Early Estimation of Heavy Rain Damage at the Municipal Level Based on Time-Series Analysis of SNS Information

Abstract

To carry out natural disaster response, restoration, and reconstruction, it is important to efficiently and quickly assess the damage caused by the natural disaster. The existing evidence demonstrates that when a natural disaster occurs, social networking services (SNS) information is amplified significantly, compared to normal times. Specifically, the damage caused by a natural disaster tends to cover a wide area and have a large scale. Additionally, it may vary considerably depending on the municipality. Thus, this study investigates whether the utilization of this amplified SNS information can offer an effective approach for real-time evaluation and monitoring of the damage caused by a natural disaster in municipal units. To this end, focusing on time-series changes in SNS information, we propose a general-purpose analysis method of SNS information for evaluating the damage caused by a natural disaster in real time in municipal units. Using real-world data twitter data, we investigate the case of Kumamoto Prefecture, which experienced heavy rain in July 2020 and July 2021, to verify the proposed analysis method.