Guidance for Improving Weight of Evidence
Through Identification of Additional Emission Reductions,
Not Modeled
by
U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards
Emissions, Monitoring, and Analysis Division
Air Quality Modeling Group
Research Triangle Park, NC 27711
November 1999
Introduction
This paper provides guidance for using information from photochemical grid modeling and
ambient air quality monitoring to estimate additional levels of emission reductions needed to support
the 1-hour NAAQS for ozone beyond the reductions contained in the demonstrations submitted by
the States in 1998. Procedures for estimating improvements expected with the implementation of
the Tier 2 low sulfur program and benefits towards attainment are also provided. Two techniques
are described for estimating additional emission reductions, each with its own strengths and
weaknesses. Use is made of the fact that, since 1999 is more than half-way from the model base year
(1990) to the attainment year (2005 or 2007 in most areas), air quality data from 1990 to 1999 allows
modelers the opportunity to determine the representativeness of the modeled predictions. These,
techniques identify the additional percentage reduction in NOx and VOC
from the emissions.
General Procedures for Improving Weight of Evidence Through Identification of
Additional Emission Reductions, Not Modeled, Including Tier 2.
To strengthen the weight of evidence and account for high modeled peaks, estimate
additional measures that at a minimum bring the model estimated future design value to 124 ppb
or below. This is done by first estimating a future design value using the model predicted peaks.
Multiply the base design value by a ratio (average of model predicted peaks (across all days),
after controls divided by before controls). The base design value is an average of three years of
monitored design values that represent the modeled base case emissions. If the model estimated
future design value is at or below 124 ppb, additional emission reductions should not be needed.
If the model estimated future design value is greater than 124 ppb, estimate additional
measures by using two ratios 1) modeled change in design values to modeled change in
1
-------
emissions and 2) air quality design value changes to NET/local emissions changes between two
reference years (e.g., 1990 and 1996). Do not include biogenic emissions. First, subtract 124
ppb from estimated future design value to identify additional ozone reduction needed. Then
multiply each ratio by the ozone reduction needed to estimate additional VOC and NOx emission
reductions needed to strengthen the weight of evidence argument for attainment. This results in
the additional percent reduction needed from the 1990 emissions.
To calculate the level of emission reductions needed (in tons per day) multiple the 1990
base emissions by the percent reductions. This results in the total tons per day reduction which
are additional reductions needed in the attainment year. To incorporate the impact of Tier 2
subtract the emission reduction estimates being applied towards attainment for Tier 2 from the
additional reductions. The remaining reductions may be adjusted to reflect other unmodeled
control measures which have been quantified. The following are more details of the procedures
with examples.
Estimating Additional Emission Reductions
Each of the methods described in the remainder of this paper begins with a monitored
ozone concentration which can be extrapolated to the attainment year and compared with the
standard. If the attainment year concentration is over 124 ppb, the methods described in this
paper can be used to estimate what would constitute additional emission reductions needed to
support a weight of evidence argument for attainment. The differences among the methods lie
in the factors used for this extrapolation. These are summarized in Table 1.
Both methods are based on the assumption that we can estimate the relationship between
ozone and its precursors (VOC and NOx). We can estimate this relationship by either (1)
comparing changes in model predicted ozone to changes in modeled emissions or (2) comparing
changes in observed air quality to changes in actual emissions. Both methods for estimating a
relationship are equally valid. Both have inherent uncertainty in estimates of emissions
inventories and estimates of the change in ozone air quality. Utility of either method is
dependent on the availability of data which shows a response in ozone due to a decrease in VOC
and NOx emissions. For example, if an area wants to apply method 2 using the NET inventories
for the 1990 and 1996 reference years, the VOC and NOx totals for the nonattainment area must
show a decrease in VOC and NOx between 1990 and 1996. If this is not the case then use of the
NET data for those two reference years in not appropriate.
