• E-mail
• Print
• Comment
• Font Size
• Digg
• del.icio.us
• Discuss article

Modeling the Distribution of Diffuse Nitrogen Sources and Sinks in the Neuse River Basin of North Carolina, Usa1

Posted on: Sunday, 23 October 2005, 03:01 CDT

By Lunetta, Ross S; Greene, Richard G; Lyon, John G

ABSTRACT:

This study quantified nonpoint source nitrogen (NPS-N) sources and sinks across the 14,582 km^sup 2^ Neuse River Basin (NRB) located in North Carolina, to provide tabular data summaries and graphic overlay products to support the development of management approaches to best achieve established N reduction goals. First, a remote sensor derived, land cover classification was performed to support modeling needs. Modeling efforts included the development of a mass balance model to quantify potential N sources and sinks, followed by a precipitation event driven hydrologic model to effectively transport excess N across the landscape to individual stream reaches to support subsequent labeling of transported N values corresponding to source origin. Results indicated that agricultural land contributed 55 percent of the total annual NPS-N loadings, followed by forested land at 23 percent (background), and urban areas at 21 percent. Average annual N source contributions were quantified for agricultural (1.4 kg/ha), urban (1.2 kg/ha), and forested cover types (0.5 kg/ha). Nonpoint source-N contributions were greatest during the winter (40 percent), followed by spring (32 percent), summer (28 percent), and fall (0.3 percent). Seasonal total N loadings shifted from urban dominated and forest dominated sources during the winter, to agricultural sources in the spring and summer. A quantitative assessment of the significant NRB land use activities indicated that high (greater than 70 percent impervious) and medium (greater than 35 percent impervious) density urban development were the greatest contributors of NPS-N on a unit area basis (1.9 and 1.6 kg/ha/yr, respectively), followed by row crops and pasture/hay cover types (1.4 kg/ha/yr).

(KEY TERMS: nonpoint source pollution; nitrogen modeling; surface water hydrology; geographic information systems; remote sensing.)

Lunetta, Ross S., Richard G. Greene, and John G. Lyon, 2005. Modeling the Distribution of Diffuse Nitrogen Sources and Sinks in the Neuse River Basin of North Carolina, USA. Journal of the American Water Resources Association (JAWRA) 41(5):1129-1147.

INTRODUCTION

Excessive nitrogen (N) loading is considered the critical factor for stimulating algal growth and eutrophication in estuarine and coastal marine waters (Ryther and Dunstan, 1971; Nixon, 1975). It has been demonstrated that estuarine eutrophication in the continental United States has increased since the onset of European settlement (Cooper and Brush, 1991). Symptoms of excessive N loading to estuarine systems include algal blooms, which increase biological oxygen demand (BOD) through decomposition processes, leading to bottom water hypoxia and under extreme conditions can culminate in fish kills (Paerl et al., 1998). Nutrient enrichment can also alter the structure and function of microbial communities, resulting in increased frequency and magnitude of phytoplankton blooms (Paerl, 1998). Pinckney et al. (1998) determined that annual cycles of phytoplankton community structure in the Neuse River of North Carolina were influenced by high inorganic N loadings, resulting in summer blooms of cryptomonads, chlorophytes, and cyanobacteria. Over the past two decades, the total N load to the Neuse River has increased by at least 30 percent (Stanley, 1988; Harned and Davenport, 1993; Dodd et al, 1993). Consequently, the water quality of the Neuse River Estuary has also declined (Paerl et al., 1998).

In the Neuse River Basin (NRB), N loading has increased coincident with increases in chemical fertilizer use in the early 1960s and animal feeding operations in the early 1970s (Stow et al, 2001). Stow et al. (2001) estimated that nonpoint source (NPS) loadings accounted for 75 percent and point source N 25 percent of total N loadings. Although NPS-N is the largest allochthonous N source, these sources are subject to soil and stream cycling processes, which include nitrogen fixation, denitrification, and ammonification, all of which affect ecosystem N budgets (Band et al., 2001). The proportion of NPS-N from fertilizer applications, animal feeding operations, and atmospheric deposition that actually enters the aquatic system (watercourses) is an ongoing research question of significance for evaluating the relative importance of NPS-N sources. Whitall et al. (2003) estimated that wet N deposition contributed up to 50 percent of the new NPS-N sources to the Neuse River Estuary from 1996 to 2000. Terrestrial nutrient cycling and hydrologie transport mechanisms represent important ecosystem processes that regulate denitrification and affect regional water quality (McMahon and Woodside, 1997).

The amended Clean Water Act of 1987 (section 319) requires states to perform an assessment of water quality problems, including those associated with diffuse (nonpoint) sources. Currently, high priority NPS issues are focused on nutrient and sediment transport from the landscape to receiving streams. These NPS loadings are used to support the development of total maximum daily loads (TMDLs) determinations of streams and rivers (USEPA, 1999). Ecosystem scale NPS processes are dynamic and function at multiple analytical scales. At the landscape scale, spatially explicit elements can function as both sources and sinks of nutrients and sediment. High resolution geospatial data can be used to support nutrient watershed scale modeling efforts. Landscape parameters commonly required to support these spatially distributed modeling approaches include the identification and delineation of individual land cover (LC) elements or patches. Landscape "patches" typically represent the primary modeling unit of a spatially distributed landscape model. They are defined in this study as contiguous and relatively homogeneous cover types that can be repetitively mapped using remote sensor data.

