Applications of some statistical techniques to weathered rock geochemical data from the Broken Hill area, New South Wales, Australia

Abstract The objective of this study was to evaluate the contribution of computer-based techniques to the interpretation of weathered rock geochemical data from the Broken Hill area of New South Wales, Australia. The evaluation was conducted at two levels: (1) simplistic data management and plotting techniques; and (2) multivariate statistical techniques. Computer-based programs and procedures have been applied, as necessary, to data for approximately 7000 weathered rock samples from thirteen different survey areas, all within a 30-km radius of Broken Hill. Field information was recorded at each sample site and the samples were analyzed for Pb, Zn, Cu, Ag and Mn, plus Co, Cr and Ni from some of the survey areas. In each of the survey areas, the presence of mineralization is indicated, to varying degrees, by higher values of Pb and/or Zn. Concentrations of Cu and Mn also reflect mineralization, but less consistently and with less contrast. Higher concentrations of Ni and Cr appear to reflect mafic rock types, where present. Simpler univariate programs proved essential for the recognition of patterns in the voluminous data set. Multivariate statistical procedures, including factor analysis and discriminant analysis, were used to identify element associations and generate scores which quantify these associations in individual samples. Three basis associations were identified in each data set and are interpreted as representing variation in (1) mineralization, (2) parent rock type, and (3) weathering. The associations so identified confirm and quantify most relationships identified during basic single-element interpretation. The composite factor and discriminant scores generally reflect the distribution(s) of the most heavily weighted variable(s) in the associations and, in this case, do not contribute significantly more to exploration than do the individual elements.