Fig. 4: California data-driven fire prediction for 2025.

Five days forecast leading to the 7 January 2025 when major fire outbreaks affected the Los Angeles metropolitan area. The data-driven model including all information of fuel status shows better localization of areas where fires ignited. California experienced two unusually wet spring seasons leading up to the 2025 fires, resulting in an increased accumulation of fuel in the region. These anomalies led to a unique pattern of heightened flammability at the wildland–urban interface. The data-driven model successfully identified this pattern by accounting for the abnormal fuel conditions.