Design of stochastic trigger boundaries for rural communities evacuating from a wildfire

Abstract

Well-planned evacuation is an effective and often necessary tool to protect life in communities threatened by wildfire. In previous incidents, community evacuations have been called too late resulting in entrapment, such as in Mati, Greece in 2018. The most reliable method to determine the safe time to call an evacuation is through trigger boundaries, perimeters on the landscape surrounding a community where the approaching wildfire is an amount of time away from the community equal to the evacuation time. This paper presents a new tool, k-PERIL, that calculates stochastic trigger boundaries, based on the variability of wildfire behaviour around a community due to the influence of historic wind, weather and vegetation variations on the wildfire. k-PERIL is applied to the rural community of Roxborough Park, Colorado, USA, producing probabilistic trigger boundaries and showing the model's ability to find and quantify areas of elevated uncertainty of evacuation. The concept of uncertainty rosettes is introduced, which show the areas where incoming wildfires cause larger variation to the boundary location because of higher sensitivity to changes in fuel, wind or evacuation. The k-PERIL tool can be used to inform effective evacuation preparation and enhance long term planning, improving community safety and wildfire resilience.