Die Identifikation hydrologischer Prozesse im Einzugsgebiet des Dürreychbaches (Nordschwarzwald)

The first part of this work deals with the most important runoff generation processes and their spatial and temporal distribution in the study area, the catchment of the Duerreychbach, Northern Black Forest, Germany: The perennial springs generate the entire base flow in the study area. A direct interaction of base flow and surface runoff could not be proven. Only for extreme events ( > HQ1) and high ground water levels a substantial contribution of deep ground water to runoff could be measured in the lower part of the catchment. The calculation of a closed water balance for the catchment outlet is not possible: Large amounts of groundwater (approx. 400 mm a) are exported. Runoff from the alluvial plain and the tracks near the stream (saturation excess flow and Hortonian overland flow) are the fastest runoff components. But, with a maximum contributing area of 1.2%, runoff generated in the alluvial plain dominates the hydrograph only for dry pre-event conditions. The well drained plateau area shows a fast runoff reaction. For medium moisture conditions these contributions dominate the runoff at the catchment outlet (50-80%). Caused by its high DOC concentration (brown colour, scum) water from the plateau areas can be visually detected. With longer event duration the contributing area grows and also areas with lower connectivity to the drainage network start to contribute to runoff. The abundance of iron pans in the hillslope soils (mostly Podsols) reduces the percolation rate of the soils and causes fast interflow in the macroporous humic layer for larger rainfall events. In depressions or where the soil is cut by tracks this flow process (“Pipe Flow”) becomes visible. Growing event duration and high intensities rises the portion of interflow within the total runoff at the catchment outlet. However, the portion of runoff from the plateau region decreases. While runoff from the plateau region dries up quite fast, the tendency of interflow generation on the hillslopes remains for a certain time: Relatively slow interflow on top of the iron layer keeps the saturation of the soil high. The infiltration capacity of the upper horizons of the forest soils is very high. Only for very wet pre-event conditions or extreme events a strong reaction of the hillslope can be expected. Runoff production on the hillslope showed a strong correlation to TDR-measurements in depths of 22 and 7 cm in a soil profile on a steep hillslope (Podsol). This location can be seen as representative for areas able to produce fast interflow. A threshold for runoff generation exists also for the saturation areas of the plateau region. Prior to saturation at this location, all rain infiltrates and no runoff is generated. Saturation measured by TDR-probes in 20 cm depth at one location correlates very well with the starting point of runoff at the local gauging station. A dense drainage network delivers soil water very fast. Using isotope and DOC measurements (showing a stable isotope composition and a high and stable DOC-concentration) it could be concluded that (for wet pre-event conditions) the runoff consists mainly of pre-event water which is very well mixed with event-water during the runoff process. But nevertheless the storage capacity (for dry pre-event conditions) of the plateau area is very high. Locations without horizons of low infiltration capacity sometimes show extremely high infiltration rates, caused by its sandy, macroporous soil properties (mostly Cambisols or Cambisols-Podsols). Surface runoff could only be observed in the case that large contributions of hillslope water and return flow in depressions cumulates. For these conditions different erosion processes were active: Besides subrosion (wash out within the upper soil layer) also rill and gully erosion along natural channel structures or tracks was observed. During the largest observed flood (HHQ, 28./29.10.1998) small landslides were generated. The main channel was completely rebuilt. Large amounts of stones and blocks were transported over long distances. A 150 year old track has been destroyed. The structure of the main channel remains stable for events with a recurrence interval of 5 or less years. In the second part of this study three numerical simulation models with a different degree of abstraction were applied to determine their principal suitability for the simulation of the observed processes: Concerning the fast processes (e.g. saturation excess flow and fast interflow) the two conceptual approaches showed disappointing results. On the one hand this was caused by their temporal resolution (one of the models uses daily time steps). But on the other hand this was caused by the underlying concepts which do not allow to maintain a spatial relation between model and reality (spatial resolution too coarse) or which do not reflect the real processes at all (e.g. interflow in hillslope caused by layered soil). Simulations based on the TOPMODEL-approach by BEVEN & KIRKBY (1979) were very disappointing. Neither the spatial distribution of the Index values fitted the observed runoff generation processes nor the simulation of saturation excess flow independently from the local groundwater table was possible. The physically based approach was very flexible. The relation between process and model scale was closer to reality. It was possible to find a strong correlation between observed processes and simulation results. For more satisfying simulation results further parameter studies are necessary. The results depended very much on the spatial discretisation of the model (length and width of the hillslopes; distances between simulation nodes). In addition this simulation model lacks a groundwater part. Only a detailed process knowledge enables the modeller to evaluate simulation results. But an increasing degree of model abstraction results in loss of spatial relation. This makes point measurements, tracer data or process observations less useful for model parameterisation, calibration and evaluation. Because of the high complexity of natural systems the transition from explanation of system behaviour (by “physically based” models) towards a pure behaviour imitation (by “black box” approaches) is necessary, especially for the operational application of models. But in this case it must be guaranteed that those models are not used for extrapolation purposes.