A detailed understanding of the distribution of terrestrial landscape elements is required to support spatially distributed mass balance modeling of N source allocations. In particular, the characterization of riparian buffer zones is essential to evaluate their function and denitrification value. Typically, riparian buffer zones are defined as areas directly adjacent to the top of the stream bank and extending outward in a perpendicular direction for a distance of approximately 20 to 30 m. Riparian buffer zones play important functional roles in nutrient cycling and erosion, and sedimentation deposition processes (Verchot et al., 1998). Characteristics associated with high quality riparian buffer zones include the presence of well established natural vegetative cover to provide stream bank stabilization; shading; and a physical and biological barrier to the migration of sediment, nutrients, and microbes from the surrounding landscape to receiving water courses. Vegetated riparian buffers also function as nutrient processors through the absorption and assimilation of nitrogen and phosphorous compounds into soils and vegetative structures (Peterjohn and Correll, 1984). Associated microbial communities fix and process nutrients associated with both surface water flow and shallow ground water seepage (Verchot et al., 1997).

The North Carolina Department of Environment and Natural Resources (NCDENR) has recommended a 30 percent total reduction in N loadings from an established baseline (1995) for the NRB (NCDENR, 2001). Allocation of N loading reductions was in part based on historic increases in N loading (Stanley, 1988), experimental work (Paerl, 1987; Paerl and Bowles, 1987), and eutrophication models (Bowen and Hieronymus, 2002), as well as scientific consensus (NCSS, 1996). McMahon and Woodside (1997) estimated that annual instream loads of N at two NRB stations representing agricultural dominated landscapes (1990) averaged 6.3 kg/ha. Using a mass balance approach, the NCDENR (2001) estimated overall NRB in-stream loading at 1.7 kg/ ha. Based on the dimensions of the NRB and N monitoring data collected at Kinston, North Carolina, Paerl et al. (1998) estimated total annual N loading at 1.6 kg/ha for 1996. Qian et al. (2000) concluded that only minor increases in N concentrations in upstream NRB locations had occurred over the past 20 years in response to increased loadings, while the ratio of nitrogen to phosphorus concentrations in the estuary has increased considerably over the past 10 years.

Study Overview

The objectives of this study were to delineate the spatial distributions of N sources and sinks, and quantify loadings to receiving streams within the NRB. Past studies have used geographic information system (GlS)-based approaches to develop statistical relationships between LC and various water quality parameters at the watershed scale of analysis (Poiani et al., 1996; Basnyat et al, 1999; Johnson et al, 2001; Tong and Chen, 2002). However, none has detailed the relationships between potential nutrient sources and sinks across the landscape and their juxtaposition (adjacency) to receiving streams. Here, the authors performed an initial LC characterization to map landscape "patches" throu\ghout the NRB, at multiple resolutions, to provide an optimized product to directly support N mass balance and hydrologie transport modeling. Next, an N mass balance model was applied using the remote sensing derived LC data on a pixel wise basis to generate quantitative estimates of potential excess N distributions that provided the baseline for subsequent precipitation event driven hydrologie transport modeling. Lastly, the hydrologie transport model was implemented in a GIS modeling environment to provide estimates (spatial) of transported excess N corresponding to a daily time step. Daily N transport values were then aggregated to characterize biologically important seasonal variations. Output products included the aggregated seasonal statistics quantifying N loadings by major LC types by 14- digit hydrologie unit codes (HUCs) and GIS coverages of N sources and sinks. A tabular N loadings database was developed as an input parameter for stream N decay modeling efforts. Lastly, spatially distributed N source sink graphics were developed to support the development of best management practices (BMPs) to obtain N reduction goals established for the NRB.

Study Area

The 14,582 km^sup 2^ NRB is contained entirely within the boundaries of the state of North Carolina (Figure 1). By definition, basin boundaries correspond with the U.S Geological Survey (USGS) 6- digit HUC, Number 030202. The upper, northwestern third of the Basin is located in the Piedmont physiographic region and the remainder in the mid-Atlantic coastal plain. The Piedmont portion of the Basin is characterized by highly erodible clay soils, rolling topography with broad ridges and stream valleys, and low gradient streams composed of a series of sluggish pools separated by riffles and occasional small rapids. In contrast, flat terrain, "blackwater streams," low lying wetlands, and productive estuarine areas characterize the coastal plain. Elevations within the NRB range from 276 m in the western part of the basin to sea level at the confluence of the Neuse River and Pamlico Sound. The Pamlico Sound represents the southern extent of the Albemarle-Pamlico Sound estuary system, which is bordered by a series of barrier islands known as the North Carolina's Outer Banks (NCDEM, 1993).

Figure 1. Neuse River Basin Location Map.

