Transcript Slides - IIASA

Using Air Quality Models for
Emissions Management Decisions
S. Trivikrama Rao
U.S. Environmental Protection Agency,
Research Triangle Park, NC
and
Christian Hogrefe
University at Albany, Albany, NY
Air Quality Standards
• The NAAQS focus on the extreme values of the
distribution of observed concentrations:
• Ozone
– 1-hr standard: 4th highest 1-hr ozone concentration over a
consecutive 3-year period must not exceed 0.12 ppm (124 ppb).
– 8-hr standard: 4th highest 8-hr ozone concentration in each year,
averaged over a consecutive 3-year period, must not exceed 0.08
ppm (84 ppb).
• PM2.5 mass
– 24-hr standard: The 98th percentile of 24-hr concentrations in each
year, averaged over a consecutive 3-year period, must not exceed
65 mg/m3
• PM10 mass
– 24-hr standard: The 99th percentile of 24-hr concentrations in each
year, averaged over a consecutive 3-year period, must not exceed
150 mg/m3.
• Disparity between the extreme-value focused NAAQS
and the performance of regional-scale photochemical
modeling systems in reproducing extreme values:
– Poor performance in predicting individual extremes (uncertainty
is on the order of 20% - 40%)
– This disparity warrants against the direct use of model-predicted
extreme concentration values in the attainment demonstrations
• Promulgation of “relative reduction factors” (RRF) in
attainment demonstrations by the US EPA (EPA, 1999)
– RRF is site-specific and is defined as the relative change in
model-predicted average daily maximum ozone concentrations
due to emission reduction.
– RRF is used to scale the observations to determine if the scaled
design value is in compliance with the NAAQS
 Examine model-to-model differences in the RRFs
 Integrate observations and model predictions into a
probabilistic framework using appropriate statistical
techniques (e.g. extreme values statistics, Bootstrapping)
Uncertainties in the Attainment
Demonstration (Modeling) Process
• Inherent uncertainty: caused by processes that
operate on space-time scales not resolved by the
grid-based models
• Reducible uncertainty: caused by our incomplete
scientific understanding of the various dynamical
processes (manifested in the different predictions
from different modeling systems, different
meteorological models or parameterizations,
different chemical mechanisms, or uncertainties in
the emissions inventories)
• Interannual variability: US EPA’s attainment
demonstration process focuses on a fixed 3-year
base window (e.g. ENSO affects pollutant levels)
RRF Differences Between Models
• Relative ozone response to two hypothetical emission
reduction scenarios is very similar between two different
modeling systems
Estimating Exceedance
Probabilities
• The CDF of the k-th highest value of a distribution
with CDF F(x) is given by equation:
n
j
n j
G( x )      1  F( x )  F( x )
j 0  j 
k -1
• Extreme value theory is applicable to the U.S. 1-hr
ozone design value
• For the U.S. 8-hr ozone design value: Apply the
combination of exponential tail fit and bootstrap
(resampling method) to estimate the design value
CDF
Predicted NAAQS Exceedance Probabilities
• Predicted 8-hr ozone NAAQS
exceedance probabilities for 1994 –
1996 base period (top) and 25%/50%
NOx/VOC emission reduction scenarios
simulated by two different modeling
systems (middle and bottom)
• Combining the extreme value approach
with model-predicted relative responses
to emission reductions yields very
similar spatial patterns of NAAQS
exceedance probabilities for different
modeling systems
Real World Emission Control Strategies
• The probability of exceeding the 8-hr ozone NAAQS for
(a) base period, b) 2007 SIP scenario, c) 2007 SIP + Tier 2
scenario, d) 2030 SIP scenario, and e) 2030 SIP + Tier 2
scenario
Summary
• The predicted relative changes in the daily maximum
ozone concentrations and the estimated probabilities of
violating the NAAQS stemming from emission controls
are more similar between different modeling simulations
than the predictions of absolute daily maximum ozone
concentrations
• The presented approach enables policy-makers to examine
the probability of achieving compliance with the NAAQS
under varying emission control options
• The results reveal that emission reductions proposed in the
US EPA’s SIP call can substantially reduce the probability
of violating both the 1-hr and 8-hr ozone NAAQS over a
large portion of the eastern United States