Statistical methods applied to geochemical analysis of data from the Guaritas Allogroup, Camaquã Basin – RS, Brazil

Abstract

In this work, we use statistical methods for the analysis of geochemical data obtained through the technique of sample fusion mass spectrometry with inductively coupled plasma (FUS-ICP: Fusion-Inductively Coupled Plasma). The statistical treatment of geochemical data provides clusters and correlations of major, minor and trace chemical elements present in the chemical composition of the rocks of the Guaritas Allogroup, belonging to the Camaquã sedimentary basin, located in southern Brazil, in the state of Rio Grande do Sul. In addition to contributing to the consolidation of knowledge about the Camaquã Basin - RS/Brazil, the focus of this study is the statistical analysis of geochemical data available for the Guaritas Allogroup, from Denalle (2013). The following statistical methods were applied: straight correlation by approximating the points of one or more variables by the method of least squares, with calculation of the statistical significance of the correlation established for the sampled points and Principal Component Analysis – PCA. The statistical approach aims to optimize the data available for the Guaritas Allogroup, with the purpose to refining the information and possible interpretations of the data, which better establish the compositional characteristics and genesis relationships for these rocks. At the sampled points there are subtle variations in chemical composition, but no significant macroscopic variations were observed. The establishment of correlation lines helps to understand the specific relations of the composition of a chemical element, which can be a major element in the rock, for example, with respect to the variation of one or two minor elements or traces. The Principal Component Analysis method – PCA contributes to the understanding of the influence of which sets of chemical elements are decisive in the differentiation of lithological types. This technique proves to be advantageous for the analysis of extensive databases, as it reduces the excessive amount of data, eliminating overlaps, but with little loss of significant information. PCA points to choices of the most representative forms of data through linear combinations of the original variables. This technique provides advantages in the management of geochemical data or other areas of research in geosciences, and can be used in an analogous way in several available databases. For the Guaritas Allogroup, this technique proved to be useful in pointing out compositional relationships between major elements, main rock formers, such as SI, Ca, Al and Mg, and minor elements, such as Mn, Zn, Ti, Rb and Sr. Therefore, the preliminary results obtained represent an advantageous tool, used in an additional way, for the interpretation of geochemical data in support of the analytical techniques traditionally employed in chemical stratigraphy.