Triggering factors and threshold analysis of baishuihe landslide based on the data mining methods

The analysis of landslide monitoring data is important to the study and prediction of landslide deformation but is very challenging. In this research, a data mining method combining two-step clustering, Apriori algorithm and decision tree C5.0 model are proposed, and the Baishuihe Landslide in the Three Gorges Reservoir area is taken as the study case. 6 hydrologic factors related to rainfall and reservoir water level are chosen to carry out the data mining analysis. First, 6 hydrologic triggering factors and the deformation rate of the landslide are clustered by the two-step clustering. Then, the Apriori algorithm is used to mine the association rules between triggering factors and deformation rate. A total of 173 association rules are generated based on the data mining, and 20 rules are selected to be analyzed. At last, the decision tree C5.0 model is built to carry out threshold analysis of hydrologic triggering factors. The results show that monthly cumulative rainfall plays an important role in controlling landslide deformation, and 73.9 mm can be regarded as its threshold. Monthly average water level is the second factor to control landslide deformation. While the monthly maximum daily rainfall has no direct control over the acceleration stage of landslide deformation. The data mining method proposed in this paper has a high accuracy in the study of Baishuihe landslide, which could provide a significant basis for the data analysis and prediction of the accumulative landslide in the Three Gorges Reservoir area.

Automated summary

The study proposed a data mining method to analyze monitoring data of the Baishuihe Landslide, revealing that monthly cumulative rainfall is a key factor in controlling landslide deformation. Additionally, a decision tree model was established for threshold analysis of hydrologic triggering factors.