1 Introduction Rockslides are one of the most hazardous natural events, because of their velocity and rarely detectable precursors. Possible strategies to prevent the risk associated to these events consist in: a) studying the gradual creep-driven deformation of rock mass by setting a proper monitoring system (Loew et al. 2016); b) realizing numerical models also considering progressive failure of rock slopes (Eberhardt et al. 2004); c) coupling microseismic monitoring systems and numerical models, taking in account rock mass damage during time (Xu et al. 2014; Tang et al. 2015); d) analyzing variations of ambient noise as indicators of imminent failure (Got et al. 2010; Lévy et al. 2010). To investigate long term rock mass deformations due to temperature, wind and rainfalls, the abandoned Acuto quarry rock wall (Frosinone – Central Italy) was selected in Autumn 2015 as test site for the installation of a multi-sensor monitoring system on a rock block prone to failure. The experimental site is included in the carbonatic Monti Ernici ridge. More in particular, Mesozoic wackestone with rudists crops out on the sub-vertical quarry wall with a height ranging from 15 m up to 50 m (Accordi et al. 1986). A geomechanical characterization of the rock mass led to the recognition of four joint sets, here indicated according to dip direction/dip convention: S0 (130/13) corresponding to the limestone strata, S1 (270/74), S2 (355/62) and S3 (190/64) (Fantini et al. 2016). The monitored sector is located in the NW portion of the 500 m long quarry front, which is characterized by the presence of a 64 m 3 densely cracked protruding block, separated from the back quarry wall by a main fracture. The multi-sensor monitoring system consists in two weather stations with air-thermometer, hygrometer, pluviometer and anemometer for wind speed and direction, installed at foot and top of the slope wall, one thermometer for the rock mass temperature, six strain-gauges installed on micro-fractures of the rock mass, four extensimeter installed on open fractures and one optical device. Moreover, a railway track was reproduced near the rock mass, to simulate a real rockfall hazard scenario (Fig. 1).

1 Introduction Understanding progressive rock failure and monitoring the stability of underground excavation zones, such as tunnels and underground mines, are critical to their operational safety. Many tunnels and underground structures are constructed under high-stress conditions, and stress redistribution may occur during and after the excavation. Stress redistribution generates energy imbalance in the rock mass, resulting in damage of the rock that impact the stability of underground structures (Chang and Lee 2004). Therefore, evaluating the progressive failure and damage mechanisms of rock at different stress levels are of great importance to geo-hazard assessment and operational safety of underground structures. In this study, we address this problem using an experiment at the laboratory scale, coupled with ultrasonic tomography (UT) and numerical simulation. A time lapse two-dimensional (2D) UT observation was conducted on a granite slab under uniaxial compression. This test was then reproduced using the combined finite-discrete element method (FDEM). The entire deformation and failure processes were studied using this combination of technologies at the macroscopic and microscopic scales. 2 Material and method 2.1 Sample and compression test The rock sample investigated in this study was a coarse-grained granite (Fangshan granite) slab 220 mm long, 110 mm wide, and 30 mm thick. This granite consisted of three main mineral phases: Feldspar (67%), Quartz (23%) and Biotite (10%), with an average grain size of 2.6 mm. The sample was tested by applying a uniaxial stress and a UT survey was conducted initially (0 MPa) and at intervals of 20 MPa applied stress.

1 Abstract The main goal of this study is to describe the spatial and temporal evolution of rock fall phenomena triggered by rapid slope deformation. To this end, we combine low cost seismic sensors and image processing to study a large instability adjacent to the Great Aletsch glacier in the Swiss Alps, i.e. the Moosfluh slope, which is undergoing an acceleration phase since the late summer 2016. With this analysis, we aim at a better understanding of the relationship between the kinematic behavior of rock slope instabilities and progressive rock mass damage, which may lead to catastrophic failure. 2 Introduction The Great Aletsch Region is a prominent site located in Switzerland, hosting the largest glacier of European Alps. In this area, glaciers have undergone to several cycles of advancement and retreat, which have deeply affected the geomorphological evolution of the surroundings (Grämiger et al. 2017). The glacier is experiencing a progressive retreat in the order of 50 meters every year, consequently, load is released from the adjacent rocks previously constrained by the ice mass, and slope instabilities might be triggered (e.g., Cossart et al. 2008). In the Aletsch region, a deep-seated slope instability called Moosfluh shows since the 90's a slow but progressive increase of surface displacement (Strozzi et al. 2010). The moving mass associated to Moosfluh affects an area of about 2 km 2 and entails an estimated volume of about 150–200 Mm 3 (Kos et al. 2016). In the late summer 2016, an unusual acceleration of the Moosfluh rockslide was observed.

