CBCatherine Bertrand

Catherine Bertrand

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Catherine Bertrand has a total of 75 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    When slopes awaken: Detecting landslide activation phases with InSAR and optical offset tracking

    by Floriane Provost, Alexander L. Handwerger, Jean-Philippe Malet, Catherine Bertrand, Federico Di Paolo, Matteo Dall Amico

    The Aiguilles - Pas de l’Ours landslide located near the village of Aiguilles (South-east French Alps) is one of the largest and continuously active landslides in France, with an estimated area of 18 ha. The landslide was activated in late 2016, raising concerns that the slope collapse may cause a sediment dam on the Guil River downslope. We investigate the kinematics of the Aiguilles - Pas de l’Ours landslide from 2014 to 2020 using a multi-sensor approach combining satellite-based radar interferometry (InSAR), satellite-based multi-pairwise optical image correlation (OIC), ground-based synthetic aperture radar (GB-InSAR), and ground-based Global Navigation Satellite Systems (GNSS). We propose a strategy to invert the 3D displacement time series by combining satellite-based InSAR and OIC results. We show that merging these datasets into a 3D displacement time series significantly improves the accuracy of satellite-based displacement time series and facilitates the interpretation of complex landslide motion, in particular to detect the onset and duration of episodic accelerations. We determine the amount of rain on the ground, Rain-on-Snow (RoS), and snowmelt using in situ meteorological data, satellite-based maps of the snow cover and snow water equivalent (SWE) modelling to explain the onset of two successive episodic accelerations in winter 2016–2017 and spring 2018. Our findings highlight the importance of systematically combining several monitoring approaches to accurately capture large-scale landslides with local estimates of hydrological forcings to determine the onset of slope acceleration. These results offer perspectives for regional detection of slope episodic accelerations.

  • Open Access

    Acquiring, sharing and exploiting long-period and multi-parametric instrumental data for landslide science: the French EPOS-France vision, and its implementation at the European level.

    by Jean-Philippe Malet, Séverine Bernardie, Catherine Bertrand, Muriel Gasc, Stéphanie Gautier-Raux, Clément Hibert, Pascal Lacroix, Thomas Lebourg, Mathilde Radiguet, Maurin Vidal

    Documenting landslide activity over long periods using monitoring standards (sensors, acquisition rates, quality-control) is critical for understanding landslide forcing factors, validating process-based models, identifying the effect of climate change on their behavior, and ultimately defining warning thresholds. These goals underline the mission of the Thematic Group “Landslides” currently being set up among several French institutes (CNRS, BRGM, CEREMA, IRD) within the national Solid Earth Research Infrastructure EPOS-France.The thematic group has two objectives organized in two Specific Actions (SAs): SA#1 - setting up a permanent observatory of continuously moving large landslides using a multi-instrumented approachSA#2 – setting up an observatory of landslide hot moments, corresponding to monitoring campaigns of specific landslide acceleration periods in order to learn from such specific extreme events.The two SAs are built upon the sensor technologies and information system of the French Landslide Observatory (Observatoire Multi-Disciplinaire des Instabilités de Versants) which is a service of the French Research Institute (CNRS) in charge of deploying, acquiring, exploiting and disseminating multi-parametric sensor data over several large landslides. OMIV has developed, since nearly 20 years, standards in terms of sensor types, using both high-grade and low-cost sensing in order to construct reference and spatially dense monitoring time series. The service provides open access to records of landslide kinematics, landslide micro-seismicity, landslide hydro-meteorology and landslide hydro-geophysics. Combined, these categories of observations are unique worldwide for long-term landslide observations. OMIV is currently supervising the acquisition and dissemination of sensor data on 8 permanent unstable slopes (Avignonet/Harmallière, La Clapière, Séchilienne, Super-Sauze/La Valette, St-Eynard, Pégairolles, Vence, Villerville) and on unstable slopes currently experiencing gravitational crises (Viella, Marie-sur-Tinée, Aiguilles). The service is organized around the dissemination of qualified data (in international reference file format) and products for 5 categories of observation (Geodesy, Seismology, Hydrology, Meteorology, Hydrogeophysics). For each category of observation, specific FAIR data repositories and access portals are available and automated processing methods have been proposed to meet the needs of the landslide research community. The products being generated are time series of GNSS and total station positions, catalogue of endogenous landslide micro-seismicity, resistivity tomography datasets, and hydro-meteorological parameters. These observations aim at contributing at identifying the key controlling parameters of different landslide types (e.g. soft/hard rock, cohesion/friction, slip/fracture, localized/diffuse damage) and at monitoring their evolution in time and space (deceleration or acceleration according to the triggering factors, sliding- flowing transition).The objective is to present the strategy of data acquisition, qualification, dissemination and exploitation of the EPOS-France landslide thematic group, and discuss ideas to set up (at mid-term) a European landslide thematic core service (landslide-TCS) within EPOS ERIC.

