Photochemical Modeling of the Ozarks Isoprene Experiment (OZIE): Comparison of MEGAN and BEIS

1. Photochemical Modeling of the Ozarks Isoprene Experiment (OZIE): Comparison of MEGAN and BEIS Kirk Baker and AnnmarieCarlton U.S. Environmental Protection Agency October 12, 2010

2. Biogenic VOCs • Biogenic VOC comprise approximately 75-80% of the North American annual 2005 NEI • Biogenic emissions largely a function of plant type, leaf area index, temperature, and solar radiation (PAR) • Biogenic emissions for regional and global scale photochemical are typically developed from either MEGAN or BEIS • Model of Emissions of Gases and Aerosols from Nature (MEGAN), currently developed by NCAR • Biogenics Emission Inventory System (BEIS), currently developed by US EPA • July 1998 OZarks Isoprene Experiment (OZIE) field study designed for evaluation of biogenic models

3. MOTIVATION • Compare MEGAN and BEIS emission estimates to surface and upper air measurements from OZIE field study • Evaluate sensitivity of CMAQ predictions when using alternative biogenic emissions models for • Primary gas phase species: e.g., isoprene, monoterpenes, sesquiterpenes • secondary species: e.g., O3, PM2.5, SOA • Primary and secondary: formaldehyde (HCHO) • Predictions of BVOCs are highly dependent on meteorology e.g., temperature, solar radiation • What is the bias in met model predictions of these variables? • How do predictions differ when alternative source of solar radiation are employed, e.g., how photosensitive are MEGAN and BEIS emission estimates?

4. Ozarks Isoprene Experiment: July 1998

5. OZIE Project Details • Surface • TNMOC, isoprene, a-pinene, b-pinene, HCHO, O3 • Solar radiation, soil temperature, wind speed, wind direction, temperature, relative humidity • Aloft • Balloon: Isoprene, a-pinene, b-pinene • Aircraft (7 flights):, O3, isoprene, TNMOC, HCHO Other Available Observation Data: • AIRS (1-hr avg) • Ozone, NO2/NOX, PM10, CO • CASTNET/IMPROVE (24-hr avg) • Speciated PM2.5 • FSL RAOB Springfield, MO • Temperature, wind vector • U.S. Airways (DS472) (Hourly reported) • Temperature, mixing ratio, wind speed, wind direction

6. CMAQ Modeling Overview • Modeled June 15-July 31, 1998 episode • CMAQ v4.7.1 (N2a) • CB05 & AERO5 • WRF v3.1 • MCIP v3.4.1.1 • BEIS v3.14 • MEGAN v2.04 • 2001 v2 based anthropogenic emissions • SMOKE inline emissions • 36 km – continental US • 12 km (blue) • 34 and 14 vertical layers

7. Biogenic Modeling • BEIS v3.14 and MEGAN v2.04 • WRF 2m Temp. and shortwave downward radiation (98a) • WRF 2m Temp. and satellite estimated photosynthetically activated radiation (PAR) (98b) • PAR is the visible light fraction of shortwave downward radiation • Satellite PAR estimates taken from GEWEX Continental Scale International Project (GCIP) and Americas Prediction Project (GAPP) Surface Radiation Budget (SRB) [http://www.atmos.umd.edu/~srb/gcip/cgi-bin/historic.cgi] • Satellite PAR resolution 0.5◦ x 0.5◦ and covers the continental U.S. • Missing hours and days replaced by hourly monthly average satellite estimated PAR

8. Monthly domain (12OZIE1) total emissions 98a: WRF PAR 98b: satellite PAR Moles/Day/10000 Moles/Day/10000 98a 98b 98a 98b 98a 98b 98a 98b BEIS BEIS MEGAN MEGAN

9. Isoprene Emissions BEIS: Isoprene Emissions (moles/sec) MEGAN: Isoprene Hour of the day (LST)

10. Monoterpene Emissions BEIS: Monoterpenes Emissions (moles/sec) MEGAN: Monoterpenes Hour of the day (LST)

11. Sesquiterpene Emissions BEIS: Sesquiterpenes Emissions (moles/sec) MEGAN: Sesquiterpenes Hour of the day (LST)

12. Temperature & PAR at OZIE Sites Temperature Bias: OZIE sites PAR Bias: OZIE sites • Warm bias over all hours & days • Morning and early afternoon PAR underestimated • Satellite PAR estimates compare better to ground observations Watts/m2 Temperature (C) Hour of the day (LST) Hour of the day (LST)

13. Isoprene Performance12km grid, 34 vertical layers Modeled (ppbC) Modeled (ppbC) Measured (ppbC) Measured (ppbC)

14. Isoprene Bias: July Episode BEIS: Isoprene MEGAN: Isoprene ppbC

15. Monoterpene* Performance Modeled “TERP” (ppbC) Modeled “TERP” (ppbC) Measured (ppbC) Measured (ppbC) α-pinene + β-pinene α-pinene + β-pinene α-pinene + β-pinene are approx. 50-70% of total monoterpenes (Sakulyanontvittaya et al, 2008). There should be an overprediction.

