Justification of seismic triggering of large prehistoric rockslides that originated on the slopes of anticlinal ridges armoured by thick carbonate units has been performed by examples of the gigantic Seimareh rockslide in Zagros (Iran) and two structurally similar, though much smaller rockslides in Dagestan (Greater Caucasus, Russia). Such structural and geomorphic conditions allow precise reconstruction of the pre-slide topography of the studied sites that increases reliability of their back analysis significantly. Linear dimensions of landslides are much larger than thickness of the siding block that makes the simplified 2D numerical modelling of these slopes quite realistic. The pseudostatic analysis performed at the first step confirmed that the study slopes could not fail without strong earthquakes. However, further dynamic analysis performed by use of the Newmark method allowed estimating characteristics of strong motions that could result in formation of rockslides that had converted in long runout rock avalanches. Possible uncertainties and open problems are discussed as well.
Finding of the actual triggering factors of large prehistoric landslides is a real challenge and its reliability is critically important for both landslide and seismic hazard assessment, as far as they are often interpreted as evidence of strong past earthquakes. However, numerous historical case studies demonstrate that very large landslides, those in the hard rock in particular, can either be triggered by strong seismic shaking, or can occur without any earthquake. Various criteria have been proposed to prove seismic origin of large prehistoric rockslides [1, 2, 3, 4], but neither size (volume) of a landslide, nor runout can be considered as indicator of their seismic origin [5, 6].
One of the most promising ways to reveal their actual trigger is the back analysis of slope stability at the sites where such landslides had occurred [7]. Such analysis, however, faces three major problems: 1) uncertainty of the mechanical properties of rocks and rock massifs used as an input data for the numerical modelling; 2) difficulties in reconstructing the pre-slide topography that predetermines mass distribution in the slope model; 3) impossibility in most of cases to determine in advance what earthquake could trigger the slope failure in question.