Rethinking Wildfire Risk Assessment: An Efficient and Uncertainty-Aware Probabilistic Framework

Wildfire hazard emerges from the interplay of stochastic ignitions, evolving atmospheric conditions, and heterogeneous fuel landscapes, producing large spatial and temporal variability that is rarely captured by currently available risk assessment frameworks. We introduce a probabilistic framework for wildfire risk analysis that treats wildfire losses as spatially distributed random variables and explicitly accounts for uncertainty evolution throughout the wildfire hazard-to-loss continuum. This perspective provides a richer description of wildfire risk beyond single-value risk indicators. To efficiently propagate uncertainty in key drivers such as ignition likelihood, wind conditions, and fuel properties, the framework adopts a deterministic uncertainty propagation strategy based on the Generalized Unscented Transform. This approach captures the nonlinear nature of fire behavior models while avoiding the computational burden associated with generating large Monte Carlo ensembles. The framework is organized in a modular manner, allowing individual hazard, damage and loss components to be coupled consistently while remaining adaptable to alternative data sources, wildfire models, and future climate. An important outcome of the proposed formulation is the derivation of spatially explicit exceedance-rate and hazard curves for wildfire behavior variables, providing probabilistic metrics that are well suited for natural hazards assessment and comparative risk analysis. The methodology is demonstrated using the 2018 Camp Fire in California, where it reproduces observed burn probability patterns and reveals the spatial distribution of exceedance rates for multiple fire behavior indicators with substantial computational efficiency. By emphasizing computational efficiency and systematic uncertainty treatment, this framework contributes to advancing wildfire risk assessment within the natural hazards community and supports novel uncertainty-informed approaches to wildfire hazard mapping and mitigation planning.