Relationship between pre-fire climate conditions with summer wildfire burned area highlights the important role of snow drought

NASA satellites have observed a growing trend in wildfire activity in recent decades across the western United States (US), and the Intergovernmental Panel on Climate Change (IPCC) projects this trend will continue with high confidence. It is imperative to understand the relationship between pre-fire (winter and spring) climate and fire season severity to effectively prepare for the severe impacts of wildfire on society: each year hundreds of homes are burned, natural resources are destroyed, and fire suppressions costs can exceed more than one billion dollars. In this study, we leverage a newly developed snow drought index from Huning and AghaKouchak (2020) to create ensembles of machine learning models that relate winter and spring climate conditions to summer wildfire burned area across the western US. Generalized additive models informed with antecedent climate conditions construct a summer burned area time series from 1984-2020 that explains more than 80% of the interannual variability in satellite-observed burned area. In a leave-one-year-out cross validation (i.e., model robustness test), antecedent climate conditions explain more than 55% of interannual wildfire variability. Among 16 winter and spring climate variables that we evaluate, annual snow drought area is arguably the most valuable burned area predictor. These findings advance the understanding of the relationship between drought and wildfire and can be effectively integrated into forecasts that inform resource allocation decisions.