Analyzing extremely rapid, flow-like landslides using laser scanning and numerical modeling

Abstract

Debris flows are extremely rapid, flow-like landslides and can impact settlements and infrastructure far from their source. Since they occur on steep slopes in mountainous terrain, debris flows are particularly important for alpine countries like Switzerland, where they have caused major damage and lead to several fatalities in the past. To reduce the destructive impact of debris flows in the future, we need accurate models, which can predict their behavior. However, field-scale data of actual debris-flow events, required to test and improve such models, are rare. This study tested a novel LiDAR sensor (Ouster OS1-64 Gen 1), which could be installed in the field to address the lack of field data. The sensor was assessed in a laboratory experiment, where a sediment mixture was released on an inclined plane. During the experiment, the sensor recorded point clouds of the moving sediment mixture at a rate of 10 Hz. The point clouds were then used to measure the following four essential parameters of the simulated landslide: slope angle, front velocity, flow depth, material volume. The sensor was able to measure all of these parameters reasonably well. However, the accuracy of the measurements of the flow depth and material volume strongly depended on the thickness of the slide: Measurements for "thick'' (ca. 3.5\,cm) flows were much more accurate than for ``thin'' (ca. 2\,cm) flows since the latter were very close to the noise level ($\pm$3\,cm) of the point clouds. Possible locations for sensor placement in the field were suggested by using the numerical model DAN3D to simulated debris flows in the Illgraben. The Illgraben is Europe's most active debris flow catchment and is located in southwestern Switzerland. Since the laboratory experiment found that the LiDAR sensor's accuracy increased with flow depth, the results of the debris-flow modeling were analyzed with respect to maximum flow depth. In two simulations of large-volume debris flows the greatest flow depths occurred in the central part of the Illgraben catchment (close to check dam CD1). Hence, this region was suggested for sensor placement. However, a comparison of flow depths at the laboratory scale and at the field scale revealed that the LiDAR sensor's position should not be determined based on the maximum flow depth only. The assumption from the laboratory experiment that ``thicker'' flows automatically lead to a better accuracy is not valid at the field scale due to the much greater flow depths in the field. In the Illgraben, even small debris-flow events have flow depths of ca. 0.5\,m and thus are well above the noise level ($\pm$ 1.5--10\,cm). Therefore, the sensor could be installed everywhere along the Illgraben channel and not just in the location suggested by the model results. An alternative location for sensor placement is located on the toe of the debris fan (check dam CD29), where the sensor could be installed next to existing measurement instruments. In this location, the sensor is expected to record point clouds with a similar or even higher precision compared to the laboratory experiment. However, the processing steps and the measurements of different parameters might be more challenging due to the large extent of debris flows in the Illgraben and due to atmospheric effects like rain or fog. Further work is required to estimate the field performance of the LiDAR sensor, but if the sensor were successfully installed and operated in the Illgraben, it would have the potential to record several moving debris flows a year and thus significantly increase the amount of field-scale debris-flow data.