Accueil > Actualités > Séminaires > Séminaire de Alexander Gluhovsky au LMD-ENS


Titre : Progress in atmospheric time series analysis
Nom du conférencier : Alexander Gluhovsky
Son affiliation : Department of Earth, Atmospheric, and Planetary Sciences Department of Statistics Purdue University
Laboratoire organisateur : LMD
Date et heure : 05-04-2013 14h30
Lieu : salle E314 du Département des Géosciences de l'ENS
Résumé :

Analysis of atmospheric data is problematic, since observed and modeled series are often prohibitively short with only one record typically available, and common methods and models are frequently inadequate as they involve strong assumptions rarely met in real data. Based on new results, the talk will discuss how these problems can be handled by resampling methods, which permit obtaining valid inference without making questionable assumptions about data generating mechanisms, and by incorporating governing equations in the construction of time series models. Due to recent advances in statistical properties of dynamical systems, such novel models may provide a viable alternative to those borrowed from standard time series analysis.