v"'f.
SEPA
United States
Environmental Protection
Agency
Environmental Sciences Research
Laboratory
Research Triangle Park NC 27711
Research and Development
EPA-600/S3-82-015 May 1982
Project Summary
Validation of the
EKMA Model Using
Historical Air Quality Data
John Trijonis and Stan Mortimer
Historical air quality data were used
to study the validity of the Empirical
Kinetic Modeling Approach (EKMA)
model for relating ozone concentra-
tions to precursor control, as a method
for evaluating ozone control strate-
gies. Using both emission and ambient
air pollution data from the Los Angeles
Basin for 1964 to 1978, trends in the
ozone precursors nonmethane hydro-
carbons (NMHC) and nitrogen oxides
(NO,) were estimated. The estimated
trends were entered into the EKMA
model to predict historical ozone
trends; the results were then compared
with actual ozone trends documented
from measurements to test the EKMA
model. Emission trend calculations
showed continual decrease (net reduc-
tion of 29%) in basinwide hydrocarbon
emissions during the period studied.
Basinwide NOX emissions rose rapidly
from 1965 to 1973 and then leveled
off, showing a net increase of 34%
over the entire period studied. Ambi-
ent ozone precursor trends agreed
well with the emission trend estimates.
Sensitivity analyses indicated that
predicted ozone trends were moder-
ately sensitive to the specific EKMA
simulation conditions and were extreme-
ly sensitive to the NMHC/NO, ratio.
The EKMA validation studies compar-
ing predicted with actual ozone trends
showed that the two trends agreed
within the error bound designated.
However, predicted ozone trends gen-
erally underestimated decreases in
actual ozone trends. Errors in the
estimated precursor trends, or in the
meteorological fluctuations in the
actual ozone trends, or in the sensi-
tivity to varied EKMA simulation con-
ditions did not account for this discrep-
ancy; it apparently resulted from
error in the NMHC/NO. ratio. The
study indicates the importance of
accurately estimating the NMHC/NO,
ratio, as well as using the most realistic
area-specific EKMA simulation
conditions (e.g., inclusion of post-8
a.m. emissions). Further research is
recommended to test the EKMA chem-
ical-kinetic mechanism and to check
the equivalency of the ambient and the
EKMA NMHC/NO, ratios.
This Project Summary was develop-
ed by EPA's Environmental Sciences
Research Laboratory. Research Trian-
gle Park, NC, to announce key findings
of the research project that is fully
documented in a separate report of the
same title (see Project Report ordering
information at back).
Introduction
The Empirical Kinetic Modeling Ap-
proach (EKMA) has recently been pro-
posed by the U.S. Environmental Protection
Agency (EPA) as a method for evalua-
ting ozone control strategies. To be
accepted as a reliable technique for
control strategy analysis, it should be
subjected to validation studies. This
study tested the EKMA method using
-------
historical-trend data for the Los Angeles
area. Specifically, trends in the ozone
precursors nonmethanehydrocarbons(NMHC)
and nitrogen oxides (NO,) were esti-
mated fro en emissions and ambient air
pollution data for the period 1964 to
1978. These precursor trends were
entered into the EKMA model to predict
historical ozone trends; the predicted
ozone trends were then compared to
actual ozone trends to test the EKMA
method.
This study showed that analysis of
historical-trend data from Los Angeles
is an appropriate and useful way of
validating the EKMA method. Historical
trend analysts also provides a very
convenient format for illustrating the
sensitivity of EKMA to various inputs,
such as the NMHC/NO, ratio and the
choice of specific photochemical-simu-
lation conditions. However, the histor-
ical-trend data also have intrinsic inter-
est. By analyzing precursor emission
trends in detail, emissions for individual
source categories were shown to have
changed due to controls and source
growth, and how total emissions have
changed as a net response to trends for
individual source categories. By exam-
ining trends in ambient precursor data,
emission trend estimates were checked
and it was verified that control programs
and source growth had the anticipated
effects. By analyzing historical ozone
trends, investigations showed whether
the ozone trends relate rationally to
precursor changes and meteorological
fluctuations. Thus, the study was useful
for validating EKMA and for understand-
ing the air quality effects actually being
produced by on-going control programs
in Los Angeles.
Based on historical-trend data the Los
Angeles region is uniquely excellent for
EKMA validation studies. Only Los
Angeles can provide nearly two decades
of high-quality, spatially-resolved, long-
term trend data for ambient concentra-
tions of ozone, hydrocarbons, and NOX.
