Volume 30, Issue 12
Free Access

Global SST/SLP waves during the 20th century

Warren B. White

Warren B. White

Scripps Institution of Oceanography, University of California, San Diego, La Jolla, California, USA

Search for more papers by this author
Yves M. Tourre

Yves M. Tourre

Lamont Doherty Earth Observatory, Columbia University, Palisades, New York, USA

Search for more papers by this author
First published: 27 June 2003
Citations: 58

Abstract

[1] A discrete number of global signals in covarying sea surface temperature (SST) and sea level pressure (SLP) dominated climate variability from 40°S to 60°N during the 20th Century. They are the quasi-biennial (∼2.2-year period), interannual (3- to 7-year period band), quasi-decadal (∼11-year period), and interdecadal (∼17-year period) signals. A joint frequency-domain analysis of SST and SLP anomalies over the global ocean finds these signals composed of mixed global standing modes and traveling waves that are similar in pattern and evolution. On each period scale, the global traveling wave is composed of global zonal wavenumbers-1 and -2, directed eastward with a phase speed that takes 1 to 2 cycles to traverse the global ocean between 40°S and 20°N. The existence of these global traveling waves may enhance the predictability of regional climate variability.

1. Introduction

[2] A set of 7 signals in global and hemispheric surface temperature and pressure during the 20th Century have been revealed by the multi-taper method/singular value decomposition (MTM/SVD) analysis of Mann and Park [1994, 1996, 1999]. This set is composed of the quasi-biennial signal near 2.2-year period, known as the quasi-biennial oscillation or QBO [e.g., Trenberth, 1975], the interannual signals near 2.9-, 3.5-, and 5.5-year period known as the El Niño-Southern Oscillation or ENSO [e.g., Rasmusson and Carpenter, 1982], the quasi-decadal signal near 11-year period known as the quasi-decadal oscillation or QDO [e.g., Mann and Park, 1994, 1996], and the interdecadal signal from 15- to 25-year period [Mann and Park, 1994, 1996] known as the bi-decadal oscillation or BDO [Minobe et al., 2002]. Subsequently, Allan [2000] applied MTM-SVD analysis to of covarying SST and SLP variability over the global ocean during the 20th Century, finding global standing modes on each period scale to be similar to one another. Superimposed on these global standing modes are global traveling waves in covarying SST and SLP anomalies that propagate eastward across the global ocean, in the tropics known as the global biennial wave or GBW [White and Allan, 2001] and the global ENSO wave or GEW [White and Cayan, 2000], and in the Southern Ocean known as the Antarctic circumpolar wave or ACW [e.g., White and Peterson, 1996].

2. Data and Results

[3] Here we apply the MTM/SVD methodology [Mann and Park, 1999] to the 93-year record (1900–1992) of SST and SLP gridded datasets from 40°S to 60°N [Kaplan et al., 1998]. It produces a local fractional variance spectrum (top, Figure 1) that yields 7 discrete signals, significant at the 99% confidence level. They are the BDO and QDO signals near 16.7-year and 10.5-year periods, respectively, the ENSO signals near 5.5-, 4.4-, 3.5-, and 2.9-year periods, and the QBO signal near 2.2-year period. These signals are the same as those determined over the continents [Mann and Park, 1994], the Northern Hemisphere [Mann and Park, 1996], the Arctic Ocean [Venegas and Mysak, 2000], the Atlantic Ocean [Tourre et al., 1999], the Pacific Ocean [Tourre et al., 2001], and the Global Ocean [Allan, 2000]. Here we supplement discussion of the global standing modes by Allan [2000] with discussion of the global traveling waves.

Details are in the caption following the image
(top) Spectrum of the joint SST-SLP local fractional variance as a function of frequency explained by the first joint MTM/SVD mode. The horizontal dashed line represents the 99% confidence limit in the spectrum. Frequencies are given in cycles per year. (1 through 7) Time sequences of amplitudes for each signal (in units of standard deviation ×10) given in the local fractional variance spectrum (see text for details). (8) Sum of quasi-biennial and interannual amplitude time sequences 3-through-7 (solid line) together with Niño 3 SST index (dashed line) band-pass filtered from 2- to 7-year period.

[4] We compute the time sequence of amplitudes corresponding to the complex MTM-SVD mode of the BDO signal, the QDO signal, the ENSO signals, and the QBO signal (1 though 7, Figure 1). These time sequences of amplitudes for each narrow-band signal (1 through 7, Figure 1) display peaks (troughs) corresponding to the eastern tropical Pacific warm (cool) phase (Figure 2b).

