Geohazards of Szent Mihály Hill, Danube Bend, North Hungary: UAV monitoring to prevent landslide caused accidents

In the last decade the area of the Szent-Mihály Hill at the Danube Bend (North Hungary) has become widely known for geohazards, e.g., due to landslide risk of the slope, etc. Numerous mass movement-related accidents have taken place, risking human life and making significant damages and interruption in railway and car traffic. The continuous formation of hillslope debris is due to the geodynamic setting: active uplift of the study area and the incision of the Danube River contribute to increasing slope angles. The near-surface debris is prone to produce voluminous gravitational movements. Due to the uninterrupted soil creep, the vegetation is also sparse, having a little role in stabilizing the slopes. Consequently, monitoring and continuous modeling of hazardous slopes is indispensable. From regular drone-borne and multispectral data acquisition, we can investigate the drainage networks and its temporal changes in order to map the high-risk areas of the slopes.

To model where sediment volume can be accumulated developing potentially dangerous hot spots for landsliding, runoff modeling has been performed. To obtain high resolution, and up-to-date DTMs of the area several UAV measurement campaigns have been made in various (leaf-on and leaf-off) seasons taken in similar acquisition angle, with 60-70% overlapping. The relative high relief of the area represents a challenge to achieve an approximately identical image resolution. The quasi-circular shape of the study area introduces the effects of different styles of shading even in one acquisition. Data processing has been made with Agisoft Metashape Professional® software package, with digital photogrammetry techniques resulting a dense point cloud for each acquisition date. Ground Control Points (GCP) were fixed on the field at various elevations, to reduce error due to uncertainty of camera locations. Using these processed data, a Digital Elevation Model (DEM) have been carried out. The obtained DEMs were filtered to get DTMs. Since part of the material flow follows the drainage network, with the processed DEM, we could investigate these features using the of flow modeling via SAGA. After creating the drainage system, we located the main pourpoints and the catchment areas, belonging to each one. By picking the biggest drainage areas, which also have the biggest risk of landslide events, potential accident causing features could be located.

Due to the huge computer capacity requirement of the applied software, we have run the processing in medium resolution for optimal procession time. To find the ideal resource-quality ratio, the same dataset has been processed multiple times, each with different quality. This way a medium and high-resolution model have been created. To compare the results, runoff modeling also has been computed with both qualities. The results show, that at least high, or ultra-high quality processing method is required to reach the necessary level of details. With this method we were able to locate the most hazardous section of the hill and numerous methods for accident prevention has been suggested.

Funding: F.V. is supported by Project TKP2021-NVA-29, support provided by the Ministry of Innovation and Technology, Hungary, National Research, Development and Innovation Fund (TKP2021-NVA funding scheme).