Factor and discriminant analysis to lithogeochemical prospecting in an area of central Sweden

Abstract

Abstract Alteration zones in host rocks surrounding mineral deposits have long been used as indicators of ores, and recent studies have stressed the importance of the chemical distribution patterns or haloes. These may be more pronounced than mineral alteration and therefore detectable over greater distances. There are, however, several problems involved in this type of prospecting. One is the possible presence of several geochemical populations in the analyzed samples. Only some of these populations reflect the haloes which one wants to detect, while others only disguise the haloes. One solution is to use multivariate statistical analysis to discriminate between the different populations. In the Stollberg orefield, central Sweden, these prospecting methods are presently in use. Here, located within Precambrian volcanic rocks, several small and large stratiform zinc-, lead- and iron ores are situated in the same stratigraphic zone and within 6 km of each other. Drifts in the area go down to 575 m. Approximately 3000 rock samples collected from outcrops, drill cores and drifts were analyzed for 35 elements by XRF and IDES (Image Dissector Echelle Spectrometer). These samples represent an area of 3 × 9 km to a depth of 600 m. Due to the presence of several geochemical populations, the raw data are difficult to interpret. After reducing the skewness of the data, factor analysis was performed and, in general, distinct multi-element zonations were detected. Factor scores were used as the input for discriminant analysis. In this manner chemical haloes were obtained which have an extension of more than one kilometer in all directions around the various ore types in the area. The ores can thus be detected at considerable distances. On the basis of these findings, some promising new prospects have been discovered.