Abstract:Affected by extreme weather (heavy rainfall) and high intensity earthquake, river blocking events caused by landslides often occur in high mountain canyon area, forming a barrier dam with obvious characteristics of “more overtopping and short life”. After the landslide dam failure, a dam-break flood with strong destructive force will be formed. As an important hydraulic parameter, flood peak discharge directly determines the impact of downstream disasters. In this paper, 67 cases of landslide dam failure from the world with complete records are widely collected, the existing landslide dam database is expanded, the data sample interval is increased, and then the prediction model of the peak discharge of landslide dam failure is modified based on the new database. The mean square error (RMSE) and correlation coefficient (R2) are used to compare and analyze the prediction effect of each model. Finally, the modified model was applied and compared by two landslide dam break events (one on October 10 and one on November 3)at the junction of Jiangda County in Tibet and Baiyu County in Sichuan Province in 2018. The results show that the overall calculation effect of the modified model is improved to varying degrees compared with that before the modification, and the scope of application is wider. However, most models do not fully consider the key influencing factors of complex dam failure problems under water-soil coupling due to the single selection index. After comparison, Peng M’s model (revised RMSE and R2 are 6070.52 m3/s and 0.9909, respectively) is preferred to calculate the peak discharge of landslide dam. Costa’s (1985) dam height Hd single parameter model (revised RMSE and R2 are 61231.95 m3/s and 0.0746, respectively) is not recommended, and other models can be used as reference. The conclusion is consistent with the application comparison results of Baige landslide dams failure cases. The research results can provide important theoretical basis for disaster prevention and mitigation and emergency rescue of landslide dams.