Machine learning prediction of the mass and the velocity of controlled single-block rockfalls from the seismic waves they generate

Aug 5, 2026

Upload by Violet

Understanding slope-instability dynamics is essential for hazard mitigation, yet direct observation is often constrained by remote locations and the spontaneous nature of events. Seismology provides distinctive information, but the relationship between event properties—mass and kinematics—and the seismic signals remains limited. A controlled single-block rockfall experiment in the Riou Bourdoux torrent used a dense seismic network and 3D trajectory reconstruction via high-resolution stereophotogrammetry. Machine learning predictions estimate block mass and velocity with average errors near 10% for velocity and 25% for mass, comparable or better than other methods, without requiring source localisation or a high-resolution velocity model.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages16

Reading Time40 min

AccessFree to download

Abstract

Introduction

Controlled experiment and data acquisition

Machine learning prediction approach

Results and comparison

Implications for seismic correlations and future models

Earth Surf. Dynam., 12, 641–656, 2024 [https://doi.org/10.5194/esurf-12-641-2024](https://doi.org/10.5194/esurf-12-641-2024)[ ](https://doi.org/10.5194/esurf-12-641-2024)© Author(s) 2024 . This work is distributed under the Creative Commons Attribution 4 .0 License. Machine learning prediction of the mass and the velocity of controlled single-block rockfalls from the seismic waves they generate Clément Hibert 1 , François Noël2,3 , David Toe5 , Miloud Talib 1 , Mathilde Desrues 1 , Emmanuel Wyser2 , Ombeline Brenguier6 , Franck Bourrier4 , Renaud Toussaint 1,7 , Jean-Philippe Malet 1 , and Michel Jaboyedoff2 1Institut Terre et Environnement de Strasbourg/ITES, CNRS & University of Strasbourg, 67084 Strasbourg, France 2Institut des Sciences de la Terre/ISTE, University of Lausanne, Géopolis, 1015 Lausanne, Switzerland 3 Geological Survey of Norway, 7491 Trondheim, Norway 4Université Grenoble Alpes, INRAE, ETNA, 38000 Grenoble, France 5Université Grenoble Alpes, INRAE, LESSEM, 38000 Grenoble, France 6 Société Alpine de Géotechnique/SAGE, 38160 Gières, France 7 SFF Porelab, The Njord Centre, Department of Physics, University of Oslo, P. O. Box 1048, Blindern, 0316 Oslo, Norway Correspondence: Clément Hibert ([hibert@unistra.fr](hibert@unistra.fr)) Received: 21 June 2022 – Discussion started: 12 July 2022 Revised: 18 July 2023 – Accepted: 9 March 2024 – Published: 3 May 2024 Abstract. Understanding the dynamics of slope instabilities is critical to mitigate the associated hazards, but their direct observation is often difﬁcult due to their remote locations and their spontaneous nature. Seismology allows us to get unique information on these events, including on their dynamics. However, the link between the properties of these events (mass and kinematics) and the seismic signals generated is still poorly understood. We conducted a controlled rockfall experiment in the Riou Bourdoux torrent (southern French Alps) to try to better decipher those links. We deployed a dense seismic network and inferred the dynamics of the block from the reconstruction of the 3D trajectory from terrestrial and airborne high-resolution stereophotogrammetry. We propose a new approach based on machine learning to predict the mass and the velocity of each block. Our results show that we can predict those quantities with average errors of approximately 10 % for the velocity and 25 % for the mass. These accuracies are as good as or better than those obtained by other approaches, but our approach has the advantage in that it does not require the source to be localised, nor does it require a high-resolution velocity model or a strong assumption on the seismic wave attenuation model. Finally, the machine learning approach allows us to explore more widely the correlations between the features of the seismic signal generated by the rockfalls and their physical properties, and it might eventually lead to better constraints on the physical models in the future. Published by Copernicus Publications on behalf of the European Geosciences Union. 642 C. Hibert et al.: Prediction of the mass and the velocity of rockfalls 1 Introduction not have a volume large enough to generate those longperiod waves, thus precluding the use of inversion methods Slope instabilities are complex natural phenomena that posea threat to humans and infrastructures in many regions of the world. Landslides, rockfalls, rock avalanches, and surface collapses generating pit craters are natural disasters that can affect our societies. They also play a major role in the Earth surface dynamics as important erosion processes, whose occurrence might be caused by external factors such as earthquakes, intense precipitation, or the thawing of ice in the joints and fractures of large rocky masses. Understanding the triggering mechanisms and their dynamics and quantifying and documenting their properties and their spatiotemporal occurrences are of paramount importance to

FAQs

Why is the dynamics of slope instabilities difficult to study directly?

Events are often remote and spontaneous, which makes direct observation challenging. Hazard mitigation also requires understanding dynamics that are not easily measured on site.

How were block dynamics obtained in the controlled rockfall experiment?

A dense seismic network recorded signals while the 3D block trajectory was reconstructed using terrestrial and airborne high-resolution stereophotogrammetry.

What does the proposed machine learning approach predict, and with what accuracy?

It predicts the mass and velocity of each controlled single-block rockfall. The average errors are about 10% for velocity and 25% for mass.

Aug 5, 2026

Upload by Violet