Continuous field estimation of dissolved organic carbon concentration and biochemical oxygen demand using dual‐wavelength fluorescence, turbidity and temperature. Issue 3 (28th November 2016)

• Record Type: Journal Article
• Authors: Khamis, Kieran; Bradley, Chris; Stevens, Rob; Hannah, David M.
• Is Part Of: Hydrological processes. Volume 31:Issue 3(2017)
• Page Start: 540
• Page End: 555
• Publication Date: 2016-11-28
• Subjects: in‐situ monitoring -- optical sensors -- urban river systems -- Water quality; Hydrology -- Periodicals; Hydrology -- Research -- Periodicals; Hydrologic models -- Periodicals; Hydrological forecasting -- Periodicals; 631.432
• DOI: 10.1002/hyp.11040
• Languages: English
• ISSNs: 0885-6087

Abstract

Abstract: Dissolved organic matter (DOM) quality and quantity is not measured routinely in‐situ limiting our ability to quantify DOM process dynamics. This is problematic given legislative obligations to determine event based variability; however, recent advances in field deployable optical sensing technology provide the opportunity to address this problem. In this paper, we outline a new approach for in‐situ quantification of DOM quantity (Dissolved Organic Carbon: DOC) and a component of quality (Biochemical Oxygen Demand: BOD) using a multi‐wavelength, through‐flow fluorescence sensor. The sensor measured tryptophan‐like (Peak T) and humic‐like (Peak C) fluorescence, alongside water temperature and turbidity. Laboratory derived coefficients were developed to compensate for thermal quenching and turbidity interference (i.e., light attenuation and scattering). Field tests were undertaken on an urban river with ageing wastewater and stormwater infrastructure (Bourn Brook; Birmingham, UK). Sensor output was validated against laboratory determinations of DOC and BOD collected by discrete grab sampling during baseflow and stormflow conditions. Data driven regression models were then compared to laboratory correction methods. A combination of temperature and turbidity compensated Peak T and Peak C was found to be a good predictor of DOC concentration ( R 2 = 0.92). Conversely, using temperature and turbidity correction coefficients provided low predictive power for BOD ( R 2 = 0.46 and R 2 = 0.51, for Peak C and T, respectively). For this study system, turbidity appeared to be a reasonable proxy for BOD, R 2 = 0.86. However, a linear mixed effect model with temperature compensated Peak T and turbidity provided a robust BOD prediction ( R 2 = 0.95). These findings indicate that with careful initial calibration, multi‐wavelength fluorescence, coupled with turbidity, and temperature provides a feasible proxy for continuous, in‐situ measurement of DOC concentration and BOD. This approach represents a cost effective monitoring solution, particularly when compared to UV – absorbance sensors and DOC analysers, and could be readily adopted for research and industrial applications.