Regresar

Cluster analysis and other multivariate statistical methods applied in the identification of chemofacies: Example in the tertiary falcon basin, Venezuela

Abstract:

The application of multivariate statistical methods (cluster, correspondence and factorial analysis) in chemostratigraphic data was studied, with the aim of testing and developing a suitable method for the identification and correlation of chemofacies. For instance, a matrix composed by 232 rock samples and 18 chemical attributes was done, from Cerro Pelado Formation, an Oligocene-Age unit from the Falcon Central Basin, in three sections outcropping in El Troncón stream, La Paloma stream and La Cuesta coalfield. Constrained cluster analysis allowed the identification of several chemofacies in the three sections; criteria for the definition of boundaries between chemofacies were defined by cut-off lines in the dendrogram, after a reduction of variables. In general, these boundaries are consistent with changes in the intensity of the sedimentary processes (marine influence, redox conditions). By cluster, correspondence and factorial analysis were detected two main geochemical associations:(a) redox-organic matter association, and (b) clastic association. The redox-organic matter association is more relevant at the top of the sequence; for that reason, the frequency of bed coals and marine influence increase towards the top. On the basis of discriminant function analysis, the section of La Cuesta coalfield was correlated with the middle zone of the La Paloma section. Obtained results allow concluding that the cluster analysis is stronger and robust statistical tools than correspondence and factor analysis, in the geochemical study of stratigraphic sections. The constrained cluster analysis was a simple, efficient and fast technique in the determination of chemofacies.