Using a statistical model of past wildfire spread to quantify and map the likelihood of fire reaching assets and prioritise fuel treatments

Abstract

We present a method to quantify and map the probability of fires reaching the vicinity of assets in a wildfire-prone region, by extending a statistical fire spread model developed on historical fire patterns in the Sydney region, Australia. It calculates the mean probability of fire spreading along sample lines around assets, weights the probability according to ignition probability and also estimates the change in spread probability that fuel reduction in treatment blocks would achieve. We have developed an R package WildfireRisk to implement the analysis and demonstrate it with two case studies in forested eastern Australia. The probability of a fire reaching the vicinity of an asset was highest in the heavily forested parts of each case study, but when weighted for ignition probability, the high probability shifted to the wildland–urban interface. Further, when weighted by asset location, high-priority areas for treatment were in blocks next to the wildland–urban interface. This method is objective, fast and based on the behaviour of real historical fires. We recommend its use in wildfire risk planning, as an adjunct to heuristic methods and simulations. Additional functionality can be incorporated into our method, for instance via a function for building impact.