UNSUPERVISED PATTERN-RECOGNITION TECHNIQUES TO INVESTIGATE MICROBIOLOGICAL AND PHYSICOCHEMICAL VARIATIONS IN CORRIENTE RIVER

MORESI, L; FECHNER, DIANA C.; RUIZ DIAZ, J. D.; VAZQUEZ, F.A; MARCHEVSKY, E; PELLERANO, R.G.

San Luis

Congreso; SOCIEDAD DE BIOLOGIA DE CUYO; 2015

SOCIEDAD BIOLOGIA DE CUYO

The aim of this study was to obtain a deeper understanding of temporal trends and sources of water pollution in different functional zones by using unsupervised pattern recognition techniques with a case study of the Corriente River, Argentina. Microbiological and physicochemical measurement data were obtained from 5 monitoring sites in three categories of functional zones during the period 2012?2014. Microbiological analysis was performed for faecal indicator bacteria (FIB) including Escherichia coli (E. coli) and Enterococcus (ENT). The higher concentrations of FIB determined in the present study were 1.6 × 103 and 2.7 × 103 CFU 100 mL−1 for E. coli and ENT, respectively. In addition, unsupervised chemometric techniques were applied to the result dataset, namely, Principal component (PCA) and Cluster analyses (CA). The PCA demonstrated that the samples collected in summer shown higher concentrations of FIB strongly correlated with high levels of temperature, ammonium-N, chemical oxygen demand, and electrical conductivity. Finally, some results achieved by the CA were also found by the PCA technique, demonstrating the complementariness of the methods.