Estimation of stream-aquifer exchanges at regional scale using a distributed model: Sensitivity to in-stream water level fluctuations, riverbed elevation and roughness

Abstract

Several studies on stream-aquifer interactions focus on the local scale. However, the estimation of stream-aquifer exchanges for a regional river network remains challenging. This study aims at assessing the sensitivity of distributed stream-aquifer exchanges to in-stream water level fluctuations, riverbed elevation and Manning roughness coefficient.An integrated distributed surface-subsurface model is applied to the Loire river basin (117,480km2, France), where in-stream water level fluctuations are taken into account with a simplified Manning-Strickler approach. The stream-aquifer exchanges are analyzed at pluri-annual and annual scales, as well as during short-term hydrological events.The model simulates the spatio-temporal variability of in-stream water levels accurately, with Nash coefficients up to 0.96 for the Loire river. The river network mainly drains the aquifer system. The average net exchanged flow is 2·10-2m3s−1km−1, which corresponds to 12% of the averaged discharge at the outlet of the basin.The assumption of constant river stages significantly impacts the total infiltration (−70%) and exfiltration (−10%) in the basin, whereas it has a negligible influence on the average net flux. The river fluctuations increase the time variability of the stream-aquifer exchanges and may determine flow reversals during flood events and also more frequently for river stretches at equilibrium with its nearby aquifer.This study highlights the importance of accounting for river stage fluctuations in the modeling of regional hydrosystems. Moreover, a sensitivity analysis indicates that it is mandatory to develop new methodologies to better estimate the riverbed elevation at high resolution for a river network at regional scale. In a lesser extent, errors on Manning coefficient affect the timing of infiltration and exfiltration, leading to temporally localized discrepancies. However it does not affect the estimates of the global net exchanges significantly.