Impact of Saline Tracer Mixing on Upscaled Electrical Resistivity Under Partially Saturated Conditions: Insights From a Pore-scale Fluorimetry Study

International audienceTime-lapse electrical resistivity tomography is a widely used geophysical method to remotely monitor water saturation and the migration of contaminant plumes. This is achieved by driving a known electrical current between an electrode pair while measuring the resulting voltage between another electrode pair. The electrical resistivity structure of the subsurface can be estimated by inversion of multiple current injection and voltage pairs. The effects of heterogeneous solute concentrations (i.e., fingering of a saline tracer) below the resolution limits of the tomogram are commonly ignored (i.e, the solution is assumed to be perfectly mixed below this scale). We have adapted an experimental set-up to study the effects of sub-resolution solute heterogeneities on the effective bulk electrical resistivity. We used a 2D analogous porous medium consisting of a Hele-Shaw cell containing a single layer of 4500 cylindrical solid grains built by soft lithography. We monitored the bulk electrical resistivity at a temporal resolution of 2 s. At the same time, we monitored the spatial distribution of the water/air phases and the saline solute concentration field in the water phase using a fluorescent tracer injected together with the saline solute and a high-resolution camera (27 pixels per mm, 12 bit images). We performed saline tracer tests under full and partial water saturations. The unsaturated flow was imposed by jointly injecting air and water. The bulk resistivities measured at the scale of the medium were confronted to resistivities computed numerically from the measured spatial distributions of the fluid phases and the salinity field. We find that the air distribution, saline tracer channeling and fingering, and mixing phenomena all result in large changes in the measured and simulated bulk resistivities by creating preferential flow paths or barriers for electric current at the pore scale. Based on these findings, we postulate that sub-resolution effects contribute to one of the most important inconsistencies in conventional time-lapse ERT imaging, namely the apparent loss of tracer mass