Multivariate statistical analysis for estimating surface water quality in reservoirs

Abstract

Regular water quality monitoring programs are an important aspect of water management. Different multivariate statistical techniques were applied for interpretation and evaluation of the data matrix obtained during a six-year monitoring program (2006 to 2011) in the principal reservoirs of the central region of Argentina. Eleven sampling sites located in two reservoirs were surveyed each climatic season for 18 parameters. Cluster analysis grouped the sampling sites into three clusters and classified the different climatic seasons into two clusters based on their similarities. Principal component analysis/factor analysis showed the existence of five significant varifactors (VF) which account for 79.3% of the variance, related to soluble salts, nutrients, physico-chemical parameters, and non-common source. Source contribution was calculated using multiple regression of sample mass concentration on the absolute VF scores. This study demonstrates the usefulness of multivariate statistical techniques helping managers to get better information about surface water systems.