  • Open Access

    Scalable XAI-based forecasting of landslide surface velocities from environmental forcings

    by Olivier Béjean-Maillard, Catherine Bertrand, Jean-Philippe Malet, Laurent Dubois, Claire Batailles, Laurent Lespine, Olivier Maquaire, Mathieu Fressard, Joshua Ducasse

    Forecasting the evolution of slow-moving landslides is a challenge because landslide motion is modulated by hydrometeorological forcing (rainfall, snowmelt, groundwater fluctuations) acting across multiple timescales, resulting in complex and strongly non-linear forcing–response relationships. By leveraging long-term multi-parameter monitoring, AI-based models can help characterise and simulate these dynamics. However, two limitations persist. First, many approaches rely on deep-learning architectures (RNNs, GRUs, LSTMs) that successfully reproduce non-linear dynamics, but do not constrain landslide physics and have limited interpretability and transferability. Second, few AI applications address landslides governed by the combined influence of multiple hydrometeorological drivers. Existing applications remain largely site-specific, relying on tailored predictor sets and local calibration. Addressing these limitations requires interpretable modelling frameworks capable of operating across multiple landslide sites including data-scarce settings.Here, we introduce a scalable and eXplainable Artificial Intelligence (XAI) modelling framework using eXtreme Gradient Boosting (XGBoost) and based on a set of 248 and physically grounded, non site-specific hydrometeorological predictors computed from net rainfall, effective rainfall, and groundwater level time series. Predictors are designed to represent three complementary aspects of landslide water-related forcing: (i) the hydrological state of the system, (ii) hydrological memory effects, and (iii) short-term hydrological transient processes. To capture multi-timescale hydromechanical dependencies, predictors are computed over multiple time windows ranging from 1 to 90 days. The approach simulates daily landslide velocities, evaluates predictive skill using RMSE and MAE metrics, and provides interpretable and explainable constraints on the predictor influence using features importance ranking and SHAP-based attribution tools.We evaluate the framework on three slow-moving landslides in France: Séchilienne (fractured miscaschist), Viella (morainic and colluvial deposits), and Villerville (chalk, sand and colluvial deposits ovelying marl substrate) spanning contrasting lithologies, deformation mechanisms and kinematics to demonstrate the scalability of the approach.The XAI framework accurately reproduces landslide velocity time series across sites and testing periods with small residual errors relative to the amplitude of observed velocity variations (Séchilienne, 0.005-0.015 cm.d-¹ ; Viella, 0.01-0.035 cm.d-¹ ; Villerville, 0.02-0.06 cm.d-¹). The identified predictors per landslide align with contrasting physical processes, including delayed pore-water pressure build-up driven by slow matrix infiltration in impermeable slope material (Villerville) and rapid responses to rainfall in more permeable (Viella) or fractured (Séchilienne) slope materials. Together, these results show that XAI frameworks can recover site-specific landslide behaviour while preserving physical interpretability across diverse settings, and demonstrate one of the first applications of a common model structure and non-site-specific predictor set across multiple distinct landslide case studies.

