CM o (NOCT>f~- ccrsjrM c\jojfNj*r f"- -^ CN
o Tj-r--cocorNjcocof~\r--
CNrN(Nir\irsirMCNCMCNir\r-jtNfMrM
I I I I
, 'j*\ootT'-^-cor-iOhi
D. oo cxic\if'"i'J^rtino
- CTiCT O'O^C7>!>CT'O^CT1C7>O
! OOOOCJOOOOO
15
-------
CD
o
o
COlTiCOCO "cT CO in (N
^ tN ro (\i n fi CM ^
o in o *<
o
o
PQ
<
H
^§5 S55* = S^55
UJ X i
-t- i |
[? _ SKSSK3^°S
i a i t i i
IX -- '
CO -'
j , >- O OoOOOOO OOOO
! .x o OOQOOOO oooo
T: | iinliiiiiioiiii
> UJ > O OOOOOOOCNOOOO
OX "^ u~\ u-> in in o O o mooo
; 1 i ~ (Nl (N (N CN(NCN
in cr> r- o r- o r-
LJ" r«-| ^ f~i *"* in in in
| l: £? ^[^(iil'tfinininiii^
KI K^ "^ f*"i in in in
i
0 ~ (
a... ^^^^^^^m^^r-r-*^
' h- OOOOOOOOOOO
T ^, OOOOOOOOOOO
O liJ ^ OOOOOOOOOOO
ji u-\if\mooooinooo
' I-NJ rj fxj rg INJ rsi rj
; _ S§SS|SSSS|RS
_ O ' [^ | 1 1 1 1 1 1 1 1 1 1 1
S?,?,SSP;S?,?,S
5SSlssl§55
1 1 1 1 1 1 1 1 1 1
[Nj rO r^\ -^ ^ in- O O O
OOOOOO
oooooo
oooooo
1 t 1 1 1 1 O O O O
oooooooooo
OOOOOin
CM CN O4 rsl (N
siss! s
i i i i in i m
r- CD rsj o KI en
oooo o
- rs -^ K, * ^ ^ o a, «
oooooooooo
oooooooooo
rOK~i^K>roK^K\pnKi^r
t 1 1 ( ( 1 ( 1 1 1
OOOOOOOOOO
ooooomomoo
sss;t;'^KSS =
s^sgsliP
oooooo
PPPPPSS = ^ = ^ = = = = = = =
Sg!S88RRSiR5££K£i=5K£5
><( 1 1 1 1 1 1 1 1 1 [ 1 1 1 1 1 1 t
*in^*^K>KA fSlK>rjK>K'K>vOm
OOOOO OOO
OOOOO OOO
OltlllO iTlOOOOO
mooino Ooo
(N (N (N K\K1fO
f~> O (N r i (N
O "J CN r*
I-T1IICN III CN""*
r^oocorsi »>* r~- KIK>K"I^»O
^ O fN (Si
O ""N fN IN
^^mm^^^fNCN .D^r^^^oofN,^
ooooo oo ooooo ooo
OOOOO OO OOOOO OOO
O f i * f r o ; i f i i i f O i i i
K^OOOOO^OO OOOOO^OOO
^noomo oo ooooo ooo
OOO CN(Sit*~>w"1 K"\K>K>
lt|OII(N(N(N rill III
1^1^ f\J (N W^K*:
OOO (SifN^nfO K>r^f'
r\ir~-vor\i
CM
>D vo (N nD
1/1 o o ~ K^
^j in tn in in
JtNjf-- COfNO^T CO''^CT'OOO^D vOCMCOCDvO i£) CO O ^ C1 35
rNK>""iTr*jinooo r^-^o (NCN^^mm rsjo CN "^
I OOOOOO
16
-------
O
u
PJ
-3
PQ
s
CL
fjj
LU
X
UJ
Q
>
Q
_J
O ^
LU CO
LU -N.
0_ S
t/1 »_
f
X rt
UJ *~*
X
?!
Q ~
UJ (/)
£L S:
CO
jj ""^
~ Q
UJ CO
LU *-x
CL 2
LO
5~
X "
*l
Q ~
LU (.1
LU *-^
a. s:
LU "~"
X
E Q
f
Q- X """
O 1
< en ^
t/l Q v
- i
:> uJ
<£ i/) '
2 CE 9
1- LU Z
CN
O
1
o
(N T-
tO (N
CO CO
O *
\ 1
it in
m
O
i
r~~
0
cncoco vo ^r o o m o in r- o a> co in
roro^K^
'fSj(Nrororo in-^^^r
rvj [^ i J o
f^^jtNirNCN oooo
ro -d- m TT
"
,
(1
"^
OOOO
O 0 O O
O 1 1 1 1
CM O O O O
in in
O 0
0 O
1 1
o o
o o
-a
i i
tn in
rO fO
\o in
O~>
O O O CO
O^ CTi 0s
O O O
rr- co o n. in CN
fN romm^
.-,.--,000 oooo
§ " 0 0 0 0 CJ O O
in in 7 7 7 V 7 ? 7
A A o o o oooo
oo^^^ amOLn
^ ^ tN rsj
Rjgs |_|
O (xj (SJ tN O 0
^^fNrsiiN (NrsirsirM
OfN\D _r^ co
^ ^ [ i.ri o Mil r-j P-I ro TJ
1 ' ' CO v£) O ' ' ' fi-
Csj rj CN CN CN 0 O C 0
.«;» »
C'
O
O
_
in
o
C'
o
in
O
fN
*O
5
T
S
**
^r
n
I
^
O
o
o
^r
o^
o
cr-
o
_
VC
0
O
1
CD
Tf
O
CN
»J
s
*J
f )
s
^
*
o
u
o
0
1
o
0>
«T
fN
^
0
f )
O
0
in in
n ^r
O
t t
in in in
OOO
OOO
OOO
O
o
PO m
in i o
o>
O
0 O
r- o vo
OOO
OOO
fN fO tO
1 1 1
OOO
m o o
o
o .
ON
0
in
ro
O
I
o
in
O
o
1
o
o
1
o
0
10
o
o
fN
o
tn
S
fN fN rsi ro (N
s
iri
^r
0
CO U3
,7,oS
1 [ 1
in in o
BSS
S
r-i
K1
s
inmin
r-K^
OOO
1 I I
OOO
O O fN
o m
inm
roK,
O
( 1
o o
vo vo in in in
OOO
000
i i t
OOO
fN fN
*T ^T in
OOO
OOO
0 rom
f- r- in
§o o
O O
»-o CM h">
i l l
OOO
o m o
0 0
0 O
fN fN
Ssl
*j ^t in
'. > C J O
1 1 1
to T «cr
SSB
o o
o o
1 1
o o
CN
in C
O
1 t
O O
*f _
tn vo
O O
0 0
1 <
O 0
in o
INI
O
in
O
r-
sS
in r-i
o
in in
gg
:
r-
in
0 ,
fN
mov-ro^r.j-fNrf'0
^ « tn in in in in ^ ^ IT.
M ro to ro ro ^ ro ro ^ TO
e = z^^^ji^r
l l l l i i t ( I TJ
*j inmtomvcii'"' o in i*- vi?
