The potential of UAV data to capture sediment dynamics — A case study from the Gesäuse National Park

This thesis deals with the potential of high-resolution UAV data for the acquisition of sediment dynamics. For this purpose, a test area (Langgriesgraben, NP Gesäuse, Austria) was selected in a coagulation with episodic water flow and high morphodynamics. From the high-resolution UAV aerial images (medium GSD 2-3 cm) of three flying times, digital point clouds, terrain models and orthophotos were created using the SfM-MVS method. The software Agisoft Photoscan Professional 1.2.4 was used and the georeference was carried out via dGPS measured pass points. The height accuracy of the generated raster terrain models and the location accuracy of the orthophotos were estimated on the basis of dGPS single points and the calculation of various statistical parameters. The accuracy achieved (e.g. height accuracy DGM: 4-7 cm RMSE) lies in the accuracy range that can be expected from the literature. Shadows in the aerial images subsequently lead to areas with greater uncertainty in the DGM. The digital point cloud and the DGM were compared for three test areas with timely recorded TLS data, with the deviation between UAV and TLS depending on the chosen comparison method and test area 6-12 cm (percentile 95 of the deviations). Based on the generated UAV terrain models, the surface change in the coagulation could be quantified by the DoD method using a minLOD. Due to the determined minLOD (14 cm) a significant change could be found to only 12 % of the total area. The deposition (295 m) clearly outweighs the erosion (163 m) of sediment during the period considered (53 d). Based on the results, it can be concluded that the method used SfM-MVS in combination with a UAV is suitable for detecting higher rates of change, but small rates of change cannot be determined due to the accuracy achieved.