From root zone modelling to regional forecasting of nitrate concentration in recharge flows – The case of the Walloon Region (Belgium)
Introduction
The groundwater recharge quality is a key factor in the sustainability of groundwater resources (Döll and Fiedler, 2008). Different sources may influence the recharge quality. This paper will focus on diffuse agricultural pollution of nitrate. According to Petry et al. (2002), the diffuse pollution from agricultural sources is one of the most important sources of groundwater quality perturbation. But it is also among the most difficult to manage since mitigation techniques that are effective in a site are not always effective if applied in another pedo-climatic context (Morari et al., 2004). Furthermore, fighting diffuse pollution from agriculture is complex due to the temporal and spatial lags between the management actions taken at the farm level and the environmental response (Schröder et al., 2004). The European community regulation on nitrate (EEC 1991/676) launched important research about mitigation options, and sets of BMP (best management practices) were proposed (Weightmann, 1996, D’Arcy and Frost, 2001). To move on to the next step and help to decide on the best management practices in order to match the EU expectations, we need modelling systems that can generalize field results regardless of the agro-pedological context.
Mathematical models simulating the complex processes in a watershed are useful tools to understand the problems and to find solutions through land-use changes and best management practices (Borah and Bera, 2003). For the past three decades a number of agricultural watershed models have been developed to support soil erosion assessment, water resource analysis, and water quality management in agricultural watersheds. In the 1970s and 1980s most of the commonly used models were formulated. Some of the commonly used hydrologic and nonpoint source pollution models include AGNPS (Young et al., 1987), ANSWERS (Beasley et al., 1980), EPIC (Williams et al., 1994), SWAT (Neitsch et al., 2002), SHETRAN (Birkinshaw and Ewen, 2000).
Since the early 1990s most modelling research has focused on the development of graphical user interfaces (GUI) and the integration with geographic information systems (GIS) and remote sensing data (Borah and Bera, 2003).
Several modelling methodologies are available to assess the fate of agrochemicals. These methodologies vary in complexity and are tailored to specific problems and specific scales (Schoumans and Silgram, 2003). The complexity level depends, among other things, on the approach used for the mathematical representation of water flow in the unsaturated zone. Hydrological modelling systems are usually classified within two main categories: numerical resolutions of Richard’s equation, like those proposed by Varado et al. (2006), and reservoir cascade schemes (Williams et al., 1994, Neitsch et al., 2002). The second type of model is particularly useful in basin scale study when long time horizons are considered and when repeated simulations are required like for sensitivity analysis (Gandolfi et al., 2006).
Gandolfi et al. (2006) compared the results from these two model types concerning soil moisture in the first meter, recharge volume, and recharge flows. They concluded to a good agreement of recharge volumes for the cascade scheme model, especially when recharge is accumulated over a monthly time basis. They also noted that reservoir models needed to be improved under the root zone by allowing the number of reservoirs in cascade to vary with changing profile depth (Gandolfi et al., 2006) to capture the daily recharge dynamics.
To model nitrogen dynamics, Cannavo et al. (2008) proposed a review of the main mathematical expressions modelling nitrogen flows. The main trend over the last 15 years has been the shift from mechanistic models to functional models, with a simplification of equations according to specific objectives or specific contexts.
In the Walloon Region, the depth of the vadose zone varies between 1.5 m and 104 m (Monjoie et al., 2000, Hallet and Barbier, 2007). It is then essential to take into account the dynamics of water and nutrient transfer through the vadose zone in the modelling of recharge water quality (Dautrebande et al., 1999) and, when doing so, to fill the gap between field cropping practices linked to nitrogen use and their expected effects on the nitrate concentration in groundwater and surface water. The first aim of this paper is to describe how the unsaturated soil under the root zone was introduced into the EPIC model (1D water and nutrient flows) (Williams, 1995).
As the final objective is to help decision on the BMP to adopt at field level to match the EU expectations in the various pedo-climatic contexts of the whole Walloon Region, the second aim of this study is to link the extended model to all the territorial databases using a geographical information system (GIS). This way, the nitrate concentration of the recharge water will be modelled in a spatially distributed way. Moreover the sensitivity analysis will allow us to validate proposed cropping practices.
