Source: AGU Advances
Urbanization, climate change, and fire suppression practices are contributing to increased wildfire risk at the densely populated wildland-urban interface. These factors make fires more unpredictable and harder to manage. In January 2025, this was made devastatingly clear in Los Angeles, when massive wildfires engulfed entire hillsides and canyons, destroying neighborhoods and damaging surrounding ecosystems.
The Mediterranean climate region of California, which stretches up most of the state’s coastline, is a naturally fire-prone landscape because its dry conditions support vegetation growth and also allow for fire to spread easily. As wildfires become more intense, better modeling and understanding of their drivers is crucial in efforts to predict risk.
Ward-Baranyay et al. looked at three of the January 2025 Los Angeles wildfires by analyzing preburn conditions, such as fuel characteristics, topography (including elevation and slope), and wind speed. Satellite observations gathered from the Ecosystem Spaceborne Thermal Radiometer Experiment on Space Station (ECOSTRESS) and the Earth Surface Mineral Dust Source Investigation (EMIT)—precursors to a recently announced NASA mission, the Explorer for Artemis Geology Lunar and Earth (EAGLE)—provided detailed information about the vegetation’s condition before the fires began. The researchers then built a random forest regression model to predict burn severity based on these conditions, ultimately demonstrating that prefire fuel conditions were a key driver of the destructive wildfires’ immediate effects on wildlands.
The model used in the study was able to accurately capture about 60% of the patterns in burn severity. It was most accurate for the Palisades and Hughes fires, but less accurate for the Eaton Fire. This discrepancy could be because the area burned by the Eaton Fire was more topographically variable, meaning its burn severity drivers may not have been fully captured by the model, the researchers suggest. Vegetation type was also a strong performance indicator: Terrain with shrub or scrub cover, the dominant vegetation type, offered the most accurate predictions for burn severity. The burn patterns of forests and other landscape types were less accurately captured.
Fuel conditions emerged as the dominant driver of burn severity, more so than topography or weather. In particular, how abundant, wet, dry, or stressed vegetation is can hint at how severe future fires may be. Tracking and monitoring these fuel conditions, researchers suggest, may be a way to monitor wildfire hazard in California and other fire-prone regions. (AGU Advances, https://doi.org/10.1029/2025AV002179, 2026)
—Rebecca Owen (@beccapox.bsky.social), Science Writer
