Introduction
Debris flows are a common, dangerous and destructive natural hazard in mountainous regions in Switzerland and worldwide. Monitoring and recording of debris flows is important to analyse long term development, to better process understanding, and to improve numerical models in order to ultimately improve preparedness for such hazardous events. ETH Zurich has developed a system using LiDAR and cameras to record ongoing debris flows in the field. Stations for example placed in Illgraben record the debris flow, from which depth and velocity are computed during post processing [1][2]. A new version of the system is being developed to enable real-time computation of debris-flow parameters. This would help to immediately assess the severity of debris flows and would also allow for direct measures, such as closing roads.
Implementation
The updated system uses a LiDAR to collect 3D point clouds at a 10 Hz frequency. In the 100 ms between two scans, the computer evaluates flow depth and velocity of the debris flow. Because the station is deployed in a remote location, solar panels and batteries are used to provide electric energy. To conserve energy, the main computer is in a low power state, and the LiDAR is turned off. Geophones further up the mountain are monitoring ground vibrations and if a debris flow occurs a signal is sent to wake up the main computer and LiDAR to start data recording and processing.
As a base for the processing pipeline, Robot Operating System 2 (ROS 2) is used. ROS 2 provides a framework to separate processing steps into modular components and offers standardized messaging between them and connected sensors. In addition, each modular component runs in its own docker container, where containers communicate with each other via ROS messages. This modularisation allows for simple management of dependencies, easier addition of new sensors and processing steps, and a more robust system where individual containers can be restarted without affecting the full system.
Figure 1 displays the processing pipeline. The LiDAR retrieves point clouds which get cropped and aligned. From the point cloud a digital elevation model (DEM) is derived, which is a 3D representation of the surface. Based on the DEM, a shaded relief map (hillshade) is generated by simulating how light and shadow fall across the terrain. Feature tracking is performed to evaluate how much the debris has flown between the previous and the current hillshade. With the tracked distance and elapsed time, the velocity of the debris flow can be estimated. In addition, the point cloud is also used to evaluate the cross-sectional area of the flow. Using these quantities, we can estimate discharge and volume.
Further work
The updated system with real-time and on-site processing provides a good foundation to monitor debris flows in remote areas to measure and record their severity. The system is currently undergoing final implementations and laboratory tests and is expected to be deployed in early summer 2026. Initial field results will reveal further improvements regarding, triggering, processing, and system design, which will be addressed in future versions of the system.
References
[1] J. Aaron, R. Spielmann, B. W. McArdell, and C. Graf, “High‐Frequency 3D LiDAR Measurements of a Debris Flow: A Novel Method to Investigate the Dynamics of Full‐Scale Events in the Field,” Geophys. Res. Lett., vol. 50, no. 5, p. e2022GL102373, Mar. 2023, doi: 10.1029/2022GL102373.
[2] J. Aaron et al., “Detailed observations reveal the genesis and dynamics of destructive debris-flow surges,” Commun. Earth Environ., vol. 6, no. 1, p. 556, Jul. 2025, doi: 10.1038/s43247-025-02488-7.