Exploring the Application of Field-Deployed Optical Sensorsfor Real-Time Indication of Fecal Contamination in the Clinton River Watershed

Abstract

Fecal contamination is a wide-spread impairment to riverine and near-shore environments across the state of Michigan. Where fecal contamination is present, harmful human pathogens are likely to exist, potentially resulting in infectious diseases such as giardia, hepatitis, cholera, and other gastrointestinal upsets. Fecal contamination may also carry with it contaminants such as nutrients, pharmaceuticals, endocrine disruptors, and toxic compounds that cause adverse disruptions to aquatic ecosystems. Water quality indicators, such as Escherichia coli (E. coli), are used by state and local governments to characterize the extent of fecal contamination. While E. coli water quality standards have laid the groundwork for monitoring fecal contamination, they are limited by the poor timeliness between E. coli sampling and lab analysis results. Additionally, wet-weather monitoring of E. coli requires substantial personnel availability and poses safety risks for those sampling. However, emergent technologies present a potential solution to these restrictions. Particularly, optical signals such as tryptophan-like-fluorescence (TLF) and optical brighteners (OB), along with nephelometric turbidity, have been correlated with fecal contamination in previous studies. These parameters can be monitored more easily than E. coli, and in a continuous fashion to provide high volumes of data. My research demonstrates the applicability of optical sensors for real-time indication of fecal contamination and provides specific recommendations for using this technology in the Clinton River Watershed. In this study, a team of researchers conducted a six-month pilot study to evaluate the validity and practicality of using these parameters across various hydrologic conditions at three unique monitoring locations. I tested the predictive ability of TLF, OB, and turbidity to detect the magnitude of fecal contamination by correlating these parameters with discrete E. coli samples enumerated through culture-based methods. I demonstrated that all three parameters were significant predictors of E. coli concentration across hydrograph phases, and prediction strength was strongest using data from the ascending limb of the hydrograph. Turbidity and TLF demonstrated either linear or log-linear regression relationships with E. coli in all models. The relationship between OB and E. coli was best fit for a cubic curve using data from the ascending limb of the hydrograph, yet the model for the descending limb was linear. Using TLF and turbidity as predictor variables, values for correlation strength were comparable with previous studies. Correlation strength using OB was highest among all parameters (r = 0.77). Model fit was relatively weak and highly dependent on sample size, demonstrating the need for additional data from multiple storms in the same season. Hydrology patterns influenced seasonal ranges for all three parameters. Additionally, strong diurnal patterns demonstrated the need for instrument-specific temperature correction algorithms. In addition to validating the predictive ability of TLF, OB, and turbidity to detect E. coli, I provide extensive “lessons-learned” on the practicality of using optical sensors for continuously detecting fecal contamination. This study demonstrated noteworthy constraints associated with long-term deployments, compelling practitioners to optimize deployment configurations, maintenance plans, and data post-processing protocol in future applications. Despite these constraints, this study reveals how optical sensors could be used to provide a rapid indication of microbial pollution, and aide practitioners in source tracking of fecal contamination.