Integration of Ultraviolet-Visible spectral and physicochemical data in chemometrics analysis for improved discrimination of water sources and blends for application to the complex drinking water distribution network of Barcelona

In some cities, concerns over the sustainability of the water distribution system have led to different strategies to cope with the demand in terms of quantity and quality. The Barcelona Metropolitan Area drinking water distribution system is supplied with water from different sources (two rivers, the Mediterranean Sea and groundwater) and is treated with different technologies (classical chlorination, membrane filtering, etc.). Different blending processes occur along the network to homogenize the water quality, both chemically and organoleptically. An appropriate technology that can recognize the water source and blends in real time along the network is crucial for global system management. This study presents a principal component analysis method capable of discriminating samples with respect to their original source and blends using Ultraviolet Visible data from a spectrophotometric probe and a small number of physicochemical parameters. The study began with principal component analysis of 37 physicochemical parameters obtained through standard laboratory procedures to distinguish among sources and blends. Further, the study investigated the possibility of discriminating the same sources and blends using only fingerprints obtained by a spectrophotometric probe. A high correlation coefficient of 0.85 revealed that the information contained in the spectra is comparable to the information provided by more expensive laboratory analyses. Moreover, the discrimination capacity of this method on Ultraviolet Visible data was slightly improved by adding three additional physicochemical parameters: conductivity, fluoride and boron concentrations. Generally, the new model could distinguish the two primary water origins of the Barcelona Metropolitan Area - the Llobregat and the Ter Rivers. The contribution of desalinated seawater and groundwater was also distinguishable in the blends containing river water. Moreover, during the method development, the influence of the water sources and blending on the occurrence and speciation of different trihalomethanes along the city was evaluated. Water from the Llobregat River tends to form brominated trihalomethanes, while water from the Ter River tends to form chlorinated trihalomethanes. The results of this study confirm that the chemometric analysis of a number of selected parameters can provide a sufficient amount of information to operate complex distribution systems using water blending. This method will contribute to the early identification of events, as well as to the operation and management of the network and evaluation of the sustainability of the system by providing information on water quality. (C) 2015 Elsevier Ltd. All rights reserved.