Skip to main content
U.S. flag

An official website of the United States government

Water-quality and streamflow datasets used in Weighted Regressions on Time, Discharge, and Season (WRTDS) models to determine trends in the Nation’s rivers and streams, 1972-2017

December 2, 2021

In 1991, the U.S. Geological Survey (USGS) began a study of more than 50 major river basins across the Nation as part of the National Water-Quality Assessment (NAWQA) project. One of the major goals of the NAWQA project was to determine how river water quality has changed over time. To support that goal, long-term consistent and comparable monitoring has been conducted by the USGS on streams and rivers throughout the Nation. Outside of the NAWQA project, the USGS and other Federal, State, and local agencies also have collected long-term water-quality data to support their own assessments of changing water quality. In 2017, data from these multiple sources were combined to support one of the most comprehensive assessments to date of water-quality trends in the United States (Oelsner and others, 2017; De Cicco and others, 2017). This data release updates these water quality trends, which ended in 2012, with 5 more years of data and now end in 2017.

This USGS data release contains all the input and output files necessary to reproduce the results from the Weighted Regressions on Time, Discharge, and Season (WRTDS) models, using data preparation methods described in Oelsner and others, 2017. Models were calibrated for each combination of site and parameter using the screened input data. Models were run on Yeti, the USGS supercomputer, in 3 separate runs. See readMe.txt for more details. Once calibrated, the WRTDS models were initially evaluated using a logistic regression equation that estimated a probability of acceptance for each model (e.g., "a good fit") based on a set of diagnostic metrics derived from the observed, estimated, and residual values from each model and data set. Each WRTDS model was assigned to one of three categories: “auto-accept,” “auto-reject,” or “manual evaluation". Models assigned to the latter category were visually evaluated for appropriate model fit using residual and diagnostic plots. Models assigned to the first two categories were automatically included or rejected from the final results, respectively. Twenty-two water-quality parameters were assessed, including nutrients (ammonia, nitrate, filtered orthophosphate, total nitrogen, total phosphorus, and unfiltered orthophosphate), major ions (calcium, bromide, fluoride, chloride, magnesium, potassium, sodium, and sulfate), salinity indicators (total dissolved solids and specific conductance), sediment (total suspended solids and suspended sediment concentration), carbon (dissolved organic carbon, total organic carbon, and particulate organic carbon), and alkalinity. Trends are reported for six trend periods: 1972-2017, 1982-2017, 1987-2017, 1992-2017, 2002-2017, and 2007-2017.