Improved hydrogeophysical imaging with ERT using direct push data as priors and geostatistical regularization

Imaging saltwater/freshwater interfaces is of importance to understand flow and transport in coastal aquifers. For hydrogeophysical imaging the method of electrical resistivity tomography (ERT) yields high-resolution images and is suited for monitoring. However, there is an intrinsic ambiguity in the inversion of the data that limits accurate quantification. In contrast, direct push (DP) data provide accurate point information. Moreover, DP fluid sampling helps to transfer the measured electrical conductivity into salinity by computing a spatially variable formation factor. Both data show typically the same structures but contradict in detail.
We present a methodology to combine both methods using joint inversion. To this end, DP data are treated like geophysical data with standard deviations derived from statistics so that the resistivity distribution strives to match the DP data at the given point. Additionally, the spatial distribution of DP data can be used to derive geostatistical correlation lengths in the horizontal and vertical directions that are incorporated into the ERT inversion using an anisotripic geostatistical regularization operator. Synthetic modellings with geostatistical media show that the ERT image is improved, not only in the vicinity of the sampling points.
We present data from the northern beach of the North Sea island of  Spiekeroog where we want to image the circulation cell that formes under the influence of the tides (upper saline plume). There is a significant improvement of the classical smoothness-constrained inversion compared to the DP-guided inversion. As a result, we observe a split into several circulation cells with local groundwater discharge zones. Our hypothesis is that these are mainly driven by the changing beach morphology.