  • Open Access

    Probabilistic Simulation of Monthly Landslide Velocity Under Hydro-meteorological Variability

    by Catherine Bertrand, Olivier Maillard-Bejean, Jean-Philippe Malet, José Moya, Olivier Maquaire

    Hydrometeorological forcing (rainfall, snowmelt, groundwater fluctuations) acts across multiple timescales and is a primary driver of surface velocity dynamics in slow-moving landslides. Many studies use trained AI-based models to simulate daily-to-monthly velocities over validation periods defined by specific historical hydrometeoroligical contexts. Although these models achieve accurate predictive skill, they are typically deterministic and therefore provide limited insight into the range of plausibly velocity responses under alternative, yet realistic, forcing conditions.To address this gap, we introduce a probabilistic framework built around two axes. Forcing variability is represented by generating 500 plausible meteorological time series using a modified Richardson-type weather generator (rainfall and air temperature). These series are then propagated through a transfer-function hydrological model to simulate groundwater-level variability driven by generated effective rainfall. Second, daily velocities are simulated using a trained XGBoost model based on a set of hydrometeorological predictors. The resulting ensemble is summarised as monthly velocity distributions over a one-year horizon, thereby capturing distinct dynamics across a full hydrological cycle. Distributional performance is evaluated using the Prediction Interval Coverage Probability (PICP) and the Mean Interval Score (MIS).We evaluate the framework on three slow-moving landslides spanning contrasting lithologies, deformation mechanisms and kinematics: Viella (morainic and colluvial deposits ; France), Villerville (chalk, sand and colluvial deposits ovelying marl substrate ; France), and Vallcebre (clayey siltstone and colluvial debris overlying limestone substrate), to demonstrate the scalability of the approach.The modified Richardson-type generator reproduces key statistical properties of historical meteorological records. Calibrated groundwater models capture the main dynamics of groundwater fluctuations, with R2 values of 0.84 (Viella), 0.76 (Villerville) and 0.53 (Vallcebre). The simulated monthly velocity distributions exhibit clear seasonality, with more contrasted annual cycles at Viella and Villerville, consistent with site-specific hydrogeological behaviour. On average, prediction intervals encompass a substantial fraction of observed monthly velocities (mean PICP: 53% for Viella, 40% for Villerville and 76% for Vallcebre), with strong variability across months. Remaining discrepancies mainly reflect data availability, limitations in groundwater simulations, and constraints in the learned forcing–velocity relationships within the XGBoost model, highlighting priorities for further methodological improvements. Overall, the proposed framework provides a first practical tool to quantify the range of probable landslide-velocity responses under multiple plausible hydro-meteorological scenarios.

  • Hydrogeological forecasting of deep-seated landslides dynamics: structure and sensitivity of tank models

    by Olivier Béjean-Maillard, Catherine Bertrand, Jean-Philippe Malet, Guillaume Cinkus, Pierre Nevers, Joshua Ducasse, Davide Bertolo, Patrick Thuegaz


  • Open Access

    Forecasting landslide motion with EXplainable Machine Learning models: the use case of Séchilienne landslide (French Alps) to identify the relevant predicting variables