16. Surface Formaldehyde Performance Modeled (ppbC) Measured (ppbC)

17. Balloon Measurements: Isoprene • Balloon: isoprene (right) • Model estimates (using satellite PAR) compared to vertical balloon and aircraft measurements • Measurements made in an open field Model Layer Concentration (ppbC)

18. Aircraft Measurements Isoprene Formaldehyde Ozone Model Layer Concentration (ppbC) Concentration (ppb) Concentration (ppb) • measurements made over tree canopy

19. Model Performance: Ozone • MEGAN predicts slightly higher regional O3 • Satellite PAR MEGAN estimates result in slightly less O3 than using WRF PAR Bias (ppb) Observed Ozone Bin (ppb)

20. Model Performance: speciated PM2.5 • Similar PM2.5 performance using BEIS and MEGAN at rural speciated monitors • Sulfate performance related to SO2 controls between 1998 and 2001 • More OC using MEGAN emissions Concentration (μg/m3) 98a 98b 98a 98b OBS BEIS MEGAN

21. Daily CMAQ PM2.5 SOC Estimates µg/m3 µg/m3 *semi-emipirical EC tracer method to estimate SOC (Yu et al 2007)

22. Conclusions and Future Directions • MEGAN BVOC estimates are substantially higher and more photosensitive than BEIS • CMAQ O3and PM2.5 estimates are similar with BEIS or MEGAN • Accept either for SIP modeling based on current state of chemical mechanisms • Future modeling study to investigate if control strategies would differ • Simulations with process analysis to investigate importance of VOCs, oVOCs • Test with different chemical mechanisms, e.g, low NOxisoprene chemistry • SOC predictions much higher with MEGAN but rural SOC still substantially underpredicted • Field Study to re-visit OZIE • Have BVOC emissions/ambient mixing ratios changed in response to changes in emissions and/or climate? • Measure SOA tracers to investigate which individual precursors/pathways contribute most to the SOC underprediction

23. Acknowledgements • Allan Beidler, James Beidler, and Chris Allen (CSC) for BEIS, MEGAN, and SMOKE emissions modeling • Lara Reynolds (CSC) for WRF application • Rob Gilliam (US EPA/ORD) for assistance and troubleshooting for 4 km WRF application • Marc Houyoux and Rich Mason (US EPA/OAQPS) for assistance with anthropogenic emissions inventory • Christine Wiedenmyer (NCAR) for observation data • Mike Koerber (LADCO) for observation data • George Pouliot and Tom Pierce (US EPA/ORD)

24. END

25. Future Field Study Directions What/where would a new “OZIE” experiment include? • Ozarks • Surface and aloft measurements • Isoprene, monoterpenes, and sesquiterpenes • SOC tracer measurements • OH radical • Speciated PM2.5; Ozarks get a lot of acidic aerosol in the summer from the Ohio Valley region • Carbon dioxide, NOX (is this a low NOX environment?) • Summer and non-summer periods like winter/fall/spring? • Roving measurements to different land use types like crops and grasslands? Urban biogenic assessment?

26. Landuse Information • MEGAN uses MODIS landuse and plant species information to make gridded emission factors • BEIS uses BELD3, a combination of plant species information and USGS landuse • Different isoprene emission factors for high emitting species (top right) and USGS deciduous tree category (bottom right) in BEIS • Difficult to differentiate plant coverage and emissions with MEGAN’s gridded emission factor product (one plant functional type shown below) BELD3: Oak+Others MEGAN: Broadleaf Trees BELD3: USGS Deciduous Trees *Others = Sycamore+Sweetgum+Willow+Populus

27. Nitric Oxide Emissions BEIS: Nitric Oxide • Biogenic nitric oxide (NO) estimated with BEIS3.12 comprises approximately 9% of the total continental United States 2005 annual NO emission inventory MEGAN: Nitric Oxide