OOO r^or~lO'"~!O o 0 O
O ^O 'C' ^~- O
tN ro rr u>
O CN f*l-3-*J
SsSssSsaSSS
gJ&SSSi^Si
(j -_ {sj fsj (NJ fO i-> TT m 10
1 1 t l 1 1 l l l l l
0 - N |2 ± n V r; 5 j, o
17
-------
CD
CO
(S/W) Q33dS QNIM
(S/W) Q33dS QNIM
I co
CD
s
Q O t/1 CQ
-------
millivolts different from the other side, the traverse was con-
sidered to be invalid due to an incomplete traverse of the plume
or another sampling problem. This background value was then sub-
tracted from each 20-meter average for the traverse. When the
resulting values were negative (due to instrument noise) , the
result was considered zero.
The next step was to multiply the adjusted millivolt read-
ings by the calibration factor. This factor was determined from
a calibration curve that was made up of data from all calibrations
made during the test period. The calibration factor is time-
dependent because the COSPEC's response varies markedly with the
sun angle. This variation was considered to be consistent during
the week of field work; therefore, the same time-dependent calib-
ration curve was used each day.
Because of the non-linear response above 600 ppmM of the
particular COSPEC used in this study*, an additional step was
required to prepare the COSPEC data for flux calculations.
Extensive tests were made in 1975 of the linearity of the two
COSPECs used in this study. From the information gathered in
these tests a curve was constructed to estimate readings for
values about 600 ppmM S02. Values from the curve were then
used in a polynominal regression to determine a conversion for-
mula to obtain true ppmM values for those readings over 600 pppM.
When this was completed, the COSPEC results were ready to be used
for flux calculations.
GFC
The Gas Filter Correlation instrument data had not been pre-
viously processed because the sensitivity (ppmM S02/mv) was not
available.1 Using a calibration curve provided by the Project
Officer6 for 5 August 1976 (judged the most suitable data to be
processed), the sensitivities were found to be:
5.88 ppmM S02/mv (day)
6.58 ppmM S02/mv (night)
Following the same procedure used with the COSPEC the zero
reference level was determined for the millivolt readings for
each of 33 GFC plume profiles. This was more difficult for the
GFC because the profiles were less distinct than the typical
COSPEC profile. In addition there was greater drift from one
edge of the plume to the other, necessitating the subtraction
*Note added in review: It is now possible to execute special fine tuning to
eliminate this high concentration nonlinearity for COSPEC II and III.
19
-------
of sloped zero-reference lines. (No non-linearity correction
was required.) The next step was to apply the sensitivities to
yield the ppmM SC>2 optical depth values for processing into mass
emission rates.
Mass Flux
The mass emission rates were calculated from the three remote
sensor SC>2 optical depths, the four sets of wind velocities, and
the geography (X-Y coordinates) provided by the MAP System. The
procedure, described in detail elsewhere2, is summarized below.
The S02 Mass Flux is calculated by the formula:
S02 FLUX = E (C x sin a x I x u x F)
(summation of individual segments of traverse)
where C = COSPEC optical depth reading in ppmM (average
value during one segment)
I - length of road segment
a = angle between road segment and wind direction
u = wind speed
F = conversion factor used to obtain MT/D SCU.
As the COSPEC passes under the plume, it measures the total
burden of SO 2 which is output by the MAP System as an average
value each 20 meters along the road. To obtain the flux of gas
across any segment the optical depth in ppmM is multiplied by the
length of the segment to obtain the total gas above the segment.
This value is multiplied by the sine of the angle between the
wind direction and the road segment to account for the fact that
the road may not be perpendicular to the flow of the gas. Finally,
multiplying by the wind speed and a conversion factor to convert
to metric tons per day (MT/D) gives the final result. The sum of
these calculations over one traverse gives a value for total mass
flux. The sine a term equals unity if the X-Y values are projected
onto a line perpendicular to the wind direction. This step
effectively shortens the segment length in proportion to the line
of the wind/road angle and allows for calculation of the center of
mass and movements about the center of mass.
Because the direction of the wind can vary dramatically
during a short time, the wind direction was derived from the
center of mass of each traverse rather than as measured by one
of the wind sensors. This value gives the best estimate of
true plume direction but does not account for possible wind
shear. 7
20
-------
Flux Calculations
Once the necessary calibration and nonlinearity factors had
been applied to the data and the appropriate wind speed had been
chosen, the data were ready for the actual flux computer calcula-
tions. The following outlines the methodology of the flux calcu-
lation program. Once the data to be used were loaded into the
computer memory, the center of mass of the COSPEC data was calcu-
lated.
Because the gas flowing across the surface perpendicular to
the wind direction is of interest, the X-Y points of the traverse
are projected onto the line perpendicular to the stack-traverse
midpoint segment, using the assumption that the wind is blowing
parallel to the stack/midpoint vector. An example of the compu-
ter printout for the flux calculation is shown in Figure 5.
T:,-IE
*
:2y
.21
1 22
123
124
125
125
1"
123
129
1 30
1 V 1
172
'. 23
i 34
1 ?5
1 1 6
. . -
'I .2
'.:>-
r
; : T
h IC,.
' -i *-'-'jU?T "6
: ^*l- 3j: 03^
:.-:EHT: 1
£'* U. P
.UnTIO
-EG
4: SC2
TC2
iNGLE LENGTH P-M
103
10:
103
163
103
103
163
103
103
103
101
101
103
10!
101
101
1C3
10?
I n 1
=;; "LU . 4;. .-:
,. ,, .,ft:E ^..^
;-p£:7 C OF H:
. 'EliT:
2 -0 =T NEHFEST
£ ': IMI'IfiL FT:
-\-»
_> i
20
19
21
20
19
21
28
19
28
20
21
19
20
- 1
20
,2. ft
1 *
1
It"
"i - 7 o "' ~'
,:.
I .riTE:
12°
: OF -i
3
1
3
3
0
0
12
41
247
3t 1
555
123
6
9
9
1
;
0
~
0
t'.ET-
- -"
: -12
-12
UINE
fiPEfl
PPMH2
113
132
191
256
256
256
49?
1314
5377
13126
24J4P
Z 42c
2 C4 ^
' 2 71-
2 o97
2 ?1S
2 ^"9
1 -"
2 . J1:.
2i: ;.:
M Tlr)3 DM,'
HOI'?
~2i --"5..j-Sc
--«.3?10
-" ^ ~ ~ ~ P »-"=.
'"-'" M5:'-
TPfl',EF,SE * It
7FEEIH 6.5 UPS
FLUX SUM
n~/r>
0.
0.
0.
0.
e.
0.
0.
i.
7.
11.
16.
4.
0.
0.
0.
8.
'?.
0.
y.
0.
_.:.?c
-0-:
BY
o
0
1
1
0
0
4
3
3
2
3
B
2
3
;
0
1
IJ
^
^1
e^'.o-is
,-?4.1J
(IT.
0
0
0
0
9
0
0
2
9
20
23
42
42
42
4:
4;
4}
4 J
4~
4 :
DHY
, 2
.2
.3
.4
.4
.4
.8
.1
.3
.5
.3
f c
.a
. 3
. 6
. ,;
f 7
_ 7
a
. ?
0.
0.
0.
0.
0.
0.
0.
1.
11.
IS.
4.
0.
0.
0.
0.
3.
j.
