Abstract: Traditional inspection methods for collapsing riverbanks are limited by the scope of monitoring, resulting in limited efficiency and an inability to promptly detect dangerous situations. Extracting water shoreline from satellite images is of great significance to the monitoring and early warning of river bank collapse. Three water body extraction methods based on DeepLabV3+image semantic segmentation technology, traditional NDWI index, and SAR technology are used with Gaofen 1 and 2 satellites, Landsat 8 satellites, and Sentinel-1 satellite images as data sources. This paper discusses the application ability of multi-source satellite data in the identification of river bank collapse, taking the example of the bank collapse with a length and depth of about 100 m in Xiaopan of Jiayu County, Hubei Province in December 2021.The research results show that the domestic Gaofen 1 and Gaofen 2 satellites combined with the DeepLabv3+method can extract the water edge. Through the comparison of different series image, the river bank collapse can be reflected. In the actual work practice, it is recommended to use Gaofen satellite and DeepLabv3+method for automatic water shoreline extraction. In cloudy and rainy weather, SAR method is used to supplement. If necessary, visual discrimination can be carried out manually in combination with the images of the Gaofen satellite and Sentinel-1 satellite.