Assessing Simulated Transmissivity in Numerical Flow Models of Complex Hydrogeology

Accurately extracting a meaningful transmissivity, a target value within one order of magnitude of field estimates, in numerical models poses a significant challenge when modeling complex groundwater systems. Aquifer transmissivity is directly proportional to the aquifer thickness and the estimated aquifer hydraulic conductivity. In complex geologic conditions (especially in fractured systems) with multiple heterogeneous and anisotropic hydrogeologic units, transmissivity can vary over several orders of magnitude. To extract a meaningful value of transmissivity from a numerical model, a simple five-layer MODFLOW model was constructed. Each layer in the model was assigned a fixed hydraulic conductivity and thickness. The model simulates multiple pumping scenarios with varying combinations of hydraulic conductivity, aquifer thicknesses, and locations of the well screen to simulate aquifer tests. Two extraction methods, the Screen Interval and Layer Summation, were used to extract transmissivity values from the five-layer model, and first compared to model-assigned transmissivity values and then to transmissivity estimated from simulated pumping tests using the Cooper-Jacob Approximation. A similar process was repeated with a more complex groundwater model, the Death Valley regional groundwater-flow system v. 2.0 numerical model. However, in the Death Valley regional groundwater-flow system v. 2.0 numerical model analysis, the Screen Interval and Layer Summation values were compared to field estimated values, not model-assigned transmissivity values. The results of these comparisons show that transmissivity values extracted from a numerical model can vary many orders of magnitude from model assigned transmissivity values, and in the case of the Death Valley regional groundwater-flow system v. 2.0 numerical model, field estimated transmissivity. However, out of the 30 pumping scenarios run, the Screen Interval method transmissivity values were all within half an order of magnitude with the model input transmissivity and was found to be the most sensitive to the variation in transmissivity in both the simple five-layer model and Death Valley regional groundwater-flow system v. 2.0 numerical model. The Screen Interval extraction method may provide the most meaningful comparison of transmissivity between model results and field estimate