 |
| Titel |
Forecasting river level using Data Based Mechanistic models and online data assimilation |
| VerfasserIn |
Paul Smith, Dave Leedal, Keith Beven  , Peter Young |
| Konferenz |
EGU General Assembly 2010
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 12 (2010) |
| Datensatznummer |
250035774
|
|
|
|
|
|
| Zusammenfassung |
| The Data-Based Mechanistic modelling framework uses parsimonious time series models to
represent the dominant modes of response of natural systems. The models used are readily
transferred into a State-Space form allowing the Kalman filter to be used for data
assimilation. Multiple case studies have demonstrated the effectiveness of this framework in
providing probabilistic forecasts in many hydrological situations. Recent work on
the prediction of water levels during flood events, presented here, has introduced
state dependant covariance matrices in the Kalman filter formulation. This allows
recognition of the fact that the structural and observational error associated with the
model may relate to the model state or input. Using case studies from UK rivers
we show that the parameters introduced by the definition of the state dependant
covariance matrices can be optimised to minimise the cost of issuing false warnings. |
|
|
|
|
|
|