Analysis of geophysical time series and information theory: Independent Component Analysis

With the aim of identifying the physical causes of variability of a given d ynamic system, the geophysical community has made extensive use of the stat istical component extraction techniques. We introduce here a recently devel oped algorithm based on information theory: the Independent Component Analy sis. This technique presents two major advantages over classical methods. F irst, it aims at extracting statistically independent components where clas sical techniques search for decorrelated components (i.e, a weaker constrai nt). Secondly, the linear hypothesis for the mixture of components is not r equired. This new method is presented in the context of geophysical time se ries analysis. The ICA algorithm is applied to the study of the variability of the tropical Sea Surface Temperature, with a particular emphasis on the analysis of the links between El-Nino/Southern Oscillation and the Atlanti c SST variability. ((C) Academie des sciences/Elsevier, Paris.).