EPA/450/5-79/009
EPA-450/5-79-009
September 1Q7J
States Office of Air Quality
imental Protection Planning and Standards
' Research Triangle Park NC 27711
Empirical Studies of the
Relationship Between
Emissions and Visibility
in the Southwest
-------
EPA-450/5-79-009
Empirical Studies of the Relationship
Between Emissions and Visibility
in the Southwest
X
by
Marilyn Marians and John Trijonis
Technology Service Corporation
Route 7, Box 124-K,
Santa Fe, New Mexico 87501
RECCN tf (j?Rn
AGENCY
1445 ROSS AVENUE
', TEXAS 75m
Grant No. 802815
EPA Project Officers: John Bachmann (OAQPS),
William E. Wilson, and Thomas G. Ellestad (ESRL)
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Air, Noise, and Radiation
Office of Air Quality Planning and Standards
Research Triangle Park, North Carolina 27711
September 1979
-------
This report has been reviewed by the Strategies and Air Standards Division
of the Office of Air Quality Planning and Standards, EPA, and approved for
publication. Approval does not signify that the contents necessarily reflect
the views and policies of EPA. Mention of trade names or commercial products
is not intended to constitute endorsement or recommendation for use. Copies
of this report are available through the Library Services Office (MD-35) ,
U. S. Environmental Protection Agency, Research Triangle Park, N. C.
27711, or from National Technical Information Services, 5285 Port Royal
Road, Springfield, Virginia 22161.
Publication No. EPA-450/5-79-009
11
-------
ABSTRACT
Historical emission trends of SOX, NOX, and NMHC are determined year-
by-year from 1948 to 1975 for four Southwestern states (Arizona, Colorado,
Nevada, and Utah). Trends in visibility levels (medians and other percen-
tiles) are documented for the period 1948-1976 at 12 airports in the South-
west.
Two analyses are used to relate emission changes to variations in
regional extinction levels. The first analysis examines the air quality
changes associated with a 90% reduction of Southwestern SOx emissions dur-
ing a nine-month copper strike and estimates the extinction produced by
SOx emissions on various spatial scales. The second analysis involves
regression studies relating historical changes in emissions to historical
extinction levels from 1948-1975. Because of limitations in the analytical
methods, there is a high degree of uncertainty in many of the results.
However, the studies do provide insights into the effects of aerosol pre-
cursor emissions on extinction at various distances from sources. In the
case of mesoscale effects of SOx in the Southwest, quantitative coefficients
are proposed which link emissions to regional extinction.
This report was submitted to the U.S. Environmental Protection Agency
by Technology Service Corporation in partial fulfillment of Grant 802815.
Work on this project started in May 1978 and was completed in March 1979.
-------
-------
CONTENTS
ABSTRACT iii
FIGURES vii
TABLES ix
1. INTRODUCTION AND SUMMARY 1
Summary of Conclusions 3
Future Work 5
2. HISTORICAL EMISSION TRENDS IN THE SOUTHWEST 7
General Magnitude of Emissions 7
Historical Changes in Emissions 8
Primary Sulfate Emissions 21
Emission Trends for Individual Air Basins 22
3. HISTORICAL VISIBILITY TRENDS IN THE SOUTHWEST 23
Changes in Observation Locations 23
Historical Visibility Trends 24
4. REGIONAL EXTINCTION/EMISSION COEFFICIENTS BASED ON
THE 1967-1968 COPPER STRIKE 42
Methodology 42
Extinction/Emission Coefficients for SOX 48
5. REGIONAL EXTINCTION/EMISSION COEFFICIENTS BASED ON
HISTORICAL EMISSION CHANGES FROM 1948 TO 1975 55
Methodology 55
Regression Analysis for Arizona Sites 57
Regression Analyses for Other Locations 65
6. CONCLUSIONS AND ILLUSTRATIVE APPLICATIONS 70
Extinction/Emission Coefficients for the Southwest 70
Illustrative Applications 72
REFERENCES 74
-------
APPENDIX A 75
APPENDIX B 94
APPENDIX C 100
APPENDIX D 106
-------
FIGURES
Number Page
1. Sulfur oxide emission trends in Arizona, 1948-1975 9
2. Sulfur oxide emission trends in Colorado, 1948-1975 10
3. Sulfur oxide emission trends in Nevada, 1948-1975 11
4. Sulfur oxide emission trends in Utah, 1948-1975 12
5. Nitrogen oxide emission trends in Arizona, 1948-1975 13
6. Nitrogen oxide emission trends in Colorado, 1948-1975 .... 14
7. Nitrogen oxide emission trends in Nevada, 1948-1975 15
3. Nitrogen oxide emission trends in Utah, 1948-1975 16
9. Nonmethane hydrocarbon emission trends in Arizona,
1948-1975 17
10. Nonmethane hydrocarbon emission trends in Colorado,
1948-1975 18
11. Nonmethane hydrocarbon emission trends in Nevada,
1948-1975 19
12. Nonmethane hydrocarbon emission trends in Utah,
1948-1975 20
13. Long-term visibility trends at Phoenix 26
14. Long-term visibility trends at Tucson 27
15. Long-term visibility trends at Denver 28
16 Long-term visibility trends at Salt Lake City 29
17. Long-term visibility trends at Fort Huachuca 30
18. Long-term visibility trends at Prescott 31
19. Long-term visibility trends at Winslow 32
20. Long-term visibility trends at Colorado Springs 33
-------
Number Page
21. Long-term visibility trends at Grand Junction 34
22. Long-term visibility trends at Pueblo 35
23. Long-term visibility trends at Ely 36
24. Long-term visibility trends at Cheyenne 37
25. Airport weather stations and NASN monitoring sites used
to determine air qulaity changes during the 1967-1968
copper strike 43
26. Seasonally adjusted changes in extinction during the
copper strike, units of (104 meters)-!, compared to
the geographical distribution of smelter SOX emissions. ... 45
27. Seasonally adjusted changes in sulfate during the copper
strike, units of yg/m3, compared to the geographical
distribution of smelter SOX emissions 46
28. Extinction/emission coefficients based on airport data
during the 1968-1968 copper strike 49
29. Sulfate/emission coefficients based on NASN data
during the 1967-1968 copper strike 50
30. Extinction/emission coefficients from airport and NASN
data during the 1967-1968 copper strike 51
31. Log-log plot of extinction/emission coefficient versus
distance to smelters 53
32. Tucson time series data on extinction and hours of
restricted visibility versus historical smelter SOX
emissions for Arizona 60
33. Phoenix time series data on extinction and hours of
restricted visibility versus historical smelter SOX
emissions for Arizona 61
34. Yearly median extinction at Tucson versus smelter
SOX emissions for Arizona 62
35. Yearly median extinction at Phoenix versus smelter
SOX emissions for Arizona 63
vim
-------
TABLES
Number Page
1.
2.
3A.
3B.
3C.
4.
5.
6.
7.
Magnitude of Precursor Emissions in the Southwest,
Tons/Day, 1970-1972
Analysis of the Statistical Significance of Observation
Site Relocations on Reported Airport Visibilities
Net Percent Changes in Visibility, 1953-1955 to 1970-
1972, Original Analysis (Trijonis and Yuan, 1978a) ....
Net Percent Changes in Visibility, 1953-1955 to 1970-
1972, All Site Relocations Discounted
Net Percent Changes in Visibility 1953-1955 to 1970-
1972, All Significant Site Relocations Discounted
Pairing of Airports and NASN Sites with Smelters
Most Likely to Affect Those Sites
Intercorrelation Among Statewide Total Emission Trends,
1948-1975
Historical Extinction-Emission Correlations for
Arizona Trend Sites
Regressions of Historical Extinction Levels Versus
Historical Smelter SOx Emission Levels
8
. 25
. 39
. 40
. 41
. 47
. 56
. 58
. 59
-------
-------
CHAPTER 1
INTRODUCTION AND SUMMARY
Several previous studies have developed models of visibility degrada-
tion based on empirical analysis of ambient data (Trijonis and Yuan, 1978a,
b; White and Roberts, 1977; Cass, 1976). These empirical models involve mul-
tiple regression equations relating extinction levels to corresponding air
quality levels. The key parameters determined by the regressions are ex-
tinction coefficients per unit mass for individual aerosol species (sulfates,
nitrates, organics, and remainder of TSP); these coefficients allow the
computation of extinction budgets allocating visibility loss among the
various aerosol components. The results of the empirical studies have
proven to be consistent with one another and with theoretical predictions
based on the Mie theory of aerosol optics (Trijonis and Yuan, 1978a, b;
Latimer et al, 1978; Ursenbach et al, 1978; White and Roberts, 1977; Waggoner
et al, 1976).
The catch with the existing empirical models is that, even if we know
with confidence the extinction coefficients per unit mass for each aerosol
species, we need an additional model to predict aerosol concentrations from
emissions. That is -- we still must address the problem of determining
secondary aerosol concentrations from gaseous precursor emissions and
primary aerosol concentrations from particulate emissions. This latter
problem is itself quite complex and difficult, especially in the case of
secondary aerosols (e.g. sulfates and nitrates), especially for the pheno-
menon of regional haze (as opposed to an isolated plume), and especially in
situations of complex terrain (as is often found in the Southwest United
States).
This report uses a new type of empirical model to address the problem
of regional haze caused by secondary aerosols in the Southwest. The
novelty of the approach is that we attempt to relate extinction not to am-
bient aerosol data, but directly to emissions of aerosol precursors. The
key outputs of the analysis are extinction coefficients per unit emissions
4-1 *
of aerosol precursors, with units of (10 meters) /(1000 t.ons prr d.iy).
Hereafter abbreviated as (104 m)"1/(1000 TPD).
-------
We attempt to derive these coefficients for SOX, NOX, and NMHC emissions on
various spatial scales, ranging from the air basin scale (£ 50 miles from
the source) to the mesoscale (^ 50-200 miles from the source) to the synoptic
scale (^ 200-500 miles from the source). Once they are determined, these
extinction/emission coefficients can serve as the basis for first approx-
imation calculations concerning the impact of emissions on regional haze in
the Southwest.
Two analyses are conducted in attempting to derive extinction/emission
coefficients for regional haze in the Southwest. The first analysis is
based on visibility and sulfate changes during the 1967-1968 copper strike,
when SOx emissions in the Southwest were reduced over 90% for a nine month
period. The second analysis is based on regressions relating historical
trends in SOx, NOx, and NMHC emissions to corresponding trends in airport
visibility data. Because of limitations in these analyses (especially in
the regression studies), a high degree of uncertainty exists in our results.
In fact, the only extinction/emission coefficient in which we have a fair
degree of confidence is the mesoscale coefficient for sulfur oxides. Sever-
al of our results, however, should serve as useful checkpoints for future
models of regional haze in the Southwest.
This report is organized in six chapters. The remainder of this
chapter summarizes our conclusions and recommends future research areas.
Chapter 2 presents detailed, year-by-year emission trends from 1948 to 1975
for SOx. N0xğ and NMHC in the four study states (Arizona, Colorado, Nevada,
and Utah) for which we have data on ambient visibility trends. Chapter 2
is supported by three appendices: Appendix A -- explaining the bases for the
emission trend calculations; Appendix B -- discussing the issue of primary
sulfate emissions from copper smelters and power plants; and Appendix C --
explaining how air basin emission trends can be computed from statewide
emission trends. Chapter 3 updates our previous visibility trend study for
the Southwest (Trijonis and Yuan, 1978a) and presents visibility trends from
1948 to 1976 at twelve locations. Chapter 4 uses data during the 1967-1968
copper strike to estimate regional extinction/emission coefficients for SOX.
Chapter 5 performs regression studies relating historical visibility trends
-------
at various locations to historical emission trends. Chapter 6 discusses
our conclusions and illustrates how extinction/emission coefficients might
be applied to estimate the impact of emission sources on regional haze.
SUMMARY OF CONCLUSIONS
Historical Emission Trends in the Southwest (Chapter 2)
Copper smelters have historically been the overwhelming source
of SOx emissions and, on a tonnage basis, the major source of
aerosol precursor emissions in the Southwest. Total smelter
SOX emissions in the Southwest approximately doubled from 1948
to 1973 as increased copper production outdistanced the effects
of the air pollution controls that were installed on a few
smelters. The recent proliferation of air pollution controls
among the Arizona smelters (the dominant group of smelters in
the Southwest)helped to reduce Arizona smelter SOx by 38% from
1973 to 1975 and by 66% from 1973 to 1977. Power plants have
recently been a rapidly growing source category for SOx and
now rank second to copper smelters in terms of total Southwest
SOx emissions.
NOx emissions have climbed steadily in the Southwest, with an
especially rapid rise in the late 1960's and early 1970's.
Total NOx emissions increased by a factor of 5 from 1948 to
1975. The two major source categories for NOx, power plants
and gasoline vehicles, account for nearly all of this rising
trend.
Total NMHC emissions in the Southwest increased by a factor of
3 from 1948 to 1972 and then started a gradual decline. This
basically reflects the trend in the dominant source category
for NMHC, gasoline vehicles.
Historical Visibility Trends in the Southwest (Chapter 3)
Visibility exhibited a decreasing trend in the Southwest from
the middle 1950's to the early 1970's. Nearly all of the 12
trend study sites showed a net decline in visibility on the
order of 10 to 30% from 1953-1955 to 1970-1972. This decrease
in visibility is equivalent to an increase in extra extinction
(above-and-beyond blue-sky scatter) on the order of 10 to 70%
(Trijonis and Yuan, 1978a).
Since the early 1970's, the declining trend in visibility has
apparently reversed. Several trend sites show an increase in
median visibility on the order of 5 to 15% from 1970-1972 to
1974-1976.
-------
Regional Extinction/Emission Coefficients Based on the 1967-1968
Copper Strike (Chapter 4T
t The nine-month copper strike of 1967-1968 was accompanied by
large decreases in sulfates and increases in visibility in
many parts of the Southwest. Statistically significant drops
in sulfate occurred out to distances greater than 300 miles
from the smelters, and statistically significant increases in
visibility occurred out to distances greater than 150 miles
from the smelters. Variations in weather were apparently not
a factor contributing to these air quality improvements
(Trijonis and Yuan, 1978a).
Both the airport visibility data and NASN sulfate data during
the copper strike imply that the mesoscale extinction/emission
coefficient for SOX is .01 to .03 (10* m)-l/(1000 TPD). In
other words, on a scale of 50 to 200 miles from the source,
each 1000 TPD SO/ was apparently associated with an extinction
of .01 to .03 (104 m)-l. The extinction/emission coefficients
based on the airport data may be slightly overestimated (on
the order of 10-30%) because of the neglect of concurrent
changes in primary particles during the copper strike.
The copper strike analysis also provided some evidence in sup-
port of the following conclusions: (1) regional extinction pro-
duced by an SOX emission source in the Southwest may tend to be
inversely proportional to distance from the source; (2) at dis-
tances 250-375 miles from the source (within the synoptic
scale), the extinction/emission coefficient may be approximately
.01 (104 m)-l/(1000 TPD SOX); and (3) at distances within 10
to 15 miles from the source (within the air basin scale), the
extinction/emission coefficient may be as high as .1 to .25
(104 m)-I/(1000 TPD SOX). As indicated by the qualified word-
ing, however, we regard these latter three conclusions as very
tenuous.
Regional Extinction/Emission Coefficients Based on Historical
Emission Changes from 1948 to 1975 (Chapter 5)
There are several important limitations to the regression ana-
lysis relating yearly extinction levels (based on airport data)
to yearly emission levels from 1948 to 1975. One of the most
important drawbacks is the intercorrelation that exists among
the "independent" emission variables. The historical emission
variables (SOX, NOX, and NMHC trends for various states)
typically correlate with one another at levels of R - .50 to
.85 (correlation coefficients). This intercorrelation and the
limited number of data points (28 years) make it extremely dif-
ficult to isolate the individual effects of all three emission
-------
variables (SOX, NOX, and NMHC). Another important drawback
is the lack of data on primary participate emission trends.
The regression analysis for Arizona is conducted using six data
bases (four airports, but two time periods for two of the air-
ports). In all six cases, historical extinction levels cor-
relate best with Arizona smelter SOX emissions, the R values
ranging from .68 to .91. Multiple regressions select Arizona
smelter SOX as (by far) the most significant variable in all
six cases and as the only significant variable in five of the
cases. These results make sense physically, considering that
copper smelters have historically been the overwhelming source
of secondary aerosol precursor emissions in Arizona.
All six Arizona regression analyses yield about the same value
for the mesoscale extinction/emission coefficient for SOX: .04 +
.005 (104 m)-l/(1000 TPD). This value is slightly higher than
the .01 - .03 (104 m)-l/(1000 TPD) values determined in the
copper strike analysis. We tend to place greater faith in the
.01 - .03 values because the regression coefficients in the
trend analysis may be slightly inflated by the effect of con-
current changes in NOX, NMHC, and primary particulate emissions.
Because of problems with respect to intercorrelated emission
variables, we could not isolate the effect of NOX and NMHC on
extinction trends in Arizona. However, there are several in-
dications that, although these effects appear to be secondary
to the impact of the large SOX emissions in Arizona, they
nevertheless are significant.
The regression studies for the other four sites (Salt Lake
City, Denver, Grand Junction, and Ely) did not fare as well
as the Arizona regressions. The analyses did provide some in-
dications that growth in the photochemical precursors (NOx and
NMHC) was the key factor related to visibility changes in the
urban areas of Salt Lake City and Denver, and that a measurable
impact from the Arizona smelters may extend well north of
Arizona. However, because of the lack of consistent regression
(extinction/emission) coefficients from site to site, and be-
cause of rather low correlations (except for Salt Lake City),
we cannot place great faith in the conclusions reached for the
non-Arizona sites.
FUTURE WORK
There are several research areas where the work conducted in this pro-
ject might be fruitfully extended. The most important area for future work
is to compare the regional extinction-emission relationships we have found
-------
to the results of other approaches, such as the Systems Applications Incor-
proated regional grid model for visibility (Latimer et al, 1978), the
Teknekron psuedo-spectral regional transport model (Niemann, 1978), simple
box model calculations, and other empirical data (e.g. extinction and air
quality data at large distances in plumes from isolated sources). In fact,
because of uncertainties in our analyses, it may be best to regard our
results as a checkpoint for future studies rather than as a useable tool
for regional visibility assessment.
