Inter-Calibration of Optical Sensors and Implications for Water Quality Monitoring

Abstract: Expected changes in water quality due to a changing climate associated with variabilities in rainfall, runoff and diffusive spreading of pollution pose an emerging risk for the drinking water supply. Extensive monitoring of water quality parameters is of importance and the possibility of sudden extreme events brings up the need of high-frequency measurements. Dissolved organic carbon (DOC) has frequently been used as indicator for quality due to its capability of transporting contaminants and heavy metals, and its negative influence on the water treatment process. Hence, two sensors equipped with optical probes measuring fluorescence at an excitation of 365 nm and emission at 480 nm (EXO) and absorbance from 220 to 720 nm (s::can) were deployed in the Fyris river in Uppsala, central Sweden to approximate DOC concentrations. This is possible as the wavelength at which absorbance and fluorescence occur are unique for each molecule. Furthermore, the EXO carried a temperature and a turbidity probe. Measurements were conducted every 15 minutes for a period of about four months. Supplementary experiments improved the development of a calibration method for compensating temperature and turbidity effects on fluorescence measurements, as well as the generation of an algorithm to estimate DOC and turbidity from spectral absorbance data. Very stable weather and discharge conditions allowed calibration only for a small data range in all variables. However, this condition favored the chance of observing a change in DOC character throughout the measuring period, which was confirmed by performing monthly laboratory analysis of DOC, TOC (total OC), fluorescence and absorbance at four different stations along the river. It became apparent that both sensors do not capture this change in composition very well, whereupon the s::can performs slightly better. The rather simple algorithm delivers reasonable estimates of DOC and turbidity, which can be improved when data with a greater variability are available. Inter-calibration of the sensors demonstrated systematic errors between the two and laboratory measurements. The dry-spell in Uppsala revealed how the parameters from different stations become more alike as the groundwater fraction of the stream water increases. The change in character depicted by the specific absorbance in the ultraviolet range suggests that during water treatment greater effort in the flocculation process will be needed.