Investigation of tropical Pacific upper-ocean variability using cyclostationary EOFs of assimilated data

Abstract

Variability of the subsurface temperature, current, and heat content in the tropical Pacific Ocean has been extracted in association with the two dominant modes of the sea surface temperature anomaly (SSTA): the low-frequency mode and the biennial mode. In a recent paper, these two modes were identified as the major modes of El Nino-Southern Oscillation (ENSO). The low-frequency mode, which explains about 36% of the total SSTA variability, represents the dominant component of SSTA variability in the tropical Pacific, and is associated not with a fast physical evolution but with a slow stochastic undulation. The biennial mode, which is the second dominant component and explains about 12% of the total variability exhibits, on the other hand, a strong physical evolution. The space–time patterns of the subsurface variability were derived from an assimilated data set via a cyclostationary empirical orthogonal functions (CSEOF) analysis and the regression of the resulting principal component (PC) time series on the target PC time series of the surface modes. Extracted space–time patterns describe the detailed evolution of the physical changes in the upper ocean of the tropical Pacific that are associated with the corresponding surface modes. Specifically, they clearly show the surface and subsurface connection of the physical changes during ENSO events, and the role of equatorial waves in the manifestation of physical changes at the surface. The derived patterns of heat content, subsurface temperature, and zonal current anomalies realistically depict the detailed temporal changes of those variables and are consistent with our understanding of the physics in the tropical Pacific Ocean. The biennial mode appears to depict faithfully the phase progression of El Nino and La Nina. The propagation of equatorial Kelvin waves along the thermocline is clearly visible during El Nino and La Nina events in the cyclostationary representation of the physical modes in the tropical Pacific Ocean. Although the low-frequency mode explains three times more SSTA variability than the biennial mode, the former does not induce strong equatorial wave activity. This observation is significant considering that both El Nino or La Nina are often viewed simply in terms of a significant SST change in the tropical Pacific. The results of the present study indicate: (1) that the two ENSO modes represent significantly different physical evolutions; (2) that the amount of SST warming or cooling does not dictate the physical evolution of ENSO; and (3) that the two modes play essentially different dynamical roles including the generation of equatorial waves.