2
-------
Table 1. Summary of Methods for Estimating Additional Emission Reductions
Method
Ozone Concentration Being
Kxtrapolated
Kxtrapolation Uatio
(normalized reduction factor)
1
Future Air Quality Design Value
Change in modeled emissions
From base to attainment vear
Change in modeled concentration
2
Future Air Quality Design Value
Change in actual emissions
Between two reference vears
Change in monitored concentration
Estimate a Future Air Quality Design Value
Both methods make use of the results of past modeling to derive a modeled response of
ozone design values to VOC and NOx controls to estimate a future air quality design value.
Relative reduction factors are derived and used similarly to what is described in U.S. EPA,
(1999), Draft Guidance on the Use of Models and Other Analyses In Attainment Demonstrations
for the 8-Hour Ozone NAAQS, EPA-454/R-99-004. If the estimated future design value is < 124
ppb, no additional emission reductions are needed to strengthen the weight of evidence argument
for attainment.
(1) Calculate an average (over all modeled days) predicted daily maximum (within the
nonattainment area and the down wind plume, or domain wide) 1-hour ozone concentration, first
with the base emissions (e.g., 1990) and then with the future emissions (e.g., 2007).
(2) Using results from step 1, calculate the relative reduction factor, RRF, by taking the ratio of
the average daily maximum 1-hour ozone concentration obtained with future emissions to that
obtained with the base emissions.
RRF = AVGf / AVGc
(1)
where
AVGf = average (across all days) predicted daily maximum 1-hour ozone concentration
for future emissions, ppb.
AVGc = average (across all days) predicted daily maximum 1-hour ozone concentration
for base emissions, ppb.
3
-------
(3) Calculate the base design value, DVB, as the average of 3 nonattainment area ozone design
values that represent the period used to predict ozone for base emissions (e.g., if 1990 emissions
are used, average design values for 1990, 1991 and 1992)1. The nonattainment area ozone
design value is the maximum monitored design value from all sites in the nonattainment area.
(4) Estimate the future design value, DVF, for the nonattainment area as the product of the
relative reductions factor (step 2) and the base design value (step 3). If the future design value
is < 124 ppb additional emission reductions should not be needed, no additional steps are
required. If the future design value is > 124 ppb proceed to apply method 1 and method 2
(below) to identify additional emission reductions.
Example 1: Estimate Future Air Quality Design Value
Given: Past results from modeling indicate predicted peaks (for three days) before controls in
1990 are 195, 180, and 165 ppb and after controls in 2007 are 155, 150 and 145 ppb. There are
two monitor sites in the nonattainment area. The monitored air quality design values for each
site are 185 and 176 in 1990, 145 and 152 in 1991, and 155 and 140 in 1992.
Find: Estimate the future air quality design value in 2007.
Solution:
(1) Compute the base and future average 1-hour daily maximum concentration. The average of
the model predicted peaks (in and downwind of the nonattainment area) for the base before
controls is: (195 + 180 + 165) / 3 = 180 ppb and for the future after controls is: (155 + 150 +
145)/3 = 150 ppb.
(2) Using the results in step 1 the relative reduction factor is: 150/180 = 0.83.
(3) Determine the nonattainment area design values representative of the episode used in the base
emissions and calculate the base design value. The nonattainment area design value for 1990 is
MAX(185, 176) = 185, for 1991 is MAX(145, 152) = 152, and for 1992 is MAX (155, 140) =
155 ppb. The base ozone design value is (185 + 152 + 155) / 3 = 164 ppb.
(4) The estimated future design value is (0.83) (164) = 136 ppb
This is > 124 ppb, so we need to apply the following methods to determine additional emission
reductions.
JNote, 1990, 1991 and 1992 design values reflect observations for 1988-90, 1989-91, and
1990-92, respectively. All of these periods include "1990", the year of the base emissions.
4
-------
Method 1: Estimate Additional Emission Reductions Using Modeled Responses
Method 1 uses the change in nonattainment area monitored base ozone design value and
estimated future ozone design value along with changes in modeled emissions before controls
(base emissions) and after controls (future emissions) to estimate additional emission reductions.
