Chemostratigraphy Enables Correlations and Reservoir Characterization with High Resolution Elemental Data

Abstract Previous studies have shown that by applying multivariate statistical analysis to chemostratigraphy, indistinct sequence stratigraphic correlations can be enhanced. Chemofacies and correlatable chemozones can be defined within highly homogenous strata, using specially designed statistical algorithms. In this study, we first investigated the better performing of linear and non-linear dimensionality reduction techniques in analyzing geochemical datasets for chemofacies and chemozones development. The general applicability of this conceptual model for sequence stratigraphic correlations, was subsequently tested. The results show that the linear method was able to account for 63% of input data variance while the non-linear technique accounted for 100% of the variance. In addition, the linear techniques are better utilized to establish chemofacies, whereas the non-linear techniques considerably perform better in establishing correlatable chemozones, while also improving accuracy.