WRAP 2002 Visibility Modeling: Emission, Meteorology Inputs and CMAQ Performance Evaluation

1. WRAP 2002 Visibility Modeling:Emission, Meteorology Inputs andCMAQ Performance Evaluation Gail Tonnesen, Bo Wang, Chao-Jung Chien, Zion Wang, Mohammad Omary University of California, Riverside Zac Adelman, Andy Holland University of North Carolina Ralph Morris et al. ENVIRON Corporation Int., Novato, CA WRAP Attribution of Haze Meeting, Denver, CO July 22, 2004

2. Annual MM5 Simulations run at the RMC Emissions processed with SMOKE Preliminary 2002 Scenario C used here. CMAQ version 4.3 (released October 2003) Data summaries, QA, results are posted on the RMC web page: www.cert.ucr.edu/aqm/308 Summary of RMC 2002 Modeling

3. MM5 Modeling Domain (36 & 12 km) • National RPO grid • Lambert conic Projection • Center: -97o, 40o • True lat: 33o, 45o • MM5 domain • 36 km: (165, 129, 34) • 12 km: (220, 199, 34) • 24-category USGS data • 36 km: 10 min. (~19 km) • 12 km: 5 min. (~9 km)

4. MM5 Physics

5. Subdomains for 36/12-km Model Evaluation 1 = Pacific NW 2 = SW 3 = North 4 = Desert SW 5 = CenrapN 6 = CenrapS 7 = Great Lakes 8 = Ohio Valley 9 = SE 10 = NE 11 = MidAtlantic

6. Evaluation Review • Evaluation Methodology • Synoptic Evaluation • Statistical Evaluation using METSTAT and surface data • WS, WD, T, RH • Evaluation against upper-air obs • Statistics: • Absolute Bias and Error, RMSE, IOA (Index of Agreement) • Evaluation Datasets: • NCAR dataset ds472 airport surface met observations • Twice-Daily Upper-Air Profile Obs (~120 in US) • Temperature • Moisture

7. METSTAT Evaluation Package • Statistics: • Absolute Bias and Error, RMSE, IOA • Daily and, where appropriate, hourly evaluation • Statistical Performance Benchmarks • Based on an analysis of > 30 MM5 and RAMS runs • Not meant as a pass/fail test, but to put modeling results into perspective

8. Evaluation of 36-km WRAP MM5 Results • Model performed reasonably well for eastern subdomains, but not the west (WRAP region) • General cool moist bias in Western US • Difficulty with resolving Western US orography? • May get better performance with higher resolution • Pleim-Xiu scheme optimized more for eastern US? • More optimization needed for desert and rocky ground? • MM5 performs better in winter than in summer • Weaker forcing in summer • July 2002 Desert SW subdomain exhibits low temperature and high humidity bias 2002 MM5 Model Evaluation 12 vs. 36 km Results Chris Emery, Yiqin Jia, Sue Kemball-Cook, and Ralph Morris (ENVIRON International Corporation) & Zion Wang (UCR CE-CERT), Western Regional Air Partnership (WRAP) National RPO Meeting, May 25, 2004

9. WRAP 36km/12km July Wind Performance Comparison 120 100 80 Wind Direction Error (degrees) 60 PacNW SW DesertSW 40 North 20 0 0 0.5 1 1.5 2 2.5 3 3.5 Wind Speed RMSE (m/s) Benchmark 12 km Subdomains MM5/RAMS Runs 36 km Subdomains

13. The RMC is continuing to test alternative MM5 configurations – to be completed at the end of 2004. Expect some reduction in bias &error in the WRAP states, however even in the best case we will have error & bias in MM5 that must be considered when using CMAQ for source attribution. MM5 Implications for AoH

14. Preliminary 2002 Scenario C based on the 1996 NEI, grown to 2002, with many updates by WRAP contractors and other RPOs. Processed for CMAQ using SMOKE. Extensive QA plots on the web page Both SMOKE QA and post-SMOKE QA Emissions Inventory Summary

15. Emissions Sources by Category & RPO

16. WRAP 2002 Annual NOx Emissions Area Biogenic On Road Non Road Road Dust Point Rx Fire Ag Fire Wildfire Offshore

17. 2002 WRAP NOx Emissions by Source & State Ag Fire 1400000 Rx Fire 1200000 Wildfire 1000000 Area [Tons/Yr] Point 800000 Nonroad 600000 Onroad 400000 200000 Utah Idaho 0 Oregon Nevada Arizona Montana Wyoming Colorado California Washington New Mexico North Dakota South Dakota