METHODS

Land Cover Classification

Two complete System Pour l'Observation de la Terre (SPOT 4) multispectral (XS) data acquisitions (20 scenes) and three complete sets of Landsat 7 Enhanced Thematic Mapper Plus (ETM+) level-lG (12 scenes) data, collected between fall 1998 and summer 1999, were used for the LC classification. An LC classification was performed using a hybrid approach that combined supervised (constrained), unsupervised (unconstrained), and rule based classification techniques (Lunetta et al., 2003). The final classification included forests (deciduous, evergreen, and mixed), agricultural land (fallow, hay and pasture, corn, cotton, soybeans, and tobacco), herbaceous vegetation (natural and maintained grasses), barren land (i.e., nonvegetated), wetlands (herbaceous and woody), open water, and urban cover types. Agricultural crop type identifications corresponded to calendar year 1999. Urban areas were defined based on percentage of impervious surfaces corresponding to: low (10 to 35 percent impervious); medium (36 to 70 percent impervious; and high (equal to or greater than 71 percent impervious) (Table 1). The LC map had a nominal landscape "patch" size of 0.4 ha for the general watershed areas, while locations adjacent to streams (riparian buffer zones) were mapped at 0.1 ha.

TABLE 1. Neuse River Basin Land Cover Classification System and Final Classification Results by Percent [ ] for Each Class Type Corresponding to Classification Levels 1 Through 3.

Unique aspects of this LC classification versus the available National Land Cover Data (NLCD) included: (1) better representation of current conditions (i.e., 1999 versus 1991), (2) characterization of impervious surfaces, (3) identification of major crop types, and (4) finer mapping resolution of riparian buffer zones. These enhancements were incorporated to support both the mass balance and hydrologie modeling efforts. For example, the identification of major crop types was used to quantify fertilizer application rates and impervious surfaces percentages that were then used to calculate water infiltration and surface flow rates. The mapped landscape "patches" were used to support the N mass balance and hydrologie modeling efforts described below.

Mass Balance Modeling

Nitrogen mass balance modeling provided "potential excess" nitrogen values for all landscape "patches" in the NRB. These values were subsequently used in the precipitation event driven hydrologie simulation modeling described below. Potential excess N was calculated as the difference between inputs to and outputs from the inorganic N pool. If inputs exceeded outputs, then the difference was assumed to represent N at risk of loss from the landscape to surface receiving waters and ground water. Potential excess N for a particular LC type "patch" was calculated according the following mass balance equation.

X = (I + F + M) - (U + D + V) (1)

where X is potential excess N, I is atmospheric N deposition, F is fertilizer N inputs, M is net soil N mineralization, U is uptake of N by plants, D is denitrification, and V is volatilization of fertilizer N, all in grams of N per square meter per day. To account for the variable rates of soil N mineralization, the model partitioned the seasonal fractions of annual net soil N mineralization, based on total N soil measurement data from the National Soils Characterization Database (NSCD) and soils taxonomy in the State Soil Geographic (STATSGO) database (USDA, 1994). The temporal allocation pattern was based on previous studies that indicated a spring maximum and a winter minimum (Garten and Ashwood, 2003). Seasonal factors for atmospheric N deposition in eastern North Carolina were derived from the National Atmospheric Deposition Program/National Trends Network (NADP/NTN) data. Seasonal factors for N fertilization were estimated from commercial fertilizer tonnage shipped to North Carolina from July 1988 through June 1999. Variations in N uptake by plants were assumed to have tracked expected seasonal differences in plant tissue production. The largest portion of the annual denitrification flux was allocated to winter and spring (Garten and Ashwood, 2003). Spatially distributed "potential excess" N maps were developed on a seasonal (winter, spring, summer, fall) time step to support the hydrologie transport modeling discussed below. For a detailed accounting of all N sources, readers are referred to Garten and Ashwood (2003).

Hydrologic Modeling

Figure 2. Conceptual Model for the Precipitation Event Driven Hydrologie Transport Model Incorporating Evapotranspiration Processes.

Figure 3. Conceptual Model for the Routing of Transported N Across the Landscape to Receiving Stream Reaches.

RESULTS

Land Cover Products

Results of the 1998 to 1999 LC classification are presented in Table 1. For the NRB woody vegetation (39 percent) was the predominant cover type, followed by agricultural (29 percent), wetland (14 percent), and urban (14 percent). Woody vegetation (forest) was further differentiated as deciduous (52 percent), evergreen (34 percent), and mixed (14 percent). Lands in agricultural production were further classified as row crop (61 percent), pasture/hay (39 percent), or fallow (0.6 percent). Row crops were further classified as soybeans (42 percent), cotton (25 percent), corn (23 percent), or tobacco (10 percent). The majority of urban areas were classified as low density (70 percent), followed by medium density (16 percent) and high density (14 percent), based on percent impervious surface area (10 to 35, 36 to 70, and greater than or equal to 71 percent, respectively). All urban areas with less than 50 percent impervious surfaces were further labeled with their representative dominate cover type. For example, low density urban areas were dominated by woody vegetation (43 percent), followed by herbaceous vegetation (30 percent), water (16 percent), and agricultural land (9.0 percent). Separate accuracy assessments were performed for both the riparian buffer zone and general watershed areas across all hierarchal classification levels. Level I accuracies were 82 percent (n = 825) for general watershed areas and 73 percent (n = 391) within riparian buffer zones (Lunetta et al., 2003)

Model Performance

Daily precipitation event data were modeled and aggregated to seasonal time periods corresponding with spring, summer and fall 1998, and winter 1999. The hydrologie model was compared to three USGS surface water gauging stations that were selected because their drainage areas directly coincided with areas defined by multiple 14- digit HUCs within the Neuse River Basin. Daily streamflow data from the Eno River at Hillsborough, Flat River at Bahama, and Contentnea Creek at Hookerton were used for the analysis. A comparison of modeled and observed streamflow volumes for three USGS stations is illustrated in Figure 4 (a-c). There was a good correspondence between seasonal patterns of the modeled versus observed streamflow volumes across all three sites. The observed difference between the lower modeled versus higher observed volumes represents the ground water and direct precipitation contribution to the streamflow volumes. The correspondence evaluation showed that an average of approximately 54 percent of transport volumes originated from the terrestrial surface of watersheds that contributed to the Eno River (58 percent), Flat River (58 percent), and Contentnea Creek (47 percent) (Table 2). Although the N transport was only modeled through winter 1999, the hydrologie model was run through summer 1999 to support the analysis described a\bove.

