The state of the science and practice in groundwater modeling brings to mind highly sophisticated computer models that are running in parallel on many multi-processor machines. These models are expected to incorporate many different processes of both saturated and unsaturated groundwater flow and transport and possibly the media to which it connects, like surface waters and the atmosphere. We are increasingly aware we cannot study groundwater flow in isolation if we are to make useful predictions of, for instance, the impacts of climate change on the groundwater regime. We have come a long way.

Today we are no longer limited to equations for flow toward a well, perhaps near an infinitely long straight canal (method of images), to sandbox models in the laboratory, or to simple steady state models of flow in a single aquifer. We now have computer models that solve groundwater flow and transport in multi-aquifer settings under transient conditions and with a user-friendly graphical user interface that allows widespread use. Additionally, multi-media models are now leaving the research environment and becoming available to mainstream consultants. So in that sense the science of groundwater modeling has matured.

The practice of groundwater modeling, however, has also matured. We have come to realize that model output, being a necessary simplification of an unknowably complex natural world, has inherent limitations. That is, a model of reality is not reality itself. There is uncertainty associated with all facets of our model—parameterization, aquifer geometry and discretization, boundary conditions, and future hydrologic drivers such as future pumping regimes and climates. Today a model is now more appropriately seen as a tool that provides a quantitative framework to make supportable forecasts rather than an oracle that gives us all the answers.

In this chapter we set out to briefly review the state of the science and practice in modeling. In doing so, we augment existing assessments from the journal Groundwater (e.g., Hunt and Zheng 2012; Langevin and Panday 2012; Molz 2017a,b; White 2017), specifically in terms of modeling approach. An effective modeling approach is critical. If a modeler does not decompose the societal problem correctly, the model will not be fit-for-purpose, no matter how sophisticated the code’s capabilities. Moreover, capabilities of codes will be ever improving; good modeling practices have a timelessness that is more robust.

How best to decompose the problem and provide models that are accepted? We lay out here some approaches for today’s applied groundwater modeling. Specifically, we suggest: (1) a step-wise modeling process; (2) including a two-dimensional areal model within this process; (3) keeping abreast of industry standards; and (4) ways to increase acceptance of the models we produce.