Carolina Seguí , Hadrien Rattez, Emmanouil Veveakis

Deep-seated catastrophic landslides are among the most powerful natural hazards on earth. These devastating events are not possible to be prevented yet, because of their large volumes and sudden acceleration phase. The present study suggests a new method to detect when a landslide will turn unstable, giving both a time-window to evacuate the area that is going to be affected and critical values for measurable variables (velocity and basal temperature) up to which remediation measures can be deployed. This work focuses on large ancient landslides reactivated due to human interaction, like the construction of a dam in the vicinity of the landslide that causes water table variations and affects the stability of the landslide. The main hypothesis of this work is that most of the deformation of deep-seated landslides is concentrated on a thin, basal shear-band forming the sliding surface. That allows deep-seated landslides to be approximated as elastic/rigid blocks sliding over a viscoplastic shear band, featuring weak phases like expansive clays. When the landslide creeps, it causes friction in the shear band to raise the temperature and increase the pore pressure of the clays until they reach a point of near-zero friction and collapse catastrophically. This study deploys an energy-based approach, accounting for the heat generated due to friction, to find the critical point where the landslide turns unstable. The theoretical model consists in a stability analysis of the landslide using a pseudo-arclength continuation method. The model is applied to the famous Vaiont landslide in Northern Italy and Shuping landslide in Three Gorges Dam in China. The results of the model reproduce with great accuracy the behavior of both landslides, thus, finding the critical point of stability of the slide.