Date of this Version
The Theis method and its derivative Cooper-Jacob method are commonly used for determining hydraulic conductivity (K) for aquifer studies using two-dimensional (2-D) groundwater modeling. Theis and derivative equations assume isotropic, homogeneous aquifers and horizontal flow. In nature, saturated permeable layers of an aquifer are often separated by less permeable layers or lenses, which commonly have K's that are lower by several orders of magnitude. The presence of such geologic features induces vertical flow during aquifer tests violating assumptions for Theis based methods. This study assesses the appropriateness and error of using Theis based equations to determine K for use in 2-D groundwater modeling, which assumes horizontal flow.
For this study, a 3-D groundwater model and constructed hypothetical aquifer configurations were used to simulate pumping tests and produce hydraulic head measurements in three dimensions. A regression analysis computer program based upon the Cooper-Jacob method was used to determine calculated transmissivity (T) and K values in the 3-D framework.
The fully and partially screened pumping well simulations for isotropic, homogeneous aquifers yielded calculated K's within a 15% error margin. Calculated K's for partially screened aquifer configurations were slightly lower due to vertical flow induced by the partially screened pumping well. Analyses of constructed anisotropic aquifers were performed with a partially screened pumping well and partial and complete low K layers (three orders of magnitude lower) resulting in lower calculated K's and error in screened layers and significantly higher calculated K's and error in unscreened layers.
This study demonstrates that assigning K for a saturated thickness using traditional aquifer pump tests and based upon traditional Theis based methods remains inappropriate for use in 2-D groundwater models because of anisotropy and heterogeneities common in most aquifers. This is especially true when data collected are dependent upon the location of the observation point and consequently error can be quantified for calculated K at these locations.
Advisor: Darryll T. Pederson