Objectives On December 18, 2023, an Ms 6.2 earthquake occurred in Jishishan County, Gansu Province, China. The earthquake caused more than 150 deaths and extensive collapse of buildings, and triggered a mass of coseismic landslides. In addition, the earthquake will also accelerate the deformation of active landslides, seriously threatening the safety of people's lives and property and infrastructure. Therefore, it is imperative to carry out research on the rapid identification and dynamic deformation monitoring of active landslides in earthquake-affected areas.

Methods We propose a novel framework for the automatic identification and dynamic deformation monitoring of active landslides in earthquake areas based on interferometric synthetic aperture radar (InSAR) technology. First, the InSAR phase-gradient rate, deformation rate and time series are calculated by utilizing both the ascending and descending Sentinel-1 synthetic aperture radar (SAR) images acquired during March 2017 to December 2023; and then, an approach for the automatic identification of active landslides is established using DeepLabV3 deep learning algorithm. As a result, the inventory map of active landslides in the study area was produced using the proposed method, and the spatial distribution characteristics of the landslides were investigated. Second, newly acquired SAR images were rapidly processed using the sequential InSAR method, thus achieving the dynamic deformation monitoring of landslides and the timely capture of earthquake-induced acceleration deformation signal.

Results The InSAR results revealed that there are 2 021 active landslides with varying dimensions within a 70 km radius of the epicenter, and six regions exhibited particularly dense landslide distributions. The spatial distribution of the mapped landslides exhibited a clustering pattern: primarily within 22 km of faults, 28 km of rivers, and 10 km of roads. Additionally, they were predominantly located at elevations below 3 400 m and on slopes ranging 15°-35°, and also mainly distributed along the north, north-east, and east directions.

Conclusion The deformation of some detected landslides within Jishishan County and along the Yellow River were significantly accelerated as a result of the earthquake event, thus reducing the stability of the slope. The reliability of the InSAR-derived results was verified through the field geological investigation. The proposed method offers significant technical value for the rapid investigation and dynamic monitoring of active landslides in similar seismic events, and the research findings provide scientific data support for post-earthquake reconstruction and secondary landslide disaster risk assessment.