Estimating solute travel times from time series of nitrate concentration in groundwater: Application to a small agricultural catchment in Brittany, France

Stream flow and concentration dynamics are usually supposed to integrate the diverse hydrological and biogeochemical processes occurring within catchments and their variability in space and time. Proxies like transit times and young water fractions are often used to understand this variability. However, this procedure might not be suitable in catchments with ephemeral streams. The objective of this paper is to test the hypothesis that in catchments where groundwater flow is the main source of water output, solute travel times can be inferred from the monitored groundwater levels and solute concentration time series of piezometer networks. We used data from a well-studied small agricultural catchment in Brittany, belonging to the French network of Critical Zone Observatories. The data set includes long-term nitrate concentration time-series measured both at the stream outlet and the network of piezometers. We applied a parsimonious, conceptual dual-permeability mixing modelto calibrate our dataset and found that the nitrate travel times estimated in mid to upslope pie-zometers were consistently higher than the one at the stream outlet. We further observed an asynchronicity in seasonal concentration-discharge dynamics between the piezometers and the stream. We hypothesized that either due to riparian denitrification and uptake by vegetation, a pronounced seasonality is being generated which is making our model underestimate the nitrate travel times at the stream, or the piezometers are acting as proxies to the slower conceptual store of our mixing model, and shallow subsurface flow or some hydrologically significant heterogeneities are contributing to the faster route which is not tapped into by the piezometer network. Latter hypothesis is further supported by the fact that nitrate travel velocity along two different transects are in agreement with the slower conceptual store of our model. Nevertheless, we established that to specifically estimate the groundwater travel time of a catchment, a piezometer free from riparian influence alongside a stream is very beneficial.