It is often necessary to develop or apply complex calibrated groundwater models in order to learn how a hydrologic system works, and narrow down hydrologic parameters. However, when a model is being applied administratively to protect the river, it is necessary to make sure that simulation of highly uncertain phenomena, such at ET salvage, or induced inflow from speculative aquifers, does not have too much influence on model results. Rather than an complex model, that includes highly uncertain elements, we are often better off with a straightforward approach that does not hide hydrologic impact in a multitude of complex boundaries.   

In addition, just because a model is more complex, that does not mean that it is right.  Groundwater models are non-unique.  In theory, the non-uniqueness problem can be attacked by developing multiple models of single system, and applying statistical methods to the resulting suite of models. However, a regulatory agency does not often have the time and resources to apply this approach.  Nevertheless, useable administrative tools are needed, and regulators must develop them for many areas.   

One way to address this kind of problem is to convert a calibrated groundwater model into a superposition version, and modify uncertain boundary conditions that are not of immediate interest to ensure that they do not play a significant role in model predictions.  An added advantage to such an approach is the relative simplicity of the resulting model, which becomes important in a regulatory setting, where non-modelers often need to make calculations of hydrologic effect.  Misapplication of models is constant danger, which becomes more acute with more complex models.