Evolution of deep-seated bedrock landslides in a deep canyon

Deep-seated bedrock landslides are characterized by large scales and long runout distances, frequently resulting in the formation of dammed lakes within deep canyons and causing serious damage to downstream residential areas, ecological and hydrological environment. However, deep-seated bedrock landslides are usually located in steep and inaccessible mountain area, posing a significant challenge to their effective observation and failure warnings. In this study, we utilized optical and synthetic aperture radar (SAR) satellite remote sensing, unmanned aerial vehicle (UAV) mapping, and field investigation to analyze three closely located deep-seated bedrock landslides (Baige, Xiaomojiu, and Guili) at different evolution stages along the Jinsha river in the eastern Tibetan Plateau. We analyzed the geological structures of these landslides and conducted evolution analysis to explore the potential ability of remote sensing technology in failure warning of landslide. The results indicate that the three landslides are situated within the Jinsha suture zone and are composed of broken gneiss, serpentinite, and mylonit. The complex materials and structure resulted in differential deformation along the slope, manifested by varying density of tensile cracks, size of fractured zone, and cumulative displacements in different parts of the slope. In the 20 years before the 2018 Baige landslide event, the coverage area of fractured zone changes from 0.09 km² to 0.38 km², accounting for 43.6 % of the total landslide area. Deformation data from SAR images (2017–2021) reveal periodic patterns during landslide evolution, though the influence of seasonal flood level variations appears limited. The periodic deformation infers that deep-seated bedrock landslide failures may be related to precipitation infiltration and slope erosion. In conclusion, the evolution of deep-seated bedrock landslides is a prolonged surface process, and multi-source remote sensing provide effective solution for detecting surface damage and deformation at various stages of evolution.