Using Smart Phone Location Data to Analyze Wildfire Evacuation Behavior

The 2020 Apple Fire in Riverside County California ( images)

The 2020 fire season has been one of the worst in decades destroying millions of acres in the western part of the United States. It has ruined homes, affected the lives of people, and destroyed natural habitats – home to a variety of flora and fauna. But what is the decision process taken by those affected by these wildfires in determining when and where to evacuate? Smartphone technologies may provide some answers.

“Wildfire evacuation is a highly under-studied but pressing issue in the U.S. now,” said Dr. Xilei Zhao, an assistant professor in the Department of Civil & Coastal Engineering at the University of Florida. “To better understand people’s decision-making process during wildfire, we propose to analyze wildfire evacuation behavior by using an emerging and innovative data source such as mobile phone location data.”

Dr. Zhao was recently awarded a grant by the National Institute of Standards and Technology (NIST) to do just that – analyze wildfire evacuation behavior by using data sources such as smart phones. The actual project title is “Analyzing Wildfire Evacuation Behavior with GPS Data”. Specifically, her research team will look at 1) analyzing people’s wildfire evacuation behavior using GPS data; 2) investigating the dynamic patterns of wildfire evacuation across key socio-demographic groups; 3) modeling and forecasting wildfire evacuation demand with GPS data; and 4) identifying important factors (especially those related to fire spread and the land-use) that are associated with wildfire evacuation demand.

The project is expected to generate a set of tools for analyzing GPS data for evacuation, which will be available as an open source on GitHub. The project could help increase the resilience of Wild Urban Interface (WUI) communities, which fall between wildlands and urban development, to future wildfire events. More specifically, the research will provide input for the development of innovative and accurate evacuation modeling tools that can be used to improve evacuation plans for WUI communities in California and across the United States.

The information gained from this study can also be used to create public information campaigns, training protocols, and emergency communication strategies to better prepare WUI households for future wildfires.

“In particular, our analysis will reveal which households to target and can help community leaders generate more refined and personalized pre-and-during event information for dissemination in future wildfire emergencies,” Zhao said. “This research will also act as a case study in identifying event-specific evacuation issues that can be shared with emergency officials in Sonoma County, California, as well as identifying needs to be addressed in future research.”

The research team includes five international experts across different disciplines. The PI, Dr. Xilei Zhao, is a data scientist who specializes in applying big data analytics and machine learning to tackle problems in evacuation and transportation. Co-PI, Dr. Ruggiero (Rino) Lovreglio of Massey University, New Zealand, has demonstrated expertise in fire safety engineering and pedestrian and evacuation dynamics simulation. Co-PI, Dr. Daniel Nilsson of University of Canterbury, New Zealand, works primarily in human decision making, the interaction between people and evacuation systems, and risk management. Co-PI, Dr. Kate Nguyen, RMIT, Australia, specializes in developing new fire engineering technologies to ensure fire safety and protect lives. Dr. Erica Kuligowski, RMIT, Australia, is a sociologist and fire protection engineer, who has an established career in studying human behavior in disasters.