Séminaire
Spectral methods are widely used for the analysis of climatic time series and are important for the detection and characterization of climatic variability. On the other hand, linear and nonlinear oscillators are often proposed as conceptual models for the understanding of variability and its periodic features. As an illustration, the delayed action/recharge oscillator (DAO) is a classic model for the El Niño Southern Oscillation (ENSO). While it is typically assumed that such oscillators are forced by noise, eg that oscillators are stochastic, the detailed spectral properties implied by their stochastic nature remain to some extent unexplored. Here, we propose an analysis of some fine spectral properties of stochastic oscillators. We apply those findings to study the ENSO 3.4 SST time series in an attempt to shed light on some currently debated ENSO features such as the low frequency modulation of its amplitude. We also highlight that stochastic oscillators may be used as building blocks for the development of novel spectral methods.
Ce séminaire a lieu dans le cadre du groupe de Statistiques pour l'Analyse, la Modélisation et l'Assimilation de l'IPSL (SAMA:
http://sama.ipsl.jussieu.fr/indexFr.html
).