Session: 16-01-01: Government Agency Student Poster Competition

Paper Number: 150649

150649 - Optimizing Wildfire Evacuation Warnings: An Agent-Based Approach for Effective Protective Action Recommendations 

Extreme heat due to climate change is a significant factor that facilitates wildfires. According to the National Interagency Fire Center, 72 large active wildfires were raging in the U.S. during the heat wave in July 2024, burning 783,521 acres as of July 14th. Emergency agencies are responsible for providing protective action recommendations (PARs) to the public to help save lives and reduce property loss. However, determining the appropriate PARs is challenging. Issuing PARs at the right time is crucial – they should be neither too late nor too early. In addition, not all the people is required to evacuate. So, the warning notification zone should be neither too wide nor too narrow. To address this problem, this paper investigates the appropriate timing for wildfire evacuation warnings and the warning notification zones through two steps: (1) simulating wildfire spread behavior using the FARSITE model, and (2) integrating the outcomes of step (1) with environmental, traffic, and societal data to feed the Agent-Based Monte Carlo Simulation Model (ABMCSM).

 

This study focuses on the city of Ashland, Oregon, using it as a case study. A small workshop involving Ashland's emergency managers was conducted to design wildfire scenarios. During the workshop, hazard information and geo-data were collected. To simulate the worst-case and likely wildfire spread behavior, scenarios were designed using the FARSITE model and reviewed by Ashland's emergency agencies, whose feedback was incorporated. The confirmed scenarios were then loaded into the ABMCSM, which integrates environmental, traffic, societal, and natural hazard data. A sensitivity analysis via the ABMCSM was conducted to examine the impact of wildfire evacuation notification time, wind speed, and milling time on evacuation efficiency. The results indicate that wind speed significantly influences wildfire evacuation notification times and zones. Additionally, milling time dramatically affects evacuation efficiency, and the evacuation warning time and zones should be adjusted when considering the uncertainty of milling time.

 

This prototype model can be verified by feeding with real-time data from various weather sensors, despite some assumptions about wildfire scenarios, such as ignition point, wind speed and direction, and hourly weather data. It represents a meaningful innovation from an academic perspective, leveraging interdisciplinary knowledge from engineering, social sciences, and natural hazards. From an application perspective, this research provides quantitative evidence to support emergency managers in making informed decisions and issuing PARs. The application is transferable to other communities and other natural hazards research. Furthermore, it enhances the understanding of wildfire evacuation dynamics.

Presenting Author: Chenqiang Liu Oregon State University

Presenting Author Biography: Chenqiang is a third-year Ph.D. Student in the Civil and Construction Engineering program at the Oregon State University (OSU). His research interests are natural hazards evacuation and Agent-Based Model (ABM) simulation. He served as the treasurer of the OSU ITE student chapter during 2022-2023. His hobbies are running and basketball.

Authors:

Chenqiang Liu Oregon State University
Louisa Wildman Oregon State University
Ashley Bosa Boise State University
Brittany Brand Boise State University
Haizhong Wang Oregon State University