Figure 4. A Comparison of Modeled and Observed Streamflow Volumes for Three Locations in the Neuse River Basin, North Carolina.

TABLE 2. Seasonal Computed Volumes as a Percentage of the Observed Values for the Three Neuse River Basin Assessment Stations.

Modeling Products

Spatially distributed N source-sink products were processed by individual 14-digit HUCs (n = 188), then assembled to also provide a seamless coverage for the NRB. Table 3 provides a summary of the overall NRB modeling results. Agriculture, which represented 29 percent of the basin, contributed 55 percent of the N to streams. The average contribution of N from agricultural land was 1.4 kg/ha. Row crops, which represented 61 percent of all land in agricultural production, were responsible for 61 percent of N contributions, followed by pasture/hay and fallow contributing 38 percent and 1.5 percent, respectively. Urban areas represented 14 percent of the NRB and contributed 21 percent of total N, or an average of 1.2 kg/ha/ yr. Low density urban was 70 percent of urban areas and contributed 54 percent, followed by medium-density (23 percent), and high (23 percent). Forested lands represented 38 percent of the NRB and contributed 23 percent of the total N, or an average of 0.5 kg/ha/ yr. Forested LC represents the dominate pre-European landscape condition, and thus is considered to represent background (natural) N loadings. Combined, the barren, herbaceous, and wetland LC types contributed only 0.9 percent of the total annual N budget to receiving streams. Combined (1999) annual loadings of NPS-N to NRB streams averaged 0.8 kg/ha/yr.

Seasonal N contributions varied considerably by LC type (Table 3). Approximately 72 percent of total NPS-N inputs occurred during the winter (40 percent) and spring (32 percent), followed next by summer (28 percent), and only minimal fall (0.3 percent) contributions. Agricultural contributions were greatest during the summer (45 percent), followed by spring (36 percent), winter (19 percent), and fall (0.2 percent). Urban contributions peaked in winter (73 percent), followed by spring (20 percent), summer (6.4 percent), and fall (0.5 percent). Similarly, forested contributions were highest in the winter (60 percent), followed by spring (40 percent), summer (7.3 percent), and fall (0.3 percent). However, considerable seasonal variability between deciduous and conifer forests were documented. Deciduous forest N contributions peaked in winter (90 percent), while conifer forest contributions tended to peak in the spring (51 percent), but also had a significant winter contribution (40 percent). Barren contributions were distributed throughout the winter (38 percent), spring (35 percent), and summer (25 percent) periods, coinciding with precipitation maxima. On an area weighted basis, urban loadings were two times greater than agricultural loadings for winter. However, agriculture loadings were five times and 18 times greater than urban loadings during the spring and summer, respectively.

Modeling outputs also differentiated between surface and subsurface N fractions (Table 4). Overall, subsurface versus surface N contributions were 53 percent and 47 percent, respectively. Forested land had the greatest total subsurface contributions (70 percent), followed by herbaceous (63 percent) and urban areas (54 percent). Surface N contributions were greatest for barren (77 percent) and agricultural (53 percent) areas. Seasonally, spring and summer N delivery was predominately surface vector at 45 percent and 36 percent, respectively. Subsurface contributions tended to be highest during the winter (58 percent) season.

TABLE 3. Annual and Seasonal Transported Total Nitrogen (N) by Cover Types for the Neuse River Basin (NRB), North Carolina, 1999.

The HUCs (n = 188) were also analyzed based on their relative annual NPS-N contributions to the NRB. Table 5 lists the area weighted rankings of HUCs representing both the highest and lowest 10 percent contributors of N. Interestingly, three forest dominated HUCs (LD. Nos. 183, 188, and 185) were ranked among the highest contributors. However, the majority of HUCs in the top 10 percent were dominated by agriculture (n = 9), followed by urban (n = 5), and mixed (n = 2). The seasonal contributions of the urban and agriculture dominated HUCs followed the typical patterns of winter peak urban contributions with agricultural peaking during the spring/ summer. The two mixed HUCs all exhibited seasonal patterns most closely corresponding to agriculture dominated systems. Conversely, the lowest N loadings were estimated for forest (n = 14) and wetland (n = 4) dominated systems, with one urban (LD. No. 169) inclusion.

