Geophysical inversion and machine learning of dense 3D seismic and resistivity models for imaging deep landslide structures

Myriam Lajaunie (1) , Céleste Broucke (2) , Jean-Philippe Malet (1, 2) , Julien Gance (3) , Clément Hibert (1, 2) , Catherine Bertrand (4) , G. Sénéchal (5) , Dominique Rousset (5) , Cathy Truffert (3) , Claire Bataillès (6) , Laurent Lespine (6)

1 ITES - Institut Terre Environnement Strasbourg
2 EOST - Ecole et Observatoire des Sciences de la Terre
3 IRIS Instruments
4 LCE - Laboratoire Chrono-environnement (UMR 6249)
5 LFCR - Laboratoire des Fluides Complexes et leurs Réservoirs
6 ONF - Office national des forêts

Geophysical imaging methods are often applied for analysing landslides; they can help detecting and localizing the landslide structures, distinguishing between lithologies, layers and slip surfaces of variable mechanical responses, and identifying the fluid storage and circulation paths. We present a joined analysis of two 3D geophysical models acquired from a dense electrical resistivity tomography (Fullwaver nodes) and a dense seismic travel-time tomography (DENSAR nodes) acquired at the Viella landslide (Hautes-Pyrenees, France). Fusion of the two geophysical models and of ancillary geotechnical and hydrogeological information from boreholes is carried out using machine learning, in order to predict the location of the different slope structures.

Mots clés: RESIF; Seismology; France; Epos-France

Mots clés: RESIF; Sismologie; France; Epos-France

Domaines: Géophysique [physics.geo-ph]

Myriam Lajaunie, Céleste Broucke, Jean-Philippe Malet, Julien Gance, Clément Hibert, et al.. Geophysical inversion and machine learning of dense 3D seismic and resistivity models for imaging deep landslide structures. 5èmes Rencontres Scientifiques et Techniques Résif, Nov 2021, Obernai (67210), France. ⟨hal-03481172⟩