A data-driven approach to improving evacuation time estimates during wildfires for communities with part-time residents in the wildland-urban interface

Abstract

Wildfires pose a significant threat to the residents living in the wildland-urban interface. Computerized modeling of wildfire evacuation could facilitate protective action decision-making and improve wildfire public safety. This study aims to leverage different types of data, traffic simulation model, and geographic information systems to develop a data-driven wildfire evacuation model to improve evacuation time estimates in resort areas. Specifically, we take into account household vehicle ownership data and the occupancy rate of second homes based on a variety of data in model construction. We used the Tahoe Donner neighborhood in Truckee, California in the case study and derived a series of evacuation time estimates. The results indicate that the evacuation time estimates vary significantly with the mean number of vehicles per home and second homes' occupancy rate in resort areas. The proposed method could help incident commanders better understand the dynamics of travel demand of the fire-prone communities with part-time residents during wildfire evacuation and increase their situational awareness.