JRJoachim Rimpot

Joachim Rimpot

*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


Joachim Rimpot has a total of 45 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Automated exploration and analysis of seismological data streams with advanced AI-based technologies

    by Joachim Rimpot

    Exploration automatisée et analyse de flux de données sismologiques avec des technologies avancées basées sur l’IA: La sismologie environnementale étudie les signaux produits par des structures géologiques telles que glissements de terrain, volcans et glaciers. Les réseaux denses de stations fournissent aujourd’hui des enregistrements très détaillés de la sismicité, mais génèrent des volumes de données si importants que les méthodes d’analyse traditionnelles deviennent insuffisantes. Cette thèse propose des approches d’exploration exhaustive de la donnée, sans connaissance a priori, fondées sur l’apprentissage auto-supervisé et le clustering. En représentant la donnée en images ou en séries temporelles multicomposantes, les méthodes développées permettent de traiter efficacement des jeux de données massifs issus de réseaux denses ainsi que des ensembles multi-environnement. Ces travaux ouvrent de nouvelles perspectives pour l’analyse des données sismologiques et pour une compréhension plus fine et plus globale de la complexité des phénomènes environnementaux.

  • Open Access

    Unveiling Exotic Seismic and Hydroacoustic Events Associated with the Activity of the Fani Maoré Submarine Volcano, Mayotte, Indian Ocean

    by Lise Retailleau, Joachim Rimpot, Jean-Marie Saurel, Clément Hibert

    On 10 May 2018, a seismic crisis occurred near Mayotte (Comoros Archipelago, Indian Ocean), and was followed by a significant ground deformation and eastward displacement of the island. Combined with the records of very-long period (VLP) earthquakes, these observations were the first signs of a major submarine volcanic eruption, Fani Maoré, discovered in 2019 approximately 50 km east of Mayotte. The deployment of onshore seismometers and Ocean Bottom Seismometers (OBS) network since 2019 has allowed a better characterisation and classification of the seismicity, including various types of seismic and acoustic signals as volcano-tectonic (VT) earthquakes, long-period (LP) earthquakes, VLP, and hydro-acoustic signals.In this study, we further investigated the seismic and hydroacoustic activity associated with the Fani Maoré submarine volcano by focusing on the characterization of rare and unconventional seismic signals that are poorly or previously not documented. We concentrated on the October–November 2019 period, when the OBS network was dense and included an OBS station deployed close to the active lava flows. To analyze this dataset, we applied a self-supervised learning (SSL) approach to four-channel time series data, comprising three-component seismometer recordings and a colocated hydrophone.The SSL-based analysis resulted in the identification of multiple clusters, revealing distinct groups of seismic and acoustic signals. By aggregating these clusters into broader families, we distinguished signals originating from non-volcanic sources (e.g., motor induced signals and whale vocalizations), and families of events clearly associated with the volcanic activity of the Fani Maoré submarine volcano. These include VT and LP earthquakes, impulsive HA signals, drumbeat-like events, tremor-like events and additional signal types whose sources remain uncertain but are likely related to volcanic processes. These results bring new insights in the dynamic of the Fani Maoré volcano, and will allow a better characterization of the involved mechanisms in the seismo-acoustic activity, and improve its monitoring.

  • Open Access

    A Time-Series-Based Self-Supervised Learning Approach for the Exploration of Complex Seismological Datasets: Application to the Fani Maoré Submarine Volcano

    by Joachim Rimpot, Lise Retailleau, Jean-Marie Saurel, Clément Hibert, Jean-Philippe Malet, Germain Forestier, Jonathan Weber

    The exploration and characterization of complex continuous seismological datasets remain challenging, particularly for highly active and/or noisy environments. Recently, several artificial intelligence based approaches have been proposed to facilitate the analysis of seismological data, either by characterizing detected events or continuous streams. Among these, we introduced an image-based self-supervised learning framework to explore continuous seismic records without requiring prior labeling or event detection. However, image-based representations may result in a loss of information, as they are derived transformations of the original raw seismic time series and may impact the discrimination of seismic events.In this study, we adapted a self-supervised learning based clustering workflow to operate directly on multichannel seismic time series. The main challenge when using self-supervised contrastive learning approaches with time series is adapting the data augmentation techniques to ensure sufficient transformation without losing the physics contained in the seismological records. We leveraged the contrastive learning framework to analyse two months of continuous records from Ocean Bottom Seismometers deployed near the Fani Maoré submarine volcano, using data augmentation strategies consistent with seismological records, such as channel masking and window masking in the time and frequency domains. The model was trained using four-channel time series derived from the raw data (three-component seismometer and one hydrophone) using 60 s sliding windows with a 50% overlap, enabling the network to learn meaningful latent representations of the data. Clustering was then performed directly within the learned latent space, allowing the identification of distinct signal groups. Applied to the Fani Maoré dataset, this approach revealed several families of clusters, including very rare and previously undocumented events likely associated with the activity of the Fani Maoré submarine volcano.