The historical trend analysis would be made more complete by adding
data on primary particulate emissions. This additional analysis should in-
clude fugitive dust sources as well as conventional sources, should consider
the size distribution of emitted particles, and should give special attention
to the apparently significant contributions of fine silicon particles to
Southwestern visibility (Macias el al, 1979). Because of uncertainties and
gaps in data on fugitive dust emissions, particle size distributions, and
fine silicon emissions, an adequate analysis of primary particulate emission
trends would be very difficult to conduct given the present state of knowledge,
Both the copper strike analysis and the trend analysis in this report
might be improved by stratifying the data by meteorological class. The im-
portant meteorological parameters include precipitation, relative humidity,
wind speed, and especially, wind direction. Although a meteorological stra-
tification might strengthen the observed relationships between extinction
and emissions, it should be noted that the results based on such a strati-
fication would no longer represent the average regional impacts under all
meteorological conditions, but rather the impacts associated with special
(e.g. worst-case) meteorology.
The analysis of the copper strike would be improved by obtaining ad-
ditional data on visibility changes during that period. Our study has been
restricted only to data available in computerized form; there are other air-
ports for which hard copy data ave available. Additional data for northern
Arizona, southern Utah, northwest New Mexico, and southwest Colorado should
help to resolve the apparent discrepancy between the spatial scales of the
sulfate changes and visibility changes during the copper strike.
-------
CHAPTER 2
HISTORICAL EMISSION TRENDS IN THE SOUTHWEST
This chapter describes historical emission trends for secondary aerosol
precursors in the Southwest. Emissions are presented year-by-year from 1948
to 1975 for sulfur oxides, nitrogen oxides, and nonmethane hydrocarbons. The
trends are provided separately for four states -- Arizona, Colorado, Nevada,
and Utah -- in which we also have data on ambient visibility trends.
Appendix A documents the derivation of the emission trend estimates.
Many sources of data -- reports of various federal agencies, state agencies,
industrial firms, and trade associations -- were considered in compiling the
emission trends. As explained in Appendix A, the emission trends were deter-
mined only for major source categories: copper smelters, power plants, gas-
oline vehicles, diesel vehicles, railroads, petroleum industries, industrial
boilers, commercial/residential sources, gasoline evaporation, and solvent
evaporation. These major source categories account for 80-100% of the SOX,
90-100% of the NOX, and 70-90% of the NMHC emissions in the four states
studied. Appendix A includes detailed tables listing year-by-year emissions
for each source category in each state.
GENERAL MAGNITUDE OF EMISSIONS
Before discussing the historical changes in emissions, it is useful
to examine the general magnitude of emissions in order to gain an overall
perspective. Table 1 lists average emissions by pollutant and state during
1970-1972, years in which emissions tended to be at or near their histor-
ical peak in the Southwest states. It is apparent that, because of a large
copper industry, the Southwest tends to be relatively rich in SOX emissions;
in 1970-1972 the four states accounted for only 2%°/o of nationwide NMHC and
NOX but for 8% of nationwide SOX. In particular, the copper smelters in
Arizona -- which emitted approximately 10 times as much SOX as the Los Angeles
Basin in 1970-1972 (AQMP, 1976) have historically been, on a mass basis,
the predominant single source of precursor emissions in the Southwest.
-------
TABLE 1. MAGNITUDE OF PRECURSOR EMISSIONS IN
THE SOUTHWEST, TONS/DAY, 1970-1972
SOX NOX NMHC
total/(copper smelters) total total
Arizona 5,8407(5,600) 570 570
Colorado 200 480 760
Nevada 8307(760) 200 210
Utah 6507(430) 290 330
Four State Total 7,5207(6,790) 1,540 1,870
United States
Total (EPA, 1976) 92,000 58,000 82,000
HSITORICAL CHANGES IN EMISSIONS
Figures 1 through 12 summarize the historical changes of SOx, NOx,
and NMHC emissions in the four states under study. The top line in each
figure depicts the year-by-year changes in total emissions for each pol-
lutant and each state. The distances between the lines measure contributions
from individual source categories.
As indicated in Figures 1 to 4, copper smelters have historically been
the dominant source of SOx emissions in the Southwest. The major factors
governing historical changes in smelter emissions have been production
levels (including the effects of labor strikes) and air pollution controls.
Over the entire study period, copper production more than doubled in Arizona
and Nevada but remained approximately constant in Utah. Downward kinks in
emissions are evident during years of major labor strikes: 1948-1949, 1954-
1955, 1959, and especially the 9 month strike of 1967-1968. Air pollution
controls reduced Utah smelter SOx emissions in the 1950's. Some control of
the Arizona smelters started in 1966, but most of the Arizona controls were
phased in after 1973. Because of emission controls and because of slightly
reduced copper production, Arizona smelter emissions dropped from 5800 TPD
in 1973 to 3600 TPD in 1975 (and continued falling to 2000 TPD in 1977).
8
-------
C7>
C
i- co
C CD
r- U
ra s-
, E 3
r 01 O
eC/a: oo
CO
S_
cu
oo
s_
O)
C-
O-
o
un
r-^
en
i
CO
s-
re
OJ
_0
I/I
-a
c
O)
S-
-M
c:
O
OJ
OJ
T3
i
X
O
OO
cu
en
u_
o
o
o
10
o
o
o
o
o
o
o
o
o
o
o
o
CM
o
o
o
SUOLSSLU3 XQS
-------
, LO
o
i-D
ro
01
o
en
ti
CO
o
TD
S-
o
o
o
e
LO
OJ
01
s_
^
C7>
O
in
SUOLSS 11113 XOS
10
-------
cu
LT>
1-^
CTl
I
co
CTi
ro
-a
OJ
:.T*-
C
-a
S
s-
OJ
a
x
o
CO
OJ
11
-------
O)
un
r^.
cr>
I
00
01
03
CO
-a
c
OJ
to
CO
QJ
OJ
'O
X
O
i-
3
t-
IT
CO
*3-
-------
CO
^r
01
o
IM
s- -o
(O C
O) CU
>- S-
OJ
Ol
-a
x
o
CD
o
s_
CU
S-
CD
r
o
o
o
o
ID
o
o
un
I
o
o
I
o
o
ro
I
o
o
CM
O
O
(Aep/suoi) SUOLSSLLU3
13
-------
S-
ra
O1
iI
CO
CT>
o
-a
i-
o
o
o
cu
S-
c:
O
O)
ai
T3
r
X
o
a>
CJ)
o
4-J
O)
(Aep/suo;) SUOLSSLW3 XQN
14
-------
co
d)
O
to <-
i- -!->
O C
s- cu
O) T3
S-.
eo
CU
LT)
r^
CTi
I
CO
ra
-o
to
>
O)
-a
c
o;
s-
co
co
OJ
T3
X
o
suoLSSLiiq
15
-------
oo
<3-
CTl
to
fO
O)
0)
S-
CO
to
OJ
-a
x
o
0)
CD
O
s_
co
O)
SUOLSSLIU3 XQN
16
-------
U o
OJ
O
UD
L_ un
O
un
un
r-~
CTl
ii
oo
=ħ
CTi
o
N
=c
c:
CO
T3
OJ
£1
O
CO
to
O
JO
i-
fO
CJ
o
S-
TD
OJ
sz
o>
c
O
CTl
OJ
S-
CT)
SUOLSSLLU3 3HWN
17
-------
rOl
O)
t<3
Ol -!->
I 00 4 C
I fO 1 OJ I
c:
s-
rc
O)
o
'.O
o
LO
LO
r-^.
en
rt
oo
<*
CTl
o
-a
S-
o
c
0)
to
to
S-
(C
o
o
S-
rd
)->
-C
OJ
c
o
^~y
o
I1
O)
S-
^3
cr>
o
o
0
o
2
o
0
0
0
0
oo
o
o
C\i
0
o
SUOLSSLUI3
18
-------
O)
un
r-^
en
i
co
CT>
-a
n3
>
0)
O)
a>
CO
O
O
i.
Ol
ro
4->
x:
O)
c:
o
HI
S-
(X'ep/suo^) suoissiuq
19
-------
its
CL>
I
CO
res
-o
C
CU
s_
sr
o
i
CO
(/I
r
E
O)
c
o
-Q
S-
o
s_
-a
>>
CD
c
"3
4->
-C
CU
SUOLSSLUI3
20
-------
The other significant sources of SOx emissions in the Southwest in-
clude railroads, power plants, and industrial/commercial/residential fuel
combustion. Railroading, a very substantial source of SOx in the late 1940's,
became a negligible source of SOX by the middle 1950's as diesel locomotives
were substituted for coal locomotives. Power plants have been a rapidly
growing source category and now rank second to copper smelters in total
Southwest SOX emissions. SOX emissions from industrial/commercial/residential
fuel combustion have remained approximately constant over the 2% decades,
as switches to cleaner fuels (from coal and residual oil to distillate oil
and gas) compensated for increased energy demand.
As shown in Figures 5 through 8, NOx emissions climbed steadily in
the four states from 1948 to 1975, increasing by a factor of 7 in Arizona
and Nevada, a factor of 4 in Colorado, and a factor of 3 in Utah. The two
major source categories for NOX, power plants and gasoline vehicles, ac-
counted for most of this rising trend. These two source categories pro-
duced a very pronounced upward trend in NOX during the late 1960's and early
1970's due to increased electricity demand, greater coal use by utilities,
traffic growth, and increased automotive NOx emission factors (a result of
the "leaning-out" of engines which was used as an initial control measure
for hydrocarbons and carbon monoxide). The other significant source cate-
gories for NOx -- industrial boilers, diesel vehicles, railroads, and
commercial/residential fuel combustion -- produced, in toto, a very gradual
rise in NOX emissions which contributed slightly to the overall NOx increase.
Figures 9 through 12 indicate that NMHC emissions in the Southwest
rose steadily from the late 1940's to a peak in the early 1970's and then
started to decrease. This basically reflects the trend in the dominant
source category for NMHC, gasoline vehicles. Emissions rose with traffic
growth for most of the period, began to level off with the automotive con-
trols of the 1960's, and started downward as the automotive standards be-
came more stringent in the 1970's. The especially pronounced dip in 1974
was produced in part by the "gasoline crisis" of that year.
PRIMARY SULFATE EMISSIONS
During the initial phases of this project, a special issue was brought
to our attention concerning possible differences between coal-fired power
plant SOX emissions and smelter SOX emissions with respect to sulfate and
21
-------
visibility impacts. Specifically, the question was raised whether power
plant SOx emissions might be less important (per ton) to ambient sulfates and
visibility than are smelter SOX emissions because power plant emissions
might contain a much smaller fraction of primary sulfates. In order to
help resolve this question, we have reviewed the literature on primary sul-
fate emissions from power plants and have gathered data from various re-
searchers on primary sulfate emissions from smelters. The information we
obtained is presented and discussed in Appendix B. Although there are
some questions concerning the various sampling methods that have been used
to test smelters for primary sulfates, the existing data indicate that the
fraction of primary sulfates from smelters is either about the same as, or
somewhat less than, the fraction of primary sulfates from coal-fired power
plants.
EMISSION TRENDS FOR INDIVIDUAL AIR BASINS
The emission trends presented in this chapter are organized by state
partly because a state-to-state breakdown will be useful in later analyses,
and partly because much of the underlying data for emission trends are only
available on a state-to-state basis. For some of the analyses performed
later in this report, it will also be necessary to have emission trends for
individual Air Quality Control Regions (AQCR's). In Appendix C, we show
that it is a fairly good approximation to assume that emission trends in
several AQCR's are just a constant times the emission trends for the cor-
responding state. The factors for computing AQCR emission trends from
state emission trends are presented in Appendix C.
22
-------
CHAPTER 3
HISTORICAL VISIBILITY TRENDS IN THE SOUTHWEST
A previous study (Trijonis and Yuan, 1978a) included data on visibility
trends at 12 locations in the Southwest. The trends were based on visibility
percentiles (e.g. medians), and the study examined a time period from 1948
to 1972. This chapter updates the previous visibility trend study in two
ways: (1) the time period is extended through 1976, and (2) previously un-
known changes in observation locations are accounted for.
CHANGES IN OBSERVATION LOCATIONS
As part of our previous study of visibility trends in the Southwest,
we conducted a survey designed to check the quality of airport data by in-
quiring about the visibility markers, reporting procedures, observation
locations, and other factors affecting the visual range measurements. Al-
though the survey proved very useful in many respects, there were problems
with a question regarding whether the observation locations had changed
"significantly" since 1948. Possibly because of the vagueness of the term
"significantly", we have found that the responses sometimes did not identify
all changes in location that had occurred.
We recently discovered that it is possible to obtain, from the Nation-
al Climatic Center, forms which provide very detailed histories of the
locations of the weather observation stations. After ordering these
forms, it was determined that several changes in observation location had
occurred which might affect the previous trend study. Accordingly, the
trend study was performed again, accounting for all relocations in ob-
servation sites.
Before redoing the trend study, we applied statistical tests to in-
vestigate whether the changes in observation locations significantly altered
reported visibility levels. Two types of test were used to estimate both the
net jump in visibility produced by the relocation and the statistical signi-
ficance of that jump. These two tests, a t-test on the net jump in quarterly
medians and a multivariate regression test on the jump in seasonally adjusted
quarterly medians, are described in Appendix D.
23
-------
Table 2 lists all of the relocations that occurred at the 12 trend
sites from 1948 to 1976 and summarizes the results of the statistical tests.
The relocations are organized into three groups. The first group of four
were accounted for in the previous visibility trend study. The second group
of two were not previously accounted for, but they could not affect the
major conclusion of our prior report that visibility had deteriorated
significantly in the Southwest from the middle 1950's (1953-1955) to the
early 1970's (1970-1972). The last group of five were not previously ac-
counted for and could affect our prior conclusions.
For each statistical test, Table 2 lists the estimated net jump in
median visibility (both in miles and percent), the t-statistic for the jump,
and the significance level for the jump (the confidence level that the t-
statistic is different from zero). Based on the estimated net percentage
jump in visibility, the significance level of the jump, and the agreement
between the two statistical tests, the last column of Table 2 presents our
overall conclusion as to whether or not the relocation produced a significant
change in reported visibilities. The conclusions are listed as either
"significant", possibly significant", or "apparently not significant".
Because of limitations in the statistical tests and associated un-
certainties concerning the real significance of the relocations, we have
decided to adopt the following policy throughout this report: For those
sites where the relocation produced a "significant" change in reported visi-
bilities, the data will be restricted to a period excluding the relocation
(e.g. Phoenix from 1959 to 1976). For those sites where the change is
"possibly significant" or "apparently not significant", the analysis will
always be done twice -- once restricted to a period excluding the relocation
(e.g. Prescott from 1948 to 1969), and once including all available data
(e.g. Prescott from 1948 to 1976).
HISTORICAL VISIBILITY TRENDS
Figures 13 through 24 summarize historical visibility trends at the
12 study sites considered in our previous report (Trijonis and Yuan, 1978a).
All site relocations, as well as our qualitative conclusions concerning the
significance of those relocations, are noted in the Figures. Also noted are
24
-------
OO
o
cz.
UJ
OO
CO
o
oo
U_ UJ
O i-.
UJ i i
es i
-. a:
oo o
D.
_j a:
t Q
00 UJ
<: o
| Q.
oo uj
a:
I O
U_ OO
o z
o
OO ii
I
00 -=C
I O
2: uj
Cxi
UJ
co
\SONALLY 1
Ğ* 1
uj Ğa:
00 ->
a
Z. LU
a. >-
s: -i
=> o:
"3 LU
1
REGRESSION TEST ON
ADJUSTED QUAR
EDIANS*
s:
>-
i
1
z
a.
s:
-3
z
0
H-
UJ
1
1
4->
00
<_>
0
o
Significance
level of
t Value
s|
"ro °
Jump in Visibility
;s) (percent)
^ 3-
Significance
level of
t Value
OJ g"
^ s-
tump in Visibil ity
ğs) (percent)
cu i
1 1 1
5 LU CU
111
Q '
11
STATION
c
tJ
u
CO
cn
A
0,
CO
o
1*
ID
CO
1
CO
r~'
LO
1
00
o
CM
LD
CM
Phoenix, AZ
rO
U
C
CD
CO
en
cn
A
CO
o
CO
CO
A
Lfl
CO
t-H
CO
H?
CO
O
CO
LD
Phoenix, AZ
c
4-
01
CO
en
cn
A
CO
S
^
i 1
t I
i 1
v
r-l
t 1
1
Colorado Springs, CO
c
u
c
C71
OO
LO
cn
A
t t
CM
UT>
LD
ID
cn
A
0
CO
cn
+
t-H
0
cn
o
CJ
o
JD
(U
13
Q_
0
C
cn
oo
01
cn
A
O
**
cn
i
oo
oo
1
T-H
oo
CM
t
Cn
CM
r-H
CM
0
S
Winslow, AZ
^
^ 0
i- T-
Q. cn
Q.T-
=r oo
t 1
o
t t
LT>
?
1 1
O
CM
i i
-4-
t i
O
0
Ln
CM
Grand Junction, CO
c >> c
rO i T3
^" c"
_Q 4_ Qj 4-
r- . i_ -
VI C fO C
i/i 01 CL Oi
0 - O.JT--
LO r**.
0 t-i
O CO
tjD CM
r- 1 O
5-S S^
cn CM
CM 1
CO CO
cn .I
1 t 1
co cn
0 0
O CO
r*- Ln
CM O
r .
