1. 1 NOAA-EPA’s National Air Quality Forecast Capability: Progress and Plans October 17, 2006 Paula M. Davidson 1, Nelson Seaman 1, Jeff McQueen 1, Rohit Mathur 1,2, Roland Draxler 1, Richard Wayland 2 1 National Oceanic and Atmospheric Administration (NOAA) 2 US Environmental Protection Agency (EPA)

2. 2 Outline Overview of Current Capability Progress in 2006 toward expanded capabilities Looking ahead

3. 3 Model Components: Linked numerical prediction system Operationally integrated on NCEP’s supercomputer NCEP mesoscale NWP: WRF-NMM NOAA/EPA community multiscale model for AQ: CMAQ Observational Input: NWS real-time weather observations EPA emissions inventory National Air Quality Forecast Capability End-to-End Operational Capability AQI: Peak Jul 28 Gridded forecast guidance products On NWS Telecommunications Gateway and EPA servers Updated 2x daily Verification basis EPA/AIRNow compilation: ground-level ozone observations Customer outreach/feedback State&Local AQ forecasters coordinated with EPA Public and Private Sector AQ constituents EPA Monitoring Network

4. 4 Operational AQ forecast guidance www.weather.gov/aq Further information www.nws.noaaa.gov/ost/air_quality

5. 5 Experimental Products: Coast-to-coast Ozone, Smoke Forecast Guidance www.weather.gov/aq-expr

6. 6 166 142 268 grid cells 259 grid cells Northeast “1x” Domain East “3x” Domain Testing Domain: Summer 2006 Testing Domain: Summer 2006 Experimental: CONUS “5X” 442 grid cells 265 grid cells IOC “1x” EUS “3X” Operational: EUS “3x” CONUS “5x” Domain

7. 7 Testing Summary Summer, 2006 Ozone: Experimental testing beginning June, 2006 over CONUS –New WRF-CMAQ linkage, improved vertical coupling with σ-P adopted in CMAQ/WRF –Updates to emissions (esp mobile and EGU sources) –ACM mixing in clouds Smoke: Experimental testing beginning March, 2006 –Fire Locations and verification based on satellite observations –Fire emissions estimates from USFS (BlueSky) –HYSPLIT/NAM transport Aerosols: Developmental testing providing comprehensive dataset for diagnostic evaluations –CMAQ (aerosol option) –Qualitative; underprediction consistent with missing source inputs

8. 8 Preliminary Analysis of Ozone Predictions: Summer, 2006

9. 9 Testing Challenges: Summer, 2006 1. Comparison of Operational and Experimental testing: Before Isoprene correction for 5X E US (3X) vs. CONUS (5X): After Isoprene correction for 5X

10. Ozone Episode in Eastern US: July 16-19, 2006 July 17 July 16 July 20 July 18 July 19

11. 11 Experimental Testing Summary: Ozone Summer, 2006 Diagnosing and addressing issues not uncovered with retrospective testing  WRF implementation into NAM: June 20WRF implementation into NAM: June 20 Relative to Eta: Similar performance in Eastern US, but cooler, wetter in Western US  Lower surface O3 –WRF upgrade/correction: August 15 Increased mixing: more vertical mixing in stable PBL, horizontal mixing in complex terrain (mountains and along coast) Correction impacts: localized; more diffuse O3 gradientsCorrection impacts: localized; more diffuse O3 gradients Operational (3X) ozone: WRF-Post Processor.Operational (3X) ozone: WRF-Post Processor. T-Interpolation (vertical; in WRF-post) error  Slightly higher ozone Corrected July 28. Impact: slightly reduced surface O3Corrected July 28. Impact: slightly reduced surface O3 Experimental (5X) ozone: PREMAQ/CMAQ correctionsExperimental (5X) ozone: PREMAQ/CMAQ corrections Experimental (5X) Ozone and Developmental Aerosols: Biogenic VOCs incorrectly underestimated  Much lower surface O3 Corrected August 5. Impact: Much increased surface O3Corrected August 5. Impact: Much increased surface O3 (3X and 5X) PREMAQ/CMAQ corrections(3X and 5X) PREMAQ/CMAQ corrections Some emissions components left from 2005; numerics: (arrays undefined, Jacobian error) Corrected August 5; September 5. Minor (localized) impactsCorrected August 5; September 5. Minor (localized) impacts

