Use of mixing models to explain groundwater quality time and space variation in a narrowed fluctuating alluvial aquifer

Hydrogeochemical studies can provide insights to pursue a flexible and responsive groundwater resources management.Origin of water, mixing and reactive processes are three key parameters that allow to understand hydrological systems and resulting groundwater quality.However, those parameters may evolve strongly and quickly over time and space, especially for hydrological system impacted by pumping activities and seasonal variations.Within this framework, the current study aims to pursue a survey of the hydrological parameters in a small dynamic alluvial aquifer fed by 3 contrasted end-members and used for water supply.The variations of the end-members and the alluvial groundwaters quality (Ca2+, Mg2+, K+, Na+, Cl-, CO32-, NO3-, delta H-2 and delta O-18) will be used as a record of those parameters. 26 sampling campaigns were carried out fortnightly from October 2017 to October 2018 and data processed with two distinct mixing models: GLUE-EMMA model, which define likelihood of each mixing component, and PHREEQC inverse modeling, which account for the reactive processes that occur as water chemistry evolves. For each studied groundwater, appropriate time variant end members signatures as well as analytical uncertainties were accounted and models constrained to fit within 5.0% for the major ions and 0.5 parts per thousand for water stable isotopes.With suitable models for all alluvial groundwater samples, both GLUE-EMMA and PHREEQC highlight the mixing process between two end-members as the main parameter of the alluvial groundwater quality. Nonetheless, Mg2+ and K+ reactive fluxes are needed to fit with the required constraints.Hence, the alluvial groundwater quality variations are either associated with the typical seasonal variations of the end-member signature and the change of the mixing processes.For both models, mixing process exhibits slow and small as well as strong and sudden changes. The first can be related with the relative capacity of the end members to feed the aquifer according to their individual high and low water periods. The second to the effects of alternative pumping regime between sampling campaign.The capacity of both models to evaluate and monitor main hydrological processes, over time and space, confirms the interest to combine them as a flexible multi process investigation suited for dynamic hydrological system.