 |
| Titel |
Comparing the performance of different model structures with respect to different hydrological signatures |
| VerfasserIn |
T. Euser, H. C. Winsemius, M. Hrachowitz, F. Fenicia, H. H. G. Savenije |
| Konferenz |
EGU General Assembly 2012
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 14 (2012) |
| Datensatznummer |
250063922
|
|
|
|
|
|
| Zusammenfassung |
| Correctly representing the dominant flow generation processes in conceptual rainfall-runoff
models is crucial for ensuring adequate predictive power of the models. Recent
work showed that on the small scale uniqueness of place requires different model
structures for different catchments and that different calibration strategies frequently
result in a wide range of model parameter sets. In this study we investigate the
following research questions: (1) What is the effect of different calibration objective
functions on the model performance? (2) Can the difference in performance of specific
objective functions be related to hydrological signatures and physical catchment
characteristics.
Data from four experimental (approx. 1000 km2) sub-catchments (Alzette, Kyll, Orne and
Seille) of the Moselle were used in this study. Eleven conceptual model structures (HBV,
GR4J and 9 SUPERFLEX (flexible) model structures) of varying level of complexity are
applied on each of the four study catchments. Besides classical objective functions (eg.
Nash-Sutcliffe efficiency), additional objective functions are defined based on several
hydrological signatures, such as the flow duration curve, rising limb density and
auto-correlation. A multi-objective optimization is performed on all the objective functions
for each catchment and each model structure considered. The results of the multi-objective
optimization are then compared using Principle Component Analysis in order to
identify the causes for differences in performance in the objective functions and
relate these to physical catchment characteristics such as elevation, shape of the
catchment and the height distribution above the nearest drain within a catchment. If such
relationships are found then they can help to a priori identify suitable model structures and
hydrological signatures in a catchment, given its spatial scale and physical characteristics. |
|
|
|
|
|
|