Geochemical typology validation of Paraná igneous province in the state of Paraná using statistical multivariate analysis

Resumo

Over the last decades, authors have produced geochemical datasets with samples collected from the whole Paraná Igneous Province (PIP). These results were compiled and standardized creating a database containing analysis of SiO₂, TiO₂, Al₂O₃, FeO_Total, MnO, CaO, Na₂O, K₂O, P₂O₅, Ba, Ce, Cr, La, Ni, Rb, Sr, Zr, and Y from outcrop and subsurface samples and the rock classification methods have used geochemical content as their criteria. In 2018 a new rock classification model for PIP was proposed using the natural gaps of four discriminant variables: SiO₂, Zr, TiO₂, and P₂O₅. This research introduces for the first time the use of Multivariate Analysis as a technique to verify and check this new model. This research used Exploration Analysis to extract two subsets: (a) one with 774 samples and 73 variables; and (b) other with 998 samples and 61 variables. Factor Analysis was used to reduce the datasets dimensions and extract Factor Scores. From the first dataset 18 Factors were extracted and from the second dataset 21 Factors were extracted. Then, Cluster Analysis was performed to group the Factor Scores in 7 clusters for the first dataset and 4 clusters for the second dataset and this analysis was followed by a Pattern Recognition and Classification Analysis using Neural Networks to corroborate the clustering. Additionally, Principal Component Analysis was performed on the samples grouped by the geochemical types proposed on the new 2018 model to extract the Principal Component Values and Weights. Finally, the Canonical Correlation Analysis correlated the Factor Scores grouped by K-Medias with the Principal Component Values and Weights grouped by the geochemical types and the result was the validation of the new classification model.