Managed Aquifer Recharge (MAR) can play an important role as a measure to control over-abstraction and to restore the groundwater balance under climate change scenarios. MAR may also be applied to sustain or improve the functioning of ecosystems. One of the major difficulties in evaluating the applicability of MAR is to model the infiltration process through variably saturated media with the presence of perched aquifers in alluvial systems. This is the case of the Medina del Campo groundwater basin within the Duero River basin.
In this study, we apply the unsaturated finite difference flow code VS2DTI to simulate MAR to the deep aquifer system used for groundwater exploitation. We estimated average natural recharge for a typical 10-year distribution of wet, medium and dry years based on the data of the period 1940-2010. Eight textural layers with their corresponding hydraulic parameters (saturated hydraulic conductivity, specific storage, porosity, residual moisture content, and alpha and beta van Genuchten parameters) were defined according to reference values in a 2000 m x 55.5 m axisymmetric rectangular domain. A variable grid with increasing cell size from the top left corner (MAR point) was specified. In the top boundary, artificial recharge was simulated as a constant specified flow condition through a 4 m riverbed in the top left corner and natural recharge was evenly distributed along the 2000 m of this boundary. A constant pressure head (p=0) was used to define the deep groundwater level in the bottom boundary. We used flow source points as a Neumann type boundary condition inside the domain to simulate the top aquifer located between 2 m and 11 m deep. The amount of flow in these points as well as the initial moisture conditions were derived from the steady flow model.
The average volume of available water for MAR as estimated by the Duero basin Water Authority was set to 0.3 m/d and distributed proportionally as a function of average annual precipitation for each type of year. For wet years, MAR equaled 0.4 m/d, 0.3 m/d for medium years, and 0 during dry years. A sensitivity analysis of the hydraulic parameters was also carried out.
The results show low water balance errors in the simulations (< 2%) and that the MAR is inefficient to recharge the deeper aquifer due to the low transmissivity and water retention capacity of the aquitard underneath the top aquifer. The main interest of the proposed approach is that it provides a basis for simple assessment of MAR feasibility through unsaturated flow modeling. This is an important contribution since multilayer aquifer systems where there are perched aquifers are difficult to tackle with standard saturated groundwater flow programs.