[ << Band/volume 23 ] [ Freiburger Schriften zur Hydrologie ] [ >> Band/volume 25 ]

Band/volume 24: Leibundgut Ch. & Uhlenbrook S. (Eds) (2007):

Abflussbildung und Einzugsgebietsmodellierung

The interplay of different runoff generation processes in a catchment is highly complex and can not be evaluated realistically without experiments. At the same time, process knowledge is insufficient due to accentuated spatial and temporal variability, which make process-based modelling difficult. In combination with traditional experiments, tracer methods have shown a great potential to assess runoff generation processes in the catchment-scale. But for a long time, systematic analyses following a nested approach have not been undertaken.
In the field of catchment modelling much progress has been achieved in recent years. Thereby, the supply with catchment-wide data, sufficient in spatial and temporal detail, is a great problem. In practice often simple, conceptual rainfall-runoff models with a reduced data demand are used. In these models, however, concepts for runoff generation processes are often too simple to serve as a basis for process-based solute transport modelling.
Another great difficulty in catchment modelling is model uncertainty, which has several reasons. Essentially, calibration of parameters, which are not available or measurable in the meso-scale, has to be done. Resulting parameter uncertainty can cause model equifinality, i.e. similar model results are reached by different parameter sets. Such uncertainties do not correspond to model requirements. Therefore, the validation of models is of special importance.
The overall projects aim of the research project „Abflussbildung und Einzugsgebietsmodellierung –“ Runoff generation and catchment modelling” was to accept the above challenges and to provide a better, process oriented description of runoff generation in meso-scale catchment models. Experimental methods, e.g. tracer-tests combined with classical hydrometric measurements, had to be developed to identify runoff generation processes in the micro- and meso-scale. As a result, realistic descriptions of runoff-generation processes were thought to enable improved (less uncertain) rainfall-runoff modelling in the meso-scale. Finally these models were foreseen to contribute to the reduction of uncertainties in predictions of global change effects.
To reach this goal, four project teams (PTs) with specific knowledge and well equipped experimental sites covering a wide geographical range were combined: PT Freiburg i. Br., PT Zittau, PT Potsdam/Wien and PT Bochum. Investigation areas were located in the southern Black Forest (Brugga and Dreisam), in the “Sauerland” (Husten), in the Ore Mountains (Rotherdbach, Wernersbach and Mandau) and in the Eastern Alps (Limbergalm, Löhnersbach and Saalach).
In these investigation sites, all belonging to the lower meso-scale, the four project teams carried out process-hydrological research and applied enhanced catchment models. Thereby, every project team introduced its own expertise, based on long term knowledge in experimental methods and modelling techniques. A special challenge was caused by the high variability of catchment  properties between the “Sauerland” and the Alps. Thus, in the natural diversity of runoff generation processes as well as in experimental and modelling methods, significant synergy could be reached.
Major findings in hydrological reactions of small-scale subsystems like periglacial drift covers, riparian aquifers, saturated areas and channels were obtained by experimental investigations. This was reached by the parallel application of different methods (e.g. groundwater and soil water monitoring, artificial and natural tracers, hydrometric methods). Within the entire project, more than one hundred tracer experiments (natural and artificial) were carried out to study water flows and to parameterize models later on. In all research areas the special relevance of subsurface flow processes (e.g. lateral interflow, different groundwater components) for flood generation could be proved. Depending to local conditions like snow melt, variation of hydrological conductivity with depth, shape of slopes, etc., the generation of lateral hillslope runoff varied. For the generation of overland flow, the distribution and location of saturated areas was found to be of greater importance than Hortonian overland flow (infiltration excess runoff). This conclusion was based on observed events during the four years project period, during extreme events, a higher relevance of Horton overland flow is probable.
Additionally to small-scale findings, the effects of catchment size on spatial heterogeneity of natural tracers and on the systematic hydrological response to events was measured and described by meso-scale experiments. Processes only relevant in the meso-scale could be detected and compared between meso scale sites. Overall insights into the hydrological response of catchments, which are composed of many different subsystems, were reached. In summary, results underline the importance of both micro and meso-scale process knowledge for process-oriented, meso scale rainfall-runoff modelling. The experimentally acquired process knowledge was regionalized by the WBS-system and integrated into process-oriented catchment models (see below).
Resulting from investigations of water fluxes in hill slopes of the Brachtpe catchment, a physically based approach to describe lateral water flows in hill slopes based on the kinematic wave could be developed, which was based on a combination of experimental and modelling approaches. Soil, landuse, geology, and finally the relief (shape of slope and gradient) played major roles.  The numeric simulation of water flows integrated geomorphometric factors and was carried out by a finite-differences-method, discretized following the explicit Lax-method.
Two process-oriented, distributed catchment models for the meso-scale were developed (especially by the working groups Freiburg i. Br. and Potsdam/Wien). Both are discretized with regionalized parameters based on GIS and WBS and simulate hydrological processes spatially and temporally in high resolution (50 x 50 m² and 200 x 200 m², respectively, with temporal scales of hours or shorter). In mathematical concepts and in space-time discretization both models are comparable. In comparison to the approach of PT Freiburg, the model of PT Potsdam/Wien describes overland flow more physically based, while lateral subsurface storm flow is less complex and process-oriented. Both models could be validated including additional experimental data (multi-response validation). In particular, in the model TACD, reduced the uncertainty in runoff simulations, due to integrating additional data. To check the plausibility of model structures, comprehensive sensitivity analyses were undertaken. Results suggested that the quality of hydrological models depends strongly on adequate input-data, especially precipitation. To reduce model uncertainty by multi-response or multi-scale data, different objective functions were combined.
Despite considerable progress in the knowledge of runoff generation, further research is needed. In meso-scale catchments experimental methods are now well developed, but can still be improved by new technologies. The vision, to apply models to catchments of completely different characteristics without complex experiments, has yet not been realized.