Application of discriminant analysis and manova to grain-size data on a study of the distribution and movement of dredged sediment

A dredged material disposal operation was monitored at a location in Lake Erie, 8 km offshore at Ashtabula, Ohio, in 1975. Approximately 200 sediment cores were collected from 12 experimental and 4 control locations before and after dredging and analyzed for the grain-size distribution and related heavy-metal content. The dredged sediments were similar to those from the lake bottom at the disposal and control sites. Because of this similarity, it was extremely difficult to distinguish between the dredged material and the lake bottom sediments without tagging the material with dyes or radioactive isotopes. A sequence consisting of a linear discriminant analysis followed by a univariate and multivariate analysis of variance was successfully applied to discriminate between the dredged and the original lake sediments. Results indicate that 4 months after the disposal operation some stations had returned to predisposal conditions, a probable result of currents stripping dredged material off the lake bottom. The analysis of variance indicated that the clay-size fraction was responsible for initial changes in the grain-size distribution. Storm induced scouring caused an eventual return of the grain-size distribution to predisposal conditions. In support of this observation, the concentrations of iron and zinc, which were statistically correlated to the clay size fraction, also exhibited the same trends.