A Bayesian approach to exploring the influence of climate variability modes on fire weather conditions and lightning-ignited wildfires

摘要: Understanding the relationships between large-scale, low-frequency climate variability modes, fire weather conditions and lighting-ignited wildfires has implications for fire-weather prediction, management conservation. This article proposes a Bayesian network framework quantifying influence of modes on occurrence lightning-ignited wildfires. The main objectives are to describe demonstrate probabilistic identifying joint individual that comprise climate-wildfire system; gain insight into potential causal mechanisms pathways; gauge lightning-ignition relative local-scale alone; assess predictive skill network; motivate use techniques intuitive, flexible which user‐friendly software is freely available. A case study illustrates application forested region in southwest Australia. Indices six considered along with two hazard variables (observed prescribed burn area), 41-year record wildfire counts. Using data set, we proposed framework: (1) based reasonable assumptions provided density converted multivariate normal; (2) generates parsimonious interpretable architecture; (3) identifies known or partially variables; (4) be used setting conditions; (5) more directly related than