Comment on nhess-2022-52

Abstract

Wildfires pose a significant risk to people and property which is expected to grow with urban expansion into fire-prone landscapes and climate change causing increases in fire extent, severity and frequency. Identifying spatial patterns associated with wildfire activity is important for assessing the potential impacts of wildfires on human life, property and other values. Here, we model the probability of fire ignitions in vegetation across Victoria, Australia to determine the key drivers of human- and lightning-caused wildfire ignitions. In particular, we extend previous research to consider the role fuel moisture has in predicting ignition probability while accounting for environmental and local conditions previously identified as important. We used Random Forests to test the effect of variables measuring infrastructure, topography, climate, fuel and soil moisture, fire history, and local weather conditions to investigate what factors drove ignition probability for human- and lightning-caused ignitions. Human-caused ignitions were predominantly influenced by measures of infrastructure and local weather. Lightning-sourced ignitions were driven by fuel moisture, average annual rainfall and local weather. Both human- and lightning-caused ignitions were influenced by dead fuel moisture with ignitions more likely to occur when dead fuel moisture dropped below 20 %. In future, these models of ignition probability may be used to produce spatial likelihood maps which will improve our models of future wildfire risk and enable land managers to better allocate resources to areas of increased fire risk during the fire season.