^ .
0.
13
03
09
10
00
08
33
28
2?
15
3S
i';"-
1 Q
T T
' p
.-)--
i ,".
00
^^.
."' f'
Figure 5. Typical flux calculation printout
21
-------
Plume Profiles
At the time each flux calculation was made, an individual
plume profile was plotted. Figure 6 shows eight representative
plume profiles. The traverse route is shown as a straight or
curved line, and the projection line is drawn normal to the plume
axis (not shown, but the stack is indicated by a dot). The plume
profile is plotted "away" from the stack, parallel to the wind
flow. These plume profiles were useful as an editing tool and
played a major part in the analysis of the flux results.
EVENT 67
GFC
EVENT 19
COSPEC I I
EVENT * 17
COSPEC I I
Figure 6. Typical plume profiles
(continued)
22
-------
EVENT * 3S
COSPEC I I I
EVENT HH
COSPEC I I I
EVENT * 7H
COSPEC I I I
EVENT 39
COSPEC I I I
EVENT 22
COSPEC I I I
Figure 6. (continued)
23
-------
SECTION 5
RESULTS
REMOTE SENSORS
A total of 308 flux calculations were made from the three
instrument/four wind system data matrix. The results for each
instrument are presented in the following graphs and tables for
the four measurement days.
COSPEC III
The 176 S02 fluxes for the COSPEC III are presented in Fig-
ure 7 and Table 2. The averages and standard deviations for all
values are summarized below:
S02 MASS FLUX
MEAN a
DATA SETS (MT/D) (% of MEAN)
COSPEC III/LDV 78.9 40.9
COSPEC III/TS 64.9 37.0
COSPEC III/PB 68.4 38.9
The plots (Figure 7) show the individual results as connected
lines (except dots appear where more than 30 minutes passed between
traverses). The horizontal line represents the mean for each set
of results.
The tabulations (Table 2) give the day, time, traverse number,
stack distance in meters, plume width (the approximate 6a width of
the projected plume profile (in meters), the wind speed in meters
per second, and the three columns of fluxes where wind data were
available (MT/D 802). The coding (a, b, c, d, e) is discussed
under Analysis of Results.
24
-------
200
150
100
50
0
L.DV
u
3 AUG
4 AUG
OJ
o
CO
x
CO
CO
3 AUG
4 AUG
200
50
00
50
383 %
3 AUG
4 AUG
5 AUG
5 AUG
\
6 AUG
6 AUG
t
5 AUG
6 AUG
Figure 7. COSPEC III SC>2 Mass Flux Results
25
-------
TABLE 2. S02 MASS FLUX RESULTS - COSPEC III
DATE
AUG
1976
3
4
TIME
(MDT)
0929-0933
0936-0941
1020-1024
1026-1029
1032-1035
1047- 049
1057- 102
1 103- 108
1 109- 112
1 1 43- 1 46
1 147- 150
1 157- 159
0856-0901
0904-0908
0908-0913
09 1 3-09 1 6
0917-0920
0925-0929
0931-0934
0935-0938
0938-0942
0954-0959
1000-1003
1007-1010
101 1- 015
1016-1022
1022-1027
1029-1034
1034-1038
1038-1042
1047-1051
1052-1055
1055-1059
1 1 38- 1 1 4 1
1 1 41-1 146
1 149-1 154
1 1 58- 1 204
1247-1251
1257-1304
1305-1310
1316- 321
1 32 1 - 1 328
TRAV-
ERSE
NO.
1
2
3
4
5
6
7
8
9
1 1
12
13
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
STACK
DIST.
(M)
225
225
200
225
225
250
275 e_
300 e.
275 e.
275
300 e
250
200
225
225
225
225
225
225
225
225
250
225
575
575 e.
550 e
525 £_
550 e
575 <-
550 e
550 e.
550 e.
575 _
275
225^.
300 e.
275 _
250
275
250
250
300
PLUME
WIDTH
(M)
150
150
175
150
175
125
300 c.
150
175 c.
225
250
250
200
175
175
175
200
150
175
175
175
225
175
325 c
350 c
350 c
350 c
350 c
375 c
300 c
325
350 c_
325 C.
225 C
300
275
400
^450
325
150
275
400
WIND
SPEED
(M/S)
10.7
9.6
8.4
7.7
8.0
7.8
7.5
7.6
6.2
8.5
8.6
1 1.9
5.9
6.8
6.4
6. 1
5.8
6.0
5.7
5.7
5.7
4.4
4.2
3.6
3. 1
3.6
3.8
3.8
3. 1
3.0
4.0
4.0
3.7
5.9
5.9
4.3
4.6
2.9
3.4
5.5
4.9
4.9
S02 MA
LDV
121.0
80.70
SS FLUX
TS
29.0
69.4
35.5
56.5
76.7ft
87. 4«
89.2
78.6
44.4a
41 .Oa
82. la.
43. Oa
37.3
-------
TABLE 2. (continued)
DATE
AUG
1976
4
5
6
TIME
(MDT)
I 330- 1 336
I 338- 1 34 1
1342- 1 347
1444-1448
' 1449-1452
1452-1456
1456-1501
1506-1512
1533-1538
1538-1543
0924-0930
0932-0937
0937-0941
0941-0948
0948-0952
0952-0957
1000-1004
1004-1008
1008-1013
1013-101 8
1019-1023
1036-1039
1040-1044
1 1 00-1 108
1 1 16-1 120
1 121 -1 126
1 126-1 132
1 132-1 138
II 38-1 143
II 50-1 155
1 156-1201
121 1-1216
1221-1227
1308-1310
131 0-1 31 3
1314- 131 8
1319-1323
1328-1331
1 33 1 - 1 334
1351-1354
0918-0923
TRAV-
ERSE
NO.
45
46
47
49
50
51
52
54
5*.
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
80
81
82
83
84
85
86
88
94
STACK
DIST.
(M)
250
275
325
275
275
300
275
2506
475
475
225
225
225
225
225
225
225
225
225
575fi
5506
225
250
250
225
225ft
500«=,
475a
575e
450e
300e
575e
575e
250
250
300
275
275
275
325
200
PLUME
WIDTH
(M)
i450
200
300
275
225
300
1425
250
250 C
400 C
200
200
200
200
175
200
200
200
175
350 C
300
275
275
200
225
275
450 C
450 C
475
700 C
750
450 C
475 C
425
375
425
650
500
375
450
175
WIND
SPEED
(M/S)
6.2
6.1
6.5
5.2
5.0
4.3
5.8
7.1
6.7
6.6
5.1
5.2
4.3
4.3
4.6
5.2
3.7
3.9
3.8
3.4
4.3
5.3
5.5
4.5
5.4
4.4
3.6
4.2
4.2
2.5
1 .7
3.8
4.6
5.4
5.4
4.0
5.0
4.2
4.6
5.9
3.9
S02 MASS FLUX (MT/D)
LDV
77.2
56.3
67.7
1 16.0
83.3
82.5
137.1
72.5
35.2
89.6
60.9
85.0
58.7
103.8
44.8
1 16.0
77.1
110.5
54.7
62.7
87.0
81 .3
95. 7 a
39. Ob
36.8
92. Ob
104.9
86.70*
208.1
154.3
40.7
TS
60.04.
