Revision History for OFR 2017-1120 Casey J. Lee, Jennifer C. Murphy, Charles G. Crawford, and Jeffery R. Deacon Methods for computing water-quality loads at sites in the U.S. Geological Survey National Water Quality Network ----------------------------------------------------- Posted online October 24, 2017 ----------------------------------------------------- Revised and reposted January 2020, version 1.1 The text version was modified to reflect updates to the network, to describe an adjustments to methods based on newly available reports, and to clarify existing language. Adjustments are as follows: Adjustments to made on page 1 and in table 1 to reflect the removal of the the Orestimba Creek near Crows Landing, CA site in 2018. Text was added on page 1 to indicate that loads and trends are computed using the Weighted Regressions on Time, Discharge, and Season (WRTDS) and Weighted Regressions on Time, Discharge and Season with Kalman Filtering (WRTDS-K) methods. Text was added on page 5 to indicate that water-quality loads will also be reported for dissolved ammonia, dissolved silica, and dissolved organic carbon. Text was added on page 6 to indicate that loads and trends are computed using the Weighted Regressions on Time, Discharge, and Season (WRTDS) and Weighted Regressions on Time, Discharge and Season with Kalman Filtering (WRTDS-K) methods and to update references to these methods to Lee and others (2019) and Zhang and Hirsch (2019). Text was added on page 8 to update the reference to Robertson and others (2018). Text was added on page 11 to clarify that loads are reported in English tons and that daily nitrate sensor values are reported in milligrams per liter, and to clarify that periods of record without sensor data are computed using LOADEST procedures described previously. Text was added on page 12 to indicate that loads and trends are computed using the Weighted Regressions on Time, Discharge, and Season (WRTDS) and Weighted Regressions on Time, Discharge and Season with Kalman Filtering (WRTDS-K) methods. Text was also added to that summarizes findings by Lee and others (2019) and to describe the difference among the two methods. Text was added on page 13 to describe the flow-normalization procedures used within WRTDS and to point readers to previously published methods described by Hirsch and DeCicco (2018). Text was added on page 17 to describe how trends are computed using WRTDS and to list previously published methods by Oelsner and others (2017) and Hirsch and DeCicco (2018). Text was added on page 18 to indicate that flow-normalized loads and trends computed by WRTDS will also be published at NWQN sites. References were updated to include new reports published by Hirsch and DeCicco (2018), Lee and others (2019), Oelsner and others (2017), Robertson and others (2018), and Zhang and Hirsch (2019). ----------------------------------------------------- Revised and reposted November 17, 2020, version 1.2 The text version was modified. On page 13, text was changed from “The WRTDS and WRTDS–K estimates are only published for water years with at least 4 observations in the respective year.” To “WRTDS and WRTDS–K estimates are computed for years with at least four observations in the respective year and for years with less than four observations that are bounded by years with at least four observations within 2 years before and after the respective year.” On page 13, the callout to table 2 was changed to table 3. On page 18, the sentence “The current iteration of NWQN data was published by Deacon and others (2017). Although the release of data from the most recent water year.” was removed. On page 18, the text “Two websites, the USGS Water-Quality Tracking website (Deacon and others, 2015) and the Nutrient Loading for the Mississippi River Basin and Subbasins website (Lee and others, 2017), provide graphics that aid in the interpretation of water-quality concentration and load data at NWQN sites. The USGS Water-Quality Tracking website provides information on multiple water-quality data types collected in the NWQN with a current (2019) focus on nutrients, sediment, and pesticides. The Nutrient Loading for the Mississippi River Basin and Subbasins website is focused on nutrient loading in the Mississippi River Basin, specifically with respect to loads contributed to the Gulf of Mexico. Graphics indicating flow-normalized loads, measures of water-quality trends, and preliminary nutrient loads to the Gulf of Mexico from October to May of the current water year are published on this website (as opposed to through ScienceBase) in early June of each year.” was changed to “The USGS Water Quality Tracking website (https://nrtwq.usgs.gov/nwqn; Lee and Henderson, 2020) provides maps and graphics to aid in the interpretation of concentrations, loads, and trends at NWQN sites, as well as links to the most recent ScienceBase release.” On page 19, the reference "Deacon, J.R., Lee, C.J., Toccalino, P.L., Warren, M.P., Baker, N.T., Crawford, C.G., Gilliom, R.G., and Woodside, M.D., 2015, Tracking water-quality of the Nation’s rivers and streams, U.S. Geological Survey data release, accessed September, 4, 2017, at https://doi.org/10.5066/F70G3H51." was removed. On page 19, the reference "Lee, C.J., Henderson, R.J., and Deacon, J.D., 2017, Nutrient loading for the Mississippi River Basin and subbasins: U.S. Geological Survey web page, accessed September 4, 2017, at https://nrtwq.cr.usgs.gov/mississippi_loads/." was changed to “Lee, C.J., and Henderson, R.J., 2020, Tracking water quality in U.S. streams and rivers—USGS National Water Quality Network data, water-quality loads, and trends: U.S. Geological Survey web page, accessed October 1, 2020, at https://nrtwq.usgs.gov/nwqn.” ----------------------------------------------------- Revised and reposted August 26, 2021, version 1.3 The text version was modified. On p. 13, WRTD was changed to WRTDS in the first paragraph. On p. 13, the sentence "The WRTDS and WRTDS–K models are computed with default values including 7-year half-window widths and a minimum of 100 total observations with at least 50 uncensored observations." was changed to "The WRTDS and WRTDS–K models are computed with 7-year half-window widths and a minimum of 50 total observations with at least 25 uncensored observations." Also on p. 13, the following two paragraphs were added: "As noted on the “Tracking Water Quality in U.S. Streams and Rivers” website (U.S. Geological Survey, 2020), estimates of trends in flow-normalized, water-quality concentrations and loads are provided among the most recent water year with available data (2020 for example), and consecutive decades before the most recent water year (that is, 1980, 1990, 2000, and 2010) are provided as a table underneath graphs when adequate data are available. Tables indicate the percentage change in flow-normalized concentrations and loads for the selected water-quality constituent among the water years selected. Tables also indicate a 90-percent confidence interval about the trend, the part of the total trend attributed to trends in streamflow conditions, and the part of the total trend attributed to other changes in the upstream drainage basin. The magnitude of the trend attributed to other changes in the upstream watershed sources is the stationary flow-normalized estimate (Choquette and others, 2019; Murphy and Sprague, 2019), which assumes that the distribution of streamflow is constant over a given trend period (a windowSide argument of 0; Hirsch and DeCicco, 2018). The magnitude of the trend attributed to trends in streamflow conditions is computed by subtracting the generalized flow-normalized estimate, which incorporates both trends in streamflow and upstream constituent sources (Murphy and Sprague, 2019), from the stationary flow-normalized estimate. Water-quality constituents and (or) trend-related information are only provided when the underlying sample data met completeness thresholds for WRTDS nutrient trends based on procedures defined by Oelsner and others (2017). WRTDS trends are only published among years in which (1) quarterly data were available during the first 2 years and last 2 years of the trend period, (2) where at least 70 percent of years during the trend period had quarterly samples, and when (3) at least 14 percent of samples were categorized as high-flow samples for at least half the decades during the trend period, and at least 10 percent of samples were categorized as high-flow samples during the remaining half of decades in the trend period. However, unlike the criteria defined in Oelsner and others (2017), high-flow samples are defined as any sample in the 80th percentile (Oelsner and others [2017] used the 85th percentile) of flows for a given month and trend period for large coastal and inland river sites. This slightly reduced threshold for high-flow samples is consistent with findings in Oelsner and others (2017) that indicated lower high-flow thresholds were suitable for larger streams (page 72). See pages 38–39 in Oelsner and others (2017) for more details on data adequacy thresholds." Three references were added to the "References Cited" section: Choquette, A.F., Hirsch, R.M., Murphy, J.C., Johnson, L.T., and Confesor, R.B., Jr., 2019, Tracking changes in nutrient delivery to western Lake Erie—Approaches to compensate for variability and trends in streamflow: Journal of Great Lakes Research, v. 45, no. 1, p. 21–39. [Also available at https://doi.org/10.1016/j.jglr.2018.11.012.] Murphy, J., and Sprague, L., 2019, Water-quality trends in US rivers—Exploring effects from streamflow trends and changes in watershed management: Science of the Total Environment, v. 656, p. 645–658. [Also available at https://doi.org/10.1016/j.scitotenv.2018.11.255.] U.S. Geological Survey, 2020, Tracking water quality in U.S. streams and rivers: U.S. Geological Survey web page, accessed August 21, 2020, at https://nrtwq.usgs.gov/nwqn/#/.