Picturing and modelling catchments by representative hillslopes.

Despite the numerous hydrological models existing in hydrology we are limited to a few forms of conceptualization when abstracting hydrological systems into different model frameworks. Speaking in black and white terms, in most cases hydrological systems are either represented spatially lumped with conceptual models or spatially explicit with physically-based models. Physically-based models are often parameter-rich, making the parametrization challenging, while conceptual models are parsimonious, with only a few parameters needing to be identified. But this simplistic mathematical expression is often also their drawback since their model states and parameters are difficult to translate to the physical properties of a catchment. It is interesting to note that both hydrological modeling approaches often start with the drawing of a perceptual model. This follows the hydrologist's philosophy to separate dominant patterns and processes from idiosyncratic system details. Due to the importance of hillslopes as key landscape elements perceptual models are often displayed as 2D cross-sections. In this study we examine whether we can step beyond the qualitative character of perceptual models by using them as blueprint for setting up representative hillslope models. Thereby we test the hypothesis if a single hillslope can represent the functioning of an entire lower mesoscale catchment in a spatially aggregated way. We do this by setting up and testing two hillslope models in catchments located in two different geological settings, Schist and Marl, using a two-dimensional physically-based model. Both models are parametrized based on intensive field data and literature values without automatic calibration. Remarkably we are able to not only match the water balance of both catchments but further have some success in simulating runoff generation as well as soil moisture and sap flow dynamics. Particularly, our findings demonstrate that both models performed well during the winter season and clearly worse during the summer period. Virtual experiments revealed that this was most likely either due to a poor representation of the onset of vegetation in the Schist catchment or due to emergence of soil cracks in the Marl area. Both findings underpin that a static parameterization of hydrological models might be problematical in case of emergent behavior. Additional virtual experiments indicate that the storage of water in the bedrock and not so much the topographic gradient is a first order control on the hydrological functioning of the Schist catchment. We conclude that the representative hillslope concept is a feasible approach in data rich regions and that this form of abstraction provides an added value to the established conceptualization frameworks in hydrology. [ABSTRACT FROM AUTHOR]