Sources, Instream Transport, and Trends of Nitrogen, Phosphorus, and Sediment in the Lower Tennessee River Basin, 1980-96

Appendix C. Methods for quantifying inputs of nitrogen and phosphorus from point and nonpoint sources

Wastewater Discharge

Nitrogen and phosphorus inputs from wastewater discharge were calculated from effluent monitoring data reported to State agencies by permitted wastewater dischargers in the LTEN River Basin. State agencies provided discharger-reported monitoring data (effluent-quality sampling data and effluent flow-measurement data for 1992 or 1995) and Standard Industrial Codes (SIC) for 729 permitted wastewater dischargers (J. Hughes, TDEC, written commun., 1998; M. Rief and T. Cleveland, ADEM, written commun., 1998; V. Prather, KDEP, written commun., 1998; G. Odom, MDEQ, written commun., 1998). Annual mean concentrations of total nitrogen and total phosphorus were estimated from self-reported concentrations or, where complete data were not available, were estimated using one of the following methods.

Where ammonia-nitrogen (NH3-N) concentration data were reported but total nitrogen (TN) data were lacking, a regression equation was used to calculate TN from NH3-N. The regression equation was developed from more than 800 observations of effluent concentrations from municipal wastewater treatment plants in Virginia and North Carolina, and thus applies only to municipal wastewater. This equation took the form:

TN = 11.97 + 0.55 (NH3-N)          (2)

where concentrations are given in milligrams per liter, as nitrogen (McMahon and Lloyd, 1995, p. 70-71).

In the absence of ammonia-nitrogen and total nitrogen concentration data, the average value of 15 mg/L, as nitrogen, was assumed for total nitrogen concentration of municipal wastewater effluent. In the absence of phosphorus data, a concentration of 3.5 mg/L, as phosphorus, was assumed for total phosphorus concentration of municipal wastewater effluent (S. Fishel, TDEC, oral commun., 1998). Values from literature were used for industrial wastewater when data were lacking. National Oceanic and Atmospheric Administration (1993) provides tables with average wastewater effluent concentrations of total nitrogen and total phosphorus based on the type of industry and the SIC of the facility.

Nutrient loads were estimated as the product of effluent annual mean concentration (estimated or measured) and effluent annual mean discharge (estimated or measured). Effluent discharge data were obtained from self-reported information from 264 of the 729 dischargers in the LTEN River Basin (representing the 264 sites for which effluent discharge data were available in digital format). At these sites, the annual mean discharge for calendar year 1992 (or 1995 in some cases) was calculated from daily, monthly, or semi-annual effluent discharge measurements. Of the 264 dischargers with digital discharge data, 64 were classified as major dischargers [that is, they discharged more than 1 million gallons per day (Mgal/d) each, or were industrial facilities with specific process wastewater of concern]. These 64 major dischargers contributed 13,000 Mgal/d (90 percent) of the total wastewater discharge, 9,000 tons/yr (68 percent) of the nitrogen load, and 640 tons/yr (83 percent) of the phosphorus load from all 264 dischargers.

Most of the remaining unestimated discharges (465 sites, fig. 4) are small domestic and commercial dischargers, such as trailer parks and schools, discharging less than 0.1 Mgal/d each. The contributions of discharge and nutrient load from these dischargers should be negligible in comparison with the contributions from the 264 estimated dischargers (S. Fishel, TDEC, written commun., 1998).

Atmospheric Deposition

Deposition data were obtained from the National Atmospheric Deposition Program/National Trends Network (NADP/NTN), a national system of precipitation chemistry monitoring stations operated cooperatively by State agricultural experiment stations, USGS, U.S. Department of Agriculture, and numerous other governmental and private entities. Data from several NADP/NTN monitoring stations in proximity to the LTEN River Basin were selected to calculate atmospheric deposition calculations: these included data from Dixon Springs Agricultural Center, Illinois (IL63); Land between the Lakes, Kentucky (KY38); Walker Branch Watershed, Tennessee (TN00); Hatchie National Wildlife Refuge, Tennessee (TN14); and Wilburn Chapel, Tennessee (TN98). Deposition data were retrieved for each of these NADP/NTN monitoring stations for 1992 to coincide with the most recent data for other sources.