  • Open Access

    Dynamics and precursors of the 2025 Blatten rock–ice avalanche: Integrating seismic analysis, granular flow simulations, and field observations

    by Jiahui Kang, Antonie Lucas, Anne Mangeney, Johan Gaume, Kate Allstadt, Clément Hibert, Liam Toney, Hervé Vicari, Michael Dietze, Mylène Jacquemart, et al.

    Cascading slope failures in high mountain environments are observed with increasing frequency as glaciers retreat and slope stability is impacted by warmer conditions. On 28 May 2025, a large rock-ice avalanche (~9.3x106 m3) originating from Birch Glacier, Switzerland, destroyed parts of the village of Blatten, and provided a rare, well-documented case of a rapid, highly mobile mass movement.We combine seismic observations, geomorphological mapping, grain size and permeability measurements, and granular flow modelling to reconstruct the evolution of this event, from precursory instabilities to the main collapse. Seismic data scanned with machine learning algorithms reveal a two-week period of increasing rockfall and small glacier failures preceding the main collapse. The main collapse was reconstructed using force history inversion of low-frequency seismic signals from Switzerland’s national seismic network. Numerical simulations constrained by both seismic data and observed deposit extents indicate that an exceptionally low effective basal friction was required to reproduce the observed deposit extent and force history. This and the field observations of low-permeability deposit materials indicate that frictional weakening contributed to the unexpectedly high mobility of the main event.Our results highlight the value of integrating seismic monitoring with field and modelling approaches to constrain the dynamics of complex rock-ice avalanches. The Blatten event illustrates how large alpine slope failures can transition into highly mobile flows. Our study provides one of the first detailed reconstructions of this hazard cascade, including precursory failure activity, and the dynamics and frictional characteristics of the main event. The frictional weakening inferred here provides a much-needed mechanistic basis for predicting runout and deposit geometry in large debris avalanches.

  • Open Access

    Unsupervised Learning for the Comprehensive Exploration of Continuous‐DAS Data

    by C. Huynh, J. Rimpot, C. Hibert, A. Turquet, T. Stangeland, J.‐P. Malet, V. Lanticq

    Distributed acoustic sensing (DAS) enables cost‐effective, dense detection of seismic activity. However, the vast amount of data produced by DAS systems presents a significant challenge for labeling and analysis. Traditional supervised machine learning approaches require extensive labeling, which is time‐consuming and prone to user bias. Our approach meets the challenge of reducing the workload required for annotation, regardless of the size of the data set and without any a priori of the content of the data set, while preserving rare seismic events. We propose a two‐step processing chain. The first step constructs a latent data representation from several hundred features. We compare two approaches: one using signal processing metrics commonly used in seismology (human‐engineered features) and the other using self‐supervised learning with common DAS data representations (image‐BYOL, as known as Bootstrap Your Own Latent). The second step applies unsupervised clustering to reduce the data set. We first apply K‐Means to obtain 5,000 clusters followed by hierarchical clustering, merging them into 500–700 clusters using an inconsistency criterion. This dual‐step approach capitalizes on the computational efficiency of K‐Means and the hierarchical granularity offered by agglomerative clustering. This method is applied to DAS data from two experiments in the Pyrenees with different configurations: a 6‐week measurement on an 800‐m cable at Viella and 19 ten‐minute measurements on a 91‐km cable. For Viella, we detected all events with . The image‐BYOL approach produces more false positives than human‐engineered features. The results highlight the potential of clustering for DAS analysis while emphasizing the need to reduce false positives especially for smaller seismic events.