24.60.
54. 7 a.
45.0 a.
8l.3a.
59.1 a.
62.3
-------
TABLE 2. (continued)
DATE
AUG
1976
6
TIME
(MDT)
0931-0934
0934-0937
0937-0941
0941-0944
0944-0947
0947-0951
0951-0954
0957- 001
005- 008
033- 038
039- 043
043- 045
046- 050
053- 056
057- 102
103- 105
105- 110
112- 115
115- 120
120- 126
126- 129
130- 135
135- 139
143- 147
147- 151
151- 156
TRAV-
ERSE
NO.
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
1 10
1 1 1
112
1 13
1 14
1 15
1 16
1 17
1 18
1 19
120
STACK
DIST.
(M)
225
250
225
225
225
225
225
550
1 125
225
550e
225
425
225
550
250
575C
225
5506
5506
250
575C
250 e
275
250
275«.
PLUME
WIDTH
(M)
200
150
200
250
175
150
150
275
225
150
275
250
275
175
300
200
250
200
275C
3006
125
275c
175
200
150
150
WIND
SPEED
(M/S)
5.5
5.2
4.2
5.1
6. 1
7.2
7.0
6.0
6.1
6.0
7.0
7.3
5.5
5.4
6.1
5.8
6.3
5.6
6.4
5.5
5.7
6.1
5.5
6.2
6.4
7.0
S02 MASS FLUX (MT/D)
LDV
50.1
37.9
82.2
82.2
69.7
36.8
62.6
47.0
62.6
39.4
92.0
77.5
91.9
53.9
144.2
70.0
73.7
102.3
65.6
62.4
68.8
39.2
58.8
70.0
71.6
124.7
TS
40.4
43.2
71 .2
80.2
72.9
43.8
59.4
47.9
45.6
89.0
74.9
97.20.
52.8
144.20.
68.5
62.3cL
91.5
55.2
-------
COSPEC II
The 54 S02 fluxes for the COSPEC II are presented in Figure
8 and Table 3. The averages and standard deviations for all values
are summarized below:
S02 MASS FLUX
MEAN 0
DATA SETS (MT/D) (% of MEAN)
COSPEC II/LDV 63.4 28.2
COSPEC II/TS 50.6 37.3
COSPEC II/PB 51.7 32.0
The plots and tabulations are identical to the preceding'
COSPEC III presentations.
GFC
The 57 S02 fluxes for the GFC are presented in Table 4. The
averaged and standard deviations for all values are summarized
below:
S02 MASS FLUX
M~EAN a
DATA SETS (MT/D) (% of MEAN)
GFC/LDV1 204.8 20.2
GFC/TS 139.4* 50.3
GFC/PB 109.8 50.4
*Includes four night traverses
The GFC flux calculations were limited to one day only,
5 August 1976, because of an accumlulation of dust on the in-
strument mirror and other problems that occurred in the field.
The tabulation is the same as the preceding COSPEC III pre-
sentation. No plot of the GFC results is offered as they are
clearly divergent by a factor of two or more from the COSPEC/
Method 6 values.
29
-------
CN
D
JO
CO
CO
200
150
100
50
0
200
150
100
50
L-Dv/
3 AUG
TS
3 AUG
200
50
00
50
-------
TABLE 3. S02 MASS FLUX RESULTS - COSPEC II
DATE
AUG
1976
4
TIME
(MDT)
0856-0901
0904-0908
0908-0913
0913-0916
0917-0920
0925-0929
0931-0934
0935-0938
0938-0942
0954-0959
1000-1003
1007-1010
101 1-1015
1016-1022
1022-1027
1029-1034
1034-1038
1038-1042
1047-1051
1052-1055
1055-1059
1 138-1 141
1 141-1 146
1 149-1154
1 158-1204
1247-1251
1257-1304
1305-1310
131 6- 1321
1321-1328
1 330- 1 336
1338-1341
1342-1347
1444-1448
1449-1452
1452-1456
1456-1501
1506-1512
1533-1538
1538-1543
TRAV-
ERSE
NO.
15
16
17
18
19
20
21
7°
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
49
50
51
52
54
55
56
STACK
DIST.
(M)
225
225
225
225
225
225
225
225
225
225
225
575 e
575 e
550 6
475 e
525 e
5756
550 C.
550
550^
575
250
250^
300
450
500
PLUME
WIDTH
(M)
200
200
225
175
200
125
125
150
175
200
175
275 C
300 C
3000
350 c
300 C
275 c.
300 C,
325
325 C.
3000
2000
275
175
375
1450
275
I50C,
200
400
(1450
175
275
275
250
250
c*425
225
175 C.
400 C
WIND
SPEED
(M/S)
5.9
6.8
6.4
6.1
5.8
6.0
5.7
5.7
5.7
4.4
4.2
3.6
3.1
3.6
3.8
3.8
3.1
3.0
4.0
4.0
3.7
5.9
5.9
4.3
4.6
2.9
3.4
5.5
4.9
4.9
6.2
6.1
6.5
5.2
4.3
4.3
5.8
7.1
6.7
6.6
S02 MASS FLUX (MT/D)
LDV
72. 1
69.2 Ou
63.0
44.6
60.2
75.6
58.5
51.2
95.8
81.4
25.5
57.2
TS
97.2
47.3
61 .5
72.70,
80. 1 0,
80.2
65.0
39.00,
40. 1 0,
64.1
-------
TABLE 4. S02 MASS FLUX RESULTS - GFC
DATE
AUG
1976
5
TIME
(MDT)
0932-0937
0937-0941
0941-0948
0948-0952
0952-0957
1000-1004
1008-1013
1013- 10 1 8
1019-1023
1036-1039
1 100-1108
1 1 16-1 120
1 121 -1 126
1 126-1 132
1 132-1 138
1 138-1 143
1 150-1 155
1 156-1201
121 1-1216
1221-1227
1308-1310
131 0-1 31 3
1314-1318
1319-1323
1328-1331
1331-1334
1351-1354
2032-2035
2035-2038
2052-2055
2059-2102
TRAV-
ERSE
NO.
58
59
60
61
62
63
65
66
67
68
70
71
72
73
74
75
76
77
78
SO
81
82
83
84
85
86
88
89
90
92
93
STACK
DIST.
(M)
225
200
225
200
225
200
200
500
500.2,
200
250
200
200 e.
475 e
475 ^
525 £
350 «.
375^.
500 e
500fL
250
250
375
275
300
275
300
250 ^
200
50 c.
50
PLUME
WIDTH
(M)
250
250
275
350
250
275
200
250
350
375
275
350
375
425 c.
475 c
575 C
675
575
500
475
400
325
300c
525
700
400
500
275
450
350
375
WIND
SPEED
(M/S)
5.2
4.3
4.3
4.6
5.2
3.7
3.8
3.4
4.3
5.3
4.5
5.4
4.4
3.6
4.2
4.2
2.5
1.7
3.8
4.6
5.4
5.4
4.0
5.0
4.2
4.6
5.9
2.0
1.9
1.8
1.8
S02 MASS FLUX (MT/D)
LDV
140.3
128.3
176.8
167.5
222.9
169.0
28.5
223.0
250.6
120.6
120.9
372. 2Sr
207.1
94.8A-
181 .8
73.80.
