Abstract: Under the impacts of the operations of upstream cascade reservoirs, the river channel in the Middle Yangtze River has suffered from intensive erosion, resulting in a frequent occurrence of bank erosion event that threatens the safety of riparian levees. However, the monitoring and early-warning system of bank erosion is not well constructed in the MYR. Therefore, this study proposed an early-warning technology of bank erosion event, which combines a bank erosion prediction module with a warning-level grading module. In the prediction of bank erosion, potential bank erosion sites are firstly selected relying on a prior assessment of the river regime. The occurrence probabilities of bank erosion are then calculated by a well-trained Random Forest Model for bank erosion, and the erosion widths at these sites are calculated by a dynamic model. In terms of the grading of warning-level, two additional indicators of floodplain width (between the water and the levee) and riparian residential area are adopted to represent the importance of a selected site to the riparian society. A warning- level matrix is constructed to combine these four different indicators into a final warning-level of bank erosion. The proposed technology is used to conduct the early-warning process of bank erosion event of the Lower Jingjiang Reach of the MYR in 2020.