CHEMOMETRIC EVALUATION OF SURFACE-WATER QUALITY AT REGIONAL LEVEL

To reduce the volume of information needed for the evaluation of the quality of some surface waters designated for purification, data gathered in a previous study on three rivers (Simeto, Alcantara and Oreto) in Sicily have been subjected to statistical multivariate analysis. The data base comprised twentyfive variables from eight sampling points at monthly intervals for one year. Such a complex matrix cannot easily be interpreted and a statistical package have been used to put the data into a more useful form. The three adopted methods (KNN, LDA and SIMCA) gave partially different results as regarding confidence limits and basically convergent results as regarding the geographic classification of the surface waters examined. This analysis also shows that the measured variables cluster into different groups, according to their information content, and provide a reliable criterion for the choice of optimal parameters.