Uncertainty in monitoring E. coli concentrations in streams and stormwater runoff

Authors: Harmel, Daren; HATHAWAY, J - University Of Tennessee; WAGNER, K - Texas Water Resources Institute; WOLFE, J - Blackland Research And Extension Center; KARTHIKEYAN, R - Texas A&M University; FRANCESCONI, W - International Center For Tropical Agriculture (CIAT); MCCARTHY, D - Monash University

Submitted to: Journal of Hydrology
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 1/19/2016
Publication Date: 2/16/2016
Publication URL: http://handle.nal.usda.gov/10113/61984
Citation: Harmel, R.D., Hathaway, J.M., Wagner, K.L., Wolfe, J.E., Karthikeyan, R., Francesconi, W., McCarthy, D.T. 2016. Uncertainty in monitoring E. coli concentrations in streams and stormwater runoff. Journal of Hydrology. 534:524-533.

Interpretive Summary: Microbial contamination in surface waters is a substantial public health concern throughout the world. Contamination is typically identified by, and restoration plans developed upon, fecal indicator bacteria often one called E. coli. Thus, monitoring of E. coli concentrations is critical to evaluate current conditions, determine restoration effectiveness, and inform model development and calibration. An often overlooked component of these monitoring and modeling activities is understanding the inherent uncertainty present in measured data. In this research, a review and subsequent analysis was performed to identify, document, and analyze the uncertainty of stream flow and stormwater runoff E. coli data. Uncertainty contributed by sample collection, sample preservation and storage, and laboratory analyses were compiled and presented, and differences in sampling method and data quality scenarios were compared. These analyses yielded several interesting results, including: 1) bridges with moderate to high nesting bird populations potentially introduce large increases in uncertainty for E. coli data collected immediately downstream; 2) manual integrated sampling produced the lowest uncertainty in individual samples, but automated sampling typically produced the lowest uncertainty when sampling whole runoff events; 3) sample collection often contributed the highest amount of uncertainty, although laboratory analysis introduced substantial uncertainty and preservation/storage introduced substantial uncertainty under some scenarios; and 4) the uncertainty in measured E. coli concentrations was greater than that of sediment and nutrients, but the difference is not as great as is commonly assumed. This research provides a comprehensive analysis of uncertainty in E. coli concentrations measured in streamflow and runoff. The results are valuable in designing E. coli monitoring projects, in reducing uncertainty in quality assurance efforts, in regulatory and policy decision making, and in fate and transport modeling.

Technical Abstract: Microbial contamination in surface waters is a substantial public health concern throughout the world. Contamination is typically identified by, and restoration plans predicated upon, fecal indicator bacteria such as E. coli. Thus, monitoring of E. coli concentrations is critical to evaluate current conditions, determine restoration effectiveness, and inform model development and calibration. An often overlooked component of these monitoring and modeling activities is understanding the inherent random and systematic uncertainty present in measured data. In this research, a review and subsequent analysis was performed to identify, document, and analyze the uncertainty of stream flow and stormwater runoff E. coli data. Uncertainty contributed by sample collection, sample preservation and storage, and laboratory analyses were compiled and presented, and differences in sampling method and data quality scenarios were compared. These analyses yielded several interesting results, including: 1) bridges with moderate to high nesting bird populations potentially introduce large increases in uncertainty for E. coli data collected immediately downstream; 2) manual integrated sampling produced the lowest random and systematic uncertainty in individual samples, but automated sampling typically produced the lowest uncertainty when sampling whole runoff events; 3) sample collection often contributed the highest amount of uncertainty, although laboratory analysis introduced substantial random uncertainty and preservation/storage introduced substantial systematic uncertainty under some scenarios; and 4) the uncertainty in measured E. coli concentrations was greater than that of sediment and nutrients, but the difference is not as great as is commonly assumed. This research provides a comprehensive analysis of uncertainty in E. coli concentrations measured in streamflow and runoff. The results are valuable in designing E. coli monitoring projects, in reducing uncertainty in quality assurance efforts, in regulatory and policy decision making, and in fate and transport modeling.