More than 70% of the German population rely on groundwater as their daily drinking water supply. Hence, the understanding of processes like fluid flow and solute transport in porous media are of broad relevance. However, subsurface hydrology faces two problems. On the one hand, information about the subsurface structure and hydraulic properties is scarce. On the other hand, many aquifer parameters like the permeability which controls the flow velocity, are heterogeneously distributed in space. This thesis is dedicated to the question of how aquifer heterogeneity impacts on subsurface flow and transport processes at different scales. We first present a large-scale numerical model of the Thuringian basin in order to investigate the mechanisms of saltwater transport. The results underline the fact that permeability and its heterogeneous spatial distribution are decisive factors for the evolving flow and salt patterns. Our results show that the correlation structure and the degree of heterogeneity (variance) impact significantly on the amount of dissolved salt and the location where salt reaches near-surface regions. Thermally induced convection is not present. However, variations in fluid density due to dissolved salt can lead to significant changes in the distribution and amount of salt in the Thuringian Basin. We further focus on a method to determine the statistical parameters which describe the heterogeneous spatial distribution of permeability by analyzing well flow at local scale. This is of interest, because pumping tests are a widely used tool to infer on porous medium characteristics. We derive an analytical solution which describes the drawdown of a steady state pumping test in 3D heterogeneous anisotropic media by making use of upscaling theory. By combining the analytical solution with an inverse estimation strategy we determine the statistical parameters, like mean permeability, variance and correlation length from on-site pumping tests.