Nutrient Loss Reduction Strategy Research Library

Introduction

Illinois is a highly agricultural state but with several major metropolitan areas. There are more than 22 million acres of corn and soybeans (60 percent of the state’s land area), much of it tile drained, and a population of nearly 13 million people (fifth nationally). Consequently, both point and non-point sources of nitrogen and phosphorus are added to the streams and rivers of the state, with these nutrients being transported to the Mississippi River and the Gulf (David and Gentry, 2000; David et al., 2010; Jacobson et al., 2011). The Mississippi River/Gulf of Mexico Watershed Nutrient Task Force has a goal to reduce the hypoxic zone in the northern Gulf to a five-year running average of 5,000 km² (approximately 1,900 sq. mi) by 2015 (Mississippi River/Gulf of Mexico Watershed Nutrient Task Force, 2008). To meet this goal, the U.S. Environmental Protection Agency (U.S. EPA) Science Advisory Board recommended a 45 percent reduction from the 1980-1996 average total nitrogen and phosphorus stream loads in the Mississippi River Basin. Because nitrate-nitrogen is thought to be the primary nutrient leading to formation of the hypoxic zone each summer, with total phosphorus secondary, the focus for reducing total nitrogen is to reduce nitrate-nitrogen loads in the Mississippi River Basin (U.S. EPA, 2007). This report provides the scientific basis for a nutrient reduction strategy for Illinois by: (1) determining the current conditions of nutrient sources in Illinois and the export from both point and non-point sources by rivers in the state, (2) describing practices that could be used to reduce these losses to surface waters and providing estimates for the effectiveness of these practices throughout Illinois, and (3) estimating the costs of the statewide application of these methods to reduce nutrient losses and meet Gulf hypoxia goals.

In this analysis, we used U.S Geological Survey (USGS) stream flow data and nitrogen and phosphorus concentrations from the Illinois Environmental Protection Agency (Illinois EPA) and USGS to estimate major watershed stream loads for the state for the 1980-2011 water years. In addition, we directly estimated nitrogen and phosphorus point source loads, while nitrogen and phosphorus non-point sources were calculated by subtracting point sources from total nitrogen and phosphorus loads. These estimates were compared to values previously published by David and Gentry (2000) to provide perspective from earlier studies in Illinois. Urban non-point sources were estimated using published values and urban land areas in Illinois. We then applied a 45 percent reduction target or goal to the 1980-1996 stream loads of nitrate-nitrogen and total phosphorus for the state to determine the goal of a state nutrient reduction strategy. The next step was estimating nitrate-nitrogen and total phosphorus yields (both point and non-point) for the eight-digit Hydrologic Unit Code (HUC) watersheds in Illinois to connect nutrient yields with listed watersheds (stream segments and lake acres that do not meet water quality criteria for dissolved oxygen, total phosphorus, nitrate-nitrogen, aquatic plants, or aquatic algae) at this scale. This allowed us to rank and determine critical watersheds for nutrient reductions. We then estimated reductions in nutrient loads for various point and non-point practice changes, estimated costs per acre for agricultural practices and nutrient reductions per pound for both point and non-point sources, and scaled our estimates to the entire state. Finally, we developed various scenarios to reduce nitrate-nitrogen and total phosphorus loads by either 20 or 45 percent. Twenty percent was chosen to be roughly half of the 45 percent reduction target.

