An application of lithogeochemistry to the evaluation of the Ni-sulphide ore potential of weathered serpentinites in the Fortaleza de Minas Greenstone Belt, Minas Gerais, Brazil

Abstract

Oostindi~r, J., Taufen, P.M. and Vriend, S.P., 1988. An application of lithogeochemistry to the evaluation of the Ni-sulphide ore potential of weathered serpentinites in the Fortaleza de Minas greenstone belt, Minas Gerais, Brazil. J. Geochem. Explor., 31: 57-73.A practical application of lithogeochemistry to the classification of weathered serpentinites as to whether or not serpentinites are host rocks to Ni-sulphide ore is presented. An effort is made to apply existing concepts used to distinguish between fertile and sterile unweathered ultramafics to explain the observed chemical differences between fertile and sterile weathered serpentinites in southwestern Minas Gerais, Brazil.The data comprise total-attack chemical analyses for A1, Ca, Cr, Cu, Fe, Mg, Mn, Ni, Si, Ti, V, and partial-attack determinations of Cu and Ni as ascorbic acid, hydrogen peroxide-soluble metal. Student's t-tests are used to identify element concentration differences between the two serpentinite groups. Discriminant analysis is used to classify the weathered fertile and sterile serpentinites. Non-linear mapping and fuzzy c-means clustering help to assess the validity of the discriminant analysis. Finally, chemical element associations provide extra information on geological and mineralogical causes that can be responsible for the identified differences.It is found that a clear separation and classification of weathered serpentinites is possible using the procedures presented. The observed chemical differences between fertile and sterile serpentinites are interpreted in terms of original magmatic differentiation, the degree of sulphur saturation of the original magma and the behavior of the derived ultramafic rocks in a tropical weathering environment.+Present address: Shell Internationale Petroleum Maatschappy. B.V.,

Hydrogeochemistry of the shallow dutch groundwater: Interpretation of the National Groundwater Quality Monitoring Network

“…It is often advantageous (or even mandatory) to split a large data set, such as the LMG, into more homogeneous subsets, prior to statistical treatment, such as trend analysis [Howarth, 1983;Vriend et al, 1988]. This avoids the problem of dealing with multimodal distributions for the individual elements and provides a basic understanding of the groundwater geochemistry.…”

“…As an alternative for these classification methods, unsupervised fuzzy c means cluster analysis is used here to discover natural groups of groundwater compositions in the LMG. This clustering technique is a powerful tool for separating data sets in homogeneous groups [Vriend et al, 1988]. The calculated clusters are related to a typical macrochemistry based on the complete set of macrochemical parameters.…”