Modeling Escherichia coli in the Missouri River near Omaha, Nebraska, 2012–16

Authors

Brenda K. Densmore, U.S. Geological Survey
Brent M. Hall, U.S. Geological Survey
Matthew T. Moser, U.S. Geological Survey

Document Type

Article

Date of this Version

2020

Citation

Densmore, B.K., Hall, B.M., and Moser, M.T., 2020, Modeling Escherichia coli in the Missouri River near Omaha, Nebraska, 2012–16: U.S. Geological Survey Scientific Investigations Report 2020–5045, 24 p., https://doi.org/10.3133/sir20205045.

Abstract

The city of Omaha, Nebraska, has a combined sewer system in some areas of the city. In Omaha, Nebr., a moderate amount of rainfall will lead to the combination of stormwater and untreated sewage or wastewater being discharged directly into the Missouri River and Papillion Creek and is called a combined sewer overflow (CSO) event. In 2009, the city of Omaha began the implementation of their Long Term Control Plan (LTCP) to mitigate the effects of CSOs on the Missouri River and Papillion Creek. As part of the LTCP, the city partnered with the U.S. Geological Survey (USGS) in 2012 to begin monitoring in the Missouri River. Since 2012, monthly discrete water-quality samples for many constituents have been collected from the Missouri River at four sites. At 3 of the 4 sites, water quality has been monitored continuously for selected constituents and physical properties. These discrete water-quality samples and continuous water-quality monitoring data (from July 2012 to 2020) have been collected to better understand the water quality of the Missouri River, how it is changing with time, how it changes upstream from the city of Omaha to downstream, and how it varies during base-flow conditions and during periods of runoff. The purpose of this report is to document the development of Escherichia coli (E. coli) concentration models for these four Missouri River sites. Analysis was completed using the first 5 years of data (through 2016) to determine if the current approach is sufficient to meet future analysis goals and to understand if proposed models such as Load Estimator (LOADEST) models will be able to represent water-quality changes in the Missouri River. Multiple linear regression models were developed to estimate E. coli concentration using LOADEST as implemented in the rloadest package in the R statistical software program. A set of explanatory variables, including streamflow and streamflow anomalies, precipitation, information about CSOs, and continuous water quality, were evaluated for potential inclusion in regression models. The best model at Missouri River at NP Dodge Park at Omaha, Nebr. (USGS station 412126095565201; hereafter “NP Dodge”) included basin explanatory variables of upstream antecedent precipitation index measured at Tekamah, Nebr.; decimal time; season; and turbidity. The best model at Missouri River at Freedom Park Omaha, Nebr. (USGS station 411636095535401; hereafter “Freedom Park”) included the same explanatory variables as the NP Dodge model with the addition of turbidity anomalies and flow anomalies. The best models at the two downstream sites (Missouri River near Council Bluffs, Iowa, USGS station 06610505 and Missouri River near La Platte, Nebr., USGS station 410333095530101) included the same explanatory variables as the Freedom Park model with the addition of local antecedent precipitation index as measured at Eppley Airport in Omaha, Nebr., and additional turbidity and flow anomalies. The final selected models were the best models given our modeling design constraint in which explanatory variables included in the model for the upstream site were included in the downstream models. Explanatory variables currently (2020) being collected and included in the selected models through 2016 explained 64–75 percent of the variability of E. coli concentration in the Missouri River. Explaining 64–75 percent of the variability might be considered low when working with physical constituents (total nitrogen or sediment), but with the natural variability of biological constituents such as E. coli, the uncertainty of E. coli laboratory measurements, and the added complexity of modeling in a large drainage basin with multiple sources, these results are adequate and indicate that the explanatory variables being collected and models such as LOADEST can represent water-quality changes in the Missouri River for E. coli concentration from 2012 to 2016.