Reservoir Landslide Displacement Prediction Under Rainfall Based on the ILF-FFT Method

Landslide prediction and early warning have always been the emphasis and difficulty of geological disaster prevention and mitigation. The prediction and early warning method is more difficult because the landslide deformation process under the joint action of rainfall and reservoir water is complex. The inverse logistic function (ILF) is used to construct the landslide displacement trend term function, and the Fast Fourier Transform (FFT) principle is used to construct the landslide displacement fluctuation term function contributed by rainfall and reservoir water level. The ILF-FFT prediction and early warning model of reservoir landslides under rainfall is proposed, and the four-stage quantitative division method of landslide evolution is realized based on the ultimate curvature of velocity and acceleration time-history functions. The ILF-FFT prediction and early warning model based on the least-squares method is applied to Gapa landslide and Sanmendong landslide, respectively. The results indicated that the prediction accuracy of the ILF-FFT model is high so that the model could reflect the overall development trend and local fluctuation characteristics of landslide displacement well. The applicability of the ILF-FFT model is good as the model is suitable for the analyses of evolution stages and landslide types of various parts of reservoir landslides.