Heterogeneity strongly affects water dynamics in soils and therefore local measurements of hydraulic state variables. Inverse modeling is used to investigate hydrological data obtained from local measurements in heterogeneous virtual soils. Variability, estimated hydraulic properties, and predicted water balances are interpreted in the light of the known truth.

Soils are structured on multiple spatial scales, originating from inhomogeneities of the parent material, pedogenesis, soil organisms, plant roots, or tillage. This leads to heterogeneities that cause variability of local measurements of hydraulic state variables and affects the flow behavior of water in soil. Whereas in real-world systems, the true underlying structures can never be absolutely known, it is appealing to employ synthetic or “virtual” experiments for assessing general properties of flow in porous media and grasping the main physical mechanisms. With this aim, three two-dimensional virtual realities with increasing structural complexity, representing cultivated soils with hierarchical spatial heterogeneity on multiple scales were constructed by the interdisciplinary research group Virtual Institute of the Helmholtz Association (INVEST). At these systems, numerical simulations of water dynamics including a heavy rain, a redistribution, and a long-lasting evaporation period were performed. The technical aspects of the construction of the virtual soils and results of the forward simulations have been presented in a paper by Schlüter et al. (2012). In this follow-up paper, we use inverse modeling to investigate measurements in virtual vertical soil profiles, mimicking typical field monitoring campaigns with moisture content and matric potential sensors placed at five depths. Contrary to the real situation, we can interpret observed data, their variability, estimated hydraulic properties, and predicted water balance in the light of the known truth. Our results showed that measurements, particularly those of water contents, varied strongly with measuring position. Using data from single profiles in systems similar to our virtual soils thus will lead to very different estimates of the soil hydraulic properties. As a consequence, the correct calculation of the water balance is rather a lucky coincidence than the rule. However, the average of the predicted water balances obtained from the one-dimensional simulations, and the estimated soil hydraulic properties agreed very well with those attained from the two-dimensional systems.