Abstract. Forecasting the occurrence of large, catastrophic slope failures remains very problematic. It is clear that in order advance this field a greater understanding is needed of the processes through which failure occurs. In particular, there is a need to comprehend the processes through which a rupture develops and propagates through the slope, and the nature of the inter-relationship between the stress and strain states of the landslide mass. To this end, a detailed analysis has been undertaken of the movement records for the Selborme Cutting slope failure, in which failure was deliberately triggered through pore pressure elevation. The data demonstrate that it is possible to determine the processes occurring in the basal region of the landslide, and thus controlling the movement of the mass, from the surface movement patterns. In particular, it is clear that the process of rupture development and propagation has a unique signature, allowing the development of the rupture to be traced from detailed surface monitoring. For landslides undergoing first time failure through rupture propagation, this allows the prediction of the time of failure as per the "Saito" approach. It is shown that for such predictions to be reliable, data from a number of points across the landslide mass are needed. Interestingly, due to the complex stress regime in that region, data from the crown may not be appropriate for failure prediction. Based upon these results, the application of new techniques for the detailed assessment of spatial patterns of the development of strain may potentially allow a new insight into the development of rupture surfaces and may ultimately permit forecasting of the temporal occurrence of failure.
Dilatancy toughening of shear cracks and implications for slow rupture propagation