12. 12 HYSPLIT Analysis Sample Smoke Evaluation - April 13, 2006 NAM/HYSPLIT 1-d Forecast & HMS Analysis 12 -14 UTC HYSPLIT 24-hr Forecast

13. 13 WRF-HYSPLIT 24-h Smoke Predictions: Preliminary Evaluation, 2006 14 September 2006 WRF-HYSPLIT 24h (blue) vs HMS (orange) Preliminary Verification statistics: FMS Threshold: 1ug/m3 in Column Initial target: 10%

14. 14 Preliminary Aerosol Predictions: Summer, 2006 July 25, 2006

15. 15 Testing Summary Summer, 2006 Ozone: Extensive retrospective testing with developmental NAM-CMAQ system during Winter and Spring still left some surprises: Experimental (5X) ozone:Experimental (5X) ozone: –Large under-predictions especially in the west; systematically lower ozone than operational (3) in East Operational (3X) ozone:Operational (3X) ozone: –Some over-predictions for cool, cloudy conditions Smoke: Experimental testing providing basis for systematic near-real time evaluations with satellite observations: –Results promising for predicted smoke transport, compared to HMS smoke analyses (independent, analyst-prepared) –Development of objective smoke analyses for near-real time verification nearly complete Aerosols: Developmental testing providing comprehensive dataset for diagnostic evaluations: Underprediction common; consistent with missing source inputsUnderprediction common; consistent with missing source inputs

16. 16 National AQF Capability: Looking ahead Ozone forecast guidance produced operationally (WRF- CMAQ) over CONUS (near-term) Improving day-to-day performance, especially in the westImproving day-to-day performance, especially in the west Transitioning experimental products to operationsTransitioning experimental products to operations Expanded products to be tested: Ozone:Ozone: –Further development: to more closely couple AQ with WRF weather output; examine impacts of vertical resolution, vertical mixing treatments, horizontal boundary conditions… –Testing over all 50 states Particulate matter components:Particulate matter components: –Smoke from large fires: objective verification baseline; transition to operations –Aerosols predicted from anthropogenic source emissions in inventories: continued development/testing/analysis

17. 17 National Air Quality Forecast Capability: Phased Growth Early Implementations: 1-day forecast guidance for ozone Developed and deployed initially for Northeastern US, September 2004 Developed and deployed initially for Northeastern US, September 2004 Expanded over Eastern US, Aug 2005 Expanded over Eastern US, Aug 2005 Near-Term targets: Implement ozone guidance over CONUS: Summer 2007 Deploy Nationwide (AK & HI) target: by 2009 Longer range (within 10 years): Develop and implement capability to forecast particulate matter (PM) –Particulate size < 2.5 microns Data assimilation for air qualityData assimilation for air quality Extend air quality forecast range to 48-72 hoursExtend air quality forecast range to 48-72 hours Include broader range of significant pollutantsInclude broader range of significant pollutants 2005: O 3

18. 18 Appendix

19. 19 National AQF Capability Status October, 2006 Ozone forecast guidance produced operationally (WRF-CMAQ): Eastern US. 2X daily, 12km grid resolution, hourly predictions, through midnight next day. Ground-level ozone concentrationsEastern US. 2X daily, 12km grid resolution, hourly predictions, through midnight next day. Ground-level ozone concentrations –On NOAA/NWS servers: www.weather.gov/aq www.weather.gov/aq –On EPA’s AIRNow website, displayed as AQ Index values –Eta was replaced in the NAM by WRF(NMM) in June, 2006 –Achieving performance targets for accuracy, reliability Expanded products being tested in FY06: Ozone:Ozone: CONUS domain, experimentally available: from June, 2006 –Further developmental work to couple AQ more closely with WRF weather output Particulate matter components:Particulate matter components: Smoke from large fires, experimentally available (CONUS): from March 2006 Transport prediction; qualitative. USFS fire emissions, passive transport. Aerosols produced/transported; sources from anthropogenic emissions in climatologic inventories; in developmental testing (CONUS): from June, 2006