Abstract <p>A photogrammetric field mapping technique is employed to derive the best-fit planes of fractures from photogrammetric point cloud data. Joint sets are then grouped based on the orientations of the best-fit planes. The acquired joint set groups are used, by applying a power law fractal transformation, to convert 2.5<i>D</i> rough surface outcrop fractures into three-dimensional (3D) fracture radius distributions. We also use graph representations to provide truncation probabilities in discrete fracture network (DFN) modeling to investigate the topological behavior of in-situ fracture arrests. Further analysis involves obtaining two-dimensional (2D) joint surface roughness parameters from the point cloud data. 2D joint surface roughness parameters are correlated with the joint roughness coefficient (JRC). The transformed fractal power law data, joint set orientations, and topological truncation probabilities are input to generate stochastic DFN models. Using two different software programs, synthetic rock masses (SRM) are created based on a validated DFN model. Finally, the block shape and size distributions of two types of SRM are investigated. The in-situ block size distribution (IBSD) results of the SRM models show that successive block-cutting algorithms can either underestimate or overestimate the in-situ block size distribution (IBSD). This discrepancy arises from the underlying block-cutting algorithms, which somewhat overlook non-persistent fractures. A multi-dimensional spacing (MDS) algorithm is written using Python scripting to account for the influence of non-persistent fractures on IBSD. Our research utilizes a photogrammetric field data acquisition technique to reveal the underlying topological behavior of complex in-situ fracture networks through graph representations. The most important contributions include the results of the MDS algorithm for IBSD-referenced rock mass block shape distribution (RBSD) curves and investigations into the block shape and block size characteristics of the DFN.</p> Highlights <p><UnorderedList Mark="Bullet"> <ItemContent> <p>A photogrammetric field mapping is conducted with a focus on reducing sampling biases.</p> </ItemContent> <ItemContent> <p>Collected 2.5<i>D</i> fractal traces are transformed into a 3D fracture radius population through reconstructed point cloud triangulation.</p> </ItemContent> <ItemContent> <p>Truncation probabilities are estimated from fracture graph representations on a 2D topological sampling plane.</p> </ItemContent> <ItemContent> <p>Synthetic rock mass’s block formations are interrogated based on the DFN model, with block shapes and IBSDs determined.</p> </ItemContent> <ItemContent> <p>A multi-dimensional spacing (MDS) algorithm is used to define the IBSD curves.</p> </ItemContent> <ItemContent> <p>A novel rock mass block shape distribution (RBSD) is proposed.</p> </ItemContent> </UnorderedList></p>