Prescott, AZ
Tucson, AZ
c c c
fO ro rtj
>ğ U >> U U
JO M- JD 4- M-
t/1 C (.0 C C
(/I CT> */) CD Ol
o ._ o .,_ .,
Q- CO d. OO CO
^ &5
cn cn
cn cn
A A
cn oo co
^o o en
r-l 1 1 LO
i i i
r--. ^r ^~
i i -
LO <4> CO
O *-H 1
CM un o
+ CM CM
CO r-H r^
-t CO ^J-
f 1 1
^-H O O
Ol <i Ğ=J-
U3 *JO LO
LD cn r^.
i
o
o
O QJ
-XL
i- *Z 1
CU
C >> i
(1) tTJ
Q UJ CO
O
CO
2
O
s
o
o
f-
OJ
Q.
01
> O
Ol
1C
i- >>
"O '
cu uj
-o
O)
E
25
-------
~o *
01 +->
> c
O fl3
s: u
c £
o !-
4-> CTj
00
^^
cu
[
£,
£
^
70
60
50-
40-
30-
20
10-,
10th Percent!' le
1 . >^, y- -*v. . .
^^*--* ^^/ ^v^ <|X^
50th Percent!' le
1 *^ -^y^~-^^^^
90th Percent!' le
*^*\ğ--ğ *-^_^_X*~^Nfc__>___. ^^-^*
'I 1 1 1 1 ! '
48 50 55 60 65 70 75 76
Year
Figure 13. Long-term visibility trends at Phoenix.
26
-------
80
70 -
^ 60 -
Ol
1 50 I
£ 40
-D
t/1
30 I
20 -
10 I
* V
48 50
O) -l->
> >, c
O r- to
5C 4-> 0
C !-
c: o> 4-
O S- T-
r- n3 c:
4-> Q. CJ)
(13 Q.T-
I
55
I
60
Year
50th Percent!le
90th Percent!le
65
I
70
I '
75 76
Figure 14. Long-term visibility trends at Tucson,
27
-------
to
1}
100-
90-
80-
70-
60-
50-
40-
30-
20-
10-
~o *
OJ 4->
> C
O res
s: >, u
C .Q M-
O - T-
i- tO C
4-> to C7>
ra o -I-
4-> O- (S~>
10th Percent!le (Estimated)
90th Percent!le
^ I
48 50
55
60
Year
I
65
70
75 76
Figure 15. Long-term visibility trends at Denver.
23
-------
~o *
CD +->
> c
O ra
s: o
O T-
i- C
4-> 01
OO -
70
60
50
30
20
t/>
> 10
10th Percentile
90th Percentile
48 50
I
55
60
Year
I
65
I I '
70 75 76
Figure 16. Long-term visibility trends in Salt Lake City.
29
-------
O)
en
c:
O)
80
70
a)
v5 50-
$ 40-
£ 30-
i/i
> 20-
10-
I I
48 50
-M <_>
C
03 O)
O C
i- S_
E O
CD Q.
r- Q;
oo o;
Ol
4-)
"3
10th Percentile (Estimated)
50th
90th Percentile
55
I
60
Year
Percentile
65
I
70 71
Figure 17. Long-term visibility trends at Fort Huachuca.
30
-------
~O --
O> -M
> c:
o ro
s: >, o
n !"
E -Q M-
O T- T-
i- to C
(-> CO C7)
(T3 O -r-
-l-> D_ OO
GO
901
80-
-, 70-
S 60-
E
>, 50-1
-P
i 4°-i
a 30-
>
20-
10-
75th Percentile (Estimated)
90th Percent!le
~ir
48 50
55
r
60
Year
65
70
75 76
Figure 18. Long-term visibility trends at Prescott.
31
-------
I/)
0)
90-
80-
70-
60-
50-
40-
30-
20-
10-
80th Fercentile (Estimated)
90th Percent!le
1 T
48 50
55
60
Year
65
I
70
O) +-5
> c
O fC
sr o
o -r-
4-> 01
CO <-
75
Figure 19. Long-term visibility trends at Winslow.
32
-------
to
O)
140-
130-
120-
110-
100-
90-
80
70-
60-
50-
40-
30 -
20 -
10 -
~O '-
QJ +J
> c
o ra
s: o
o -r-
- c
-(-> CD
03 !-
4-> GO
CO
10th Percentile
50th Percentile
90th Percent-He
-" T
48 50
55
60
65
70
Year
Figure 20. Long-term visibility trends at Colorado Springs.
33
-------
Ol
i
120-
110-
100-
90-
80-
70-
60-
50-
40-
30-
20-
10-
o z:^
O) 4->
> >> c
O i fC
s: -(-> u
C 'I-
C CU 4-
O S_ T-
r- fO C
4-> Q. C7)
(B Q.T-
->
48 50
10th Percentile (Estimated)
90th Percentile
I
55
60
Year
i
65
70
T i
75 76
Figure 21. Long-term visibility trends at Grand Junction.
34
-------
100
90
80
70
60
50
40
30-
20
10
^ I
48 50
QJ
4->
rO
O)
(O
4->
re
Q
"O ^
0) 4->
> E
O n3
5! O
O
fC
00
OO
10th Percentile (Estimated)
50th Percentile
90th Percentile
I
55
60
Year
I
65
70
75 76
Figure 22. Long-term visibility trends at Pueblo.
35
-------
to
OJ
JD
r
(/I
80 -
70 -
60 -
50 -
40 -
30 _
20 _
10 -
~nr
48 50
CD
C7>
to
u
i- S_
E O
CD Q.
r- CU
oo o;
-o
O)
c: .a 4-
O -r- !-
r- in C
!-> CO C7)
res o !-
+-> tx oo
80th Percentile (Estimated)
90th Percentile
55
~T~
60
Year
60
~r
70
75 76
Figure 23. Long-term visibility trends at Ely.
36
-------
1/5
(1)
E
130-
120-
110-
100
90-
80-
70-
60-
50-
40-
30-
20-
10-
o c
r- -I
<+- 4-> O)
r- i_ CD
C O C
C71 Q. (O
r- a) _c
00 Oi O
10th Percent!le (Estimated)
90th Percent!le
48 50
55
60
Year
65
"1 P
70 75 76
Figure 24. Long-term visibility trends at Cheyenne.
37
-------
*
three significant changes in reporting practices.
A major conclusion of our previous report was that visibility deterior-
ated throughout the Southwest on the order of 10 to 30% from the middle 1950's
(1953-1955) to the early 1970's (1970-1972). This conclusion was based on
the trend summaries presented in Table 3A. Because of the new information
concerning site relocations, we have modified these trend summaries as shown
in Tables 3B and 3C. Table 3B includes data only for fixed site locations;
Table 3C also includes data for the site relocations that were not "signifi-
cant". It is obvious that whichever way one calculates the trends (Table 3B
or Table 3C), the same conclusion remains: visibility deteriorated on the
order of 10 to 30% at nearly all sites from the middle 1950's to the early
1970's.
Examing data from 1948 to 1970, Latimer et al. (1978) also noted a
decreasing trend in visibility at Southwestern sites. In addition, they
found that visibility tended to increase significantly from 1970 to 1976.
Our results tend to confirm this latter conclusion; the eight sites for
which we have data through 1976 exhibited the following changes in median
**
visibility from 1970-1972 to 1974-1976: Phoenix (+12%), Prescott (+7%) ,
***
Tucson (+12%), Denver (+7%), Grand Junction (+3%), Ely (+18%) , Salt Lake
City (0%), and Cheyenne (-12%).
These three changes occurred at Fort Huachuca (causing a downward jump in
visibility from 1955 to 1956), Ely (causing a downward jump in visibility
from 1954 to 1955), and Cheyenne (causing an upward jump in visibility from
1958 to 1959). The trends for these sites are started after the reporting
changes.
**
75th percentile instead of median.
***
'80th percentile instead of median.
38
-------
TABLE 3A. NET PERCENT CHANGES IN VISIBILITY,
1953-1955 TO 1970-1972, ORIGNIAL
ANALYSIS (TRIJONIS AND YUAN, 1978a)
LOCATION
CHANGES IN THREE-YEAR AVERAGES, 1954 TO 1971
Best (10th %) Median Worst (90th %)
Visibility Visibility Visibility
URBAN
Phoenix, AZf
Tucson, AZ
Denver, CO
Salt Lake City, UT
NONURBAN
Fort Huachuca, AZ
Prescott, AZ
Winslow, AZ
Colorado Springs, CO
Grand Junction, CO
Pueblo, C0f
4-
Ely, NVT
Cheyenne, WY
0%
N.A.
-22%
-24%
-12%
N.A.
N.A.
-17%
-9%
-9%
N.A.
-28%
-23%
-22%
-13%
-27%
-27%
*
-25%
**
-17%
-12%
-4%
+35%
**
-42%
-23%
-20%
-11%
-29%
-19%
-28%
-21%
-27%
-17%
-3%
0%
-33%
-19%
N.A.
Not available
t
Trends for these sites are extrapolated from data covering most,
but not all, of the period 1954-1971.
**
75th percentile visibility instead of median visibility.
>e
80th percentile visibility instead of median visibility.
39
-------
TABLE 3B. MET PERCENT CHANGES IN VISIBILITY,
1953-1955 TO 1970-1972, ALL SITE
RELOCATIONS DISCOUNTED
LOCATION
CHANGES IN THREE-YEAR AVERAGES, 1954 TO 1971
Best (10th%) Median Worst (90th%)
Visibility Visibility Visibility
URBAN
Phoenix, AZf
Tucson, AZ
Denver, C0f
Salt Lake City, UTf
NONURBAN
Fort Huachuca, AZ
Prescott, AZf
Wins low, AZ
Colorado Springs, CO
Grand Junction, CO
*f*
Pueblo, COT
+
Ely, NVT
Cheyenne, WYf
0%
N.A.
-12%
-17%
-12%
N.A.
N.A.
-17%
- 9%
-11%
N.A.
-28%
-23%
-32%
- 8%
-22%
-27%
*
OQ°/
-L.O/0
**
-17%
-12%
- 4%
+34%
**
-13%
-23%
-20%
-24%
-50%
+ 3%
-28%
-18%
-27%
-17%
- 3%
- 5%
+10%
-19%
N.A.
t
**
Not available
Trends for these sites are extrapolated from data covering most, but
not all, of the period 1954-1971. The periods for trend analysis at
these sites are as follows: Phoenix, 1959-1972 (site moved in 1958);
Tucson, 1959-1972 (site moved in 1958); Denver, 1953-1968 (site moved
in 1969); Salt Lake City, 1955-1972 (site moved in 1954); Fort Huachuca,
1956-1970 (reporting practices change in 1955, data terminate in 1971):
Prescott, 1953-1969 (site moved in 1970); Colorado Springs, 1953-1966
(site moved in 1967); Pueblo, 1955-1968 (data start in 1954, site
moved in 1969); Ely, 1962-1972 (site moved in 1961); and Cheyenne,
1959-1972 (reporting practices change in 1958).
75th percentile visibility instead of median visibility.
V
80th percentile visibility instead of median visibility.
40
-------
TABLE 3C. NET PERCENT CHANGES IN VISIBILITY
1953-1955 TO 1970-1972, ALL SIGNI-
FICANT SITE RELOCATIONS DISCOUNTED
LOCATION
CHANGES IN THREE-YEAR AVERAGES, 1954 TO 1971
Best (10th%) Median Worst (90th%)
Visibility Visibility Visibility
URBAN
Phoenix, AZ
Tucson, AZ
Denver j CO
Salt Lake City UT1"
NONURBAN
Fort Huachuca, AZ
Prescott, AZ
Wins low, AZ
Colorado Springs, CO
Grand Junction, CO
Pueblo, CO1"
Ely, NV1"
Cheyenne, WY1"
0%
N.A.
-22%
-17%
-12%
N.A.
N.A.
-17%
- 9%
-11%
N.A.
-28%
-23%
-22%
-13%
-22%
-27%
*
-25%
**
-17%
-12%
- 4%
+34%
-42%
-23%
-20%
-11%
-29%
+ 3%
-28%
-21%
-27%
-17%
- 3%
- 5%
-33%
-19%
N.A.
t
**
Not available
Trends for these sites are extrapolated from data covering most, but
not all, of the period 1954-1971. The periods for trend analysis at
these sites are as follows: Phoenix, 1959-1972 (site moved in 1958);
Salt Lake City, 1955-1972 (site moved in 1954); Fort Huachuca, 1956-
1970 (reporting practices change in 1955, data terminate in 1971);
Colorado Springs, 1953-1966 (site moved in 1967); Pueblo, 1955-1968
(data start in 1954, site moved in 1969); Ely, 1955-1972 (reporting
practices change in 1954); and Cheyenne, 1959-1972 (reporting practices
change in 1958).
75th percentile instead of median visibility.
80th percentile instead of median visibility.
41
-------
CHAPTER 4
REGIONAL EXTINCTION/EMISSION COEFFICIENTS
BASED ON THE 1967-1968 COPPER STRIKE
The nine month copper strike of 1967-1968 produced a very large re-
*
duction of sulfur oxide emissions in the Southwest. Compared to average
daily emission levels during surrounding years (1965, 1966, 1969, 1970,
and the other 15 months of 1967-1968), sulfur oxide emissions dropped 5550
TPD in southeast Arizona, 700 TPD at McGill (Nevada), and 580 TPD at Magna
(Utah). By relating these SOX emission reductions to corresponding air
quality changes, this chapter estimates regional extinction/emission co-
efficients for sulfur oxides in the Southwest.
METHODOLOGY
Our previous report (Trijonis and Yuan, 1978a) documented airport
**
visibility changes and NASN Hi-Vol sulfate changes at the sites shown in
Figure 25 during the 1967-1968 nine month copper strike. That report dis-
cussed in detail the large percentage improvements in air quality during
the copper strike (e.g. visibility improved 8-29% at the 5 airports in
Arizona, and sulfates dropped 60-67% at the 4 NASN sites in Arizona), the
high statistical significance of the observed improvements, and evidence
that meteorological variation did not contribute significantly to the im-
provement. For the purposes of the present study -- derivation of empirical
extinction/emission coefficients the air quality change of interest dur-
ing the strike is the absolute change in the extinction coefficient (B),
with units of [10 m]~ . This can bi
using the Koschmeider relationship,
with units of [10 m]~ . This can be readily derived from the airport data
(i)
where V is the median visibility in [miles]. The change in absolute
~*
The overall SOX reduction in the four study states (Arizona, Colorado,
Nevada, and Utah) was 95%. If we also include New Mexico, the overall
reduction was 91%.
**
National Air Surveillance Network
42
-------
cr
(O
O)
c:
0)
+J
QJ
-o
O
4-)
T3
01
ID
r
to
CD
i- CD
O 'I-
i i r_
C V)
o
O)
Z. CL.
CO Q.
< O
~s" (j
-o co
(T3 CTl
tl
to i
O **Q
r- CTl
4-> I
(C
4-> O)
tO JZ
5-
OJ O)
ro i-
O) 3
2 -O
4-> LO
S- OJ
O CD
Q- C
S- 03
r- _C
< O
(XI
OJ
s_
o
43
-------
extinction can also be estimated from ambient sulfate data by assuming an
extinction coefficient per unit mass for sulfates. For this purpose, we
have chosen an extinction coefficient per unit mass of .04 (10 m)~ /(yg/m )
based on empirical regression models for Phoenix and Salt Lake City (Trijonis
and Yuan, 1978a). This value is slightly less than the typically reported
values of .06 (+.03)(l04m)~1/(yg/m3) calculated from the theory of Mie
scattering for aerosols in the accumulation size range (Latimer et al.,
1978; White and Roberts, 1977; Ursenbach et al., 1978; Waggoner et al.,
1976). However, a value of .04 (10 m) /(yg/m ) is consistent with theore-
tical predictions for the Southwest, if one accounts for the low relative
humidity in that area (Latimer et al., 1978; Trijonis and Yuan, 1978a).
* **
Figures 26 and 27 present seasonally adjusted changes in extinction
(airport data) and sulfates (NASN data) during the 1967-1968 copper strike.
In order to relate these air quality changes to emissions, we have chosen
based on prevailing wind directions (NOAA, 1977) and on the proximity of
monitoring sites to smelters -- to couple the emission sources and monitor-
ing sites as shown in Table 4. Alamogordo, Albuquerque, Grand Junction,
Denver, and Cheyenne have been excluded from the analysis because they ap-
pear to be less likely impact areas based on distance to the smelters and
prevailing wind directions.
For each airport, an extinction/emission coefficient is derived simply
by dividing the extinction change during the strike by the reduction in SOx
emissions from the smelters paired with that airport. A similar computation
*
Seasonally adjusted changes in air quality during the strike are deter-
mined by comparing average (or median) levels during the strike to
average (or median) levels during equivalent 9 month periods of sur-
rounding years (usually 6 years symmetrically around 1967-1963, but
sometimes fewer years depending on data availability).
**
In most cases the extinction changes are based on median visibility
levels, but in some cases higher percentiles are used to avoid extra-
polation of the visibility frequency distributions. The reader should
note that the extinction changes for Denver and Prescott turn out to
be the same whether or not we account for the relocation of those sites
(Denver in June, 1969 and Prescott in June, 1970). That is using
six surrounding years to do the seasonal adjustment happens to give the
same result as using 2 surrounding years for Denver and 4 surrounding
years for Prescott.
44
-------
to
E
O
CO
to
O E
CD
CO
t-> X
r- O
E CO
i-
Ğ CD
CD 4->
r- CD
s- E
+-> to
S- O
CD
O- C
Q. O
O !-
O *->
3
CD JD
^: -i-
4-> S-
-i->
en co
c: -i-
- -a
i Q_
*-> 03
O S-
E CD
i- O
4-) CD
X CD
CD
CD
E -E
r- 4->
CO O
CD 4->
CD
E "O
it! CD
-E £-
O t3
Q.
-O E
CD O
+-> U
to
>> S-
r CD
i 4->
rt3 CD
E E
O
ro CZ)
CD i I
CO' - 1
CD
S_
45
-------
46
-------
TABLE 4. PAIRING OF AIRPORTS AND NASN SITES WITH
SMELTERS MOST LIKELY TO AFFECT THOSE SITES.
