Using hillslope-scale groundwater model to bridge the gap between basin inputs and river concentrations. The case of nitrates in Brittany (France).
Résumé
EGU21-10962https://doi.org/10.5194/egusphere-egu21-10962EGU General Assembly 2021© Author(s) 2021. This work is distributed underthe Creative Commons Attribution 4.0 License.Using hillslope-scale groundwater model to bridge the gap betweenbasin inputs and river concentrations. The case of nitrates inBrittany (France).Luca Guillaumot1, Luc Aquilina2, Jean-Raynald de Dreuzy3, Jean Marçais, and Patrick Durand1International Institute for Applied System Analysis (IIASA), Laxenburg, Austria (luca.guillaumot@live.fr)2Université de Rennes 1, CNRS - Géosciences Rennes, Rennes, France (luc.aquilina@univ-rennes1.fr)3Université de Rennes 1, CNRS - Géosciences Rennes, Rennes, France (jean-raynald.de-dreuzy@univ-rennes1.fr)Over the past decades, intensive agriculture has altered surface water and groundwater resourcesquality. Nutrient surplus increased nitrate concentrations in groundwater and rivers resulting ineutrophication or drinking water risk having ecosystem, sanitary and economic repercussions.Legislations led to a reduction of agricultural inputs of nitrogen since 1990’s followed by adecrease of nitrate concentrations in rivers, but still difficult to predict and evaluate. Indeed, theincomplete knowledge of the spatial variability of climate and nitrogen inputs, cumulated to theunknown groundwater heterogeneity, leads to hydrological and biogeochemical processesdifficult to model. This study deals with the long-term variations (~decades) of nitrateconcentrations in three rivers (~30 km² catchment) located in Brittany. Thus, we focus ongroundwater modelling because they constitute the bigger hydrological reservoir. We developed aparsimonious equivalent hillslope-scale groundwater model. The model parameterization, whichcontrols hydrological functioning such as mean groundwater residence times, young watercontribution to the river or denitrification, relies on long-term monitored streamflow and nitrateriver concentrations. In addition, dissolved CFC were sampled in the catchments. Finally, we foundthat uncertainty on simulated nitrate river concentrations is low. The physically-based model alsobrings information on temporal and spatial variability of groundwater residence times highlightingthe relative importance of young (1-5 yr) and old waters (~decades) for nitrate riverconcentrations. Moreover, calibrated models show similar trends looking at two fictive inputscenarios from 2015 to 2050.
