Comparison of Multi-temporal Elevation Models of a Debris-Flow Channel

Abstract

Debris flows are among the most hazardous landslides. Additionally, their relevance increases as a consequence of global climate change, a phenomenon, which has already been observed by other authors. In June 2015 two debris flows occurred due to a radar-inferred rainfall intensity of about 90 mm in 45 min near Oberstdorf (Bavaria, Germany) in the Northern Alps. The debris flows lead to damage costs of several million euros and more than 300 people were evacuated. To monitor geomorphic changes and a potential channel refill after the event, high resolution (5 cm grid size) digital elevation models (DEMs) were derived from terrestrial laser scanning (TLS) in May, July and August 2017 in the Roßbichelgraben, one of these channels. Therefore, the 800 m long channel was observed with 72--78 laser scan positions on each date and DEMs of difference (DoDs) with spatially variable uncertainty were calculated. Simultaneously, DEMs were derived by photogrammetric surveys using an unmanned aerial vehicle (UAV). Photos taken by the UAV were photogrammetrically reconstructed into point clouds using the structure from motion---multi-view stereo workflow (SfM-MVS). The results show a good correlation of detectable areas with geomorphic changes between both methods. However, the accuracy of the SfM-MVS derived DEMs remains uncertain in steep parts of the channel. Nevertheless, the current contribution demonstrated that SFM-MVS offers great potential for topographic surveying and is a promising tool to quantify geomorphic changes in the near future.