How to obtain alert velocity thresholds for large rockslides/rock avalanches

A reliable forecast of the failure phase of landslides is still a difficult task, expecially when coping with large rockslides (potentially rock avalanches), because of uncertainly known geometry, non-linear time-displacement relationships and superposition of seasonal effects. The analysis of monitoring data is fundamental to evaluate the geometry and kinematics of a landslide. Anyway, the most important goal of the monitoring activity on landslides is the prediction of the time to failure, more relevant when rockslides threaten villages and lifelines. In this work, a new method is suggested, starting from existing models, to forecast slope movement and failure and to prepare alert thresholds. The series of adopted data concern the 30 Mm3 “Ruinon” rockslide (Valfurva, Central Alps, Italy), active during the last three decades with an acme period starting from 1996. The slide consists of a “compound” movement on a 50-70 m deep surface, suspended on the valley and suitable to originate a large, fast moving rock avalanche. The slide was analysed through photointerpretation, field survey, DEM analysis, geomechanical characterisation and numerical modelling of slope deformations and rock avalanche spreading. These steps allowed a better interpretation of the data provided by the monitoring network set up by the Regione Lombardia Geological Survey since 1996. The network includes distometers, wire extensometers, GPS benchmarks, inclinometers and borehole extensometers. Series of absolute and cumulated displacement data have been analysed. This allowed to recognise the trend of displacements and to distinguish four different sectors, each one characterised by a different state of activity. Representative data were fitted by power-law theoretical curves, according to the “accelerating creep” theory by Voight (1989), in order to evaluate a suitable failure time of the rockslide. Unfortunately, large seasonal deviation from theoretical curves (acceleration during spring and summer and constant, low velocity during winter) hampered the computation of the exact time of failure. Then, the Voight’s method was applied to single accelerating phases to find values of the controlling parameters (A and a) typical of the involved rock mass. This allowed to obtain theoretical curves representing rock mass behaviour. The corresponding “fictitious” times of failure allowed the computation of velocities at different time intervals before the final collapse. These velocities represent threshold values (pre-alert, alert and emergency) to be used for emergency management