Skip to Main content Skip to Navigation
Theses

Modélisation des structures locales de covariance des erreurs de prévision à l'aide des ondelettes

Abstract : The spatio-temporal representation of background error covariances is one of the major problems in data assimilation algorithms. In this thesis, the diagnosis of geographical variations of the local correlation is introduced through the local length-scale diagnosis. The length-scale estimation and the properties of this estimation are studied in details. In this work spherical wavelets are used, according to the formulation introduced by Mike Fisher (ECMWF), in oder to model the local correlation functions "of the day". It is shown that this formulation offers a spatial average of the local correlation that reduces the sampling noise. Moreover, this wavelet formulation provides a robust estimation even for a small ensemble. This formulation is also able to catch the spatio-temporal dynamic of correlation, it is illustrated with the length-scale.
Document type :
Theses
Complete list of metadatas

https://tel.archives-ouvertes.fr/tel-00285515
Contributor : Olivier Pannekoucke <>
Submitted on : Thursday, June 5, 2008 - 4:37:48 PM
Last modification on : Friday, April 5, 2019 - 8:14:09 PM
Long-term archiving on: : Friday, September 28, 2012 - 3:35:26 PM

Identifiers

  • HAL Id : tel-00285515, version 1

Collections

Citation

Olivier Pannekoucke. Modélisation des structures locales de covariance des erreurs de prévision à l'aide des ondelettes. Océan, Atmosphère. Université Paul Sabatier - Toulouse III, 2008. Français. ⟨tel-00285515⟩

Share

Metrics

Record views

290

Files downloads

1632