ABSTRACT:

Wildfire burn severity manifests as impacts on above-ground vegetation, relevant to ecological processes, or on soil, including below-ground organic matter, with direct relevance to hydrological and geomorphological processes. While Vegetation Burn Severity (VBS) is typically estimated using remote sensing proxies, Soil Burn Severity (SBS) is generally assessed through expensive, time-consuming manual field measurements. We introduce the Burn Severity of Soil And Vegetation (BurnSAVe) dataset to not only enable descriptive understanding of SBS and VBS interconnections but also facilitate prescriptive models that translate VBS to SBS. BurnSAVe integrates 456 spatially and temporally harmonized attributes describing fire severity, vegetation condition, terrain, climate, weather, soil properties and moisture, and daily fire progression for 532 large fires mainly across the Western United States from 2012 through 2024. We evaluated dataset quality and consistency using multiple complementary approaches, including cross-source comparisons and statistical assessments. BurnSAVe provides analysis-ready, pixel-level information suitable for statistical and machine learning applications, supporting systematic investigation of wildfire severity patterns, fire progression, and post-fire ecosystem responses across broad spatial and temporal scales.