 |
| Titel |
Ensemble Kalman filter-based discharge observation sensitivity within a spatially distributed hydrological model |
| VerfasserIn |
Oldrich Rakovec, Albrecht Weerts, Pieter Hazenberg, Julius Sumihar, Paul Torfs, Remko Uijlenhoet |
| Konferenz |
EGU General Assembly 2011
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 13 (2011) |
| Datensatznummer |
250052948
|
|
|
|
|
|
| Zusammenfassung |
| Data assimilation is a useful tool in applied hydrology to obtain more reliable and skillful
forecasts by merging the model estimates with observations. Currently, most hydrological
forecasts employ lumped hydrological models (with deterministic or manual state updating),
but there is a clear tendency to move towards spatially distributed models with
hydrological ensemble forecasting because of the increased availability and better
quality of spatially measured data. Nevertheless, there has not been so much attention
paid in the literature to discharge data assimilation into spatially distributed model
states and evaluating the sensitivity of individual discharge gauges. In this study
demonstrate how recent developments in other disciplines (e.g. meteorology and
oceanography) can be applied to hydrological models used for flood forecasting. We aim to
optimize the number of discharge gauges and their locations within the catchment
domain using the Ensemble Kalman filter-based observation sensitivity technique.
The contribution of individual discharge gauges is evaluated using a cost function
difference between the model run with and without data assimilation. Twin experiments
(synthetic and real world) for stratiform and convective rainfall events are carried out
within the Upper Ourthe (1600Â km2), a quickly responding catchment in the Belgian
Ardennes. The first results show that assimilation of discharge into the model states
at interior points does improve hydrological simulations for both rainfall types.
Additionally, observation bias at observation stations can also be easily identified. |
|
|
|
|
|
|