Incorporating fine-scale drought information into an eastern US wildfire hazard model

Abstract

Wildfires in the eastern United States are generally caused by humans in locations where human development and natural vegetation intermingle, e.g. the wildland–urban interface (WUI). Knowing where wildfire hazards are elevated across the forested landscape may help land managers and property owners plan or allocate resources for potential wildfire threats. In an earlier paper, we presented a model showing monthly hazards of wildfire across Ohio, Pennsylvania and New Jersey at a 30-m resolution. Here, we refine the spatial resolution of drought conditions by incorporating a 4×4-km gridded self-calibrated drought index, include mean winter temperature to restrict hazards from being modelled into locations possibly covered by snow, and compare the performance of the updated models with the original ones. The area under the curve values for the updated models were within 10% of the values for the original models, but the refinement of drought conditions resulted in a less generalised probability of hazards, potentially increasing the applicability of these models. Among the 12 monthly models, the wildland–urban interface had the highest contribution followed by a weighed drought frequency index.