UA Researchers Develop More Effective Model for Wildfire Evacuation

TUSCALOOSA, Ala. — At the year’s halfway point, the National Interagency Fire Center reported active fires in 10 states, marking the highest number of individual fires in a decade. Some of the more vulnerable homes lie at the intersection of forest and town — the wildland-urban interface. In a study funded by the National Science Foundation, researchers at the University of Alabama developed a new model designed to help these communities develop more effective wildfire evacuation plans.

Living on the (Forest) Edge

Especially in the West, many communities not only occupy this space where buildings mingle with wildfire fuel, but they have widely varying occupancy levels throughout the year. Many homes in towns like Truckee, California are second homes, and the Tahoe Basin sees many visitors in the fire season.

Dr. Dapeng Li, an assistant professor of geographic information science in the Department of Geography and the Environment, led the recently published research. He said these transient populations can pose a special challenge in wildfire evacuation planning. The number of people in town can vary greatly by time of year or even by day of the week.

Assessing the effectiveness of evacuation planning for communities like these has typically focused on total evacuation time. “We found that doesn’t tell the whole story,” Li said. “We also needed to take into account things like individual travel times and how close the evacuation traffic is to the fire front during evacuation.”

In the Zone

Many counties in the American West use wildfire evacuation solutions from companies like Genasys Inc., which assigns homes to predefined evacuation zones. These PEZs help emergency managers evacuate residents in stages so that not everyone is rushing for the exit at once.

Li and his team wanted to learn whether staged evacuation based on PEZs would provide any benefit to WUI communities with high transient populations in these areas. Furthermore, there was no available software model that emergency planners could adapt to their own communities.

To fill these gaps, the UA team developed a model to simulate staged wildfire evacuations based on the use of PEZs. The evacuation model needed to be fully adaptable to either proprietary PEZs or to ones developed by county or state authorities. It also had to be able to adjust to the varying occupancy rates of the second homes in these communities.

“This was a very community-engaged project, and we collaborated with the Town of Truckee to secure the funds from National Science Foundation to work on this project” Li said. “After the 2021 Caldor Fire in the Lake Tahoe Basin, community stakeholders recognized that this was a huge issue.”

The team worked closely with the Town of Truckee to gather data on the seasonal variations in occupancy in Tahoe Donner. Proprietary PEZ data from Genasys, Inc. was used in this model, and other data was gleaned from public databases.

The evacuation model takes into account trip generation, or how many cars leave each home; departure time, based on proximity to the fire and how long people would need to get ready; and the distance to the nearest outlet to safety. They ran simulations for the community with varying occupancy rates of the second homes.

Answers for Community Managers

The team found that a staged evacuation provides a more effective strategy compared with a simultaneous evacuation.

“We found staged evacuation reduces the number of vehicles close to the fire front,” Li said.

This reduces the risk of death and potential harm from smoke inhalation. Staged evacuation also reduced the amount of time spent in traffic for individual vehicles. Hours spent in unmoving traffic can often add stress to an already volatile situation.

“Wildfire evacuations are inherently complex, as they involve rapidly changing conditions,” said Madusha Maha Gamage, a graduate student in UA’s DIGIT Lab and the article’s lead author. “Our staged wildfire evacuation model can be calibrated to any WUI community using local data, and it can be easily adapted to any system that already used predefined wildfire evacuation zones.”