Tuesday, 30 January 2024
Hall E (The Baltimore Convention Center)
Wildfires in North America have become more prevalent over the past decades. In 2020, over ten million acres of land were burned from wildfires in the United States, setting many records on both the magnitude and spatial extent of wildfire emissions. These wildfires led to widespread health concerns, including many air quality alerts across the United States. Many organizations have created wildfire emission datasets that quantify the amount of emissions, specifically PM2.5, from these wildfires. There, however, is a degree of disagreement among these datasets. In this study, we compare eleven different datasets (ECCC, EPA, FEER, FINN, FLAMBE, GBBEPX, GFAS, HRRR, QFED, RAP, and RAVE) and create an ensemble that better represents the PM2.5 emissions during the 2020 Gigafire period. The individual datasets as well as the ensemble mean are then used to drive the HYSPLIT model to predict concentration outputs and trajectories to assess the spatial and temporal differences in the modeled PM2.5. The HYSPLIT runs were configured with the Sofiev et al. (2012) plume rise scheme and multiple meteorologies (GDAS, NAM, WRF). First, we compared the 11 different emissions datasets and found that these datasets differ by a factor of 10. All datasets tend to show the same emissions trends over time with peaks occurring near the end of August and beginning of September. Next, we ran HYSPLIT with different emission datasets to simulate the surface PM2.5 concentrations and we found that spatially, the HYSPLIT output using the GBBEPX emission showed the highest concentrations as well as the largest areas of impact by the emissions. We also conducted a statistical analysis including area hit ratio and area false alarm ratio. For the period of August 15 – August 30, 2020, the GBBEPX and QFED datasets had the highest Area Hit Ratio over the entire timeframe, ranging from 55-90% over the time period. RAP tended to have the lowest Area Hit Ratio falling to zero on many occasions. For area false alarm ratio, values for all datasets were high at the beginning of the time frame, but fell to nearly zero around August 19. Between August 19 and August 28, the GFAS dataset consistently had Area False Alarm Ratio values of zero. The EPA dataset also showed low values for Area False Alarm Ratio. This work shows that each dataset tends to have its own strengths and limitations in quantifying the emissions from the 2020 August Complex Fire.

