Targeted inversion of airborne electromagnetic data to inform the development of a groundwater model: Example from the Kaweah Subbasin, California

A groundwater model is the foundation on which to build effective groundwater management. We are exploring ways in which data acquired using the airborne electromagnetic (AEM) method can be used to support the development of the groundwater model and improve the accuracy of the model predictions. We incorporate the AEM data by designing the inversion of the data to meet specific needs in model development, using as an example, the development of the groundwater model in the Kaweah groundwater basin, located in California's Central Valley. The region's existing groundwater model was constructed using primarily lithologic logs resulting in uncertainty in the subsurface structure.

AEM data were acquired, covering the Kaweah subbasin, in 2018. The two questions which we addressed: 1) What is the extent, i.e. depth and thickness, of the Corcoran Clay, a major confining unit? 2) What is the depth of the interface between sediments and the underlying bedrock in the foothills, an important control on mountain recharge entering the valley? A conventional AEM inversion approach (using an L2 norm) yielded a smoothed resistivity model in this area. Transforming resistivity to sediment texture/lithology allowed us to identify the presence of units dominated by sand/gravel or clay, and bedrock. However, due to the smoothed model, the Corcoran clay was mapped to be much thicker than was realistic and there was no clearly defined bedrock surface. Improvements can be made by incorporating prior information and using a more targeted inversion approach. Knowing that the thickness of the Corcoran clay was much thinner (~ 20 m) than the surrounding aquifers (>200 m), we designed a regularization function constraining the resistivity distribution to be dominated by a resistivity value corresponding to the aquifers and allowing for outliers that could be either more conductive or resistive (the use of sparse norms). This improved our estimates (based on comparison with well data) of the depth and thickness of the Corcoran Clay layer. Our ability to determine the depth to the bedrock was improved by recognizing that there would be a large resistivity contrast at the bedrock surface. This surface was identified by allowing a sharp boundary only if there was a need to have this resistivity contrast for fitting the AEM data.