An Analysis of Factors Influencing Disaster Mobility Using Location Data from Smartphones: Case Study of Western Japan Flooding

Soohyun Joo,Takehiro Kashiyama,Yoshihide Sekimoto,Toshikazu Seto,,
DOI: https://doi.org/10.20965/jdr.2019.p0903
2019-09-01
Journal of Disaster Research
Abstract:Western Japan was hit by heavy rain from June 8 to July 28, 2018. Record-breaking rain caused nearly all rivers to flood in Hiroshima and other areas. Over 200 people died following this disaster. Authorities attempted to understand why evacuation was not conducted swiftly enough to stop these deaths. They mentioned that normalcy bias and cognitive dissonance are two primary causes of significant damage [1]. Moreover, an effective alert system is necessary to ensure that evacuation behaviors and procedures are incited at the appropriate time. To understand the factors that influence people’s behavior, we estimated the probability of irregular behavior by unit changes in external condition. We chose 500 m mesh as a unit of analysis to consider individual singularity and classified 3 classes of mesh to identify abnormal behavior. We verified that as the number of residents in each mesh increases, the likelihood of a person in that region to exhibit normalcy bias increases as well. Owing to data, the accuracy of this method is somewhat low. However, several implications may still be drawn from our results, such as the demand for an adequate alert system. Using the results of people’s mobility and disaster risk information, approaches to dangerous situations such as the examined case may be improved in the future.
What problem does this paper attempt to address?