EMISSION SOURCE
AIRPORTS
NASN SITES
Arizona Smelters
Fort Huachuca, Tucson,
Phoenix, Prescott, Winslow,
Farmington, and Las Vegas
Tucson, Phoenix,
Maricopa County,
Grand Canyon, Mesa
Verde, and Las Vegas
McGill, NV Smelter
Ely, Wendover, and
*
Dugway
White Pine
Magna, UT Smelter
Salt Lake City and
*
Dugway
Salt Lake City
Because of uncertainty concerning the source most likely to affect
Dugway, Dugway has been coupled with both the McGill and Magna smelters.
47
-------
is made with the NASN sulfate data. The SOX extinction/emission coefficients
based on the airport data may be overestimated because of simultaneous re-
ductions in primary particulates from smelters during the copper strike.
Based on light scattering contributions from sulfate and fine silicon par-
ticles reported by Macias et al (1979) for nonurban areas and a smelter
plume, we conclude that the overestimate may be on the order of 10-30%.
The above analysis neglects superposition of effects from the three
smelter areas. It is probably safe to neglect superposition in considering
sites paired with the Arizona smelters; however, because of the extremely
large emissions of the Arizona smelters, it may not be accurate to neglect
their impact on sites paired with the McGill and Magna smelters. The con-
sequences of this possible source of error will be addressed in the next
section which presents the results of our analysis.
EXTINCTION/EMISSION COEFFICIENTS FOR SOX
Figure 28 presents the extinction/emission coefficients based on visi-
bility changes during the copper strike. Change in extinction divided by
change in SOX emissions is plotted as a function of the distance between
the airport and the associated smelters. For those airports paired with
the Arizona smelters, a weighted average distance (by emissions) is used.
Figure 29 presents changes in sulfate divided by changes in emissions,
plotted as a function of distance between the NASN sites and the smelters.
Figure 30 is a composite of Figures 28 and 29, with sul fates converted to
extinction using a factor of .04 (10 m)~ /(yg/m ). Figure 30 indicates
that the extinction/emission coefficients based on the airport visibility
data agree fairly well with the extinction/emission coefficients based
on the NASN sulfate data. At distances of 50 to 200 miles from the smelters,
both data sets consistently yield extinction/emission coefficients on the
order of .01 to .03(10 m) /(1000TPU SOX). Also, both data sets seem to imply
that the extinction/emission coefficient rises considerably as the distance
to the source decreases from 50 to 10 miles. The only major disagreement
.
I he only exception to this conclusion among the 11 sites in the 50 to 200
mile range is the Wendover airport data.
48
-------
.40-
.35
o
GO
Q
O.
§ .30-
E
*o .25-
.20-
o
GO . 15-
co
o
5 .ion
.05-
a: versus Arizona smelters
n: versus Nevada smelter
u: versus Utah smelter
Wendover
n
Salt Lake City1
Dugway
"T
50
Tucson9 * Phoenix
Dugway11 Prescott3
Fort Huachucaa winslow9
~~1 T
100 150 200 250
DISTANCE TO SMELTERS (MILES)
,
Farminqtond
T* -
Las
Vegas3
300
350
Figure 28. Extinction/emission coefficients based on airport
data during the 1967-1968 copper strike.
49
-------
9-
x
o
VI
£ 8-
o
o
o
en
6-
IT 5H
o
o
2 4-
oo
oo
oo
3-
2-
l-
O Salt Lake City
0 White Pine"
TucsonaO
Phoenix
Maricopa Countyc
a: versus Arizona smelters
n: versus Nevada smelter
u: versus Utah smelter
Grand Canyon'
Las
Vegas3
O
O Mesa Verde
50
100 150 200 250
DISTANCE TO SMELTERS (MILES)
300
350
Figure 29. Sulfate/enn'ssion coefficients based on NASN data during
the 1967-1968 copper strike.
50
-------
.40_
.35 -
o
IS)
O
Q.
§ .30
.25-
.20-
oo
I
z.
LU
u_
o
CJ
2 .15
oo
00
.10-
.05 -
Based on airport data
O Based on NASN sulfate data
O
o
o
1
50
I [ I I
100 150 200 250
DISTANCE TO SMELTERS (MILES)
300
350
Figure 30. Extinction/emission coefficients from airport and NASN
data during the 1967-1968 copper strike.
51
-------
between the two data sets is that the airport data (Farmington and Las
Vegas) indicate negligible changes in extinction at distances of 300-375
miles, while three NASN sites imply statistically significant changes
(Trijonis and Yuan, 1978a) at distances of 250 to 375 miles and yield ex-
tinction/emission coefficients on the order of .01 (10 m)~ /(1000 TPD) at
those distances.
The higher extinction/emission coefficients at distances within 40
miles of smelters deserves further comment. This effect can be due to any
of the factors listed below:
The larger extinction/emission coefficients at sites near
individual smelters may be due to the tendency for air pol-
lution sources to produce greater effects at smaller dis-
tances from the source. Figure 31 replots the data in
Figure 30 on a log-log scale; the log-log regression line
indicates that the effect of SOx emissions on regional ex-
tinction in the Southwest may be inversely proportional to
the distance from the source (i.e. an x"* dependence).
The sites that are nearest to individual smelters and that
yield the high extinction/emission coefficients are all
coupled with the Utah and Nevada smelters. The extinction/
emission coefficients at these sites might be artificially
high because we have neglected emissions from the Arizona
smelters which might produce significant impacts even at
very long distances.
0 The extinction/emission coefficients at the Utah and Nevada
sites are more sensitive to potential errors in estimates
of the extinction changes because we are dividing by much
smaller emission changes. Relatively small absolute errors
in the extinction data at these sites can produce rather
large errors in the extinction/emission coefficients.
The extinction/emission coefficients might tend to be slight-
ly higher at the northern Utah and Nevada sites because
relative humidity is higher at those locations. Relative
humidity averages 50-60% in northern Utah and Nevada but
only 30-50% at the other study locations (NOAA, 1977).
Higher relative humidity could tend to produce greater
sulfate formation rates as well as greater extinction per
unit mass of ambient sulfates.
It is tempting to attribute most or all of the increased extinction/emission
coefficients at the northern Utah and Nevada locations to the closer prox-
imity of the smelters (i.e. the x~l dependence). However, because of the
52
-------
O
CL
o
o
o
E
5J-
O
UJ
II
o
LlJ
fl
>-
.3 _
.2 _
.1
.08 ,
.06.
.05 .
.04.
.03-
.02_
.01-
.008-
.006
.005
.004
.003
.002-
.001
10
y = 1.66x"'95
Correlation Coefficient = -0.83
o
Based on airport data
O Based on NASH sulfate data
Note: Farmington data excluded because
of negative extinction/emission
change.
I I I I I I 1
20 30 40 50 60 80 100
X, DISTANCE TO SMELTERS (miles)
I
200
300 400
Figure 31. Log-log plot of extinction/emission coefficient
versus distance to smelter.
53
-------
other complicating factors involved, this conclusion cannot be stated with
great confidence.
Our analysis of emission and extinction changes during the 1967-1968
copper strike leads to following principal findings: (1) at distances 50 to
200 miles from the source, 1000 TPD of SOX emissions in the Southwest ap-
parently produces an increase in regional extinction on the order of .01 to
.03 (10 m) ; (2) regional extinction produced by an SOX emission source in
the Southwest rna/ tend to be inversely proportional to the distance from
the source; (3) at distances of 250 to 375 miles from the source, 1000 TPD
of SOx in the Southwest may produce an increase in regional extinction on the
/i -i
order of .01 (10 m) ; and (4) at distances 10 to 15 miles from the source,
1000 TPD of SOx in the Southwest may produce an increase in regional ex-
tinction as high as .1 to .25 (10 m)-l.
54
-------
CHAPTER 5
REGIONAL EXTINCTION/EMISSION COEFFICIENTS BASED ON
HISTORICAL EMISSION CHANGES FROM 1948 TO 1975
As demonstrated in Chapter 2, large changes in SOX, NOX, and NMHC
emissions have occurred in the Southwest over the period 1948 to 1975.
This chapter attempts to derive regional extinction/emission coefficients
by relating the historical emission changes to corresponding visibility
changes through a correlation/regression analysis.
METHODOLOGY
The basic methodology of this chapter is to regress yearly values of
median extinction (based on airport data) against yearly emission levels
from 1948 to 1975. Multiple linear regressions are run of the form
B = a + CjSOx + c2NOx + c3NMHC (1)
where the units of extinction (B) are [ 10 meters]" , and the units of
emissions are 1000 TPD. The emission variables are usually chosen as state-
wide totals or AQCR totals (which are adjusted statewide values, see Appen-
dix C), but sometimes we also use multi-state emission totals or emissions
of individual source types (e.g. copper smelters). The regressions are run
step-wise, retaining only those coefficients that are greater than zero at a
95% significance level. The regression coefficients, c^, represent extinc-
tion/emission coefficients with units of [(104 m)~l/(lQQQ TPD)].
There are several limitations to our historical regression analysis.
One very important drawback is the intercorrelation that exists among the
"independent" variables (see Table 5); we are faced with the classical prob-
lem that everything correlates with everything. The high intercorrelations
and limited number of data points (at most 28 years) make it an insurmount-
able problem to weed out, with certainty, the individual impacts of all
three emission variables (SO. NO. and NMHC).
X X
A second important drawback is the omission of primary particulate
emissions from the analysis. Macias et al (1979) found that, in nonurban
areas of the Southwest, approximately h of the aerosol light scattering is
presently due to primary fine silicon, while approximately 12 is due to
55
-------
TABLE 5. INTERCORRELATION AMONG STATEWIDE
TOTAL EMISSION TRENDS, 1948-1975
ARIZONA COLORADO
SOx NOX NMHC SOX NOX NMHC
SOX 1.00 0.78 0.91 1.00 0.49 0.15
NOX 1.00 0.86 1.00 0.87
NMHC 1.00 1.00
NEVADA UTAH
SOX NOX NMHC SOX NOX NMHC
SOX 1.00 0.76 0.72 1.00 -0.68 -0.81
NOX 1.00 0.77 1.00 0.85
NMHC 1.00 1.00
sulfates. Because of the apparently significant contribution from primary
particles, and because of possibly important intercorrelations between pri-
mary particulate trends and gaseous precursor trends, it would be worthwhile
to expand our analysis to include primary aerosols when adequate data be-
come available (see discussion at the end of Chapter 1). If primary aerosol
emission trends are highly intercorrelated with the gaseous precursor trends,
the extinction/emission coefficients derived in this chapter would tend to
be somewhat overestimated.
Another significant limitation is that our linear regression form neg-
lects nonlinear interactions and synergistic effects among the three precur-
sor emissions. It is well known that photochemical smog depends on NOX and
NMHC emissions in a rather complicated, nonlinear manner. Also, the visibi-
lity impacts of SOX may depend on the level of the photochemical precursors
(NOX and NMHC), because photochemical oxidant can be an important factor go-
verning the conversion of S02 to sulfate. For the same reason, the visibility
impacts of changes in the photochemical precursors may depend on the exist-
ing SOX level. A statistical analysis of 28 yearly data points cannot hope
56
-------
to provide a complete characterization of these nonlinear and synergistic
effects.
Other limitations include errors in the data base, i.e. variations in
the visibility data (produced by weather fluctuations or unaccounted for
changes in observation procedures) and possible errors in our emission
trend estimates. Uncertainty is also introduced in the process of attempt-
ing to select the appropriate emission source area to be linked with each
set of visibility data.
Despite the potentially severe drawbacks in the analysis, we will pro-
ceed with the regression study for eight visibility trend sites. The inter-
pretation of the results will be guided by physical reasoning and the
limitations of the method will be borne in mind. Although we cannot hope
to derive definitive extinction/emission coefficients for SOX, NOX, and
NMHC on various spatial scales, the analysis should help to identify the
emissions that are most significant to visibility in various areas and may
yield approximate estimates of the extinction attributable to those emissions.
The eight trend sites selected for the regression study are Tucson,
Phoenix, Prescott, and Winslow in Arizona; the large metropolitan areas of
Denver and Salt Lake City in the northern part of the study area; and two
non-metropolitan sites (Grand Junction and Ely) in the northern part of the
study area. Cheyenne was excluded because it is not within the four study
states; Fort Huachuca, Colorado Springs, and Pueblo were excluded because
they had limited time periods of continuous data that did not extend into
the 1970's. With the exception of Prescott (75th percentile), Winslow (80th
percentile), and Ely (90th percentile), the yearly extinction values are
all based on median visibilities. Whenever a site relocation occurred that
was "apparently not significant" or "possibly significant" (see Chapter 3),
the regression analysis was performed twice -- both for the entire time
period and for a sub-period excluding data possibly affected by the relocation.
REGRESSION ANALYSIS FOR ARIZONA SITES
The regression analysis for Arizona involves six data sets: Tuscon
(1950-1975), Tuscon (1959-1975), Phoenix (1959-1975), Winslow (1948-1973),
Prescott (1948-1975), and Prescott (1948-1969). Tuscon and Prescott are
considered twice because of site relocations (see Chapter 3). The emission
variables considered in the analysis include total Arizona SOX, Arizona
57
-------
smelter SOX, Arizona non-smelter SOX, total Arizona NOX, and total Arizona
NMHC.*
Table 6 summarizes the correlation coefficients between historical ex-
tinction levels and corresponding emission levels. It is noteworthy that all
six data sets correlate best with total Arizona SOX and Arizona smelter SOX.
The correlations are especially strong at Tucson and Phoenix. The results
in Table 6 appear to be consistent with the finding in Chapter 2 that the
copper smelters have historically been the overwhelming source of secondary
aerosol precursors in Arizona. In fact, if we hypothesize that smelter SOx
emission trends were the basic cause of the extinction trends, the correla-
tion (or lack of correlation) between extinction and the other emission
variables might be partly explained as artifacts of the intercorrelation of
smelter SOX with those other emission variables (.999 with total Arizona SOX,
-.31 with non-smelter SOX, .71 with Arizona NOX, and .89 with Arizona NMHC).
TABLE 6. HISTORICAL EXTINCTION-EMISSION CORRELATIONS
FOR ARIZONA TREND SITES
CORRELATION COEFFICIENTS
DATA SET
Tucson (1950-1975)
Tuscon (1959-1975)
Phoenix (1959-1975)
Winslow (1948-1973)
Prescott (1948-1975)
Prescott (1948-1969)
Arizona
SOX
.92
.89
.82
.67
.70
.71
Arizona
Smelter
SOX
.91
.88
.81
.68
.70
.70
Arizona Non-
Smelter SOX
.03
.10
-.01
-.48
-.12
-.25
Arizona
NOX
.74
.54
.37
.49
.49
.59
Arizona
NMHC
.85
.83
.55
.59
.63
.59
Stepwise multiple regressions of extinction versus Arizona smelter
, Arizona non-smelter SOX> Arizona NOX, and Arizona NMHC select Arizona
Note that considering Phoenix/Tucson AQCR totals for SOX, NOX> and NMHC would
yield identical correlations because we have assumed that the AQCR emissions
are a constant fraction of total Arizona emissions. If we wish to consider
Phoenix/Tucson AQCR emissions, we need only make an adjustment in the re-
gression coefficients according to the parameters given in Appendix C.
58
-------
smelter SOX as (by far) the most significant term for each of the six data
sets. For five of the data sets, Arizona smelter SOX is the only significant
*
variable. Table 7 summarizes the results of regressing extinction versus
Arizona smelter SOX. The consistency of the extinction/emission (regression)
coefficients from site-to-site- and the high t-statistics at all sites are
very encouraging. It is also encouraging that the extinction/emission coef-
ficients are of approximately the same magnitude as (actually slightly higher
than) the .01 to .03 (104 m)~1/(1000 TPD) values obtained in the copper strike
analysis at sites on the order of 50 to 200 miles from smelters. The reader
should note that the historical trend analysis is essentially independent of
the 1967-1968 copper strike analysis. The 1967-1968 copper strike just par-
tially affects 2 of the 28 yearly data points, and these two data points are
not outliers in the regression (note that yearly emission levels for 1967 and
1968 were at or above the levels for the entire 1950's). We have redone the
historical regression analyses eliminating the data for 1967-1968 and have ob-
tained nearly identical correlation coefficients, regression coefficients, and
t-statistics.
TABLE 7. REGRESSIONS OF HISTORICAL EXTINCTION LEVELS
VERSUS HISTORICAL SMELTER SOX EMISSION LEVELS
DATA SET REGRESSION COEFFICIENT t-STATISTIC
(10* m)-l/(1000 TPD) (t^l.7 for 95% confidence)
Tucson (1950-1975)
Tucson (1959-1975)
Phoenix (1959-1975)
Winslow (1948-1973)
Prescott (1948-1975)
Prescott (1948-1969)
.035
.038
.041
.047
.031
.039
11.1
7.2
5.4
4.5
5.0
4.4
Figures 32 through 35 illustrate graphically the especially strong re-
lationships between smelter emissions and visibility at Tucson and Phoenix.
**
*
The only exception is Tucson (1959-1975) where Arizona NMHC enters as a
secondary variable.
**
Note that the data point for 1976 is added to these figures. All of our re-
gression analyses extended only to 1975 because we could obtain data on emis-
sions for most sources only through 1975. Data for both visibility and smel-
ter emissions were, however, available through 1976.
59
-------
.45-
o
IH
t
5 .40-
.35-
Smelter SO Emissions
/\
Median Extinction
I1III11IIIIIIII1IIIIf\
rir~|r~
70 75
oo
-6000 o
t i
oo
00
U4000
X
O '
00 >-
h- 00
_ I 1Z.
UJ O
-2000
50
Note: o versus denotes the "apparently not significant"
site relocation in 1958 (see Chapter 3).
80 _
£ 60 _
c< o
o
u. >-
O h-
UJ CO
CJ ><
d.
