Skip to Main content Skip to Navigation
Theses

MODÉLISATION SPATIO-TEMPORELLE D'UNE VARIABLE QUANTITATIVE À PARTIR DE DONNÉES MULTI-SOURCES APPLICATION À LA TEMPÉRATURE DE SURFACE DES OCÉANS

Abstract : In this thesis, an important oceanographic variable for the monitoring of the climate is studied: the sea surface temperature. At the global level, this variable is observed along the ocean by several remote sensed sources. In order to treat all this information, statistical methods are used to summarize our variable of interest in global daily map. For that purpose, a state-space linear model with Gaussian error is suggested. We begin to introduce this model on data resulting from having an irregular sampling. Then, we work on the estimation of the parameters. This is based on the combination of the method of moments and the maximum likelihood estimates, with the study of the EM algorithm and the Kalman recursions. Finally, this methodology is applied to estimate the variance of errors and the temporal correlation parameter to the Atlantic ocean. We add the spatial component and propose a separable second order structure, based on the product of a temporal covariance and a spatial anisotropic covariance. According to usual geostatistical methods, the parameters of this covariance are estimated on the Atlantic ocean and form a relevant atlas for the oceanographers. Finally, we show that the contribution of the spatial information increases the predictive behaviour of the model.
Complete list of metadatas

Cited literature [86 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-00582679
Contributor : Pierre Tandeo <>
Submitted on : Sunday, April 3, 2011 - 5:28:26 PM
Last modification on : Friday, October 23, 2020 - 4:52:00 PM
Long-term archiving on: : Thursday, November 8, 2012 - 1:10:52 PM

Identifiers

  • HAL Id : tel-00582679, version 1

Collections

Citation

Pierre Tandeo. MODÉLISATION SPATIO-TEMPORELLE D'UNE VARIABLE QUANTITATIVE À PARTIR DE DONNÉES MULTI-SOURCES APPLICATION À LA TEMPÉRATURE DE SURFACE DES OCÉANS. Mathématiques [math]. Agrocampus - Ecole nationale supérieure d'agronomie de rennes, 2010. Français. ⟨tel-00582679⟩

Share

Metrics

Record views

478

Files downloads

1519