Aquifers are the primary sources of clean drinking water. Pollution in aquifers is one of the most challenging and important environmental problems. It is not only extremely complex to map but also difficult to remediate. Flow and transport of water and pollutants in porous media requires detailed characterization of the properties of the media. The main property which controls the flow and transport is hydraulic conductivity (K), which can be defined as the ability of the media to let the water flow through. Intensive studies to map the distribution of hydraulic conductivity are necessary to model the plume migration. Conventional in-situ aquifer characterization techniques are invasive and lack the necessary high resolution. Therefore, novel methods are required to improve the methods to monitor and simulate the flow and transport through aquifers.This study introduces a combination of novel techniques to provide the necessary information related to porous media. The proposed method was tested at a highly heterogeneous site called the Macro Dispersion Experiment (MADE) site in Mississippi. The MADE site is a very well studied site where several large scale tracer tests were conducted in the 1980s and 1990s. The tracers used for these tests were monitored using more than 300 multi-level sampler (MLS) wells. Concentration measurements showed that the majority of the mass stayed near the injection area, whereas minute concentrations were measured further down-gradient. This behavior is significantly different from the simulations created using models based on the Advection-Dispersion Equation (ADE). This behavior and the inability to explain this using most models has led to a major debate in the hydrologic science community.The hypothesis of this study is that the ADE based models can reproduce simulations of the measured transport when the models are parameterized with sufficient high-resolution hydraulic conductivity data. Two novel high resolution characterization methods, the direct-push high resolution hydraulic conductivity (HRK) tool and 3D full-resolution ground penetrating radar (GPR) were combined to generate 3D K fields using fractal stochastic methods. This study demonstrated that the complementary geophysical data can be used to reduce the K variance by dividing the aquifer into hydrofacies. This approach, in combination with a fractional differencing filter, simplifies the statistically complex distribution of K. Fractional differencing was also capable of removing the long range dependence in vertical K profiles to investigate the underlying K distribution. The 3D K fields were then used to test the ADE based modeling approach at the site and resulting concentrations were compared to one of the large scale tracer experiments. The simulations in this study resulted in mass distributions comparable to those measured during the tracer test experiments. They successfully reproduced the extent of the plume in both 1D and 2D using K fields based on solely field data. Additional tests emphasized the importance of high-resolution data to parameterize K models to successfully simulate flow and transport using the ADE model.
Read
