Cross-gradients Joint Inversion of Time-lapse Crosshole ERT and GPR Data

Time-lapse geophysical monitoring is a valuable tool in hydrogeology, but the resulting models can suffer from inversion ambiguities. These ambiguities may be reduced by structural joint inversion of multiple geophysical methods. Joint inversion using cross-gradients constraints has been successfully applied to static geophysical measurements. Here we present the first application to time-lapse data. We show on a synthetic and field vadose-zone water-injection experiment how joint time-lapse inversion of crosshole ERT and GPR data can improve the estimated plume shape and the mass recovery. The 3-D layout and coverage of the ERT measurements help to constrain the GPR model, which is based on 2-D measurements; the GPR data are able to resolve sharper boundaries and allow better estimation of changes in water content. The recovered mass fraction within a plume defined by thresholding is 91% for the individual and 99% for joint inversion of the synthetic data. For the field example, the recovered mass fractions are 84% and 91% for individual and joint inversions, respectively. The method appears to works well for water-tracer experiments, but needs further development before it can be applied to flow and transport problems in unsaturated media.