Modeling wildfire spread in wildland-industrial interfaces using dynamic Bayesian network

Global warming and the subsequent increase in the frequency and severity of wildfires demand for specialized risk assessment and management methodologies to cope with the ever-increasing risk of wildfires in wildland-industrial interfaces (WIIs). Wildfires can jeopardize the safety and integrity of industrial plants, and trigger secondary fires and explosions especially in the case of process plants where large inventory of combustible and flammable substances is present. In the present study, by modeling the WII as a two dimensional lattice, we have developed an innovative methodology for modeling and assessing the risk of wildfire spread in WIIs by combining dynamic Bayesian network and wildfire behavior prediction models. The developed methodology models the spatial and temporal spread of fire, based on the most probable path of fire, both in the wildland and in the industrial area.