Who's in Charge: Model or Modeler?

Of course, the modeler must be in charge, but the ease of using numerical groundwater codes in the 21^st century can tempt a modeler to suspend thinking and surrender control to the code. Certainly, the modeling process in the 21^st century is radically different and more efficient than in the mid-20^th century.  Advances in modeling logistics (including the availability of high speed portable computers, GUIs and parameter estimation codes) have dramatically increased the efficiency of the modeling process.  These improvements have the potential to increase the effectiveness of models.  Yet, GUIs, which facilitate data input and the viewing of output, tempt the modeler to input data rapidly without critical thinking, to change parameters blindly until the model converges to a seemingly reasonable solution, and to accept results uncritically without performing quality assurance checks on both input and output.  Parameter estimation codes, while straightforward to run, require a sophisticated modeler who can guide the “automated” calibration process to ensure that the model is properly constrained by hydrogeological realities.  Finally, the speed of today’s computers encourages the modeler to run and re-run the model without the necessary thinking and evaluation of output in between runs.

Safeguards against surrendering control to the model include taking the time to design a good conceptual model and performing a quality assurance (Q/A) check on model input and output.  The Q/A check on results should include checking the water balance against the conceptual model, examining the head solution for anomalies and consistency with the conceptual model, and checking the final set of parameter values for agreement with the hydrogeological conceptual model.