BARRIERS AND DISINCENTIVES TO QUALITY GROUNDWATER MODELING IN PRACTICE

Modern, computer-based groundwater flow and transport models have regularly been applied in practice for the last thirty years. Three types of application are common. These are the design and operation of facilities, the development and testing of regulations and policies, and dispute resolution. In all applications there are barriers inhibiting quality modeling practices, especially the use of more recently developed concepts and methods. Even in the absence of barriers there are often finanical, business, social, and other disincentives to quality modeling. The propensity for barriers and disincentives increases where parties have a prejudiced interest in the modeling outcomes; such an interest is highest in dispute resolution. Barriers and disincentives enter the modeling processes at all stages, starting with selecting the modeling team, posing the issue to be addressed, compiling data, formulating and testing the conceptual model, choosing the computer program(s) to implement the conceptual model, interpreting and presenting the results, and using the results to make recommendations or decisions.

This talk focuses on barriers and disincentives that are implicit, intrinsic, and insidious. Consider some examples. Disincentives include a desire to keep the client happy, minimizing costs or liability by taking advantage of the gray of uncertainty. You shade it almost white or almost black, whichever most benefits the client, but either case is the very definition of bias. Even more insidious is when the modeling team and client become essentially symbiotic, with the modelers loosing any perspective that what they are presenting is biased. Barriers include the level of training and resources available to the modeling team, and the restriction of computer programs to their own proprietary programs or perhaps only to widely used community programs, regardless of limitations they may have in algorithms, programing, or conceptual breadth and fitness. Barriers can be as silly as a decision to consider only the client's data, ignoring what may be a plethora of off-site information, much of it in the public domain. Or they can be as treacherous as a narrow view of an issue or conceptual model that ignores the broader context. Perhaps the most crippling barrier is that applied modellers get just "one shot" and then the project is over.