ABSTRACT
Often, the largest obstacles to groundwater flow and transport modeling are incorporation of the related aspects of heterogeneity and uncertainty. Hydrogeologic heterogeneity is the primary source of spreading of contaminant plumes, and therefore of paramount importance in estimating travel times of contaminants. Descriptions of heterogeneity in three-dimensional model space are extremely data intensive and cannot be adequately satisfied given the extraordinarily small volume of subsurface sampled by aquifer testing. Uncertainty is introduced to the analysis of groundwater transport through lack of data for flow and transport parameters, as well as uncertainty in their spatial distribution. Though characterization efforts present the opportunity to reduce uncertainty, it cannot be eliminated, any more than heterogeneity can be fully described.
With the inevitability of heterogeneity and uncertainty, defensible analysis of groundwater contaminant transport problems must incorporate these features in the analysis. To ignore them is to feign an impossible level of knowledge about the natural system that may prove unacceptable to regulators and the public. A variety of tools for handling heterogeneity and uncertainty have been developed in the literature in recent years, enhanced by advances in numerical computing abilities. The tools chosen can be tailored for the specific conditions (and uncertainties) presented by a particular subsurface environment.
Three widely varying environments can be used to illustrate the approaches. The sites have in common the contaminant transport problem presented by the presence of an underground nuclear test detonated in contact with groundwater. The problem to be addressed is the prediction of transport of radionuclides in the subsurface over the next one thousand years. Each site builds on the complexity of the previous. A summary of the site background, the modeling approach, representative results and similarities and differences between the sites are presented in this 224paper. The common focus is that spatial variability and parametric uncertainty propagate through the flow and transport solutions and result in a large range of output that places wide bounds on the expected results at a particular site.