406.4
601 .6
TS
157.0
164.5
204.0
176.3
186.6
151 .6
77.5
39.7
61 .0
22.7
1 18.3
244.4
42.6
I24.8a.
141. 5a
234. Od
267. 5a
212. 6a
372. 2«
139. 8a
92.2«
1 13.00.
44. 3&
147. 8a
423. 3a
99.6oL
124. 7a
85. la.
25.7a.
10.4
15.5
PB
147.0
144.8
199. 5J^
13.7
8I.5&-
89.7
108.5
93.6
32
-------
REFERENCE METHOD 6
The 25 stack sampling results are presented in Figure 9 and
Table 5. The plot is to the same scale as the remote sensor for
ease of comparison of the two sets of data. The overall average
and standard deviations are:
DATA SETS
Method 6 / Method 2
S02 MASS FLUX
MEAN a
(MT/D) (% of MEAN)
67.4
8.7
It must be noted that no corrections for moisture have been
applied to these reference method data. Any further analysis that
compares the Method 6 results with the COSPEC results should first
make the necessary correction before making the comparisons.
The "annulus" results quantify the SC>2 flux between the inner
and outer stacks; less than 1% of the S02 was found in the annulus,
200
150
i
'lOO
X
CO
CO
CM 50
o
CO
t*»T/D
3 AUG
4 AUG
5 AUG
6 AUG
Figure 9. Reference Method 6 Mass Flux
33
-------
TABLE 5. STACK SAMPLING RESULTS - EPA METHOD 6
DATE
AUG
1976
3
4
5
6
ANNULUS
4
5
6
TIME
(MDT)
1020-1 040
153- 113
1 125-1 145
1 320.- 1 340
0851-091 1
0923-0943
1005- 025
041- 101
143- 203
319- 339
358- 418
500- 520
0912-0932
0942-1002
1039-1 059
1 105- 125
1131- 15!
1210- 230
1304- 324
1343- 403
091 5-0935
0945- 005
1 015-1 035
1056-1 1 16
1 129-1 149
1 130-1 150
1240-1 300
1008-1028
1328-1 348
1 106-1 126
RUN
NO.
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
20
21
22
23
24
25
1
2
3
4
5
STACK GAS
FLOW RATE
(ScFm)
782,000
782,000
782,000
782,000
758,000
758,000
758,000
765,000
765,000
765,000
771 ,000
77 1 ,000
761 ,000
766,000
766,000
766,000
770,000
770,000
769,000
767,000
752,000
752,000
756,000
756,000
756,000
43,500
57,400
52,300
52,700
44,500
SULFUR
CONG.
(ppmv)
830.9
744.3
551 .8
856.0
822.9
840.2
835.3
846.5
812.8
885.5
837.7
856.3
820.9
847.6
899.3
776.4
866.2
825.2
807.9
773. 1
864.9
769.0
702.7
780.7
758.2
140.6
53.7
94.5
1 39.0
34.2
DIOXIDE
FLUX
(MT/D)
70.8
63.4
47.0
72.9
67.9
69.4
69.0
70.5
67.7
73.8
70.3
71.9
68.0
70.7
75.0
64.8
72.6
69.2
67.7
64.5
70.8
63.0
57.9
64.3
62.4
0.67
0.34
0.54
0.80
0.17
34
-------
ANALYSIS of RESULTS
Measurement Errors
The initial analysis of the results was an attempt to iden-
tify known errors in the collection and processing of the remote
sensor and wind data that could contribute to erroneous calculated
SC-2 flux values. Five potential errors were identified:
ERROR CODE DESCRIPTION
a Altitude of the selected wind speed was outside the
assumed plume vertical dispersion (±75 meters from
plume axis).
b Time of the selected wind velocity was outside the
time window for the traverse.
G The Wind/Road angle was greater than ±50°. (Tra-
verse road exactly normal to the plume axis is 0°.)
d Double plume measurement based on criteria of one
or more instruments returning to a 0-10% S02 opti-
cal depth reading, creating distinct double peaks.
e Traverse route was on a corner, thus increasing
the chances of non-normal plume crossing and double
plume measurement.
The wind velocity errors (a,£>) were coded into the wind sum-
mary (Table 1) and carried through to the flux summaries (Tables
2, 3, 4). The plume profile and traverse road geometry errors
(c,d,e} were coded into the flux results (Tables 2, 3, 4).
Figure 10 presents four examples of Error o (Wind/Road angle
> ±50°). These happen to occur mostly at a corner, used for tra-
versing when the wind was from the southeast. (Note the wind
arrows labelled with wind speed in meters per second and letters
identifying the source of the data.)
Figure 11 shows four examples of Error e, corner measurements.
(Though many traverses had both c and e Errors, not all did.) Two
different corners are shown. (Note different stack distance.)
Error d, double plumes, is described below.
Plume Bifurcation
Another phenomenon that must be considered in the processing
of remote sensor data is the bifurcated plume, the division of the
plume profile into two (or more) distinct peaks with a differing
degree of separation.7 Figure 12 shows four examples of bifur-
cated plumes, and Figure 13 shows two truly double plume profiles.
This was also noted in 1975 work at the same power plant.
35
-------
n
*
i-
o
UJ
D_
CO
O
O
0>
T3
O
u
JH
O
o
w
* O
. LU
z Q-
U CO
a 8
U
O
LU
CL
CO
O
O
tfl
0
S
O
O
LO
+1
A
bO
H
36
-------
eg
N
~ o
I- UJ
gQ-
oo
5 8
#
O
^
O
tu
D-
CO
O
s u
s
CO
0)
T3
O
o
!H
O
fH
J-l
w
V /
w
CD
0)
c
M
o
37
-------
o
LU
CL
CO
o
O
N _
PJ
* O
LU
t Q-
10
10
o
LU
D_
CO
O
O
o
LU
Q_
CO
o
o
0)
a
(U
P
ai
u
H
PQ
DO
H
38
-------
EVENT 7
COSPEC I i
EVENT t MS
COSPEC I I I
Figure 13. Double plumes
The problem caused by bifurcated plumes is difficulty in dis-
crimination between cases in which the stack emissions have been
physically divided by the mechanics of buoyant gases and those in
which a plume has actually been measured twice. Because of velo-
city and/or directional shear, the plume may have separated into
multiple parcels that reappear over the traverse route during a
single survey.
Plume Sorting
The errors were studied to determine their relative impact
on overall average S02 Mass Flux. Tables 6, 7, and 8 present COS-
PEC III and COSPEC II results for all four days with averages, ex-
cluding the individual Errors (a ,b ,c,d,e~) ; finally, averages ex-
cluding all four Errors are given.
The differences in the four-day S02 Mass Flux averages
caused by excluding these Errors range from -10.9 to +19.6
MT/D. No single Error caused the highest difference consis-
tently, though Errors a and e usually had more impact than
Errors a and £>. Error o (orientation of the plume and route)
had the largest influence by a slight amount on results ob-
tained from LDV winds and the second largest effect on TS
winds as judged by reduction in the standard deviation of the
average, expressed as "I". The a error (altitude of wind
data) had the greatest influence on TS data, as might be ex-
pected because of tethersonde altitude constraints. Because
39
-------
of the noncontinuous nature of the pibal and profiles, the b
Error dominated the results from this data set by a signifi-
cant amount. The difference caused by excluding all four
Errors was about equal to or less than that caused by indi-
vidual Errors, indicated an expected cancelling effect.
