Trial of Detection for Landslide Areas from Wide Range of Image Obtained by L-band SAR (PALSAR-2)

Abstract

Data from the Phased Array type L-band Synthetic Aperture Radar 2 (PALSAR-2), an L-band SAR on board the Advanced Land Observing Satellite-2 (ALOS-2), were used to detect landslides caused by the 2016 Kumamoto earthquakes. Differences in α angle and HH-VV coherence obtained before and after the disaster were used. Some landslides occurred in a forested area before the disasters were detected by these parameters. Misidentifications were prominently observed. A forest mask was applied to the image taken before the disaster, and it was confirmed that the forest mask worked well to reduce misidentification of landslide areas. Detection rates are estimated to be 63%. The main reasons for failures of detection were radar shadow, and landslide areas consisting of either a vacant piece of land or land with low vegetation before the disaster.