Abstract ID: 3.12894 | Accepted as Talk | Talk/Oral | TBA | TBA

Different land cover types have different effects on the occurrence and processes of shallow landslides (translational earth and debris slides with a depth < 2 m). For a thorough shallow landside susceptibility and hazard assessment it is important to have a comprehensive understanding of the effects of land cover on slope stability. However, especially the effects of forest cover and its structure on the processes of shallow landslides are still largely unknown. Due to the bias of existing inventories against landslides in forests and the inability of commonly used analysis datasets (e.g., aerial imagery) to penetrate the forest canopy, most studies are unable to capture differences between landslides in forests and open land. Previous studies on landslide mapping instead used laser scanning data datasets, which are capable of capturing landslide signs beneath a forest canopy. In recent years it also became possible to use bi-temporal datasets for these analyses. Use of such datasets could provide not only a more complete picture of landslides that occurred under the forest canopy, but also provide more accurate representations of their topographic signatures. This study provides insights into how forest cover and structure affect the topography (e.g., slope values within the landslide) and morphology (e.g., their area and shape characteristics) of shallow landslide scarps. This is investigated using bi-temporal Airborne Laser Scanning (ALS) datasets. Based on the resulting Difference of Digital elevation models (DoD), a semi-automatic detection algorithm was developed to map landslides under forest canopy. The resulting mapped scarp segments were classified based on their canopy cover and used to analyse differences between landslides in forests and open land, using their topographic profiles and morphological features. The results show that there are significant differences between landslides in forests and open land. Landslides in forests within the study area are on average significantly smaller (23 m2) than those outside forests (33 m2) and have a narrower profile, with a mean cross-sectional curvature of -0.24 against -0.19 for landslides in open land. These are valuable insights which can be used to better inform shallow landslide modelling approaches in forested areas.