Parallel factor analysis and multivariate curve resolution as data fusion tools to supervise a stream

Abstract

In this work, a new method is proposed to monitor the distribution, evolution and correlation of dissolved organic matter on the superficial water of a stream with respect to physicochemical variables that characterize the basin and season sampling of each campaign. The method is based on measuring fluorescence emission-excitation matrices and some physicochemical parameters of water samples through both time and space. In a first phase, parallel factor analysis (PARAFAC) or multivariate curve resolution with alternating least-squares (MCR-ALS) were applied to extract the information on the relative proportions of each fluorophore on each sample. Then, MCR-ALS was applied again to the entire database, in order to study the spatial and time distribution. This methodology was used to study the behavior of a basin stream that is significantly modified by anthropic activities.