This data set includes input data for the development of regression models to predict chloride from specific conductance (SC) data at 56 U. S. Geological Survey water quality monitoring stations in the eastern United States. Each site has 20 or more simultaneous observations of SC and chloride. Data were downloaded from the National Water Information System (NWIS) using the R package dataRetrieval. Datasets for each site were evaluated and outliers were removed prior to the development of the regression model. This file contains only the final input dataset for the regression models. Please refer to Moore and others (in review) for more details. Moore, J., R. Fanelli, and A. Sekellick. In review. High-frequency data reveal deicing salts drive elevated conductivity and chloride along with pervasive and frequent exceedances of the EPA aquatic life criteria for chloride in urban streams. Submitted to Environmental Science and Technology.

This data set contains 18 metrics used to describe patterns in specific conductance (SC) and chloride concentrations in 93 streams located across the eastern United States. These data were quantified for an analysis described in Moore and others (in review). All metrics were quantified for a water year and a median was taken across all years for which data were available to provide a single value for each site. High-frequency SC and chloride were measured or estimated at sub-daily time steps from 2-minute intervals to hourly intervals (e.g., high-frequency) depending on the site. Moore, J., R. Fanelli, and A. Sekellick. In review. High-frequency data reveal deicing salts drive elevated conductivity and chloride along with pervasive and frequent exceedances of the EPA aquatic life criteria for chloride in urban streams. Submitted to Environmental Science and Technology.

This data set contains 18 metrics used to describe patterns in specific conductance (SC) and chloride concentrations in 93 streams located across the eastern United States. These data were quantified for an analysis described in Moore and others (in review). All metrics were quantified for a water year and a median was taken across all years for which data were available to provide a single value for each site. High-frequency SC and chloride were measured or estimated at sub-daily time steps from 2-minute intervals to hourly intervals (e.g., high-frequency) depending on the site. Moore, J., R. Fanelli, and A. Sekellick. In review. High-frequency data reveal deicing salts drive elevated conductivity and chloride along with pervasive and frequent exceedances of the EPA aquatic life criteria for chloride in urban streams. Submitted to Environmental Science and Technology.

The file "Chloride_specific_conductance_regression_model_forms_for_estimating_high-frequency_chloride_concentrations.csv" contains the regression equation forms for two types of regressions: 1) single linear (SLR) and 2) piecewise (or segmented; SEG) regression between specific conductance (SC) and chloride (Cl) concentrations for 56 USGS water-quality monitoring stations across the eastern United States, plus four regional regressions developed by pooling data for sites within a region (see Moore and others (in review) for more information). Some sites, and all regions, have both SLR and SEG models reported in this table. The analysis included in the Moore and others (in review) study used results from the SLR models if those are the only model reported, or the SEG models are both reported. This data set includes the slope, intercept, model correlation coefficient (R2) and the number of observations used to develop the regression equation for the SLR models. For the SEG models, information on the slopes and intercepts for the two linear segments are included in addition to the coefficient of determination (R2) and number of observations. Breakpoint estimates (the point separating the two line segments) and uncertainties are only reported for the SEG models. Moore, J., R. Fanelli, and A. Sekellick. In review. High-frequency data reveal deicing salts drive elevated conductivity and chloride along with pervasive and frequent exceedances of the EPA aquatic life criteria for chloride in urban streams. Submitted to Environmental Science and Technology.

The file "Chloride_specific_conductance_regression_model_forms_for_estimating_high-frequency_chloride_concentrations.csv" contains the regression equation forms for two types of regressions: 1) single linear (SLR) and 2) piecewise (or segmented; SEG) regression between specific conductance (SC) and chloride (Cl) concentrations for 56 USGS water-quality monitoring stations across the eastern United States, plus four regional regressions developed by pooling data for sites within a region (see Moore and others (in review) for more information). Some sites, and all regions, have both SLR and SEG models reported in this table. The analysis included in the Moore and others (in review) study used results from the SLR models if those are the only model reported, or the SEG models are both reported. This data set includes the slope, intercept, model correlation coefficient (R2) and the number of observations used to develop the regression equation for the SLR models. For the SEG models, information on the slopes and intercepts for the two linear segments are included in addition to the coefficient of determination (R2) and number of observations. Breakpoint estimates (the point separating the two line segments) and uncertainties are only reported for the SEG models. Moore, J., R. Fanelli, and A. Sekellick. In review. High-frequency data reveal deicing salts drive elevated conductivity and chloride along with pervasive and frequent exceedances of the EPA aquatic life criteria for chloride in urban streams. Submitted to Environmental Science and Technology.

This data set includes estimated chloride concentrations for the 93 USGS water quality monitoring stations located across the eastern United States. Chloride concentrations were predicted using regression equations that established the relationship between simultaneous measurements of chloride and specific conductance (SC). Site-specific models were developed and applied when data were available, and regional regression models were used where there were insufficient data available to establish a site-specific regression model. These models were applied to high-frequency SC data sets to produce high-frequency predicted chloride concentrations at 2-minute to 1-hour intervals, depending on the frequency at which SC is measured at each site. Up to four chloride concentration estimations may be available for a particular site, depending on whether a simple linear regression was developed (SLR), a piecewise regression was developed (SEG), and whether or not site-specific models were developed and applied. The best-fitting regional regression model was applied to all 93 sites. Regions include the southeast, Mid-Atlantic, Mid-Atlantic carbonate (e.g., underlain with carbonate bedrock), and New England, and correspond to the regions included in Moore and others (in review). Each file is labeled by its short name. Please see "Chloride_site_information.csv" for the corresponding site IDs and the USGS station numbers. Moore, J., R. Fanelli, and A. Sekellick. In review. High-frequency data reveal deicing salts drive elevated conductivity and chloride along with pervasive and frequent exceedances of the EPA aquatic life criteria for chloride in urban streams. Submitted to Environmental Science and Technology.

