A method for detecting season-dependent modes of climate variability: S-EOF analysis

Anomalous climate is often regulated by the annual cycle. Based on this physical consideration, we propose a new method termed as Season-reliant Empirical Orthogonal Function (S-EOF) analysis to detect major modes of climate variability. The S-EOF analysis of Indo-Pacific SST during the past 54 years reveals two statistically significant leading modes, which are not obtainable by using conventional EOF analysis. These modes represent the Indo-Pacific Low-Frequency (LF) and Quasi-Biennial (QB) modes associated ENSO, and reveal the fundamental differences between the LF and QB modes in their seasonal evolution, fractional variance structure, and interdecadal variation and trend. The interdecadal variability is coupled with the LF mode, suggesting that decadal-interdecadal SST variation is primarily represented by the ENSO-like interdecadal regime shift in the late 1970s. A warming trend is mingled with the QB mode, accounting for a large portion of the local SST variability in the Maritime Continent-western Pacific horseshoe region.