[ << Band/volume 9 ] [ Freiburger Schriften zur Hydrologie ] [ >> Band/volume 11 ]

Band/volume 10: UHLENBROOK S. (1999):

Untersuchung und Modellierung der Abflußbildung in einem mesoskaligen Einzugsgebiet

One objective of this study was to increase the knowledge of the runoff generation processes in the Brugga basin. Therefore source areas of runoff were characterized and the contribution of runoff components during different hydrological conditions were determined. Another objective was to integrate the results of the experimental investigations of the runoff generation into a catchment model. This lead to the development of a more process oriented runoff generation routine. Based on the experimental and preliminary work, the new developed model was applied to the Brugga basin. An evaluation of the modelling results was possible by using additional data (multipleresponse validation).
The Brugga basin is a mountainous basin (438 - 1493 m a.s.l.) with nival runoff regime, located in the southern Black Forest, Southwest Germany. 75 % of the area is wooded, 22 % is used as pasture land. Urban land use is dominant in 3 % of the area. The mean annual precipitation amounts to 1750 mm, generating a mean annual discharge of 1220 mm. The crystalline bedrock consists of gneiss and anatexits. The bedrock is covered by a debris cover, which consists of moraines and periglacial deposits.
In the experimental part of this study hydrograph separations were performed for different events using 18O, dissolved silica and chlorid as tracers. The concentrations of the main anions and cations in discharge and in wells provided further information about the runoff generation processes. Using the environmental tracers 18O and 3H the residence time ofthe water in the different flow systems were evaluated and the amounts of runoff components were determined for aperiod of three years. Three main runoff components were identified:
Direct runoff is generated on saturated areas, sealed areas and boulder trains. It consists of event water and water which was stored near the surface. During short periods of a few ho urs this component can contribute as much as 50 % of total stream discharge, for longer periods (several years) the contribution amounts to somewhat more than 10 %. The aquifers of the slopes contribute about 70 % of total discharge (so-called flow system-2).
18O measurements showed that the mean residence time of the water in these reservoirs is between two and three years. With mechanisms like the piston flow effect and the groundwater ridging effect these reservoirs contribute to flood formation, however they are also important for base flow. The so-called jlow system-l originates from the hilly uplands and crystalline hard rock aquifer and generates mainly base flow. The mean residence time of the water is approximately 6 - 9 years, which was determined using 3H and freons measurements. For a periodof three years the contribution of this component was estimated at 20 %.
Based on the experimental investigations and using different spatial information (i.e. geology, properties ofthe debris cover, topography and further maps) zones with the same dominating runoff generation processes were delineated in the Brugga basin. 'In order to achieve this, a specific method was developed, which accounted for the characteristics of Brugga basin and the available data. The application of this method resulted in a spatial delineation of zones, where it is assumed that the same runoff generation processes dominate. This is the basis of the spatial discretization in the newly developed catchment model TAC. The semi-distributed catchment model TAC (tracer aided catchment model) was developed based on the experimental investigations. The model is a conceptual model, which implies that complex hydrological processes are conceptualized using relatively simple storage routines. The snow routine is based on the degree day method. The soil routine was adopted from the HBV model. The main objective was to develop an improved, process oriented model routine of the runoff generation. Therefore, specific routines were created for all zones with the same dominating runoff generation processes. Concentrations of patural tracers can be attributed to the different runoff components modelled by T AC. The concentrations must be determined by tracer hydrological investigations. Consequently, the simulation of the tracer concentration in the discharge is possible. The quality of the TAC results can be assessed from the agreement ofthe simulated and the observed tracer concentration in the relation to the efficiency of the runoff simulation. An application of TAC in other basins is possible, but a delineation of zones with the same dominating runoff generation routines is required.
The application of TAC in the Brugga basin produced reasonable results. The rainfall runoff modelling on a daily basis was at least as good as the simulations using other conceptual models (i.e. TOPMODEL, HBV, PRMS). The model was validated using an independent period and the quality of the runoff simulation was equal to the calibration period. In a next step, a model validation on internal stages and flows was tried using additional information (multiple-response validation). Therefore, simulations of the snow routine were compared with snow height measurements at the station Feldberg (1480 m a.s.l.) of the German Weather Service. The general dynamic of the snow cover was weIl modelled, however a detailed analysis of the routine was not possible, because the data was insufficient. Additionally, the modelling of the discharge and silica concentrations at the most frequent runoff zone (zone with periglacial debris cover) was examined.
Therefore the simulations of TAC (discharge and silica concentrations) were compared with measurements of aspring, which has a catchment that is dominated by the periglacial debris cover. The discharge of the spring was weIl modelled, and the general dynamic of the silica concentrations was simulated adequately. Furthermore, the modelling results of TAC were validated at the outlet with tracer measurements. A good agreement of simulated and observed silica concentrations was reached for some periods. Also, a comparison of the portions of the simulated runoff components with the calculated portions using 18O and 3H measurements was performed. The portions of the runoff components agreed far a period of almost three years. Both methods showed the dominance ofthe flow system-2.
The modelling results, the simulation of the different hydrological processes and the model validation using additional information (discharge, snow height measurements, discharge at aspring, silica concentrations and runoff components determined by environmental isotopes) lead to the following conclusion:
The modelling approach of TAC, which is based on the spatial delineation of zones with the same dominating runoff generation processes, and the conceptualization of the runoff generation processes was suitable for an improved process oriented modelling in the Brugga basin. In addition, the potential of tracer methods was demonstrated. They are powerful tools for identifying the runoff generation on catchment scale. On this basis, better process oriented modelling concepts can be developed. The information from tracers (e.g. tracer concentrations, calculated runoff components) can be used to validate or disprove a modelling concept.

PDF-Download (9,0 MB)