46
conductances, and then the K, values. The PEST runs also helped find targets that were in
the wrong location or layer. The elevations of the well screens for the target wells were
estimated from depth below ground surface and surface elevations, which did not necessarily
correspond to the Troy Valley model surface elevations. Therefore, these wells were then
moved so that they pumped from the correct hydrofacies unit. Two wells were measured
only once and one well had over 10 feet of variance in the measurements. The PEST runs
indicated these targets needed to have a weight of zero, which means they are highly
uncertain. Hence, they were removed for the manual calibration. The PEST results also
suggested that using a uniform anisotropy ratio of Kz/K, < 1 where K, is the vertical
hydraulic conductivity and Kx is the horizontal hydraulic conductivity in both the x and y
directions, for all four hydrofacies, was appropriate. This was based on the parameter
sensitivities calculated by PEST, which consistently had Kx values 1-4 orders of magnitude
higher than the K, values. Finally, Kx and K, values for the three bedrock units that were
determined by PEST were used and held constant during manual calibration. More
information on the PEST simulations can be found in Appendix 3.
During manual calibration, Kx of each unit was held constant and values of the
anisotropy ratio Kz/Kx of 1:1, 1:10, 1:100, and 1:1000 were tested. The final calibrated
model used an anisotropy ratio of 1:10 for the four hydrofacies. Testing of river and lakebed
conductance indicated the model was not sensitive to these and the calibrated K, values of
hydrofacies 1 and 2, respectively, were used in the final calibrated model. The calibrated
model uses and error closure criterion of 0.001 ft.