An experiment of non-invasive characterization of the vadose zone via water injection and cross-hole time-lapse geophysical monitoring

The characterization of the vadose zone, i.e. the part of the subsurface above the water table, is a challenging task. This zone is difficult to access with direct methods without causing major disturbance to the natural in-situ conditions. Hence the increasing use of geophysical methods capable of imaging the water presence in the vadose zone, such as ground-penetrating radar (GPR) and electrical resistivity tomography (ERT). This type of monitoring can be applied both to processes of natural infiltration and to artificial injection (tracer) tests, by collecting multiple data sets through time (time-lapse mode). We present the results of a water-injection experiment conducted at a test site in Gorgonzola, east of Milan (Italy). The site is characterized by Quaternary sand and gravel sediments that house an extensive unconfined aquifer, potentially subject to pollution from industrial and agricultural sources. ERT and GPR profiles were acquired in 2D cross-hole configuration and time-lapse mode over a period of several days preceding and following the injection of 3.5 m(3) of fresh water in a purpose-excavated trench. A 3D model of the water- infi Itrati on experiment was calibrated against the time-lapse cross-hole data, particularly focusing on the ability of the model to reproduce the vertical motion of the centre of mass of the injected water as imaged by GPR and ERT. This model calibration provided an estimate of the isotropic hydraulic conductivity of the sediments in the range of 5-10 m/d. However, all isotropic models overpredict the measured excess of moisture content, caused by water injection, as imaged by GPR. The calibration of anisotropic models for the vertical hydraulic conductivity, with the horizontal hydraulic conductivity determined by direct measurement, also leads to a good fit of the sinking of the centre of mass, with a better mass balance in comparison with field data. The information derived from the experiment is key to a quantitative assessment of aquifer vulnerability to pollutants infiltrating from the surface.