On October 17 and 29, 2018, two rock and glacier avalanches occurred on the western slope of the Sedongpu Basin upstream of the Yarlung Zangbo River in the Tibetan Plateau, forming the disaster chains and causing damage to many bridges and roads. Based on the comparative analysis of multiple pre-and post-remote sensing images, the initial sliding body, which was composed of rock and glacial material, was located on a steep slope above an elevation of 6000 m. Under the coupling effect of multiple factors such as gravity, rainfall, and weather changes, the initial sliding body detached from the source zone and then transformed into a debris flow after impact and fragmentation. The debris flow traveled downstream and scraped loose glacial till in its path, causing the volume of the sliding body to increase. In addition, the debris flow traveled 10 km under low frictional resistance, as a result of the lubrication via early rainfall and glacial meltwater. Eventually, the debris flow rushed out onto the valley floor, forming a landslide dam and blocking the Yarlung Zangbo River. The deposit volumes on October 17 and 29 were 20.4 million m3 and 10.1 million m3, respectively, with a total mean thickness of ~22m. This study provides an insight into the dynamic process as they unfolded, through multitemporal satellite imagery and numerical simulation. Furthermore, we also discuss the potential cause of rock/ice avalanche and disaster scenarios, as well as the tendency of the rock and glacier avalanches are discussed.
In recent years, rapid and reliable interpretation for emergency response to landslides and other geological hazards has become increasingly important. This paper presents DFmamba, an improved deformable dual-branch visual state-space network, to address engineering challenges such as missed large landslide bodies, boundary shifts, and loss of small-scale details. DFmamba mitigates the limited effective receptive field and window-partition constraints that often prevent existing methods from balancing large-area semantic consistency, multi-scale detection, precise boundary delineation, and computational efficiency. It employs a parallel encoder with a convolutional branch and a Visual State-Space Model (VSSM) branch to jointly capture local textures and global context. In the decoder, deformable residual blocks (DRB) enhance geometric modeling of irregular boundaries, while multi-scale feature alignment and a shallow high-frequency injection (MFP) mechanism strengthen boundary responses and preserve fine details. Experiments on the public CAS dataset against representative CNN-, Transformer-, and SSM-based baselines show that DFmamba achieves improved Precision, Recall, F1-score, and IoU, with stable performance across multi-scale scenarios, demonstrating strong robustness for landslide segmentation.