Tuesday, 11 February 2003: 4:30 PM
Optimization of a macroscale hydrological model for flood forecasting in the Odra watershed
Joachim Geyer, GKSS Research Center, Geesthacht, Germany; and M. Klein and H. T. Mengelkamp
Poster PDF
(165.2 kB)
The hydrological model SEROS, which is a combination of the land surface scheme SEWAB and a large-scale routing scheme, is applied to the Odra watershed covering 120,000 kmē with a horizontal grid size of 7x7 km2. The routing network and sub-catchments of each gauging station are determined from a DEM. Forcing data from 50 synoptic stations and about 1250 precipitation stations are used to force the model. Daily discharges of 29 gauging stations and of 13 reservoirs in the mountainous region are used to calibrate and verify the model. The calibration period covers 3 years from 1992 to 1994. The verification period from 1995 to 1999 includes the extreme flooding event in 1999 during which hundreds of cities and villages were inundated, more than 100 casualties occurred and vast areas of land were flooded for weeks.
SEROS is optimized/calibrated by use of the Shuffled complex evolution approach (SCE-UA) which was developed at the University of Arizona. The calibration parameters which include parameters for surface and subsurface runoff generation, for the control of evapotranspiration and for the horizontal water transport are allowed to vary in prescribed boundaries. As a measure of the efficiency the Nash-Sutcliffe coefficient is used.
The coincidence of measured and simulated hydrograph is not necessarily a criterion for the adequate description of physical processes. Therefore the parameter sensitivity is investigated in relation to sub-catchment characteristics (orography, landuse, soil type). Results indicate the relevance of different physical processes for different catchments. The findings are elaborated towards a regionalization for parameter estimation.
Supplementary URL: http://w3.gkss.de/KSH