Hydrologie unit codes were also ranked based on predominate cover type to evaluate the variability associated with the most homogeneous watersheds within the NRB. Table 6 lists and ranks the HUCs dominated by agriculture (n = 10), urban (n = 10), and forest (n = 10). Agriculture dominated HUCs ranged from 58.0 to 71.3 percent agricultural cover, and ranked from among the highest to lowest N contributors. Urban dominated HUCs ranged from 57 to 77 percent urban, and consistently ranked among the highest contributors. While forest dominated HUCs ranged from 69 to 81 percent forest, and consistently ranked in the lowest 20 percent. However, one HUC (LD. No. 188) was ranked as the fifth greatest contributor of N on a per unit area basis within the NRB.

TABLE 4. Surface and Subsurface Transported Total Nitrogen (N) by Cover Types for the Neuse River Basin (NRB), North Carolina, for 1999.

To better understand the observed variability between the HUCs described above, the predicted N source and sink products were overlaid on ETM^sup +^ 15 m panchromatic imagery to provide the spatial context for further evaluation. Examples of agriculture dominated HUCs are presented to illustrate the spatial patterns corresponding to both the highest (LD. No. 158), and among the lowest (LD. No. 3) annual agricultural source within the NRB (Figure 5). The spatial juxtaposition of agricultural N sources to nonbuffered receiving streams is clearly evident in the Bear Creek River (LD. No. 158) catchment (Figure 5a). While the stream network of the Lower Little Contentnea (LD. No. 3) catchment tended to have more extensive riparian buffers (Figure 5b). Stream reach forested riparian buffers were determined to be 51 percent for HUC No. 3 and 37 percent for HUC No. 158. Additionally, row crop production in the Bear Creek River catchment was dominated by corn (45 percent), and in the Little Contentnea, dominated by soybeans (42 percent). Corn N fertilizer applications rates were approximately 12 times those applied for soybean row crops (Garten and Ashwood, 2003). The model results indicated that average agricultural N sources were 4.8 kg/ ha for the lower buffered corn dominated HUC (LD. No. 158), compared to 0.4 kg/ha for better buffered soybean dominated HUC (LD. No. 3) (Tables 7 and 8).

Figure 6 illustrates changes in seasonal patterns of N sources across the NRB by HUC. During the winter months, urban sources including the Raleigh-Durham metropolitan area and the cities of Goldsboro, Wilson, and Kinston were the greatest NPS-N sources ranging from 0.5 to 1.3 kg/ha (Figure 6a). Spring was a transition period between the urban and agricultural sources (Figure 6b). Agricultural N sources clearly dominated during the summer months with values ranging from 0.3 to 4.5 kg/ha (Figure 6c). Agriculture also tended to dominate during the fall months, although at very low values, less than 0.1 kg/ha (no graphic). High annual N contributors were primarily agricultural, with the inclusion of some urban areas (Figure 6d).

TABLE 5. Rankings of NRB 14-Digit HUCs (n = 188) Corresponding to the Highest and Lowest Annual Nitrogen (N) Sources to Receiving Water Courses.

Model Comparisons

The HUC-based N loading products were compared to other available NRB modeling results and instream monitoring station based estimates for index periods ranging from 1990 to 1999. The study 1999 reference year instream loading value of 0.8 kg/ha was approximately 50 percent of that reported by Paerl et al. (1998) at 1.6 kg/ha for reference year 1996, and the NCDENR (2001) value of 1.7 kg/ha for reference year 1999 (Table 9). Similarly, the total anthropogenic NPS-N loading value of 21 kg/ha (Garten and Ashwood, 2003) was approximately 50 percent less than the Stow et al. (2001) estimate of 39 kg/ha for reference year 1999. Although the Stow et al. (2001) value was higher, it corresponded only to NRB agricultural (cropland) N sources. McMahon and Woodside (1997) reported a total N loading value of 6.3 kg/ha for an agriculture dominated NRB watershed for reference year 1990. This value was approximately 4.5 times greater than the average NRB agricultural value of 1.4 kg/ha (Table 3), but compared reasonably well with the highest contributing agriculture dominated HUC (LD. No. 158) at 4.8 kg/ha (Table 7). Additionally, the McMahon and Woodside (1997) value was based on instream monitoring measurements that included ground water, animals, and point source N inputs that were not accounted for in this modeling effort.

TABLE 6. Rankings and Seasonal Loadings (percent) of NRB 14- Digit HUCs (n = 188) Grouped by Predominate Land Cover Type.

Figure 5. Data Visualization of the Spatial Distribution of N Sources and Sinks for Two HUCs Dominated by Agricultural Cover Types: (a) Bear Creek River Catchment and (b) Lower Little Contentnea Catchment

DISCUSSION

The single largest N inventory pool in the mass balance model was contained in the organic soils throughout the NRB. This large potential N source was calculated based on an annual net N mineralization rate to arrive at an annual flux. The soils inventory data used in this study originated from the NSCD and information on soil taxonomy was derived from STATSGO database (\Garten and Ashwood, 2003). The NSCD-STATSGO data had a spatial resolution of 100 ha, compared to the variable high resolution LC data (0.1 to 0.4 ha). This disparity in data resolutions potentially represented a major source of modeling error because of the relatively large proportion of N associated with the organic soil N pool. The significance of the organic soils pool was clearly illustrated by the unexpectedly high contributions of N from the Rattlesnake Branch HUC (LD. No. 188). Although this watershed was approximately 77 percent forested cover, it was the single highest annual contributor of N within the NRB. Forest N loadings were 4.5 kg/ha, or approximately nine times greater than the NRB average annual rates of 0.5 kg/ha. These elevated loadings were attributed solely to the high N mineralization rates predicted by the model and represented an elevated, yet naturally occurring (background) N source in the NRB.