    by Olivier Maillard, Catherine Bertrand, Jean-Philippe Malet

    Recent works on landslide displacement forecasting using machine learning or deep learning models show relevant performance. However, they are mostly based on the use of historical displacement information and do not provide information on the most predictive features in terms of meteorological and hydrogeological variables for the forecast, and thus the identification of possible precursory factors. In this context, providing approaches based on EXplainable Machine Learning (XML) is essential for landslide forecasting as it concerns making decisions about risk mitigation actions, it supports the identification of possible precursory factors and it increases confidence in the predictions.The proposed XML-based landslide forecasting approach is developed and tested using ensemble learning methods such as Random Forest and XGBoost. It relies on the use of multi-year and multi-parameter data chronicles to analyse the relationships between surface displacements (target data) and hydro-meteorological conditions (predictor data). Displacement and meteorological data are acquired through the landslide monitoring network. Hydrological data, when not available, are simulated discharge calculated with reservoir based-model; the simulations allow to construct water level time series for each water reservoirs identified in the unstable slope. The predictive time series are decomposed into a set of 340 descriptive features (mean, variance, difference, number of rainy days, number of consecutive rainy periods of X days, …). The displacement time series are detrended using the multiplicative decomposition method.This method has been applied to several use cases, such as the Séchilienne landslide located southwest of the Belledonne massif (French Alps). The Random forest and XGBoost models are trained and tested over periods of 12 and 5 years respectively, and applied to three automatic extensometers located in the most active part of the landslide. The results indicate that the main features used include variations in water levels over past 10 to 30 days, as well as the number of consecutive rainy period during the month. These results are associated with accurate predictions for the three extensometers, with coefficients of determination ranging between 0.37 and 0.46.We show that these models have high predictive power while informing about the most important hydro-meteorological features. The application of the models to trendless displacement time series significantly improves prediction accuracy.

  • Open Access

    The OMIV service: acquiring and sharing long-period instrumental time series for documenting landslide activity

    by Jean-Philippe Malet, Catherine Bertrand, Clément Hibert, Mathilde Radiguet, Thomas Lebourg, Stéphanie Gautier, Grégory Bièvre, Maurin Vidal, Xavier Wanner, Candide Lissak, et al.

    Documenting landslide activity over long periods and monitoring standards (sensors, acquisition rates, quality-control) is critical for understanding the landslide forcing factors, develop process-based models, identify the effect of climate change on their behavior, and ultimately define warning thresholds.The French Landslide Observatory (Observatoire Multi-Disciplinaire des Instabilités de Versants) OMIV is the service of the French Research Institute (CNRS) in charge of deploying, acquiring, exploiting and disseminating multi-parametric sensor data over several large landslides in France. OMIV has developed, since more than 15 years, standards in terms of sensor types, using both high-grade and low-cost sensing in order to construct reference and spatially dense monitoring time series. The service provides open access to records of landslide kinematics, landslide micro-seismicity, landslide hydro-meteorology and landslide hydro-geophysics. Combined, these four categories of observations are unique worldwide for long-term landslide observations. OMIV is currently supervisizing the acquisition and dissemination of sensor data on 8 permanent unstable slopes (Avignonet/Harmallière, La Clapière, Séchilienne, Super-Sauze/La Valette, St-Eynard, Pégairolles, Vence, Villerville) and on unstable slopes currently experiencing gravitational crises (La Clape, Viella, Marie-sur-Tinée, Aiguilles). The service is organized around the dissemination of qualified data (in international reference file format) and products for 5 categories of observation (Geodesy, Seismology, Hydrology, Meteorology, Hydrogeophysics). For each categories of observation, specific FAIR data repository and access portals have been developed and automated processing methods have been proposed to meet the needs of the landslide research community. The products being generated are time series of GNSS and total station positions, catalogue of endogeneous landslide micro-seismicity, resistivity tomography datasets, and hydro-meterological parameters).OMIV provides consistent and harmonized landslide monitoring data in order to identify the physical processes that control the landslide dynamics, both for slopes affected by slow-moving slides and cliffs affected by rockfalls, use these datasets to develop and validate landslide deformation/propagation models, extract (from the long-term observations) the patterns that may characterize changes in the landslide dynamics (annual, seasonal, event) and propose possible forerunners. The OMIV observations aim at contributing at identifying the key controlling parameters of different landslide types (e.g. soft/hard rock, cohesion/friction, slip/fracture, localized/diffuse damage) and at monitoring their evolution in time and space (deceleration or acceleration according to the triggering factors, sliding- flowing transition).The objectives are to present the OMIV datasets, sensing standards and automated processing methods that has been developed, both for the science community and for operational partners in charge of landslide risk management (ONF-RTM, BRGM, CEREMA), for some of the monitored landslides. The objectives are also to present the future directions of the service with a focus on the modelling of the landslide processes using both process-based and machine learning approaches.

