![Hier klicken, um den Treffer aus der Auswahl zu entfernen](images/unchecked.gif) |
Titel |
Nonlinear measurement function in the Ensemble-Kalman filter and a practical application of the Sigma-point Kalman filter |
VerfasserIn |
Y. Tang, D. Chen |
Konferenz |
EGU General Assembly 2012
|
Medientyp |
Artikel
|
Sprache |
Englisch
|
Digitales Dokument |
PDF |
Erschienen |
In: GRA - Volume 14 (2012) |
Datensatznummer |
250063756
|
|
|
|
Zusammenfassung |
In this study, we explored several computational schemes of the Kalman gain of
Ensemble Kalman Filter (EnKF) for nonlinear measurement function. Emphasis is
placed on a comprehensive interpretation of the current algorithm and an extension
of it based on statistically rigorous derivations. It was mathematically proven
that the modified Kalman gain formulas can remove the implicit assumption in the
current algorithm. A simple Lorenz model was used as a test bed to compare these
algorithms. Experiments showed that the modified Kalman gain could perform better
than the current one for the Lorenz model parameter estimate, which involves a
highly nonlinear measurement function.
Another issue addressed in this study was the computational cost of the Sigma-point
Kalman filters (SPKFs). The truncated single value decomposition (TSVD) method was
used to construct a reduced rank SPKF. A realistic ENSO (El Nino and Southern
Oscillation) forecast model was used to test the reduced rank SPKF. The performance
of the reduced rank SPKF was compared to the square root Ensemble Kalman filter
(EnSRF) that was designed parallel to the SPKF. The reduced rank SPKF was found to
be very computationally feasible and led to smaller errors compared to the EnSRF, in
terms of ENSO simulation. |
|
|
|
|
|