Topographic profile and morphology analysis of shallow landslides inside and outside of forests with a semi-automatic mapping approach and bi-temporal airborne laser scanning data

Abstract

Investigating the effects of forest land cover and forest structure on shallow landslide characteristics such as their morphology (e.g., area and mean depth) and topographic profiles could provide a better understanding of how forests affect landslide processes. Landslides located under the forest canopy, which are often overlooked by conventional landslide mapping methods (e.g., using aerial imagery), can be captured using airborne laser scanning (ALS). In this study we investigated forest effects on landslides by developing a well-performing semi-automated workflow for mapping landslide scars and analysing their characteristics in relation to the forest canopy cover, using bi-temporal ALS data and a random forest model. The mapped landslide scars were analysed with a forest canopy cover mask and forest structure parameters, such as the closest tree distance and the number of trees surrounding the scar. The investigated scars within the forest have significantly larger depths, thicknesses and higher pre-failure slope values than scars located outside the forests. Additionally, the closest tree distance showed a clear relationship with the scar volume or landslide magnitude. This enhances our understanding of forest impacts on landslide processes and their protective function. Furthermore, it shows that inventories which neglect landslides in forests also misrepresent their characteristics.