Current Conditions

Point Source Nutrient Loads

Nitrogen and phosphorus point source data are available through the U.S. EPA Integrated Compliance Information System (ICIS). We began with Illinois EPA’s analysis (Mosher, 2013) of total phosphorus data in ICIS, from which Mosher received information on 1,660 point sources of phosphorus in Illinois for 2009. Mosher (2013) concluded that the ICIS tools did not allow an accurate estimation of point source phosphorus loadings in Illinois. As a result, Illinois EPA used phosphorus data from 42 facilities provided by the Illinois Association of Wastewater Agencies (IAWA), including data provided after Mosher’s report, along with discussions with cooling water dischargers to recalculate phosphorus concentration and loads for the largest 108 dischargers listed in ICIS. For our analysis we added data from Decatur’s publicly owned treatment works (POTW). Illinois EPA found some important errors in the ICIS output and recalculated the top 108 sources in the data from the ICIS output (Mosher, 2013). The 108 sources included the 100 largest phosphorus sources in the state—and therefore most of the point source phosphorus load—in addition to eight sources provided by IAWA (Mosher, 2013). Mosher (2013) used total phosphorus concentrations either from values reported by facilities, Illinois EPA’s knowledge of the facility, or the ICIS database. In our analysis of phosphorus, we examined the other major discharging facilities (hereafter referred to as majors) in the ICIS database, a total of 263 facilities that included the top 108 previously analyzed. Majors are nearly all treatment works with design flows >1 million gallons per day (MGD), but they also include a few treatment works that score >80 points on the National Pollutant Discharge Eliminating System (NPDES) Permit Rating Worksheet. As Illinois EPA had done for the top 108 sources, we used Illinois EPA’s best estimate of the total phosphorus concentration for many of the industrial and agricultural facilities and a few POTWs that had very high total phosphorus concentrations in the ICIS database. For all others, we used the U.S. EPA ICIS value for total phosphorus, which was typically between 2.5 and 3 mg/L. Similar to Mosher (2013), we found that the original ICIS output overestimated the total phosphorus load by a large percentage. The ICIS estimate for the 263 majors was 29.4 million lb total P yr^-1, whereas our estimate was 16.6 million lb yr^-1. The ICIS major point source total phosphorus estimate was, therefore, 1.8 times too high (we believe our estimate is more accurate because we used actual data from dischargers in Illinois instead of the modeled values used by U.S. EPA). Based on this over prediction, we used U.S. EPA estimates multiplied by 0.565 for the other 1,397 total phosphorus point sources in the ICIS database. Because these 1,397 point sources were a small proportion of the overall point source phosphorus estimate and no other data were available for the wide range of sources in the data set, this was the best estimate we could make. A median of total phosphorus from the POTWs would not be appropriate to use for these varied point sources.

There are fewer measurements available for nitrogen because many facilities have been monitoring only ammonia concentrations. We made a request through IAWA for nitrogen data for the 2008-2012 period and received data from 34 major facilities (requests went out by email to all IAWA members on March 5, 2013, with a reminder on April 8, 2013 from Robin Ellison of IAWA). Three of the facilities only reported ammonia, but 31 reported total nitrogen or nitrate-nitrogen, with most reporting both. Some had five years of data, some only one. All reported flow. Facility size ranged from 1.3 to 712 MGD, with a median of 12 MGD. For typical plants (large Chicago plants excluded), the average total nitrogen concentration was 16.8 mg/L, with a nitrate-nitrogen concentration of 14.9 mg/L. This average is based on data from 26 facilities, mostly for 2008-2012. Illinois EPA made a request to ICIS for all nitrogen data, and 392 sources were reported, all POTWs. These are the only point sources in Illinois with a permit for nitrogen, far fewer than the 1,660 permitted for phosphorus. Because ICIS reported only ammonia data for nearly all plants, only flow data could be used from this source, and no ICIS nitrogen concentration data were used in our analysis. Annual loads were directly calculated for the 31 plants that reported nitrate-nitrogen or total nitrogen data. For the other 361 plants in the ICIS database (392 total sources minus the 31 that reported concentrations to us), the flow from ICIS and the average total nitrogen and nitrate-nitrogen concentrations reported above were used to estimate the source. Data for both the phosphorus and nitrogen estimates were available for all seven Metropolitan Water Reclamation District of Greater Chicago (MWRDGC) plants, which was important given that MWRDGC operates the largest plants in the state. 

Table 3.1. Point source total phosphorus loads for the entire state and by major river basins. The category “all other basins” includes point sources outside the eight major basins.

Table 3.2. Point source total nitrogen and nitrate-nitrogen loads for the entire state and by major river basin. The category “all other basins” includes point sources outside the eight major basins.

Point source total phosphorus was estimated at 18 million lb yr^-1, with most from the major facilities in the state (16.6 million lb yr^-1) (Table 3.1). The estimated point source load of total nitrogen was 87.3 million lb yr^-1, with 75.2 million lb as nitrate-nitrogen (Table 3.2). Most of the point source nitrogen is from northern Illinois, with large loads in the Illinois and Rock rivers.

A comparison was made with previously published point source estimates for Illinois to see if our prior understanding of the importance of these nutrient sources was correct. Using completely different estimation techniques (per capita nitrogen in effluent) for the 1990s, David and Gentry (2000) reported a very similar total nitrogen estimate of 86 million lb yr^-1 from point sources. More recently, David et al. (2010) estimated 123 million lb N yr^-1 consumed in food, which is expected to be larger than the nitrogen discharged due to gaseous nitrogen losses during wastewater treatment and the nitrogen removed in biosolids. The estimate of point source phosphorus loads is larger than predicted by David and Gentry (2000) or Jacobson et al (2011) based on food consumption (13 million lb yr^-1). This is likely due to the inclusion of industrial point sources in the current study that were not considered in the earlier work. 

