Regression models developed to estimate Escherichia coli concentrations using 2019-24 data at four stream sites in Chester and Delaware Counties, Pennsylvania

Regressions were developed to predict Escherichia coli (E. coli) concentrations at four U.S. Geological Survey (USGS) gages on streams in Chester and Delaware Counties in southeastern Pennsylvania. The regression analysis was based on 2019-24 discrete E. coli concentrations measured in samples collected approximately monthly by USGS from April to October and concurrent available streamflow and water-quality variables. Available water-quality variables included turbidity (Turb), water temperature, specific conductance (SC), pH and dissolved oxygen concentration measured continuously by monitors operated by USGS at the gages. The data used for regression and results of regression analysis are described in the Model Archive Summary for each site. Through the regression analysis, the variables that best predicted E. coli (EC) relative to water-quality standards were identified. Log10(Turb), seasonality terms, sin2piD_365 and cos2piD_365, and for some sites, SC, selected as explanatory variables in linear regression equations as the best predictors of log10(EC) based on residual plots and regression statistics. Sensitivity, specificity, positive predictive values, and negative predictive values were computed for the regression dataset to assess model reliability for predicting concentrations above and below a Pennsylvania Department of Environmental Protection (PADEP) standard of 410 colonies forming units (CFU) per 100 milliliters (mL) for E. coli in recreational water for the swimming season of May 1 through September 30. The data collection and regression analyses were done by USGS in cooperation with Chester County Health Department (CCHD) and Chester County Water Resources Authority (CCWRA).

[
  "010:12"
]

[
  "geospatial"
]