Understanding solute and heat transport processes in aquifers is crucial for monitoring and protecting the groundwater critical zone. Crosshole ground-penetrating radar (GPR) is a useful method for enhancing transport characterization in aquifers. Particularly, GPR full-waveform inversion (FWI) can provide subsurface images with resolution at the subwavelength scale, making it a well-suited method for monitoring tracer plumes. While previously applied to a salt tracer test, its effectiveness for other tracer types with different
conductivity and permittivity contrasts remains unexplored. Here, we apply this method to two tracer tests conducted at the Krauthausen test site in northwest Germany: (1) a natural gradient heat tracer test and (2) a forced-gradient combined heat-salt tracer test. Both tracers use hot water, which provides a contrast in both electrical conductivity and dielectric permittivity with the groundwater. This should permit improved monitoring capabilities with GPR
FWI in comparison to the salt tracer, which only provides a contrast in electrical conductivity. However, this comes with new challenges in the processing workflow, specifically, in the starting model strategy. The results illustrate that using the ray-based permittivity tomogram and
a homogeneous conductivity as FWI starting models for each time-lapse dataset is the best strategy. Additionally, we applied an amplitude analysis approach to improve the starting model in regions with low ray coverage. The effects of the heat tracer were detected over the entire depth range of the aquifer, from 3-11 m, and especially at mid-aquifer depths from 6-8.5 m. For the heat-salt tracer, we were able to detect separate effects from the salt and heat: the salt was observed at depth in the aquifer, in accordance with the salt tracer test, and the
heat was observed approximately two days after the salt effects at mid-aquifer depths (from 5.5-7.5 m). In regions where minimal effects from the tracer were observed, specifically for the heat tracer, the consistent results between independent time-lapse datasets demonstrate the repeatability of the method, indicating the suitability of GPR FWI for hydrogeophysical time-lapse studies.

Ground penetrating radar (GPR) data has proven to be a powerful tool for non-invasively estimating the electrical properties of the shallow subsurface. Petrophysical relationships allow them to be converted into hydrological parameters such as soil water content (SWC) and porosity. In the presence of thin layers, complex wave phenomena can occur that are usually ignored in standard processing. Recent developments in modeling tools and computation power allow us to include detailed modeling in advanced inversion algorithms, such as full-waveform inversion (FWI).
In this thesis, GPR wave phenomena occurring due to the presence of boreholes in crosshole tomography and the presence of permittivity gradients in surface GPR data acquisition will be investigated. First, these phenomena need to be identified and understood so that they can be effectively included in advanced inversion algorithms.
Using the finite difference time-domain solver gprMax, various models with different borehole radii and different permittivity distributions are evaluated. Additionally, a resistor loaded finite-length antenna model is built to compare its radiative properties with a point dipole source in crosshole GPR applications. The results reveal that the differences between these two antenna types are small, but not equal. The dominant factor that influences the radiated wavefields is found to be the properties of the surrounding medium. For increasing vertical offsets, elongated wavetrains are observed, which are probably caused by waveguiding effects of the water-filled boreholes. The permittivity gradient models for surface GPR reveal scenario-dependent phenomena, notably the presence of a lower halfspace ground wave and dispersive wave propagation for increasing and decreasing permittivity gradients, respectively. Overall, the results suggest that integrating parameters into the FWI workflow, which account for surface gradients, could be beneficial and lead to improved imaging capabilities.