From Points to Lines to Surfaces: A Novel Approach to Reconstructing Field-Scale Three-Dimensional Shale Roof Models Using Image Processing

Most coal mine roofs in the Appalachian region consist of laminated shale, which is prone to brittle failure. Accurate modeling of shale roofs is vital for studying cutter failures of these shale roofs, and ordinary and formational bedding planes play a significant role in the response off the shale roof to mining-induced stresses. This paper presents a robust method for constructing three-dimensional shale roof models using limited borehole cores and leveraging image processing techniques. The approach begins with image extraction from shale cores. Then, we extract the portions that feature the core and employ image-stitching techniques to merge them seamlessly. Once the images are stitched, they are duplicated horizontally and merged to create a cross-sectional plane image representing a shale roof. This roof cross-sectional plane image serves as an input to our proprietary Digital Image Processing (DIP) algorithm, which is adept at extracting ordinary bedding planes (interfaces that represent color transitions between a single shale unit). The extraction process involves transforming discrete core points (ordinary bedding plane pixels) into continuous lines representing ordinary bedding planes. Finally, extending this method into the third dimension, the model is further extrapolated to form a comprehensive three-dimensional shale roof. Through this process, the geometry information of ordinary bedding planes can be extracted for each type of shale in cores, including sandy shale, gray shale, dark shale, etc. This work bridges the gap between limited core data and comprehensive 3D geological modeling, offering valuable insights into identifying and quantifying the variations in different shale roof geologies.