Slope movement classification and new insights into failure prediction based on landslide deformation evolution

Abstract

This study aims to investigate the features of landslide evolution from the perspectives of statistics and nonlinear systems. First, to understand the complex processes of landslide evolution intuitively, a new landslide classification based on deformation patterns is presented. Then, frequency-size statistics of velocity events are performed in five historical landslides, including four failure cases and one non-failure case. The results show that velocity events follow power-law distributions within a certain size range. The power-law behaviors are usually associated with a ubiquitous phenomenon of self-organized criticality (SOC). A statistical test indicates that some outliers with extreme sizes significantly exceed power-law extrapolation. One possible explanation is that they are “dragon-kings” (DKs), possessing different formation mechanisms from their peers in power laws. Herein, the role of SOC and DKs in landslide evolution is analyzed in detail. The two theories provide new ways to understand the underlying mechanisms of landslide evolution. Meanwhile, power-law distributions and DKs provide new insights to predict landslide deformation.