Earthquake-induced reactivation of landslides. Recent advances and future perspectives

Earthquake-induced landslides are generally responsible for severe damages and losses as proved by records of last seven years which demonstrate that more than 50 % of the total losses due to landslides in the World are due to co-seismic slope failures (Petley 2012). Moreover, as reported by Bird and Bommer (2004), the greatest damage caused by earthquakes are often related to landslide events. Several historical earthquake-induced landslides demonstrated the severity of such events as they often involved areas which have been intensely damaged by the seismic shaking. This was, among others, the case of Las Colinas landslide, triggered by the January 13th 2001 Mw 7.6 El Salvador earthquake, which caused about 585 losses (Evans and Bent 2004). Earthquake-induced landslides can also trigger co-related phenomena, by a sort of “domino-effect”, among which river damming and tsunamis. An extraordinary example of such an effect is reported in some historical chronics that testify the catastrophic scenario due to earthquake-induced Scilla rock avalanche triggered by the 6th February 1783 earthquake in Southern Italy (Bozzano et al. 2011b) which killed almost 1500 persons as it produced a 16 m height tsunami wave along the coastline where people found refuge after the mainshock occurred one day before (Mazzanti and Bozzano 2011). The last published version of the database of earthquake-induced ground effects in Italy CEDIT by Martino et al. (2014; http://www.ceri.uniroma1.it/cn/gis.jsp), whose main peculiarity is to be constructed on the basis of several historical documents over a period of approximately one millennium from 1000 AD to present, demonstrates that landslides represent the most documented type of ground failures. These effects correspond to more than 40 % of the inventoried ones, which include ground-cracks, liquefactions and superficial faulting. Earthquake-induced landslides should be distinguished in “first time slope failures” and “reactivated landslides” as very different approaches are requested for providing failure scenarios as well as for identifying the main predisposing conditions. Earthquake-induced first time slope failures mainly involve jointed rock masses since they are often represented by disrupted landslides (Keefer 1984), i.e. rock-falls, topples or block-sliding. Such effects are expected to occur up to some tens of kilometers also in case of 4.0 ≤ Mw ≤ 5.0 earthquakes (Keefer 1984; Rodriguez et al. 1999); several studies demonstrated the significant role of local seismic effects (i.e. due to topography, slope orientation, near field conditions) in triggering disrupted landslides (Sepulveda et al. 2005; Alfaro et al. 2012 and references therein). Nevertheless, a part for back-analysis approaches, several critical features still remain in forecasting first time rock failures as: (i) diffused geomechanical characterization should be available or suitable criteria for extrapolating them, (ii) very detailed geological features should be collected on generally impervious areas, (iii) high resolution digital elevation models (DEM) should be provided for steep slopes or cliffs. Earthquake reactivated landslides more commonly involve coherent soils or debris; generally, it is possible to inventory these landslides as they are active or quiescent phenomena, which can be recognized by typical landforms or historical chronics that document their past activations. Although already existing landslide masses are a priori recognizable, a great effort is requested to evaluate how their stability conditions change due to earthquake occurrence as well as to quantify their co-seismic or post-seismic mobility in terms of expected displacements. This difficulty depends on the complex interactions between the seismic waves and the landslide mass that is conditioned by several features among which the slope geometry, the landslide mass properties, the physical characteristics of the seismic waves (Lenti and Martino 2012, 2013). With respect to the curves of magnitude versus maximum epicentral distance that are available in literature, the most of the outliers (i.e. far field landslide events) inventoried in the “instrumental age” (i.e. with available accelerometric records) are represented by coherent landslides which involve stiff clays or soils (Delgado et al. 2011). The majority of the documented outlier cases regards the reactivation of large to very large landslides, whose volume is in the order of ten millions of cubic meters. Such a high landslide intensity, combined with the unexpected far-field occurrence, is particularly relevant for risk management in high seismicity areas. This paper is focused on earthquake-induced landslide reactivation and report a state-of-art and some recent advances in slope stability analysis as well as in co-seismic displacement evaluation for pre-existing landslide masses. As discussed in the following, most of the criticism in analyzing slope stability under seismic action regards the dynamic inputs considered for computing the safety factor (SF) and the uncertainness in assuming mechanical properties necessary for evaluating the available soil strength. On the other hand, the main criticism in computing co-seismic landslide displacements depends on the rheological assumptions (i.e. rigid vs. deformable soil masses) as well as on the complexity of the physical interactions between seismic waves and landslide masses (i.e. 1D or 2D seismic amplification, incidence angle of seismic waves related to the slope geometry, seismic wave polarization within landslide mass).