Suggested Citation:
Grajdura, Sarah and Miguel Jaller (2022)

Mixed Methods Approaches to Wildfire Evacuation: Modeling Behavior, Simulation, and Equity

. Institute of Transportation Studies, University of California, Davis, Dissertation UCD-ITS-RR-22-126

Abstract:

This dissertation presents several aspects of short-notice wildfire evacuation, using empirical findings from the 2018 Camp Fire in Butte County, California. I examine the manner and timing in which people find out about and begin evacuating in a short notice wildfire. Using these findings, I build a simulation model of such a disaster, and examine different worst-case scenarios. Lastly, I use thematic analysis to reveal findings from first-person interviews with fire evacuees. This topic is important due to the prevalence of wildfires in California and the chance of future no/short-notice wildfires occurring in the future. In particular, the Camp Fire was extremely deadly and destructive. It is imperative that I study these large-scale events to improve response and planning. In this dissertation, I rely on data from two post-evacuation surveys as well as interview data taken at post-fire shelters. This unique dataset allows us to answer several questions about this specific event. I use the qualitative findings to add context to our quantitative results. The first paper addresses the timing of awareness, departure, and preparation in short and no-notice wildfire events. Much of the literature has focused on the timing of when people choose to stay at their property, but no literature to our knowledge empirically analyzes awareness and departure in a short or no-notice evacuation. I also analyze the evacuation notice data sent out during the 2018 Camp Fire event. I find that quicker awareness is associated with higher income, smartphone ownership, seeing the fire firsthand, and familiarity with the local evacuation plans. Departure times were delayed for those living in the community longest, among other findings. The second paper addresses how to simulate a short or no-notice wildfire evacuation by building an agent-based model. I use empirical data to inform the timing of when evacuees become notified of the disaster and begin to depart. I use this model to study different worst-case scenario outcomes, namely delayed awareness time, limited smartphone access, and reduced vehicle access. I find that these scenarios lead to longer evacuation times. This model provides a strong basis for future wildfire-related scenario modeling. The final paper shares qualitative interview findings from 26 in-person shelter interviews post Camp Fire. These interviews share information on several areas of evacuee experience from evacuation through a month post-evacuation. By centering accounts from those living in shelters, I gain a new perspective unique to disadvantaged communities. I coded the interviews based on several topics: evacuation, evacuation traffic conditions, fears/problems, financial aid/assistance, finding out about the fire, and shelter/housing.