Discriminant Analysis for Predicting Post-Earthquake Municipal Road Recovery

Abstract

In the event of a widespread natural disaster, the speed of recovery is greatly affected by the priority given to road recovery, not only in terms of assessing the extent of the damage but also in terms of providing rapid assistance to the affected areas and carrying out recovery activities. The purpose of this study is to simulate the recovery of local roads in the Tokai region (Shizuoka, Aichi, and Mie prefectures) in Japan after the Nankai Trough earthquake, which is expected to occur in the future. From the previous analysis of driving data in the Tohoku region (Fukushima, Miyagi, and Iwate prefectures) in Japan during the first six months of the 2011 Tohoku earthquake, we discovered regional characteristics and other factors that affect road recovery. In this study, we continued to analyze and quantify the influencing factors in each pattern of road recovery and applied discriminant analysis to construct a predictive road recovery model. The model showed an accuracy rate of 72.4%. In addition, we used this model to predict municipal road recovery in the Tokai region after the Nankai Trough earthquake in the future and mapped it on a geographic information system (GIS).