Objectives In the research work of reservoir landslide displacement prediction, due to the lag of reservoir water level response, it is difficult for the traditional landslide displacement prediction model to analyze the monotonically increasing step deformation characteristics, which seriously affects the prediction results, and it is necessary to establish a landslide displacement prediction model that can consider the time lag effect.

Methods We analyze the time lag effect of reservoir levels separately through grey correlation, account for the cumulative effect of earlier rainfall, and consider the effect of earthquake on landslide deformation, and establishe an autoregressive distributed lag landslide displacement prediction model that can be applied to engineering sites.

Results The results show that: (1) The engineering case study concluded that rising reservoir levels and earthquakes were the main triggering factors for the increased deformation of the landslide, and the lag time of reservoir levels acting on landslide deformation was 8 days. (2) The correlation coefficient between the cumulative landslide displacement and the actual displacement calculated by the new model is as high as 0.992 7, with a root mean square error of 14.11 mm. (3) The calculation of trend speed ratio indicators can provide a new sensitivity evaluation parameter for landslide monitoring and early warning.

Conclusions The study establishes a physically significant prediction model for reservoir bank landslide displacements, provides a comprehensive analysis of landslide displacements, achieves a quantitative calculation of the seismic contribution to landslide displacement evolution, and provides new technical support for the safety risk management of the whole process of reservoir bank landslide evolution during the water storage period.