Although independent data sets regard-
ing the NMHC/NO, ratio are available
from several monitoring programs, the
quality of the Los Angeles data for
emission-trend analysis is unexcelled.
The area has unusually good informa-
tion regarding existing emission levels,
source growth rates, and source control
levels. Furthermore, substantial
changes in historical precursor levels
are required to adequately test EKMA,
knowing that Los Angeles has under-
gone significant decreases in hydrocar-
bons and increases in NO, since the
middle 1960s. Finally, the severe photo-
chemical smog problem in Los Angeles
and the current controversy over the
recent lack of air'quality improvement
enhance the interest of a study in that
area.
The EKMA validation studies con-
ducted.herein examined the time period
1964 to 1978. To provide robust (statis-
tically stable) data sets for the analysis,
3-year averages (1964-1966, 1967-
1969, 1970-1972, 1973-1975, and
1976-1978) of air quality data were
used, with 1964-1966 serving as the
base period. To test the EKMA model,
predicted ozone trends were compared
to actual ozone trends for each sub-
sequent 3-year interval.
This study uses four basic data sets:
information of the 6 a.m. to 9 a.m.
ambient NMHC/NO. ratio, estimates of
historical emission trends, ambient pre-
cursor trend data, and ozone trend data.
The base period (1964-1966) 6 a.m. to 9
a.m. NMHC/NOx ratio was determined
by examining ambient data for several
early seventies monitoring programs
and by extrapolating the results back to
1964-1966 based on historical precur-
sor trends. Historical trends in NMHC
and NOX emissions were documented
through a new and comprehensive
analysis of all major source categories.
Ambient precursor trends were exam-
ined at 9 sites for NMHC, 10 sites for
NOx, and ambient ozone trend data from
13 sites. The ambient data for NMHC,
and NO,, and ozone were subjected to
several correction and adjustment pro-
cedures (discussed herein) to ensure
consistent data suitable for trend anal-
yses.
The basinwide EKMA isopleth model
(based on the maximum ozone during a
10-hour irradiation) was tested against
historical trends for the yearly basin-
wide 1 -hour ozone maximum. To provide
greater generality in validating the
isopleth approach and to increase the
number of test cases, four individual
sites for analysis were chosen: Azusa,
Downtown Los Angeles (DOLA), Ana-
heim, and San Bernardino. The valida-
tion studies at these four sites were
conducted using EKMA isopleths spe-
cific to the time of maximal ozone
occurrence at the locations. Two ozone
air quality indices, yearly maximum 1-
hour concentration and 95th percentile
of daily maximum concentrations, were
used at each of the four sites.
Conclusions
The objectives of this study were (1) to
characterize historical precursor trends
in the Los Angeles region using both
emissions data and ambient data; (2) to
document historical ozone trends; (3) to
investigate the sensitivity of EKMA
predictions to the NMHC/NOX ratio and
to the choice of specific simulation
conditions; and (4) to test the EKMA
isopleth model by comparing predictions
of historical ozone trends with actual
ozone trends. The following subsections
summarize our findings and conclusions
with respect to each of the above four
objectives.
Historical Precursor Trends
Estimated basinwide emissions of
hydrocarbons decreased continually dur-
ing the study period, with a net reduction
of 29% from 1965 to 1977. This reduc-
tion was predominantly due to de-
creases in emissions from light-duty
vehicles (the largest source category).
Light-duty vehicle hydrocarbon emis-
sions decreased 40% from 1965 to
1977 despite a 54% increase in traffic
levels. Organic solvent emissions also
significantly (30%) decreased, with this
reduction basically occurring between
1965 and 1974.
Estimated basinwide NOX emissions
rose rapidly from 1965 to 1973 and then
essentially leveled off. The net increase
over the entire study period, 1965 to
1977, was 34%. This rise was predom-
inantly due to a 55% increase in NO,
emissions from light-duty vehicles,
again the largest source category. The
net increase in NO, from light-duty ve-
hicles from 1965 to 1977 basically
represented traffic growth. Light-duty
vehicle NOX emission factors for new
cars increased sharply in the late 1960s,
but by 1977 the fleet-averaged NO,
emission factor was reduced to the
1965 level due to the new car NO,
emission standards of the 1970s. NO,
emissions from heavy-duty vehicles and
from residential, commercial, and indus-
trial fuel burning also increased signif-
icantly from 1965 to 1977, reflecting
growth in traffic and natural gas usage,
respectively. Power plant NO, emissions
decreased slightly from 1965 to 1977.
