| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Environmental Resources Research Institute, 001 Land and Water Building, Pennsylvania State Univ., University Park, PA 16802
b School of Forest Resources, 311 Forest Resources Lab., Pennsylvania State Univ., University Park, PA 16802
c School of Forest Resources, 201 Forest Resources Lab., Pennsylvania State Univ., University Park, PA 16802
* Corresponding author (sas371{at}psu.edu)
Received for publication June 29, 2001.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Abbreviations: CASTNet, Clean Air Status and Trends Network • CBW, Chesapeake Bay watershed • GIS, geographic information system • NADP, National Atmospheric Deposition Program • USGS, United States Geological Survey
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Atmospheric deposition is of particular interest since a number of studies have indicated that atmospheric deposition may account for as much as 25 to 80% of the total nitrogen entering the bay (Fisher et al., 1988; Tyler, 1988; Hinga et al., 1991; Fisher and Oppenheimer, 1991; Boynton et al., 1995; Jaworski et al., 1997). However, questions regarding the accuracy of atmospheric deposition inputs, nitrogen retention estimates, and in-stream nutrient uptake rates used in these studies have been raised in a number of assessment reports (Baker et al., 1993; Hicks et al., 1995, 1996; Gardner et al., 1996). More specifically, these uncertainties include:
1. Wet deposition estimates to the entire watershed were based on a relatively small number of point measurements. For example, Fisher et al. (1988) used four National Atmospheric Deposition Program (NADP) sites, and Russell et al. (1998) used data collected from only one NADP site.
2. Both wet deposition and export estimates were based on annual values, seasonal variation in both deposition and export were not considered (Fisher et al., 1988; Tyler, 1988; Hinga et al., 1991; Fisher and Oppenheimer, 1991).
3. Deposition and export calculations were generally based on only one year of data (Fisher et al., 1988; Tyler, 1988; Hinga et al., 1991; Fisher and Oppenheimer, 1991).
4. Dry deposition measurements were assumed to be a fixed percentage (generally 100%) of wet deposition, no adjustments were made relative to land cover (Fisher et al., 1988; Tyler, 1988; Hinga et al., 1991; Fisher and Oppenheimer, 1991).
5. Retention and export coefficients for various land cover classes were estimated and assumed to be constant within land cover classes and within regions with different deposition rates (Fisher et al., 1988; Tyler, 1988; Hinga et al., 1991; Fisher and Oppenheimer, 1991).
6. All nitrogen exported to streams draining forested watersheds was assumed to be from atmospheric sources (Fisher et al., 1988; Tyler, 1988; Hinga et al., 1991; Fisher and Oppenheimer, 1991).
7. Deposition and export calculations were based on a basin-wide aggregate of land cover classes (e.g., all forest lands, all agricultural lands, etc.), not the distribution of land uses within specific subbasins of the CBW (Fisher et al., 1988; Tyler, 1988; Hinga et al., 1991; Fisher and Oppenheimer, 1991).
8. In-stream removal estimates were based on coefficients that were held constant for all surface waters throughout the watershed (Fisher et al., 1988; Tyler, 1988; Hinga et al., 1991; Fisher and Oppenheimer, 1991).
9. Verification of export estimates against measured export data was not undertaken (Fisher et al., 1988; Tyler, 1988; Hinga et al., 1991; Fisher and Oppenheimer, 1991).
This study has been designed to address many of the uncertainties listed above. The contribution of atmospheric nitrogen deposition to the Chesapeake Bay was reassessed on a seasonal basis from 1984 through 1996 using a high resolution precipitation model, measured concentrations of nitrogen species in precipitation, and hydrologically important watershed characteristics. Using a geographic information system (GIS), land use, soil hydrologic groups, and drainage basin grids of the Chesapeake Bay watershed were assembled. An interpolated surface illustrating nitrogen export potential was created using these grids and nitrogen export coefficients reported in the literature (Nizeyimana et al., 1997; Mitsch, 1977). Wet and dry deposition loads to each of 64 subbasins in the Chesapeake Bay watershed were estimated on a seasonal basis from 1984 through 1996 using a high resolution wet deposition model (Grimm and Lynch, 2000) and dry to wet deposition ratios determined from USEPA Clean Air Status and Trends Network (CASTNet) deposition monitoring data (USEPA, 1995). Stream export data compiled by Langland et al. (1995)(1998) and partitioning coefficients determined by Nizeyimana et al. (1997) and Mitsch (1977) were used to estimate watershed nitrogen retention coefficients for each of the subbasins. Distance-weighted in-stream nitrogen utilization rates were obtained to estimate the amount of nitrogen exported from each subbasin that reaches the Chesapeake Bay. Where possible, estimates of nitrogen export based on model calculations were compared with existing stream discharge-concentration measurements.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
 |
The elevation (e) and slope and aspect parameters (N, S, E, W) were derived from 1 by 1 degree United States Geological Survey (USGS) digital elevation model (DEM) data sets. Each of the four slope and aspect values for a given point was determined using five, 16.1-km (10-mile) radial transects starting from the given point along bearings 0, ±22.5, and ±45 degrees from the major compass bearing. A mean slope value was calculated for each transect as:
 |
The five transect slope values were then averaged to produce the corresponding direction slope and aspect value for the given point.
