2007 GSA Denver Annual Meeting (28–31 October 2007)

Paper No. 25
Presentation Time: 1:30 PM-5:30 PM

COMPARISON OF A FINITE-DIFFERENCE MODEL TO AN ANALYTIC ELEMENT MODEL: RESULTS, COSTS, AND BENEFITS


RAYNE, Todd W., Department of Geosciences, Hamilton College, 198 College Hill Road, Clinton, NY 13323 and BRADBURY, Kenneth R., Wisconsin Geological and Natural History Survey, 3817 Mineral Point Rd, Madison, WI 53705, trayne@hamilton.edu

Selecting an appropriate simulation method for specific problems is part of the “art” of groundwater modeling. This selection can be particularly important in studies of stream-aquifer interaction. For such situations, analytic element (AE) models offer excellent, scale-independent resolution of small hydrologic features but are limited in their ability to represent hydrogeologic heterogeneity and 3D flow. In contrast, finite-difference (FD) models like MODFLOW offer significant power to represent heterogeneity, but suffer from grid discretization issues near small features like streams. We compared a three-dimensional FD model (MODFLOW) and a two-dimensional AE model (GFLOW) in simulations of groundwater flow and groundwater – surface water interactions along a small trout stream in central Wisconsin. While the models were not constructed simultaneously, they were constructed independently (i.e. the AE model was not used as the basis for the MODFLOW model). After the models were completed, we compared them on the basis of the time and ease of model construction, the results of the models, and the difficulty of model calibration.

In the test drainage basin, the models produced very similar results. Both models produced head distributions and stream flow values that showed good agreement with field data. The MODFLOW model had more parameters that could be adjusted to achieve a calibrated model, but the calibration took longer because of this. The GFLOW model took substantially less time to create and calibrate than the MODFLOW model, but lacked three-dimensional flow and detailed heterogeneity of hydraulic conductivity and recharge that we incorporated into the MODFLOW model.

We conclude that the choice of modeling method depends on the amount of time and data the modeler has and the goals of the modeling effort. Three-dimensional FD models are appropriate in cases when a large amount of spatial variability in hydraulic conductivity or recharge exists, or when the model is used to study vertical flow in layered aquifers with different properties. An analytic element model is appropriate when the spatial variability of parameters is not known or when the scale of horizontal flow is much greater than the vertical.