Abstract
Synthetic aperture radar（SAR）installed in an artificial satellite can observe ground conditions even at night and in bad weather conditions when disaster surveying by helicopters is difficult. Using SAR in the Japanese artificial satellite ALOS（Daichi）to detect landslides applies correlation coefficient difference analysis and NDPI（Normalized Difference Polarization Index）difference analysis to the area damaged by the Iwate-Miyagi inland earthquake（Iwate-Miyagi）and sediment-related disasters caused by heavy rain in Hofu city, Yamaguchi Pref.（Yamaguchi disaster）. The results are summarized as follows. The analysis methods have almost identical landslide detection capability. Both analysis methods can detect landslides with size over 10,000 m2 with above 60％accuracy, and over 40,000 m2 with above 80％ accuracy. There may be a minimum detectable landslide size. The size of the Iwate-Miyagi is 900 m2 and Yamaguchi disaster is 2,500 m2. In all analyzed views of SAR satellite images, detection accuracy was better in the Iwate-Miyagi case than the Yamaguchi disaster case. Landslide size is a more effective determinant of capability to detect landslides than slope gradient. The smaller a landslide's size, the more effective slope direction is as a determinant of landslide detection difficulty.