In order to evaluate the effectiveness of the LDV under
the desired conditions of the transport wind carrying the
plume near the instrument, when the plume was directed toward
the corner northwest of the plant, the times were preselected.
These traverses had been highlighted as having a potential
Error, e. The average of the twelve mean flux determinations
with COSPEC III under these conditions is 70.11 (±37.2%)MT/D.
The standard deviation of this subset, ±37.2%, is smaller than
any of the other LDV subsets shown in Table 6, indicating
that, even though it is a smaller sample, it is more homo-
geneous than the rest of the determinations by LDV winds.
Looking at the seven Method 6 determinations in the same
time frame as the e traverses, the average SC>2 flux is 66.2
(±8.1%) MT/D. Thus the average of twelve e. traverses is
within 6% of the seven Method 6 determinations -- the best
agreement with Method 6 of any subset of the field measure-
ments .
Additional error analyses were performed. Correlations
were sought between traverse time, wind speed, and plume width
and excessively high flux values; none were found.*
*Note added in review: Millan Millan suggested that error analyses might
extend to consideration of "plume aspect", i.e.: Whether or not it was a
cohesive or non-cohesive plume, in light of the unstable condition during
the study. Such conditions may well require a larger number of profiles
in a set to improve the correlation with stack measurements.
40
-------
TABLE 6. ANALYSIS of RESULTS - LDV. WINDS
TRAVERSES
INCLUDED
ALL
ALL EXCEPT
a ERRORS
ALL EXCEPT
b ERRORS
ALL EXCEPT
a ERRORS
ALL EXCEPT
d ERRORS
ALL EXCEPT
e ERRORS
ALL EXCEPT
a,b3c3d3e
ERRORS
NO.
59
56
57
51
57
47
42
COSPEC 1 1 1
AVG S02 FLUX (MT/D)
78.9 (±40.951)
78.5 (±42.5?)
79.4 (±41 .2%)
82.2 (±39.8?)
77.9 (±41.2?)
81.2 (±41.7?)
79.6 (±43.6?)
NO.
12
11
10
10
10
11
6
COSPEC 1 1
AVG S02 FLUX (MT/D)
63.4 (±28.2?)
62.8 (±29.8?)
None - See ALL
67.8 (±14.8?)
60.2 (±27.1?)
61.7 (±28.8?)
61.4 (±18.2?)
Error Codes:
a - altitude of wind data outside limits (±75m)
b - time of wind data outside limits
c - plume axis/traverse route angle outside +50° limits
d - double plume measured
e - plume traverse on corner
41
-------
TABLE 7. ANALYSIS of RESULTS - TS WINDS
TRAVERSES
INCLUDED
ALL
ALL EXCEPT
a ERRORS
ALL EXCEPT
b ERRORS
ALL EXCEPT
c ERRORS
ALL EXCEPT
d ERRORS
ALL EXCEPT
e ERRORS
ALL EXCEPT
a^b^c^d^e
ERRORS
NO.
87
39
87
68
84
58
36
COSPEC 1 1 1
AVG S02 FLUX (MT/D)
64.4 (±37.0$)
64.6 (±33.4$)
See ALL
66.9 (±35.7$)
64.4 (±37.0$)
66.7 (±36.5$)
65.4 (±39.3$)
NO.
31
5
31
21
28
20
5
COSPEC I I
AVG S02 FLUX (MT/D)
50.6 (±37.3$)
70.2 (±21.2%)
See ALL
53.9 (±37.5$)
49.6 (±37.4$)
54.7 (±39.2$)
70.2 (±27.2$)
Error Codes:
-------
TABLE 8. ANALYSIS of RESULTS - PB WINDS
TRAVERSES
INCLUDED
ALL
ALL EXCEPT
a ERRORS
ALL EXCEPT
b ERRORS
ALL EXCEPT
c ERRORS
ALL EXCEPT
d ERRORS
ALL EXCEPT
e ERRORS
ALL EXCEPT
a^^Oyd^B
ERRORS
NO.
30
31
9
28
28
24
6
COSPEC 1 1 1
AVG S02 FLUX (MT/D)
68.4 (±38.9)
None See ALL
63.9 (±18.2*)
68.5 (±40.3?)
68.4 (±40.0*)
66.5 (±38.3*)
64.3 (±21.5)
NO.
11
3
3
10
9
11
2
COSPEC 1 1
AVG S02 FLUX (MT/D)
51.7 (±32.0$)
None See ALL
42.1 (±13.4?)
53.0 (±31.8*)
51.8 (±35.2*)
See ALL
40.8 (±17.9*)
Error Codes:
a -
b -
c -
d -
e -
altitude of wind data outside limits (±75m)
time of wind data outside limits
plume axis/traverse route angle outside +50° limits
double plume measured
plume traverse on corner
43
-------
SECTION 6
COMPARISON of RESULTS
TIME-AVERAGED RESULTS
The COSPEC III results were chosen for comparison with the
Method 6 data because there was a larger data base for statisti-
cal analysis. The sets of values were first prepared by aver-
aging over common time intervals. The 20-minute Method 6 runs
defined the time intervals for which COSPEC III fluxes were av-
eraged; two to five traverses were averaged for each Method 6
test. The 20-minute average are tabulated in Tables 9, 10, and
11. There are seven to eighteen resulting sets of 20-minute av-
erages, depending on which of the four sets of wind data were
used.
Similarly, 60-minute averages were calculated for three to
thirteen COSPEC III traverses. These are listed in Tables 12 and
13. There are seven to twelve resulting sets of 60-minute aver-
ages, depending on which wind data were used. No 60-minute aver-
ages were done for the pibal wind data because of insufficient
data.
(These averages were calculated without making any correc-
tion for moisture in the stack sampling S02 mass fluxes in this
first-order analysis.)
These five tables reveal considerable variability in the S02
mass flux results with respect to the three wind measuring sys-
tems. The individual 20-minute remote sensing averages differ up
to +95%. The greatest differences occur in the LDV data (repro-
cessed), followed by the tethersonde (TS) results. The pibal (PB)
results have the best agreement with Method 6 among the 20-minute
averages. The range of the percentage differences for each data
set (after a single worst value was discarded) are summarized be-
low:
DIFFERENCES BETWEEN REMOTE and STACK S02 FLUXES
DATA SET 20-Minute Averages 60-Minute Averages
COSPEC III/LDV +421, -451 +241, -19%
COSPEC III/TS +31%, -54% +18%, -20%
COSPEC III/PB +23%, -17%
44
-------
TABLE 9. COMPARISON of RESULTS
COSPEC III/LDV vs METHOD 6 - 20-MINUTE AVERAGES
DATE
AUG
1976
3
4
5
6
REMOTE SENSING STACK SAMPLING
TIME
(MDT)
1020-1032
' 1057-1 1 12
II 43-1 150
0856-0916
0925-0942
1007-1027
1047-1059
1 141 -1 158
1316-134!