Wet deposition data for the WLLSR algorithm were derived from daily precipitation records from the National Oceanic and Atmospheric Administration's (NOAA) National Climatic Data Center and from weekly precipitation chemistry data from the National Atmospheric Deposition Program/National Trends Network (NADP/NTN). The NOAA precipitation measurements are available from approximately 8000 cooperative sites across the USA (France, 1994). The NADP/NTN precipitation chemistry data are obtainable from a relatively sparse network of approximately 220 sites across the USA (National Atmospheric Deposition Program, 2000). Because of the sparseness of NADP/NTN sites, their data cannot be used directly to model the effects of topography on wet deposition. To obtain adequate wet deposition sample density, the concentration data from the NADP/NTN sites were interpolated using a multiquadic equation (MQE) (Hardy, 1971) to each of the NOAA sites and the corresponding concentration estimates and precipitation values were used to calculate deposition as:
 |
Daily precipitation and weekly concentration records were both summarized into seasons (December–February, March–May, June–August, and September–November) for each of 13 years used in this analysis. Precipitation was summarized as total volume measured and precipitation ionic concentrations as volume-weighted means. Wet deposition estimates for each season of each year used in this study were modeled separately (Fig. 2) .
|
Estimating Dry Deposition to the Chesapeake Bay Watershed
Dry to wet deposition ratios were based on measured dry and wet depositions at 20 CASTNet (USEPA, 1995) and NADP (National Atmospheric Deposition Program, 2000) monitoring sites located within or on the periphery of the CBW. Since the amount of dry deposition received at a given point is a function of land cover among other variables, CASTNet dry deposition and NADP wet deposition estimates were grouped by land cover classification and season. For comparison, dry deposition of nitrogen was assumed to equal the sum of nitric acid nitrogen, ammonia nitrogen, and nitrate nitrogen, while wet deposition of nitrogen was assumed to equal the sum of ammonia nitrogen and nitrate nitrogen. Data collected between June 1988 and May 1999 were used in this analysis. The values in each category were averaged, and then compared to determine a dry to wet deposition ratio for each season and available land cover type. The results of the sorting and analysis of the wet and dry deposition network data are shown in Table 1 .
|
Estimating Watershed Retention and Export of Nitrogen Deposition in the Chesapeake Bay Watershed
The amount of nitrogen exported from a specified land cover type is highly variable. This variability can be attributed, in part, to other physical basin characteristics including the slope, underlying geology, and soil type. To account for the variability found in these features, a nitrogen export potential grid was created using ARC/INFO (Environmental Systems Research Institute, 2000) software. Regional information on soil hydrologic group (which accounts for slope, geology, soil type, and soil permeability) was assembled from the State Soils Geographic (STATSGO) database (United States Geological Survey, 1995). This information was converted to grid format, then combined with the land cover information to account for this variability (Fig. 3)
. This method of representing nutrient export variability has been commonly used in modeling efforts, including the Generalized Watershed Loading Function (GWLF), the Soil and Water Assessment Tool (SWAT), a Distributed Rainfall-Runoff Model (F2D), and others (DeBarry et al., 1999).
|
|
Export from each subbasin was determined on a cell-by-cell basis using ARC/INFO (Environmental Systems Research Institute, 2000). Seasonal nitrogen export from each grid cell was calculated as:
 |
All deposition falling directly on surface water was assumed to be transported directly from each subbasin. Using the export grids, the amount of nitrogen exported from a watershed was determined by summing the cell values contained within the basin delineation.
