Published in J. Environ. Qual. 32:2026-2032 (2003).
© ASA, CSSA, SSSA
677 S. Segoe Rd., Madison, WI 53711 USA
TECHNICAL REPORTS
Ground Water Quality
Modeling Flow and Nitrate Fate at Catchment Scale in Brittany (France)
Céline Conan*,a,
Fayçal Bouraouia,
Nadine Turpinb,
Ghislain de Marsilyc and
Giovanni Bidoglioa
a Joint Research Centre of the European Commission, Inst. for Environment and Sustainability, Soil and Waste Unit, TP 460, 21020 Ispra (VA), Italy
b Cemagref Rennes, URE Gere, 17 avenue de Cucillé, 35044 Rennes Cedex, France
c Laboratory of Applied Geology, Univ. of Paris 6, T.26-0 E.5 case 123, 4, place Jussieu, 75252 Paris Cedex 05, France
* Corresponding author (conan{at}eau-rhin-meuse.fr).
Received for publication September 19, 2002.
 |
ABSTRACT
|
|---|
In the intensive pig-farming (Sus scrofa) area of Brittany (western France), many surface and subsurface water resources are contaminated by nitrate (NO3) with concentrations that chronically exceed the European Community 50 mg L-1 drinking standard. To ensure sustainable water supply, the fate of NO3 must be considered in both surface water and ground water. The fate of N was investigated in a Britain catchment, the Coët-Dan watershed, with an integrated management tool: the hydrological SWAT model coupled with the ground water model MODFLOW, and its companion contaminant and solute transport model MT3DMS. The model was validated with respect to water quantity during a 6-yr period and for the NO3 concentration during a 44-mo period, at two gauging stations in the catchment. The coupled models reproduced accurately the measurements. At the basin outlet, the Nash-Sutcliffe coefficients were 0.88 for monthly flow for the entire period and 0.87 for monthly N load. Alternative scenarios were simulated and showed potential benefits of decreasing manure application from 210 to 170 kg N ha-1 as required by the European Commission Nitrates Directive.
Abbreviations: FIA, flow injection analysis • IFEN, French Institute for the Environment • NVZ, nitrate vulnerable zones
|
INTRODUCTION
|
|---|
ALTHOUGH N IS ESSENTIAL for healthy plant and animal populations, high concentrations of this nutrient can degrade water quality. Excessive concentrations of nitrate (NO3) (the most common form of N dissolved in streams and ground water) in surface water can stimulate the growth of algae and nuisance organisms. A significant part of the total N load to streams is carried by ground water that discharges to streams. This is particularly true for the Brittany region in France where the base flow is a dominant component of the stream flow.
To reduce and prevent water pollution caused or induced by NO3 from agricultural sources, the Nitrates Directive (European Community, 1991) requires the Member States to designate nitrate vulnerable zones (NVZ), i.e., areas where surface waters contain an excessive concentration of NO3, where ground water NO3 levels are likely to, or already exceed 50 mg L-1, or where waters are eutrophic. The entire Brittany region has been designated as NVZ, implying that farming practices must be changed to reduce NO3 losses in leaching and runoff. By implementing Codes of Good Agricultural Practice and Action Programmes, Member States were required to limit the amount of manure N applied to the land (including by the animal themselves) to 210 kg N ha-1 yr-1 by 1999 for each farm or livestock unit. In 2003, this value will be reduced to 170 kg N ha-1 yr-1.
Furthermore, as pressures on water resources increase in Europe (Scheidleder et al., 1999) and in Brittany in particular (IFEN, 1999), the Water Framework Directive (European Community, 2000) attempts to protect and improve water quality and resources across the European Union. An analysis of pressures and impacts on water resources must be completed by the end of 2004. This created the need to develop integrated predictive tools for evaluating the impact of various land-use and climate scenarios on water quantity and quality at basin scale.
A combined model based on SWAT and MODFLOW was originally developed to address water management issues concerning how ground water pumping for irrigation affects stream flows in two basins in Kansas (Perkins and Sophocleous, 1999; Sophocleous et al., 1999). This approach was expanded to simulate NO3 concentration in surface water and the fate of leached nitrate in ground water with the use of a reaction/transport model. A simple model using MODFLOW and MT3D (Zheng, 1990) was developed on a catchment in Brittany, to investigate the impact of NO3 leaching on NO3 concentration in surface water (Molénat and Gascuel-Odoux, 2002). Both uniform recharge amount and NO3 concentration of the recharge were input to the models. The methodology presented here is based on modeling of the general hydrology of the region and the fate of N in the unsaturated zone using SWAT (Arnold et al., 1998), ground water flow using MODFLOW (Harbaugh and McDonald, 1996), and the fate of leached NO3 from the unsaturated zone into the aquifer system using the MT3DMS (Zheng and Wang, 1999).
The objectives of this study were to validate this integrated tool for the prediction of water quantity and quality in the Coët-Dan watershed, localized in Brittany, a heavily fertilized region. The sensitivity of the model was also evaluated by testing alternative management scenarios.
|
MATERIALS AND METHODS
|
|---|
Study Area
The Coët-Dan watershed covers an area of 12 km2 in the middle of Brittany, west of France (Fig. 1)
. The landscape is gently sloping with elevations that range from 67 to 131 m above sea level. The northern upper part of the catchment is characterized by an open-land plateau. The southern part is characterized by steeper slopes (5%). The watershed presents an oblong shape with the axis of the north-northwest flowing river.
The climate of the region is typically oceanic, with an average annual rainfall of 950 mm for the study period (1994–1999) and an average annual temperature of 11.5°C. Rainfall amounts >20 mm d-1 are exceptional and represent 2% of rainfall events. Rain intensity is <4 mm h-1 for 80% of rainfall events (Carluer, 1998). The Coët-Dan River flows from north to south for 7 km. Dry in summer, it presents a mean annual flow of 345 mm. The area close to the stream is characterized by silty hydromorphic soils; elsewhere soils are silty loam. Two aquifer formations can be distinguished: weathered rock underlain by fractured schists with small inclusion of sandstone and containing pyrite minerals (Martelat et al., 1997). Since 1994, about 90% of the catchment is used for agriculture purposes, dominated by cereal production [mostly wheat (Triticum aestivum L.) and corn (Zea mays L.)] and pasture. The watershed is characterized by heavy fertilization that consists of applied swine-fresh manures and mineral fertilizer. Average N application on the agricultural land was around 315 kg ha-1 for 1994 (Bouraoui et al., 1999) and the annual amount of manure represented 230 kg N ha-1 (Bordenave and Merceron, 1999).
The two gauging stations of the watershed (Fig. 2)
were equipped with a broadcrested weir and a limnimetric station to measure stream flows, which were recorded on a data logger. An automatic sampler (ISCO) was set up to take 24 samples of 0.5 L during 24 h once a threshold stream depth was exceeded. The water samples were kept in a refrigerated room before analysis and were analyzed for NO3 content within 3 d following collection. Filtered and diluted water is sent to an automated ion QuikChem 8000 (Lachat Instruments) analyzer using a peristaltic pump, based on the technology of flow injection analysis (FIA). Nitrate is determined with the Method 10-107-04-1-A, Lachat Instruments (USEPA Ref 353.2). Nitrate is quantitatively reduced to nitrite by passage of the sample through a copperized cadmium column. The nitrite (reduced NO3 plus original nitrite) is then determined by diazotizing with sulfanilamide followed by coupling with N-(1-naphthyl)ethylenediamine dihydrochloride. The resulting water soluble dye has a magenta color, which is read at 520 nm. Additional details about the analytical methodology and the experimental setup are given by Cann (1993).
Model Principles
The SWAT–MODFLOW–MT3DMS linkage used SWAT v.99.2 (Arnold et al., 1998), MODFLOW-96 v.3.3 (Harbaugh and McDonald, 1996) and MT3DMS (Zheng and Wang, 1999). The integrated model consisted of SWAT for water and N fate in the unsaturated zone; MODFLOW as ground water flow using SWAT predicted recharge as time-variant input; and MT3DMS for assessing the fate of NO3 leached from the topsoil as predicted by SWAT.
SWAT Model
The SWAT (Soil and Water Assessment Tool) model menu is a semidistributed watershed model with a GIS interface (DiLuzio et al., 2002) that outlines the subbasins and stream networks from a digital elevation model and calculates daily water balances from meteorological, soil, and land-use data. SWAT simulates each subbasin separately according to the soil water budget equation taking into account daily amounts of precipitation, runoff, riverbed transmission losses, percolation from the soil profile, and evapotranspiration. Modifications were made to better simulate variable surface area of the saturated runoff contributing zones by adding the direct transformation of rain on saturated zones to surface runoff, and lateral subsurface flow according to Darcy's law (Conan et al., 2002).
SWAT simulates the movement and transformation of N in the catchment. Basic processes simulated are mineralization, denitrification, volatilization, and plant uptake. The simulation of N mineralization considers two different organic pools: fresh organic N pool, associated with crop residue and microbial biomass; and the stable organic N pool, associated with the soil humus. Mineralization from the fresh organic N pool is controlled by the C/N and C/P ratios. Organic N associated with humus is divided into active and stable pools, which are in equilibrium. Only the active pool of organic N is subjected to mineralization. Mineralization of N is adjusted according to soil moisture and temperature conditions. SWAT considers the addition of organic and/or inorganic fertilizers, and atmospheric deposition for N. Additional details about the N cycle simulated by SWAT can be found in Arnold et al. (1998). Plant uptake of N is estimated using a supply and demand approach. The N demand is computed on a daily basis, based on the optimal N crop concentration for each growth stage. Nitrogen can be lost from the watershed in particulate or dissolved forms.
MODFLOW Model
MODFLOW is a fully distributed model that calculates ground water flow from aquifer characteristics. It solves the three-dimensional ground water flow equation using finite-difference approximations. The finite-difference procedure requires that the aquifer be divided into cells, where the aquifer properties are assumed to be uniform. The unknown head in each cell is calculated at a point or node at the center of the cell. MODFLOW is designed to simulate aquifer systems in which saturated-flow conditions exist, Darcy's Law applies, the density of ground water is constant, and the principal directions of horizontal hydraulic conductivity or transmissivity do not vary within the system.
The hydrologic terms simulated by SWAT for each subbasin are transformed from cumulative volumes per unit area (L) to monthly flow rates (L3/T) in the system of units specified for MODFLOW's simulation (Perkins and Sophocleous, 2000). Grid cells of MODFLOW are associated with geographical extent of subbasins simulated by SWAT. Hydrologic components simulated by SWAT for each subbasin are combined to specify fluxes for MODFLOW's solution in each time step and distributed to each corresponding grid cell (Perkins and Sophocleous, 2000). These fluxes include ground water recharge, tributary inflow, and a maximum rate for evaporation from shallow ground water.
MT3DMS Model
MT3DMS is a three-dimensional ground water contaminant and solute transport model that can simulate advection, dispersion, dual-domain mass transfer, and chemical reactions of dissolved constituents in ground water. MT3DMS uses the output head and cell-by-cell flow data computed by MODFLOW to establish the ground water flow field. The partial differential equation describing the fate and transport of contaminants of species k in 3-D, transient ground water flow systems can be written as follows (Zheng and Wang, 1999):
 | [1] |
where, 
is porosity of the subsurface medium, dimensionless; Ck is dissolved concentration of species k (M L-3); t is time (T); xi,j is distance along the respective Cartesian coordinate axis (L); Dij is hydrodynamic dispersion and diffusion coefficient tensor (L2 T-1); vi is seepage or linear pore water velocity (L T-1); it is related to the specific discharge or Darcy flux through the relationship, vi = qi/; qs is volumetric flow rate per unit volume of aquifer representing fluid sources (positive) and sinks (negative) (T-1); Cks is concentration of the source or sink flux for species k (M L-3); 
Rn is chemical reaction term (M L-3 T-1).
The concentrations of NO3 percolation simulated by SWAT for each subbasin and each time step are distributed over the grid, according to the association of subbasin domains with grid cells, and used as monthly recharge concentration terms for MT3DMS over the simulation period.
Model Calibrations
The coupled models SWAT–MODFLOW–MT3DMS were applied to the Coët-Dan watershed for the period 1994–1999. The modeling consisted of a three-step approach. The available data included measured daily stream flows from the outlet gauging station for the entire simulation period and piezometric levels from three boreholes from January 1994 to May 1997 for the water quantity, and measured monthly NO3 concentrations at the outlet of the catchment from January 1994 to August 1997 for the water quality. Boreholes and gauging station locations are shown in Fig. 2. The calibration was performed using the measured time series for 1995 and 1996. The year 1994 was not considered in the calibration and was used as a warm up period for the models. Model parameters were adjusted by trial-and-error to reduce the differences between simulated and measured values using both efficiency criterion (the Nash-Sutcliffe coefficient) and graphical considerations. At the gauging station of Villeneuve (south-east of the catchment), measured discharges were available from January 1994 to February 1999 and were used to ensure the spatial validation.
First, the digital elevation model, with a cell-size of 5 m, was used to discretize the watershed into 44 subbasins. The local dominant soil/land-use scheme was chosen to characterize each subbasin. Daily precipitation and temperature data were available from one meteorological station. The fertilizer applications consisted in mineral fertilizers and swine-fresh manures on corn and wheat, and beef-fresh manure on pastures where cattle (Bos sp.) grazed. The total annual amounts of applied fertilizer were 282 kg N ha-1 on the agricultural subbasins. The parameters of the SWAT model adjusted in calibration and their ranges are briefly listed in Table 1.
Within MODFLOW, the aquifers were represented as two layers, discretized into a regularly spaced grid of 115 rows and 81 columns. Each row and column of the model represented a 50-m wide strip of aquifer. Ground water limits for the model corresponded to those of the surface water basin. These boundaries were designated as no-flow cells. Two isotropic layers represented the unconfined weathered rock and the confined schists. The hydraulic conductivity of the weathered rock used in the model ranged from 10-5 to 10-4 m s-1, the vertical conductance used in the model was 10-3 s-1, and the specific yield used in the model ranged from 0.01 to 0.15. The transmissivity of the schists used in the model ranged from 6.10-6 to 2.10-5 m2 s-1, and the specific storage ranged from 10-3 to 10-4. All theses calibrated values are within the range of literature values given for this watershed (Molénat, 1999). Recharge was distributed according to SWAT simulation outputs for each month. Stream–aquifer interaction was simulated using a stream-routing package for MODFLOW. Stream flows assigned to the upstream reach of each stream segment were values input from monthly SWAT simulation. The model was initially calibrated under steady state conditions by adjusting input data including hydraulic conductivity, and vertical conductance. Calibration under transient conditions also included adjustment of specific yield and evaporation.
Finally, the MT3DMS model was based on the grid and results of MODFLOW. Distributed NO3 concentrations calculated by SWAT were input for each stress period (month). Since NO-3 does not form insoluble minerals, neither precipitates nor is adsorbed, the only natural way of removing nitrate from aquifers is by reduction (Chapelle, 1993). The concentration and availability of O2, organic matter and/or pyrite (FeS2) determine the latter process. Nitrate reduction mostly by pyrite in the dissolved phase was modeled by a first-order irreversible reaction:
 | [2] |
where C0 is initial NO3 concentration (M L-3), Ct is day t NO3 concentration (M L-3), and k is first-order rate constant (T-1).
The first-order rate constant was about 10-3 d-1 for the weathered rocks and 0.06 d-1 for the schists. This coefficient for schists, corresponding to a half-life time of 11 d, was in agreement with the half-life time of 8 d calculated by Pauwels et al. (1998). Initial NO3 concentration was calibrated at 27.10-3 kg m-3.
For the scenario analysis, all calibrated physical and hydrogeological parameters of the models were kept the same as in the baseline case presented in the previous section, i.e., with an even application of fertilizer on the agricultural subcatchments. Two scenarios, concerning the Codes of Good Agricultural Practice of the Nitrates Directive, were tested on the Coët-Dan catchment. The annual amount of applied organic fertilizer was decreased from the actual amount to 210 to 170 kg N ha-1. Another scenario was dedicated to simulating the time-effect of changing practices on the N cycle. For the period 1994–1998, measured climate data were used. From 1999 to 2013, precipitation were assumed to be those of the period 1972–1986, temperatures were the 1994–1998 temperatures repeated three times according to available data.
|
RESULTS AND DISCUSSION
|
|---|
Figure 3
shows the measured and simulated daily discharges at the outlet using only the SWAT model with its simple exponential ground water lag. The range of data is 0 to 2.5 m3 s-1. Predicted peak flow rate and daily flow compared favorably with the measured values. The mean measured discharge of the Coët-Dan River was 0.133 m3 s-1 for 1994–1999 and the simulated value was 0.140 m3 s-1. The Nash-Sutcliffe coefficients for daily flow at the outlet were 0.79 for the calibration period (1995–1996), 0.42 for the validation period (1994 and 1997–1999), and 0.66 for the entire period, respectively. The efficiency for the calibration period is higher than that for the validation period because the rainfall pattern is different for these periods. During the calibration, the year 1995 was particularly wet (Fig. 3), while the validation period is much drier. In most cases, the order of magnitude of the measured and predicted flood peaks and recession curves were in good agreement. A statistical test for comparing the paired mean monthly flow (measured and observed) was performed to detect any systematic bias. The only significant differences (5% level) were detected for June and December. The coupled models tend to underpredict the flow during the month of June and overpredict it in December. Possible reasons for these differences are the rough management description used in the model and also problems with the simulation of the behavior of fractured media. When using the coupled model SWAT-MODFLOW the monthly predicted discharges were slightly improved: the Nash-Sutcliffe coefficients for monthly flow at the outlet for the entire period were 0.87 with the SWAT model and 0.88 with the coupled model SWAT–MODFLOW. The Nash-Sutcliffe coefficients for monthly flow at the Villeneuve gauging station were 0.83 with SWAT and 0.84 with the coupled model (Fig. 4)
. The representation of the piezometric levels was adequate with indexes of agreement for the three boreholes of 0.41, 0.49, and 0.68. However, some discrepancies were noted during the simulation, probably due to the difficulty of representing the saturated flow in fractured media (Fig. 5)
.
View larger version (19K):
[in this window]
[in a new window]
|
Fig. 5. Measured (closed symbols) and simulated (open symbols) piezometric levels (m) between January 1994 and June 1997 for the three different wells.
|
|
The SWAT simulated NO3 yield in the surface water was always underestimated, even if it had the same trend as the measured one (Fig. 6)
. This was expected because the contribution of base flow for nutrient load was not originally taken into account by SWAT v.99.2: this version did not track NO3 through the ground water system. On the other hand, the coupled models estimation for NO3, which represented the base flow, was very accurate. The Nash-Sutcliffe coefficient for monthly N load for the 44-mo study period was 0.87 at the outlet of the catchment, and 0.60 at the Villeneuve gauging station (Fig. 7)
, respectively. The main difference between the SWAT and coupled models estimation for NO3 was the ground water contribution. The models also well represented the seasonality of NO3 losses and the higher loads occurring from November to March–April. Thus, the ground water transport and NO3 reduction were accurately modeled, which was facilitated by the good modeling of water quantity (for both surface water and ground water).
View larger version (23K):
[in this window]
[in a new window]
|
Fig. 6. Measured, simulated by simple SWAT, and simulated by the coupled models SWAT–MODFLOW–MT3DMS monthly NO3 load (103 kg) in the surface water at the outlet.
|
|
View larger version (22K):
[in this window]
[in a new window]
|
Fig. 7. Measured, simulated by simple SWAT, and simulated by the coupled models SWAT–MODFLOW–MT3DMS monthly NO3 load (103 kg) in the surface water at Villeneuve.
|
|
The average annual N balance over the entire watershed was 100 kg N ha-1 of humus mineralization, a fertilizer application of 245 kg N ha-1 (equivalent to an application of 282 kg N ha-1 on agricultural subbasins), with a plant uptake about 130 kg N ha-1 (equivalent to a mean plant uptake of 150 kg N ha-1 for agricultural subbasins), 165 kg N ha-1 percolated with an amount of 100 kg N ha-1 reduced in the ground water. The estimated ground water denitrification by MT3DMS represented about 65% of the NO3 that percolated to the aquifers, and the denitrification rate was 5.39 mmol NO3 L-1 yr-1 during the period of simulation. These results were consistent with both those found during a small field tracer test on the Coët-Dan catchment (Pauwels et al., 1998) and those concerning a delta deposit aquifer in North Dakota (Korom et al., 2001). These last authors estimated with tracer tests that 61% of the denitrification could be explained by pyrite oxidation with a rate up to 5.47 mmol NO3 L-1 yr-1. Furthermore, these rates were comparable in particular with the study of Tesoriero et al. (2000), which assumed pyrite oxidation as a primary source of electrons, and in general with those from other field-based studies in sand and gravel aquifers reported by Puckett and Cowdery (2002). The annual average amount of NO3 reduction in the ground water predicted by the models was 450 Mg yr-1. Under the hypothesis of a complete denitrification (Eq. [3]), 3 mol of NO-3 are consumed by 1 mol of FeS2:
 | [3] |
Thus, the annual amount of oxidized pyrite due to denitrification would be 290 Mg yr-1 (according to the molar masses of NO3 and pyrite). The pyrite content is very small in the schists and varies between 5 and 40
, according to Pauwels et al. (1999). For a bulk density of 2600 kg m-3 and a pyrite content of 5, the volume of schists containing 290 Mg of pyrite would be 22 300 m3. As the surface area of the catchment is 12 km2, if only the first meter of schists can reduce NO3 by pyrite oxidation, then this process could be sustained for 500 yr, under the stated hypothesis, with the same NO3 pressure as during the 6 yr of simulation and a homogeneous distribution of the pyrite among the schists and the flow path. By assuming the upper 25 m of the schists and a mean pyrite content of only 3, Pauwels et al. (2000) found that the amount of pyrite corresponds to >4000 times the annual NO3 excess for 1994. This result agreed with the simulation of coupled models by taking the same amount of pyrite and half of the annual excess that leached to the ground water. Variations in spatial directions (geographical and with depth) of pyrite content were detected (Pauwels et al., 2000). To establish the sustainability of this process, detailed measurements of the pyrite content would be required. However, the schist contribution represents only 25% of the base flow (Pinault et al., 2001); thus, high NO3 concentration can still be observed in the Coët-Dan River.
Scenarios
The plant uptake for both scenarios concerning the Codes of Good Agricultural Practice of the Nitrates Directive (reduction of organic fertilization to 210 and 179 kg N ha-1) did not change from the baseline, about 130 kg N ha-1 yr-1 for the entire catchment (143 kg N ha-1 for the agricultural part of the catchment). According to the coupled models a reduction of fertilization would reduce the amount of N that percolates to the ground water up to 25 kg N ha-1 for the subbasins dominated by cereal production (Fig. 8)
. The NO3 concentration in the ground water near the outlet would decline by up to 15 and 22 mg L-1 between the present state and the applications of 210 and 170 kg N ha-1 yr-1, respectively, as prescribed by the Nitrates Directive (Fig. 9)
. These two scenarios reproduced ideal cases: they represented a sudden and drastic change in agricultural practices and N-soil content. In fact, N-soil content or structural excess would not change so suddenly. Thus, a last scenario was dedicated to simulating the time-effect on the N cycle. The beneficial impact of the application of the Nitrates Directive can be observed with time (Fig. 10)
. Nitrate concentration in ground water is higher under the present scenario of application of organic fertilizer (282 kg N ha-1 yr-1) than for the application according to the Nitrates Directive. Substantial reduction in NO3 concentration in ground water can be observed after 2004 when the amount of applied manure does not exceed 170 kg N ha-1 yr-1. However, even if some improvements regarding the water quality can be noted, NO3 concentrations are still in the order of 50 mg L-1.
View larger version (29K):
[in this window]
[in a new window]
|
Fig. 8. Nitrogen percolation to ground water for current application and two scenarios concerning applications of 210 and 170 kg N ha-1 yr-1 as prescribed by the Nitrates Directive.
|
|
View larger version (22K):
[in this window]
[in a new window]
|
Fig. 9. Nitrate concentration in ground water near the outlet for present applications of 282 kg N ha-1 yr-1 and for scenarios as prescribed by the Nitrates Directive (210 and 170 kg N ha-1 yr-1).
|
|
View larger version (24K):
[in this window]
[in a new window]
|
Fig. 10. Nitrate concentration in ground water near the outlet for present applications of 282 kg N ha-1 yr-1 during a 20-yr period and for the scenario as prescribed by the Nitrates Directive (amount of manure would be 210 kg N ha-1 yr-1 by 1999, and 170 kg N ha-1 by 2003).
|
|
|
SUMMARY AND CONCLUSION
|
|---|
The proposed methodology for modeling the general hydrology of the region and the fate of N using the coupled models SWAT–MODFLOW–MT3DMS, for the Coët-Dan catchment, was successful. In particular the simulated water quantity and NO3 yield presented very good Nash-Sutcliffe coefficients (0.68 for the daily water discharge, and 0.88 for the monthly NO3 load). Hence, the lack of ground water NO3 representation in the SWAT model could be addressed by the use of a ground water model using MODFLOW; and the fate of leached nitrate from the unsaturated zone into the aquifer system using MT3DMS. A major advantage of the coupled system is that it takes into account denitrification that represents in this catchment up to 60% loss of the leached N. The ground water models had the advantage of being fully distributed. Manure applications as required by the Nitrates Directive reduced the N percolation to the ground water without affecting the plant growth. It was shown that the integrated SWAT–MODFLOW–MT3DMS models constitute a valuable management tool that can be used to assess the impact of different policies on water quality.
|
REFERENCES
|
|---|
- Arnold, J.G., R. Srinivasan, R.S. Muttiah, and J.R. Williams. 1998. Large area hydrologic modeling and assessment: I. Model development. J. Am. Water Resour. Assoc. 34:73–89.
- Bordenave, P., and M. Merceron. 1999. Présentation des bassins versants ateliers (Coët-Dan et Kerharo) et de la baie de Douarnenez. p.107–124. In Pollutions diffuses: Du bassin versant au littoral. Ifremer Centre de Brest, Saint-Brieuc, Ploufragan, France.
- Bouraoui, F., N. Turpin, and P. Boerlen. 1999. Trend analysis of nutrient concentrations and loads in surface water in an intensively fertilized watershed. J. Environ. Qual. 28:1878–1885.[Abstract/Free Full Text]
- Cann, C. 1993. Suivi de la qualité de l'eau. Etude menée sur le bassin versant expérimental du Coët-Dan. CEMAGREF, Rennes, France.
- Carluer, N. 1998. Vers une modélisation hydrologique adaptée à l'évaluation des pollutions diffuses: Prise en compte du réseau anthropique. Application au bassin versant du Naizin (Morbihan). Ph.D. diss. Univ. of Pierre et Marie Curie, Paris, France.
- Chapelle, F.H. 1993. Ground-water microbiology and geochemistry. John Wiley & Sons, New York.
- Conan, C., F. Bouraoui, G. de Marsily, and G. Bidoglio. 2002. A modelling of the water quality in the Coët-Dan catchment (France). p. 222–226. In G.H. Schmitz (ed.) Water Resources and Environment Research. Proc., ICWRER, Dresden, Germany. 22–25 July 2002. Eigenverlag des Forums für Abfallwirtshaft und Atlasten e. V., Pirna.
- DiLuzio, M., R. Srinivasan, J.G. Arnold, and S.L. Neitsch. 2002. ArcView Interface for SWAT2000 User's Manual GSWRL Rep. 02–03, BRC Rep. 02–07. Rep. TR-193y. Texas Water Resources Inst., College Station, TX.
- European Community. 1991. Council directive 91/676/EEC concerning the protection of waters against pollution caused by nitrates from agricultural sources. Official Journal of the European Communities L 375. Office for Official Publications of the European Communities, Luxembourg.
- European Community. 2000. Directive 2000/60/EC of the European Parliament and of the council, establishing a framework for community action in the field of water policy. Official Journal of the European Communities L 327. Office for Official Publications of the European Communities, Luxembourg.
- Harbaugh, A.W., and M.G. McDonald. 1996. User's documentation for MODFLOW-96, an update to the U.S. Geological Survey modular finite-difference ground-water model U.S. Geol. Surv. Open-File Rep. 96–485.
- IFEN. 1999. The environment in France. La Découverte, Paris.
- Korom, S.F., A.E. Kammer, A.J. Schlag, and P.A. Skubinna. 2001. In-situ study of denitrification in the Elk Valley Aquifer, North Dakota. Eos Trans. AGU 82:S168 (abstr.).
- Martelat, A., J.-C. Foucher, P. Lachassagne, and H. Pauwels. 1997. Processus de dénitrification au sein d'un aquifère de socle (schistes à pyrite): Caractérisation du fonctionnement hydrogéochimique à l'échelle locale, dans le cadre d'un petit bassin versant breton. p. 29–36. In T. Pointet (ed.) Hard rock hydrosystems. Issue 241. Int. Assoc. of Hydrological Sciences, Rabat, Morocco.
- Molénat, J. 1999. Rôle de la nappe sur les transferts d'eau et de nitrate dans un bassin versant agricole. Etude expérimentale et modélisation. Ph.D. diss. Univ. of Rennes 1, Rennes, France.
- Molénat, J., and C. Gascuel-Odoux. 2002. Modelling flow and nitrate transport in groundwater for the prediction of water travel times and of consequences of land use evolution on water quality. Hydrol. Proc. 16:479–492.
- Pauwels, H., J.C. Foucher, and W. Kloppmann. 2000. Denitrification and mixing in a schist aquifer: Influence on water chemistry and isotopes. Chem. Geol. 168:307–324.
- Pauwels, H., W. Kloppmann, J.-C. Foucher, A. Martelat, and V. Fritsche. 1998. Field tracer test for denitrification in a pyrite-bearing schist aquifer. Appl. Geochem. 13:767–778.
- Pauwels, H., W. Kloppmann, P. Lachassagne, A. Martelat, J.-L. Pinault, and H. Talbo. 1999. Dénitrification des eaux souterraines du bassin versant du Coët-Dan et transfert vers les eaux de surface. p.198–209. In Pollutions diffuses: Du bassin versant au littoral. Ifremer Centre de Brest, Saint-Brieuc, Ploufragan, France.
- Perkins, S.P., and M. Sophocleous. 1999. Development of a comprehensive watershed model applied to study stream yield under drought conditions. Ground Water 37:418–426.
- Perkins, S.P., and M.A. Sophocleous. 2000. Combining SWAT and MODFLOW into an integrated watershed model Open-File Rep. 2000–67. Kansas Geol. Surv., Lawrence, KS.
- Pinault, J.-L., H. Pauwels, and C. Cann. 2001. Inverse modeling of the hydrological and the hydrochemical behavior of hydrosystems: Application to nitrate transport and denitrification. Water Resour. Res. 37:2179–2190.
- Puckett, L.J., and T.K. Cowdery. 2002. Transport and fate of nitrate in a glacial outwash aquifer in relation to ground-water age, land use practices, and redox processes. J. Environ. Qual. 31:782–796.[Abstract/Free Full Text]
- Scheidleder, A., J. Grath, G. Winkler, U. Stärk, C. Koreimann, C. Gmeiner, S. Nixon, J. Casillas, P. Gravesen, J. Leonard, and M. Elvira. 1999. Groundwater quality and quantity in Europe. European Environment Agency, Copenhagen.
- Sophocleous, M.A., J.K. Koelliker, R.S. Govindaraju, T. Birdie, S.R. Ramireddygari, and S.P. Perkins. 1999. Integrated numerical modeling for basin-wide water management: The case of the Rattlesnake Creek basin in south-central Kansas. J. Hydrol. (Amsterdam) 214:179–196.
- Tesoriero, J.A., H. Liebscher, and S.E. Cox. 2000. Mechanism and rate of denitrification in an agricultural watershed: Electron and mass balance among groundwater flow paths. Water Resour. Res. 36:1545–1559.
- Zheng, C. 1990. MT3D, a modular three-dimensional multispecies transport model for simulation of advection, dispersion, and chemical reactions of contaminants in groundwater systems. USEPA, Robert S. Kerr Environ. Res. Lab., Ada, OK.
- Zheng, C., and P.P. Wang. 1999. MT3DMS: A modular three-dimensional multispecies transport model for simulation of advection, dispersion, and chemical reactions of contaminants in groundwater systems; documentation and user's guide SERDP-99–1. U.S. Army Corps of Engineers, Washington, DC.
TOP
ABSTRACT
INTRODUCTION
