Estimation of Hourly 3km Biomass Burning Emissions by Fusing 2km ABI and 375m VIIRS Fire Observations

Biomass burning from wildfires is an important source of atmospheric trace gases and aerosols and the emitted smoke significantly degrades regional air quality. Accurate and timely estimates of fire emissions are critical to air quality forecast models that require high temporal resolution (e.g., hourly) emissions. However, most existing emissions inventories provide daily emissions at a coarse resolution (e.g., 0.1°, 0.25°, and so on). Hourly emissions are obtained usually by distributing daily emissions to predefined diurnal cycles, which has large uncertainties. To improve air quality forecast, we developed an algorithm to estimate hourly fire emissions at 3km by fusing 5-minute 2km fire radiative power (FRP) from GOES-16 and GOES-17 ABI (Advanced Baseline Imager) and daily 375m FRP from Suomi NPP and NOAA-20 VIIRS (the Visible Infrared Imaging Radiometer Suite) across the Conterminous United States (CONUS). Specifically, ABI and VIIRS FRP are first stratified and aggregated separately by five ecosystems (i.e., forest, shrubland, savannas, grassland, and cropland) in 3km grids. The 3km ABI FRP is then calibrated against 3km VIIRS FRP, which is to mitigate the underestimation of ABI FRP due to omission of small and/or cool fires relative to VIIRS FRP. Fused the ecoregion and ecosystem specific FRP diurnal climatology, the calibrated ABI+VIIRS FRP is used to reconstruct FRP diurnal cycles. Finally, hourly 3km biomass consumption and emissions are calculated using the reconstructed FRP diurnal cycles for one year from April 2020 to March 2021. The preliminary results show that fires consumed 222 Tg dry biomass, with 56%, 9%, 13%, 10%, and 12% contributed by fires in forest, shrubland, savannas, grassland, and cropland, respectively. Accordingly, fires emitted a total of 361 Tg CO2, 18 Tg CO, and 2.3 Tg PM2.5 emissions. Seasonal emissions indicate that the maximum monthly emissions occurred in September 2020 when record-breaking wildfires raged in western coastal States, and explained 42% of annual emissions. Diurnal variation reveals that emissions were emitted mainly during daytime, generally peaking around 3:00 PM, while nighttime emissions were also significant in forest fires. This algorithm is expected to operationally generate hourly 3km emissions products for air quality forecast models.