Vers une prise en compte des erreurs-modèle en assimilation de données 4D-variationnelle.
Application à un modèle réaliste d'océan

Abstract : Data assimilation is a class of mathematical methods widely used in meteorology or oceanography. the aim of these methods is to redraw the state of the system using information provided by model in one hand and observation in the other hand. Among these, variational data assimilation methods have recently known a strong development until providing operational suites in the main weather forecast centres. However these methods do not generally account for model errors. The goal of this thesis is to provide "inexact model" algorithm that can be used in the framework of realistic configuarations. Two approaches are presented and developped. First, the Optimal Nudging which consists in, in the framework of the classical 4D-Var, adding to the model a Newtonian relaxation toward the observations. The amplitude of the correction is estimated through optimal control.
Second, in the "Control of the systematic part of the error" the model error is supposed to be a stationnary term. This term is estimated by control along with the initial condition.
Preliminary studies with simplified model and simulated data on these methods are presented and the improvment relative to the classical 4D-Var are pointed out. Finally, the method of control of the systematic part of the error is applied to a primitive equation model in the framework of a realistic experiment in order to validate the good results obtained in the academic configurations.
Document type :
Theses
Complete list of metadatas

https://tel.archives-ouvertes.fr/tel-00325349
Contributor : Arthur Vidard <>
Submitted on : Monday, September 29, 2008 - 9:32:28 AM
Last modification on : Thursday, January 11, 2018 - 6:14:32 AM
Long-term archiving on: Thursday, June 3, 2010 - 7:40:52 PM

Identifiers

  • HAL Id : tel-00325349, version 1

Collections

IMAG | CNRS | UJF | INSMI | UGA

Citation

Arthur Vidard. Vers une prise en compte des erreurs-modèle en assimilation de données 4D-variationnelle.
Application à un modèle réaliste d'océan. Mathématiques [math]. Université Joseph-Fourier - Grenoble I, 2001. Français. ⟨tel-00325349⟩

Share

Metrics

Record views

451

Files downloads

252