Regional landslide identification using the slope unit containing deformation information

Using Interferometric Synthetic Aperture Radar (InSAR) to identify potential landslides on a regional scale requires defining their boundary ranges and shape characteristics. However, InSAR data in grid form can only delineate the deformation area within a slope and cannot identify its boundary and geometric features. To address this issue, a morphological image analysis and homogeneous slope unit-based method was adopted to extract slope units. Using Geographic Information System spatial analysis technology, the slope unit (surface vector) and InSAR (raster data) datasets were superimposed (hereinafter referred to as SU_InSAR). This can delineate specific slope boundaries for InSAR data scattered on a regional scale, enhancing readability and yielding a slope unit which contains deformation information for the identification of potential landslides. SU_InSAR was used to identify potential landslides in Minjiang River basin in Aba Prefecture, Sichuan Province, China. The results derived from the new mapping unit were compared to those of a hydrological-mechanical model using the Receiver Operating Characteristic evaluation method. The comparison results showed that the accuracy of the SU_ InSAR-based model was 24.4% higher than that of the hydrologic-mechanical model, which is expected to improve landslide mitigation.