This data set includes estimated chloride concentrations for the 93 USGS water quality monitoring stations located across the eastern United States. Chloride concentrations were predicted using regression equations that established the relationship between simultaneous measurements of chloride and specific conductance (SC). Site-specific models were developed and applied when data were available, and regional regression models were used where there were insufficient data available to establish a site-specific regression model. These models were applied to high-frequency SC data sets to produce high-frequency predicted chloride concentrations at 2-minute to 1-hour intervals, depending on the frequency at which SC is measured at each site. Up to four chloride concentration estimations may be available for a particular site, depending on whether a simple linear regression was developed (SLR), a piecewise regression was developed (SEG), and whether or not site-specific models were developed and applied. The best-fitting regional regression model was applied to all 93 sites. Regions include the southeast, Mid-Atlantic, Mid-Atlantic carbonate (e.g., underlain with carbonate bedrock), and New England, and correspond to the regions included in Moore and others (in review). Each file is labeled by its short name. Please see "Chloride_site_information.csv" for the corresponding site IDs and the USGS station numbers. Moore, J., R. Fanelli, and A. Sekellick. In review. High-frequency data reveal deicing salts drive elevated conductivity and chloride along with pervasive and frequent exceedances of the EPA aquatic life criteria for chloride in urban streams. Submitted to Environmental Science and Technology.

High frequency estimated chloride (Cl) and observed specific conductance (SC) data sets, along with response variables derived from those data sets, were used in an analysis to quantify the extent to which deicer applications in winter affect water quality in 93 U.S. Geological Survey water quality monitoring stations across the eastern United States. The analysis was documented in the following publication: Moore, J., R. Fanelli, and A. Sekellick. In review. High-frequency data reveal deicing salts drive elevated conductivity and chloride along with pervasive and frequent exceedances of the EPA aquatic life criteria for chloride in urban streams. Submitted to Environmental Science and Technology. This data release contains five child items: 1) Input datasets of discrete specific conductance (SC) and chloride (Cl) observations used to develop regression models describing the relationship between chloride and SC 2) The predicted chloride concentrations generated by applying the sites-specific and regional regression models to high-frequency SC datasets 3) The regression equations for 56 USGS water quality monitoring stations across the eastern Unite States, as well as three regions 4) Response variables describing temporal patterns in SC and chloride, calculated by using the estimated high-frequency chloride time series datasets and high-frequency SC datasets 5) Watershed characteristics describing the land use, geology, climate, and deicer application rates for the 93 watersheds included in the Moore et. al (in review) study.

High frequency estimated chloride (Cl) and observed specific conductance (SC) data sets, along with response variables derived from those data sets, were used in an analysis to quantify the extent to which deicer applications in winter affect water quality in 93 U.S. Geological Survey water quality monitoring stations across the eastern United States. The analysis was documented in the following publication: Moore, J., R. Fanelli, and A. Sekellick. In review. High-frequency data reveal deicing salts drive elevated conductivity and chloride along with pervasive and frequent exceedances of the EPA aquatic life criteria for chloride in urban streams. Submitted to Environmental Science and Technology. This data release contains five child items: 1) Input datasets of discrete specific conductance (SC) and chloride (Cl) observations used to develop regression models describing the relationship between chloride and SC 2) The predicted chloride concentrations generated by applying the sites-specific and regional regression models to high-frequency SC datasets 3) The regression equations for 56 USGS water quality monitoring stations across the eastern Unite States, as well as three regions 4) Response variables describing temporal patterns in SC and chloride, calculated by using the estimated high-frequency chloride time series datasets and high-frequency SC datasets 5) Watershed characteristics describing the land use, geology, climate, and deicer application rates for the 93 watersheds included in the Moore et. al (in review) study.

This data release contains data used to develop linear regression models to calculate continuous chloride concentrations in groundwater and runoff at an urban rain garden in Gary, Indiana. A python script was developed to perform the regression analysis and estimate sodium chloride (NaCl) loading to 5 rain-garden flumes as the product of: (1) the estimated continuous (one-minute) chloride concentration, (2) the mass ratio of sodium chloride to chloride, and (3) continuous flume discharge. The python source code used to execute the regression model and loading calculations is included in this data release in the zipped folder named Model_Archive.zip. The regression input (discrete specific conductance and chloride concentration data), chloride concentration model input (continuous specific conductance data) and NaCl loading model input (continuous flume discharge and specific conductance data) that were used to estimate chloride concentrations at 3 USGS monitoring wells (413610087201001, 413612087201301, 413611087201004) and NaCl loading at 5 USGS rain garden flume monitoring stations (413611087201101, 413611087201001, 413612087200901, 413611087200901, 413611087201002) are also included as part of the model archive (Model_Archive.zip). The model output consists of 3 .csv files for the USGS monitoring wells with estimated continuous (hourly) chloride concentrations (mg/L) and 5 .csv files for the USGS flume monitoring sites with estimated continuous (1-minute) chloride concentrations (mg/L) and NaCl loading (grams) that are presented in the zipped folder Model_Output_Data.zip.