Fire management decision support
The modelling framework presented here has several features that are useful at regional scale for fire managers. It can incorporate long-term field-based fuel hazard surveys. It can use a comprehensive selection of predictors including climate, terrain, soil and vegetation. It can produce output that is interpretable and useful to fire managers and at a resolution that is relevant for operational management (Penman et al., 2022). Despite the high resolution, it has relatively low computational demand for regional projections (i.e., does not require high-performance computing). The comprehensiveness separated this approach from previous empirical models that only consider a subset of predictors (e.g., Pierce et al., 2012; Jenkins et al., 2017; McColl‐Gausden et al., 2020) while the high resolution (90 m) is what current process-based modelling cannot provide (Rabin et al., 2017).
The random forest model predicted increased P4_5 in the future for areas currently under semiarid climates (Figure 4). This prediction is consistent with a previous analysis on historical burnt area and aridity (Kelley et al., 2019). Combining the finding with current knowledge on key limiting factors in different regions (Archibald et al., 2013; Bradstock et al., 2014; Boer et al., 2016), our predictions indicate changes in future fire regime as current semiarid regions showing increasing fuel load in elevated stratum. Specific fire management measures should be thus based on the actual fuel hazard and its change in all fuel strata (Figure 3; Figure S4 and S5).