Skip to Main content Skip to Navigation

Détection statistique des changements climatiques

Abstract : According to the International Panel on Climate Change (IPCC), detection is the statistical demonstration that an observed change cannot be explained by natural internal variability alone. This PhD Thesis deals with regional climate changes detection and in particular with the statistical methods well suited to it. Several statistical hypothesis testing procedures are introduced and studied. The first method considered involves looking for a climate change signal in the observations, assuming that its spatial distribution is known. In this case, a new adaptation of the optimal fingerprint method is proposed. It is based on the use of a well-conditioned covariance matrix estimate of the internal climate variability. The second approach proposes to look for a signal with a prescribed temporal pattern. This investigated pattern can be evaluated from climate model runs by using smoothing splines. A third strategy involves the study of an undefined climate change signal but one which satisfies a space-time separability assumption. Its time component also need to be regular. A functional statistical framework can be used in this case to construct a test of significance for the first smooth principal component, based on the penalised likelihood ratio. Applying these different methods to observed datasets covering France and the Mediterranean basin has led to new sets of results regarding the current climate changes over these regions. Significant changes are found in the mean annual and seasonal temperatures as well as in the annual precipitation over France. These changes are not spatially uniform, and modify the spatial distribution of the variable considered. Finally, comparing the various methods proposed allows to discuss the ability of numerical climate models to properly represent the spatial and temporal features of climate changes.
Complete list of metadatas
Contributor : Aurélien Ribes <>
Submitted on : Tuesday, December 8, 2009 - 4:54:06 PM
Last modification on : Friday, April 5, 2019 - 8:14:08 PM
Long-term archiving on: : Thursday, June 17, 2010 - 11:28:37 PM


  • HAL Id : tel-00439861, version 1



Aurélien Ribes. Détection statistique des changements climatiques. Mathématiques [math]. Université Paul Sabatier - Toulouse III, 2009. Français. ⟨tel-00439861⟩



Record views


Files downloads