TABLE 7. Annual and Seasonal Transported Total Nitrogen (N) by Land Cover Types for the Bear Creek River Catchment (No. 158) in the Neuse River Basin, North Carolina.

TABLE 8. Annual and Seasonal Transported Total Nitrogen (N) by Land Cover Types for the Lower Little Contentnea Catchment (No. 3) in the Neuse River Basin, North Carolina.

Figure 6. Series of Equal Frequency Distributions Illustrating the Spatial Distribution of N Sources Across the Neuse River Basin Corresponding to (a) Winter, (b) Spring, (c) Summer, and (d) Annual.

Figure 6. Series of Equal Frequency Distributions Illustrating the Spatial Distribution of N Sources Across the Neuse River Basin Corresponding to (a) Winter, (b) Spring, (c) Summer, and (d) Annual (cont'd).

TABLE 9. Annual Nitrogen Loading Estimates Derived From Anthropogenic Source and Instream Measurement Data for the Neuse River Basin, North Carolina.

An additional shortcoming of this modeling effort was a result of the limitations associated with the available agricultural crop type data. Although the LC cover data described in this article were optimized to characterize specific crop types, there was a limit to the number of crop types that could be mapped. This limitation was mainly due to the very diverse crop types within the NRB and the temporal resolution limitations associated with the space borne ETM^sup +^ sensor. Crop types were mapped corresponding to what was present during the summer growing season for reference year 1999. Most notably, winter wheat was unrepresented by this mapping effort. Based on crop statistics tabulated by the North Carolina Department of Agriculture and Consumer Services (NCD/CS), approximately 10 percent of the study would have been planted with winter wheat during the spring and fall months of 1999. The inclusion of wheat would have likely resulted in a discernible increase in the contribution of agricultural N sources and a slight shift in the seasonal N source distributions, from summer to winter and spring.

The modeling approach presented in this paper is applicable to highly valued ecosystems like the NRB. If only a relative ranking of watersheds N contributions is required, a statistical-based regression modeling analysis of LC percentages versus monitoring station data would typically be sufficient to provide a rapid assessment. However, if a detailed understanding of the distribution of N source and sink across the landscape is required, a more comprehensive, spatially distributed modeling approach is usually required. Advantages of the modeling approach presented in this paper include: (1) a detailed understanding of the spatial distribution of N sources and sinks corresponding to individual landscape "patches" to support the development of detailed BMPs, (2) differentiation of N source contributions on a seasonal basis to develop linkages with important ecological processes (e.g., summer algal blooms), (3) quantitative stream reach loading data to initialize instream N decay models, and (4) partitioning of surface versus ground water N to provide mechanisms for linkages with ground water N transport models. Also, this type of detailed modeling approach can provide a better understanding of the complex ecological processes that cannot be achieved using only LC type as the principal modeling variable. For example, in the case of the Rattlesnake Creek HUC described above, a simple landscape-based regression analysis would have ranked this HUC among the lowest N sources, where the distributed modeling approach predicted it to be among the highest because of the high soil N mineralization potential.

The nutrient modeling effort presented by the authors represents an initial effort to link spatially distributed LC and soils data with a daily precipitation event driven hydrologie transport model to quantify N source contributions of individual landscape "patches" to receiving stream reaches on a seasonal time step. The model was implemented using standard GIS analysis tools, multiple spatial data coverages, and process-based rate coefficients. Output products included both a tabular (spreadsheet) database and high resolution graphic overlays. The tabular database was developed to initialize instream N decay modeling, and provide inventory data to support the development of NRB TMDL N reduction targets. Graphic overlays were optimized to provide a detailed understanding of the spatial distributions of N sources and sinks in the context of their juxtaposition to receiving stream reaches. Graphics were specifically developed to provide data visualization products to support the development of BMPs strategies for implementing N reduction goals. Individual GIS data coverages could be updated to reflect current conditions, or modified to predict the results of various BMP implementation scenarios. Also, process rate coefficients can be modified to reflect advances in the understanding of ecosystem N processes. The availability of higher resolution soils data, detailed soils taxonomic information, and improved (seasonal) agricultural crop type data would likely enhance NPS-N model performance.

CONCLUSIONS

The modeling results presented in this paper demonstrated both the strengths and weaknesses of spatially distributed modeling capabilities for NPS-N. Currently available moderate resolution satellite imagery can be used to derive sufficiently detailed maps of landscape condition to support distributed nutrient mass balance and hydrologie modeling efforts, and provide graphic products detailing the spatial distribution of NPS-N sources and sinks. However, the capability to validate modeling results were severely limited by the placement of NRB water quality monitoring stations at sites associated with higher order streams and reservoir outflows. Monitoring data obtained from these sampling locations represent a mixture of (1) surface flow N inputs across multiple cover types and watersheds, (2) ground water N contributions that function at a vastly different time step compared to surface water flows, and (3) downstream and reservoir nitrification and denitrification processes that subsequently alter N concentrations. Additional water quality monitoring stations located on first-order streams and at the outflows of isolated watersheds (functioning largely independent of outside surface flows) would greatly enhance future modeling validation efforts.

