Core Time

[P-20] The analysis of tsunami evacuation behavior considering tsunami victim’s data from a case study in Kesennuma city

*Anna Shinka¹, Shosuke Sato², Mizutani Daijiro², Fumihiko Imamura² (1. Graduate School of Engineering, Tohoku University, 2. International Research Institute of Disaster Science)

Keywords:tsunami evacuation behavior, missing data, tsunami victim

Many researches have been conducted to clarify what make people evacuate earlier using questionnaires and interviews. In most of those researches, only tsunami survivors behavior are analyzed but tsunami victim’s behaviors haven’t been considered. However, if we analyze the evacuation behavior excluding data of victims, analyzed evacuation status would be biased. In addition, conducting interview is difficult in some cases and survivor’s impresseion about victims would be different from those of the actual victims. So, complimenting methods of victim’s data are required to consider their actual evacuation behaviors. In this research, we develop a method of complementing victims to estimate their tsunami evacuation behaviors based on the data of the survivor’s. Specifically, the evacuation behaviors are estimated as evacuation start times which vary depending on preliminary preparation and risk recognition for tsunami. To complement the data of victims, we used the hypothesis that preliminary preparation and risk recognition for tsunami of survivors who encountered tsunami and that of victims are same. The propability of early evacuate is empathzed by a logit model and Markov chain Monte Carlo method is used to estimate the parameters in the model to complement the victim’s behavior considering uncertenly. The method of complementating victim’s evacuation behavior is developed and we found that the probability of early evaucation was decreased when we consider about the victims in frequent family meeting.