THE STOCHASTIC FACIES MODEL OF HETEROGENEITY: A NECESSARY TOOL OR UNNECESSARY TEDIUM?

For decades geologists have characterized patterns of geologic facies based on subsurface data and modern analogs. Hydrogeologists were largely unable to incorporate most of the facies characterizations in flow and transport models, until the advent of stochastic facies simulation methods based on the likes of indicator geostatistical and geologic process techniques. As these facies simulation methods improve, it is becoming increasingly feasible for geologists to conduct the facies characterization entirely with quantitative tools such as Markov chain models of transition probability in geostatistical indicator simulation (TPROGS) within a deterministic (e.g., sequence stratigraphy) or stochastic (sedimentary process simulation) framework. This approach can more fully integrate the geologic characterization into the groundwater model. On the other hand, many hydrogeologists, often citing parsimonious considerations, see it as an additional step that they have been getting along without for some years through more traditional means such as interpolation of pumping test data and inverse procedures. I will discuss the role of the facies characterization paradigms with examples of actual applications illustrating (1) cases where the heterogeneity cannot be feasibly characterized without the facies approach, (2) cases where absence of such heterogeneity invalidates the governing transport equation, and (3) cases where the facies model may be unnecessary. The issue of parsimony and how one can assess whether a model is too complex will also be discussed.