Urban Runoff Nutrient Loads

Urban runoff was estimated using Illinois land cover data and published tables of nutrient loss per acre. We used two sets of land cover maps. The first was Illinois Land Cover: An Atlas (1996), a 1991-1995 analysis by the Illinois Department of Natural Resources (IDNR) using Landsat 4 and 5 Thematic mapper satellite imagery acquired during the 1991-1995 spring and fall seasons, with most of the data from 1992. This dataset has six urban land uses: high density, medium/high density, medium density, low density, transportation, and urban grassland. The newer 1999-2000 land cover data was a joint effort by the U.S. Department of Agriculture National Agricultural Statistics Service, the Illinois Department of Agriculture, and IDNR using Landsat 5 TM and Landsat 7 RTM+ satellite imagery acquired during the spring, summer, and fall seasons of 1999 and 2000. However, these newer data only divided urban areas into high-density, medium/low-density, and urban open space categories.

Nitrogen and phosphorus yields for urban areas were obtained from the report Preliminary Data Summary of Urban Storm Water Best Management Practices (U.S. EPA, 1999). Table 4-3 was used from the 2015 Illinois NLRS Strategy Report, but is not shown here. These estimates were derived from several different studies of typical urban area nutrient yields originally from Horner et al. (1994). We then multiplied the published estimates for nutrient loads per acre by the actual acres of each land cover type. For the 1999-2000 land cover, we used nutrient yield averages from different land cover classes listed in Table 4-3 to match the three categories of land cover data available. We also used data for nitrate-nitrogen and total nitrogen urban runoff loads from a study conducted in Baltimore (Groffman et al., 2004), total nitrogen from a study in Seattle (Herrera Environmental Consultants, 2011), total phosphorus from estimated inputs to the DuPage River (DuPage River Salt Creek Workgroup, 2008), and total phosphorus in urban runoff from an Illinois EPA summary (Illinois EPA, 1986). Each of these data sources were combined with the land cover data described above.

Land cover data indicated that there are about 2.3 million acres of urban land in Illinois. We estimate that urban runoff is a source of about 1.5 million lb total P yr^-1, 6 million lb nitrate-N yr^-1, and 8.3 million lb total N yr^-1. These are approximate values given the approach used but are likely around the right order of magnitude. There was little difference in the estimates using the two land cover databases. 

Riverine Nutrient Loads

We used stream flow and nitrogen and phosphorus concentrations for the eight major rivers in the state with available data, which represents 74 percent of the state area (Table 3.3 and Figure 3.1). USGS flow data and Illinois EPA and USGS data were used to calculate annual fluxes during 1980-2011 for nitrate-nitrogen, total nitrogen, dissolved reactive phosphorus (DRP), and total phosphorus. The results were extrapolated to represent the state (56,371 sq. mi). This generally follows methods used by David and Gentry (2000). For the Rock River, 54 percent of the drainage at Joslin, where the gage is located, is in Wisconsin. David and Gentry (2000) estimated the Illinois load as 46 percent of the load at Joslin, but we used a different method. We calculated the load for the Rock River at Rockton, Illinois, which is mostly drainage from Wisconsin. We then subtracted the Rockton load from that at Joslin, giving us the load from Illinois sources (3,187 sq. mi) only.

Table 3.3. River systems, location and station number of discharge and water quality data, drainage area, and fraction of drainage area in Illinois used in estimating export of nitrogen and phosphorus by surface water from Illinois.

A variety of methods can be used to determine the annual load for a river using continuous daily flow and infrequent nutrient concentration measurements. There has been much discussion in the literature about the advantages and disadvantages of each method. Based on our assessment of the literature and current techniques available, interpolation is thought to be the best method for highly soluble nutrients such as nitrate-nitrogen in larger rivers. And because nitrate-nitrogen is a large percentage of total nitrogen, interpolation can be used for total nitrogen as well. However, for phosphorus in smaller rivers, there is a strong concentration response to flow, and high flow loads can be underestimated with interpolation when sampling is infrequent. The USGS Weighted Regressions on Time, Discharge, and Season (WRTDS) technique (Hirsch et al., 2010) fits a relationship that includes flow and, therefore, better estimates the high flow days that are critical to estimating phosphorus loads (Royer et al., 2006).

Delete

Edit embedded media in the Files Tab and re-insert as needed.

align image leftalign image centeralign image right

We conducted linear interpolation to estimate daily nutrient concentrations between sampling days using SAS version 9.2 and the Proc Expand procedure. Daily flow and measured nutrient concentrations were the input data, with daily flow and daily concentration the output. With this procedure, the observed values are present in the final data set as they are not replaced with estimated values.

