Resident-led agent-based modeling approach for capturing residents’ heterogeneity: toward feasible tsunami evacuation strategies

Agent-Based Modeling (ABM) has been recognized as an effective approach for modeling human responses under anticipated disaster situations. The strength of ABM lies in its ability to represent the heterogeneous attributes, behaviors, and decision-making processes of individual agents, which is essential for identifying feasible countermeasures. This study focuses on residents’ tsunami evacuation behaviors, where such heterogeneity is particularly significant, and explores the direction of ABM approach that integrates the disaster situations residents are concerned about and their corresponding behavioral responses — elements that have been insufficiently captured in conventional expert-driven ABM. Specifically, junior high school students in tsunami-prone areas are positioned as modelers, and an approach is proposed that enables them to independently model and visualize disaster situations and evacuation behaviors of concern to residents. To realize this, three modeling skills are defined: (1) customization of evacuation spaces, (2) design of basic evacuation behaviors, and (3) behavior design based on situational recognition. A support tool was also developed to prepare road network data and implement it in the Scratch environment. An evaluation experiment with seven first-year junior high school students showed that five students successfully acquired the defined skills and created situation-dependent evacuation behavior models. These findings demonstrate the potential for local students to mediate resident-led tsunami evacuation modeling, reducing dependence on external experts and offering a new direction for strengthening community-based disaster preparedness.