Environmental data, including river pollution data, are characterized
by high variability. Much information is lost by using only univariate
graphical or statistical methods for data evaluation and interpretati
on. Chemometric methods, in particular methods of multivariate data an
alysis, help to extract the latent information in such data. The combi
nation of cluster analysis as the first step and multivariate analysis
of variance and discriminant analysis as the second step enables iden
tification of similar locations in a river. Pollution sources and disc
hargers can be detected by means of factor analysis. The deposition-re
mobilization behavior of metals in a river can be described using part
ial least squares regression. Summarizing, it can be stated that metho
ds of multivariate data analysis are powerful tools for the evaluation
and interpretation of river pollution data. (C) 1998 Academic Press.