DAMAGE CLASSIFICATION OF URBAN AREAS IN THE 2016 KUMAMOTO EARTHQUAKE USING TEXTURE MEASURES FROM ALOS-2 PALSAR-2 IMAGES

摘要

After a natural disaster strikes, it is necessary to assess the amount of damage in vulnerable areas immediately. Synthetic aperture radar (SAR) is independent of time and weather conditions for capturing images of ground surface. In this research, post-event polarized data from ALOS-2 PALSAR-2 with 3.12-m resolution were used to classify the damaged areas in Mashiki town, Kumamoto prefecture, Japan, which was severely affected by the April 14, 2016 (Mw6.2) earthquake and the April 16, 2016 (Mw7.0). Accordingly, the texture measures of the SAR backscatter data set were prepared and used for supervised classification using the Support Vector Machine (SVM) algorithm. This study aims to explore the potential of texture features for detecting damaged regions after earthquakes.