Emission trends differ among the
various individual source areas affecting
the four ozone study sites, reflecting
differences in growth rates and source
types within those areas. Comparing
estimated emission trends with ambient
-------
precursor trends for each source area,
we found very good agreement, espe-
cially when viewing the data overall
from 1965 to 1977. In light of this
agreement, we can state with a high
degree of confidence that there was a
moderate (15 to 30%) red uction i n hydro-
carbons and a moderate (25 to 35%)
increase in NO, within the Los Angeles
Basin from 1965 to 1977.
Historical Ozone Trends
For the EKMA validation studies,
using an air quality index that repre-
sents high ozone days but has a low
year-to-year meteorological variance is
desirable. Unfortunately, a trade-off
exists in that the indices which repre-
sent the highest ozone days tend to be
the least robust indices. A satisfactory
compromise is the index "95th percen-
tile of daily maximal 1-hour ozone,"
which was used for the four study sites.
To represent worst-case conditions, the
index "yearly maximum 1-hour ozone"
was used at the four study sites and for a
basinwide analysis.
Various adjustments and corrections
were applied to the historical ozone and
oxidant data base in the Los Angeles
area to ensure consistent data for trend
analysis. Fortunately, a single monitor-
ing technique was used during the entire
period studied at all but one of the four
study sites. Thus, ozone trends at these
study sites required relatively few ad-
justments and corrections.
Ozone concentrations at Azusa,
DO LA, and Anaheim decreased from
the middle 1960s to the early to middle
1970s and then leveled off. All sites in
Los Angeles exhibited this historical
pattern in ozone trends, except those in
the extreme eastern parts of the basin
where, as at San Bernardino, ozone
increased at a very slight rate during the
entire study period.
Selection of Specific
Conditions for EKMA Validation
One critical input to the EKMA iso-
pleth model is the base period ambient 6
a.m.toSa.m. NMHC/NO* ratio, and the
best estimate for the 1964-1966 base
period in Los Angeles was 13:1. This
estimate was derived by examining data
from several monitoring programs dur-
ing the early 1 970s, a nd by extrapolati ng
back to 1964-1966 based on historical
trends in ambient precursor levels. To
make both atmospheric and EKMA
NMHC concentrations more nearly equiv-
alent, the ambient NMHC data were
adjusted to correspond to measurements
made with propane calibration. The
contributions from the nearly nonreac-
tive compounds ethane and propane
were also subtracted from the ambient
NMHC data. Because of discrepancies
concerning the NMHC/NOX ratio among
various monitoring programs, there was
considerable potential for error in our
estimate of the NMHC/NO, ratio. In
sensitivity studies, we considered a
range of the ratio from 7:1 to 25:1.
The EKMA simulation conditions
chosen for this study were standard
except for including post-8 a.m. emis-
sions and using a special dilution pat-
tern. For the studies at individual moni-
toring sites, isopleths that represented
ozone at fixed irradiation times rather
than maximum ozone over the entire
irradiation period were also useful. A
sensitivity analysis using six types of
EKMA isopleths indicated that predicted
ozone trends were moderately sensitive
to the specific simulation conditions. In
particular, adding post-8 a.m. emissions
to the model significantly improved the
agreement between predicted and
actual ozone trends in Los Angeles.
For the specific EKMA simulation
conditions that were used, and for the
type of historical precursor changes that
were examined (decreasing NMHCwith
increasing NO,), predicted ozone trends
were extremely sensitive to the base
period NMHC/NO, ratio. Potential errors
in the NMHC/NOx ratio alone obviously
could account for any discrepancy be-
tween predicted and actual ozone
trends. Bearing this conclusion in mind,
the EKMA validation studies were stud-
ied under stringent test conditions which
did not allow for potential errors in the
NMHC/NOx ratio.
Validation of the EKMA
Isopleth Method
In the validation studies, historical
precursor trends of either emission or
ambient precursor trends and the EKMA
isopleth model were used to predict
historical ozone trends. These predicted
ozone trends were then compared to
actual ozone trends. Error bars repre-
senting the error in historical precursor
trends were included on the predicted
ozone trends; error bars representing
meteorological fluctuations were in-
cluded on the actual ozone trends. The
validation studies were conducted for
basinwide maximum ozone as well as
for two ozone air quality indices (yearly
maximum 1-hour concentration and
95th percentile of daily maximum con-
centrations) at the four individual study
sites.
