A WATER INJECTION EXPERIMENT IN THE VADOSE ZONE: THE USE AND VALUE OF NON INVASIVE CROSS-HOLE DATA FOR MODEL CALIBRATION

Abstract

The characterization of the deep vadose zone is traditionally a difficult task, since access without disturbance of samples or modification of in-situ conditions is practically impossible. On the other hand, the presence and flow of water in the vadose zone controls a number of phenomena of great environmental interests, such as contamination of water resources, catchment hydrology, floods and slope stability. As a consequence, non invasive – geophysical - techniques, such as electrical resistivity tomography (ERT) and ground-penetrating radar (GPR), are being increasingly used to investigate the vadose zone. The use of cross-hole methods is particularly effective and increasingly applied. The dependence of the geophysical response on changes in soil moisture content, e.g. via changes in electrical resistivity or dielectric properties, is the key mechanism that permits the use of non-invasive techniques to monitor the vadose zone in time-lapse mode, i.e. via repeated measurements over time. The data from non-invasive techniques can be subsequently used to calibrate physical-mathematical models of water flow in the unsaturated zone. In this paper, we present the results of a forced infiltration experiment conducted at the research site in Gorgonzola, close to Milan, Italy. The site has been equipped to monitor the unsaturated flow dynamics in the local alluvial quaternary sediments of the Po river plain, characterized by a fairly coarse sandy-gravel grain size distribution. Four 20 m deep boreholes were drilled at the site; three of them are permanently equipped with stainless steel borehole electrodes for ERT imaging. Each borehole has a set of 24 electrodes, spaced 0.8 m vertically. The same boreholes are also being used for cross-hole GPR tomography. Both MOG (multiple offset gathers) and ZOP (zero offset profiles) have been adopted as configurations for GPR surveys. Conversion from GPR traveltimes to moisture content values has been provisionally performed using the semi-empirical relationships. ERT has been used so far only to provide qualitative images of changes in resistivity due to water saturation changes. Inversion of ERT data has been performed using the CRTomo 2D code (Kemna, 2000), while the code Migratom (Jackson and Tweeton, 1994) was used for the inversion of GRP MOG data. ZOP profiles have been analyzed under the assumption of straight ray horizontal propagation. Monitoring of natural infiltration processes started in January 2005. In July 2005 an artificial water injection experiment was performed using a 2 m deep, 2.60 m long trench. A total of 3 cubic meters of tap water infiltrated in about 2 hour time. A number of ERT and GPR MOG datasets were acquired from July 5 (date of water injection) to July 11, 2005. The data were used to calibrate a 3D unsaturated flow model, thus helping the identification of field-saturated hydraulic conductivity and other unsaturated soil parameters. Strengths and limitations of the methodology are analysed and discussed.