The WRDTS load estimates were calculated using software developed and provided by USGS (available at github.com/ USGS-CIDA/WRTDS/wiki). WRDTS estimates are based on regressions with discharge, time, and seasonality. The user can specify the relative weightings for each of these factors by changing the value of three variables: windowY for time, windowQ for discharge, and WindowS for seasonality. The model developers recommend default values of 10, 2, and 0.5, respectively, for these parameters. Daily load estimates produced by WRDTS with the default weightings were compared to the observed loads on the days when sample concentrations were measured by two different methods. First, a linear regression was conducted between observed and model-estimated loads, with the intercept set to zero. If the slope of the regression line deviated substantially from 1, or if the coefficient of determination was less than 0.8, alternative values for weightings were considered. Secondly, the WRDTS software calculates a flux bias statistic, which estimates the average deviation of the model load from the measured loads. If the flux bias statistic indicated a bias of 10 percent or greater, we used WRDTS with variables appropriate for the model to estimate loads. The weighting values that produced load estimates with the lowest flux bias statistic and the greatest correspondence between observed and model-estimated daily loads were considered the best estimates.

This analysis was informed by communication with USGS model developers. For the Illinois River, the seasonality parameter was reduced to 0.25. Sprague et al. (2011) conducted and published an analysis of Illinois River nitrate-nitrogen loads and found a seasonality weighting of 0.25 was appropriate. The default weightings produced an unusually large flux bias on the Embarras River, and Hirsch recommended reducing the discharge weighting from 2 to 1 in personal communication. This substantially reduced the flux bias and improved the correspondence between the estimated and observed loads. This weighting was also found to reduce the bias and improve the correspondence for both DRP and total phosphorus loads in the Kaskaskia and Vermilion rivers and for estimating DRP loads for the Rock River at Joslin and the Green River.

For all eight rivers, we calculated and compared annual loads for 1980-2011 using both interpolation and WRDTS for nitrate-nitrogen, DRP, and total phosphorus and using interpolation alone for total nitrogen. For nitrate-nitrogen, interpolation and WRDTS gave results that differed by less than 10 percent for most rivers. The Embarras River was an exception for which WRDTS produced estimates that were 15 to 20 percent larger than interpolation. For DRP and total phosphorus, both methods gave similar loads for the larger rivers, such as the Illinois. For smaller rivers, WRDTS gave larger loads in years with higher flows. This comparison supported our use of interpolation for nitrate-nitrogen and total nitrogen and WRDTS for DRP and total phosphorus. 

There was a small (1 percent) increase in water flow from the early to later period, with small increases in nitrate-nitrogen (1.4 percent) and total nitrogen (1.7 percent) loads. These changes are within the errors of our estimation methods and suggest little change with time. However, total phosphorus increased by 9.3 percent, with most of this increase in DRP (20.1 percent). The David and Gentry (2000) estimate for the total phosphorus load during 1980-1997 was likely lower due to interpolation being used to estimate loads. Total nitrogen and water loads were similar to what David and Gentry (2000) estimated. Point sources were 18.4 percent of the nitrate-nitrogen loads for 1997-2011, 16.3 percent of total nitrogen, and 48 percent of total phosphorus, which is nearly identical to previous estimates by David and Gentry (2000) for an earlier time period. Nutrient sources that contribute to the riverine load for the state are shown as a percent of the total in Figure 3.2.

The increase in water flow was due to unusually high flows compared to the long-term average during 2008-2011, which followed relatively lower flows during 1997-2007 (Figure 3.3). Linear regression indicated no significant trend in annual flow for the 1980-2011 period. Figure 3.3 includes a Locally Weighted Scatterplot Smooth (LOESS) curve calculated using SAS Ver. 9.2 that can be used to describe the relationship between Y and X without assuming linearity or normality of residuals and is a robust description of the data pattern (Helsel and Hirsch, 2002). Annual nitrate-nitrogen and total phosphorus loads had different temporal patterns, with nitrate-nitrogen having no trend through time but total phosphorus increasing (Figure 3.4). A linear regression of annual total phosphorus loads with annual water flux and year had an R² of 0.97, with both water and year significant at the p <0.0001 level. Annual loads of DRP had a similar result, with an R² of 0.96. Therefore, the increase in annual phosphorus flux appears to be related to water flux and possibly factors such as changing point source inputs or agricultural practices (e.g., fertilizer form, placement, and timing, manure practices, and tillage changes), although these were not evaluated.

