Implementation of Multivariate Clustering Methods for Characterizing Discontinuities from Scanlines and Oriented Boreholes
In geological engineering, discontinuities are typically analyzed by grouping (clustering) them into subsets based on similar orientations, and then characterizing each set in terms of position, spacing, persistence, roughness and other parameters. Multivariate analysis can be used to incorporate some of these other parameters directly into the cluster analysis. The implementation of four methods of cluster analysis that consider orientation, spacing and roughness are described here: nearest neighbor, k-means, fuzzy c-means, and vector quantization. The net result is a better grouping of discontinuities, so that members of a subset might be more uniform in terms of mechanical or hydrological properties. This paper presents the implementation of this analysis in a Windows® based program CYL that also serves as a graphical visualization tool.
W. Zhou and N. H. Maerz, "Implementation of Multivariate Clustering Methods for Characterizing Discontinuities from Scanlines and Oriented Boreholes," Computers & Geosciences, Elsevier, Aug 2002.
The definitive version is available at http://dx.doi.org/10.1016/S0098-3004(01)00111-X
Geosciences and Geological and Petroleum Engineering
National Science Foundation (U.S.)
University of Missouri Research Board
Keywords and Phrases
Euclidean Norm; Discontinuity Sets; Discontinuous Rock; Multivariate Cluster Analysis Algorithms; Visualization Tools
Article - Journal
© 2002 Elsevier, All rights reserved.