Abstract:

Detection and mapping of landslide and rockfall events using remote sensing products has been proved to be an effective approach to provide landslide inventories. However, most of the studies are lacking valuable semantic information about landslide elements and how they react with the surrounding environment; natural and man-made primitives. In addition, post classification object-based approaches have been proved to result in better accuracy compared with the pixel-based. Lately, innovative close-range remote sensing technology such as Unmanned Aerial Vehicle (UAV) photogrammetry and Terrestrial Laser Scanning (TLS) are widely applied in the field of geoscience due to their efficiency in collecting data about terrain morphology rapidly. This research aims to demonstrate the applicability of UAV technology for automated semantic labeling in managing landslide and rockfall hazard in mountainous environments during emergency situations. SfM photogrammetry in addition to high accuracy RTK-GNSS ground control point establishment, is used to provide detailed 3D point clouds describing the surface morphology of the landslide and rockfall elements. The proposed methodology was divided in five main working phases. The first phase includes designing and execution of an optimal UAV flight planning to collect accurate 3D data. During the second phase, the pre-processing and raw data preparation such as point cloud filtering and elimination of ambiguities is taking place, while at the next phase an image segmentation using the 3D point cloud RGB information is created. The main task was focused on identifying the specific landslide elements by using an object-based approach. Based on Object-Based Image Analysis (OBIA), a sequence of image-based processes was applied, including multi-scale object segmentation, spectral, morphometric and contextual information extraction aiming to detect the landslide among other features. The next phase was set up for object classification in meaningful and homogeneous landslide classes (e.g. scarp, depletion zone, accumulation zone) which are spatially connected by introducing contextual information in the ruleset. The proposed methodology presents the effectiveness and efficiency of UAV platforms to acquire accurate photogrammetric datasets from intense relief environments and complex surface topographies by providing a holistic assessment and characterization of the failure site based on semantic classification of the landslide and rockfall objects. Results have demonstrated the capabilities of combining UAV platforms with computer-based methods for rapid and accurate identification of valuable semantic information subjectively and even from inaccessible areas of the landslide and rockfall body.