PRACTICAL APPLICATION OF REGULARIZED INVERSION IN GROUNDWATER MODELS

In every scientific discipline, there is a gap between the development of new analysis techniques and the application of those techniques in solving real problems. The field of groundwater modeling exhibits this lag between research and application most significantly in the area of model calibration or inverse modeling. One of the most important new techniques in inverse modeling is called regularized inversion. The first commonly used regularized inversion model, called PEST, was developed by John Doherty. Regularized inversion coupled with pilot points (Doherty 2003) provides the groundwater modeling community with a sophisticated technique for model calibration.

Lack of understanding on the part of most groundwater modelers has kept inverse models in general, and regularized inversion in particular, from being more widely used. Thus the challenge for researchers and developers of these new techniques is to not only create a robust methodology for calibrating models but also to present these techniques in a way that modelers with limited mathematical backgrounds can appreciate.

The use of regularized inversion is presented for a variety of real-world models to illustrate how the technique can assist in model calibration. Suggestions for improving transfer of technology from researcher to applied modeler are discussed. Common pitfalls are also highlighted, along with an assessment of regulatory and client acceptance of the technique. One example, from southwestern Florida shows the problem of model non-uniqueness and the importance of a sound conceptual model.