Modeling evacuation decisions in the 2019 Kincade fire in California

Communities around the world are increasingly exposed to larger and more intense wildfires. A common method that officials use to protect community members from harm is evacuation. Data on how people behave during wildfires is critical when planning for evacuation and deciding when and how to evacuate entire communities during an event. Using a similar method to the 2016 Chimney Tops 2 Fire study, an online survey was conducted with households impacted by the 2019 Kincade fire in Sonoma County, California. The survey measured pre-event and event-based factors to 1) predict household perceptions and evacuation decisions and 2) compare results across fire events. Regression analysis identified the factors that influenced risk perception at the time of evacuation decision, i.e., pre-fire perceptions of safety, household makeup (of adults, pets, and livestock), income, education and threat assessment. Logistic regression analyses found that risk perception, length of residence, household makeup, income, education, evacuation order, fire cues, pre-fire perceptions of the safety, and homeownership influenced evacuation decisions. These results differ widely from the 2016 fire due to differences in fire conditions and experiences across populations. Results from this work bring the field closer to a generalized theory of human behavior during wildfire evacuation and improve community-wide evacuation planning and real-time decision-making.