0924-0937
0941-1004
1040-1 108
1 1 00-1 132
1 132-1 155
121 1-1227
1308-1323
0918-0941
0944-1008
1057-1 120
1 130-1 156
NO. OF
TRAVERSES
3
2
2
4
4
4
3
3
3
2
4
2
4
3
2
4
4
5
5
5
SOa FLUX
(MT/D)
86.0
73.0
85.4
58.7
81 .7
37.9
122.9
52.7
55.7
91 .2
72.9
TIME
(MDT)
1020-1040
1053-1 1 13
1 125-1 145
0851-091 1
0923-0943
1005-1025
1 041-1 101
1 143-1203
1319-1339
0912-0932
0942-1002
1039-1059
1 105-1 125
1 131-1 151
1210-1230
1304-1324
0915-0935
0945-1005
1056-1 1 16
1 129-1 149
RUN
NO.
1
2
3
5
6
7
8
9
10
13
14
15
16
17
18
19
21
22
24
25
S02 FLUX
(MT/D)
70.8
63.4
47.0
67.9
69.4
69.0
70.5
67.7
73.8
68.0
70.7
75.0
64.8
72.6
69.2
67.7
70.8
63.0
64.3
62.4
45
-------
TABLE 10. COMPARISON of RESULTS
COSPEC III/TS vs METHOD 6 - 20-MINUTE AVERAGES
DATE
AUG
1976
3
4
5
6
REMOTE SENSING STACK SAMPLING
TIME
(MDT)
1020-1032
1057-1 | 12
1 143-1 150
0856-0913
0935-0942
1007-1027
1047-1059
1 141-1 158
1 32 1 - 1 34 1
0924-0937
0941-1008
1040-1 108
1 1 00- 1 1 32
1 132- II 55
121 1-1227
J 308- 1323
0918-0941
0944-1001
1057-1 120
1 130-1 151
NO. OF
TRAVERSES
3
2
2
3
2
4
3
3
2
2
5
2
4
3
2
4
4
4
5
4
S02 FLUX
(MT/D)
53.8
82. 1
52.8
67.6
56.0
27.5
82.7
83.0
44.9
66.3
93.2
31 .9
69.6
49.5
56.0
84.3
52.9
TIME
(MDT)
1020-1040
1053-1 1 13
1 125-1 145
0851-091 1
0923-0943
1005-1025
104!-! 101
1 143-1203
1319-1339
0912-0932
0942-1002
1039-1059
1 105-1 125
II3I-I 151
1210-1230
1304-1324
0915-0935
0945-1005
1056-1 1 16
1 129-1 149
RUN
NO.
1
2
3
5
6
7
8
9
10
13
14
15
16
17
18
19
21
22
24
25
S02 FLUX
(MT/D)
70.8
63.4
47.0
67.9
69.4
69.0
70.5
67.7
73.8
68.0
70.7
75.0
64.8
72.6
69.2
67.7
70.8
63.0
64.3
62.4
46
-------
TABLE 11. COMPARISON of RESULTS
COSPEC III/PB vs METHOD 6 - 20-MINUTE AVERAGES
DATE
AUG
1976
3
4
5
6
REMOTE SENSING STACK SAMPLING
TIME
(MDT)
1020-1035
131 6- 131 4
1444-1501
0924-0948
1036-1044
1 121-1138
0947-1001
NO. OF
TRAVERSES
3
4
4
4
2
3
3
SOa FLUX
(MT/D)
67.8
61 .5
76.3
83.3
38.0
63.4
53.1
TIME
(MDT)
1020-1040
1319-1339
1500-1520
0912-0932
1039-1059
1 105-1 125
0945-1005
RUN
NO.
1
10
12
13
15
16
22
S02 FLUX
(MT/D)
70.8
73.8
71 .9
68.0
75.0
64.8
63.0
TABLE 12. COMPARISON of RESULTS
COSPEC III/LDV vs METHOD 6 - 60-MINUTE AVERAGES
DATE
AUG
1976
3
4
5
6
REMOTE SENSING STACK SAMPLING
TIME
(MDT)
0936-1035
1047-1 150
1 103-1 159
0856-0959
1001-1059
1052-1 154
1444-1543
0924-1004
1036-1 132
1 126-1227
0918-1008
1033-1 135
1057-1 156
NO. OF
TRAVERSES
4
5
5
10
1 1
5
6
7
6
3
10
13
12
S02 FLUX
(MT/D)
79.9
78.0
82.0
57.2
57.2
75.5
79.3
TIME
(MDT)
1020-1040
1053-1 145
1053-1 159
0856-0942
1005-1 101
1041-1203
1500-1520
0912-1002
. 1039-1125
1 131-1230
0915-1005
101 5-1 1 16
1056-1 149
RUN
NO.
1
2,3
2,3
5,6
7,8
8,9
12
13,14
15,16
17,18
21,22
23,24
24,25
S02 FLUX
(MT/D)
70.8
55.2
55.2
68.7
69.8
69.1
71 .9
69.4
69.9
70.9
66.9
61.1
63.4
47
-------
TABLE 13. COMPARISON of RESULTS
COSPEC III/TS vs METHOD 6
- 60-MINUTE AVERAGES
DATE
AUG
1976
3
4
5
6
REMOTE SENSING STACK SAMPLING
TIME
(MDT)
0936-1035
1047-1 150
1 103-1 159
0856-0959
1001-1059
1052-1 154
1257-1347
0924-1023
1036-1 138
1 126-1227
1308-1354
0918-1001
1033-1 135
1057-1 151
NO. OF
TRAVERSES
4
5
5
6
1 1
5
4
1 1
7
7
7
9
13
1 1
SO 2 FLUX
(MT/D)
-
-
69.1
55.7
62.1
43.3
77.7
67.7
76.3
71.2
55.8
72.0
68.1
TIME
(MDT)
1020-1040
1053-1 145
1053-1 159
0856-0942
1005-1 101
1041-1203
1319-1339
0912-1002
1039-1 125
1 131-1230
1304-1403
0915-1005
101 5-1 1 16
1056-1 149
RUN
NO.
1
2,3
2,3
5,6
7,8
8,9
10
13,14
15,16
17,18
19,20
21.22
23,24
24,25
S02 FLUX
(MT/D)
70.8
55.2
55.2
68.7
69.8
69.1
73.8
69.4
69.9
70.9
66. 1
66.9
61 . 1
63.4
The LDV wind data corrected most of the bias, which would
have been part of the LDV data set. However, the tethersonde
(TS) results have generally better agreement with the stack sam-
pling values, and the pibal (PB) results are an improvement on
all three other wind measuring systems for 20-minute averages.
MEANS and DIFFERENCES
Further analysis of the 20-minute and 60-minute averages
elucidated the relative accuracy of the remote sensing mass
fluxes. In this simple, first-order analysis, the means of the
time-averaged data were calculated; the percent differences from
the stack sampling averages over the same time period were also
figured. These means reduce all of the results to single values;
they represent seven to twelve hours of measurements or 25 to 75
traverses.
The results of this analysis are tabulated in Tables 14 and
15. These two tables offer the most succinct summary of the fin-
dings of this study. By comparing the two columns of S02 mass
flux numbers and studying the third column of percent differences
it is clear that:
48
-------
The LDV wind data were a significant improvement over
LDV values, showing agreement to the reference method
within +7.6% (and 6% for winds over the instrument).
