Wildfire Loss Modeling: A Flexible Semiparametric Approach

Abstract

Wildfires have emerged as one of the most devastating natural disasters worldwide, which, in addition to a loss of life, results in billions of dollars in annual losses. This study aims to identify factors that contribute to the frequency and severity of wildfire losses. We propose a semiparametric modeling approach to analyze wildfire property losses in the United States from 1995 to 2020. Our methodology incorporates flexible generalized additive models for location, scale, and shape (GAMLSS), which integrate climate, socioeconomic, and geographical variables into the modeling process. Through empirical analysis, we demonstrate the critical role of spatial effects in modeling wildfire-rela..