Water quality assessment using SVD-based principal component analysis of hydrological data

Abstract

Principal component analysis (PCA) based on singular value decomposition (SVD) of hydrological data was tested for water quality assessment. Using two case studies of waste- and drinking water, PCA via SVD was able to find latent variables which explain 80.8% and 83.7% of the variance, respectively. By means of scatter and loading plots, PCA revealed the relationships among samples and hydrochemical parameters which were also confirmed by factor analysis (FA). In the case of wastewater, these latent variables clearly displayed changes of water composition over time. Drinking water samples were clustered into four groups which were characterised by their typical water composition. On the basis of these results PCA was found to be a suitable technique for water quality assessment.