Enhanced statistical evaluation of fluorescence properties to identify dissolved organic matter dynamics during river high-flow events.

Fluorescence spectroscopy has become a widely used technique to characterize dissolved organic matter (DOM) and organic hazardous micro-pollutants in natural and human-influenced water bodies. Especially in rivers highly impacted by municipal and industrial wastewater treatment plants' effluents, the fluorescence signal at low-flow is mainly dominated by these discharges. At river high-flow, their influence decreases due to dilution effects, and at the same time, other compounds of DOM, stemming from diffuse inputs, can increase or even dominate. Therefore, whereas the analysis of DOM is little informative on the changing sources and pathways of emissions, fluorescence spectroscopy can enhance our understanding and our possibilities of monitoring such dynamics in river catchments. This paper analyzed samples from seven high-flow events in an Austrian river. Firstly, independent DOM components were discriminated using a parallel factor analysis (PARAFAC) to show the varying composition of DOM during different phases of high-flow events. Furthermore, partial least squares (PLS) and sparse PLS (sPLS) regression were applied to identify excitation and emission wavelengths, serving as proxy parameters for quantifying dissolved organic carbon (DOC) and chloride. The PLS models show the best prediction accuracy but use the entire excitation-emission matrix in exchange. In selecting predictors, the use of excitation and emission wavelengths adjusted via sPLS is superior to the extracted PARAFAC components. The sPLS model yields 16 wavelength combinations for DOC (RMSE

Informations de copyright

Copyright © 2022 The Authors. Published by Elsevier B.V. All rights reserved.

Déclaration de conflit d'intérêts

Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.