  • Open Access

    Self‐Supervised Learning Based Clustering Workflow for Exploring Seismological Data From Dense Networks

    by Joachim Rimpot, Clément Hibert, Jean‐Philippe Malet, Germain Forestier, Jonathan Weber

    A key challenge in environmental seismology is processing seismic data to study source physics, natural and human‐induced forcings, and geological structures such as landslides, glaciers, and volcanoes. Seismic arrays with dozens of stations have expanded data set sizes, and this, combined with signal complexity and high noise levels, makes it difficult to analyze using traditional event detection and labeling methods especially for low‐energy or rare events. Clustering continuous data offer a comprehensive method for exploring the data sets and detecting all relevant events. In this study, we present a clustering workflow based on self‐supervised learning (SSL) designed to handle data sets ranging in size from thousands to millions of events. This approach enables automated clustering of continuous data from seismic arrays containing dozens of stations. When applied to the “Marie‐sur‐Tinée” landslide data set, our workflow processed 10 millions 30‐s windows and identified four main families: Potential Endogenous Landslide Seismic Events, Potential Regional Earthquakes, Potential Rainfall‐Induced Signals, and Noise. Despite the overall consistency, some noise remained in the event‐related clusters highlighting areas for further improvement in clustering methods. Nevertheless, the proposed SSL‐based clustering workflow shows great potential for an efficient exploration of seismic data sets of millions of events and could be a solution for the blind exploration of similarly large data sets.


  • Open Access

    Unsupervised Learning for the Comprehensive Exploration of Continuous-DAS Data

    by Camille Huynh, Joachim Rimpot, Clément Hibert, Antoine Turquet, Tord Sture Stangeland, Jean-Philippe Malet, Vincent Lanticq

    Distributed Acoustic Sensing (DAS) enables cost-effective, dense detection of seismic activity. However, the vast amount of data produced by DAS systems presents a significant challenge for labeling and analysis. Traditional supervised machine learning approaches require extensive labeling, which is time-consuming and prone to user bias. To address this, we propose a two-step processing chain. The first step constructs a latent data representation from several hundred features. We compare two approaches: one using signal processing metrics commonly used in seismology (human-engineered features) and the other using self-supervised learning with common DAS data representations (image-BYOL). The second step applies unsupervised clustering to reduce the dataset. We first apply K-means to obtain 5000 clusters, followed by hierarchical clustering, merging them into 500–700 clusters using an inconsistency criterion. This method is applied to DAS data from two experiments in the Pyrenees with different configurations: a 6-week measurement on an 800-m cable at Viella and 19 ten-minute measurements on a 91-km cable. For Viella, we detected all events with Mw >2.0. The image-BYOL approach produces more false positives than human-engineered features. The results highlight clustering’s potential for DAS analysis while emphasizing the need to reduce false positives, especially for smaller seismic events.


  • Open Access

    Uncovering environmental and other exotic seismic sources with machine learning

    by Clément Hibert, Joachim Rimpot, Camille Huynh, Charlotte Groult, Jean-Philippe Malet, Germain Forestier, Jonathan Weber, Camille Jestin, Vincent Lanticq, Floriane Provost, et al.

    Seismology, beyond the study of earthquakes, has become an indispensable tool for understanding environmental changes, offering unique insights into a wide range of phenomena and natural risks, from slope instabilities to glacial dynamics and hydrological hazards. However, the sheer volume and complexity of modern seismic datasets, amplified by the emergence of dense seismic networks and technologies such as Distributed Acoustic Sensing (DAS), pose significant challenges. Recent advances in Artificial Intelligence (AI) and Machine Learning (ML) have revolutionized our ability to analyze these datasets, enabling a deeper exploration of seismic data to find rare and exotic environmental seismic sources. Supervised learning approaches have been successfully applied to create large-scale instrumental catalogs of landslides and other environmental processes, at different spatio-temporal scales, from short-term datasets recorded on dense local seismic stations networks, to chronicles spanning decades on seismic networks covering whole regions of the world (Alaska, Alps, Greenland). These techniques achieve high detection rates and robust classification of seismic events, even for low-magnitude or rare signals that traditional methods might overlook. Supervised learning approaches also allow us to advance our capability to estimate physical properties from seismic waves, such as the use of machine learning to infer mass and kinematics of slope instabilities, which provide critical inputs for understanding the dynamics of these events and their associated hazards. These methodologies not only allow us to document environmental processes more exhaustively but also open up possibilities for studying poorly understood or previously undetectable seismic sources. Going beyond supervised learning, we have developed workflows based on self-supervised and unsupervised approaches to analyze continuous seismic data, uncovering unexpected patterns and revealing hidden environmental seismic sources recorded by dense seismic stations networks. Distributed Acoustic Sensing represents another frontier, turning fiber optic cables into dense seismic networks. By combining DAS with innovative AI-driven methods, we have demonstrated the potential to detect and classify low-magnitude earthquakes and anthropogenic sources, even in noisy environments, paving the way for real-time seismic monitoring on unprecedented scales.By applying these AI-driven approaches, we are enhancing the field of environmental and exotic sources seismology, improving our ability to analyze vast seismic archives, and offering new tools to monitor, understand, and mitigate geohazards in a changing environment. This talk will highlight the latest methodological advances and showcase how they are applied to various geological and environmental contexts, from landslides, avalanches and glaciers in the Alps to fiber optic networks at different scales, underscoring the far-reaching implications of AI for seismological sources identification.

