Rapid Disaster Response to A Landslide Dam Using Multi-Sensor Optical and SAR Satellite Observations in Hualien, Taiwan

Remote sensing techniques, including optical and Synthetic Aperture Radar (SAR) imagery, offer an effective and rapid method for emergency disaster monitoring, particularly in areas that are difficult to access. In this study, multi-sensor satellite observations from Pleiades, TerraSAR-X, Capella, and Sentinel-1 are utilized to monitor the evolution of a landslide dam formed in Wanrung Township, Hualien, Taiwan, following intense rainfall associated with Tropical Storm Wipha. The landslide was initially detected by seismic monitoring on July 21, 2025. Subsequently, a high-resolution TerraSAR-X image acquired on July 30 revealed a landslide area of approximately 16 ha. Stereo optical images from Pleiades were used to generate a digital surface model (DSM), which enabled the estimation of landslide volume. Additionally, the water volume of the barrier lake was also derived from the lake surface elevation relative to the DSM. As the barrier lake gradually expanded, multi-temporal Pleiades imagery was applied to monitor changes in lake area. In parallel, Interferometric SAR (InSAR) analysis based on deep learning–assisted scatterer selection was conducted using Sentinel-1 data to investigate slope deformation around the landslide body. On September 23, 2025, additional heavy rainfall induced by Typhoon Ragasa caused a rapid rise in water levels, resulting in dam failure and subsequent downstream flooding. Owing to the all-weather, day-and-night imaging capability of active SAR systems, TerraSAR-X and Capella continued to acquire post-event data, providing critical information on embankment and bridge failures. The resulting inundation and sediment deposition affected approximately 382 ha in Guangfu Township. These results demonstrate that integrated multi-sensor satellite observations not only enable rapid tracking of landslide-dam evolution but also provide an operational and transferable monitoring framework covering dam formation, stability assessment, failure detection, and post-event impact evaluation. Such a comprehensive remote sensing strategy is particularly valuable for emergency management under increasing extreme rainfall events and can be applied to future landslide-dam crises worldwide.