Thèse Evolution Mécanique des Glissements de Terrains Profonds Sous Forcage Sismique et Superficiel H/F - Doctorat.Gouv.Fr

Établissement : Université Côte d'Azur École doctorale : SFA - Sciences Fondamentales et Appliquées Laboratoire de recherche : Laboratoire GEOAZUR Direction de la thèse : Louis DE BARROS ORCID 0000000255419162 Début de la thèse : 2026-10-01 Date limite de candidature : 2026-04-24T23:59:59 Les glissements de terrain constituent un processus géomorphologique majeur en milieu montagneux et représentent un risque naturel important, pouvant entraîner des effets en cascade comme la formation de barrages naturels ou des tsunamis en lacs alpins. Leur déclenchement est lié à des forçages de surface (pluie, température) qui diminuent progressivement la stabilité des pentes via des mécanismes hydrologiques et mécaniques, et à des forçages profonds, notamment les séismes, qui induisent des contraintes dynamiques favorisant l'endommagement interne des roches. Des facteurs anthropiques, comme les vibrations liées au trafic, peuvent également jouer un rôle local.

Si les déclencheurs principaux (précipitations intenses, séismes) sont bien identifiés, l'interaction entre forçages de surface et profonds dans la dégradation progressive des glissements profonds reste mal comprise. Des études récentes suggèrent que leur action combinée pourrait accentuer l'endommagement mécanique, mais les mécanismes précis demeurent incertains.

Les méthodes actuelles de surveillance reposent surtout sur les mesures de déplacement, efficaces pour détecter les phases d'accélération, mais limitées pour caractériser les processus internes comme l'endommagement ou la perte de rigidité. La sismologie offre une approche complémentaire, permettant d'observer l'état mécanique interne de manière continue. Les variations de vitesse des ondes sismiques (dv/v) renseignent sur les évolutions de l'état mécanique, liées notamment à l'endommagement diffus et aux conditions hydrologiques, tandis que les changements de fréquences de résonance (df/f) traduisent les variations de rigidité à l'échelle de la structure. Leur combinaison permettrait de mieux distinguer les effets locaux et globaux, mais reste encore peu développée.

Cette thèse vise à construire un cadre physique combinant ces deux approches afin d'étudier comment les forçages externes interagissent dans la dégradation mécanique des glissements profonds. Les hypothèses principales sont que l'action conjointe des forçages hydrologiques et sismiques amplifie les variations de dv/v, que les perturbations sismiques ont des temps de récupération distincts des effets saisonniers, et que les variations de df/f apportent des informations complémentaires sur la rigidité effective globale su système.

Les objectifs sont : (1) caractériser la géométrie et les propriétés mécaniques des glissements via l'analyse modale, (2) suivre les variations temporelles des propriétés élastiques en combinant dv/v et df/f, et (3) évaluer l'impact différencié des forçages climatiques, sismiques et anthropiques.

Deux sites sont étudiés. Le premier, Spitze Stei (Suisse), est un glissement profond, très instrumenté, situé à une altitude comprise entre 2000 et 2900 m. Il est soumis à des variations saisonnières marquées, favorisant les cycles de gel-dégel, la contribution de la fonte nivale et la dégradation du pergélisol, ainsi qu'à des séismes régionaux. Le second, Sainte-Marguerite (France), est un glissement actif affectant une infrastructure routière, influencé par des conditions météorologiques, un trafic intense et une activité sismique locale. Ces sites offrent des conditions idéales pour analyser les interactions entre forçages et améliorer la compréhension des mécanismes de déstabilisation. Landslides are among the most significant geomorphic processes shaping mountainous regions and constitute a major natural hazard. Beyond direct slope failures, they may generate severe cascading effects, including natural dam formation and tsunamis in alpine lakes. Surface forcings associated with meteorological and climatic processes progressively reduce slope stability through hydrological loading, pore-pressure increases, thermoelastic stress variations, and mechanical weathering. In contrast, deep forcing associated with seismic shaking impose transient dynamic stresses from depth, potentially inducing damage accumulation and stiffness reduction within the rock mass. In some settings, additional anthropogenic perturbations, such as persistent surface vibrations from road traffic, may locally contribute to slope dynamics.

Although the principal triggers of landslides, including intense precipitation and seismic shaking, are well established, it remains unclear whether surface and deep forcings act independently or interactively in driving the progressive mechanical degradation of deep-seated landslide. Recent observations (Bontemps et al., 2020) suggest that the combined action of hydraulic and repeated seismic perturbations may weaken slope materials more effectively than either forcing acting alone. However, the respective roles of surface and deep perturbations, their recovery dynamics, and the mechanisms through which they interact remain insufficiently constrained.

Current monitoring strategies rely primarily on displacement measurements and empirical triggering thresholds. While effective for detecting acceleration phases, these approaches provide limited insight into subsurface mechanical degradation, progressive damage accumulation, and stiffness evolution within unstable rock masses. As a result, the mechanical processes governing the transition from steady to accelerating creep remain poorly constrained within a physics-based framework.
Seismology offers new means to probe the internal mechanical state of unstable slopes in a continuous and non-invasive manner. Variations in seismic wave velocities derived from ambient noise interferometry (dv/v, Chmiel et al., 2019; Le Breton et al., 2021; Clements and Denolle, 2023). provide sensitive indicators of distributed changes in elastic properties and damage within the rock mass. Similarly, shifts in resonant frequencies (df/f, Mercerat et al., 2021; Finnegan et al., 2022; Grechi et al., 2025) obtained through modal analysis reflect structural-scale stiffness variations governed by geometry and boundary conditions.

