Applying machine learning to Denmark’s Landslide Inventory

Landslides have been an underestimated geohazard in Denmark, directly impacting infrastructure and settlements (Svennevig et al., 2020). Climate change is expected to accelerate landslide activity in Denmark, yet, relatively few scientific studies exist (Svennevig et al., 2024). The previously published national landslide inventory documents more than 3200 landslides that were manually mapped based on morphological expression in a national 0.4 m LiDAR DEM (Luetzenburg et al., 2022). However, the inventory requires updates and further development to provide a more complete record of landslide occurrences across Denmark. Here we take the next step by applying machine learning to expand, refine, and conduct quality-control on the national inventory. Further efforts focus on developing AI approaches to detect paleo-landslides and assign landslide activity to recent landslides. The expected outcome is a more comprehensive landslide inventory that will enhance understanding of slope instability mechanisms and provide insights into how landslides may evolve under a changing climate, while supporting enhanced geohazard and risk assessments.