Abstract
The Macrodispersion Experiment (MADE) at Columbus Air Force Base (MS, USA) was initiated in the mid-1980s and aimed to study solute transport in highly heterogeneous porous media by conducting large-scale natural-gradient tracer experiments. A review of the original field tracer experiments reveals several issues that were not addressed in most modeling efforts. These issues include: non-stationary flow; significant questions regarding the reliability of reported hydraulic conductivity values; a significant mass imbalance (23-50%) between the injected and observed tracer; a three-dimensional architecture based on sedimentological information; and vertical hydraulic head gradients. This paper demonstrates how these issues can be integrated into a knowledge framework that systematically assesses the knowns, unknowns, and confidence levels. Using the knowledge framework, we generate a set of multi-conceptual models as a way forward for a holistic approach for an improved understanding of the processes that affect the interpretation of measured tracer concentrations at the MADE site. Our purpose for applying the workflow at the MADE site is twofold. First, to provide a constructive dialogue towards untangling several unresolved issues associated with modeling the MADE tracer experiments. Second, to illustrate how the application of a knowledge framework coupled with multi-conceptual models can support a holistic approach for understanding flow and transport at highly heterogeneous sites.