| Abstract: | The present study demonstrates the usefulness of principal component analysis in condensing and interpreting multivariate time‐series of water quality data. In a case study the water quality system of the lock‐regulated part of the River Neckar (Germany) was analysed, with special emphasis on the oxygen budget. Pooled data of ten water quality parameters and discharge, which had been determined at six stations along a 200 km reach of the river between the years 1993 and 1998, were subjected to principal component analysis. The analysis yielded four stable principal components, explaining 72% of the total variance of the 11 parameters. The four components could be interpreted confidently in terms of underlying processes: biological activity, dilution by high discharge, seasonal effects and the influence of wastewater. From analysing the data of single stations separately, these processes were found to be active throughout the complete reach. Considering the oxygen budget of the river, the variance of biological activity, representing the counteracting processes of primary production and microbial degradation, was found to be most important. This principal component explained 79% of the observed variance of oxygen saturation. In contrast, the analysis of a reduced data set from the 1970s showed that oxygen saturation was then dominated by discharge and temperature variations. The findings indicate that the oxygen budget used to be governed directly by the emission of degradable matter, whereas nowadays eutrophication is most important for extreme oxygen concentrations. Therefore, controlling eutrophication has to be the primary goal, in order to mitigate the rare episodes of pronounced oxygen over‐ and undersaturation in the future. Copyright © 2002 John Wiley & Sons, Ltd. |
