ABSTRACT

A multivariate statistical strategy employing cluster analysis, ordination, and discriminant analysis was used to maximize the environmental information extracted from the Barataria Bay, Louisiana, grain-size data tabulated in the classic study by Krumbein and Aberdeen (1937). The weight percent whole phi variables were first tested for redundancy using R-mode cluster analysis. Then the samples were partitioned into five environmentally significant facies using Q-mode cluster analysis: (1) beach sand, (2) foreshore sand, (3) silty channel sand, (4) silty channel margin sand, and (5) organic silt and mud. Wilk's lambda and discriminant analysis tested the statistical significance of these facies. They were interpreted by using a two-dimensional ordination and verified by examining the field distribution of the five facies and the x-ordination coordinates. Discriminant analysis can be used to extend the classification to other Barataria Bay samples of unknown affinity.

This approach provides the investigator with a powerful tool for sedimentary environmental analysis, since it can be applied to obtain geologically significant results independent of any knowledge concerning the spatial distribution of the samples. It should be applicable to the study of ancient as well as modern sedimentary environments.