Atmospheric deposition of total nitrogen was calculated as the sum of nitrate wet and dry deposition and ammonia wet deposition. Nitrate dry deposition rates were calculated from nitrate wet deposition rates, by multiplying the wet values by a dry/wet deposition ratio determined by Sisterson (1990). Because high-elevation (> 610 meters) terrain and urban areas were of very limited areal extent in the LTEN River Basin, nitrate droplet deposition, nitrate urban wet deposition, and nitrate urban dry deposition were assumed to be negligible and were not included in the analysis. Organic-nitrogen deposition was not monitored at the NADP/NTN stations and therefore was not included in this analysis; however, monitoring studies in other parts of the Nation (for example, Harned, 1995) indicate that this may be a significant component of wet deposition of nitrogen.

Wet deposition rates for the selected NADP/NTN monitoring stations ranged from 0.33 to 0.68 kilogram per hectare (kg/ha) for ammonia, and from 0.53 to 0.73 kg/ha for nitrate. This variability among stations preempted selection of a single basin-wide average deposition rate for the LTEN River Basin. Instead, the total nitrogen deposition rates calculated for each NADP/NTN monitoring station (in tons per square mile per year) were weighted to each of the tributary basins and major hydrologic units, based on the station's distance to the centroid of each basin and major hydrologic unit.

Fertilizer Application

Estimates of fertilizer inputs to agricultural lands in the LTEN River Basin were calculated using two sources of information: fertilizer sales data and fertilizer application recommendations. County-level estimates of the amount of nitrogen fertilizer sold were computed by Jerald Fletcher (West Virginia University, written commun., 1992), by disaggregating state-level estimates of the amount of fertilizer sold in 1991 (obtained from the National Fertilizer and Environmental Research Center of TVA) to county-level estimates. Disaggregation was done by multiplying the state-level estimates by a ratio of county-to-state expenditures based on the 1987 Census of Agriculture (U.S. Department of Commerce, 1989).

Fertilizer application recommendations were selected after comparing published recommendations for Tennessee (Savoy and Joines, 1998) and Alabama (Adams and others, 1994) with the recommendations made by county agricultural extension agents. Published fertilizer application recommendations are based on crops and soil-test results, which are used to determine the availability of common agricultural nutrients, including nitrogen, phosphorus, and potassium. Agricultural extension agents from a number of counties across the LTEN River Basin were interviewed to determine typical soil ratings for the basin. Because of the great variability in soil-test results within individual counties, agents suggested using a medium soil rating to arrive at a basin-wide average fertilizer application recommendation for each crop.

Application recommendations for selected crops (corn, wheat, tobacco, soybeans, cotton, and hay) were multiplied by 1992 county-level data on harvested acreage (U.S. Department of Commerce, 1994) to derive an estimate of applied fertilizer for nitrogen and phosphorus for each crop in each county for 1992. The estimates were then summed by crop to provide a single fertilizer application estimate for each county.

County-level input estimates from both methods were weighted to provide estimates for selected basins. The weighting algorithm apportions the county input to each basin based on the portion of cultivated land, by county, encompassed within each basin. Estimates of nitrogen and phosphorus fertilizer inputs were also calculated for each major hydrologic unit using the same land-use weighting approach. This weighting approach may give inaccurate results in areas where cropping practices vary greatly across the agricultural land within a county, but the error introduced in this step is not significant for larger basins (those that include large parts of one or more counties).

