Incorporating Prior Information In Time-Lapse ERT Inversions Using Conditional Regularization Constraints 744753

Electrical geophysical imaging methods have become commonplace for characterizing and monitoring subsurface processes according to their electrical properties or changes therein. The inherent non-uniqueness in geophysical imaging data requires that some form of additional constraints be provided to the inverse problem to achieve meaningful results. That information most commonly comes in the form of regularization constraints that impose parsimonious solutions. Often, further information exists concerning the subsurface exists that is conditional in nature or cannot be quantified. For example, it may be known that the electrical conductivity in some region of the subsurface is greater than some lower bound over some time period. In this talk, we demonstrate how such conditional information can be used to constrain time-lapse ERT imaging results using iteratively reweighted least squares. The algorithm accommodates multiple types of simultaneous constraints to enable multiple types of prior conditional information to inform the inversion. We demonstrate the utility of the approach in terms of improving resolution and image interpretability with three field scale examples; 1) 4-D image of stage driven groundwater/surface water interaction, 2) 4D imaging of fluid flow through a stimulated fracture zone, and 3) real-time imaging of chemical amendment injections for remediation of subsurface contamination.