(1) Calculate the change in air quality design value by subtracting the estimated future design
value (e.g., 2007) from the base air quality design value (e.g., 1990). Estimate the percent
reduction in NOx emissions and VOC emissions which occurred within the nonattainment area
before and after controls. Do not include biogenic emissions. Divide the percent reduction in
NOx emissions by the change in the air quality design value and divide the percent reduction in
VOC emissions by the change in the air quality design value. This step results in two reduction
factors, one for changes in NOx emissions and one for changes in VOC emissions.
(2) Estimate the amount of additional ozone reduction needed by taking the difference between
the future design value and 124 ppb, the maximum ozone design value consistent with meeting
theNAAQS.
(3) Calculate additional necessary emission reductions by taking the product of each of the
reduction factors (step 1) and the amount of ozone reduction needed (step 2).
Example 2: Calculate reduction factor using model predictions and apply to model estimated
future design value
Given: Results from modeling used in Example 1 indicate an estimated future design value is
136 ppb and the monitored air quality ozone base design value representative of the
nonattainment area is 164 ppb. The control strategy reflects a 30% reduction in VOC and a
35% reduction in NOx emissions. These reductions were obtained by comparing the modeled
1990 base emissions to the modeled 2007 attainment year emissions for the nonattainment area.
Find: The amount of additional VOC and NOx reduction needed to reduce the model estimated
future design value to 124 ppb, so that a convincing weight of evidence argument can be made
for attainment.
Solution:
(1) Calculate the change in air quality design value as 164 - 136 = 28 ppb. The estimated
percent reduction in VOC and NOx are given 30% VOC and 35% NOx. The reduction factor
for VOC is 30% / 28 ppb = 1%/ ppb and for NOx is 35% / 28 ppb = 1.2%/ ppb.
(2) The amount of additional ozone reduction needed is (136 - 124) = 12 ppb.
(3) Therefore, the additional reduction needed in VOC is (1%) (12) = 12% of the VOC
emissions. And, the additional reduction needed in NOx emissions is (1.2%) (12) = 14% of the
5
-------
NOx emissions.
Method 2: Estimate Additional Emission Reductions Using Observed Air Quality Changes
This method uses monitored ozone air quality design values and
emissions estimates for the nonattainment area to calculate the reduction
factors for VOC and NOx. These reduction factors are then applied to the
model estimated future design value as calculated in Example 1 to estimate
additional emission reductions.
(1) Calculate the percent reduction in NOx emissions and VOC emissions
which occurred within the nonattainment area from an earlier year (e.g.}
mqo) to a more recent year (e.g., 1996). The National Emissions
Trends (NET) inventory provides an example of these data. Do not include
biogenic emissions.
(Z) Calculate the change in the nonattainment area's ozone design value
using the same reference years. To account for fluctuations in meteorology average
three years of design values to estimate the design value for each of the reference years. The
nonattainment area average design values are used to assess the observed change in air quality
from the "early" time period to a "recent" time period. Monitors that were only online during
one of these periods may not be representative of the actual change in air quality. Rationale
for excluding a monitor should be documented.
(3) Divide the percent reduction in NOx emissions by the change in the
area's ozone design value. Divide the percent reduction in VOC emissions by
the change in the area's ozone design value. This step gives two reduction
factorsj one for changes in NOx emissions and one for changes in VOC
emissions.
6
-------
(4) Calculate the additional amount of ozone reduction needed by
subtracting 12.4 ppb from the model estimated future design value (see
Example 1).
(5) Calculate additional necessary emission reductions by taking the product of each of the
reduction factors (step 1) and the amount of ozone reduction needed (step 2).
Example 3: Calculate reduction factor using change in ozone air quality design values and
nonattainment area emissions, and apply to model estimated future design value
Given: There are two monitors in the nonattainment area. The monitored air quality design
values for each site for reference years 1990 and 1996 are presented in Table 2. Emission
reductions between 1990 and 1996 are 30% reduction in VOC and a 35% reduction in NOx
emissions. These reductions were obtained by comparing the 1990 NET inventory to the 1996
NET inventory for the nonattainment area. The model estimated future design value in 2007 is
136 ppb.