18. WRAP 2002 Annual SO2 Emissions Area Biogenic On Road Non Road Road Dust Point Rx Fire Ag Fire Wildfire Offshore

19. Onroad 2002 WRAP SO2 Emissions by Source & State Ag Fire 3.00E+05 Rx Fire Wildfire 2.50E+05 Area Nonroad 2.00E+05 Point [Tons/Yr] 1.50E+05 1.00E+05 5.00E+04 0.00E+00 Utah Idaho Oregon Nevada Arizona Montana Wyoming Colorado California Washington New Mexico North Dakota South Dakota

20. 2002 WRAP NH3 Emissions by Source Category 2.50E+05 Nonroad Ag Fire 2.00E+05 Rx Fire Point 1.50E+05 Onroad Tons/Yr Wildfire Area 1.00E+05 5.00E+04 0.00E+00 Nevada Utah Idaho Oregon Arizona Montana Wyoming Colorado California Washington New Mexico North Dakota South Dakota

21. Preliminary 2002 EI Used here. Updates for final 2002 EI will include: New EI data from other RPOs and Canada 2002 NEI to replace grown 1996 NEI Reprocess in SMOKE with final MM5 All final inputs ready now except Canada & MM5 Emissions Summary

22. CMAQ v4.3 36-km grid, 112x148x19 Annual Run CB4 chemistry Evaluated using: IMPROVE, CASTNet, NADP, STN, AIR/AQS CMAQ Simulations

23. Guidance from EPA not yet ready: Difficult to assert that model is adequate. Therefore, we use a variety of ad hoc performance goals and benchmarks to display CMAQ results. We completed a variety of analyses: Compute over 20 performance metrics Scatter-plots & time-series plots Soccer plots Bugle plots PM Performance Criteria

24. We completed a variety of analyses: Compute over 20 performance metrics Scatter-plots & time-series plots Soccer plots Bugle plots Goal is to decide whether we have enough confidence to use the model for AoH: Is this a valid application of the model? Goal of Model Evaluation

25. Plot error as as a function of bias. Ad hoc performance goal: 15% bias, 35% error based on O3 modeling goals. Larger error & bias are observed among different PM data methods and monitoring networks. Performance benchmark: 30% bias, 70% error (2x performance goals) PM models can achieve this level in many cases. Soccer Goal Plots

26. Annual CMAQ vs IMPROVE

27. Spring Summer Fall Winter

28. Annual CMAQ vs CASTNet

29. Spring Summer Fall Winter

30. Annual CMAQ vs STN

31. Spring Summer Fall Winter

32. Annual CMAQ vs NADP

33. Spring Summer Fall Winter

34. Performance Goals and Criteria- Proposed by Jim Boylan • Based on FE and FB calculations • Vary as a function of species concentrations • Goals: FE +50% and FB  ±30% • Criteria: FE +75% and FB  ±60% • Less abundant species should have less stringent performance goals and criteria

35. Performance Goals and Criteria- Proposed by Jim Boylan • PM Performance Goals • Proposed PM Performance Criteria

36. Monthly SO4 Fractional Bias

37. Monthly SO4 Fractional Error

38. Monthly NO3 Fractional Bias

39. Monthly NO3 Fractional Error

40. Monthly NH4 Fractional Bias

41. Monthly NH4 Fractional Error

42. Monthly OC Fractional Bias

43. Monthly OC Fractional Error

44. Monthly EC Fractional Bias

45. Monthly EC Fractional Error

46. Monthly PM25 Fractional Bias

47. Monthly PM25 Fractional Error

48. TSSA results are run in CMAQ v4.4 with emissions version Preliminary 2002 C Performance evaluation used CMAQ 4.3 Previous CMAQ runs used CMAQ 4.3 with Preliminary 2002 B emissions (no fires) CMAQ & EI Versions

49. CMAQ v4.3 & v4.4 versus IMPROVE July

50. CMAQ v4.3 Mean fractional bias (no filter) January +25% MFB July –20% mean MFB Slightly worse January O3 performance in v4.4 CMAQ Ozone Performance