20 _
0 -
Smelter SO Emissions
A
00
Hours Visibility < 60 miles
(II
50
55
60
Year
65
i '
70
75
6000
5000
-4000
-3000
oo
zr
o
00
oo
X
O
oo >-
I oo
_ I z
UJ O
2000 §
D;
Figure 32. Tucson time series data on extinction and hours of
restricted visibility versus historical smelter SO
emissions for Arizona. >
60
-------
t
o
-
_l
C£
.70 -
.65 -
.60
.55
Median Extinction
Smelter SO Emissions
X
| i T i i | . - .r | i r i i | i
50 55 60 65
I
70
l
75
_ 6000 oo
o
- 5000
00
00
_ 4000 S^:
fV Q
UJ \
H- 00
1- 3000 do
s: i
oo ~'
U 2000 §
Year
00
00
o: o
_
o h-
UJ CO
O HH
80-
70-
60-
50-
£ 40-
30^
Hours Visibility < 40 miles
Smelter SO Emissions
A
50
55
1 ' I
60
65
I
70
1- 6000
o
\- 5000 to
ii
s:
UJ
L 4000 o^
oo >-
- 3000 Liz
UJ O
- 2000
75
Year
Figure 33. Phoenix time series data on extinction and hours
of restricted visibility versus historical smelter
SO emissions for Arizona.
A
61
-------
E
3-
O
.45-
z
X
n .40-
Q
ex:
<
.35_
oooo
3000
4000
5000
6000
ARIZONA SMELTER SOV EMISSIONS (TONS/DAY)
A
Note: o versus denotes the "apparently not significant"
site relocation in 1959 (see Chapter 3).
Figure 34. Yearly median extinction at Tucson versus smelter
SO emissions for Arizona.
X
62
-------
.70-
E
;t-
O
.65-
X
LlJ
.60-
>-
_J
<
.55-
3000
4000
3000
6000
ARIZONA SMELTER S0v EMISSIONS (TONS/DAY)
X
Figure 35. Yearly median extinction at Phoenix versus smelter
SO emissions for Arizona.
63
-------
Figures 32 and 33 present time series plots of yearly extinction versus
smelter SOX and of yearly hours with reduced visibility versus smelter SOX.
Figures 34 and 35 present scatter diagrams of yearly extinction versus smel-
ter SOX emissions.
The above trend analysis apparently indicates that changes in copper
smelter SOX emissions were the principal factor affecting historical visi-
bility levels at Tucson, Phoenix, Winslow, and Prescott. This conclusion
makes sense physically, considering that smelters have been the overwhelming
source of secondary aerosol precursor emissions in Arizona. The analyses
for the six data sets yield extinction/emission coefficients on the order of
.04 i .005 (104 nO'VflOOO TPD SOX). This value for the extinction/emission
coefficient appears relevant to regional haze at distances on the order of
*
50 to 200 miles from the emission source.
Because of problems with respect to intercorrelated emission variables
and limited number of data points, we could not isolate the effect of NOX
and NMHC emissions on extinction trends in Arizona. However, there are in-
dications that, although these effects appear to be secondary to the impact
of the large SOX emissions in Arizona, they nevertheless are significant.
Some evidence concerning the importance of NOX and NMHC to visibility in
Arizona is as follows:
The emission/extinction coefficient for smelter SOX that we
derived from the trend analysis, .04 (104 m)-l/(1000 TPD SOX),
is higher than the emission/extinction coefficient that we
derived from the copper strike analysis, .01 to .03 (104 m)~V
(1000 TPD SOX). This could be explained in part by the fact
that the copper strike essentially involved changes only in
SOX,** while the trend analysis involved concurrent long-term
increases in NOX and NMHC. The regression coefficients in the
trend analysis may be slightly inflated because they also
partly represent the effect of intercorrelated NOX and NMHC
emissions.
Median visibility in central Phoenix is 40 miles, significantly
lower than the 60 mile median visibility found in nonurban
areas of southeastern Arizona. Some of this difference may
*
The weighted average distances between the sites and the smelter SOX
emissions are as follows: Tucson - 80 miles, Phoenix - 120 miles, Winslow -
170 miles, and Prescott - 190 miles.
**
Simple calculations convincingly indicate that reduced commuter travel by
smelter workers during the strike would produce negligible changes in NOX
and NMHC emissions.
64
-------
be due to greater suspended dust levels in the city. However,
the difference may also be explained by the presence of urban
photochemical smog (from NOx and NMHC emissions) which contains
nitrate and organic aerosols and also serves to increase the
amount of S02 converted to sulfate.
The absolute reductions of sulfate and extinction during the
1967-1968 copper strike tended to be greater in urban areas
than in suburban and nonurban areas. This probably reflects
the role of urban photochemical smog in enhancing the amount
of S02 converted to sulfate.
REGRESSION ANALYSES FOR OTHER LOCATIONS
The regression studies for the other four sites (Salt Lake City, Den-
ver, Grand Junction, and Ely) did not fare as well as the Arizona regressions.
The analyses provided some hints that the impact of the Arizona smelters may
extend well north of Arizona, and that growth in the photochemical precur-
sors (NOx and NMHC) was the key factor related to historical visibility
changes in the urban areas of Salt Lake City and Denver. However, because
of inconsistencies in the regression coefficients from site to site, and
because the correlations (except for Salt Lake City) were rather low, we
cannot place great faith in these conclusions. The lack of very large local
emission changes (such as that which occurred among the Arizona smelters) and
the problem of intercorrelated emission variables made it impossible to
quantify, with confidence, extinction/emission coefficients based on the
regressions for these four sites.
The following subsections describe the regression analyses for the
four sites in the northern part of the study region. The sites are discussed
in descending order with respect to the degree of correlation found in the
regression studies.
Salt Lake City
The Salt Lake City regressions were conducted for the years 1955 to
1975. In the first set of regressions, the emission variables included
total NOX, total NMHC, and total SOX for the Salt Lake City (Wasatch Front)
*
AQCR. In some cases the SOX variable was divided into two variables:
*
Identical correlations would be obtained using total Utah emissions, since
we have assumed that the AQCR emission trends are constant multiples of
respective trends for the entire state (see Appendix C).
65
-------
Magna copper smelter SOX and non-smelter SOX.
The extinction trends for Salt Lake City correlated significantly with
total NOX (R = .87) and total NMHC (R = .78), but insignificantly with
the various SOX emission trend variables. A colinearity problem exists be-
cause (as shown in Table 5 for Utah) NOX and NMHC trends are highly inter-
correlated, and both are significantly correlated in a negative manner with
SOx. Essentially, the extinction trends at Salt Lake City from 1955 to 1975
followed the upward trends in NOX and NMHC rather than the downward or con-
stant trends in SOX.
The multiple regressions indicate significant regression coefficients
for both NOX (t-statistic = 4.5) and NMHC (t = 2.3). Compared to our pre-
vious results, the extinction/emission (regression) coefficients are quite
high: 1.26 (104 mJ'VdOOO TPD) for NOX and 1.39 (104 mJ'VdOOO TPD) for
NMHC. The rather large magnitude of these coefficients might partly be ex-
plained by the localized (air basin) scale of the analysis. In Chapter 4
we noted that the extinction/emission coefficients for SOX seemed to become
much greater when the distance from the source was reduced to less than 50
miles.
The importance of the photochemical pollutants (NOX and NMHC) to ex-
tinction in Salt Lake City is not inconsistent with our previous findings
for that area. For example, we have noted previously that extra extinction
(extinction above-and-beyond blue-sky scatter) dropped only 8% at Salt Lake
City with the SOX reduction of the 1967-1968 copper strike (Trijonis and Yuan,
1978a). Also, an extinction budget for Salt Lake City (Trijonis and Yuan,
1978a) indicates that sulfates account for only one-third of extra extinction
in that area. Even the fraction of extinction attributable to sulfates may
depend on photochemical precursor emissions as well as SOX emissions because
of the potential role of photochemical smog in oxidizing SOz to sulfate.
A second set of regressions was run for Salt Lake City including
Arizona smelter SOX emissions as well as the local emission variables.
Arizona smelter emissions correlated significantly with Salt Lake City ex-
tinction (R = .79) and entered the regression equation as a secondary but
significant variable (t = 3.6) with a regression coefficient of .03 (10 m)~ /
(1000 TPD). Based on this result, one might speculate that the Arizona
66
-------
smelter emissions have a significant effect as far as Salt Lake City. How-
ever, because of colinearity problems (Arizona smelter SOX trends happen to
correlate well with Salt Lake City NOX and NMHC trends) and because of the
extremely questionable physical basis for the conclusion (e.g. the long
transport distances over terrain of increasing elevation), we would place
very little confidence in that speculation.
Denver
Because of a "possibly significant" site relocation in 1969 (see
Chapter 3), the Denver regression study was performed for two time periods:
1948 to 1968 and 1948 to 1975. The independent variables used in the re-
gressions were total NOX, total NMHC, and total SOX emission trends for the
Denver AQCR. For both time periods, extinction correlated at a marginal
significance with NMHC emissions (R = .39 and .50) and NOX emissions (R =
.39 and .41) but insignificantly with SOX emissions.
The stepwise multiple regressions selected NMHC emissions as the only
significant variable with regression coefficients of 0.27 and 0.29 (10 m)~ /
(1000 TPD) and t-statistics of 1.9 and 2.9 respectively for the two time
periods. There is a major colinearity problem, however, between NMHC and
NOX emissions which correlated with each other at a level of 0.87 (see Table
5). Excluding NMHC from the regressions left NOX as the only significant
variable with regression coefficients of 0.40 and 0.23 (10 m) /(1000 TPD)
and t-statistics of 1.8 and 2.3. The regression (extinction/emission) co-
efficients are much higher than those found for SOX in the Arizona study;
again this could reflect the more localized nature (air basin scale) of the
effects being considered.
The Denver results are consistent with the Salt Lake City results in
implicating NOX and NMHC as the most significant local emission variables
related to historical extinction trends. We cannot be very certain, however,
about the Denver results because the statistical significance of the cor-
relation and regression coefficients is only marginal.
In another set of regressions, we added Arizona smelter SOX to the
list of emission variables for Denver. Arizona smelter SOX correlated with
Denver extinction at levels of .60 and .61 for the two time periods, respec-
tively, and became the only variable retained in the multiple stepwise re-
gressions with regression coefficients of .027 and .021 (10 m) /(1000 TI'U)
and t-statistics of 3.2 and 3.9. This seems to imply that Arizona smelter
67
-------
SOX emissions may affect visibility at Denver. However, because of the
colinearity problems (Arizona smelter SOX trends correlate highly with
Denver NOX and NMHC trends) and the questionable physical basis of linking
Arizona smelter SOx with Denver visibility, we cannot attach confidence to
this conclusion.
Grand Junction
Because of the "apparently not significant" site relocation in 1950,
the Grand Junction analysis was performed for two time periods: 1951-1975
and 1948-1975. The emission variables chosen for the analysis were Colo-
rado plus Utah totals for SOX, NOX, and NMHC as well as Arizona smelter
SOX emissions.
The multiple stepwise regression based on 1951-1975 data selected
Arizona smelter SOX as the only significant variable with a marginal cor-
relation coefficient of 0.39. The regression coefficient, .005 (10 m)~ /
(1000 TPD), was marginally significant (t = 2.0) but seemed consistent with
the values obtained for Arizona sites % .04 (10 m)~ /(1000 TPD) consider-
ing the much greater distance to the smelters.
The regressions based on 1948-1975 data indicated that none of the
emission variables were significant at the 95% confidence level. Two of
the variables, Colorado and Utah total SOX and Arizona smelter SOX, were
almost significant at the 95% confidence level. The regression coefficients,
.02 (104 m)-l/(1000 TPD) for Colorado plus Utah SOX and .004 (104 m)-l/
(1000 TPD) for Arizona smelter SOX, are fairly consistent with the results
of the copper strike analysis, considering the transport distances involved.
These results should be viewed as very tenuous, however, because of the
lack of statistical significance and the ever-present colinearity problem.
The Ely regressions were conducted for two periods: 1962-1975 and 1955
1975 (a "possibly significant" site relocation occurred in 1961). The
emission variables used included total SOX, smelter SOX, non-smelter SOX,
total NOX, and total NMHC for Nevada as well as Arizona smelter SOX.
The extinction trends for 1962-1975 did not correlate significantly
with any of the emission variables, but the extinction trends for 1955-
1975 correlated marginally with all the variables. Multiple stepwise
68
-------
regression with the 1955-1975 data selected Nevada NMHC as the only signi-
ficant variable, but the regression coefficient, 9.3 (104 mJ'VtlOOO TPD),
was far out of line with the values obtained in our other analyses. The
analysis did not make much sense at Ely, especially considering the prox-
imity of that site to the McGill smelter and our earlier finding that Ely
showed a significant improvement in visibility during the 1967-1968 copper
strike.
We are not sure why the regression analysis fared so poorly at Ely,
but one reason might be the fact that we used a very high (90th) percentile
extinction at Ely in order to avoid extrapolation of the visibility fre-
quency distribution. The 90th percentile extinctions (lowest visibilities)
might be predominantly affected by yearly weather fluctuations (e.g. pre-
cipitation days) which could mask the effect of emission changes.
69
-------
CHAPTER 6
CONCLUSIONS AND ILLUSTRATIVE APPLICATIONS
This chapter summarizes our conclusions concerning extinction/emission
coefficients for the Southwest and demonstrates how such coefficients can
be applied. It must be emphasized that the applications presented here
should be regarded only as illustrative because high uncertainty exists in
the extinction/emission coefficients which serve as the bases for these ap-
plications. Further work needs to be done both with empirical analyses and
large-scale modeling calculations before we can be more certain about the
relationship between aerosol precursor emissions and regional haze in the
Southwest.
EXTINCTION/EMISSION COEFFICIENTS FOR THE SOUTHWEST
This section lists the conclusions we have reached with respect to
regional extinction/emission coefficients for the Southwest. The discussion
is organized according to three spatial scales: mesoscale (^ 50 to 200
miles from the source), air basin scale fa 50 miles from the source), and
synoptic scale (^ 200 to 500 miles from the source).
Mesoscale (^ 50 to 200 miles)
Airport visibility data and NASN sulfate data during the 1967-1968
copper strike both indicate that the regional extinction/emission coef-
ficient for SOX emissions is approximately .01 to .03 (104 ra)"1/(1000 TPD
SOx) on a scale of 50 to 200 miles from the source, For the same spatial
scale, the historical trend regressions with 6 Arizona data sets imply an
extinction/emission coefficient of .04 t .005 (104 mJ'VdOOO TPD SOx).
The coefficient derived from the Arizona regression analysis, however, may
be somewhat inflated because of colinearity problems (i.e. historical in-
creases in NOx and NMHC occurred concurrently with the increasing trend in
SOx). Considering all our analyses, a regional extinction/emission coef-
ficient of .01 to .03 (10 m) /(1000 TPD SOX) seems most reasonable for
the mesoscale. Unfortunately, we could not derive estimates of extinction/
emission coefficients pertinent to mesoscale effects of NOX and NMHC.
Note that all of our estimates of extinction/emission coefficients for
gaseous precursors may be somewhat inflated (possibly on the order of 10-
30% or more) by the omission of primary particulates from the analysis.
70
-------
Air Basin Scale (£ 50 miles)
There is a very high degree of uncertainty in our estimates of
extinction/emission coefficients for the air basin scale (£ 50 miles from
the source) in the Southwest. The 1967-1968 copper strike analysis in-
dicated that SOX emissions might produce extinction on the order of .1 to
.4 (10 m) /(1000 TPD) within 10 to 15 miles from a source, but this con-
clusion was made tenuous by several complicating factors in the analyses
(see Chapter 4, page 52). The regression studies for Salt Lake City and
Denver implicated photochemical precursor (NOX and NMHC) emissions as being
most closely tied to historical visibility trends in those urban areas and
implied that the extinction/emission coefficients might be on the order of
.2 to 1.4 (104 m)"1/(1000 TPD) for both NOX and NMHC on an air basin scale.
This conclusion, however, was uncertain because of statistical colinearity
problems in the regression analysis.
Synoptic Scale (^ 200 to 500 miles)
Our results concerning the effect of aerosol precursor emissions at
distances greater than 200 miles from the source are likewise very un-
certain. Sulfate data during the copper strike imply an extinction/emission
coefficient on the order of .01 (104 m)"1/(1000 TPD SOX) at distances 250 to
375 miles from the copper smelters, but visibility data at two airports fail
to confirm significant effects at these distances during the strike. The
historical regression analyses for Grand Junction, Denver, and Salt Lake City
indicate statistically significant extinction/emission coefficients of .005
to .03 (104 mJ'VdOOO TPD SOx) at distances of 400 to 500 miles from the
smelters, but this result is clouded by problems of colinearity in the re-
gressions. It may be that an extinction/emission coefficient on the order
of .01 (10 m) /(1000 TPD) is appropriate for synoptic scale SOX effects
in the Southwest, but this conclusion should be confirmed by analyzing
several more airports during the 1967-1968 copper strike before we can
*
lend it any credence.
*
The visibility analyses reported here are based only on airport data
available in computerized form. Hard-copy visibility data may be
available for several other airports in northern Arizona and southern
Utah/Colorado.
71
-------
ILLUSTRATIVE APPLICATIONS
This section presents two examples which illustrate how extinction/
emission coefficients might be used to conduct order of magnitude calcu-
lations concerning regional haze in the Southwest. In the second example
we obtain a gross check on our results by comparing our predictions to
existing geographical patterns of nonurban visibility in the Southwest.
Mesoscale Haze from a Large Source of SOX
As a simple hypothetical example, we consider the following: Assuming
a background visibility of 75 miles, what would be the effect of a 500 TPD
SOx source on regional haze at distances of 50 to 200 miles from the source?
*
From the Koschmeider relationship, the 75 mile background visibility
translates to an extinction of 0.324 (10 m)~ . Assuming an extinction/
emission coefficient of .02 1 .01 (104 mJ'VdOOO TPD SOX), the 500 TPD SOX
would increase mesoscale extinction by .01 ħ .005 (10 m)~ . Adding this
to the background extinction and substituting back into the Koschmeider
relationship, we conclude the mesoscale effect would be to reduce visibility
from 75 miles to 72.8 ħ 1.1 miles.
