Displacement Prediction of Landslide Influenced by the Periodic Precipitation, Reservoir Level and Groundwater Level Fluctuations

Abstract

Since the initial impoundment of the Three Gorges Reservoir in June 2003 and approximately 30 m of reservoir level fluctuation, numerous preexisting landslides have been reactivated. To mitigate disastrous landslides, the Baishuihe landslide in the Three Gorges region was selected as a case study in predicting such displacement using the monitoring data and a radial basis function-support vector machine (RBF-SVM) model. The landslide displacement was strongly influenced by periodic precipitation, reservoir level and groundwater level fluctuations. Primary landslide influencing factors were used as independent variables to predict the displacement using several kernel function types including polynomial function, sigmoid function, and RBF based on SVM model. Prediction results demonstrated that the RBF-SVM with the optimal parameters c of 170, 0.05 and 0.04 can provide the best predictive accuracy, with the maximum and minimum absolute error values of 9.84 and 0.47mm, respectively.