Statistical zonation of sediment samples using ratio matching and cluster analysis

A zonation technique for sediment samples based upon contaminant ratios of a series of samples is described. The technique is based on the fact that sediment samples of common origin will tend to have similar ratios of concentrations of trace pollutants such as heavy metals, PCBs, etc., despite dilution by variable amounts of inert materials such as silica or calcite. Comparison of these ratios between individual samples yields a matrix of similarity coefficients, which are then analyzed by a hierarchical agglomerative cluster analysis procedure. Sediment samples of similar origin will tend to group together in a single cluster, whereas samples influence by unique factors such as transport patterns or proximity to point source discharges will appear as separate individual clusters. The results of application of this methodology to Great Lakes samples from Hamilton Harbour and Port Hope are presented.