Baige landslide, which located on the upper reaches of Jinsha River, happened successively on October 10 and November 3, 2018 and blocked the river twice. The dammed lakes and their breaching flood brought about severe disaster to local residents along the river. After the sliding, three crack zones were formed at the back edge of the landslide, which may block the river again. In view of the uncertainty of the physical and mechanical parameters of the landslide, this paper uses the optimization algorithm based on the maximum likelihood estimation to inverse the shear strength parameters of the landslide, and uses the inversion parameters to calculate the reliability and evaluate the stability of the residual unstable block by Monte Carlo method. The results show that: (1) Combining the response surface method with the maximum likelihood based probability back analysis method, the calculation process is simple and the calculation method has strong adaptability. (2) According to the inversion results Baige 11.03 landslide, the posterior distribution of shear strength is c~N(13.05kPa, 32), tanφ~N(0.64, 0.042). The uncertainty of prior parameters can be reduced by inversion. (3) The local failure mode of each crack zone is obtained by searching the most dangerous sliding surface. The calculated scope of the most dangerous sliding area of each crack zone is approximately the same as that of the instability area obtained from the field investigation and monitoring, and the failure probability under the local failure mode is significantly greater than that under the overall failure mode, indicating that the failure of the crack zone is mainly progressive traction failure. (4) The volumes of the instability zones C1-1, C2-1 and C3-1 at the back edge of Baige landslide are 7 × 105 m3, 3.2 × 106 m3 and 1.3 × 106 m3, respectively, with the corresponding failure probability of 21.70%, 33.90% and 27.30%. All of which are extremely dangerous. It is necessary to carry out risk assessment on the damming-breaching-flooding disaster chain, and put forward reasonable disposal measures and emergency plans.
