RECENT RESEARCH ACCOMPLISHMENTS
1. How do hydraulic-conductivity measurements relate to the values required by models? How can they be used most effectively to constrain model development?
An approach to investigating hydraulic conductivities that had not previously been pursued is to combine extensive data sets from complex field systems and controlled laboratory experiments with sophisticated methods of sensitivity analysis and nonlinear regression to evaluate carefully how measured values of hydraulic conductivity relate to the in-situ hydraulic behavior of the materials.
This approach provides a new perspective because it tests directly the relation between measured hydraulic conductivities, boundary conditions, and so on, and values required to reproduce measured hydraulic heads, flows, and transport using models. This approach requires an unusual amount of information about a system to provide conclusive results.
The resulting understanding of the relation between measured and in situ hydraulic properties is needed to understand how to use measured values of hydraulic conductivity either in deterministic or stochastic models. For example, it is important to understand the error in the measured values relative to the variability that occurs in situ in subsurface systems to understand what is really being accomplished when using the measured values in either deterministic or stochastic methods.
Since 1995 I have investigated this problem with students Heidi Christiansen Barlebo of the Technical University of Denmark using an unusually extensive field data set, and Gilbert Barth of the University of Colorado using controlled laboratory experiments. The field data set is from the Columbus Air Force Base MADE experiment.
The laboratory device Gilbert Barth and I used to investigate hydraulic-conductivity measurements is, at 10-meters long, 1.2-m high, and 6-cm thick, the largest such device used to date for controlled studies of flow through porous media. Clean sands of five distinct sizes were packed in a random distribution. Results held three surprises.
Last Modified: April 14, 2008