Understanding Factors that Contribute to Wildfire Growth in the Western United States

Wildfires in the western United States are increasing in intensity and frequency, with 61% of documented wildfires having occurred since the year 2000 in a record dating back to 1950. Research supports that this trend is largely attributable to climate change and increased urbanization. Consequently, damage to public infrastructure, destruction of private property, and loss of life are increasing. One avenue to protect these assets is to further research wildfires so we can better understand how to mitigate fire hazard and predict hazardous periods and events. For example, this includes improving wildfire behavior forecasting during active wildfire events, insights on optimal prescribed burns, and estimation of the impacts of climate change and urbanization. The overarching goal of this study was to evaluate wildfire climatology by addressing two key questions: (1) which meteorological and vegetative factors are most correlated with annual burn area in the western U.S., and (2) can annual burn area be modeled with machine learning algorithms? These questions have been investigated by comparing 13 seasonally averaged fire covariates (e.g., vapor pressure deficit, surface air temperature, relative humidity, fuel moisture, etc.) with satellite observed western U.S. fire events from 1984-2022. We determined vapor pressure deficit to be most positively correlated covariate with annual burn area, while 100- and 1,000-hour fuel moistures were most negatively correlated. The random forest regression model consistently underpredicted the annual burn area. Future work will explore ways to improve the machine learning model so it is more representative of actual conditions.