Once export loads from each subbasin were determined, transport factors were applied to estimate attenuation as the nutrients travel down the tributaries to the Chesapeake Bay. The transport factors, developed for use in the Chesapeake Bay Water Quality Model (USEPA, 1998), are dimensionless coefficients ranging theoretically from 0.0 to 1.0, with coefficients decreasing with increasing distance from the bay. Data describing the Chesapeake Bay Water Quality Model segments and corresponding hydrologic unit watersheds were obtained. Transport factors from each hydrologic unit watershed were estimated by averaging the Chesapeake Bay Water Quality Model transport factors used within each HUC.
Verification of Export Estimates from Selected Subbasins within the Chesapeake Bay Watershed
The validity of the model and model estimates was assessed by comparing model output with measured stream discharge and concentration data from 64 basins within the CBW, compiled by Langland et al. (1995)( 1998). All stream discharge data used in this analysis can be obtained from the United States Geological Survey (2000).
The measured amount of nitrogen exported from these watersheds was compared with the export predictions generated using the coefficients from the watershed nitrogen export layer and atmospheric deposition inputs discussed above. Since calculation of all point and nonpoint sources of nitrogen pollution was beyond the scope of this study, export estimates could not be verified exactly. However, the model export estimates were compared with stream discharge and concentration measurements to ensure that the model calculations were within an acceptable range (e.g., model export cannot be higher than measured stream export, and should be within the percentages obtained from the literature). In cases where modeled export exceeded measured stream export, regression techniques were used to assess the source of error. Overestimation of nitrogen export (kg modeled N export - kg measured N export) was used as the dependent variable. Predictor variables included the percent of each land cover type within the drainage basin, watershed size, season, watershed location, and seasonal precipitation amount.
Estimating the Contribution of Atmospheric Nitrogen Deposition to the Chesapeake Bay
Annual nitrate and ammonium nitrogen loads to the CBW from atmospheric sources were calculated for 1984 through 1996 by multiplying modeled nitrogen export from each of the hydrologic unit code watersheds by the appropriate transport factor. Since information on seasonal variations of in-stream nutrient uptake are not available, loading to the Chesapeake Bay was estimated on an annual basis. These estimated loads are reported in millions of kilograms. The estimated 1984 atmospheric nitrogen load was compared with the estimates prepared by Fisher et al. (1988), Fisher and Oppenheimer (1991), and Hinga et al. (1991). Total load comparisons with Tyler's (Tyler, 1988) work were not made because only nitrate nitrogen was accounted for in that study.
|
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
Estimates of Dry Deposition to the Chesapeake Bay Watershed
The contribution of dry deposition to the total atmospheric nitrogen load ranged from 26 to 40% based on the approach used in this study (Table 1). This is considerably less than previous estimates, due to the lower dry to wet deposition ratios determined using CASTNet dry deposition and NADP wet deposition network data. While these numbers are lower than the estimates used in other studies, they are consistent with the findings of other studies conducted in the region including Meyers and Sisterton (1989) and the USEPA (1995).
Using the ratios developed in this investigation, dry deposition to the CBW ranged from 41.9 to 60.1 million kg per year from 1984–1996 (Table 3). These numbers account for 23 to 48% of the total nitrate and ammonium nitrogen load to the Chesapeake Bay on an annual basis. Fluctuations in the percentage result from the calculation of dry to wet deposition ratios according to land cover and the spatial and temporal variability in wet deposition loads.
When these ratios are employed, total dry nitrogen deposition loading to the CBW is substantially lower than the estimates used in previous studies. Hinga et al. (1991), Fisher et al. (1988), and Fisher and Oppenheimer (1991) estimated that 104.5 and 110.84 million kg of nitrogen were deposited to the CBW via dry deposition in 1984. By comparison, the use of the precipitation model and the dry to wet deposition ratios lead to a dry deposition loading estimate of 55.9 million kg of nitrogen during 1984. Using the dry deposition estimate leads to a 23 and 25% decrease in the total loading estimates used by Hinga et al. (1991), Fisher et al. (1988), and Fisher and Oppenheimer (1991).
