Investigation of ENSO variability using cyclostationary EOFs of observational data

Abstract

¶The variability of El Nino-Southern Oscillation (ENSO) on all time scales, ranging from months to decades, has been studied from the perspective of sea-surface temperature anomaly (SSTA), sea-level pressure anomaly (SLPA), and surface wind anomaly (SWA) fields using a statistical tool called cyclostationary empirical orthogonal function (CSEOF) analysis. The analysis indicates that a significant fraction of variability in the tropical Pacific can be explained in terms of an irregular interplay of two dominant modes: the low-frequency mode and the biennial mode. These modes, and in particular the biennial mode, are well defined coupled modes of the tropical ocean-atmospheric circulation system, as suggested by strong correlation (> 0.9) in the evolution of the different physical variables. The low-frequency mode and its PC time series are very similar to the so-called “interdecadal” mode identified by earlier investigators. The corresponding PC time series of the low-frequency mode has a broad spectral peak at 5.3 yrs. The biennial mode represents the oscillation pattern of El Nino and La Nina and has a clear biennial cycle, the amplitude of which varies on a longer time scale. There is a broad spectral peak also at 5.3 yrs in the principal component (PC) time series of this mode. Unlike in earlier studies, this biennial mode is clearly separable from the low-frequency mode, since these two modes have different physical evolution patterns. Results of the present study also indicate that the biennial component, having a clear phase-locking tendency, is not as variable as has been suggested in some previous investigations. It is shown that much of the inter-El Nino variability can be explained in terms of an irregular interplay between the biennial mode and the low-frequency mode. Finally, the role of the low-frequency mode in the tropical Pacific seems to be more important than has been suggested previously. Specifically, recent warming events since 1975 are shown to be associated with the low-frequency mode rather than with the biennial mode.