Agent-based modeling for the prediction of evacuation behavior during flooding based on EDFT model

Natural disasters, particularly floods, pose significant challenges to urban areas, affecting both infrastructure and human lives. This paper presents a comprehensive simulation model aimed at predicting evacuation behaviors in response to floods. Utilizing an agent-based modeling (ABM) approach, the study simulates human decision-making processes during flood events, specifically focusing on the city of Mantova, Italy. The model integrates dynamic environmental parameters such as flood strength, timing, and proximity, as well as demographic factors, to predict evacuation decisions. Furthermore, the Extended Decision Field Theory (EDFT) is employed to capture the diversity and unpredictability of human behaviors in disaster scenarios. The simulation also incorporates real-time flood data, using GIS tools to assess building damage based on water depth, building characteristics, and repair costs. For the first time, this model is applied to flooding scenarios, offering valuable insights into evacuation dynamics and flood impact. Additional parameters are incorporated to extend the model, making it adaptable to various flood scenarios. This approach is crucial for understanding how many people choose to evacuate, ultimately aiding in disaster preparedness and saving human lives. The model's results are validated against survey data from local residents, providing insights into the effectiveness of evacuation strategies under varying flood return periods. This study highlights the potential of ABM in enhancing disaster resilience through predictive evacuation modeling and damage assessment.

Keywords:

Dynamic-data-driven modeling , Evacuation behaviors , flooding , EDFT , Disaster management , Agent-based simulation