Models-3’s Application in FAQS Project: Impact of Urban Definition’s Change on Model Results

1. Models-3’s Application in FAQS Project: Impact of Urban Definition’s Change on Model Results Yongtao Hu, M. Talat Odman, Maudood Khan, and Armistead G. Russell Air Resource Engineering Center, Georgia Institute of Technology October 22, 2002 Georgia Institute of Technology

2. Outline • Introduction to FAQS Project • First Application of Models-3 in Georgia’s Three Metro Areas • Investigation of the Impact of Urban Definition’s Change on the Model Results • Summary Georgia Institute of Technology

3. Georgia Institute of Technology

4. Fall line Air Quality Study (FAQS) • Georgia’s Three Metro Areas Other Than Atlanta: Augusta, Columbus, and Macon may also be Experiencing Poor Air Quality. • In 2000, Georgia EPD Launched FAQS to Assess Urban and Regional Air Pollution, to Identify the Sources of Pollutants and Pollutant Precursors, and to Recommend Solutions to the Poor Air Quality in the Three Metro Areas. • Why This Name? As the Three Cities Lie Along Georgia’s “Fall Line” – the Line Dividing the Piedmont Region from the Coastal Plain. • FAQS Includes Enhanced Monitoring, Emission Inventory Development, Air Quality Modeling and Control Strategy Recommendation. Georgia Institute of Technology

5. FAQS Air Quality Modeling: First Scenario • Focus on the Three Metro Areas as well as Atlanta • First Scenario: Ozone Episode of August 11-20, 2000 • First Application: In a Attempt to Validate the models and to Improve the Models’ Performance Georgia Institute of Technology

6. Georgia Institute of Technology

7. Model Setup and Parameters • Meteorology: MM5 • MM5 grids are 3 cells larger on each side than CMAQ grids • 34 vertical layers with the top at 70mb • NCEP ETA data and ADP observational data as inputs • OSU land-surface scheme, MRF, … • One-way nesting, Surface FDDA for only winds and Gridded FDDA ( no FDDA with finest grid) • Emissions: SMOKE • SAMI inventory for 1995 with Projection Factors from EGAS • OTAG’s area spatial surrogates and USGS’s urban area and major highways • Mobile5b for applying VMT inventory • SAPRC99 • BEIS3 with BELD3 Georgia Institute of Technology

8. Model Setup and Parameters (continued) • Air Quality: CMAQ • FAQSD1, 36-km, 78x66 cells; FAQSD2, 12-km,78x66 cells; FAQSD3, 4-km, 102x78 cells • 22 vertical layers, 10 layers in the lowest kilometer • Default initial and boundary conditions • SAPRC99 • Since lacking of the spatial surrogates in some part area of FAQSD1, SMOKE ran for only FAQSD2 and FAQSD3, and CMAQ did the same Georgia Institute of Technology

9. Mean Bias Error (MBE) • Root Mean Square Error (RMSE) • Mean Normalized Bias (MNB): • Mean Normalized Error (MNE): Comparison: Simulated vs. Observed Georgia Institute of Technology

10. 36-km surface meteorological fields vs. gridded 3-hoursly NCEP ETA surface analysis data corrected by NWS surface observational data Georgia Institute of Technology

11. Georgia Institute of Technology

12. Georgia Institute of Technology

13. Hourly surface meteorological variables simulated at 12-km or 4-km cells vs. observed at NWS stations (26 falling in 12-km domain and 9 in 4-km domain) Georgia Institute of Technology

14. Georgia Institute of Technology

15. Modeled at 12-km cells vs. Observed at NWS Stations Georgia Institute of Technology

16. Diurnal Modeled vs. Observed at NWS #26 Georgia Institute of Technology

17. Hourly ozone values: Simulated at 12-km or 4-km cells vs. Observed at AIRS ozone sites (21 falling in 12-km domain and 18 in 4-km domain) Georgia Institute of Technology

18. Georgia Institute of Technology

19. Georgia Institute of Technology

20. Georgia Institute of Technology

21. Investigation Of the Impact of Urban Definition’s Change • Three different Spatial Surrogates Data sets: • 2000, Using census 2000 data of Urban Areas Definitions, Roads, Populations and Housing • 1990, EPA developed based on census 1990 data, revised Urban areas definitions • USGS, OTAG’s database but with USGS urban definitions and Major Highways • Apply each of them to the emissions spatial allocation in SMOKE. • A new set of temporal profiles used in SMOKE modeling • Apply CMAQ to the same ozone episode with these three different emissions but with same in all others • Redesigned vertical structure as 13 layers with 7 layers in the lowest kilometer • Currently finished only 12-km modeling Georgia Institute of Technology

22. Georgia Institute of Technology

23. Hourly ozone values: Simulated at 12-km cells (with three different spatial surrogates data sets) vs. Observed at AIRS ozone sites Georgia Institute of Technology

24. Georgia Institute of Technology

25. Georgia Institute of Technology

26. Conclusions • First application of CMAQ in FAQS has a good agreement between the 12-km and 4-km simulations and the observations • Considering none coarse domain was simulated and using default initial and boundary conditions • Found disagreements in results • The lack of improvement in simulation with finer spatial resolution may be due to the inaccurate spatial surrogates used in emissions allocation. • The underestimation of NOx emissions may be introduced by insufficient spatial resolution and/or inaccurate spatial allocation of surrogates even if the total emission has been estimated accurately Georgia Institute of Technology

27. Conclusions (continued) • We tried to explain the disagreements by applying three different spatial surrogate data sets to air quality modeling in 12-km resolution • We believe that using more accurate urban definitions in air quality modeling will result in better model results • Unfortunately, we were not able to prove this with the 12-km resolution grid and the limited number of data sites that are available • We expect that when we use 4-km grid resolution the difference between the spatial surrogate data sets will become more obvious • Future work will examine spatial resolution and other factors as possible cause for the disagreements Georgia Institute of Technology