A major impediment of current modeling capabilities is the ability to directly incorporate biological process to further refine estimates of NPS-N export from the landscape to receiving stream networks. To substantially advance future modeling capabilities, research is required to increase the understanding of landscape- based nutrient processes and more directly integrate satellite remote sensor measurement data to dynamically drive important ecosystem processes. The incorporation of biological processes would potentially be important to better understand why observed increases in NPS-N loadings over the past 20 years have not resulted in a proportional increase in downstream N loadings (Stow et al., 2001). Additionally, a major impediment to current and future modeling advancements that needs to be overcome is the availability of high spatial resolution soils taxonomy data to support the calculation of soil N inventories.

ACKNOWLEDGMENTS

The authors would like to thank Hans Paerl, Stephen Kraemer, and Chris Roessler for their insightful comments in the preparation of this manuscript, and acknowledge the N mass balance modeling performed by Chuck Garden and Thomas Ashwood in support of this effort. The U.S. Environmental Protection Agency funded and partially conducted the research described in this paper. It has been subject to the Agency's programmatic review and has been approved for publication. Mention of any trade names or commercial products does not constitute endorsement or recommendation for use.

1 Paper No. 04008 of the Journal of the American Water Resources Association (JAWRA) (Copyright 2005). Discussions are open until April 1, 2006.

LITERATURE CITED

Band, L.E., C.L. Tague, P. Groffman, and K. Belt, 2001. Forest Ecosystem Processes at the Watershed Scale: Hydrological and Ecological Controls of Nitrogen Export. Hydrol. Processes. 15:2013- 2028.

Basnyat, P., L.D. Teeter, K.M. Flynn, and E.G. Lockaby, 1999. Relationships Between Landscape Characteristics and Nonpoint Source Pollution Inputs to Coastal Estuaries. Environmental Management 23(4):539-549.

Bowen, J.D. and J.W. Hieronymus, 2002. Model Predicted Water Quality Response to Reductions in Inorganic and Organic Nitrogen Loading. In: Estuarine and Coastal Modeling, M.L. Spaulding (Editor). Proceedings of the 7th International Conference, American Society of Civil Engineers, New York, New York, pp. 244-265.

Cooper, R. and G.S. Brush, 1991. Long-Term History of Chesapeake Bay Anoxia. Science 254:992-996.

Dodd, R.C., P.A. Cunningham, R.J. Curry, and S.J. Stichter, 1993. Watershe\d Planning in the Albemarle-Pamlico Estuarine System. Report No. 93-01, Research Triangle Institute, Research Triangle Park, North Carolina.

Duffie, J.A. and W.A. Beckman, 1980. Solar Engineering of Thermal Processes. John Wiley and Sons, New York, New York, 762 pp.

Flanagan, D.C. and G.R. Foster, 1989. Storm Pattern Effect on Nitrogen and Phosphorus Losses in Surface Runoff. Transactions of ASAE 32(2):535-544.

Garten, Jr., C.T. and TL. Ashwood, 2003. A Landscape Level Analysis of Potential Excess Nitrogen in East-Central North Carolina, USA. Water, Air, and Soil Pollution 146:3-21.

Hargreaves G.H. and Z.A. Samani, 1982. Estimating Potential Evapotranspiration. Proceedings of the American Society of Civil Engineers 108(IR3):225-230.

Hargreaves G.H. and Z.A. Samani, 1985. Reference Crop Evapotranspiration From Temperature. Applied Engineering in Agriculture 1(2):96-99.

Harned, J.E. and M.S. Davenport, 1993. Water-Quality Trends and Basin Activities and Characteristics for the Albemarle-Pamlico Estuary NC and VA. Report No. 90-398, U.S. Geological Survey, Raleigh, North Carolina.

Jensen, M.E., R.D. Burman, and R.G. Alien (Editors), 1990. Evapotranspiration and Irrigation Water Requirements. ASCE Manual No 70, American Society of Civil Engineers, New York, New York, 360 pp.

Johnson, D.J., WL. Myers, and G.P. Patil, 2001. Predictability of Surface Water Pollution Loading in Pennsylvania Using Watershed- Based Landscape Measurements. Journal of the American Water Resources Association (JAWRA) 37(4):821-835.

Lunetta, R.S., J. Ediriwickrema, J. Iiames, D. Johnson, J.G. Lyon, A. McKerrow, and D. Pilant, 2003. A Quantitative Assessment of a Combined Spectral and GIS Rule-Based Land-Cover Classification in the Neuse River Basin of North Carolina. Photogrammetric Engineering and Remote Sensing 69(3):299-310.

McCuen, R.H., 1982. A Guide to Hydrologie Analysis Using SCS Methods. Prentice-Hall, Englewood, California, 145 pp.

McMahon, G. and M.D. Woodside, 1997. Nutrient Mass Balance for the Albemarle-Pamlico Drainage Basin, North Carolina and Virginia, 1990. Journal of the American Water Resources Association (JAWRA) 33(3):573-589.

NCDENR (North Carolina Department of Environment and Natural Resources), 2001. Phase II of the Total Maximum Daily Load for Total Nitrogen to the Neuse River Estuary, North Carolina. Raleigh, North Carolina, 57 pp.

NCDEM (North Carolina Division of Environmental Management), 1993. Neuse River Basin: Water Quality Management Plan. Raleigh, North Carolina, 140 pp. + Appendices.

