Two in situ optical sensors, a single-excitation fluorescence-based sensor (fDOM) mounted on a multi-parameter EXO2 sonde (YSI), and a stand-alone, multispectral absorbance-based instrument (spectro::lyser, scan Messtechnik GmbH), were evaluated for their capability to (i) estimate river dissolved organic carbon (DOC) concentrations and (ii) provide oversight of drinking water production. The sensors were deployed between March and November 2017 in the river Fyris, which drains a mixed forested and agricultural 2003 km² catchment and serves as a drinking water source by managed aquifer recharge. Grab samples were collected every 2 to 3 weeks and compared with logged sensor data collected at 15 minute intervals. The fDOM probe signal was used to estimate DOC concentrations in the range of 10.4 to 24.4 mg L⁻¹ using linear regression (R² = 0.71, RMSE = 2.5 mg L⁻¹), after correction for temperature, turbidity and inner-filter effects. Temporal changes in DOC character associated with the mixed land use landscape, as indicated by optical indices, reduced this sensor accuracy for estimating DOC concentration. Nevertheless, humic substance concentrations, the fraction of DOC that is preferentially removed during artificial infiltration, were well captured. The spectrolyser signal was used to establish a 2-component partial least square model that captured DOC fluctuations from 10.2 to 29.4 mg L⁻¹ (R² = 0.92; RMSE = 1.3 mg L⁻¹). This multiple-wavelength model (220 to 720 nm) effectively handled the changes in DOC composition while accurately estimating DOC concentrations. This study explores the advantages and limitations of optical sensors for their use in managed aquifer recharge and drinking water production in relation to DOC levels.

中文翻译：

---

评估两个原位光学传感器的准确性，以估算饮用水生产中的DOC浓度

两个原位光学传感器，一个安装在多参数EXO2探空仪（YSI）上的单激发荧光传感器（fDOM），以及一个基于多光谱吸收的独立仪器（spectro :: lyser，scan Messtechnik GmbH）评估了它们对（i）估算河流溶解的有机碳（DOC）浓度和（ii）提供饮用水生产监督的能力。传感器于2017年3月至11月之间部署在Fyris河中，该河排水了2003 km ²集水区，并通过有管理的含水层补给作为饮用水源。每2至3周收集一次抓取样品，并将其与以15分钟为间隔收集的记录的传感器数据进行比较。使用fDOM探针信号通过线性回归来估算10.4至24.4 mg L ^-1范围内的DOC浓度（R ² = 0.71，RMSE = 2.5 mg L ^-1），校正温度，浊度和内部过滤效果之后。如光学指标所示，与混合土地利用景观相关联的DOC特性的时间变化降低了用于估算DOC浓度的传感器精度。然而，腐殖质浓度（在人工渗透过程中优先去除的DOC分数）已被很好地捕获。光谱仪信号用于建立2分量偏最小二乘模型，该模型捕获了从10.2至29.4 mg L ^-1（R ² = 0.92; RMSE = 1.3 mg L ^-1）。这种多波长模型（220至720 nm）可有效处理DOC组成的变化，同时准确估算DOC浓度。这项研究探索了与DOC水平相关的光学传感器在受控含水层补给和饮用水生产中的优势和局限性。