AoH Work Group Weight of Evidence Framework WRAP Meeting

1
AoH Work GroupWeight of Evidence Framework
WRAP Meeting Tucson, AZJanuary 10/11, 2006

• Joe Adlhoch - Air Resource Specialists, Inc.

2
Overview

• Review of RHR visibility goals
• What do we mean by weight of evidence (WOE)
  approach?
• Review of model approach to determine reasonable
  progress
• Review of other data inputs

3
Review of RHR Visibility Goals

• Define current conditions at each Class I area
  using the 2000-04 baseline period
• Define natural conditions
• Improve visibility such that the average Haze
  Index for the 20 worst days in the baseline
  period reach natural conditions by 2064
• Ensure that visibility on the 20 best days does
  not degrade
• Periodically assess the improvement in visibility
  between the baseline period and 2064 and show
  that reasonable progress is being achieved

4
Schematic of Glide Path
From Guidance for Estimating Natural Visibility
Conditions Under the Regional Haze Rule, EPA 2003
5
WOE Definition

• Set of analyses supplemental to primary
  measurement/modeling efforts
• WRAP AoH working definition
• Review of all available analyses that bear on
  Class I area visibility
• Monitoring data
• Emissions data
• Model results
• Attribution results (combination of multiple
  methods)
• Review of trends (monitoring and emissions)
• Review of episodic (natural ?) events
• Back trajectory and other analyses
• Assigning appropriate weight to each analysis
  (based on relevance and uncertainty)
• Ultimately, this will take the form of a
  checklist of things to review and instructions on
  how to weigh each piece

6
Use of AQ Model to Estimate 2018 Visibility
(simplified)

• Assumption the AQ model is better at predicting
  relative changes in concentration than absolute
  concentrations
• Steps
• Determine the 20 worst days from the 2002
  IMPROVE data
• Model species concentrations for 2002
• Model species concentrations for 2018 base and
  scenarios
• Determine a species-specific relative reduction
  factor (RRF) for the average of the 20 worst
  days (based on step 1 above)
• RRFsulfate 2018sulfate / 2002sulfate
• Project 2018 concentrations by applying the RRFs
  to the IMPROVE data for the 20 worst days in
  each baseline year
• Projected 2018concentration Avg. RRF x
  Baselineconcentration
• Calculate projected 2018 visibility for 20 worst
  days and compare to the Glide Path

7
2002 Model Performance Agua Tibia, CA
8
2018 -2002 Model Change Agua Tibia, CA
9
2002 Model Performance Zion, UT
10
2018 -2002 Model Change Zion, UT
11
Is Model Prediction of Reasonable Progress
Reasonable?

• Determine if the major species causing visibility
  impairment are handled well by the model
• The variability in the 5-year baseline could be
  used as an uncertainty range to bound the
  projected 2018 visibility
• Which species most affect variability?
• Meteorological dependencies?
• Could this be tied to monitoring uncertainties?
• Are there episodic events that could justifiably
  be removed from the data set (e.g., large fire
  episodes during baseline period)?
• Review attribution source regions and their
  emissions
• How well do attribution methods agree?
• If source regions can be identified with
  confidence, do the projected emissions reductions
  for 2018 support the models visibility
  reductions?

12
Median Uncertainty of IMPROVE Data Across WRAP

• Uncertainty based only on lab reported
  uncertainties for daily samples (2000 2004)

• OC, EC, Soil, and CM uncertainty determined from
  standard propagation of error analysis on
  individual component terms
• Uncertainty due to flow/size cut errors not
  included

13
Glide Path for Agua Tibia, CA
14
Glide Path for Agua Tibia, CA
Baseline Variability (dv)
Baseline Variability by Species
15
Glide Path for San Gabriel, CA
16
Glide Path for San Gabriel, CA
Baseline Variability (dv)
Baseline Variability by Species
17
Glide Path for Goat Rocks, WA
18
Glide Path for Goat Rocks, WA
Baseline Variability (dv)
Baseline Variability by Species
19
Large Episodic Fire Impacts in 2002
20
SO2 Point and Area Emissions Reductions
21
NOx Point and Area Emissions Reductions
22
Expected Attribution Results

• The modeled attribution results (CAMx and PSAT
  method) will tell us how much species mass is
  likely due to specific source regions (states,
  Canada, Mexico, Pacific, etc.)
• The results can be displayed as
• Amount or percent of species mass attributed by a
  region
• Amount or percent of extinction attributed by a
  region

23
Phase I Attribution Graphics
24
Phase 2 Attribution Footprint

• The following maps show mock ups for how
  attribution results might be displayed in Phase 2
  (data shown is from Phase I)
• Helps to answer the questions
• Which states need to consult on visibility issues
• What contributions to haze might be coming from
  outside the WRAP or the U.S.

25
Phase I Sulfate and Nitrate Extinction Attributed
to Arizona (TSSA Analysis)
26
Phase I Sulfate and Nitrate Extinction Attributed
to Oregon (TSSA Analysis)
27
Phase I Sulfate Extinction Attributed to WRAP
States (excluding UT, WA, WY)
Phase I clustering based on SO4/NO3 attribution
28
Phase I Sulfate Extinction Attributed to non-WRAP
Source Regions
29
Phase I Nitrate Extinction Attributed to non-WRAP
Source Regions