In this study, we develop a surrogate model for the efficient estimation of the seawater extent in leaky offshore aquifers, in which the dispersive mixing zone is represented approximately. The model is trained using factorial sampling of a five-parameter non-dimensional parameter space. The samples are modelled with a numerical code (SEAWAT), and the results are used to correct the parameters of a two-parameter semi-analytical sharp-interface model to take into account the mixing zone, which is represented by the 5 % and 50 % seawater concentration contours (isochlors). These corrections allow both the tip and the toe (where a specified isochlor meets the top and bottom, respectively, of the aquifer) to be approximated. The method applies corrections to the freshwater-seawater density difference, the rate of freshwater inflow, and the head drop across the offshore aquitard, allowing the semi-analytical approach to match the dispersive, numerical solution. This approach builds on previous dispersive correction formulae developed for onshore coastal aquifers. This surrogate model requires that the tip is located offshore and provides a rapid method for estimating the offshore freshwater extent and volume, and any seawater intrusion (represented by a shift from one steady-state condition to another) arising from changes to the stresses of the system, e.g., due to sea-level rise or modified submarine groundwater discharge. A macro-enabled Excel workbook implementing the correction is provided to disseminate the methodology. This is available at https://github.com/ZAsiri/Surrogate-Model-.git . • A surrogate model for the dispersive mixing zone in leaky offshore aquifers. • Trained using factorial sampling of a five-parameter space. • The model is validated using independently generated testing data. • Model corrects a two-parameter semi-analytical sharp-interface model. • The model provides a rapid method for estimating the offshore freshwater extent and volume.
Asiri et al. (Sun,) studied this question.