Publication Type:Journal Article
Source:Journal of Applied GeophysicsJournal of Applied Geophysics, Volume 118, p.145-154 (2015)
Keywords:classification, clustering, downhole measurement, fuzzy k-means, logs, ocean drilling program, physical properties, physical units, rock type
Downhole density, gamma radioactivity, and magnetic susceptibility measurements in five drillholes at the Victoria property (located in the south range of the Sudbury basin) were analyzed to identify homogenous physical units. The fuzzy k-means clustering algorithm was used for unsupervised classification of the data. Four main physical units were identified in boreholes with distinct physical characteristics. Three of them were differentiated mainly based on different gamma ray and density values, and the fourth one was characterized by high magnetic susceptibility. Physical units were compared with rock types logged by geologists to determine which rock types corresponded to physical units. We found that there was a meaningful spatial and statistical correlation between physical units (characterized based on their physical property measurements) and lithological units as indicated by rock types at the Victoria property. However, not all rock types could be uniquely identified by the statistical classification, but a set of similar groups could be identified. Hence, identifying a group of rock types described by each physical unit can be used to translate physical data to/from lithological data. Alternatively, the physical log units could be used as a quality control procedure to check the geological logs, or to highlight areas where more careful logging or other investigation would be warranted. (C) 2015 Elsevier B.V. All rights reserved.
Ck3lw<br/>Times Cited:1<br/>Cited References Count:26