SEDIMENTARY ENVIRONMENTAL ANALYSIS OF LONG ISLAND SOUND, USA WITH MULTIVARIATE STATISTICS

ABSTRACT A multivariate statistical strategy, which reduces the sample-variable matrix to a set of interpretable graphical relationships, was used to maximize the environmental information extracted from bottom-sediment, grain-size data for Long Island Sound, USA. The weight-percent whole phi variables were tested for redundancy using R-mode cluster analysis. Q-mode cluster analysis partitioned 57 traverse samples into five facies. This classification was extended to the other 171 samples through discriminant analysis. Ordination was employed to depict the gradational relationships among the samples and facies, and to observe significant environmental and textural parameter gradients within the sample space. Interpretations obtained with the ordination and with other standard techniques were tested by comparing the facies map with the distribution of known environmental phenomena. Five environmentally significant facies were determined for Long Island Sound with the described strategy. These facies are: (1) clayey silt, (2) sandy-clayey silt, (3) silty-clayey sand, (4) silty sand, and (5) sands, which contain individually some-what dissimilar samples. The sand and silty-sand facies are restricted to shoal areas and the margins of the Sound. Most of Long Island Sound, however, is blanketed by sediments which contain a high portion of silt-sized particles. A scarcity of sediment sources is suggested by the fine nature of the sediment.