Orientation domains: A mobile grid clustering algorithm with spherical corrections

Joana Mencos, Oscar Gratacós, Mercè Farré, Joan Escalante, Pau Arbués, Josep Anton Muñoz

Research output: Contribution to journalArticleResearchpeer-review

2 Citations (Scopus)

Abstract

An algorithm has been designed and tested which was devised as a tool assisting the analysis of geological structures solely from orientation data. More specifically, the algorithm was intended for the analysis of geological structures that can be approached as planar and piecewise features, like many folded strata. Input orientation data is expressed as pairs of angles (azimuth and dip). The algorithm starts by considering the data in Cartesian coordinates. This is followed by a search for an initial clustering solution, which is achieved by comparing the results output from the systematic shift of a regular rigid grid over the data. This initial solution is optimal (achieves minimum square error) once the grid size and the shift increment are fixed. Finally, the algorithm corrects for the variable spread that is generally expected from the data type using a reshaped non-rigid grid. The algorithm is size-oriented, which implies the application of conditions over cluster size through all the process in contrast to density-oriented algorithms, also widely used when dealing with spatial data. Results are derived in few seconds and, when tested over synthetic examples, they were found to be consistent and reliable. This makes the algorithm a valuable alternative to the time-consuming traditional approaches available to geologists. © 2012 Elsevier Ltd.
Original languageEnglish
Pages (from-to)140-150
JournalComputers and Geosciences
Volume49
DOIs
Publication statusPublished - 1 Dec 2012

Keywords

  • Bedding orientation
  • Shifting grid
  • Size-oriented clustering algorithm
  • Square error criterion
  • Structural analysis

Fingerprint

Dive into the research topics of 'Orientation domains: A mobile grid clustering algorithm with spherical corrections'. Together they form a unique fingerprint.

Cite this