The Hesbaye area is located in the northeastern part of Belgium. The aquifer formations consist of chalk deposits. Groundwater provides about 80,000 m3 d-1. Despite 5 to 20 meters of superficial loess deposits, the groundwater quality is threatened by increasing nitrate concentrations of 0.35 mg×L-1 per year in the semi-confined part of the aquifer to 0.7 mg×L-1 in the unconfined aquifer. Presently, nitrate concentrations are between 15 and 25 mg×L-1 in the semi-confined part of the aquifer but are more than 35 mg×L-1 (reaching locally 150 mg×L-1) in the unconfined part that covers 95% of the area. Nitrate concentrations have such a high spatial variation that various statistical treatments (such as kriging used to draw iso-concentration maps) have failed. This failure is due to the fact that the concentrations are highly influenced by surface land use (grass land, culture land, villages, point source pollutants, etc.). In addition, nitrate content in the aquifer varies vertically with decreasing values at depth (gradient of 0.7 mg×L-1×m-1).
Aquifer parameters were determined by 38 pumping and tracer tests conducted in radial convergent or cylindrical flow at 11 sites. Results showed that hydraulic conductivity values ranged from 1 × 10-6 m×s-1 to 4 × 10-2 m×s-1 and effective porosities from 0.5% to 7%, showing that the aquifer was heterogeneous. Dispersivity values were affected by scale effects and varied according to chalk weathering or fracture zones. They ranged from less than 5 m in fractures to more than 60 m in weathered chalk (as in the upper part of the aquifer) and in the chalk matrix. In the chalk, transport processes were influenced by the immobile water effect due to diffusive transfer from the moving to the non-moving fluid. Non-effective porosity filled by non-moving fluid was estimated between 8 to 42%. The transfer constant ranged from 0.98 × 10-7 s-1 to 10 × 10-7 s-1.
The determination of the transport parameters allowed simulation of nitrate transport at a regional scale. The SUFT3D (Saturated and Unsaturated Flow and Transport Model), developed by the Hydrogeology Section of the Georesources, Geotechnologies and Building Materials Department of Liege University was used. The modelled groundwater zone was defined as a 2.0 x 4.5 km rectangle of 10 km². The aquifer was subdivided into 6 layers of 3350 cells (50 x 50 m wide and 3 to 15 m thick). Boundary flow conditions were defined as a prescribed head (Dirichlet conditions) to the north and the south of the area modelled. As the model simulations run for a time period of 30 years, the northern Dirichlet conditions had to be adapted to the regional and seasonal water table fluctuations that were observed during this period. At the south boundary, as the aquifer is drained by the river Geer, the water table is fixed at the river bed altitude. The eastern and western boundaries were, according to the regional piezometry, assumed to be impermeable. For the transport boundary conditions, prescribed flux (Cauchy conditions) was used for the aquifer top. Elsewhere Neumann conditions were used
Simulations were run for the period from 1963 to 1992. Nitrate inputs were averaged yearly and estimated according to actual input conditions. These conditions were calculated by simulation of nitrate flows through the non-saturated part of the aquifer using the EPIC-Model and taking into account the amount of nitrate fertilisers used by farmers (given by the Belgian government Statistical Institute). Initial conditions were calculated according to the 1963 nitrate inputs.Simulations demonstrated that it is important to distinguish the origin of the pollution as either point or non-point (diffuse) sources. For point source pollutants (such as contaminated infiltration basins), aquifer nitrate concentrations increased during low water level periods due to weaker dilution linked with a poor regional water gradient. During high groundwater levels, dilution is more important and the nitrate concentration decreases. If a point source pollutant is suppressed, aquifer quality is improved within one to two years. This demonstrates the importance of protective actions that could be applied within the framework of the protection zones around collecting galleries and pumping fields.
For diffuse contamination the mean input over the area (10 m depth below cropped areas) increased from 1.32 × 10-7 mg×m-2×s-1 in 1963 to 5.14 × 10-7 mg×m-2×s-1 (i.e., a factor of four). According to these values, concentrations ranged from 11 mg×L-1 to 22 mg×L-1 (i.e., increasing by 0.5 mg×L-1 per year) between 1963 and 1992. Predictive simulations, using 1992 input, show that it will take more or less 30 years for the aquifer to be in equilibrium with the 1992 input. At that time the mean concentration value will be around 30 mg×L-1.
The main results of the simulations clearly show that if actions are taken to decrease nitrate inputs, even if the aquifer nitrate contents rapidly react to the new input, nitrate levels will decrease slowly and take about 30 years to be in equilibrium with the new inputs. This long delay is due to the immobile water effect that is characteristic of the chalk aquifer. Thus it is important to inform environmentalists who work on action programs (such as the water directive imposed by the European Community in the vulnerable zones) that the effects of their actions must be based on 10 to 20 year scenarios. To this estimation, based on the reaction time of the aquifer to a new input, one must also add the time transfer of the pollutant through the unsaturated part of the aquifer.
Hebaye aquifer, Chalk, Nitrate contamination, Groundwater modelling, Immobile water effect.
Vincent Hallet, Département de géologie,
Facultés Universitaires Notre-Dame de la Paix,
Rue de Bruxelles, n° 61 5000 Namur Belgique