Constraining volatile organic compound emissions from western US wildfires with WE-CAN airborne observations

Abstract:

Wildland fires are a significant source of atmospheric volatile organic compounds (VOC), but their emission remains highly uncertain. Here we employ GEOS-Chem nested grid simulations at 0.25° × 0.3125° resolution to interpret WE-CAN (Western wildfire Experiment for Cloud chemistry, Aerosol absorption and Nitrogen) airborne observations in terms of new constraints on the western US VOC emissions from wildfires. Four biomass burning emission inventories, either burned-area-based (the Global Fire Emissions Database version 4 with small fires or GFED4s, and the Fire Inventory from NCAR version 1.5 or FINN) or fire radiative power-based (the Quick Fire Emission Dataset version 2.4 or QFED, and the Global Fire Assimilation System version 1.2 or GFAS), are applied in GEOS-Chem. Our initial results reveal that all inventories predict 29 out of 30 fires sampled by the C-130 aircraft during WE-CAN, and show similar spatial emission patterns with significant differences for individual fires. VOC emission estimates differ by a factor of 1-2 for benzene, toluene and xylenes, 5 for methyl ethyl ketone but more than 17 for propane, formaldehyde, acetaldehyde and acetone over the western US across the inventories, while the difference in total global emissions is smaller (a factor of within 4). Compared to WE-CAN observations, we find the VOC emission ratios relative to CO used in inventories are overestimated by a factor of more than 3 for acetone, methyl ethyl ketone, benzene, and toluene, and a factor of within 2 for formaldehyde, acetaldehyde, propane, and xylenes. We will further evaluate the simulated vertical distribution of VOCs and discuss how the fire injecting heights in the model affect atmospheric chemistry in the western US. Finally, we will compare the total VOC fire emissions in GEOS-Chem with the measured VOCs during WE-CAN and draw conclusions about the ‘missing’ reactive carbon in the current models/inventories.