Three cross-borehole geophysical methods are used to image water migration in the unsaturated zone after a point injection of water. Mass balance calculations and moment analysis highlight the differences in resolution between the methods. The choice of moisture content threshold value significantly influences results of the moment analysis.
We compare results of three cross-borehole geophysical approaches for imaging tracer migration arising from a point injection of water in the unsaturated zone: three-dimensional electrical resistivity tomography (ERT), two-dimensional ground-penetrating radar (GPR) tomography and quasi-three-dimensional GPR tomography. In the studied field experiment, a tracer was injected for a period of 5 d and was monitored both during injection and for 5 d during the subsequent redistribution. The three methods show similar characteristics of the plume development and movement, which has a strong lateral component and slow vertical migration. In addition to revealing the main tracer plume, two-dimensional GPR and the quasi-three-dimensional GPR results show development of secondary plumes at depth, which are not captured by the three-dimensional ERT due to lack of resolution. The flow patterns are compared to geological information from a coring obtained at the site and it is concluded that the diversion of water in the lateral direction can be caused by a few thin layers of contrasting geological composition. Mass balance calculations based on moment analysis of the moisture content changes reveal that two-dimensional and quasi-three-dimensional GPR results show similar results and that three-dimensional ERT underestimate the amount of tracer substantially. Our results further show that the analysis volume as well as threshold value for moisture content increase has significant impact on computed mass recovery. The choice of threshold value, in particular, should be method-dependent and needs to be considered carefully if the results of the moment calculations are to be used in constraining hydrological models.