The TS wind data had negative differences up to -9.61.
The PB showed the closest agreement (for 20-minute
averages only) at +1.0%.
TABLE 14. MEANS and DIFFERENCES of 20-MINUTE AVERAGES
S02 MASS FLUX (MT/D)
REMOTE STACK DIFF. (%)
- COSPEC III/LDV 73.4 68.2 + 7.6
COSPEC III/TS 62.0 68.6 - 9.6
COSPEC III/PB 70.3 69.6 + 1.0
TABLE 15. MEANS and DIFFERENCES of 60-MINUTE AVERAGES
S02 MASS FLUX (MT/D)
REMOTE STACK DIFF. (%)
COSPEC III/LDV 72.7 67.6 + 7.5
COSPEC III/TS 65.4 68.1 - 4.0
It is apparent that the three different sets of wind data
(LDV, TS, and PB) produced S02 mass fluxes in agreement with
Method 6 within ±101 when considering long-term (7-12 hour) aver-
ages. This reinforces earlier studies1'7; individual COSPEC de-
rived flux calculations are not representative of the true value,
but time-averaged data are within +10% of the accepted reference
stack sampling method.
The obvious conclusion is that, the longer the averaging
time, the better the remote sensing results. Thus, if only 20-
minute remote sensing tests of two to five traverses are used,
the expected spread in the results would be about ±35%; for 60
minutes of testing (four to thirteen traverses) the results would
have a spread of about ±20%. And, if seven to twelve hours of
data are gathered (25 to 75 traverses), the difference is reduced
to about ±10%.
49
-------
Further statistical analyses may sharpen the assessment of
relative accuracy by expressing the differences between the remote
and in-stack methods in terms of error intervals and confidence
limits.
WIND MEASUREMENT ACCURACY
The analysis in Section 5 showed the good agreements of LDV
related flux values to Method 6 determinations if the plume being
measured by the COSPEC was transported toward the LDV monitoring
site. This improved agreement indicates that the conditions in
which the COSPEC and LDV are in the same sector as the plume are
the best for reliable measurements. These measurements will be
improvements of the use of pibal winds.
The reason for the need for the COSPEC and LDV to be close
together to produce good results arises from the fact that the
LDV is essentially a point monitor of wind velocity aloft, as
opposed to pibal, which determines more of a velocity average
over an altitude range between readings. If the point measure-
ments (even those of the tethersonde) are not made near the plume,
they will not reflect the transport winds in the plume accurately
enough under mid-day turbulent conditions. The averaged pibal
data thus can do a better job of approximating the average trans-
port conditions, even though they may not be in the same sector
as the plume.
The expense of the LDV system would be justified if it could
be capable of readily responding to changes in wind direction in
a mobile sense. It would be sufficient to have it relocatable
and not necessarily capable of measurements while moving.
50
-------
REFERENCES
Sperling, R.B., Evaluation of Upward-Looking Spectrometers
as S02 Mass Emission Monitors, Technical Report presented
to U.S. EPA by Environmental Measurements, Inc. in
partial fulfillment of P.O.* DA-6-99-5860A, Sept. 30, 1976.
Sperling, R.B., Evaluation of the Correlation Spectrometer
as an Area S02 Monitor, EPA Report #600/2-75-077,
October 1975.
Entropy Environmentalists, Inc., Source Sampling Report:
A Western U.S. Power Plant, Performed for U.S. EPA under
Contract No. 68-01-3172, August 1976.
Krause, M.C. et al., Evaluation of LDV Techniques for
Remote Wind Velocity Measurements, Prepared for U.S. EPA
by Lockheed Co. under Contract No. 68-02-2415, October 1976.
Intera Environmental Consultants Ltd., Summary Report:
Meteorological Data from a Tethered Balloon, Prepared for
U.S. EPA under Contract No. DA6-99-6644A, September 1976.
Stern, A. C., Air Pollution, Vol. 13 Air Pollution & Its
Effects, Academic Press, 1968.
Millan, Gallant, § Turner, The Application of Correlation
Spectroscopy to the Study of Dispersion for Tall Stacks,
Atmospheric Environment, Vol. 10, pp 499-511, January 1976.
Millan, M. M. , Technical Note, A Note on the Geometry of
Plume Diffusion Measurements, Atmospheric Environment, Vol 10,
pp 665-658, February 1976.
Varey, et. al., Plume Dispersion & SO2 Flux Measurements at
Drax Power Station, England, presented at Correlation Spect-
roscopy Conference, Toronto, Canada, June 1977.
51
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
"'.-600/2-79-094
2.
3. RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE
ACCURACY OF REMOTELY SENSED S02
MASS EMISSION RATES
5. REPORT DATE
May 1979
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
R. B. Sperling, M. A. Peache, and W. M. Vaughn
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Environmental Measurements, Inc.
215 Leidesdorff Street
San Francisco, California 94111
10. PROGRAM ELEMENT NO.
1AD712 BA-040 (FY-78)
11. CONTRACT/GRANT NO.
68-02-2711
12. SPONSORING AGENCY NAME AND ADDRESS
Environmental Sciences Research Laboratory - RTF
Office of Research and Development
U. S. Environmental Protection Agency
Research Triangle Park, N. C. 27711
13. TYPE OF REPORT AND PERIOD COVERED
Final Report 6/76-12/78
14. SPONSORING AGENCY CODE
EPA/600/09
15. SUPPLEMENTARY NOTES
16. ABSTRACT
Remote sensing data of single-stack power plant emissions and local wind speed have
been analyzed to determined S02 mass flux for comparison with EPA referenced methods.
Four days of S02 data were gathered from a moving platform by three upward-viewing
remote sensors two ultraviolet absorption spectrometers and an infrared gas filter
spectrometer. Wind velocity data were gathered by a laser-doppler velocimeter (LDV);
supplemental data were obtained from a tethered balloon (telemetered) and pilot
balloons (optical theodolite). The data matrix (S02, X-Y position, wind velocity for
120 traverses) was computer processed; the end result was the S02 mass flux derived
from the remote sensing data. Comparisons were made between these S02 fluxes (aver-
ages for 20 minutes and 60 minutes) and those derived from in-stack measurements. The
results of the comparisons show the relative accuracy of the remote sensing technique
for quantifying S02 mass emission rates. The analysis shows that as averaging time
increases from 20 minutes to 12 hours the difference between the remotely measured
S02 mass flux and the stack sampling S02 mass flux decreases from about ±35% to ±10%.
In general, no single wind measuring system produced superior results over te other
two. The LDV and COSPEC, however, produced the best agreement with Method 6 (+6%)
when the plume was transported near the LDV instrument.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
c. COSATl Field/Group
* Air pollution
* Sulfur dioxide
* Remote sensing
Weight (mass)
Emissions
Rates (per time)
* Accuracy
Thermal power plants
Ultraviolet spectrometers
Infrared spectrometerjs
Wind velocity
13B
07B
14B
10B
04 B
18. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19. SECURITY CLASS {This Report)
UNCLASSIFIED
21. NO. OF PAGES
60
20. SECURITY CLASS /This page)
UNCLASSIFIED
22. PRICE
EPA Form 2220-1 (Rev. 4-77)
PREVIOUS l^
ISOBSOLETE
52
------- |