Multi-scale variability features of global sea surface temperature over the past century

Abstract

Understanding the multi-scale variabilities of global sea surface temperature (GSST) is extremely critical for deepening the comprehension of surface climate change. Great efforts have been made to study the multi-scale features of GSST, however, aiming to fully reveal the local features, here we propose a combined approach, incorporating an adaptive method named Ensemble Empirical Mode Decomposition (EEMD), and Pairwise-Rotated EOF (REOF), to separate signals on various frequency bands and eliminate the confounded EOF signatures. The results show that the explained variance of high-frequency components (HFC) in the equatorial central-eastern and south mid-latitude Pacific could reach more than 60%. The grid points where the variance contributions of low-frequency components (LFC) are greater than 40% are mainly concentrated in the subpolar North Atlantic and the Southern Ocean in both Pacific and Atlantic sectors, while that for secular trend (ST) hitting beyond 60% are displayed in the North Indian Ocean, the Southern Ocean from the tip of southwest Africa expanded to the southern side of Australia, Indo-western Pacific, east of the continents in both hemispheres and tropical Atlantic. By applying the EOF/REOF analysis, the leading modes of the HFC, LFC, and ST are then yielded. It is found that the patterns of the HFC are associated with El Niño-South Oscillation (ENSO) diversity, inferring the dominance and independence of the Eastern Pacific (EP) and Central Pacific (CP) El Niño. Meanwhile, Atlantic Multidecadal Oscillation (AMO) and Pacific Decadal Oscillation (PDO) emerge in the rotated modes of the LFC, with the former exhibiting an Atlantic-Pacific coupling.