  • Open Access

    Ambient noise shear-wave tomography for shallow landslide structural models retrieval from dense 3D seismological arrays.

    by Myriam Lajaunie, Joachim Rimpot, Dimitri Zigone, Céleste Broucke, Jean-Philippe Malet, Elise Weiskopf, Clément Hibert, Joshua Ducasse, Catherine Bertrand

    The versatility, cost-efficiency and easy deployment of seismic sensor nodes facilitate geophysical monitoring in environments that were previously inaccessible for instrumentation, and among them landslides and unstable slopes, most of the time located in remote mountains. Using nodes allows for the setup of dense arrays with sensor inter-trace distances that become compatible with the geometries and dimensions of the geological structures to image. This becomes particularly true for landslides which have complex 3D architectures (hummocky bedrocks, layering, multi-dimensional fractures, diverse geotechnical material, deep and perched aquifers and water circulations) and are shallow processes with respect to the classical investigation depths and sensitivity of most geophysical survey techniques.Here we develop a specific processing workflow to allow the computation of 3D shear-velocity models with Ambient-Noise-based tomography applied to dense arrays of seismic stations. The workflow is applied to a dataset acquired at the Viella shallow landslide (France) developed in altered schists and moraine deposits. We deployed 70 IGU-16HR-3C-5Hz SmartSolo sensors (EOST/PISE service) with inter station distances of 70 m for a period of 25 days.The processing consists in several steps, all of them being tuned to the specific case of shallow depths of investigation. In areas where only few strong (ML>4) earthquakes are triggered, with a low azimuthal distribution, surface-waves velocity fields are complex to estimate with earthquakes. Ambient noise cross-correlation tomography has the advantage of using the ambient noise to model the surface waves velocities by retrieving the interstation Green’s functions. The main hypothesis for retrieving the Green’s functions is a homogeneous noise-source distribution, which is never achieved in a natural environment. Therefore, data filtering and daily stacking are crucial to reduce the effect of non-uniform noise distributions and lead to consistent velocity models. Due to the noisy environment of Viella (torrential flows, farming activity, anthropogenic noise), several procedures were implemented to optimize the processing (reduction of the coherent noises in the processed data, use of a pseudo-topography to estimate as accurately as possible the inter-station distances and travel times). We then computed the dispersion curve diagrams for the surface waves on which we applied a strict selection to only keep the consistent part of the surface waves dispersion curves. The selection parameters were optimized for the Rayleigh and Love waves. Then, we inverted the inter-station travel times to compute group velocities maps at several frequencies. Finally, we proceed to a Markov-Chain-Monte-Carlo inversion of each of the dispersion curves extracted from the group velocity maps. We finally obtained a 3D shear velocity model, which is further combined with geological and borehole information in order to document the 3D structure.The objectives are to present the processing workflow developed specifically for shallow imaging and the retrieval of 3D heterogeneities; effects of the processing parameters will be discussed on the Viella dataset. The approach developed for Viella is generic and has been further applied to other geological processes (permafrost at the Chauvet rock glacier, Marie-sur-Tinée mudslide), and the models will be discussed.