The approach discussed above is an improvement over estimating dry deposition by simply doubling wet deposition inputs. However, considerable uncertainty still exists in these estimates. Errors associated with dry deposition collection methods and additional physiographic features including elevation, slope, aspect, and proximity to pollution sources contribute to this uncertainty. In addition, the previously used 1:1 dry to wet deposition ratio was used to assess dry deposition to land cover types lacking dry deposition data, and there is no reason to assume that these ratios are correct. Based on the ratios that were obtained, the ratio of dry to wet deposition at other land cover types may be significantly less than the 1:1 ratio used here, leading to further dry deposition loading reductions.
Estimates of Total Deposition to the Chesapeake Bay Watershed
Annual nitrogen deposition to the CBW over the 13-yr period covered by this study ranged from 100.2 to 188.9 million kg with an average annual deposition rate of 151.3 million kg (Table 3). The 1984 estimate of total atmospheric nitrogen deposition to the watershed (171.2 million kg) is approximately 23, 27, and 18% less than the estimates made by Fisher et al. (1988), Fisher and Oppenheimer (1991), and Hinga et al. (1991), respectively. A comparison of 1984 wet and dry deposition indicates that most of the difference in total deposition between this study and those of Fisher et al. (1988), Fisher and Oppenheimer (1991), and Hinga et al. (1991) is the result of large discrepancies between dry deposition estimates.
The 1984 estimate of 171.2 million kg is 13% higher than the 13-yr average deposition to the CBW. This illustrates the potential for misinterpretation that may be introduced when only one year of atmospheric data is used to assess the fate of atmospheric deposition to a watershed. Using a single year of data with either relatively high or relatively low deposition rates may lead to an over- or underestimation of the average annual contribution of atmospheric nitrogen to the CBW.
Seasonal deposition to the CBW is highly variable both within years as well as within seasons (Table 3). The high degree of seasonal and annual variation in atmospheric deposition is due in part to seasonal and annual variability in precipitation volumes. Precipitation variability affects both the amount of nitrogen that is deposited and the hydrologic characteristics of the watershed that influence the transport of nitrogen to the bay. Seasonal fluctuations in atmospheric nitrogen deposition can also be attributed, in part, to fluctuations in emissions of nitrogen species to the atmosphere. The rate of nitrogen species emission is strongly dependant upon air temperature, wind velocities, fertilizer and manure spreading practices, automobile use, and farm animal diet and density (Asman et al., 1998).
Estimates of Watershed Export of Nitrogen Deposition to the Chesapeake Bay
The wide range in nitrogen export estimates to the bay derived in previous studies and the current analysis is due, in part, to the use of different watershed nitrogen retention and export coefficients (Table 4)
. The coefficients used by Tyler (1988) are much lower than the coefficients used in the other analyses, and lead to the smallest estimate of the percent contribution of atmospheric nitrogen to the Chesapeake Bay. The coefficients used by Hinga et al. (1991), Fisher and Oppenheimer (1991), and Fisher et al. (1988) were similar (within 10%) for all three studies.
|
To compare the relative effects of employing the coefficients used in previous studies and in the current analysis, export of 1984 deposition was calculated using each set of coefficients. The results of this analysis indicate that export is highest when the coefficients used in the previous studies are employed. Using these coefficients, it was estimated that 42 to 43 million kg of nitrogen was exported from the combination of forest, pasture, agricultural, and urban lands in 1984. Using the techniques developed in this analysis, an estimated 30 million kg of nitrogen was exported. Differences in export estimates from urban and agricultural lands were relatively small, with current export equaling 10 and 17% less, respectively. Major differences in the range of export from forested land resulted in the discrepancy seen in the total export estimates. Fisher et al. (1988), Fisher and Oppenheimer (1991), and Hinga et al. (1991) estimated that 20% of the nitrogen deposited to forested lands was exported. This yields an export load that is four times larger than the estimates based on the approach used in this analysis.
Again, using 1984 data as an example, the in-stream nitrogen attenuation coefficients adopted from the Chesapeake Bay Water Quality Model suggest that the amount of nitrogen exported from forested, agricultural, and urban lands will decrease 27, 21, and 13%, respectively, as the nitrogen load travels to the bay. These estimates of nitrogen attenuation are different from previous estimates due to the spatial variability of land cover and soil hydrologic group within the watershed and the use of attenuation coefficients, which decrease with increasing distance to the bay.
