The Qianjiangping landslide occurred after the first impoundment of the Three Gorges Reservoir in July 2003. Field investigation revealed that failure occurred when the reservoir reached 135 m, but the stability of the affected slope was already reduced by pre-existing bedding-plane shears, quarrying of mudstone from the landslide toe, and previous heavy rain. A possible explanation of the rapid and long runout mechanism of the landslide is that movement on a bedding-plane shear ruptured the calcite cement and rapidly reduced the sandstone strength to residual shear strength.

No abstract

At 4:40p.m. on November 23, 2008, the Gongjiafang slope collapsed on the north bank of Yangtze River in Wu Gorge of Three Gorges Reservoir. The 380,000-m 3 sliding mass consisted mainly of cataclastic rock. A video record of the major sliding incident was analyzed using the general laws of physical motion. The analysis indicated that the maximum speed and maximum velocity of the water wave near the landslide were 31.8m and 18.36m/s, respectively. Wave run-up investigation indicated that the maximum run-up on shore was 13.1m, which declined to 1.1m at Wushan dock 4km away. The incident causes no casualties, but did result in economic losses of RMB five million. The numerical simulation model GEOWAVE was used to simulate and reproduced the impulse wave generated by the landslide; the results were in good agreement with the observed incident. The numerical simulation data were then applied to analyze the decay and amplification effects of the landslide wave in the river course. The field investigations and witness information provide valuable materials for the studies of landslide kinematics and impulse waves generated by landslides. In addition, the research results provide a useful reference for future similar waves generated by landslides in reservoirs.

The first impoundment of the Three Gorges Dam reservoir in China started from a water surface elevation of 95 m on June 1, 2003 and reached 135 m on June 15, 2003. Shortly after the water level reached 135 m, many slopes began to deform and some landslides occurred. The Qianjiangping landslide is the largest one; it occurred on the early morning of July 14, 2003 and caused great loss of lives and property. Field investigation revealed that, although failure occurred after the reservoir reached 135 m, the stability of the slope was already reduced by preexisting sheared bedding planes. To study the mechanism of the rapid motion of this reactivated landslide, two soil samples were taken from a yellow clay layer and a black silt layer in the sliding zone, respectively, and a series of ring shear tests were conducted on the samples. One series of ring shear tests simulates the creep deformation behavior, while the other series simulates different shear rates. Conclusions drawn from analysis of the ring shear tests indicate that the mechanism of the rapid motion of the reactivated landslide was caused by the rate effect of the black silt layer during the motion phase after the creep failure. The yellow clay layer did not play any important role in the rapid motion in the 2003 event.

Landslides are destructive geological hazards that occur all over the world. Due to the periodic regulation of reservoir water level, a large number of landslides occur in the Three Gorges Reservoir area (TGRA). The main objective of this study was to explore the preference of machine learning models for landslide susceptibility mapping in the TGRA. The Wushan segment of TGRA was selected as a case study. At first, 165 landslides were identified and a total of 14 landslide causal factors were constructed from different data sources. Multicollinearity analysis and information gain ratio (IGR) model were applied to select landslide causal factors. Subsequently, the landslide susceptibility mapping using the calculated results of four models, namely, support vector machines (SVM), artificial neural networks (ANN), classification and regression tree (CART), and logistic regression (LR). The accuracy of these four maps were evaluated using the receive operating characteristic (ROC) and the accuracy statistic. Results revealed that eliminating the inconsequential factors can perhaps improve the accuracy of landslide susceptibility modelling, and the SVM model had the best performance in this study, providing strong technical support for landslide susceptibility modelling in TGRA.