Identification and Analysis of Geological Patterns in the Witwatersrand Basin Using Spatial Decomposition

Deposition of gold-bearing deposits in the Witwatersrand Basin was controlled by hydraulic processes. These resulted in the formation of independent sedimentological facies areas having different statistics and semivariograms. Consequently, when there are indications that data are a mixture of statistical populations, it is important to decompose them into statistical subpopulations and analyse these independently. However, in the case of actual mining, physical position and extent of sub- populations are important and, as a consequence, any decomposition of data must take space into account if it is to be of use. An algorithm for the spatial decomposition of data into its sub-populations is described and is applied to real data taken from the Witwatersrand. Two distinct areas with completely different geostatistical parameters were revealed.