This paper describes a kernel version of empirical orthogonal function (EOF) analysis and its application to detect patterns of interest in global monthly mean sea surface height (SSH) anomalies from satellite altimetry acquired during the last 17 years. EOF analysis like principal component analysis (PCA) is based on an eigenvalue decomposition of the variance-covariance matrix of the data. The kernel version is based on a dual formulation also termed Q-mode analysis in which the data enter into the analysis via inner products in the so-called Gram matrix only. In the kernel version the inner products are replaced by inner products between nonlinear mappings into higher dimensional feature space of the original data. Via kernel substitution also known as the kernel trick these inner products between the mappings are in turn replaced by a kernel function and all quantities needed in the analysis are expressed in terms of this kernel function. Results from the last 17 years have been obtained from analysing 3 degree longitude and 2 degree latitude, 65 degrees South to 65 degrees North joint TOPEX and JASON-1 data over the period 1992–2008. Preliminary analysis shows some interesting features related to large scale ocean currents and particularly to the pulsing of the El Niño/Southern Oscillation. Large scale ocean events associated with the El Niño/Southern Oscillation related signals are conveniently concentrated in the first SSH EOF modes. A major difference between the classical linear EOF and the kernel EOF analysis is the concentration of patterns associated with large ocean currents in the first two kernel EOF modes.
|Title of host publication||MultiTemp|
|Publication status||Published - 2009|
|Event||MultiTemp - Mystic, Connecticut, USA|
Duration: 1 Jan 2009 → …
|City||Mystic, Connecticut, USA|
|Period||01/01/2009 → …|