Simulating burn severity maps at 30 meters in two forested regions in California

Abstract Climate change is altering wildfire and vegetation regimes in California’s forested ecosystems. Present day fires are seeing an increase high burn severity area patch size. The ability to predict future patterns could better support policy land management decisions. Here we demonstrate a methodology first, statistically estimate individual classes at 30 meters second, cluster smooth patches onto known landscape. Our goal here was not exactly replicate observed maps, but rather utilize maps as one realization of random process dependent on climate, topography, fire weather, fuels, inform creation additional realizations through our simulation technique. We developed two sets empirical models with different datasets test if coarse accurately model for severity. While visual acuity can be used assess the performance process, also employ Ripley’s K function compare spatial point processes scales capturing appropriate amount clustering. FRAGSTATS obtain metrics contiguity simulation. helped identify number clustering iterations showed how focal window sizes affected patches. Improving simulate may help advance understanding potential influence fuels ecosystem-level response variables that important decision-makers. Simulated managing habitat estimating risks loss, protecting infrastructure homes, improving emissions projections, mapping planning treatment scenarios.