Abstract: The topography along the Jinsha River valley is complex, and the slope toe is seriously eroded by rainwater and river water, which has become one of the hardest hit areas of landslide disasters in China. Due to the difficulty of landslide hazards full identification with single track InSAR observations, a long time-series InSAR two-dimensional deformation extraction model and method was proposed in this paper. Firstly, a time constraint factor was introduced on the basis of SBAS algorithm, combining a spatio-temporal baseline optimization with ascending and descending track solving strategies. Then, being the landslide-prone area along the Jinsha River at the junction of Baiyu and Gongjue counties, a Sentinel-1A ascending and descending imagery of the landslide-prone area from 2018 to 2022 was used to carry out the time-series InSAR deformation solution. Ten landslide hazards and their spatial distribution in the deep-cut canyon area of Jinsha River were identified and obtained. And five hidden dangers of mega-landslides were identified (the monomer exceeded 5010<sup<7</sup< m<sup<3</sup<, and the maximum deformation rate exceeded −70 mm/y), solving more the landslide direction and the amount of each landslide. Finally, seasonal evolution characteristics and long-term development trend of the landslides summarized were summarized based on combination of the precipitation data. The relevant research results can provide important reference data and information support for geological disaster prevention and major infrastructure construction in this area.