Rockfall hazard assessment based on semi-automatic point cloud analysis from UAV data

Abstract

Slettind in Flaktstad municipality, Lofoten in Nordland, has numerous rockfalls throughout the year. Rockfalls hit Fv 805 on a weekly basis, and it is estimated the most dangerous road in Nordland county. A rock avalanche hit the road winter 2017 closing it for 2 weeks, and isolating the small village Myrland. Statens Vegvesen consider to build a tunnel to protect the road but the failure mechanisms has been little understood, as it is deemed too dangerous for traditional fieldwork to access the mountain in a safe manner. Traditional fieldwork involves shear strenght testing of joint sets and infill, as well as mapping of joint orientations, roughness, and volumes of blocks, using Barthons Q method, Rock Mass Index or GSI. However as it has not been feasible to attend the wall for such mapping, due to the steepness, height and risk for rockfall, new techniques for rockfall hazard assessment have been put in use.

The purpose of this master thesis has been to use photogrammetry from UAV images, and Structure-for-Motion to create a 3D modell and identify joint surfaces, orientations and evaluate the failure mechanisms. The workflow has been compared and evaluated against traditional mapping methods. Photogrammetry from drone images has proven valuable for understanding the structures and driving forces in a rock mass, and is a more flexible and cheaper option than LiDAR or similar, to build point clouds. As such, a validated semi-automated workflow is a resource for evaluating steep, inaccessible mountainsides.