Extreme rainfall is one of the critical triggering factors for geological hazards worldwide, particularly in western China, where it has induced numerous loess-related geological disasters. On September 1, 2022, at approximately 01:05 AM (Beijing Time), a catastrophic loess–mudstone interlayer landslide occurred on the northwest slope of Hongya Village, Huzhu County, Qinghai Province, resulting in seven fatalities. This study investigates the deformation characteristics and failure mechanisms of the landslide using an integrated “air–space–ground–subsurface” approach, combining geomorphological field investigations, engineering geological drilling, high-density electrical resistivity imaging, and SBAS-InSAR time-series analysis. Results indicate significant spatial heterogeneity in deformation, with two major deformation zones (Zone #1 and Zone #2) and five secondary landslides identified. Zone #1 was characterized by a “frontal creep–middle traction–rear pushing” failure mode under the influence of extreme rainfall and terrain conditions. In contrast, Zone #2 was affected by both rainfall and lateral traction induced by the overall sliding of Zone #1, forming a “rear pushing–lateral traction–frontal settling” pattern. High-density resistivity results revealed that groundwater was primarily stored in the loess–gravel layer and fractured bedrock, with noticeable water accumulation within the sliding zone. InSAR time-series deformation analysis from 2018 to 2022 shows that intense summer rainfall accelerated slope movement and produced multi-stage deformation processes, consistent with field observations. Additionally, a regional rainfall threshold model based on 81 rainfall-induced landslide events was developed, quantifying critical rainfall durations and accumulations that trigger landslides in the region. The integration of this rainfall threshold with multi-source monitoring data enhances the understanding of rainfall-triggered landslide mechanisms and improves early warning capabilities. This study demonstrates the integrated advantages of conventional geological investigations and remote sensing monitoring, proposes a new approach for identifying and analyzing large-scale complex landslides, and provides important theoretical support for the early warning and risk mitigation of hydrodynamically triggered landslides.

Loess-mudstone landslide,Deformation characteristics,Extreme rainfall,Rainfall threshold,Electrical resistivity tomography (ERT),SBAS-InSAR