There is no way to determine whether the use of land cover and soil hydrologic group to assign nitrogen export coefficients and the incorporation of the spatially variable in-stream attenuation coefficients leads to an increase in model accuracy. However, by incorporating these variables into the model, variations in slope, soil type, soil thickness, the presence of fragipan, subbasin location, etc. are considered, all of which affect the amount of nitrogen delivered to the Chesapeake Bay.
Verification of Nitrogen Export Estimates from Selected Subbasins within the Chesapeake Bay Watershed
The cell-based modeling techniques employed in this analysis allow export from any delineated basin to be determined. In this case, the cell size was selected to be equal to the coarsest resolution of all the data sets incorporated into the model, or 422 m2 (approximately 18 ha). The cells were grouped to represent export from each of 64 gauged watersheds located within the Chesapeake Bay drainage basin. Each of the selected watersheds has long-term, seasonal nitrate and ammonium export data from years 1985 through 1996 (Langland et al., 1995, 1998).
The sorting and analysis of the gauged watershed database yielded 2140 seasonal nitrate and ammonium load estimates. Modeled export was compared with each of these seasonal measured export values. Of the 2140 cases where comparisons were made, modeled export (from atmospheric sources only) exceeded the measured export in 380 (18%) of the cases and was less than the measured export in the remaining 1760 cases. The data set was divided according to these differences, and least squares regression techniques were used to examine each of the three different cases (Fig. 4) . The plots of Regression Models 1 and 3 indicate that a transformation of the data may be useful. In these cases, data were transformed and the models were recalculated. However, the standard transformations we attempted did not greatly improve model results, so the original data values were used in the analysis.
|
Significant predictors of the difference between modeled and measured export are watershed area, seasonal precipitation amount, and the percent of the watershed covered by forested land. According to the land cover GIS dataset used in this analysis, forest cover makes up 57.4% of the total watershed area and approaches 100% in some subbasins. The predominance and high nutrient retention capabilities of forest land in the basin make the percent cover of forested land an important factor influencing nutrient export. Watershed area is significant since the larger, monitored watersheds have more diverse land uses and an increased number of nitrogen sources that were not incorporated into the model. Also, in-stream nitrogen removal processes were not accounted for at the subbasin scale and may have led to overestimation of export for larger watersheds. The transport of nitrogen species, particularly from atmospheric sources, is hydrologically driven. Precipitation not only delivers wet deposition to the watershed, it is also responsible for washing dry deposition off of the land surface, as well as transporting both wet and dry deposition through the watershed to the stream channel. The effect of storm intensity, storm duration, and soil moisture are not well represented in the model and may have contributed to the discrepancies between measured and modeled nitrogen export.
Regression Model 2 (Model Export Exceeds Measured Export)
The 380 cases in which modeled atmospheric export exceeded measured export were examined separately. In these cases it is clear that the model is overestimating nitrogen export. It is important to note that the model may overestimate the contribution of atmospheric nitrogen to stream loading in other cases, but it is not possible to identify these instances without accurately modeling all other nitrogen sources and processes in the basin.
Results from this regression model indicate that summer season, percent of the watershed covered by water, percent of the watershed covered by forest, watershed area, and seasonal precipitation amount were significant predictors of model error (Fig. 4). Model error increased during the summer season, indicating that the export coefficients assigned to the land cover–hydrologic soil group combinations may be too high, or that during summer only a small (e.g., riparian areas) part of the watershed is hydrologically active and thus contributing to nitrogen export. Lynch and Corbett (1989) found that this was the case for sulfate export from a small, forested watershed in central Pennsylvania.
An increase in the amount of surface water within a watershed increased model error. The National Land Cover Dataset (NLCD) used to determine land cover distribution has a resolution of 30 m2. At 30-m2 resolution, pixels classified as water most likely represent large water bodies such as major rivers and reservoirs. Previous studies have shown that surface water bodies are effective at removing sediments and nutrients from the water (Windolf et al., 1996; Jeppesen et al., 1998). Since the denitrifying effect of large water bodies was not accounted for in this model, the presence of water bodies in a drainage basin will increase model overestimation.
The percent of the basin covered by forest was negatively correlated with model overestimation. A very high portion of the atmospheric nitrogen that is deposited to forested land is retained by the ecosystem. This is reflected in the low export coefficients assigned to forested areas. Watersheds containing large tracts of forest cover export low amounts of nitrogen, which reduces model error.
