Dense 3D seismic traveltime tomography to understand complex deep landslide structures

Bedrock geometry, geological discontinuities, geotechnical units and shear surfaces/bands control the deformation patterns and the mechanisms of slope instabilities. Seismic P-wave refraction tomography is useful to detect these features because P-wave velocity significantly decreases in fractured and weathered rocks relative to consolidated ones, and because lateral changes of velocity can highlight alternation of dipping fracture zones and consolidated rocks. Acquiring this information at high spatial resolution is of paramount importance to model landslide behaviour. 

The Viella slope instability (Hautes-Pyrénées, France) is a complex and deep-seated (> 80 m) landslide which has reactivated in Spring 2018 as a consequence of both a 100-yr return period flash flood (Bastan torrent) which affects the lower part of the slope, and a major rockslide (> 100.000 m³) modifying the stress conditions in the upper part. The landslide, which covers an area of ca. 650 000 m², is primarily composed of schists with different degrees of weathering, forming several kinematic units with surface velocities in the range [0.5 – 5] mm.month^-1. Many buildings and infrastructures (roads, bridge) are progressively damaged (cracks, progressive tilting) and scarps and lobes develop at the surface delineating the kinematic units.  

In order to model the evolution of the landslide and design possible mitigation measures (drainage, slope reprofiling), a 3D seismic survey has been carried out in summer 2020. The survey was designed to provide a highly detailed velocity model untill 100 m depth, highlighting possible lithological and mechanical contrasts as well as water preferential flow paths. The acquisition was carried out using 71 3C miniaturized seismic sensors buried at ca. 30 cm in the ground and spaced with an average intertrace of 70 m in accordance with slope topography. IGU16HR-3C 5Hz SmartSolo geophones of the DENSAR service (EOST) were used. The seismic array was recording continuously from June, 22nd to July, 21st 2020 at a sampling rate of 500 Hz. 370 controlled seismic sources were triggered at 122 locations using blank 12-gauge shotgun cartridges, Seismic Impulse Source Systemshots, 90-kg Propelled Energy Generator shots and a Mechatronics Lightning source generating P and S-waves with mono-frequency and sweep signals between 5 and 60 Hz of maximum 80 s length.  

We present the results of this active P-wave traveltime tomography. We first discuss the quality of the recorded signals related to each different type of source, given the noise and attenuation conditions at Viella. Because the signals were challenging to detect a methodology based on manual picking was used, supported by automatic detection tools and considerations regarding the network geometry in an a priori velocity model.  

The P-wave model was obtained using the inversion library pyGIMLI, which permits an accurate description of the topography, and provides a spatial discretization adapted to the problem. To supplement and constrain the interpretation of the P-wave velocity model, borehole information as well as a 3D resistivity model of the zone are available. With regards to these data, the geometric features and physical parameters of the main geological structures of the landslide are discussed.