A lumped model to simulate nitrate concentration evolution in groundwater at catchment scale

Groundwater nitrate contamination requires implementation of measures to reduce the impact of agricultural sources. Models appear as a way to understand the nitrate transfer and its temporal trend as well as a forecasting tool useful for decision making. Given their ease of implementation compared to discretized models, lumped models seem appropriate tool for stakeholder purposes but remain rarely used in transport modelling. The lumped model BICHE, simulating both hydrologic and nitrate transport in a coupled way, was applied on three sites having contrasted hydrogeological characteristics and under various agricultural pressures. The model has an automatic calibration procedure which is used under constraints fixed by the operator. It was executed on three catchments on a long duration simulation (about 30 years) and performed again 5 years later. This extension of the dataset allowed to evaluate the robustness of the calibration. Whether for the water level or the nitrate concentrations modelling, the performance of the model was almost not altered, confirming the validity of the calibration. Results highlighted model performance for long-term trend of nitrate concentration whereas very short-term variations were less well reproduced, especially for the studied site showing large weekly fluctuations. Certain data, in particular agricultural data, are particularly difficult to obtain for the whole catchment at a detailed time step and the use of rough estimation for these data generate uncertainties. This aspect is however not prohibitive since the purpose of such modelling is to forecast global trends in nitrate concentrations under various types of scenarios. This modelling for the three sites allowed an analysis of the effect of agricultural changes in term of nitrate concentration but also on time lag between the implementation of these changes and their impact on groundwater quality. Considering the ease of implementation, the robustness of model and the quality of the simulations obtained, we confirm that this tool is particularly relevant for stakeholders to address groundwater nitrate contamination.

Automated summary

Groundwater nitrate contamination requires measures to reduce agricultural impacts. Models, especially lumped models like BICHE, can be useful for understanding nitrate transfer trends and forecasting, showing good performance in long-term trend analysis.