An overdispersed Poisson model for forest fires occurrences in Southern Italian municipalities

In recent years, the number and the magnitude of wildfires are constantly growing in southern EU countries due to extreme climate conditions. This study proposes a modeling approach to investigate the relation between fire occurrence and several potential socio-economic and environmental driven factors considering two neighboring regions in southern Italy (Apulia and Basilicata). Multiple sources of data with different spatial support are used and data were preprocessed in order to reconduct the analysis to the municipality scale. A Bayesian zero-inflated Poisson model with spatial component is proposed to accommodate the excess of zeros in the counts and to account for the neighboring structure between municipalities. Preliminary results suggest the appropriateness of such approach with some insights explaining the dependence relation.