  • Open Access

    Unsupervised Learning for In-Depth Analysis of Continuous DAS Data

    by Camille Huynh, Joachim Rimpot, Clément Hibert, Antoine Turquet, Tord Stangeland, Jean-Philippe Malet, Vincent Lanticq

    Distributed Acoustic Sensing (DAS) enables seismic monitoring by transforming fiber optic cables into dense, cost-effective sensor arrays. However, the vast data volume generated by DAS presents challenges for labeling, sometimes even making data labeling more time consuming than processing and research. Traditional supervised machine learning methods require extensive manual labeling for individual events, which is both time-consuming and susceptible to user bias.To address these challenges, we propose a clustering-based approach to group similar data, allowing for cluster-level labeling rather than event-by-event annotation. Our method employs a two-step processing chain denoted (a) and (b). In the step (a), data is represented in a latent space defined by hundreds of features. Two approaches for constructing this latent space are explored: one using human-engineered features based on seismological signal processing, and the other leveraging self-supervised learning via the image-BYOL algorithm, which utilizes bidimensional representations of DAS data. The step (b) applies unsupervised clustering, initially reducing the dataset to 5000 clusters using K-Means partitioning algorithm, followed by hierarchical clustering to condense these into 500–700 interpretable clusters using an inconsistency criterion.This method was applied to two DAS datasets collected in the Hautes-Pyrénées. The first dataset involved six weeks of continuous measurements along an 800-m cable in Viella, recorded with a temporal resolution of 400 Hz, a gauge length of 10 m, and a channel spacing of 2.4 m. The second dataset consisted of 19 ten-minute recordings along a 91-km cable, with a temporal resolution of 200 Hz, a gauge length of 10 m, and a channel spacing of 4.8 m. Using cluster-based labeling on the Viella dataset, we successfully detected 100% of earthquakes with a magnitude Mw>2.0 and identified the daily periodicity of anthropogenic events, such as those related to farming activities. Continuous and long-duration (>30 s) seismic signals, primarily generated by mechanical farming engines, demonstrated a clear periodicity, whereas impact-driven or impulsive events were less consistent in timing due to their diverse origins.These findings highlight the potential of clustering techniques to analyze DAS data efficiently, reducing reliance on manual event labeling. Nevertheless, further improvements are necessary to minimize false positives, particularly for smaller seismic events.

  • Open Access

    Self-supervised learning based clustering workflow for exploring seismological data from dense networks

    by Joachim Rimpot, Clément Hibert, Jean-Philippe Malet, Germain Forestier, Jonathan Weber

    A key challenge in environmental seismology is processing seismic data to study source physics, natural and human-induced forcings and geological structures such as landslides, glaciers, and volcanoes. Seismic arrays with dozens of stations have expanded dataset sizes, and this, combined with signal complexity and high noise levels, makes it difficult to analyze using traditional event detection and labeling methods, especially for low-energy or rare events. Clustering continuous data offers a comprehensive method for exploring the datasets and detecting all relevant events. In this study, we present an iterative clustering workflow based on self-supervised learning (SSL) designed to handle datasets ranging in size from thousands to millions of events. This approach enables automated clustering of continuous data from seismic arrays containing dozens of stations. When applied to the ”Marie-sur-Tinée” landslide dataset, our workflow processed 10 millions 30-second windows and identified four main groups: Potential Endogenous Landslide Seismic Events, Potential Regional Earthquakes, Potential Rainfall-Induced Signals, and Noise. Despite the overall consistency, some noise remained in the event-related clusters, highlighting areas for further improvement in clustering methods. Nevertheless, the proposed iterative SSL-based clustering workflow shows great potential for an efficient exploration of seismic datasets of millions of events and could be a solution for the blind exploration of similarly large datasets.