NCSS (North Carolina State Senate), 1996. Senate Select Committee on Water Quality and Fish Kills. Summary Report, North Carolina State Senate, Raleigh, North Carolina.

Nixon, S., 1975. Coastal Marine Eutrophication: A Definition, Social Causes, and Future Concerns. Ophelia 41:199-220.

Paerl, H.W, 1987. Dynamics of Blue-Green Algal (Microcystis aeruginosa) Blooms in the Lower Neuse River, North Carolina: Causative Factors and Potential Controls. University of North Carolina Water Resources Research Institute Report No. 229, Raleigh, North Carolina, 164 pp.

Paerl, H.W, 1998. Structure and Function of Anthropogenically Altered Microbial Communities in Coastal Waters. Current Opinion in Micro. 1:296-302.

Paerl, H.W and N.D. Bowles, 1987. Dilution Bioassays: Their Application to Assessments of Nutrient Limitation in Hypereutrophic Waters. Hydrobiologia 146:265-273.

Paerl, H.W., J.L. Pinckney, J.M. Fear, and B.L. Peierls, 1998. Ecosystem Response to Internal and Watershed Loading: Consequences for Hypoxia in the Eutrophying Neuse River Estuary, North Carolina, USA. Mar. Ecol. Prog. Ser. 166:17-25.

Peterjohn, WT. and D.L. Correll, 1984. Nutrient Dynamics in an Agricultural Watershed: Observations on the Role of Riparian Forests. Ecology 65:1466-1475.

Pinckney, J.L., H.W. Paerl, M.B. Harrington, and K.E. Howe, 1998. Annual Cycles of Phytoplankton Community-Structure and Bloom Dynamics in the Neuse River Estuary, North Carolina. Marine Biology 131:371-381.

Poiani, KA., B.L. Bedford, and M.D. Merrill, 1996. A GIS-Based Index for Relating Landscape Characteristics to Potential Nitrogen Leaching to Wetlands. Landscape Ecology 11(4):237-255.

Preti, F., 1999. Overland Transport of Nitrate: Algorithms Validation. Water Science and Technology 39(3):77-92.

Qian, S.S., M.E. Borsuk, and C.A. Stow, 2000. Seasonal and LongTerm Nutrient Trend Decomposition Along a Spatial Gradient in the Neuse River Watershed. Environ. Sci. Technol. 34:4474-4482.

Ryther, J.H. and W.M. Dunstan, 1971. Nitrogen, Phosphorus, and Eutrophication in the Coastal Marine Environment. Science 171:1008- 1013.

Stanley, D.W., 1988. Historical Trends in Nutrient Loading to the Neuse River Estuary, NC. In: Coastal Water Resources, W. Lyke and T. Hoban (Editors). American Water Resources Association, Middleburg, Virginia, TPS-88-1, pp. 155-164.

Stow, C.A., M.E. Borsuk, and D.W. Stanley, 2001. Long-Term Changes in Watershed Nutrient Inputs and Riverine Exports in the Neuse River, North Carolina. Wat. Res. 35(6):1489-1499.

Tong, S.T.Y. and W. Chen, 2002. Modeling the Relationship Between Land Use and Surface Water Quality. Journal of Environmental Management 66:377-393.

USDA (U.S. Department of Agriculture), 1986. Urban Hydrology for Small Watersheds ( second Edition). Soil Conservation Service, Engineering Division, Technical Report TR-55, National Technical Information Center, Springfield, Virginia.

USDA (U.S. Department of Agriculture), 1994. State Soil Geographic (STAGO) Data Base: Data Use Information. National Cartography and GIS Center, U.S. Department of Agriculture, National Resource Conservation Center, Fort Worth, Texas.

USEPA (U.S. Environmental Protection Agency), 1999. Protocol for Developing Nutrient TMDLs. EPA Report 841-B-99-007, Office of Water, Washington, D.C., 128 pp.

Verchot, L.V., E.G. Franklin, and J.W. Gilliam, 1997. Nitrogen Cycling in Piedmont Vegetated Filter Zones: I. Surface Soil Processes. J.Environ. Qual. 26:327-336.

Verchot, L.V., E.G. Franklin, and J.W. Gilliam, 1998. Effects of Agriculture Runoff Dispersion on Nitrate Reduction in Forested Filter Zone Soils. Soil Sci. Soc. Am. Journal 62:1719-1724.

Whitall, D., B. Hendrickson, and H.W. Paerl, 2003. Importance of Atmospherically Deposited Nitrogen to the Annual Nitrogen Budget of the Neuse River Estuary, North Carolina. Environment International 29:393-399.

Ross S. Lunetta, Richard G. Greene, and John G. Lyon2

2 Respectively, Research Environmental Scientist, U.S. Environmental Protection Agency, National Exposure Research Laboratory (E24305), Research Triangle Park, North Carolina 27709; NRC Associate, National Research Council, National Exposure Research Laboratory (243-05), Research Triangle Park, North Carolina 27709; and Director, ESD, U.S. Environmental Protection Agency, National Exposure Research Laboratory, 944 East Harmon, Las Vegas, Nevada 89119 (E-Mail/Lunetta: lunetta.ross@epa.gov).

Copyright American Water Resources Association Oct 2005

Source: Journal of the American Water Resources Association

More News in this Category