SUCCESSFUL AND UNSUCCESSFUL APPLICATIONS OF INVERSE METHODS ON A REGIONAL GROUNDWATER MODEL

The authors modified and recalibrated a regional groundwater model for the United Water Conservation District (United) in Ventura County, California with the assistance of inverse methods. United planned to use the model to support the Fox Canyon Groundwater Management Plan. However, the model simulated unrealistically high and increasing groundwater levels in the Forebay and Oxnard Plain subbasins that were important areas for groundwater management alternatives. Previous modeling effort included extensive use of inverse methods to calibrate the model, but failed to improve the calibration. As a result of this failure, United developed skepticism about the utility of inverse methods. The authors focused modeling efforts on evaluating and revising the water balance. Only after the water balance reflected the water level data in the area of concern were inverse methods used to refine the calibration.

The authors evaluated the water balance of the model and determined that several aspects of the water balance needed to be revised. The ocean boundary condition was inactive for some time steps, which prevented outflow of groundwater from the basin. The high volume of recharge in the Oxnard Plain did not match the model geologic structure that simulated this area as confined underneath a low permeability layer. Finally, the stream conductance values for streams in the Forebay were too low to reflect the apparent hydraulic communication between the groundwater basin and the Santa Clara River. The failure to evaluate and revise these aspects of the water balance had doomed previous efforts to use inverse methods as an automatic calibration tool of hydraulic parameters. After the water balance was revised, inverse methods had a chance to succeed.

Inverse methods were useful in this situation to quickly calibrate the model after changes to the water balance. This exemplifies how inverse methods facilitate a redefinition of the classical modeling process. The redefined process allows for more iterative revisions of the conceptual model. This use of inverse methods can improve the development and calibration of models, while the marketing of inverse methods as a technique for fixing flawed or poorly posed models will result in client dissatisfaction.