Existing Patterns in Monurban Visibility
A companion report on existing visibility levels in the United States
(Trijonis and Shapland, 1978) uses airport, nephelometry, and photographic
photometry data for 1974-1976 to characterize the geographical pattern of
nonurban visibility in the Southwest. It is found that median nonurban
visibility is on the order of 78 miles in Utah, Colorado, and Nevada but
falls to around 60 miles in central and southern Arizona. Can this pattern
be accounted for by the mesoscale effects of the copper smelter emissions
in Arizona?
Noting that the Arizona smelters averaged 3720 TPD for the period
1974-1976, and assuming a mesoscale extinction/emission coefficient of .02
(104 m)~1/(1000 TPD SOX), calculations parallel to those in the last example
indicate that the Arizona smelters would reduce a background visibility of
78 miles down to 63 miles. The agreement with the observed nonurban visi-
bility in southern/central Arizona (60 miles) is an encouraging sign that
See Equation (1), page 42.
72
-------
we may have the right order of magnitude for the mesoscale SOx extinction/
emission coefficient.
73
-------
REFERENCES
AQHP (Air Quality Maintenance Plan), Preliminary Emissions Inventory and
Air Quality Forecast 1974-1995, Final Report of the Boundary and Forecasting
Committee of the Air Quality Maintenance Planning Policy Task Force for
Southern California, April 1976.
Cass, G.R., "The Relationship Between Sulfate Air Quality and Visibility
at Los Angeles," Caltech Environmental Quality Laboratory, Memorandum #18,
August 1976.
EPA (Environmental Protection Agency), "Monitoring and Air Quality Trends
Report, 1974," EPA-450/1-76-001, February 1976.
Latimer, D.A., et al., "The Development of Mathematical Models for the
Prediction of Anthropogenic Visibility Impairment," Draft Final Report,
Prepared at Systems Applications, Inc., San Rafael, California, Under
Contract #68-01-3947, for Environmental Protection Agency, 1978.
Macias, E.S., et al., "Characterization of Visibility Reducing Aerosols in
the Southwestern United States: Interim Report on Project VISTTA", Prepared
by Meteorology Research Inc. under EPA Contract No. 68-02-2713, January 1979.
Niemann, B., Regional modeling studies in progress under contract to EPA
Office of Research and Development, Teknekron, Berkley, California, 1978.
NOAA (National Oceanic and Atmospheric Administration), Climatic Atlas of
the United States. National Climatic Center, Asheville, N.C., 1977.
Trijonis, J. and K. Yuan, "Visibility in the Southwest: An Exploration of
the Historical Data Base," EPA-600/3-78-039, April 1978 (a).
Trijonis, J. and K. Yuan, "Visibility in the Northeast: Long-Terrn Visibility
Trends and Visibility/Pollutant Relationships," EPA-600/3-78-075, 1978 (b).
Trijonis, J. and D. Shapland, "Existing Visibility Levels in the U.S.-
Isopleth Maps of Visibility in Suburban/Nonurban Areas During 1974-1976,"
Prepared at Technology Service Corporation, Santa Fe, N.M., Under Grant
#802815 for EPA Office of Air Quality Planning and Standards and EPA
Environmental Science Research Laboratory, 1978.
Ursenbach, W.O., et al., "Visibility Models for the Arid and Semi-Arid
Western United States," Paper # 78-43.6, Presented at the 71st Annual
Meeting of the Air Pollution Control Association, Houston, Texas, 1978.
Waggoner, A.P., et al., "Sulfate-Light Scattering Ratio as an Index of the
Role of Sulphur in Tropospheric Optics," Nature, Vol. 261, No. 5556, pp. 120-
122, 1976
White, W.H. and P.T. Roberts, "On the Nature and Origins of Visibility-
Reducing Aerosols in the Los Angeles Air Basin," Atmospheric Environment.
Vol. 11, p. 803, 1977.
74
-------
APPENDIX A
EXPLANATION OF EMISSION TREND ESTIMATES
This appendix describes how 1948 to 1975 emission trends were cal-
culated for sulfur oxides (SOX)> nitrogen oxides (NOX), and non-methane
hydrocarbons (NMHC) for the states: Arizona, Colorado, Nevada and Utah.
It also includes tables listing detailed trend estimates for each source
type considered.
Introduction
NEDS, the 1973 National Emissions Report (EPA, 1976), provides emis-
sion inventories which indicate that in each of the states considered, 80-
100% of the SOX, 90-100% of the NOX, and 70-90% of the HC emissions are
produced by a relatively few source types: copper smelters, steam-electric
power plants, gasoline vehicles, diesel vehicles, railroads, petroleum indus-
tries , industrial boilers, commercial/residential sources, gasoline evap-
oration, and solvent evaporation. A good estimate of the historical trends
in emissions can be made by considering the contributions from just these
sources.
Discussions in AP-42, Compilation of Air Pollution Emissions Factors
(EPA, 1977), supplied methods and insights to determine emission factors
for most of these source types. When a method for estimating emission
factors for a specific source type was not readily available, we often re-
sorted to computing an emission factor so that our 1973 emission estimate
is consistent with the corresponding datum reported in NEDS.
A major contribution to our historical data base was made by FEDS,
Federal Energy Data System: A Statistical Summary (DOE, 1978), the Depart-
of Energy's compilation of fuel use data by consumer type, state, and
year, back to 1960. Other valuable supplies of data come from the Minerals
Yearbooks (BOM, 1948-1975), annual publications of the Department of In-
terior's Bureau of Mines, which report in considerable detail the produc-
tion activities and sales of the minerals/fuels industries. Highway
75
-------
Statistics: A Summary to 1975 (DOT, 1975) provides detailed data on gaso-
line and diesel fuel consumption, essential to several of our calculations.
Additional data were supplied by many other sources; these are mentioned
.in the discussions that follow.
There are many more sources of data which we collected, considered,
and discarded for one reason or another. Usually these sources either were
discarded because they were less complete than some other data source or
they were discarded in favor of data which seemed to be a more direct in-
dication of the activity under consideration.
Copper Smelters
Historically, copper smelters have produced the preponderance of SOX
emitted in the Southwest. The smelters of relevance to this trend study
*
include seven in Arizona, one in Nevada, and one in Utah.
For 1972 and later years, detailed SOX emission data for Arizona
copper smelters are presented in the First Annual Report on Arizona Copper
Smelter Pollution Control Technology (Larson and Billings, 1977). Yearly
emissions prior to 1972 are determined by factoring according to Arizona
copper production as listed in the Minerals Yearbooks. The production re-
lated emission factor is adjusted in 1966, 1969, and 1971 to account for the
partial SOX controls installed at Phelps Dodge/Morenci, Kennecott/Hayden,
**
and Asarco/Hayden during those years (Weisenberg et al., 1976).
SOX emissions at the Kennecott/McGill, NV smelter were approximately
700 tons/day in 1973 (EPA, 1976; Weisenberg et al., 1976). Emission trends
for all other years are assumed proportional to copper production in Nevada,
as reported in the Minerals Yearbooks, because no controls were installed
on the McGill smelter.
SOX emissions at the Kennecott/Salt Lake City smelter are determined
from data on average daily emissions and number of yearly operating days
*
There are also two copper smelters operating in New Mexico, but New Mexico
emission trends are not included in this report.
**
The overall percent sulfur removed in Arizona smelters has changed his-
torically as follows: 3% in 1965, 6% in 1966, 9% in 1969, 13% in 1971,
16% in 1972, 21% in 1974, 37% in 1975, and 50% in 1976.
76
-------
supplied for each year (1948-1975) by Kennecott Copper (Heaney, 1978).
Table A-l summarizes the copper smelter SOX emission trends for the
three states.
TABLE A-l. TRENDS IN SOX EMISSIONS FROM COPPER SMELTERS
SOX (TONS/DAY)
YEAR Arizona Nevada Utah
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
2650
2540
2850
2940
2800
2790
2670
3210
3570
3650
3440
3040
3810
4150
4550
4670
4890
4970
5070
3440
4310
5320
6030
5200
5570
5780
4770
3640
340
280
390
420
430
460
520
590
600
580
490
420
570
580
620
610
500
530
590
380
570
780
800
720
750
700
630
600
990
970
1300
1180
1160
1110
860
770
900
500
400
270
460
580
680
520
500
600
600
290
420
640
520
350
410
560
510
440
77
-------
Power Plants
Sulfur contents and emission factors for power plants were derived
from the fuel consumption and emission data reported in Steam-Electric
Plant Air and Water Control Data (DOE, 1969-73). Statewide factors as well
as factors for individual plants were computed for the years 1969, 1970,
and 1971. The average sulfur contents and emission factors for these three
years were assumed to apply throughout the 28 years under study.
The "Air and Water Control Data" only cover 5 years, so emission
trends were estimated using the fuel usage data listed in Steam-Electric
Plant Construction Cost and Annual Production Expenses (FPC, 1948-74).
These data were adjusted slightly so that they are consistent with the "Air
and Water Control Data" in the years where both are available. Neither data
source covers 1975, so the change in fuel usage reported in FEDS for that
year was used to estimate the pollutants emitted in each state.
Table A-2 lists the estimated NOX and SOX emitted by power plants in
each state during the years, 1948-1975. Table A-3 presents the same infor-
mation for some of the larger power plants in the area under study.
Gasoline Vehicles
In order to estimate the NOX and NMHC emitted by gasoline powered
vehicles, it was necessary to construct an average vehicle emission factor
for each year, each pollutant, and for low and high altitudes. This was
done using the method suggested in AP-42. The data used included the year-
ly number of vehicles in operation by model year (MVMA, 1976, 1977), the
average miles traveled per year by vehicle age (EPA, 1977), and the AP-42
emission factors corrected to an average speed of 40 miles per hour. These
considerations produced a high and low altitude, fleet-averaged emission
factor for each pollutant and for each year for both light duty vehicles
and heavy duty vehicles.
The emission factors relating to vehicle miles of light duty travel
and vehicle miles of heavy duty travel were combined with vehicle use data
(i.e. percent of total vehicle miles traveled by cars and trucks, and the
average miles per gallon for cars and trucks (DOT, 1967 and 1974; MVMA, 1977)
78
-------
TABLE A-2. TRENDS IN SOX AND NOX EMISSIONS FROM
STEAM-ELECTRIC POWER PLANTS
YEAR
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
Arizona
(tons/day)
SOX NOX
0
0
0
0
0
0
0
0
0
0
0
0
0
0
19
14
11
6
7
7
8
7
1
2
6
45
97
157
4
4
5
6
5
8
11
15
21
20
20
23
31
37
49
42
39
25
31
31
34
40
41
46
51
71
101
136
Colorado
(tons/day)
SOX NOX
10
4
5
8
7
9
9
12
15
11
13
14
29
32
36
44
48
30
75
83
84
84
97
98
99
141
138
168
12
8
9
14
14
21
21
25
27
27
31
36
44
43
45
51
57
58
80
88
93
93
102
103
115
147
148
165
Nevada
(tons/day)
SOX NOX
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
5
8
9
14
17
15
33
78
85
99
103
0
0
0
0
0
0
0
0
3
4
4
5
6
9
10
9
12
18
24
24
29
36
39
68
117
124
129
128
Utah
(tons/day)
SOX NOX
2
2
2
3
8
15
13
18
18
25
27
20
32
32
24
22
23
22
23
22
23
23
25
26
25
29
34
58
2
2
2
2
7
13
16
23
24
28
25
26
27
28
23
21
21
21
25
23
22
21
22
21
23
27
31
51
79
-------
GO
1
z.
t i
O
^
> t
O
fV
U_
GO
z:
o
ğ-H
GO
t (
UJ
X
o
^^
Q
z
a;
X
0
GO
I I
GO
z.
UJ
0£.
1
OO
=c
LU
t
ca
h-
I
ID
r~
O
CT>
C
i
*-
_3
M"
<
* -
O
"-3
its
>
(O
^^
^^
c~
ro
3
f~3
C
rc
GO
p
7**r
i*-^1
O)
>
fO
^C^"
O
g
N ^
to
S-
QJ
i_ c
3 S-
o o
U_ 0
o
^
0)
CU
v
0
S-
O)
f-
C_J
o"
0
cu
o
-C
ro
O.
ro
^~.
ro
-o
in
C
0
+J
-^
rO
o
^^
(/)
c~
0
-M
^"^
* *.
^>
n3
T3
*i*^-
CO
c:
o
4-)
^_-/
x~*ğ
>^
fO
-o
^^^
to
c
0
4J
rc
-o
*^^
to
c~
O
40
> f
^
ro
"O
to
c
o
^
ra
-o
^*"
c
o
"-'
X
o
z.
X
o
GO
X
o
~^
X
o
GO
X
o
z.
X
o
GO
X
0
1ZL
X
o
GO
X
0
z
X
0
GO
X
o
"^
X
o
GO
X
0
z
X
o
GO
o;
UJ
I C\J
CSJ
1 OO
in CM
CO CO
en oo
Ğd-
CM
cxi oo
CM tn
^t- tn
CO
oo o 10 LD <-o
ii in un to <^>
icocvjro
I C\J C\J C\J CM CXJ
CM CO O CO CO
ooo- iLO
t < O
CO CTi
80
-------
to provide one fuel-use related factor for each year and for low and high
altitudes. These were then used in conjunction with the highway gasoline
data (DOT, 1975) to produce yearly estimates of emissions from 1948 to 1975.
Of the four states under study, only Colorado was taken to be high altitude.
The hydrocarbon emissions from gasoline powered vehicles are 98.5%
non-methane (Trijonis and Arledge, 1976). This factor was introduced to
produce the final NMHC estimates reported in Table A-4.
Diesel-Powered Vehicles
Trends in fuel use by diesel powered vehicles were estimated from
"Private and Commercial Highway Use of Special Fuels, By States, 1949-1975"
appearing in Highway Statistics: Summary to 1975 (DOT, 1975). These data
are not available before 1949, so the 1948 datum for each state was esti-
mated by factoring according to total fuel used (DOT, 1975).
In order to use the emission factors recommended in AP-42, the fuel
use data were converted into vehicle miles traveled using yearly average-
miles-per-gallon figures for trucks (DOT, 1967; DOT, 1974; MVMA, 1977).
These calculations produced the NOX emission trends listed in Table
A-5.
Railroads
The Association of American Railroads (AAR, 1957; 1967; 1977) provides,
in its yearly Statistics of Railroads of Class I in the United States, data
on the various fuels used by Class I railroads. The data were multiplied by
1.05 to make them more representative of railroads of all classes (AAR, 1957;
1967; 1977).
At the present time, locomotives operate on diesel fuel and AP-42
gives emission factors for them only. The fuels of the past (coal and fuel
oil) were assumed to have been burned in small, hand-fired, external com-
bustion engines, and AP-42 emission factors appropriate for such engines
were used.
Perry and Decarlo (1967) report a 2% average sulfur content for coal
produced in the United States in 1964. This was the figure assumed in our
calculations. Distillate fuel oil usually contains less than 0.3% sulfur
(EPA, 1977); 0.3% was used here.
81
-------
TABLE A-4. TRENDS IN NOX AND NMHC EMISSIONS FROM
GASOLINE-POWERED MOTOR VEHICLES
YEAR
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
Arizona
(tons/day)
SOX NOX
36
36
40
44
50
53
53
59
63
68
73
78
85
89
95
102
107
110
115
117
127
151
172
195
230
255
244
261
113
112
122
142
159
170
168
187
198
213
228
247
268
280
298
310
325
328
330
329
345
374
375
374
388
383
341
336
Colorado
(tons/day)
SOX NOX
41
44
47
52
55
57
59
62
64
67
69
74
76
80
83
90
93
97
100
103
116
128
151
173
201
221
229
248
237
253
276
300
317
332
341
360
369
386
396
424
438
439
479
510
522
535
543
545
573
582
595
596
619
582
535
535
Nevada
(tons/day)
SOX NOX
11
11
13
14
16
18
18
20
20
21
22
24
25
27
32
36
38
41
42
42
45
50
59
65
76
83
82
91
35
36
39
45
49
56
57
63
64
66
69
75
79
85
99
112
116
121
119
119
123
123
127
125
129
124
115
117
Utah
(tons/day)
SOX NOX
30
32
35
36
36
38
39
41
45
47
49
50
52
55
57
61
64
64
65
69
70
75
86
98
109
126
133
133
96
101
107
115
115
120
122
131
142
148
153
158
165
172
181
191
193
194
198
198
205
212
213
210
213
200
186
182
82
-------
TABLE A-5. TRENDS IN NOX EMISSIONS FROM DIESEL
POWERED VEHICLES
NOX (tons/day)
YEAR Arizona Colorado Nevada Utah
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
5
5
7
9
10
12
13
14
16
18
20
21
23
23
25
29
31
32
35
38
42
49
50
60
67
79
80
84
3
3
5
6
8
9
11
13
14
16
17
18
19
19
21
24
25
25
27
29
32
34
36
41
47
56
56
56
3
3
4
5
6
7
8
8
8
9
9
9
9
10
11
13
14
14
14
14
17
18
20
21
23
26
29
29
3
4
4
5
7
7
8
9
11
12
13
13
14
14
15
16
18
18
19
20
23
24
26
26
32
38
39
40
83
-------
These considerations provided an estimate of emissions produced annual-
ly on all railroads in the United States. In order to obtain an estimate of
emissions for each state, the nationwide totals were factored so that the
1973 results are consistent with the railroad emissions reported in NEDS.
Table A-6 contains the resulting figures.
.Petroleum Industry
In Colorado and Utah, petroleum industry emissions are nonnegligible.
The Minerals Yearbooks' report of yearly petroleum production by state was
used to produce estimates of emission trends. This was done by using the
petroleum industry emissions reported in NEDS as a base and causing them to
vary yearly as petroleum production.
The results are listed in Table A-7.
Industrial Sources
Combustion of coal, natural gas, fuel oils, gasoline, and diesel fuel
from industrial sources produces significant NOX and SOX. Emissions from
the use of each fuel were estimated using the methods described in AP-42.
