Calibration of a texture-based model of a ground-water flow system, western San Joaquin Valley, California

The occurrence of selenium in agricultural drain water from the western San Joaquin Valley, California, has focused concern on the semiconfined ground-water flow system, which is underlain by the Corcoran Clay Member of the Tulare Formation. A two-step procedure is used to calibrate a preliminary model of the system for the purpose of determining the steady-state hydraulic properties. Horizontal and vertical hydraulic conductivities are modeled as functions of the percentage of coarse sediment, hydraulic conductivities of coarse-textured (Keearse) and fine-textured (Kfine) end members, and averaging methods used to calculate equivalent hydraulic conductivities. The vertical conductivity of the Corcoran (Keen) is an additional parameter to be evaluated.
In the first step of the calibration procedure, the model is run by systematically varying the following variables: (1) Kcoarse/Kfine' (2) KcoarsdKcorc' and (3) choice of averaging methods in the horizontal and vertical directions. Root mean square error and bias values calculated from the model results are functions of these variables. These measures of error provide a means for evaluating model sensitivity and for selecting values of Keearse, Kfine, and Kcorc for use in the second step of the calibration procedure. In the second step, recharge rates are evaluated as functions of Keearse, Kcere, and a combination of averaging methods. The associated Kfine values are selected so that the root mean square error is minimized on the basis of the results from the first step.
The results of the two-step procedure indicate that the spatial distribution of hydraulic conductivity that best produces the measured hydraulic head distribution is created through the use of arithmetic averaging in the horizontal direction and either geometric or harmonic averaging in the vertical. The equivalent hydraulic conductivities resulting from either combination of averaging methods compare favorably to field- and laboratory-based values.