Is a simple model based on two mixing reservoirs able to reproduce the intra-annual dynamics of DOC and NO stream concentrations in an agricultural headwater catchment?

Agriculture disturbs the biogeochemical cycles of major elements, which alters the elemental stoichiometry of surface stream waters, with potential impacts on their ecosystems. However, models of catchment hydrology and water quality remain relatively disconnected, even though the observation that dissolved organic carbon (DOC) and nitrate (NO) have opposite spatial and temporal patterns seems relevant for improving our representation of hydrological transport pathways within catchments. We tested the ability of a parsimonious model to simultaneously reproduce intra-annual dynamics of stream flow, DOC and NO concentrations using 15 years of daily data from a small headwater agricultural catchment (AgrHyS observatory). The model consists of an unsaturated reservoir, a slow reservoir representing the groundwater and a fast reservoir representing the riparian zone and preferential flow paths. The sources of DOC and NO are assumed to behave as infinite pools with a fixed concentration in each reservoir that contributes to the stream. Stream concentrations thus result from simple mixing of slow and fast reservoir contributions. The model simultaneously reproduced annual and storm-event dynamics of discharge, DOC and NO concentrations in the stream, with calibration KGE scores of 0.77, 0.64 and 0.58 respectively, and validation KGE scores of 0.72, 0.58 and 0.43 respectively. These results suggest that the dynamics of these concentrations can be explained by hydrological transport processes and thus by temporally variable contributions from slow (NO rich and DOC poor) and fast reservoirs (DOC rich and NO poor), with a poor representation of the biogeochemical transformations. Unexpectedly, using the concentration time series to calibrate the model increased uncertainty in the parameters that control hydrological fluxes of the model. The legacy storage of NO resulting from agricultural history in the studied catchment supports the assumption that the main DOC and NO sources behave as infinite pools at the scale of several years. Nevertheless, reproducing the long-term trends in solute concentration would require additional information about DOC and NO trends within the reservoirs.