Skip to Main content Skip to Navigation
Theses

Filtres de Kalman réduits et efficaces pour l'assimilation de données en océanographie

Abstract : The singular evolutive extended Kalman (SEEK) filter and its interpolated variant, called SEIK, have been implemented and successfully tested in several oceanic models. However, these filters remain expensive for real operational assimilation. The main purpose of our work is to develop degraded forms of the SEEK and SEIK filters which are less costly and yet perform reasonably well. Our approach essentially consists in simplifying the evolution of the correction basis of the SEEK and SEIK filters, which is the most expensive part of these two filters. To enhance the performance of our filters in the model unstable periods, we first introduce the notions local and mixed EOFs analysis in order to improve the representativity of the correction basis. This thought us to construct a new variant of the SEEK filter with a partially local semi-evolutive correction basis. We next present several adaptive tuning schemes based on the parameters of the SEEK filter. Finally, we finish by a comparison between the performances of the SEEK and the ROEK filters to show the feasibility of the evolution of the correction basis, the latter also being introduced to compensate the non-evolutivity of the EOFs local basis. We have implemented all our filters in a realistic setting of the OPA model over the tropical Pacific zone and their performance studied through twin experiments. The SEIK filter is used as a reference for comparison. The results of these experiments show that our new filters perform nearly as well as the SEIK, but can be 2 to 10 times faster.
Document type :
Theses
Complete list of metadata

https://tel.archives-ouvertes.fr/tel-00004682
Contributor : Thèses Imag <>
Submitted on : Monday, February 16, 2004 - 2:13:01 PM
Last modification on : Wednesday, March 10, 2021 - 1:50:03 PM
Long-term archiving on: : Friday, April 2, 2010 - 7:29:34 PM

Identifiers

  • HAL Id : tel-00004682, version 1

Collections

Citation

Ibrahim Hoteit. Filtres de Kalman réduits et efficaces pour l'assimilation de données en océanographie. Modélisation et simulation. Université Joseph-Fourier - Grenoble I, 2001. Français. ⟨tel-00004682⟩

Share

Metrics

Record views

826

Files downloads

483