While dv/v is particularly sensitive to distributed damage and microcrack evolution, resonance frequency shifts primarily capture bulk structural stiffness changes. Their joint interpretation therefore provides complementary constraints on mechanical state and damage evolution, allowing discrimination between localized perturbations and large-scale structural weakening. Although both approaches have demonstrated significant potential for landslide monitoring, they are most often applied independently. A unified interpretation framework combining dv/v and df/f remains underdeveloped, particularly for deep-seated landslide.

This thesis aims to develop a physics-based framework that combines relative seismic velocity changes and resonance frequency monitoring to investigate whether, and how, the interaction between surface and deep forcings contributes to progressive mechanical degradation and instability in deep-seated landslide. Our hypotheses are that: (1) The combined action of hydro-climatic loading and moderate seismic shaking produces larger-magnitude and longer-recovery reductions in dv/v than either forcing acting independently, (2) Co-seismic perturbations exhibit recovery timescales distinct from seasonal hydro-climatic variations, enabling separation of deep dynamic forcing from surface-driven effects, (3) Resonance frequency shifts (df/f) reflect bulk structural stiffness changes and therefore display different amplitudes and recovery dynamics compared to dv/v, providing complementary constraints on damage localization and structural weakening.

The research objectives are to:
1. Characterize landslide geometry and effective mechanical properties.
Using modal analysis and resonance frequency identification, the project will estimate characteristic vibration modes of a studied landslide. These modal parameters will be integrated with available geological and geomorphological constraints to infer thickness, bulk geometry, and elastic properties of the unstable masses.
2. Quantify temporal variations in elastic properties.
Relative seismic velocity changes (dv/v) will be estimated from ambient noise cross-correlation. Concurrent monitoring of resonance frequency shifts (df/f) will provide complementary sensitivity to variations in bulk stiffness. The combined analysis will allow detection of transient co-seismic perturbations, seasonal hydro-thermal effects, and longer-term mechanical evolution.
3. Evaluate the differential impact of distinct external forcings.
The effects of: climatic forcing (precipitation, temperature variations), moderate seismic shaking, and, where relevant, traffic-induced vibrations, will be evaluated through time-series analysis and event-based approaches. Quantitative metrics (e.g., cumulative rainfall, displacement, peak ground velocity) will be compared to observed dv/v and df/f variations. Particular attention will be given to identifying co-seismic velocity drops, seasonal thermoelastic effects, and potential nonlinear coupling between seismic damage and hydrological sensitivity.

By integrating resonance analysis and ambient noise interferometry, this thesis aims to establish a physics-based framework for monitoring stiffness degradation and damage evolution in large landslide. The results will improve understanding of how external perturbations influence slope stability and may contribute to the development of mechanically informed indicators of incipient failure.

Study Sites
1. Spitze Stei, Switzerland
The first study site is the Spitze Stei rock slope in the Swiss Alps, affected by a deep-seated landslide involving fractured bedrock and degrading permafrost. The unstable mass is estimated at approximately 16 million m³. Current displacement rates reach several meters per year, with seasonal accelerations in spring and summer exceeding 20-40 cm per day. During these periods, frequent rockfall activity occurs, typically involving volumes of a few thousand cubic meters and occasionally reaching up to 30,000 m³.
The site has been instrumented since 2021 with three semi-permanent seismic stations, three seasonal seismometers, and periodic nodal array deployments to enhance spatial resolution of the slope's dynamic response. These seismic observations are complemented by multi-parameter monitoring initiated in 2018, including borehole temperature and pore-pressure measurements, automatic camera surveys, surface displacement measurements from 13 GPS stations and 28 prisms, as well as seasonal interferometric and Doppler radar observations.
Spitze Stei is located in the canton of Bern, in proximity to the seismically active Valais region. Consequently, the slope is periodically subjected to moderate regional earthquakes, providing repeated dynamic perturbations that may influence its mechanical state. This tectonic context, combined with strong seasonal hydro-climatic forcing and permafrost degradation, makes Spitze Stei an exceptional natural laboratory for investigating the interaction between surface and deep forcings. The site further offers the opportunity to evaluate whether internal mechanical indicators derived from seismic monitoring provide added insight beyond displacement-based approaches for understanding slope dynamics and rockfall hazard evolution.

2. Sainte-Marguerite landslide (South of Briançon, Massif des Écrins)
The second study site is the active deep-seater Sainte-Marguerite landslide, located south of Briançon in the Massif des Écrins (Hauts-Alpes, France). The landslide has been monitored for more than 15 years by Cerema. The Route nationale 94 (RN94) crosses the landslide in a north-south direction and has been repeatedly and cyclically damaged, requiring continuous structural monitoring. Since 2019, boreholes equipped with inclinometers have recorded several acceleration phases under varying meteorological conditions.
The lithological context consists of meta-schists and quartzites overlying Jurassic dolomitic bedrock. Several active detachment levels have been identified through geophysical surveys. Traffic loading is particularly intense during certain periods of the year due to heavy trucks travelling toward and from Italy via the Fréjus Tunnel. In addition, an active seismic swarm located approximately 20 km east of the landslide generates local earthquakes with magnitudes up to Mw 3, providing a natural source of seismic forcing.

Le/la candidat recruté/e devra avoir un master en sciences de la terre, avec des compétences en géomécanique et en stabilité des pentes. Il/elle devra être autonome et montrer de bonnes capacités d'analyse et de réflexion pour prendre en main son sujet. Il/elle devra avoir des notions en sismologie, hydrologie et hydro-mécanique, traitement du signal et outils numériques. Un niveau suffisant en anglais pour permettre la présentation des résultats dans des conférences internationales et l'écriture d'article scientifique est aussi requis.