Semi-automatic extraction of dangerous rock blocks from jointed rock exposures based on a discontinuity trace map

Using a rock slope surface model generated from unmanned aerial vehicle (UAV) digital photogrammetry, this study develops a new semi-automatic dangerous rock blocks extraction approach. It consists of four steps: (1) three dimensional trace map extraction and projection; (2) convex polygon searching using a depth-first search algorithm and a sign consistency test; (3) true block formation analysis based on the geometric topology; and (4) block removability, failure mode and stability analyses using the edge vector, active resultant vector and limit equilibrium method, respectively. Compared with existing block identification methods, the proposed approach considers the contribution of line-type discontinuities on block formation and integrates geometric and mechanical modelling processes in one program. This approach is suitable for rockfall sources quantitative assessment. The locations, geometric parameters, and stability results of all identified blocks can be automatically obtained, and these could provide a reference for rockfall hazard assessment. The proposed approach is applied to a railway bridge case in China, and the corresponding rockfall hazard is evaluated. At the end of this paper, the joint sets related to block formation, uncertainties of block shapes, and discontinuity size effect are discussed.