Watershed area was positively correlated with model error. This indicates that in cases where model overestimation occurred, the magnitude of the overestimation was greater in the larger watersheds. The causes of this result are somewhat difficult to assess, since the model underestimates export from large watersheds in the vast majority of cases. Typically, larger watersheds contain multiple point sources, which discharge quantities of nitrogen far greater than the modeled atmospheric nitrogen discharge. However, in the cases investigated in this regression analysis, the largest of the watersheds contain very few point sources, relative to other basins with measured nitrogen export information. The positive correlation between watershed size and model overestimation may be due to interactions between differences in land use distribution as basin size increases, and export coefficients that are too high for several land cover classes. The smaller watersheds used in this study are largely forested, while the larger basins tend to have more diverse land use patterns. If some of the export coefficients used for agricultural, high intensity developed, and/or low intensity developed lands are too high, modeled nitrogen export from larger basins made up of these land cover types may be greater than the measured export.
Seasonal precipitation amount was negatively correlated with model error. This indicates that during periods of drought, the model overestimates export of nutrients. During excessively dry periods only a small fraction of the watershed is hydrologically active and contributing to nutrient export. Since the model does not take into account soil moisture conditions, in cases of extreme drought the model will still predict export in response to any amount of precipitation. In reality, small precipitation events occurring during dry periods can be entirely absorbed by the soil layer and may never affect stream discharge volumes or export.
Cases in which model export exceeded measured export during the summer season were investigated independently of the other error cases to determine if error predictors were different for summer than for other seasons. During the summer, percent of the watershed covered by water, percent of the watershed in planted and/or cultivated lands, watershed area, and seasonal precipitation amount were found to be significant predictors of model error (R2 = 74.4%, n = 232). The relationships between percent of the watershed covered by water, watershed area, seasonal precipitation and model overestimation are probably similar to the explanations given above. The positive relationship between percent of the watershed in planted and/or cultivated lands and model error indicates that the export coefficients assigned to planted and/or cultivated lands for the summer months may be too high.
Regression Model 3 (Modeled Export Less Than Measured Export)
Year, watershed size, seasonal precipitation amount, and the percent of the watershed covered by forest were significant predictors for the 1760 cases in which measured nitrogen export exceeded modeled nitrogen export (Fig. 4). Watershed size was negatively correlated with the difference between measured and modeled export, indicating that the difference will increase with increasing watershed size. Larger watersheds contain many more sources of pollutants that are not accounted for in the model, so it is not surprising that as watershed size increases, measured export and the difference between measured and modeled export tend to increase. Seasonal precipitation amount and the percent of the watershed covered by forest are both positively correlated with the difference between the measured and modeled nitrogen export. This indicates that as precipitation amounts increase and the watershed becomes more hydrologically active, atmospheric deposition becomes increasingly important relative to other nitrogen sources. When this occurs, the discrepancy between the measured and modeled export values becomes less. The percent of the watershed covered by forest is significant for a different reason than in the other cases. Generally, watersheds that are partially or totally forested do not have as many sources of pollution within their boundaries as more developed watersheds. Therefore, as the percent of a watershed covered by forest increases, other sources of pollution decrease; consequently, the discrepancy between measured and modeled nitrogen export tends to decrease as well.
Model Estimates of the Contribution of Atmospheric Nitrogen Deposition to the Chesapeake Bay
Based on this analysis, the contribution of atmospheric nitrogen to the Chesapeake Bay ranged from 34.2 to 50.4 million kg for years 1984 through 1996 (Table 5)
. The average annual nitrogen export over the 13-yr period of record was 41.7 million kg. The 1984 annual export of 46.3 million kg is approximately 15% less than the amount obtained by Fisher et al. (1988) and Fisher and Oppenheimer (1991). Tyler's (1988) estimate of loading is considerably less than 46.3 million kg, since she did not include ammonium deposition in her analysis.
|
Dividing the annual atmospheric nitrogen loads into seasonal contributions yields information that should be considered in the design of management plans. On average, fall, spring, summer, and winter contribute 16, 39, 28, and 17% of the total annual nitrate and ammonium nitrogen load to the Chesapeake Bay, respectively. While there is great inter-annual variability in seasonal loading, the highest seasonal loading typically occurs during the spring and summer months. Differences in loading between these seasons can be attributed to higher export coefficients used during the spring season and higher flows. On average, the contribution to Bay loading is similar for fall and winter seasons. Winter nitrogen export coefficients are higher than the coefficients assigned for fall, and this difference is reflected in the relatively high winter loads seen in nine of the thirteen years that were studied. However, very high fall deposition rates in the other four years caused the two seasons to appear similar when all thirteen years were averaged.
