Comparison of 2D vs 3D inversions of electric resistivity tomography data over river embankments

Electric resistivity tomography (ERT) is often adopted for the geophysical characterization of river embankments thanks to its high sensitivity to water saturation, infiltration paths and other critical subsurface discontinuities that can result in electrical resistivity variations. ERT data are commonly acquired and processed following a 2D approach. However, the embankment geometry is composed by a crest (that can be not linear) and two slopes, whose topography can be steep. Therefore, the problem to be solve is truly 3D. We created a 3D realistic geometry of a typical embankment and performed 2D and 3D ERT inversions of synthetic data generated from that 3D model for localized anomalies identification. The model and mesh were accurately created in Gmsh, while the ERT forward and inverse modelling were run in ResIPy. We compared 2D and 3D inversions of ERT data generated with different electrode sequences, i.e., dipole-dipole and Wenner-Schlumberger. The main outcomes of our study proved that the anomalies inside the embankment are better characterized by means of a full 3D approach especially if the anomalies are not exactly below the ERT line and if the model geometry is not simple.