Cluster Analysis and Discriminant Analysis for Determining Post-Earthquake Road Recovery Patterns

Abstract

The transport network in eastern Japan was severely damaged by the 2011 Tohoku earthquake. To understand the road recovery conditions after a large earthquake, a large amount of time is needed to collect information on the extent of the damage and road usage. In our previous study, we applied cluster analysis to analyze the data on driving vehicles in Fukushima prefecture to classify the road recovery conditions among municipalities within the first six months after the earthquake. However, the results of the cluster analysis and relevant factors affecting road recovery from that study were not validated. In this study, we proposed a framework for determining post-earthquake road recovery patterns and validated the cluster analysis results by using discriminant analysis and observing them on a map to identify their common characteristics. In addition, our analysis of objective data reflecting regional characteristics showed that the road recovery conditions were similar according to the topography and the importance of roads.