Seasonal export trends based on this investigation may be helpful in designing strategies for decreasing nitrogen export to the bay. Large spring and summer loads have a compounding effect on eutrophication problems since they can be immediately utilized by phytoplankton and lead to algae blooms that have affected the bay in recent years. Decreasing spring and summer nitrogen loads through employment of best management practices and seasonal restrictions on combustion emissions may be more effective at controlling eutrophication in the bay than attempting to control nitrogen loads throughout the entire year.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
The spatial resolution of wet deposition was greatly enhanced through the use of a high-resolution precipitation model. Annual deposition to the CBW was found to be highly variable from 1984 through 1996, fluctuating by a factor of three during this period. Assuming that the addition of nitrogen from fertilizer, animal waste, and point sources are relatively constant from year to year, annual variability in nitrogen export to the CBW may be influenced more strongly by fluctuations in atmospheric deposition than any other source of nitrogen. Therefore, it is unreasonable to assume that the previous single-year estimated (Tyler, 1988; Fisher et al., 1988; Fisher and Oppenheimer, 1991; Hinga et al., 1991) contribution of atmospheric nitrogen could be applied for any given year.
Estimates from this study indicate that dry atmospheric nitrate and ammonium nitrogen deposition represents between 23 and 48% of the total atmospheric nitrogen deposition to the bay. While these estimates are low with respect to previous estimates, they agree with the findings of other wet and dry deposition studies conducted in the northeastern USA (Meyers and Sisterton, 1989; USEPA, 1995, 1999). Since dry deposition is assumed to equal a fixed percentage of wet deposition, annual dry deposition to the CBW is also highly variable, fluctuating by a factor of 1.4 from 1984 through 1996.
On a seasonal basis, 17, 39, 28, and 16% of the total annual nitrogen export are input to the Chesapeake Bay from atmospheric sources occur during the winter, spring, summer, and fall, respectively. Excess nitrogen loading has had a wide range of negative effects on the Chesapeake Bay ecosystem, many of which are seasonally dependant. Therefore, the timing of atmospheric nitrogen inputs and export should be considered when designing strategies aimed at reducing the input of nutrients into the bay.
The determination of a percent contribution of atmospheric nitrogen loading to the bay is somewhat unrealistic since all sources of pollution were not accounted for in this study. However, using the atmospheric deposition inputs derived from this analysis and the estimates of other nitrogen sources derived by Fisher et al. (1988), Fisher and Oppenheimer (1991), and Hinga et al. (1991), atmospheric nitrogen sources account for 30% of the total input of nitrogen to the watershed during 1984. While this estimate is slightly lower than the estimates made by Fisher et al. (1988), Fisher and Oppenheimer (1991), and Hinga et al. (1991), comparisons of measured and modeled export done here indicate that the atmospheric nitrogen contribution may still be too high. The reasons for this overestimation appear to be related to the influence of season, timing and volume of precipitation, and watershed characteristics including size, land cover sequencing, and the presence or absence of water bodies. In addition, nitrogen deposition was higher than average in 1984 according to deposition model results. Using the deposition estimates obtained for 1992 (a year of relatively low atmospheric nitrogen deposition to the CBW) and the point source, fertilizer and animal waste estimates used by Fisher et al. (1988), Fisher and Oppenheimer (1991), and Hinga et al. (1991), the percent contribution of atmospheric sources to Chesapeake Bay nitrogen loading would be 20%.
Additional research is needed to investigate methods of improving the accuracy and temporal resolution of both atmospheric deposition (wet and dry) and the export of atmospherically deposited nitrogen species from all land covers. Data that accurately describe both point and nonpoint nitrogen pollution inputs are currently available for many areas including the CBW. Geographic information systems technology can be very useful in designing, tailoring, and editing these pollution input data sets. However, combining geographic information systems input information with a dynamic watershed model that incorporates many of the watershed processes difficult to model using static relationships would improve the capabilities of the modeling effort. This is particularly true with respect to watershed retention coefficients that are assumed to be constant for all seasons for all portions of a given watershed.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS |
