Norwegian Institute for Air Research
Netherlands Institute for Ecology
Tyndall Centre for Climate Change Research
Institute for Environmental Studies, Free University Amsterdam
University of Plymouth
Centre for Social and Economic Research on the Global Environment
Land-Ocean Interactions in the Coastal Zone
 


Nutrient Dynamics in European Water Systems

Case Study 2 - Chasing after nutrients through watersheds

 
2.2 What it takes
 
The ELOISE projects INCA, EROS-2000/EROS-21 and RANR have focused on the modelling of nutrient transformation processes in the watershed and the river network. The main aim was to relate statistics of land use and human (agricultural, domestic) practices of nutrient input into the system, to the load of rivers.

Inca Model

INCA is a deterministic model that includes land and river processes, and is driven by spatially explicit input data, as shown below in Figure 2.2(a). The model accounts for stocks of ammonium and nitrate in the soil and ground water pools, and in the stream reaches. The model also simulates the flow of water through the plant/soil system from different land use types to deliver the nitrogen load to the river system, which is then routed downstream after accounting for direct effluent discharges, and in-stream nitrification and denitrification.

 
Figure 2.2(a). The structure of the land component of INCA clearly showing all the individual processes taken into consideration in this full-deterministic model (Wade et al., 2002).
Figure 2.2(a). The structure of the land component of INCA clearly showing all the individual processes taken into consideration in this full-deterministic model (Wade et al., 2002).
 
Figure 2.2(b). Schematic representation of the HYDROSTRAHLER model and its parameters. NAP0: initial level of the water-table (mm); SOLsat: water saturation level of the soil (mm); tinf: rate of infiltration (decade 1); tecs: rate of superficial runoff (decade 1); ten: Water-table runoff (decade 1); tmelt: degree-decade factor (mm C 1 decade 1); NIV0: initial snow depth (mm at the top of the basin).
Figure 2.2(b). Schematic representation of the HYDROSTRAHLER model and its parameters. NAP0: initial level of the water-table (mm); SOLsat: water saturation level of the soil (mm); tinf: rate of infiltration (decade 1); tecs: rate of superficial runoff (decade 1); ten: Water-table runoff (decade 1); tmelt: degree-decade factor (mm C 1 decade 1); NIV0: initial snow depth (mm at the top of the basin). From Garnier et al., (2002)
 

The RIVERSTRAHLER model

This model (Billen & Garnier 2000, Garnier et al., 2002), first applied to the Seine (Billen et al. 2001), was used to describe nutrient and ecological dynamics in the Danube watershed and river.

This model synthesises the hydrological network of a river basin by stream order (HYDROSTRAHLER module), which reduces the computational load to a reasonable level (Figure 2.2(b)). The river model for the different stream orders of several sub-basins integrates full ecological dynamics, including transformations of nutrients in the ecosystem (biogeochemical RIVE module).

For the different sub-basins, nutrient and organic inputs are derived from gross statistics (population density, type and intensity of industrial activity, fertiliser application, land use) ( Figure 2.2(c)).

 
Figure 2.2(c). Relationship between specific fluxes of nitrogen and phosphorus (kgN or P km -2) and population density (inhabitants km -2), Da: Danube River ; El: Elbe River ; Lo: Loire River ; Mo: Mosel River ; Rh: Rhine River ; Sch: Scheldt River ; Se: Seine River . From Garnier et al. (2002).
Figure 2.2(c). Relationship between specific fluxes of nitrogen and phosphorus (kgN or P km -2) and population density (inhabitants km -2), Da: Danube River ; El: Elbe River ; Lo: Loire River ; Mo: Mosel River ; Rh: Rhine River ; Sch: Scheldt River ; Se: Seine River . From Garnier et al., (2002).
 

Aggregating chaos in rules.

Models require input data in quantities that grow exponentially with the model complexity, and a recurrent problem confronting model-users is the lack of data at the desired spatio-temporal resolution.

Forsman & Grimvall (2003) examined for the RANR project under what circumstances spatially distributed inputs to a substance transport model can be replaced by spatially aggregated inputs without jeopardising the accuracy of spatially aggregated model outputs.

The SOIL/SOILN model was used by Forsman & Grimvall (2003) as test case according to the following rationale:

1.-Response of model estimates for leaching of nitrogen from the root zone to variation in meteorological inputs, soil and crop type, and fertiliser application.

2.-Supression of non linear features by (i)- aggregating model outputs over one or several years, and (ii) considering relationships between different components of a multivariate model output ( Figure 2.2(d)).
 
Figure 2.2(d). Regression lines fitted to annual and 30-yr means of precipitation and nitrate leaching. Soil parameters and agricultural practices were selected to correspond to cultivation of barley on loamy sand in southern Sweden
Figure 2.2(d). Regression lines fitted to annual and 30-yr means of precipitation and nitrate leaching. Soil parameters and agricultural practices were selected to correspond to cultivation of barley on loamy sand in southern Sweden
 

Finally, Forsman & Grimvall (2003) presented a simplified model of the expected total leaching of nitrogen from the root zone in agricultural areas.

<< Nutrient Dynamics Contents - back to top - 2.3 What you get >>