In groundwater flow modeling, as in any modeling problem, a certain amount of error is inevitable. Recharge or discharge wells, acting as point sources or sinks, play a key role in modeling accuracy, and the way they are treated can either reduce or increase errors. In this study, two approaches were investigated: first, transferring the well to the nearest node in its neighborhood, and second, distributing the pumping rate of the well among the closest nodes. A hypothetical aquifer was examined under two conditions-unconfined and confined-and using both triangular and square meshes. The results indicated that simplifying the model by moving the pumping well to the nearest node is justified only for unconfined aquifers with triangular meshes. For other cases-including unconfined aquifers with square meshes and confined aquifers with either mesh-the second approach is recommended, as it significantly reduces errors in groundwater flow modeling. These findings can also be generalized to real aquifer studies. Quantitative results show that Approach 2 consistently reduces modeling errors: for unconfined aquifers, MAE values are below 0.03 for both mesh types, whereas confined aquifers exhibit larger reductions, particularly with triangular meshes, where MAE reaches 0.38 and maximum errors up to 1.17. These results highlight the robustness of Approach 2 across different mesh configurations and aquifer conditions, providing an effective and reliable numerical tool for groundwater modeling.
Zeynali et al. (2025) studied this question.