| Abstract: | Multiple linear regression models calculated from readily obtainable chemical and physical parameters can explain a high percentage (70 per cent or greater) of observed sediment-trace element variance for Cu, Zn, Pb, Cr, Ni, Co, As, Sb, Se, and Hg in a widely divergent suite of 61 sediment samples. The independent variables used in the models may be single parameters, principal component scores, or principal component scores combined with their cross-products. The most useful type of variable must be determined on a case-by-case basis. The independent variables (geochemical parameters) incorporated in the models calculated during this study probably are applicable to many aquatic sediments; albeit, use of a larger data set (>61) could alter the magnitude of the calculated coefficients. The geochemical parameters included in the models were of a physical (e.g. grain size, surface area) and a chemical (e.g. organic matter, amorphous iron oxides) nature. Comparison between actual and predicted trace element concentrations obtained from the models may provide a means of defining ‘average’ sediment-trace element concentrations. In this context, the models may also help identify either naturally or anthropogenically impacted sites for additional study. |
