Articles | Volume 9
26 Sep 2006
26 Sep 2006

Multiscale investigations in a mesoscale catchment – hydrological modelling in the Gera catchment

P. Krause, F. Bäse, U. Bende-Michl, M. Fink, W. Flügel, and B. Pfennig

Abstract. The application of the hydrological process-oriented model J2000 (J2K) is part of a cooperation project between the Thuringian Environmental Agency (Thüringer Landesanstalt für Umwelt und Geologie – TLUG) and the Department of Geoinformatics of the Friedrich-Schiller-University Jena focussing on the implementation of the EU water framework directive (WFD). In the first project phase J2K was parametrised and calibrated for a mesoscale catchment to quantify if it can be used as hydrological part of a multi-objective tool-box needed for the implementation of the WFD. The main objectives for that pilot study were:

  1. The development and application of a suitable distribution concept which provide the spatial data basis for various tasks and which reflects the specific physiogeographical variability and heterogeneity of river basins adequately. This distribution concept should consider the following constraints: The absolute number of spatial entities, which forms the basis for any distributive modelling should be as small as possible, but the spatial distributed factors, which controls quantitative and qualitative hydrological processes should not be generalised to much. The distribution concept of hydrological response units HRUs (Flügel, 1995) was selected and enhanced by a topological routing scheme (Staudenrausch, 2001) for the simulation of lateral flow processes.
  2. J2K should be calibrated for one subbasin of the pilot watershed only. Then the parameter set should be used on the other subbasins (referred as transfer basins) to investigate and quantify the transferability of a calibrated model and potential spatial dependencies of its parameter set. In addition, potential structural problems in the process description should be identified by the transfer to basins which show a different process dominance as the one which was used for calibration does.
  3. Model calibration and selection of efficiency criteria for the quantification of the model quality should be based on a comprehensive sensitivity and uncertainty analysis (Bäse, 2005) and multi-response validations with independent data sets (Krause and Flügel, 2005) carried out in advance in the headwater part of the calibration basin.
  4. To obtain good results in the transfer basins the calibrated parameter set could be adjusted slightly. This step was considered as necessary because of specific constraints which were not of significant importance in the calibration basin. This readjustment should be carried out on parameters which show a sensitive reaction on the identified differences in the environmental setup.
  5. Potential scaling problems of the process description, distribution concept or model structure should be identified by the comparison of the modelling results obtained in a small headwater region of the calibration basin with observed streamflow to find out if the selected efficiency measures show a significant change.