Modelling the biophysical interaction occurring in the unsaturated zone (UZ) produces a direct effect on the groundwater dynamic. The simulation of this interaction is performed through different approaches. Conceptual models, which represent a simplified UZ dynamics, are often selected for the coupling to groundwater models when computationally intense uncertainty evaluation techniques are used. On the other hand, physically based models are widely used, particularly at the field scale, for an accurate representation of the water and solute dynamics. Finding the best model balance to improve the representation of groundwater-vegetation interaction while maintaining computational feasibility is essential for the appropriate representation of spatially distributed net-recharge.
We evaluate here the performance of two different UZ models coupled to MODFLOW. A simplified one-dimensional UZ model solves the water balance for each compartment in which the UZ is discretised. The coupling is performed with specific attention posed to the water table-vegetation interaction and to the dynamical calculation of the thickness of the UZ. Model results are spatio-temporally distributed net-recharge values, soil water content and water table levels.
The physically based model SWAP, which simulates the water, solute and heat transport in the UZ by solving the 1-D Richard’s equation, was also coupled to MODFLOW. The coupling was performed similarly to the simplified UZ model. Such coupled models were tested at different locations in the south-east of South Australia, under heterogeneous conditions of groundwater-vegetation interaction.
The simple model configuration shows good results and correlation for groundwater heads and soil moisture, particularly for the topsoil column. Because of the reduced computational load, it was found to be suitable for inverse modelling calibration and uncertainty evaluation algorithms. The coupled physical model represents the soil-water content at different depths in more details, accounting for the heterogeneity of soil parameters, capillary rise and solute transport. The latter is not simulated in the simple coupled model and should be considered for some arid environments where salinity affects plant extraction.
The coupling of the simple and flexible UZ model and the testing in a semi-arid South Australia environment have shown that it has the potential to be effectively applied for data scarce regions where remote sensing is the only source of information. Such a model has been also used for data assimilation experiments, providing useful tools for uncertainty reduction techniques.