Abstract: |
The similarity in the distribution of rock properties could be quantified by measures of data coherency that include the spatial arrangement and relationship between different rock types. Measures of coherency are introduced for categorical and continuous variables. They are used for quality control including identifying questionable facies designations, top elevations, and problems with the data. The coherency can be maximized by adjusting top elevations in stratigraphic space, and results are used to help guide the geological gridding process, that is, to optimize the transformation from physical space to stratigraphic space. Coherency measures are used as a similarity metric and applied in a hierarchical clustering scheme to assist with geological zonation. Data is clustered into geological objects having similar categorical and continuous property distributions. This is useful for segregating data into stationary domains for geostatistical modeling. Due to the local nature of the coherency measures, identifying non-linear geological objects such as fluvial channels is possible. Two other uses of clustering are identified and include detection of boundaries between different geological successions and facies proportion trend modeling. The coherency measures are applied to synthetic data extracted from training images that represent different depositional environments to demonstrate the effectiveness in quality control, gridding, and geological zonation. |
