On many a project, preparing and running the groundwater model is the largest cost item. Why is this?

A partial answer comes from Steven P. Larson of S.S. Papadopulos & Associates. Steven Larson’s biography in the conference proceedings tell us that “during his tenure with the USGS, Mr. Larson participated in the development and use of 2-dimensional and 3-dimensional groundwater models that were widely used. Just prior to leaving the USGS, Mr. Larson was the principal investigator on a project to develop a modular version of the USGS groundwater model programs. That project was continued after he left the USGS and ultimately produced the program MODFLOW.”

In his paper The Use of Complex Computer Modeling of Groundwater Systems delivered to the Rock Mountain Mineral Law Foundation in Vancouver July 2007, Larson surveys the factors leading to the ever-increasing complexity and cost of groundwater modeling. He lists these factors as contributing to increased complexity of groundwater models:

  • Increased computing power
  • Development of pre-and post-processing programs
  • Perceived need for greater resolution
  • Concerns that simplicity will be viewed as inaccuracy
  • Inexperience of the analyst.

Leaving aside the greater computing power and computer code power, we have here a classic case of the non-scientific approach. There is the layman’s confusion over precision and accuracy. He notes: “Many features of hydrologic systems can be delineated with great precision and there is a perception that, by necessity, it must be described precisely in a model in order to achieve accurate results.”

And then there is the lawyers conduct in adversarial settings. Models are targets of attack in any legal proceedings. The models typically have many parameters, involve simplifying assumptions, and are by definition, idealizations of reality. The lawyer is skilled at portraying these as flaws if he needs to demolish the model to win the case. The modeler responds by making the model more complex.

Not that the groundwater analyst is free of guilt. As noted by Larson, and as I have often observed myself, most modelers forget that many a groundwater situation could be as well represented by a simple equation or standard flow net culled from the old literature.

I most liked his advice on how to deal with a tendency to make groundwater models overly complex:

  • The principle of parsimony should be the guide in the modeling process.
  • The modeling process is data driven; calibration is a critical measure of model performance.
  • The process will always rely on the subjective judgment of the analyst for model structure and focus.
  • Objective methods including statistical measure, parameter estimates, and sensitivity studies, can help provide insight into the relative importance of model parameters and help to establish if more complex approaches are required.
  • An unwavering focus on the purpose of the model is a key ingredient in successfully determining and maintaining the necessary level of model complexity.
  • Public perception, especially in adversarial settings, is a significant factor in determining the level of complexity necessary for model acceptance.