The objective of the present Ph.D.-study is to further investigate unsaturated flow phenomena, which are generally considered complex and unpredictable. Hydrogeophysical methods are used for assessment of the flow and transport processes during four field experiments that were carried out at three different sites in Denmark. The sites at Hjelm Hede, Arrenæs and Voulund are all
characterized by thick unsaturated zones that consist almost entirely of sand. The first experiment was a dye tracer infiltration at Hjelm Hede, which was monitored using high-resolution reflection Ground Penetrating Radar (GPR). Geophysical datasets obtained before and after infiltration were compared to dye staining patterns seen in an excavated profile. It was found that water had
infiltrated in a highly irregular manner, and the GPR data showed that moisture content had increased well below the extent of the dye staining. Second, a point injection experiment was carried out at Arrenæs, in which a water tracer was added at 1.5 m depth for 5 days and subsequently monitored using three- dimensional cross-borehole Electrical Resistivity Tomography (ERT) and two- and three-dimensional GPR. A previous experiment at the site had revealed that the topsoil diverted water laterally, and therefore it was decided to inject water below the topsoil. The tracer plume as imaged by the geophysical data was compared to grain size analysis from the field site and it was found that small changes in grain size and sorting degree were responsible for initiating irregular development of the tracer plume, which expanded more laterally than vertically. Spatial moment analysis was applied and relevant moisture content threshold choices in delineating the tracer plume were discussed. The same threshold value was not found for the three types of data. Mass balance calculations revealed that two- and three-dimensional GPR show similar results and that tracer mass was severely underestimated by three-dimensional ERT. Another forced infiltration experiment was carried out at Arrenæs; this time water was added across a large area at a high rate for 14 days. Moisture content development was monitored using cross-borehole GPR and relative gravity measurements. GPR data were used as a means of ground-truthing the gravity data, which had not previously been used in similar experiments. For the first week of infiltration the mass balances match, but at that time the water front has reached the bottom of the GPR domain, which cannot account for the entire mass anymore. Likewise, an increase in gravity was no longer seen because the measurements are not sensitive to changes beneath this depth. GPR data revealed that flow in the measured domain was highly irregular. A three-dimensional hydrological model with hydraulic parameters and spatial correlation lengths similar to those found at the site was then developed. The moisture content development seen in the model during a synthetic infiltration was compared to actual GPR data to confirm that the same types of flow were present in the model. Next, a suite of similar models were set up, with different spatial correlation lengths but the same hydraulic parameterization. A particle tracking analysis was performed and the arrival time distribution of a suite of particles was compared to a homogeneous one-dimensional flow and transport model with added dispersion. For large spatial correlation lengths, it is concluded that the commonly applied assumption of one-dimensional flow in the unsaturated zone can be valid. In the last experiment reported here, we used cross-borehole ERT for estimation of natural recharge based on long-term monitoring of the movement of a saline tracer at the Voulund field site. ERT
data were compared to samples obtained from core drillings from the site, and it was found that the position of the plume with time was well-resolved in the ERT data, although concentration values were smoothed in the inversion procedure and thus lower than those found in core data. The final value of recharge of 516 mm/year is in agreement with a recharge estimate based on lysimeter drainage measurements from the same site. In order to calculate the natural recharge it was necessary to assume that unsaturated flow was vertical and homogeneous.
Such an assumption can be difficult to justify, however the results from the three-dimensional modelling at Arrenæs show that it may be valid if the spatial correlation lengths found in the (sandy) subsurface are high.