Combining multispectral UAVs with object based image analysis for landslide investigations in forested terrain

Variations in vegetation density from multispectral data and surface morphology models have been successfully used for landslide delineation and monitoring in forested areas. However, the limited spatial resolution of satellite imagery and the lack of reference elevation models have limited the application of these methods to large landslide events. This study investigates the application of small unmanned aerial vehicles (UAVs) with multispectral cameras for land cover classification and landslide mapping in forested areas. Photogrammetric terrain models and orthoimages (RGB and multispectral) obtained from repeated mapping flights between November 2023 and May 2024 were combined with an ALS reference terrain model into an object-based image classification workflow. The models enabled the differentiation between natural forest and areas affected by past mining activities, as well as the identification of variations in vegetation density and growth rates. The findings demonstrated that high-resolution multispectral datasets can be employed for the efficient characterization of landforms and landslide monitoring, even in densely vegetated areas.