WRAP 2028 Visibility Projection Methods draft April 29

WRAP 2028 Visibility Projection Methods – draft April 29, 2020 112 Western U. S. Class I areas – contiguous WESTAR-WRAP region Evaluate Modeled Visibility results to project future impairment and 2028 Reasonable Progress Goal (deciviews) to be set for Class I area(s) in your state Explore 2028 visibility projection options accounting for EPA guidance and western air quality issues: 1) fire and/or dust source impacts 2) isolate anthropogenic U. S. sources, and/or 3) account more explicitly for model performance All options to be available on TSS using selection tools EPA Default Method: (EPA Modeling Guidance for Demonstrating Attainment of Air Quality Goals for Ozone, PM 2. 5, and Regional Haze - final, November 2018) EPA Method corrected: Evaluate Most Impaired days to exclude fire and dust impacts to the greatest degree possible Mod. MID Method: Modeled Most Impaired Days due to U. S. anthro SO 2 and NOx emissions are used to project 2028 visibility directly Greatest IMPROVE Sample Day Population (GISDP) Method: use different #s of IMPROVE sample days by site that exclude fire and dust to a great degree and have best model performance vs. monitored values to project 2028 visibility

1) EPA projection method without adjustment – ROMO Simplest case: A. 2028 projections are below URP and 2014 -2018 monitored average • A. ROMO Summary 2000 -2004 baseline: 11. 1 dv 2014 -2018: 8. 4 dv 2028 OTBa: 7. 3 dv 2028 OTBb: 8. 1 dv Natural Haze: 4. 9 dv International: TBD 2028 clearest: 1. 1, 1. 3 dv The 2 projections use the 2012 -16 and 2014 -18 monitoring averages for their respective Relative Response Factors B. MID are dominated by Amm. SO 4 and Amm. NO 3 C. Amm. NO 3 and Amm. SO 4 aerosols decreasing 2000 -18 B, D. Look at aerosol Modeled to Obs for MPE and outliers C.

1) EPA Projection Method without adjustment – OLYM Minimum Evaluation: A. 2028 projections are below URP and 2014 -2018 monitored averages B. MID are dominated by Amm. SO 4 and Amm. NO 3 C. Amm. NO 3 and Amm. SO 4 aerosols decreasing 2000 -18 D. Look at aerosol Modeled to Obs for MPE and outliers E. Correct outliers? • Method 2: Check PSAT Rep. Base for SO 4 fraction due to natural (DMS) or international anthro • Method 4: correction for MPE (e. g. Amm. NO 3) B. A. OLYM Summary 2000 -2004 baseline: 14. 9 dv 2014 -2018: 11. 9 dv 2028 OTBa: 11. 4 dv 2028 OTBb: 10. 9 dv Natural Haze: 6. 9 dv International: TBD 2028 clearest: 3. 4, 3. 6 dv C.

2) EPA Method corrected for non-US Anthro contributions – OLYM Method 2 Criteria: A. Use PSAT Rep. Base to assign daily aerosols to A. - US Anthro - Canada/Mexico B. - Comm Marine - International Anthro C. - natural - Oceanic natural (DMS) PSAT to define non-US Anthro contributions to SO 4: A. Using Example 2011 WAQS PSAT. IMPROVE aerosol weighted by PSAT rrf 2011 IMPROVE MID International + Natural B. Check PSAT for SO 4 fraction from natural (DMS) or international anthro C. Go to Method 3, Rank Modeled US Anthro MID D. Go to Method 4, PSAT for all IMPROVE days C. US Anthro B, D. 2011 all IMPROVE days

2) EPA method corrected for Fire impacts on MID – SAWT Method 2 Criteria: A. SAWT 2028 a and 2028 b projections above URP, slightly below 2014 -18 monitored average B. OMC dominant on many MID. C. PSAT Rep. Base assigns OMC to fire and natural, as seen below for 2011 MID. D. Exclude MID with high fire impacts from 2028 rrf calculations. E. Go to Method 3: Rank Modeled US Anthro MID F. Go to Method 4: A. SAWT Summary 2000 -2004 baseline: 9. 6 dv 2014 -2018: 8. 6 dv 2028 OTBa: 8. 3 dv 2028 OTBb: 8. 5 dv Natural Haze: 4. 7 dv International: TBD 2028 clearest: 2. 4 dv • Expand # IMPROVE days used for evaluation B. C. 2014 2011 Natural OMC on MID

