A Study on Agent Modeling of Tourist Evacuation Behaviors in an Earthquake: A Case Study of an Evacuation Simulation of Himeji Castle

Abstract

It is an urgent issue to ensure safety of tourists under an earthquake. How do tourists behave under the disaster? This study focuses on tourists’ evacuation behavior who don’t have enough knowledge of evacuation sites and routes. This study shows an agent modeling technique using a sampling survey of the web-based questionnaire considering the information behaviors and the earthquake experiences and a tourist evacuation agent model using this technique. We analyze the tourists’ evacuation behavior using samples of 1086 respondents who made the domestic trip from 20 to 69 years old collected by the web survey in November 2016. The results were summarized into four orientation categories: transport information, safety confirmation, personal safety, and balanced. Next, when two groups, one group with and one without experience of a great earthquake, were compared using four groups, a significant difference was found in any age and sex of each group. Further, this study uses decision tree analysis to model evacuation behavior rules incorporating information behaviors and shows a tourist evacuation simulation using the tourists’ evacuation behavior rules. The results confirm an agent modeling technique using sampling survey and decision tree analysis, and a tourist evacuation behavior agent model incorporating information behaviors by the agent modeling technique is enabled.