1 The small and large deformations of geomaterials From the continuum mechanics point of view, a number of geomaterials are both damageable elastic solids in which highly localized features emerge as a result of failure and materials experiencing high, permanent strains that dissipate stresses. In this sense, modelling their deformation lies between a solid mechanics (small deformations) and a fluid dynamics (large deformations) problem. One important example is the Earth's crust, in which brittle fracturing and Coulomb stress redistribution are known to take place and for which scaling properties have been recognized for years (Kagan & Knopoff 1980; Turcotte 1992 and others). Along active faults, co-seismic fracturing activates aseismic creep, leading to deformations that can be larger than those associated with the fracturing itself (Cakir et al. 2012) and to slip rates that decrease progressively over years to decades due to various healing processes (Gratier et al. 2014). Creep relaxes a significant amount of elastic strain, retarding stress accumulation along some portions of faults and concentrating stresses on other locked portions. Hence this dissipative process should be included in earthquakes models (Cakir et al. 2012; Gratier et al. 2014). Another example is sea ice, which deforms rapidly under the action of the wind and ocean drags, in the brittle regime, and for which scaling properties have also been recently recognized (Marsan et al. 2004 and many others). In this case, much larger deformations occur once faults, or ice "leads" (see Fig. 1a, A ), are formed and divide the ice cover into ice plates called "floes" (Fig. 1a, B ), as these plates move relative to each other with much reduced mechanical resistance. In sea ice models, these large deformations must be accounted for as they set the overall drift and long-term evolution of the ice pack. In such contexts, the challenge of the continuum modelling approach lies in the representation of the discontinuities that arise within a material due to fracturing processes using continuous variables and grid-cell averaged quantities. On the numerical point of view, another challenge arises as the methods employed must allow resolving the extreme gradients associated with these discontinuities while limiting the diffusivity associated with advective processes. Here, we present a simple continuum mechanical framework called Maxwell-Elasto-Brittle (Maxwell-EB), built in the view of allowing a transition between the small deformations associated with the fracturing and the larger, permanent, post-fracture deformations, while having the capability of damage mechanics models to reproduce the observed space and time scaling properties of the deformation of brittle geomaterials. The theoretical and numerical development of this new rheological model will be discussed of in two different geophysical contexts: modelling the drift and deformation of Arctic sea ice at regional scales and representing the pre-eruption deformation of a volcanic edifice.

Some studies have suggested that the shaking and deformation associated with earthquakes would result in a temporary increase in hillslope erodibility. However very few data have been able to clarify what causes this transient state and what controls its temporal evolution. We present integrated geomorphic data constraining an elevated landslide susceptibility to rainfall following 5 continental shallow earthquakes, the Mw 6.9 Finisterre (1993), the Mw 7.6 ChiChi (1999), the Mw 6.6 Niigata (2004), the Mw 6.8 Iwate-Miyagi (2008) and the Mw 7.9 Gorkha (2015) earthquakes. We constrained the magnitude (5 to 20 fold) and the recovery time (1 to 4 years) of this susceptibility change and associated it with subsurface damage caused by the strong shaking (Marc et al. 2015). The landslide data suggest that this ground strength weakening is not limited to the soil cover but also affects the shallow bedrock. Coseismic rock damage is supported by observations of shallow (0 to ~100m) seismic velocity drops constrained with ambient noise waveform correlations within the epicentral area of four of those earthquakes (e.g., Takagi etal. 2012, Hobiger et al. 2015). For most stations we observe a subsequent exponential velocity recovery (i.e. proportional to e −t/ τ ) with a τ value in fair agreement with the one estimated based on landslide observation. This recovery dynamic is also consistent with post-seismic processes, namely GPS post-seismic displacement and aftershocks decay (Fig. 1, Marc et al., in review). We analyzed strain time series in Japan and Taiwan and it appears inconsistent with the recovery of landslide susceptibility and shallow seismic velocities. In contrast, surface dynamic strain associated with ground shaking caused by aftershocks display similar relaxation time and may control the subsurface property recovery.

1 Introduction The development of large rock slope failures within glaciated valleys is often assumed to be primed by steepening of topography from glacial erosion and by subsequent glacier retreat (McColl 2012). Glacier retreat removes confining stresses, changes hydrological, thermal, and weathering processes, and changes the interaction of co-seismic waves within the slope (e.g. Gischig et al. 2011; McColl et al. 2012; Grämiger et al. 2016), gradually lowering stability through strength degradation or a rise in destabilizing shear stresses. Few studies have documented the evolution of slope failure in glaciated valleys over historical time periods, and thus relatively little is known about how these processes interact to generate failures. For example, glacial debuttressing may cause catastrophic failures, as indicated by numerous post-glacial rock avalanche deposits in formerly glaciated valleys; but some of these failures may have occurred thousands of years following ice withdrawal (e.g. Ballantyne et al. 2014; Ostermann & Sanders, in press), suggesting that earthquake shaking or gradual strength degradation may also be responsible for triggering failure. It has also been suggested that the initial development of rock slope failures can proceed prior to the complete glacier withdrawal, and involve deformation of a buttressing glacier (McColl & Davies 2013). This study takes the opportunity to investigate the long-term development of a large, active rockslope failure that is still partially-buttressed by a glacier, with the aim to explore how failure develops over time and to assess the factors controlling movement and failure evolution. 2. Study site The study focusses on the Mueller Rockslide, in Aoraki Mt Cook National Park, in the central Southern Alps of New Zealand (Fig. 1). The rockslide, estimated to be ~150 Mm3 in volume, occupies the dip slope of an overturned anticline in Mesozoic greywacke sandstone (Lillie & Gunn 1964) (Fig. 2). The rockslide toe is partly buttressed by the Mueller Glacier, which has thinned by ~100 m since the Little Ice Age. The glacier becomes narrow adjacent to the landslide toe, perhaps in part due to squeezing by the Mueller Rockslide (McColl & Davies 2013).