In order to arrive at final emission factors, several assumptions were
made. Industrial coal was assumed to be 0.5% sulfur, while residual and
distillate fuel oils were taken to be 1% and 0.2% respectively. These
choices of sulfur contents produce emission estimates which are consistent
with those listed in NEDS for the year 1973.
Distillate and residual fuel data are from FEDS after 1959. Before
that, they are from "Mineral Industry Surveys" (BOM, 1949, 1951, 1953, 1957-
62). The lists of industrial and oil company fuels were used because these
data are most consistent with FEDS data in the years they overlap. In ad-
dition, the data were slightly adjusted to make them more comparable with
the FEDS data and to insure consistency in trend estimates.
The gasoline and diesel figures were computed from data given in FEDS
from 1960 to 1975. Before 1960, the data used are those given in Highway
Statistics: Summary to 1965 (DOT, 1967) as "Nonhighway Use of Motor Fuel".
These were factored so they match the FEDS data reasonably well in the years
84
-------
TABLE A-6. TRENDS IN SOX AND NOX EMISSIONS
FROM RAILROADS.
YEAR
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
Arizona
(tons/day)
SOX NOX
119
90
76
67
47
34
20
19
15
10
5
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
13
13
15
18
19
21
21
22
23
22
21
23
23
22
23
23
24
24
26
25
25
25
24
24
26
27
26
24
Colorado
(tons/day)
SOX NOX
139
105
89
78
55
40
24
22
17
12
6
5
4
4
4
4
4
4
5
5
5
5
4
4'
4
5
5
4
15
15
18
21
23
24
24
26
27
26
25
26
26
26
27
27
28
29
30
29
30
29
28
29
30
31
31
28
Nevada
(tons/day)
SOX NOX
60
45
38
34
23
17
10
9
7
5
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
6
7
8
9
10
10
10
11
12
11
11
11
11
11
11
12
12
12
13
13
13
13
12
12
13
13
13
12
Utah
(tons/day)
SOX NOX
139
105
89
78
55
40
24
22
17
12
6
5
4
4
4
4
4
4
5
5
5
5
4
4
4
5
5
4
15
15
18
21
23
24
24
26
27
26
25
26
26
26
27
27
28
29
30
29
30
29
28
29
30
31
31
28
85
-------
TABLE A-7. TRENDS IN NOX AND SOX EMISSIONS
FROM PETROLEUM INDUSTRIES
YEAR
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
Colorado
(tons/day)
SOX NOX
3
4
4
4
5
6
7
8
9
8
7
7
7
7
7
6
5
5
5
5
5
4
4
4
5
6
6
6
1
2
2
2
2
2
3
3
4
3
3
3
3
3
3
2
2
2
2
2
2
2
2
2
2
2
2
2
Utah
(tons/day)
SOX NOX
0
0
0
1
1
1
1
1
1
2
9
14
13
12
11
12
10
9
8
8
8
8
8
8
9
12
10
15
0
0
0
0
0
0
0
0
0
0
2
2
2
2
2
2
2
2
1
1
1
1
1
1
2
2
2
3
86
-------
1960 to 1962. This slight adjustment was again necessary to produce con-
sistency in trend estimates.
The natural gas data were taken from Minerals Yearbooks.
Results of adding together each fuel's contribution to industrial NOX
and SOX are listed in Table A-8.
Commercial and Residential
Emission factors for residential and commercial consumption of fuels
vary considerably depending on the sulfur content of the fuel burned and
the type of boiler used. The factors used here were chosen so that the
results in 1973 would be consistent with the emissions for each state re-
ported in NEDS.
Commercial and residential boilers were assumed to use coal that con-
tained 1.3% and 0.7% sulfur respectively. Minerals Yearbooks list the sulfur
contents of residual fuel oils for the years 1971-1975; these were averaged
to produce the 1.4% sulfur figure that was used here. Distillate fuel oil
was taken to be 0.2% sulfur since AP-42 reports it to be less than 0.3%.
Coal use data for the years 1960-1975 were taken directly from FEDS.
For earlier years, Minerals Yearbooks provided retail coal sales data for
the entire United States. These data were multiplied by a state/U.S. dis-
tribution factor obtained by averaging (over 1960 and 1961) the ratios of
statewide commercial and residential coal use reported in FEDS to the
United States total retail coal sales listed in the Minerals Yearbooks.
The resulting data were adjusted according to population growth for the
years 1948 to 1959.
FEDS provides fuel oil use data for the 1960-1975 years. Tables of
sales of range oil and heating oils, distillate and residual, found in
yearly "Mineral Industry Surveys" (BOM, 1949, 1951, 1953, 1955, 1957-1962)
were used to supply data for earlier years.
The natural gas use data were taken directly from Minerals Yearbooks.
The estimated emissions resulting from the residential and commercial
burning of coal, fuel oils, and natural gas were summed to produce the
results presented in Table A-9.
87
-------
TABLE A-8. TRENDS IN SOX AND NOX EMISSIONS
FROM INDUSTRIAL SOURCES
YEAR
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
Arizona
(tons/day)
SOX NOX
9
5
2
2
2
2
2
2
2
2
2
2
3
3
3
2
2
3
4
3
3
3
3
4
7
9
9
7
18
20
19
23
27
28
29
30
32
34
31
29
29
37
36
35
26
47
55
39
20
21
25
31
37
45
39
31
Colorado
(tons/day)
SOX NOX
50
45
49
53
50
54
44
53
54
53
48
46
50
52
53
46
55
53
51
47
54
50
52
44
54
56
59
63
65
61
69
72
67
73
68
79
82
77
81
80
85
81
88
92
95
104
98
83
90
86
88
94
87
98
93
91
Nevada
(tons/day)
SOX NOX
6
5
6
8
9
10
5
5
4
5
4
3
4
4
4
3
3
2
2
1
1
1
3
4
6
5
5
3
8
6
9
10
11
11
10
10
10
11
12
11
10
12
14
13
11
13
11
8
7
8
11
15
15
14
12
10
Utah
(tons/day)
SOX NOX
64
59
75
80
75
77
67
76
79
76
63
69
66
66
74
73
85
90
96
90
92
97
85
84
87
95
101
101
54
50
56
64
63
69
59
60
66
65
58
63
64
63
73
76
86
90
89
83
88
93
90
91
93
102
105
101
88
-------
TABLE A-9. TRENDS IN SOX AND NOX EMISSIONS FROM
COMMERCIAL/RESIDENTIAL SOURCES
YEAR
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
Arizona
(tons/day)
SOX NOX
1
1
1
1
1
2
1
1
1
0
1
1
1
1
1
1
1
1
1
0
1
1
1
2
2
1
2
2
4
4
4
4
4
4
3
4
4
4
6
8
8
6
6
5
8
7
7
6
8
8
9
11
11
13
14
17
Colorado
(tons/day)
SOX NOX
36
38
40
33
32
28
29
31
28
22
20
22
19
24
28
26
28
23
22
22
21
23
17
14
16
14
11
8
14
15
15
16
15
15
16
16
15
15
13
17
16
18
20
19
22
23
25
24
24
26
26
26
27
29
28
31
Nevada
(tons/day)
SOX NOX
4
5
5
4
4
4
4
3
4
3
3
3
4
5
4
4
4
6
6
3
3
2
6
4
3
2
2
1
2
2
2
2
2
2
1
1
1
1
1
2
2
2
2
2
2
3
3
2
3
3
4
5
4
4
4
5
Utah
(tons/day)
SOX NOX
33
36
26
32
31
29
26
28
27
21
18
16
19
26
30
27
22
20
18
18
17
20
17
24
18
20
27
19
11
12
10
12
11
10
10
11
11
10
9
8
10
12
14
14
13
12
11
12
11
13
12
15
13
14
15
13
89
-------
Evaporative Hydrocarbons
The amounts of hydrocarbons emitted by the evaporation of solvents
were estimated using the 1973 emissions reported in NEDS as a base. Sol-
vent evaporation in each state was then assumed to vary as the state's
population. Population figures were interpolated from data in Statistical
Abstract of the United States: 1975 (U.S. Bureas of the Census, 1975).
Estimates of emission from gasoline evaporation were also made using
the NEDS data as a base. These, however, were taken to vary as the total
amount of motor fuel consumed in each state varied according to Highway
Statistics: Summary to 1975 (DOT, 1975).
Evaporative emissions are all non-methane according to Trijonis and
Arledge (1976) so no correction was made to produce the non-methane numbers
appearing in Table A-10.
90
-------
TABLE A-10. TRENDS IN NMHC EMISSIONS FROM GASOLINE
EVAPORATION AND SOLVENT EVAPORATION
NMHC Evaporation (tons/day)
YEAR
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
Arizona
6
6
7
8
9
9
10
10
11
12
13
14
15
16
17
18
19
19
20
21
23
26
28
30
34
37
35
36
Gasoli
Colorado
10
10
11
12
13
14
14
15
16
17
17
18
19
19
20
21
21
22
23
24
26
27
30
32
35
37
35
37
ne
Nevada
2
2
2
3
3
3
4
4
4
4
4
5
5
5
6
7
7
7
8
8
9
9
10
11
12
12
12
13
Utah
5
5
5
6
6
7
7
7
8
8
9
9
9
10
10
10
11
11
12
12
13
14
15
16
17
18
18
18
Arizona
12
12
12
13
14
15
16
17
17
19
20
21
22
21
24
25
25
26
27
28
28
29
30
31
33
34
36
37
Solvents
Colorado
17
17
18
18
19
19
20
20
21
22
22
23
24
24
25
25
26
27
27
28
28
29
30
31
32
33
33
34
Nevada
1
1
1
1
2
2
2
2
2
2
2
2
3
3
3
3
4
4
4
4
4
4
4
4
5
5
5
5
Utah
8
9
9
9
9
10
10
10
10
11
11
11
11
12
12
12
12
13
13
13
13
13
14
14
14
15
15
15
91
-------
REFERENCES FOR APPENDIX A
AAR (Association of American Railways), "Statistics of Railways of Class I
in the United States: Calendar Years 1946 to 1956", Washington, D.C.,
August 1957.
AAR (Association of American Railways), "Statistics of Railways of Class I
in the United States: Years 1956 to 1966", Washington, D.C., August 1967.
AAR (Association of American Railways), "Statistics of Railways of Class I
in the United States: Years 1965 to 1975", Washington, D.C., January 1977.
BOM (Bureau of Mines), Minerals Yearbook, 1948-1975, U.S. Department of
the Interior, Washington, D.C.
BOM (Bureau of Mines), "Mineral Industry Surveys", Annual Reports: 1949,
1951, 1953, 1955, 1957-1962, U.S. Department of the Interior, Washington,
D.C.
DOE (Department of Energy), Steam-Electric Plant Air And Water Control Data,
Annual Reports: 1969-1973, Washington, D.C.
DOE (Department of Energy), Federal Energy Data System (FEDS) Statistical
Summary, DOE/EIA-0031/2, UC-13, February 1978.
DOT (Department of Transportation), Highway Statistics: Summary to 1965,
Federal Highway Administration, Bureau of Public Roads, Washington, D.C.,
March 1967.
DOT (Department of Transportation), National Personal Transportation Study:
Annual Miles of Automobile Travel, Report #2, Federal Highway Administration,
Washington, D.C., April 1972.
DOT (Department of Transportation), Highway Travel Forecasts, Washington,
D.C., November 1974.
DOT (Department of Transportation), Highway Statistics: Summary to 1975,
Federal Highway Administration, Washington, D.C.
EPA (Environmental Protection Agency), NEDS (1973 National Emissions Report),
Research Triangle Park, North Carolina, May 1976.
FPC (Federal Power Commission), Steam-Electric Plant Construction Cost and
Annual Production Expenses, Annual Reports: 1948-1975, Washington, D.C.
Heaney, R.J., Environemntal Affairs Division of the Utah Copper Division
of Kennecott Copper Corporation, Salt Lake City, Utah, 1978.
92
-------
Larson, N.I. and P.E. Billings, First Annual Report on Arizona Copper Smelter
Pollution Control Technology, Arizona Department of Health Services, Phoenix,
Arizona, 1977.
MVMA (Motor Vehicle Manufacturers Association), 1975 Automobile Facts and
Figures, Detroit, Michigan.
MVMA (Motor Vehicle Manufacturers Association), Motor Vehicle Facts and
Figures '77, Detroit, Michigan.
Perry, H. and J.A. Decarls, "The Search for Low Sulfur Coal", Mechanical
Engineering. Vol. 89, pp. 22-27, April 1967.
Trijonis, J. and K. Arledge, "Utility of Reactivity Criteria in Organic
Emission Control Strategies: Application to the L.A. Atmosphere", EPA-600/
3-76-091, August 1976.
U.S. Bureau of the Census, Statistical Abstract of the United States: 1975,
96th Edition, Washington, D.C., 1975.
U.S. Environmental Protection Agency, Compilation of Air Pollutant Emission
Factors, AP-42, 3rd Edition, Research Triangle Park, N.C., August 1977.
Weisenberg, I.J. and J.C. Serne, "Design and Operating Parameters for
Emission Control Studies: Kennecott, McGill, Copper Smelter", Prepared for
EPA Industrial Environmental Research Laboratory, EPA-600/2-76-036c,
February 1976.
Weisenberg, I.J. and J.C. Serne, "Design and Operating Parameters for Emission
Control Studies: Phelps Dodge, Morenci, Copper Smelter", Prepared for EPA
Industrial Environemntal Research Laboratory, EPA-600/2-76-036g, Feb. 1976.
Weisenberg, I.J. and J.C. Serne, "Design and Operating Parameters for Emission
Control Studies: Asarco, Hayden, Copper Smelter", Prepared for EPA Indus-
trial Environmental Research Laboratory, EPA-600/2-76-036J, February 1976.
Weisenberg, I.J. and J.C. Serne, "Design and Operating Parameters for Emission
Control Studies: Kennecott, Hayden, Copper Smelter", Prepared for EPA In-
dustrial Environmental Research Laboratory, EPA-600/2-76-036b, February 1976.
Williamson, S.J., Fundamentals of Air Pollution, Addison-Wesley, Reading,
Massachussetts, 1973.
93
-------
APPENDIX B
PRIMARY SULFATE EMISSIONS FROM COPPER SMELTERS
Over the past three decades, copper smelters have been the dominant
*
source of sulfur oxide emissions in the Southwest United States. That
the SOX emissions from copper smelters significantly affect sulfate and
visibility in the Southwest was demonstrated in an earlier TSC report
(Trijonis and Yuan 1978), which showed that the 9 month copper strike of
1967-1968 produced substantial reductions in sulfate and increases in
visibility on a large spatial scale. Lockwood and Hartman (1970), Trijonis
and Yuan (1978), and Latimer et al (1978) also demonstrated that historical
visibility trends in the Southwest closely followed trends in copper
smelter SOx emissions; visibility deteriorated as copper smelter SOx
emissions increased from the early 1950's to 1972, and visibility improved
from 1972 to 1976 as smelter SOX emissions were halved by air pollution
controls and reduced copper production.
There is concern that the recent improving trend in visibility may
be reversed in the future if SOx emissions increase greatly as a result of
new power plants and other energy developments in the Southwest. One
argument against this concern is that power plant SOx emissions may be less
important (per ton) with respect to sulfates and visibility than are
smelter SOx emissions. Specifically, power plant SOx might be less im-
portant than smelter SOx because power plant SOx might contain a much
smaller fraction of primary sulfate emissions. In order to resolve this
argument there is a need to compare primary sulfate emissions from power
plants with that from smelters.
There are numerous measurements available on the percentage of sulfur
emitted as sulfates (mostly $03) from power plants. For coal-fired power
plants various data indicate the following: 0.5-1.0% (Bailey and Ruddock
1978), 2.4% (Richards et al 1975), 1-2% (Nader 1978a; Cheney and Homolya 1978),
*
Here we define the Southwest as Arizona, Colorado, Nevada, New Mexico,
and Utah. In the late 1960's and early 1970's, copper smelters accounted
for more than 90% of total SOX emissions in these five states.
94
-------
and 0.7-1.8% (Howes 1978). For oil-fired power plants, data show 2-3% (Dietz
and Wieser 1978; O'Neal et al 1978), 2.7% (Richards et al 1975), up to 10%
(Nader 1978b; Homolya and Cheney 1978), and 3.4% (Howes 1978). The consen-
sus is that primary sulfate comprises approximately 0.5-2% of sulfur emissions
from coal fired power plants and 2-10% of sulfur emissions from oil fired
power plants (Wilson et al 1978).
There has been much less discussion in the literature concerning
primary sulfate emissions from smelters. In a search for primary sulfate
emission data for smelters, we contacted approximately 25 individuals in the
smelter industry, environmental research firms, universities, research labor-
atories, EPA, and other governmental agencies. This search uncovered the
data listed in Table B-l; these data include stack tests as well as plume
measurements taken a few kilometers downwind of smelters.
As indicated in Table B-l, almost all the stack test data indicate
that primary sulfate emissions are approximately 0.2-1.0% for both "green-
*
feed" and "roaster" smelters. The plume samples, at about 2-5 km downwind
of smelters, indicate about 1-2% primary sulfates from both green feed and
**
roaster smelters. The difference between the stack and plume measurements
is likely due to one of two factors. First, both types of data could be
correct, with the increased sulfate at 2-5 km representing secondary sulfate
formed by S02~"~S04 conversion after emission. This, however, would corres-
pond to a near-stack sulfate formation rate on the order of 5% per hour,
which is somewhat higher than the 1-3% per hour reported by Lusis and Wiebe
(1976) based on measurements between 5 and 100 km from the stack. Second,
the stack test data might represent an artificially low estimate of primary
*
We included notations of green-feed and roaster smelters in Table B-l be-
cause some individuals we contacted indicated that primary sulfate emissions
might be much higher for green-feed smelters than they are for roaster
smelters.
**
One other set of plume data (Kendall et al 1976) indicates much higher
levels of primary sulfates, 11% or greater. The measurement techniques
used for this data, however, are considered extremely unreliable (Kendall
1978). For example, data from these tests yield unbelievable sulfate
conversion rates of 50-100% per hour.