A comparison of basin-level input estimates from the two computation methods (table C1) reveals significant differences in results for some basins. Sales-based nitrogen inputs were consistently higher than crop application recommendations (ranging from 6 to 49 percent higher). Sales-based estimates for phosphorus differed from crop application recommendations by as much as 46 percent, but were not consistently higher or lower across all the basins. Several factors may contribute to these differences:

  1. Estimates based on application recommendation are calculated for fertilizer applied to harvested acres only, not to total agricultural lands. The areas for which nitrogen sales-based estimates are almost 50 percent higher than application recommendation-based estimates may correspond to areas where farmers apply commercial nitrogen fertilizer to pasture land.
  2. Application recommendation-based estimates are calculated based on medium soil-test results, but soils in some parts of the LTEN River Basin do not match that result. Application recommendation-based estimates for phosphorus may exceed sales-based estimates for soils that have high natural phosphorus content, such as in the basins for sites 5 and 6 (Duck River at Williamsport and above Hurricane Mills).
  3. The use of manure to fertilize agricultural areas is not taken into account in the sales-based estimates. This may account for smaller estimates of phosphorus based on sales for areas where manure is commonly applied.
  4. Estimates based on fertilizer sales do not account for the possibility that fertilizer may be purchased in one county but applied in another county.

Sales-based input estimates were generally viewed as the more reliable of the two estimates and were used in comparison among sources for the selected basins and for major hydrologic units.

Nitrogen Fixation

Estimates of nitrogen input were adopted from literature values for nitrogen fixation and county-level estimates of 1992 harvested acreage for soybeans (the only legume with significant acreage in the LTEN River Basin). The rate of nitrogen fixation by soybeans was based on comparison of rates reported in Tennessee, Alabama, Kentucky, and North Carolina agricultural literature. Reported rates varied little geographically. A rate of 105 lb/acre was applied throughout the study area (Craig and Kuenzler, 1983). The rate was multiplied by 1992 harvested acres for soybeans (U.S. Department of Commerce, 1994) to estimate the amount of biologically fixed nitrogen, in tons. Estimates were then converted from county to basin level using the same land-use weighting algorithm described for fertilizer application.

Crop Uptake

The mass of nutrients incorporated into crop biomass was calculated using literature values of rates of nutrient uptake and 1992 county-level data of harvested amounts. Nutrient uptake rates varied among literature sources (McMahon and Lloyd, 1995; Mitchell, 1998; Savoy, 1999), but because most of the LTEN River Basin is located within Tennessee, rates reported by the University of Tennessee Agricultural Extension Service (Savoy, 1999) were used for this analysis. To derive the mass of nutrient removal for each harvested crop, the nutrient uptake rate (pounds per acre for the harvested amount per acre) was multiplied by the harvested amount (in pounds, bushels, or tons) (U.S. Department of Commerce, 1994). Estimates of nutrient removal in tons per year were totaled by county for all crops, then converted from county level to basin level by using the same land-use weighting algorithm described for fertilizer application.

Livestock Waste

County-level estimates of the mass of nutrients from livestock waste have been compiled for all livestock categories, including cattle, hogs, and chickens (U.S. Department of Commerce, 1994). These estimates were calculated for each livestock category from county-level animal census data and from estimates of the nutrient content of daily wastes (Barth and others, 1992), by using the following equation:

[nutrient mass in livestock waste, in pounds per year] = [census estimate of number of animals] x [average weight of animal per 1,000 pounds] x [nutrient content of waste, in pounds per day per 1,000 pounds animal] x [365 days per year]           (3)

Nutrient mass estimates for 1992 were converted from county level to basin level using a land-use weighting algorithm similar to the one used for weighting fertilizer application, except that the weighting was based on the distribution of pasture land.

Estimates of nutrient input from livestock waste are associated with two separate double-counting problems in a mass-balance analysis of sources and sinks. First, if the livestock producing the waste are fed fertilized crops grown within the watershed, the nutrient input is double-counted as both applied fertilizer and produced waste. Second, the ammonia volatilized from manure contributes to atmospheric nitrogen, so that nutrient input may be double-counted as both produced waste and measured atmospheric deposition. These discrepancies cannot be accounted for with the available data.

Next Title Page Table of Contents List of Illustrations List of Tables Conversion Factors and Vertical DatumGlossary

Last modified: Thu Mar 30 13:21:47 CST 2000

Tennessee home page