Prediction of river damming susceptibility by landslides based on a logistic regression model and InSAR techniques: A case study of the Bailong River Basin, China

• Published in: Engineering geology 2022-03, Vol.299, p.106562, Article 106562
• Main Authors: Jin, Jiacheng, Chen, Guan, Meng, Xingmin, Zhang, Yi, Shi, Wei, Li, Yuanxi, Yang, Yunpeng, Jiang, Wanyu
• Format: Article
• Language: English
• Subjects: Active landslides; Logistic regression; Qinghai-Tibet Plateau; River damming susceptibility; Time series InSAR; Unstable slopes
• ISSN: 0013-7952; 1872-6917
• DOI: 10.1016/j.enggeo.2022.106562

Although a fundamental part of landslide dam risk management, the evaluation of the spatial probability of dam formation is a challenging problem. In this study, we developed a new method for evaluating the river damming susceptibility of active landslides and unstable slopes (ALUSs), which combines a logistic regression model with time series Interferometry Synthetic Aperture Radar (InSAR) techniques. The approach is applied in the Bailong River Basin, on the eastern edge of the Qinghai-Tibet Plateau, China. We established a primary assessment model using historical damming and non-damming landslide data based on a logistic regression model. Jack-knife analysis showed that the most significant factors controlling river damming susceptibility were valley width, landslide volume, internal relief, and river discharge, in order of decreasing significance. Seventy ALUSs were detected by the InSAR technique and their volumes were predicted using a Volume–Area empirical power-law relationship. The river damming susceptibility of these instabilities was assessed and the results show that 18 of them have a high or very high probability of river damming. Four active landslides with volumes >107 m3 were found to be prone to forming complete blockage conditions, as identified by the Morphological Obstruction Index (MOI). The predictive ability of the methodology was tested using two recent river damming cases and the prediction results were in good agreement with reality. Our approach can potentially help in the early identification and hazard assessment of landslide damming in the Bailong River Basin and similar alpine areas. •A new approach to predict river damming susceptibility of ALUSs is proposed.•The approach combines logistic regression model with time-series InSAR.•The prediction ability of this approach is tested by two recent landslide-damming events.•The approach is important for the spatial prediction of potential landslide damming.