[5] We test the validity of these quasi-biennial and interannual time sequences (3 through 7, Figure 1) by summing them and comparing the resulting index against the Niño-3 SST index, the latter band-pass filtered from 2- to 7-year period (8, Figure 1). This comparison finds both indices correlated at 0.77 the 5 discrete signals in the spectrum explaining ∼60% of the Niño-3 index over the 93-year record.

[6] The spatio-temporal evolution of corresponding SLP variability for the 16.7-year period BDO signal, the 10.5-year period QDO signal, the dominant 3.5-year period ENSO signal, and the 2.2-year period QBO signal are displayed side-by-side (Figure 2a). The evolution of each signal derives from the dominant complex singular value decomposition modes, each evolving from the eastern tropical Pacific SST cool phase at 0° phase to the corresponding SST warm phase at 180° phase (see Figure 2b). The spatio-temporal evolution of each SLP signal is characterized by global zonal wavenumber-1 and -2 spatial patterns (Figure 2a), propagating eastward from 40°S to 20°N, and taking ∼1/2 cycle to propagate from the eastern tropical Indian Ocean at 0° phase to the central tropical Pacific Ocean at 180° phase. This eastward propagation has not been observed in the QDO and BDO. These global traveling waves are superimposed on global standing modes, recognized by regional maxima in the eastern tropical Pacific Ocean and in the eastern equatorial Indian Ocean.

Details are in the caption following the image
Spatial evolution for SLP variability in the BDO signal at 16.7-year period, the QDO signal at 10.5-year period, the dominant ENSO signal at 3.5-year period, and the QBO signal at 2.2-year period. Seven panels are chosen to represent ∼1/2 cycle of each signal Blue (yellow-to-red) indicating negative (positive) SLP weights.
Details are in the caption following the image
As in Figure 2a but for SST variability in the 4 representative signals.

[7] The spatio-temporal evolution of corresponding SST variability over 1/2 cycle for each signal are displayed side-by-side (Figure 2b). Each is characterized by global zonal wavenumber-0, -1, and -2 spatial scales, accompanied by a meridional V-shape pattern that is symmetric about the equator, propagating slowly eastward across the tropical Indian and western Pacific oceans and across the eastern tropical Pacific and Atlantic oceans, but interrupted in the central tropical Pacific Ocean by the development of anomalous warm/cool tongues.

[8] In the QBO, ENSO, and BDO signals, covarying SLP and SST weights in the tropical Indian Ocean are preceded by those propagating northward from the Southern Ocean near 40°S (Figures 2a and 2b), indicative of the influence that Southern Ocean climate variability exerts on tropical climate variability [White et al., 2002]. In all four signals, SST/SLP variability in the tropical Indian Ocean leads that in the western tropical Pacific Ocean by ∼1/2 cycle, while that in the eastern tropical Pacific Ocean leads that in the tropical North Atlantic Ocean by a similar amount.

3. Discussion

[9] Recently, White et al. [2003a] demonstrated that the QBO and QDO signals in the Pacific basin share the same delayed action oscillator mechanism used to explain the quasi-periodicity of ENSO [e.g., Zebiak and Cane, 1987]. In Figures 2a and 2b, we can also see the potential for a global delayed-action oscillator mechanism stemming from the eastward phase propagation of the global SST/SLP wave circling the globe on each period scale, with increasing period of each global signal arises from the increasing time it takes for the global SST/SLP wave to circle the globe.

[10] Recent studies finds these global climate signals associated with global-average warming/cooling of the upper ocean by ±0.2°K [White et al., 1997, 2001, 2003b], consistent with that observed in SST weights (Figure 2b). White et al. [2001, 2003b] found global-average warming/cooling associated with the ENSO and QDO signals stemming from internal processes in the Earth's climate system. Yet, the QDO time sequence of amplitude (2, Figure 1) displays peaks that are aligned with the ∼11-year signal in the Sun's irradiance [White et al., 1997]. This suggests that the Sun's 11-year irradiance signal either excites the QDO signal in the Earth's climate system.

[11] These complex MTM-SVD modes in Figures 2a and 2b yield the average wave characteristics of the global SST/SLP modes/waves during the 20th Century, but they may not represent them properly during different epochs within (e.g., White and Cayan, 2000). Thus, the description of these global SST/SLP waves required documentation of their quasi-periodic change in characteristics occurring over the 20th Century.

Acknowledgments

[12] Warren White is supported by the Office of Global Programs of NOAA (NOAA NA17RJ1231). Yves Tourre is presently adjunct at LDEO of Columbia University.