Hierarchical delineation and multivariate statistical discrimination of chemozones of the Cauvery Basin, south India: Implications on spatio-temporal scales of stratigraphic correlation

Abstract

Success in locating oil pools in the Cauvery Basin, south India had been found to be based on the ability to delineate precisely the stratigraphic traps resulting from frequent sea level changes. However, recognition and delineation of them in terms of depositional units through conventional stratigraphic methods have been elusive owing to the limitations of such methods and lack of unified stratigraphic markers that could be traced at regional and basinal scale. This paper attempts to recognize depositional units in terms of chemozones, chronologic and lithostratigraphic units by assigning distinct geochemical signatures. Geochemical signatures were assigned through hierarchical delineation and discriminant function analysis. It is observed that individual depositional units could be recognized statistically with whole-rock geochemical composition. The strata under study show two second order chemozones comprising six major chemozones that in turn correspond to third order sea level cycles and minor chemozones at the scale of fourth order and/or further shorter sea level cycles. The geochemical signatures showed 100% distinctness between sample populations categorized according to chronostratigraphy and lithostratigraphy. The durations of these stratigraphic units range from 18 million years to less than a million years and indicate distinct geochemical compositional change at different time slices. By implication and also due to the close correspondence between sea level variations reported from this basin and global sea level cycles, it is suggested that recognition and correlation of individual depositional units with distal counterparts could be made accurately. Implication of these results is that stratigraphic units, at varying scales either temporally or spatially, could be assigned with unique geochemical signature, with which accurate prediction and correlation of similar units elsewhere is possible with measurable accuracy.