The results of the EKMA validation
studies were nearly the same whether
one uses emission trends or ambient
precursor trends to derive predicted
ozone trends. In both cases, the vali-
dation study results were not poor
considering the stringent test conditions
which did not include potential errors in
the NMHC/NO, ratio. For most sites and
years, the predicted and actual ozone
trends agree within the error bars
representing both variance in precursor
trends and meteorological fluctuations
in ozone trends. However, predicted
ozone trends generally underestimated
decreases in actual ozone trends at all
sites except San Bernardino, and the
discrepancy was rather large for certain
sites and years.
Three possible explanations for the
observed discrepancy that predicted
ozone levels which tended to under-
estimate historical improvements in
actual ozone levels were rejected. The
cause was not errors in the historical
precursor trends, because the precursor
trends error bars were small, and be-
cause the precursor trends were con-
firmed by two independent data bases
(emissions data and ambient data).
Studies of met-adjusted ozone (i.e., data
adjusted for influences of varying mete-
orology) strongly suggest that the dis-
crepancy was also not du,e to meteoro-
logical fluctuations in actual ozone
trends. Also, further refining EKMA
simulation conditions (improved dilution
patterns, improved estimates of post-8
a.m. emissions, addition of carry-over
ozone) will probably not eliminate the
discrepancy because our predicted
ozone trends were not sensitive enough
to changes in the simulation conditions.
The obvious factor that can account
for observed discrepancies between
predicted and actual ozone trends is
error in the NMHC/NOX ratio. Using a
10:1 ratio rather than a 13:1 ratio
eliminates all of the discrepancies.
There are two ways to account for error
in the NMHC/NO, ratio. First, we could
have a random error in our estimate of
the ratio, which would not necessarily
recur when applying EKMA to other
areas. Second, there could be a system-
atic error in the EKMA chemical-kinetic
mechanism or in matching sthe atmos-
pheric NMHC/NOx ratio with the EKMA
-------
NMHC/NO, ratios, which might improve
the EKMA mechanisms or the guide-
lines for adjusting atmospheric NMHC/
NO, ratios before use in EKMA.
This study has pointed out two poten-
tially significant sensitivities of the
EKMA isopleth method. The predictions
of EKMA can be moderately sensitive to
the NMHC/NO, ratio, and thus, in some
applications, it may be important to
ensure that the most realistic, area-
specific EKMA simulation conditions are
used. Also, accurately estimating the
EKMA-equivalent NMHC/NO, ratio is
important. Predictions of future ozone
trends will likely be less sensitive to
both the simulation conditions and to
the NMHC/NO, ratio, because future
emission reductions are likely in both
NMHC and NO,, and because the predic-
tions of EKMA are much less sensitive
to the simulation conditions and the
NMHC/NO, ratio when both precursors
undergo similar changes.
As a postscript to this study, the
controversial lack of improvement in
Los Angeles ozone air quality since the
middle 1970s was studied. Several
possible explanations for this apparent
lack of improvement were considered:
source growth, failure of catalytic con-
verters, meteorological fluctuations, re-
versal in the historical trend of a decreas-
ing NMHC/NO, ratio, and overall failure
of control programs. Source growth and
catalyst failures were not reasonable
explanations considering available data.
Other potential explanations seemed
reasonable to a degree, but each also
appeared to contradict certain observa-
tions. We recommend taking new data
and conducting further analyses to
resolve this controversy.
John Trijonis and Stan Mortimer are with the Santa Fe Research Corporation,
Santa Fe. NM 87501.
Basil Dimitriades is the EPA Project Officer (see below).
The complete report, entitled "Validation of the EKMA Model Using Historical A ir
Quality Data," (Order No. PB82-197187; Cost: $13.50, subject to change) will
be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Environmental Sciences Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
if U S GOVERNMENT PRINTING OFFICE. 1982 - 559-017/0727
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
Postage and
Fees Paid
Environmental
Protection
Agency
EPA 335
Official Business
Penalty for Private Use $300
CSS ENVIR?PROTECTION AGENCY
REGION 5 LIBRARY
230 S DEARBORN STREET
CHICAGO IL 60604
------- |