95
-------
TABLE B-l. DATA ON PERCENT PRIMARY SULFATES FROM SMELTERS
SMELTER
SMELTER
TYPE DAT.' SO'JRCE
HLASURfMENT DISTANCE OF PERCENT OF EMITTED
KETHODS FROI! STACK SAMPLES SULFIP THAT IS SULFATE
STACk TESTS
Phelps Dodge/ Ajo (AZ)
(Copper)
Maqsw.'San Manuel (AZ!
(Copper)
Ke"necott/Hayden (AZ)
(Copper)
Various Arizona Smelters
(Copper)
White Pine/wnite Pine (HI)
(Copper)
-------
sulfate because of measurement problems. Specifically, some of the primary
sulfate may be lost in the EPA Method 5 procedure that precedes the EPA
Method 8 procedure in many of the tests.
If S02 to sulfate conversion were the cause of the discrepancy be-
tween the stack tests and plume tests, we would conclude that primary sul-
fate comprises 0.2 to 1.0% of smelter SOx emissions (i.e. the stack samples
represent true primary sulfate). If the stack samples were in error, we
would conclude that primary sulfates represent about 1-2% of smelter SOx
emissions, possibly slightly less if we allow for some sulfate formation
between the time of emission and the time of the plume sample. In either
case, we conclude that the percentage of primary sulfate from copper
smelters is not significantly greater than the percentage of primary sulfate
from power plants primary sulfate from smelters is apparently about
the same, or significantly less, than primary sulfate from power plants.
97
-------
REFERENCES FOR APPENDIX B
Bailey, E.M. and H.A. Ruddock, "Measurements of Sulfur Trioxide at iVA
Coal-Fired Power Plants Using the Condenser Method," Proceedings of the
Workshop on Measurement Technology and Characterization of Primary Sulfur
Oxides Emission from Combustion Sources, April 24-26, 1978, Southern Pines,
NC, to be published as an EPA report.
Bowerman, L., EPA Region IX, Personal Communication, June 1978.
Cheney, J.L. and J.B. Homolya, "Characterization of Combustion Source
Sulfate Emissions Using a Selective Condensation Sampling System," Pro-
ceedings of the Workshop on Measurement Technology and Characterization
of Primary Sulfur Oxides Emission from Combustion Sources, April 24-26,
1978, Southern Pines, NC, to be published as an EPA report.
Dietz, R.N. and R.F. Wieser, "Operating Parameters Affecting Sulfate
Emissions from an Oil-Fired Power Unit," Proceedings of the Workshop on
Measurement Technology and Characterization of Primary Sulfur Oxides
Emission from Combustion Sources," April 24-26, 1978, Southern Pines,
NC, to be published as an EPA report.
Eatough, D., Brigham Young University, Personal Communication, June 1978.
Forrest, H. and L. Newman, "Oxidation of Sulfur Dioxide in the Sudbury
Smelter Plume," Atmospheric Environment, Vol. 11, 1977.
Forrest, J., Brookhaven National Laboratories, Personal Communication,
June 1978.
Homolya, J.B. and J.L. Cheyne, "An Assessment of Sulfuric Acid and Sulfate
Emissions from the Combustion of Fossil Fuels," Proceedings of the Workshop
on Measurement Technology and Characterization of Primary Sulfur Oxides
Emission from Combustion Sources, April 24-26, 1978, Southern Pines, NCS
to be published as an EPA report.
Howes, J., Battelle Memorial Laboratory, Personal Communication, June
1978.
Kendall, S.B., et al., "A Kinetics Study of the Sulfur Species in Smelter
Emissions," under NSF Grant GY - 10801, University of Arizona, 1976.
Kendall, S.B., Phelps Dodge Corporation, Personal Communication, June 1978.
Latimer, D.A., et al., "The Development of Mathematical Models for f;he Pro-
diction of Anthropogenic Visibility Impairment," Prepared at Systems Ap-
plications, Inc. for USEPA under Contract # 68-01-3947, 1978.
Lockwood, G.W. and W.K. Hartman, "Visibility Variations at Tucson, Arizona
and Kitt Peak National Observatory," Publications of the Astronomical
Society of the Pacific, Vol. 82, 1970.
Lusis, M.A. and H.A. Wiebe, "The Rate of Oxidation of Sulfur Dioxide in the
Plume of a Nickel Smelter Stack," Atmospheric Environment, Vol, 10, 1976.
98
-------
Mercer, C., Bureau of Air Quality, Arizona Department of Public Health,
Personal Communication, June 1978.
Nader, J.S. (a), "Field Measurements and Characterization of Emissions from
Coal-Fired Combustion Sources," Presented at the APCA 71st Annual Meeting,
June 25-30, 1978, Houston, TX.
Nader, J.S. (b), Chief SSERB/EMDC, Environmental Protection Agency, Personal
Communication, June 1978.
O'Neal, A.J., H. Cowherd, and F.W. Lipfert, "Use of a High-Flow Stack
Sampler for Determination of Particulate Sulfate Emissions," Proceedings
of the Workship on Measurement Technology and Characterization of Primary
Sulfur Oxides Emission from Combustion Sources, April 24-25, 1978, Southern
Pines, NC, to be published as an EPA report.
Richards, J., et al., "Stationary Source Control Aspects of the Ambient
Sulfates and Nitrates Issues," Prepared for the USEPA/ORD under Contract
# 68-02-1321, 1975.
Steiner, J., ACARAX Inc., Personal Communication, June 1978.
Trijonis, J. and K. Yuan, "Visibility in the Southwest: An Exploration of
the Historical Data Base," Prepared at Technology Service Corporation for
EPA Office of Planning and Evaluation and EPA Office of Research and De-
velopment under Grant # 802815, 1978.
Weisenberg, I.J. and J.C. Serne, "Design and Operating Parameters for
Emission Control Studies: White Pine Copper Smelter," Prepared for EPA
Industrial Research Laboratory, EPA-600/2-76-036a, February 1976.
Weisenberg, I. J. and J.C. Serne, "Design and Operating Parameters for
Emission Control Studies: Kennecott, McGill, Copper Smelter, Prepared for
EPA Industrial Environmental Research Laboratory, EPA-600/2-76-036c,
February 1976.
Weisenberg, I.J. and J.C. Serne, "Design and Operating Parameters for
Emission Control Studies: Kennecott, Hurley, Copper Smelter," Prepared for
EPA Industrial Environmental Research Laboratory, EPA-600/2-76-036d,
February 1976.
Weisenberg, I.J. and J.C. Serne, "Design and Operating Parameters for
Emission Control Studies: Phelps Dodge, Ajo Copper Smelter," Prepared for
EPA Industrial Environmental Research Laboratory, EPA-600/2-76-036f,
February 1976.
Wilson, W.E., et al., "Dependence of Ambient Sulfates on Precursor Control:
Chemical Aspects," Presented at the AIChE Meeting, June 4-8, 1978, Philadel-
phia, PA.
99
-------
APPENDIX C
ESTIMATING AQCR EMISSION TRENDS FROM STATE EMISSION TRENDS
Chapter 2 of this report presents historical emission trends from
1948 to 1975 in four Southwestern states. For some of the analyses per-
formed in Chapter 5, it is also necessary to know emission trends for in-
dividual Air Quality Control Regions (AQCR's). This appendix explains how
AQCR emission trends for Phoenix/Tucson, Denver, and Salt Lake City can be
estimated from state emission trends for Arizona, Colorado, and Utah
respectively.
There are essentially three factors which determine total emission
trends for a given pollutant in a certain region: (1) the relative contri-
butions (for some base year) of each source category to total emissions,
(2) changes in emission factors (from air pollution controls, fuel switches,
etc.) for individual source categories, and (3) growth or attrition rates
for individual source categories. It appears to be a fairly good approx-
imation to make the fairly crude assumption that, for each of the three
AQCR's, these three factors have operated in a way that kept AQCR emissions
at a nearly constant percentage of state emissions. With respect to the
first factor, Tables C-l to C-3 compare the distribution of emissions by
source type in 1973. It is evident that for each pollutant, the distri-
bution of emissions by source type is nearly the same in each AQCR as it
is in the corresponding state. With respect to the second factor, we can
expect that changes in emission factors would tend to be rather uniform
within each state. This is certainly true of the changes produced by auto-
motive controls, and it should be approximately true with respect to fuel
changes (e.g. the switch from coal to diesel oil in railroading, the switch
to cleaner fuels for residential and commercial sources, etc.). With res-
pect to the third factor, we have no data comparing individual source growth
rates in each AQCR with corresponding statewide source growth rates. How-
ever, population growth should be a fairly good indicator of overall source
growth rates in AQCR's versus states. Comparative population statistics for
metropolitan areas and states are readily available only for 1960-1973 (USBOC,
1975); these data indicate that the three metropolitan areas have grown
100
-------
TABLE C-l. COMPARISON OF SOURCE DISTRIBUTIONS FOR
PHOENIX/TUCSON AQCR EMISSIONS AND TOTAL
ARIZONA EMISSIONS (EPA, 1976)
Total Emissions
(tons/day)
SOX
Phoenix/
Tucson
1960
Arizona
5620
NOX
Phoenix/
Tucson
329
Arizona
463
NMHC
Phoenix/
Tucson
479
Arizona
630
DISTRIBUTION BY SOURCE CATEGORY (PER CENT OF TOTAL)
Cu Smelters
Gasoline
Vehicles
Diesel
Vehicles
Railroads
Power
Plants
Industrial
Commercial &
Residential
Gasoline
Evaporation
Solvent
Evaporation
Other
98
0.3
0.2
0.3
0.1
0.1
1.1
99
0.1
0.1
0.3
0.1
0.1
0.3
51
12
7
19
2
3
6
49
15
6
17
3
3
7
68
5
6
20
66
6
5
23
101
-------
TABLE C-2. COMPARISON OF SOURCE DISTRIBUTIONS FOR
DENVER AQCR EMISSIONS AND TOTAL COLORADO
EMISSIONS (EPA, 1976)
Total Emissions
(tons/day)
S(
Denver
69
)x
Colorado
131
Nl
Denver
271
)x
Colorado
508
NMf
Denver
314
1C
Colorado
602
DISTRIBUTION BY SOURCE CATEGORY (PER CENT OF TOTAL)
Gasoline
Vehicles
Diesel
Vehicles
Railroads
Power
Plants
Industrial
Commercial &
Residential
Petroleum
Industry
Gasoline
Evaporation
Solvent
Evaporation
Other
4
46
24
6
8
12
4
45
20
11
4
16
46
11
7
18
9
6
1
2
49
11
6
17
9
6
0.4
2
75
6
7
11
76
6
5
12
102
-------
TABLE C-3. COMPARISON OF SOURCE DISTRIBUTIONS FOR SALT
LAKE CITY AQCR EMISSIONS AND TOTAL UTAH
EMISSIONS (EPA, 1976)
Total Emissions
(tons/day)
SOX
Salt
Lake
City
461
Utah
501
NOX
Salt
Lake
City
194
Utah
295
NMHC
Salt
Lake
City.
166
Utah
270
DISTRIBUTION BY SOURCE CATEGORY (PER CENT OF TOTAL)
Cu Smelters
Gasoline Vehicles
Diesel Vehicles
Railroads
Power Plants
Industrial
Commercial &
Residential
Petroleum
Industry
Gasoline
Evaporation
Solvent
Evaporation
Other
49
1
5
15
3
10
18
45
1
7
15
4
9
19
___
34
9
8
9
22
7
7
4
38
12
11
10
16
5
4
4
61
6
7
25
63
6
5
26
103
-------
slightly more rapidly than their respective states -- Phoenix/Tucson (4.0%
per year) versus Arizona (3.4% per year), Denver (3.0% per year) versus
Colorado (2.6% per year), and Salt Lake City (2.1% per year) versus Utah
(1.9% per year). Rudimentary sensitivity analyses indicate that these
relative differences in growth rates produce effects that are small com-
pared to the overall effect of source growth on emissions.
Based on the above discussion, we have decided to assume that emission
trends in each of the three AQCR's are approximately a constant factor times
emission trends in the corresponding state. The factors for calculating
AQCR emissions are listed in Table C-4. Because of the crude nature of the
assumed constant relationship between AQCR emissions and state emissions,
we are not as confident of the AQCR emission trends as we are of the state
emission trends. However, the constant factors in Table C-4 should be use-
ful in allowing the extinction/emission coefficients to represent air basin
emissions (rather than state emissions) in our analyses which center on
visibility trends within metropolitan areas.
TABLE C-4. FACTORS FOR DETERMINING AQCR EMISSION
TRENDS FROM STATE EMISSION TRENDS
AQCR/STATE
Phoenix-Tucson/Arizona
Denver/Colorado
Salt Lake City/Utah
FACTOR
SOX
0.35
0.53
0.92
FOR COMPUTING AQCR
NOX
0.71
0.53
0.66
EMISSIONS
NMHC
0.76
0.52
0.61
104
-------
REFERENCES FOR APPENDIX C
EPA (U.S. Environmental Protection Agency), 1973 National Emissions Report,
National Emissions Data System (NEDS) of the Aerometric and Emissions Re-
porting System (AEROS), EPA-450/2-76-007, May 1976.
USBOC (U.S. Bureau of the Census), Statistical Abstract of the United
States 1975, U.S. Department of Commerce, Washington, D.C., 1975.
105
-------
APPENDIX D
STATISTICAL TESTS FOR DETERMINING THE
SIGNIFICANCE OF RELOCATIONS IN VISI-
BILITY OBSERVATION SITES
This appendix describes the two statsitical tests for examining the
effect of observation site relocations on reported visibilities. Both tests
are based on quarterly median visibilities, and both use approximately four
years of data (two before and two after the time of the relocation). The
analysis is restricted to only two years on either side of the relocation
in order to help minimize the possibility that long-term trends will be
confused with the jump in visibility caused by the relocation. Quarterly
values are used because they are readily computed by our data processing
programs and because yearly values would yield too few data points.
Test 1. Net Jump in Quarterly Medians
The first test is best illustrated by example. Assume a relocation
occurred during the 3rd quarter of 1965. Then the data points used are as
follows:
1963 4th quarter median
1st quarter median ------------ x~
1964 2nd quarter median ------------ x^
3rd quarter median ------------ xğ
4th quarter median ------------ x5
1st quarter median ------------ xg
1965 2nd quarter median _____-_----- \
3rd quarter median relocation
4th quarter median ------------ y
1st quarter median ------------ y
1966 2nd quarter median ------------ y
3rd quarter median ------------ y^
4th quarter median ------------ y
1st quarter median ------------ y
1967 2nd quarter median ------------ y
106
-------
Seven quarterly changes in visibility are computed as z^ = y. -x., i = 1, ..
., 7. The estimated net jump in visibility is simply the average value (z..),
and the t-statistic for the jump is zT/(az//7~), where az is the standard
deviation of the z..
Test 2. Multivariate Regression Test on Seasonally Adjusted Quarterly
Medians
The first test has the advantage that it is simple and automatically
discounts for seasonal trends in the data. The major disadvantage is that
it could easily confuse a gradual trend in the data with a jump produced by
the relocation. To account for this latter possibility, we also conducted
a simple multiple regression test.
The data for the regression test consist of two years of seasonally
adjusted quarterly medians before the relocation (z , z_7, ..., z_,) and
two years of seasonally adjusted quarterly medians after the relocation (z,,
Zp, ...ğ Zn). A multiple regression is then run with these 16 data points
as follows:
zi = a + bi + c H(i),
where H(i) is the Heavyside step function (zero for i _ 0, one for i > 0).
The coefficient "c" now represents our estimate of the net jump produced by
the relocation (discounting for the net linear trend which is represented by
"b"). The t-statistic is "c" divided by its standard deviation.
The regression test is explicitly designed to discount for the gradual
trend in the data. However, it suffers from the possibility that distortions
could be produced by the intercorrelation (R = 0.89) between the "independent"
variables, "i" and "H(i)".
*
The seasonal adjustment factors are computed as follows:
f _ z for all 16 values
nth quarter ,, ,, . ., . ,
^ z for the 4 nth quarter values
107
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1 REPORT NO.
EPA-450/5-79-009
3. RECIPIENT'S ACCESSION-NO.
4. TITLE AND SUBTITLE
Empirical Studies of the Relationship Between Emissions
and Visibility in the Southwest
5. REPORT DATE
September 1979
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
Marilyn Marians and John Trijonis
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Technology Service Corporation
Route 7, Box 124-K
Santa Fe, New Mexico 87501
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
Grant 802015
12. SPONSORING AGENCY NAME AND ADDRESS
U.S. Environmental Protection Agency
Research Triangle Park, N.C. 27711 *
13. TYPE OF REPORT AND PERIOD COVERED
Interim May 1978-March 1979
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
16. ABSTRACT
Historical emission trends of SOX, NOx, and NMHC are determined year-by-year
from 1948 to 1975 for four Southwestern states (Arizona, Colorado, Nevada, and
Utah). Trends in visibility levels (medians and other percentiles) are docu-
mented for the period 1948-1976 at 12 airports in the Southwest.
Two analyses are used to relate emission changes to variations in regional
extinction levels. The first analysis examines the air quality changes
associated with a 90% reduction of Southwestern SOx emissions during a nine-
month copper strike and estimates the extinction produced by SOx emissions on
various spatial scales. The second analysis involves regression studies relating
historical extinction levels from 1948-1975. Because of limitations in the
analytical methods, there is a high degree of uncertainty in many of the results.
However, the studies do provide insights into the effects of aerosol precursor
emissions on extinction at various distances from sources. In the case of
mesoscale effects of SOx in the Southwest, quantitative coefficients are proposed
which link emissions to regional extinction.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
Air pollution
Aerosols
Sulfates
Visibility
Sulfur oxides emissions
Southwest
18. DISTRIBUTION STATEMENT
Release to Public
19. SECURITY CLASS (ThisReport)
Unclassified
21. NO. OF PAGES
20. SECURITY CLASS (Thispage)
22. PRICE
EPA Form 2220-1 (9-73)
108
------- |