 |
| Titel |
Smoothing problems in a Bayesian framework and their linear Gaussian solutions |
| VerfasserIn |
Emmanuel Cosme, Jacques Verron, Pierre Brasseur, Jacques Blum, Didier Auroux |
| Konferenz |
EGU General Assembly 2011
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 13 (2011) |
| Datensatznummer |
250047668
|
|
|
|
|
|
| Zusammenfassung |
| Smoothers are increasingly used in geophysics. Several linear Gaussian algorithms have been
presented in the past literature: The sequential smoothers (fixed-interval or fixed-lag), the
ensemble smoother, the forward-backward smoother, also called Rauch-Tung-Striebel
smoother, and the two-filter smoother. Even if they are all equivalent in the linear Gaussian
framework, their prior Bayesian formulations are different, both in the nature and the
resolution strategies of the smoothing problem. In particular, the first two solve a joint
estimation problem. The typical joint smoothing problem is reanalysis, that is, the
estimation of a time sequence of system states based on the observations available in
the same time interval. The last two solve a marginal smoothing problem. This
problem is met, for example, when one aims to estimate the initial state of an observed
process.
In this presentation, we propose a revisit of the different smoothing algorithms, with the
following goals: (i) present a clear description of the Bayesian formulation of each smoothers
mentioned above (what they really solve, and how); (ii) exhibit the assets and drawbacks, in
particular for implementations based on the Ensemble Kalman Filter; (iii) present a few
examples of application. |
|
|
|
|
|
|