Table 2. Air Quality Design Values (ppb)
Monitor
1990 Reference Year
1996 Reference Year
1990
1991
1992
1996
1997
1998
1
185
145
155
140
146
139
2
176
152
140
135
145
130
Find: The amount of additional VOC and NOx reduction needed to reduce the future design
value to 124 ppb, so that a convincing weight of evidence argument can be made for attainment.
Solution:
(1) The estimated percent reduction in VOC and NOx are given 30% VOC
and 357° NOx.
(Z) Calculate the change in the nonattainment area's ozone design value.
7
-------
Determine the design value for each, reference year by first taking the
maximum design from the two sites for each of three years and then
averaging the three years design values. The nonattainment area's ozone
design value for 1990 is (12,5 +- 157- +- 155) / 3 = 164 and for 199
-------
Til hie 3: Nonnllninmcnl Area Emissions Siiiniiiiirv (tpd) without Tier 2
Year
VOC
NOx
Point
Area
Mobile
Total
Point
Area
Mobile
Total
1990
400
447
350
1197
300
377
250
927
2007
241
282
200
723
150
312
125
587
Estimated Tier 2
Reduction =
10
25
Find: What are the additional emission reductions in tons per day still needed after incorporating
Tier 2?
Solution:
(1) The additional reductions are (.16 * 1197 tpd) = 192 tpd for VOC and (.19 * 927) = 176 tpd
for NOx.
(2) After subtracting Tier 2 reductions the remaining reductions are (192 - 10) = 182 tpd for
VOC and (176 - 25) = 151 tpd for NOx.
9
-------
Use of Results
The results from both methods should be considered along with other weight of evidence
presented in the technical analyses for the attainment demonstration. For example, where model
predicted peaks show greater improvement when low level NOx emissions are reduced verses
VOC or elevated NOx, substituting an equal amount of low level NOx reductions for the VOC
reductions is acceptable. Also, where modeling demonstrates substantial improvements in
model predicted peaks when emission reductions are applied to adjacent counties, the area of
controls may be extended to include adjacent counties. However, if emissions from adjacent
counties are used they must be included in the total emissions for the base and future. Modeling
the additional emission reductions would normally address these two example as well as the
following: change in boundary conditions due to transport, location of emissions (such as point,
area or mobile), elevated vs low level emission reductions, chemistry and wind flow patterns.
Model sensitivity runs may be used to help identify the appropriate controls measures to fill the
additional emission reductions needed to provide for attainment in the weight of evidence
analyses.
For guidance on VOC and NOx substitution use the, "NOx Substitution Guidance", EPA
1993; "Transmittal of NOx Substitution Guidance", memorandum from John Seitz, 1993;
"Clarification of Policy for Nitrogen Oxides (NOx) Substitution", memorandum from John Seitz,
1994; and "Guidance for Implementing the 1-Hour Ozone and Pre-Existing PM10 NAAQS",
memorandum from Richard D. Wilson, 1997. The 1993 and 1994 guidance was primarily
designed for the post-1996 rate of progress (3%/year VOC reduction) requirement and allowed
NOx reductions to be substituted for the otherwise mandatory VOC reductions as long as the
NOx reductions were shown to be consistent with the attainment demonstration (in other words,
if the attainment demo relied only on VOC reductions, the area could not substitute NOx
reductions for the 3%/year requirement, and if the attainment demo relied on both VOC & NOx
reductions, NOx could be substituted in part). The 1994 guidance document (Guidance on the
Post-1996 Rate-of-Progress Plan and the Attainment Demonstration, EPA-452/R-93-015, Jan.
1994) provided equations & procedures for calculating the amount of NOx reductions that could
be substituted for VOC for the rate of progress requirements. Also, the 1997 guidance
establishes the 100 & 200 km distances for substitution of emission reductions outside the
nonattainment area. These documents are located on the EPA website:
"www.epa.gov/ttn/oarpg/tlpgm.html".
10
------- |