Spatial domaining of highly variable continuous geometallurgical data

An outcome of geometallurgical mapping and modeling is the generation of continuous down hole profiles of quantitative geometallurgical attributes at assay scale. These attributes vary from traditional routine data collection methods such as assays and geotechnical logging (e.g. RQD, Fracture Frequency), to new measurements including petrophysical attributes, EQUOtip hardness data and modelled estimates of metallurgical performance indices (e.g. A*b, BMWi). Highly variable continuous geometallurgical data with high spatial resolution can make it difficult to identify spatially continuous domains. An automated method based on the well known time series analysis technique of cumulative summation (CuSum) has been developed which uses statistical analysis to identify domain boundaries in down hole profiles. A bootstrap analysis automatically identifies potential domain boundary locations with associated confidence levels and a series of t-tests determine the significance of defined populations. A hierarchical clustering algorithm is applied to results to enable data sensitivity to be taken into account. The method produces statistically significant populations and is suitable for application on either single or multiple drill hole datasets. Domains defined preserve the inherent variability of the continuous down hole data and provide a method for scaling up variability data (i.e. assay interval scale) to mine scale (i.e. bench scale, block model). A major advantage of the method is its ability to provide suitable input parameters for use in spatial modeling of non-additive and non-linear geometallurgical attributes through conditional simulation techniques.