Nutrient Loadings and Trends in the Jordan Lake Watershed

The NC Division of Water Resources developed the B. Everett Jordan Reservoir Water Supply Nutrient Strategy in 2009 which includes a comprehensive set of rules designed to reduce nutrient over-enrichment in Jordan Lake and restore it to full use. For this report, available nutrient data from 1990-2020 from selected stations was analyzed to evaluate progress in achieving nutrient reduction goals relative the baseline period of 1997-2001. The selected stations include ambient and coalition monitoring stations from North Buffalo, South Buffalo, Reedy Fork, New Hope, Northeast, and Morgan Creeks, and the Haw River that have a co-located or nearby USGS gage station with long-term flow data (Figure below).

Nutrient loads were calculated using two methods; the first method uses the USGS LOADEST method. LOADEST creates a regression model based on stream flow, concentration, and time to develop mean load estimates with 95 percent confidence intervals on a monthly basis (Runkel et al. 2015). This method is best used to understand the amount of nutrients being pushed to the lake which can vary greatly with streamflow.

The second method calculates flow-normalized (FN) nutrient load and concentrations using the USGS Weighted Regressions on Time, Discharge, and Season (WRTDS) (Hirsch and De Cicco, 2015; Hirsch et al., 2010). This method is more useful to observe nutrient reduction progress (or lack thereof) by reducing effects of year-to-year variability in discharge on the record of trend in water quality.

The WRTDS method is a recently developed exploratory data analysis approach that provides insight about the characteristics of water quality data and can be used to evaluate changes in nutrient concentration and loads (Hirsch and De Cicco, 2015; Hirsch et al., 2010). WRTDS uses probability distributions of daily streamflow to reduce or eliminate the influence of random, year-to-year streamflow variability on estimates of trends in concentration and load while preserving the influence of long-term trends in streamflow. This approach also has the ability to identify non-monotonic trend patterns and the ability to differentiate between trends in concentration versus trends in flux (Hirsch et al., 2018a; Choquette et al., 2019). The WRTDS is being used to analyze nitrogen, phosphorus, and suspended-sediment loads and trends for the Chesapeake Bay Network Stations (Moyer and Langland 2020; Chanat et al., 2018a).

Once the Flow-normalized loads and concentrations are estimated for each ambient monitoring station, annual load reductions relative to the 1997 baseline year and five-year average load reductions relative to the 1997-2001 were computed. The recently developed WRTDS Bootstrap Test (WBT) was employed to determine the uncertainty of trend results in flow-normalized load including the 90% confidence intervals for the magnitude of trend, hypothesis tests for trend in flow-normalized concentration and flow-normalized load (α =0.1) and the likelihood that the direction of trend is correct (Choquette et al., 2019, Hirsch et al., 2015). For the selected Jordan Lake watershed stations, the likelihood of an upward or downward trend for the annual (from 1997 to 2020) and multi-year (from 1997-2001 to 2016-2020) loads and concentrations were estimated. Table 2 presents the descriptions of the likelihood designations suggested by Hirsch et al., (2015). Higher values of likelihood indicate a higher confidence that estimates of positive and negative trends are truly positive and truly negative.

The following tables and figures show a summary of the flow-normalized total nitrogen load trends from the baseline year to 2020 period. The baseline year for Reedy Fork (B0400000) is 2002 and 2001 for Northeast creek (B3670000) when the stations were established. The baseline year for the rest of the stations is 1997.

South Buffalo Creek (B0670000), Haw River at NC 49N (B1140000), and Haw River at SR 1713 (B210000) exhibited a likely to highly likely upward trend in annual total nitrogen load compared to the baseline year with increases ranging from 3% to 28%. The annual total nitrogen load for the rest of the stations showed a likely to highly likely downward trend. The annual total nitrogen load reduction for these stations ranged from 6% to 78%. Among the selected sites, North Buffalo Creek (B0540000) near Greensboro had the highest reduction (78%) and Morgan Creek at Mason (B3899180) had the lowest reduction (6%). Haw River at NC 49N had the largest increase (28%).

