Abstract -This study introduces a comprehensive simulation model using agent-based modeling (ABM) to assist local governments in developing disaster training scenarios, focusing on earthquake-induced infrastructure damage, sediment disasters, and evacuation behavior. The model comprises three key components: a disaster simulator predicting infrastructure damage, a sediment disaster model replicating damage within warning zones, and an evacuation behavior model representing individual decision-making. Developed with the SOARS toolkit and visualized using Python and ArcGIS Pro, the model effectively captures disaster scenarios. Although infrastructure damage predictions were slightly lower than historical data, they were successfully incorporated into the scenarios. The sediment disaster model accurately reflected trends in the number of affected agents, like past events, and the evacuation model demonstrated consistent behavior patterns across simulations. Additionally, new insights emerged, such as cases where the number of evacuating agents exceeded shelter capacity. Municipalities can employ this model to enhance their disaster response capabilities and conduct more realistic and diverse disaster simulations, ultimately improving the effectiveness of their disaster preparedness training programs. Keywords - Agent-based modeling, disaster simulation, evacuation behavior, SOARS toolkit, disaster tabletop drills
Tsukamoto et al. (2025) studied this question.