ASSESSING THE SPATIAL VARIABILITY OF HYDRAULIC CONDUCTIVITY AND ITS ROLE IN SOLUTE AND HEAT TRANSPORT

Ed Sudicky’s early work (1986, WRR) assessing the spatial variability of hydraulic conductivity (K) and its role in the dispersion process was a driving force in exploring the often unrealized multi-scale complexity of aquifers. This work spurred a tremendous amount of research by numerous researchers into accessing aquifer heterogeneity in porous and fractured media. However, even today, numerical flow and transport simulations are often run using sparse data sets which severely limit their accuracy and realism.

Over the years we have tried to advance both the assessment of the role of the spatial variability of K in the dispersion process and the characterization of K variability in the field. Dispersion experiments with correlated random fields have been conducted in a 6.2x1.2x0.5 m flow tank. Using custom made porous media size distributions we were able to develop 25 separate permeability classes (3400 cells) to segment the permeability field. Image processing techniques were then developed to quantify the concentration distributions at mm scale without the need for invasive sampling. While highly resolved laboratory experiments may advance dispersion theory a key limitation is the field application. Presently, it is very difficult and expensive to detail subsurface heterogeneity. Current methods of numerically quantifying aquifer heterogeneity often involve either sophisticated interpolation techniques or stochastic realizations; both methods require invasive point sampling on which to condition estimates. To address this we investigate the merits of surface ground-penetrating radar (GPR) and the development of new space-local spectral transform texture recognition techniques (applied to glaciofluvial outcrop photographic data sets) for the nearly continuous, remote determination of K and its spatial distribution. More traditional (slug, core, and pump tests) multi-scale sampling of heterogeneity in another glaciofluvial aggregate pit reveal a 3 order of magnitude difference in K estimates depending on the scale of investigation. Thus quantifying the variability and effects of sampling scale are critical in order to properly assess the impact of thermal ground water plumes on adjacent cold water streams.