Flow-normalized total nitrogen concentrations showed similar trends like that of the loads for most stations except for Morgan Creek where a highly likely upward trend is observed and South Buffalo Creek (B0670000) where the concentration trend is likely downward. The annual total nitrogen concentration reductions ranged from 6% to 89% at these stations. The highest reduction (89%) was observed for North Buffalo Creek and the lowest reduction (6%) was observed for Reedy Fork Creek at SR 2719 (B0400000). For those station with decreasing trends, the likelihood of the trend in annual flow-normalized total nitrogen concentration ranged from likely to highly likely downward. The annual flow-normalized total nitrogen concentration increased by 7%, 12%, and 19% during this period for Haw River at NC 49N, Haw River at SR 1713, and Morgan Creek at Mason, respectively. 

In addition to baseline to 2020 changes in FNTN load and concentration, the change from the baseline five-year window to the 2016-2020 window was computed. The baseline years for Reedy Fork (B0400000) are 2002-2006 and 2001-2005 for Northeast creek (B3670000). The baseline year for all other stations is 1997-2001.

South Buffalo Creek (B0670000), Haw River at NC 49N (B1140000), and Haw River at SR 1713 (B210000) showed a likely to highly likely upward trend in the five-year average total nitrogen load compared to the baseline year. For New Hope Creek at SR 1107 (B3040000) and Morgan Creek at Mason, the upward/ downward trend was uncertain. The five-year average total nitrogen load for the rest of the stations exhibited a likely to highly likely downward trend. For those stations where reduction in the five-year average load is observed, the annual flow-normalized total nitrogen loads reductions ranged from 7% to 65%. Among these the New Hope Creek at SR 1107 (B3040000) site show the lowest reduction (5%) and North Buffalo Creek site show the highest reduction (65%) in the five-year average flow-normalized total nitrogen load. The Haw River at NC 49N (B1140000) station had the largest increase (35%) in the five-year average flow-normalized total nitrogen load. The load increased by 5%, 35%, and 16% for South Buffalo Creek (B0670000), Haw River at NC 49N (B1140000), and Haw River at SR 1713 (B210000), respectively.

Flow-normalized total nitrogen concentrations showed similar trends like that of the loads for most stations except for Morgan Creek at Mason WWTP (B3899180) where a highly likely upward trend is observed and South Buffalo Creek (B0670000) where the concentration trend is likely downward. The trend for Reedy Fork Creek at SR 2719 (B0400000) is uncertain. For those sites where decreases in concentrations are observed, the relative reduction ranged from 7% to 79%. The Morgan Creek at Mason WWTP (B3899180) site had the largest increase (18%) in the five-year average flow-normalized total nitrogen concentration. The concentration increased by 11%, 9%, and 18% for Haw River at NC 49N (B1140000), Haw River at SR 1713 (B210000), and Morgan Creek at Mason WWTP (B3899180) respectively.

The following tables and figures show a summary of the flow-normalized total phosphorus load trends from the baseline year to 2020 period. The baseline year for Reedy Fork (B0400000) is 2002 and 2001 for Northeast creek (B3670000). The baseline year for the rest of the stations is 1997.

From the baseline year to 2020, except for the South Buffalo Creek (B0670000), all other tributaries show substantial decreases in total phosphorus load. The annual total phosphorus loads reduction for these stations ranged from 23% to 82%. The likelihood of the trend in the annual flow-normalized total phosphorus load at these sites ranged from likely downward to highly likely downward. The annual total phosphorous load increased by 34% for South Buffalo Creek. The likelihood of the trend for South Buffalo Creek site is likely upward. Among the selected sites, North Buffalo Creek (B0540000) near Greensboro had the highest reduction (82%) and Haw River NC 49N had the lowest reduction (23%).

During the baseline-2020 period, all tributaries except the Northeast Creek at SR1731 show substantial decreases in FN total phosphorus concentrations while the South Buffalo Creek shows a modest decrease. The annual total phosphorus concentrations reductions ranged from 7% to 89% at these stations. The highest reduction (89%) was observed for North Buffalo Creek and the lowest reduction (7%) was observed for South Buffalo Creek. The likelihood of the trend in annual flow-normalized total phosphorus concentration at these sites ranged from likely to highly likely downward except for the South Buffalo Creek site where the likelihood of the trend in annual flow-normalized total phosphorus load for is uncertain (upward/ downward trend as likely as not). Annual flow-normalized total phosphorus concentration increased by 12% during this period for Northeast Creek at SR1731. The likelihood of the trend in annual flow-normalized total phosphorus concentration for the Northeast Creek site is likely upward.

