Ecological burn severity atlas for forests of the western United States

Abstract

As wildfire activity accelerates in many regions worldwide, quantifying how wildfires affect forest ecosystems is increasingly important. In recent decades, the characterization and tracking of burn severity in forests has advanced thanks to a suite of indices derived from satellite imagery. While these approaches enable wall-to-wall mapping of fire effects across landscapes and regions, the translation of satellite-derived burn severity indices into ecological changes occurring on the ground has not yet been fully realized. To address this challenge, we developed Bayesian models connecting satellite-derived indices of burn severity to a large dataset of detailed on-the-ground measures of burn severity (1,059 field plots measuring tree basal area mortality, tree stem mortality, live canopy cover loss, needle/leaf retention, tree bole charring, and ground surface charring). We incorporated important geospatial covariates (pre-fire forest conditions, climate, topography) that mediated these relationships and improved model performance. Using the best-performing models, we produced an atlas of ecological fire effects across forests of the western United States. For each of 6,771 wildfire events, this atlas includes modeled best estimates of seven on-the-ground burn severity measures, spatially explicit estimates of uncertainty for each measure, and maps of the probability each measure exceeds thresholds typically associated with high-severity fire. By translating unitless satellite-derived burn severity indices into interpretable and field-measurable metrics, this atlas is well-positioned to catalyze new understanding of how fires are changing forest ecosystems. Characterizing and tracking specific ecological effects of wildfire can support better detection of shifting fire regimes and how those changes are driven by changing climate and land management practices, providing strong utility for the research, management, and policy communities.