Section snippets
Materials and methods
The EPIC model proposed by Williams et al. (1994) is restricted to the soil layer between the surface and 1.5 m depth. It is composed of physically based components for simulating soil water balance, crop growth, erosion, the movements of pesticide and nutrient with water and sediments and related processes for assessing management strategies (Priya and Shibasaki, 2001), using a daily time step. The EPIC crop growth module (including water and nutrient uptake) is also part of the SWAT model
Deep 1D modelling and validation
The first tests of the EPIC model in the Walloon Region were launched during the nineties (Masereel and Dautrebande, 1995, Cocu et al., 1999, Dautrebande et al., 1999). Masereel and Dautrebande (1995) worked on soil moisture modelling under wheat crop at different depths. They found that the model was reasonably accurate for the deeper layers, as can be seen in Fig. 1. Considering three measurement repetitions (crosses, squares, and diamonds), RMSE (root mean square error = square (sum (observed
Discussion and conclusions
Since the 1990s the research has focused on soil–water–plant models that applied specifically to water and nitrogen flows in the root zone. The impacts of cropping practices were monitored in this zone. Considering that the unsaturated soil layer varies between 1.5 and 104 m in the Walloon Region of Belgium, we attempted to fill the gap between root zone modelling and groundwater recharge quality. On the basis of the EPIC model, an extended model has been developed and adapted to the regional
Acknowledgements
The development of the model was carried out with the financial support of the Walloon Region and the public society of water management (Société Publique de Gestion de l’Eau).
References (40)
- et al.
Nitrogen transformation component for SHETRAN catchment nitrate transport modelling
J. Hydrol.
(2000) - et al.
An integrated modelling framework to estimate the fate of nutrients: application to the Loire (France)
Ecol. Modell.
(2008) - et al.
Modeling N dynamics to assess environmental impacts of cropped soils
Adv. Agron.
(2008) - et al.
The role of best management practices in alleviating water quality problems associated with diffused pollution
Sci. Total Environ.
(2001) - et al.
Environ. Modell. Softw.
(2006) - et al.
An integrated non-point source model-GIS system for selecting criteria of best management practices in the Po Valley, North Italy
Agric. Ecosyst. Environ.
(2004) - et al.
Hydrological controls on nutrient concentrations and fluxes in agricultural catchments
The Science of the Total Environment
(2002) - et al.
National spatial crop yield simulation using GIS-based crop production model
Ecol. Modell.
(2001) - et al.
A GIS-based modelling approach for implementation of sustainable farm practices
Environ. Modell. Softw.
(2000) - et al.
The effects of nutrient losses from agriculture on ground and surface water quality: the position of science in developing indicators for regulation
Environ. Sci. Policy
(2004)
Analysis of nitrate pollution control policies in the irrigated agriculture of Apulia Region (Southern Italy): a bio-economic modelling approach
Agric. Syst.
Global estimation of crop productivity and the impact of global warming by GIS and EPIC integration
Ecol. Modell.
Development and assessment of an efficient vadose zone module solving the 1D Richard’s equation and including root extraction by plants
J. Hydrol.
ANSWERS: a model for watershed planning
Trans. ASAE
Watershed-scale hydrologic and nonpoint-source pollution models: review of mathematical bases
Trans. ASAE
Cited by (18)
Looking at trends in high flows at a local scale: The case study of Wallonia (Belgium)
2020, Journal of Hydrology: Regional StudiesCitation Excerpt :We focused on climatic variables. Precipitation, snow, temperature and evapotranspiration data were obtained for each catchment from 1 January 1961 until 31 December 2017 using the EPICgrid model (Sohier et al., 2009). In this model, snow data is actually derived from precipitation data when temperature falls below 0 °C; snowfall is therefore calculated in mm of precipitation.
Performance assessment of nitrate leaching models for highly vulnerable soils used in low-input farming based on lysimeter data
2014, Science of the Total EnvironmentCitation Excerpt :EPIC (Williams et al., 1984, 1989) is a cropping systems simulation model, which was developed to estimate soil productivity as affected by erosion throughout the United States during the 1980s. EPIC is a field scale model, but linked to a GIS it has been applied in several regional model applications (Burkart et al., 1999; Sohier et al., 2009). Furthermore the EPIC model has been applied to study the effect of agricultural practices and biofuel cultivation on N leaching at the European scale (Bouraoui and Aloe, 2007; Van der Velde et al., 2009).
Assessing the regional and temporal variability of the topographic threshold for ephemeral gully initiation using quantile regression in Wallonia (Belgium)
2014, GeomorphologyCitation Excerpt :Average annual precipitation is around 800 mm and is fairly equally distributed throughout a year. The mean erosivity of precipitation is 680 MJ mm ha− 1 h− 1 yr− 1 (Sohier et al., 2009). The Jurassic area (#2) is characterized by a rugged relief with steep slopes.
Modelling the effects of the current policy measures in agriculture: An unique model from field to regional scale in Walloon region of Belgium
2010, Environmental Science and PolicyCitation Excerpt :These water and nitrogen flows can then be aggregated on the basis of each reporting unit (i.e. surface water body, NVZ,…) or it can be used as a raster file into a GIS. Sohier et al. (2008b) showed a good agreement between local measurements and local modelling using EPICgrid model (Sohier et al., 2009). In the first action plan, 3096 km2 were designated as NVZ.
Evaluation of a spatialized agronomic model in predicting yield and N leaching at the scale of the Seine-Normandie Basin
2018, Environmental Science and Pollution ResearchAssessing the operation rules of a reservoir system based on a detailed modelling chain
2015, Natural Hazards and Earth System Sciences