1. A semi-automatic approach for joint orientation recognition using 3D trace network analysis.
- Author
-
Mehrishal, Seyedahmad, Kim, Jineon, Song, Jae-Joon, and Sainoki, Atsushi
- Subjects
- *
FUZZY algorithms , *TRACE analysis , *FUZZY clustering technique , *DIGITAL mapping , *DIGITAL elevation models , *DIGITAL maps - Abstract
Identifying rock mass discontinuities and their plane orientation are crucial factors when determining rock mass characteristics. Rock mass discontinuity mapping is fundamentally dependent on joint trace surveying since traces of them are most often the only visible features in rock outcrops. Traditional methods for joint trace surveying using tape and a geological compass are challenging, time-consuming, and hazardous. Fortunately, non-contact measuring techniques offer the advantage of generating accurate objective records of rock masses and enabling the measurement of discontinuities from digital surface models and 3D point clouds of outcrops without requiring direct access to the rock mass and associated constraints. Herein, an innovative approach for identifying discontinuity planes in rock formations using 3D trace data is presented. We introduce the concept of curved and straight traces, with a curvature index indicating the trace's accuracy by representing its discontinuity plane. In addition, we identified co-planar traces by analyzing intersecting straight traces, thereby further contributing to discontinuity plane determination. The methodology's effectiveness was established through validation using a predefined 3D trace network of discontinuity planes with known orientations on a 3D digital rock outcrop model. The methodology was then applied to trace data collected from an actual rock outcrop. The algorithm successfully matched >66% of traces with their corresponding discontinuity planes. After clustering, 80% of the identified planes aligned with principal joint sets within the rock mass. These identified planes accurately aligned with their expected jointing pattern, thereby validating the robustness of the method. Despite challenges presented by discontinuous and complex trace segments obtained from semi-automatic trace detection techniques, the algorithm effectively processed such traces, thereby enabling a comprehensive understanding of the rock mass's jointing system. This enables swift identification of the main joint orientations. We leveraged stereonet analysis to identify principal joint sets using kernel-based and fuzzy C-means clustering techniques. Our approach represents an important advancement in the characterization of rock mass structural properties. • AI-based semi-automatic 3D trace detection method enhances safety and reliability. • Digital joint mapping yields more objective results in structural geology. • Complex 3D trace data processing identifies joints' orientation, spacing, and length. • Kernel and fuzzy c-means clustering Identifies principal joint sets on stereonet. • Machine learning can improve digital joint mapping to characterize rock mass. [ABSTRACT FROM AUTHOR]
- Published
- 2024
- Full Text
- View/download PDF