2) EPA Method corrected for fire OMC – GLAC Method 2: A. GLAC 2028 a and 2028 b projections above URP, about the same as 2014 -18 monitored average B. Several MID in summer with elevated OMC C. EPA methods overestimates anthro OMC at GLAC (see 2011 PSAT example below) D. Use PSAT Rep. Base to show fire impacts on MID E. Exclude MID with high fire impacts from 2028 rrf calculation. F. Go to Methods 3 and 4 C. B. A. GLAC Summary 2000 -2004 baseline: 15. 9 dv 2014 -2018: 13. 8 dv 2028 OTBa: 13. 0 dv 2028 OTBb: 13. 7 dv Natural Haze: 4. 9 dv International: TBD 2028 clearest: 5. 0, 5. 4 dv

2) EPA method corrected for high Dust and Carbon – SYCA Method 2: A. SYCA 2028 a and 2028 b projection above URP, but lower than 2014 -18 monitored average B. 2014 v 2 model 23 X over predicts OMC on 7/19/14, • Remove 7/19/14 in 2028 OTBa rrf calculation C. Several 2014 MID with elevated Coarse and Fine Soil • CAMx under predicts windblown (natural) dust D. Make species stacked bar plot of trend and projections? E. Use PSAT Repbase to identify US Anthro fraction in MID F. Go to Method 3, Modeled US Anthro MID G. Go to Method 4, All days D. A. SYCA Summary 2000 -2004 baseline: 12. 2 dv 2014 -2018: 11. 6 dv 2028 OTBa: 11. 0 dv 2028 OTBb: 10. 8 dv Natural Haze: 4. 7 dv International: TBD 2028 clearest: 3. 8, 4. 0 dv B. C. OMC = 232 Mm-1

2) EPA method corrected for High Dust – SACR Method 2: A. SACR 2028 a and 2028 b projection above URP, similar to 2014 -18 monitored average B. Amm. SO 4 and Amm. NO 3 dominate most MID C. Look at Modeled vs Obs for MPE or outliers D. Look at PSAT Rep. Base for natural vs US anthro contributions on MID • PSAT can identify anthro dust, CAMx under predicts windblown (natural) dust E. Go to Methods 3 and 4 B. C. A. SACR Summary 2000 -2004 baseline: 16. 5 dv 2014 -2018: 15. 0 dv 2028 OTBa: 14. 7 dv 2028 OTBb: 16. 2 dv Natural Haze: 5. 5 dv International: TBD 2028 clearest: 6. 3, 6. 5 dv

Current EPA projection method likely underestimates progress • Artifacts in the method that EPA uses to evaluate progress: • Estimate of natural haze (very large uncertainty) • Estimate of international impairment (large uncertainty) • Calculates U. S. impairment by subtracting estimates of natural haze and international impairment from the total haze. • This approach transfers uncertainty in natural and international to the U. S. impairment. • Instead, WRAP will use the model to calculate U. S. impairment in the baseline period and 2028. • WRAP will apply modeled U. S. impairment in the Weight of Evidence dynamic modeling. • The model is most reliable for estimates of U. S. impairment. • Uncertainty in natural and international will not affect the estimates of U. S. progress. 9

3) Modeled US Anthro MID, starting with MID evaluation – YELL A. Method 3: A. YELL 2028 a and 2028 b projections above URP, similar to 2014 -18 monitored average B. OMC dominant on many 2014 MID. C. OMC increasing as fraction of MID over 2000 -18 D. Go to Method 3: Rank Modeled US Anthro MID E. Go to Method 4: • Expand # IMPROVE days evaluated C. B. YELL Summary 2000 -2004 baseline: 8. 3 dv 2014 -2018: 7. 5 dv 2028 OTBa: 7. 0 dv 2028 OTBb: 7. 7 dv Natural Haze: 4. 0 dv International: TBD 2028 clearest: 6. 3, 6. 5 dv

Method 3: Rank Modeled US Anthro MID – YELL • • Rank modeled days by daily US Anthro Fraction of Total Aerosol Extinction in Rep. Base (see 2011 WAQS example) Select x number of days for rrf calculation. Use same days in 2028. Apply rrf to 2014 -2018 5 yr ave IMPROVE and compare to URP for MID 2011

Method 4: Evaluate Additional IMPROVE days - YELL • Rank and select IMPROVE days with best MPE for NO 3 and SO 4 (examples from 2011 WAQS) • Use PSAT to evaluate US Anthro fraction of Rep. Base and 2028 • Select IMPROVE days for ALL, Group 50 or other range 2011