Abstract In January 2025, some of the most destructive wildfires in California's history ignited across densely populated wildland‐urban interface regions in Los Angeles County. Given the increasing risk of intense wildfires and growth of wildland urban interface communities, improved characterization and monitoring of the drivers of high‐severity fire is essential for proactive wildfire management. Random forest modeling of biophysical variables has shown that fuels, fire weather, and topography influence burn severity, with fuels conditions such as abundance, moisture, and water stress often emerging as dominant drivers. However, the joint integration of fuels conditions derived from both spectroscopic visible‐to‐shortwave infrared and multispectral thermal infrared observations are largely unexplored for assessing and monitoring burn severity hazard. We demonstrate an empirical approach to modeling wildfire burn severity using interactions between fuels conditions, characterized jointly by the novel Ecosystem Spaceborne Thermal Radiometer on Space Station (ECOSTRESS) and Earth Surface Mineral Dust Source Investigation (EMIT), fire weather, and topography. With a single, seven‐feature random forest regression model for three of the January 2025 Los Angeles wildfires, we predicted burn severity ( R 2 = 0.60, RMSE = 0.12) using ecologically meaningful and uniquely contributing observations and determined that pre‐fire fuels conditions were key drivers of burn severity. Overall model stability was strong as indicated by Monte Carlo validation ( R 2 = 0.60, standard deviation = 0.024); however, model performance and behavior varied across vegetation categories and fires. Our results are particularly relevant for the spatiotemporal assessment of wildfire burn severity drivers, providing insights to support proactive land management decisions.
Ward‐Baranyay et al. (2026) studied this question.