Multiscale Event-Based Mining in Geophysical Time Series: Characterization and Distribution of Significant Time-Scales in the Sea Surface Temperature Anomalies Relatively to ENSO Periods from 1985 to 2009

Abstract

In this study 1D geophysical time series are regarded as series of significant time-frequency events. We combine a wavelet-based analysis to a Gaussian Mixture Model to extract characteristic time-scales of 486043 detected into the Sea Surface Temperature Anomaly (SSTA) observed from satellite at global scale from 1985 to 2009. We retrieved four low frequency characteristic time- scales in the 1.5-7 year range, and highlighted their spatial distribution, most of them known as being influenced by the Nino Southern Oscillation (ENSO). The spatial distribution of the high frequency SSTA also showed a dependency to the ENSO regimes. It pointed out that the ENSO signal also involves specific signatures at these time-scales. These fine-scale signatures can hardly be retrieved from global EOF approaches which tend to exhibit uppermost the low frequency influence onto the SSTA of ENSO. We observed particularly a major increase of the high frequency events in the SSTA during Nino periods of 11% at global scale with a local maximum of 80% in Europe. The methodology was also used to highlight an ENSO-induced shift in time-scale during the major 1997-2000 ENSO event in the inter-tropical Pacific. We observed a clear time-scale shift from the high frequencies towards the 3.36 year-scale with a maximum shift observed two months before the maximum of the ENSO event.