Spatio-Temporal Analysis for Understanding the Traffic Demand After the 2016 Kumamoto Earthquake Using Mobile Usage Data

This paper focuses on the effect of natural disasters on the population density transition. Emergency management needs demand forecasting by understanding the major demands during a disaster event. Mobile phone traffic is useful for understanding the demands because the behavioral dataset from typical surveys for ordinary demand pattern is impractical. The aim of our spatio-temporal analysis is to identify the characteristics of the population density after a disaster. To deal with this problem, we apply the latest spatial statistic approach to the aggregated mobile phone data before and after a disaster. A regression analysis clarifies the effect of the damage based on the result obtained by this approach. Our case study analyzes the population density before and after the 2016 Kumamoto earthquake. We confirm that this approach can identify the main characteristics: commuting transitions during morning and recreation population on the weekend using the data before the earthquake. The results clearly highlight some interesting spatio-temporal patterns after the earthquake: the recovery process of daily life and time variation of the refugee population density.