Peak 8-hr Ozone Model Performance when using Biogenic VOC estimated by MEGAN and BIOME (BEIS) Kirk Baker Lake Michigan Air Directors Consortium October 2007 – CMAS Conference
Background Biogenic emissions are a large contributor of regional VOC in the central and eastern United States that participate in photochemical reactions which form ozone It is important to capture the magnitude and spatial scale of biogenic VOC emissions, especially isoprene, to appropriately model high ozone episodes in the Midwest United States.
Biogenic Emissions Models 1.The Model of Emissions of Gases and Aerosols from Nature (MEGAN) was recently developed as the next generation emission model for biogenic emissions of gases and aerosols 1.MEGAN has been implemented into the CONsolidated Community Emissions Processing Tool (CONCEPT) emissions modeling framework 2.The CONCEPT emissions framework also includes the biogenic emissions model BIOME, which is a combination of BEIS3 and GloBEIS methodologies 3.A slightly different version of BIOME was implemented in the EMS-2003 emissions model
Biogenic Model Inputs The BELD3 land use dataset is input to the BIOME biogenic models for fractional land-use and vegetative speciation information Inputs to MEGAN include plant functional type emission factors, PFT area coverage, soil wilting point data, leaf area index, and additional meteorological variables including soil moisture. Soil moisture estimated by MM5 for the 1 m soil depth is used as input to MEGAN because it represents the plant root layer Other inputs to all 3 biogenic models include hourly satellite photosynthetically activated radiation (PAR) and 15 m (AGL) temperature data output from MM5
Satellite Estimates of PAR The approach is based on a physical inference scheme to derive surface spectral radiatiative components from satellite observations at the top of atmosphere A modified version of the Global Energy and Water Cycle Experiment (GEWEX) /Surface Radiation Budget (SRB) model (Whitlock et al., 1995; Pinker et al., 1995)
PAR SURFRAD sites used: Sioux Falls Bondville Goodwin Creek PSU Comparison for entire year of satellite PAR with surface measurements
Daily total isoprene emissions (tpd) for 2002
Total daily emissions (tpd) by model
Total emissions for July 1, 2002 estimated by MEGAN (top), EMS/BIOME (middle) and CONCEPT/BIOME (bottom)
Photochemical Modeling 36 km (large box) and 12 km (small dark box) modeling domain CAMx version 4.5 MM5v3.6.x CB05 gas phase chemistry Anthropogenic emissions based on 2002 State submitted inventories
Results When modeled with a photochemical transport model, each set of biogenic emissions result in a similar spatial pattern of peak ozone formation The MEGAN emissions have the highest ozone peaks of the 3 sets of biogenic emissions The 99th percentile daily maximum 8-hourly average ozone observation for each monitor over each summer is paired with model estimates The mean bias of 99th percentile 8-hourly average maximum ozone over all stations (N=303) in the 12 km domain is ppb with MEGAN biogenics, with CONCEPT/BIOME biogenics, and with EMS/BIOME biogenics The isoprene estimated by MEGAN resulted in ozone estimates closest to peak observations over the entire 12 km modeling domain.
Daily average mean gross error for all 3 simulations for 8-hr ozone > 80 ppb
MNGE (%) for O3; difference between simulations (megan – ems/biome)
Conclusions The MEGAN isoprene emissions improve model estimates of high ozone in the upper Midwest compared to 2 implementations of the BIOME model In general, each of the 3 biogenic emissions estimates resulted in similar spatial patterns of ozone formation in the region which suggests that ozone performance is more closely related to anthropogenic emissions and meteorology Future work will include a comparison of photochemical model estimated secondary organic aerosol using emissions from MEGAN and BIOME
Organic Carbon Mass Underprediction
Daily Emission Totals of PM Pre-cursors MonoterpenesSesquiterpenes
MEGAN v1 and v2 Emission Factors