  • Open Access

    Inventory and characterization of recent (<100 years) gravitational activity of the Queyras DSGSDs - South French Alps

    by Clément Boivin, Jean Philippe Malet, Catherine Bertrand, Yannick Thiery

    Deep Seated Gravitational Slope Deformation (DSGSD) are gravitational processes damaging slopes over long periods of time. These processes may be reactived with the occurrence of smaller, shallow gravitational events. Thus, a better understanding of DSGSDs, from their formation to more catastrophic phases of activity, is an important goal  for natural hazard prevention in mountainous areas. .A first inventory of DSGSD in the Western Alps has been proposed by Crosta et al. (2013) with 1057 DSGSDs identified. A similar work has been conducted more recently at the scale of the French Alps by Blondeau (2018) who identified nearly 460 DSGSDs. Despite the importance of these works, there are still many Alpine sub-massifs where high concentrations of DSGSDs (Blondeau., 2018) have been recognized but where no detailed studies have been conducted. This is the case of the Queyras Massif (South French Alps). It is in this context that this study is carried out, with both the objectives of locating and characterizing the DSGSDs observed in this area and identifying their recent activity.The proposed approach is based on quantitative geomorphological studies combining photo-interpretation of multi-date aerial imagery, analysis of DSMs and field observations. Quantitative description criteria are proposed to identify DSGSDs and discriminate them from large deep-seated landslides. Thirty DSGSDs are inventoried and their lithological and structural setting is analyzed. Analysis of multi-date aerial photographs and InSAR derived landslide velocities (NSBAS processing of Sentinel-1 observations; e.g. André et al., XX?) allow characterizing their gravitational activity.

  • Open Access

    Hydrogeological modelling of the Viella landslide (Hautes-Pyrénées) for hazard understanding

    by Joshua Ducasse, Catherine Bertrand, Olivier Maillard, Jean-Philippe Malet, Myriam Lajaunie, Sylvain Benarioumli, Claire Bataillès, Laurent Lespine

    Following a rockslide in 2018, a landslide was reactivated affecting the town of Viella in the Hautes-Pyrénées (South France). The slope movement threatens road infrastructure and buildings. The landslide is in the Bayet-Badoueil watershed. This torrential stream has its source on the heights of Viella and has the Bastan river as its outfall. The Bastan flows at the toe of the landslide. The landslide is compartmentalized and covers an area of about 50 ha. he is composed of a morainic floor on which colluvium and scree have been deposited following the dismantling of the mountain above Viella. The whole thing rests on a Devonian substratum. The colluviums are composed of schists and limestones (Devovian). The study aimed at improving the state of knowledge of the Viella landslide to better manage the natural disaster. Water circulation within the massif is the motor of the sliding. Modelling the hydrogeological conditions allow better understanding the phenomena and will help to design mitigation solutions. A three-dimensional geological model was built as a prerequisite of the hydrogeological modelling with the 3D GeoModeller software. The model was built from the geological map, the logs of the fifteen drillings, including eight piezometers and seven inclinometers, as well as 3D geophysical models (3D resistivity model, 3D P-wave velocity model). The heterogeneity of the colluvium was simplified into two layers to locate the rupture surface at the interface of these layers. The depth of the rupture surface in relation to the topographic surface varies from a few meters below the Bayet to 55 meters deep at the I10 inclinometer. The construction of the geological model makes it possible to improve knowledge of the local structures and to propose geometry for the formations and the position of the rupture surface. The realization of a three-dimensional finite element water flow model, built from the geological model and an electrical resistivity model, with the software FEFLOW (©DHI) provides an understanding of the functioning of the landslide aquifer. This integrative approach on hydrogeological modeling makes it possible to propose a robust model which made it possible to establish the piezometric map of the site at equilibrium. In the landslide, the piezometry is between 780m and 970m the general orientation of the groundwater flow is about 340° north. The hydraulic conductivities determined by the model are between 10-4 and 10-5 m.s-1 in the colluvium under the village. From the calibrated model, various simulations were carried out to estimate the impacts of mitigation works on the water storage and circulation. It further helped to simulate the piezometric response of the slope to a flood event at the toe of the landslide. Model simulations showed that the (“sealing” or “waterproofing”) of a 650m section in the lower part of the Bayet-Badoueil stream would lower the piezometric height under the village to a maximum of 30m and reduce the hydraulic load upstream of the landslide. A decrease of 5 to 10 meters seems achievable and would be sufficient to significantly reduce the sliding kinematics.