In addition to baseline to 2020 changes in FNTP load and concentration, the change from the baseline five-year window to 2016-2020 window was computed. The baseline years for Reedy Fork (B0400000) are 2002-2006 and 2001-2005 for Northeast creek (B3670000). The baseline window for all other stations is 1997-2001.

The following tables and figures show the summary of the five-year average flow-normalized total phosphorus loads and trends during the baseline to 2016-2020 period. During this period, except for South Buffalo Creek, all other tributaries show substantial decreases in FN total phosphorus load. The annual FN total phosphorus loads reductions ranged from 13% to 77%. The likelihood of the trend in flow-normalized total phosphorus load at these sites ranged from likely downward to highly likely downward. Among the selected sites the Haw River at NC 49N site show the lowest reduction (13%) and North Buffalo Creek site show the highest reduction (77%) for the five-year average flow-normalized total phosphorus load. The likelihood of the trend in flow-normalized total phosphorus load for the South Buffalo Creek site is likely upward.

Flow-normalized total phosphorus concentrations followed similar trends like that of the loads. During this period, except for South Buffalo Creek and Northeast Creek, all other tributaries show substantial decreases in FN total phosphorus concentration. The FN total phosphorus concentration reductions ranged from 6% to 99%. The likelihood of the trend in flow-normalized total phosphorus concentration at these sites ranged from likely to highly likely downward. Among the selected sites the South Buffalo Creek and the Northeast Creek sites show the lowest reduction (6%) and North Buffalo Creek site shows the highest reduction (99%) in the five-year average flow-normalized total phosphorus concentration. The likelihood of the trend in flow-normalized total phosphorus concentration for the South Buffalo Creek site is uncertain (upward/ downward trend as likely as not).

The following dashboard presents loading estimates and relative TN and TP load reduction for the selected stations for this analysis. Select a station on the Station List pane to bring up LOADEST and WRTDS flow-normalized loading information.

The top right panels below show the annual TN and TP load estimated using the LOADEST and flow-normalized load estimated using WRTDS for each individual station. The bottom right panels show annual flow-normalized nutrient load reductions computed to evaluate the relative change from the baseline year and the reduction in the five-year average flow-normalized nutrient load relative to the five-year baseline period average for TN and TP at each individual station. These plots can aid in evaluating the effectiveness of management actions implemented to reduce TN and TP loading to Jordan Lake by comparing estimated loading under long-term average flow conditions. For the five-year averages, loads of nitrate, TKN, TN, and TP were computed by five-year moving averages and compared with the corresponding value for the baseline line reference period (mainly 1997-2001, and 2001-2005 when data is not available). Negative numbers in the plots indicate decreased loading under average flow conditions.

Choquette, A.F., Hirsch, R.M., Murphy, J.C., Johnson, L.T., Confesor, R.B., 2019. Tracking changes in nutrient delivery to western Lake Erie: approaches to compensate for variability and trends in streamflow. J. Great Lakes Res. 45:21–39

Hirsch, R.M., De Cicco, L.A., 2015. User Guide to Exploration and Graphics for RivEr Trends (EGRET) and dataRetrieval: R Packages for Hydrologic Data, Version 2.0, U.S. Geological Survey Techniques Methods, 4-A10. U.S. Geological Survey, Reston, VA (93 pp. (at: doi:10.3133/tm4A10); with updates (2018) EGRET Version 3.0 at: https://cran.r-project.org/web/packages/EGRET/index.html) (revised)).

Hirsch, R.M., Moyer, D.L., Archfield, S.A., 2010. Weighted regressions on time, discharge,and season (WRTDS), with an application to Chesapeake Bay river inputs. J. Am.Water Resour. Assoc. 46, 857–880.

Hirsch, R.M., Archfield, Stacey A., De Cicco, Laura A., 2015. A bootstrap method for estimating uncertainty of water quality trends. Environ. Model. Softw. 73, 148–166. 

Moyer, D.L., and Langland, M.J., 2020, Nitrogen, phosphorus, and suspended-sediment loads and trends measured at the Chesapeake Bay nontidal network stations—Water years 1985–2018, U.S. Geological Survey data release, accessed January 14, 2021, at  https://doi.org/10.5066/P931M7FT .

Runkel, R. L.; Crawford, C.G.; Cohn, T. A. 2004. Load Estimator (LOADEST): A Program for Estimating Constituent Loads in Streams and Rivers. Techniques and Methods Book 4, Chapter A5. USGS