Fine Pattern of Natural Modes in Sea Surface Temperature Variability: 1985–2003

Abstract

Abstract A natural mode refers, in this study, to a periodic oscillation of sea surface temperature (SST) that is geophysically significant on a global, regional, or local scale. Using a newly developed harmonic extraction scheme by Chen, which has the advantage of being space–time decoupled and fully data adaptive, a variety of natural modes have been recovered from global monthly SST data for the period of 1985–2003. Among them, the eight most significant modes are identified as primary modes, whose spatial patterns are presented, along with their phase distributions. At seasonal time scales, a 4-month primary mode is uncovered in addition to the well-documented annual and semiannual cycles. At interannual time scales, the dominant El Niño–Southern Oscillation (ENSO) variability is found to be composed of at least five primary modes, with well-defined central periods around 18, 25, 32, 43, and 63 months. At time scales beyond ENSO, a decadal SST signal with an average period of 10.3 yr is observed. A unique contribution of this study is the derivation and presentation of fine patterns of natural SST modes and signals in joint dimensions of time, space, period, and phase, leading to several findings and conclusions that are of potential importance: 1) The degree of separability and regularity of the sub-ENSO modes is surprising, and thus reveals new details on the nature of this event. 2) The midlatitude counterparts of the equatorial interannual and decadal SST modes/signals are found in the two hemispheres with a frequency shift toward longer periods. The “shadows” of the Pacific Ocean’s ENSO modes are also observed with some detail in the Atlantic and the Indian Oceans. All of these provide direct evidence that teleconnections exist between the equatorial and extratropical oceans, as well as among the tropical Pacific, tropical Atlantic, and tropical Indian Oceans, possibly as a result of the “atmospheric bridge.” 3) A sharply opposite anisotropy is observed in the spatiotemporal pattern between the interannual modes and decadal signals, implying that they are potentially of a categorical difference in origin. 4) Locality or regionality is a fundamental feature for most of the SST modes. Treating the interannual or decadal variability as a single ENSO or Pacific decadal oscillation mode appears to be an oversimplification, and may lead to inappropriate interpretations. The results herein represent an improved knowledge of the natural variability in sea surface temperature, which will hopefully help to enhance the understanding of natural fluctuations of the global/regional climate system in the context of ocean–atmosphere interaction.