An Interdisciplinary Agent-based Evacuation Modeling (ABM) Framework: Convergence of Natural Environment, Built Environment, and Social System for Risk Assessment and Improving Life Safety

Keywords: Tsunami, agent-based model, resilience,
Abstract Type: Paper Abstract

Abstract

This study couples FN-curves with Agent-based Modeling and Simulation (ABMS) to assess risk for tsunamis with various recurrence intervals. By considering both expected number of casualties and the likelihood of tsunami events, multiple series of simulations and in-depth analyses determine (1) how vertical evacuation structure (VES) placement impacts mortality rate; (2) what the best evacuation strategies VES locations are; and (3) where evacuees are likely to be caught by tsunami waves. The results from utilizing FN-curves to conduct disaggregative analyses based on six tsunami scenarios indicate that choosing one tsunami scenario or averaging the risk of different scenarios may not fully articulate VES impacts due to the “levee effect,” which potentially leads to false positives. Findings show that placing VESs close to shorelines saves nearby at-risk populations, but also results in two risk increasing phenomena: “exposure to risk” (i.e., evacuees being attracted to high-risk roads by a VES
when evacuating) and “blind zones” (i.e., locations near a VES where evacuees increase their risk by evacuating to that VES). When limited to one VES, placement near a population’s centroid results in the lowest mortality rate. More than one VES may lower mortality rate further if VESs are spreading out according to community’s topography. In addition to the analysis of tsunamis, the approach of coupling FN-curves with ABMS can be used by local authorities and engineers to determine tailored hard-adaptive measures and evacuation strategies, which helps to avoid maladaptive actions in different
hazardous events.