For annual loads of nitrate-nitrogen, the Illinois, Embarras, Little Wabash, Big Muddy, and Vermilion all declined between 1980-1996 and 1997-2011, whereas the Rock, Green, and Kaskaskia increased, as did the state load (Figure 3.5). The greatest change was for the Rock River, where the load increased 66 percent between these two periods, while flow increased 12 percent. Because we estimated the load for the Rock by subtracting the station at Rockton from the load at Joslin, the resulting load is representative of Illinois only, and the increase in annual nitrate-nitrogen loads was a result of greater losses from Illinois. For total phosphorus, all rivers except the Green, Vermilion, and Embarras increased, leading to an overall 10 percent increase for the state. This analysis of major rivers indicates that the increase and decrease in nitrate-nitrogen riverine loads led to no change in the overall state export, but there were differences through time within Illinois watersheds.

We compared Illinois loads to overall Mississippi River Basin loads available from the USGS. In 1997- 2011, Illinois contributed about 20 percent of the nitrate-nitrogen load, 11 percent of the total phosphorus load, and 7 percent of the water flow to the Gulf.

Table 3.4 shows the average annual riverine water estimated for the entire state based on the eight major rivers as well as nitrogen and phosphorus loads for the two periods of interest: 1980-1996 and 1997-2011.

Delete

Edit embedded media in the Files Tab and re-insert as needed.

align image leftalign image centeralign image right

Delete

Edit embedded media in the Files Tab and re-insert as needed.

align image leftalign image centeralign image right

Delete

Edit embedded media in the Files Tab and re-insert as needed.

align image leftalign image centeralign image right

Delete

Edit embedded media in the Files Tab and re-insert as needed.

align image leftalign image centeralign image right

DRP was about half of the total phosphorus, but it has increased as a percent over the past 10 years (Figure 3.6). It was consistently about 45 percent of total phosphorus in the 1980s and 90s but has been greater and more variable since. Declines in particulate phosphorus loads are likely related to reduced erosion from the adoption of conservation tillage and possibly increased tile drainage, whereas increases in DRP could be due to the reduced incorporation of phosphorus fertilizers (and more intense winter and spring storms), increased population, and increased tile drainage. 

Delete

Edit embedded media in the Files Tab and re-insert as needed.

align image leftalign image centeralign image right

Riverine Nutrient Yields

Riverine nutrient loads are influenced by the size of a watershed. Larger watersheds typically produce larger flows and nutrient loads. Another way to examine nutrient losses from a watershed, and to compare watersheds, is to divide the nutrient load at the outlet by the area of the watershed to determine the yield. This allows watersheds of differing sizes to be compared by their nutrient loss per unit of land. Yields of total nitrogen and nitrate-nitrogen varied greatly across the state, with the tile-drained watersheds having much larger yields than the non-tiled, southern Illinois watersheds (Figure 3.7). In addition, some of the watersheds in southern Illinois are not as intensely agricultural. The state average nitrate-nitrogen yield was 11.3 lb/acre/yr averaged for the 1997-2011 period, but this varied from 1.4 (Big Muddy) to 23 (Vermilion) lb/acre/yr. Total phosphorus yields were less variable and averaged 1.1 lb/acre/yr, with a range of 0.55 to 1.18 lb/acre/yr. When yields of nitrate-nitrogen and total phosphorus are viewed by source, the importance of point sources in the Illinois River and non-point sources in the other rivers is clearly shown (Figure 3.8)

Delete

Edit embedded media in the Files Tab and re-insert as needed.

align image leftalign image centeralign image right

Delete

Edit embedded media in the Files Tab and re-insert as needed.

align image leftalign image centeralign image right

Riverine Nutrient Load Goal or Target

To meet a 45 percent reduction of the 1980-1996 average riverine loads of nitrate-nitrogen and total phosphorus, the nitrate-nitrogen load target is 222 million lb yr^-1, and total phosphorus is 18.7 million lb yr-1. Given that the 1997-2011 loads were greater than in 1980-1996, this would require a 46 percent reduction from those loads for nitrate-nitrogen and 50 percent for total phosphorus. Figure 3.9 shows loads by river and for the state and the reduction goal. To meet the nitrate-nitrogen target, the focus must be on agricultural sources, mostly in northern and central Illinois. Reductions in point sources could meet a large part of the total phosphorus target, but additional reductions from agriculture throughout the state will prob- ably be needed. As Figure 3.10 shows, the target for nitrate-nitrogen has only been met during low-flow years. Additionally, the total phosphorus target was only met during the 1988 drought, although other dry years came close. Consistently meeting the target will take major reductions from all sources. 

Delete

Edit embedded media in the Files Tab and re-insert as needed.

align image leftalign image centeralign image right

Delete

Edit embedded media in the Files Tab and re-insert as needed.

align image leftalign image centeralign image right