Sea Surface Temperature Variations

Both pairs of climate indices used to characterize in-terannual and decadal ENSO-like variability—CT/ GR and the tropical/North Pacific PCs—are derived from SSTs. Thus, initially, the global SST variations associated with the indices will be outlined. The patterns of SST associated with the CT and GR series are indicated by correlations mapped in Figs. 3a and 3b; positive correlations indicate regions where the ocean is, on average, warmer when the index is more positive and cooler when the index is more negative. As can be expected from the preceding descriptions of the indices, the CT index is more (positively) correlated than the GR index with SSTs in the tropical Pacific. Conversely, the GR index is more (negatively) correlated with SSTs of the North Pacific. Extratropical SSTs in the southern Pacific are about equally correlated to the two indices. Both indices are modestly correlated with SSTs in the tropical Indian Ocean. Despite the differences in emphasis between the CT and GR correlations, the two most remarkable aspects of the SST patterns associated with the two indices are (1) the symmetry of each correlation pattern about the equator in the Pacific basin and (2) the overall similarity between the CT and GR patterns. This similarity led ZWB to describe the decadal variations that dominate the GR index as decadal ENSO-like variability and motivated this study. The strong cross-equatorial symmetries of the SST patterns are reflected in strong symmetries of atmospheric circulation patterns and hydroclimatic responses analyzed in subsequent sections of this chapter.

The rotated tropical and North Pacific PC indices are correlated to SSTs in patterns (Figs. 3c and 3d) that are similar to CT and GR, respectively. However, SST correlations with the tropical PCs emphasize the tropical Pacific SSTs (relative to extratropical SSTs) even more than the CT SST correlations do. SST correlations with the North Pacific PCs emphasize the North Pacif

0 60 e 120 e 180 120 w 60 w 0 0 60 e 120 e 180 120 w 60 w 0

FIGURE 3 Correlation coefficients between annual-averaged sea surface temperatures (SSTs) and (a) Cool Tongue (CT) index (1903-90), (b) Global Residual (GR) index (1903 -90), (c) rotated tropical Principal Component (PC) index (1914-90), and (d) North Pacific PC index (1914-90). The contour interval is 0.2, dashed where negative. Correlations > +0.2 or < —0.2 pass a two-tail Student i-test of being different from zero at 95% significance levels.

0 60 e 120 e 180 120 w 60 w 0 0 60 e 120 e 180 120 w 60 w 0

FIGURE 3 Correlation coefficients between annual-averaged sea surface temperatures (SSTs) and (a) Cool Tongue (CT) index (1903-90), (b) Global Residual (GR) index (1903 -90), (c) rotated tropical Principal Component (PC) index (1914-90), and (d) North Pacific PC index (1914-90). The contour interval is 0.2, dashed where negative. Correlations > +0.2 or < —0.2 pass a two-tail Student i-test of being different from zero at 95% significance levels.

ic temperatures even more than the GRs do. Once again, southern SSTs are about equally correlated with the two PC series.

Thus, the two pairs of Pacific climate indices correspond to similar patterns of SST variation. The CT corresponds largely to ENSO variations by its very definition; the tropical PC emphasizes those variations as an outgrowth of the parsimony provided by all PC analyses. As representations of ENSO variations in the tropical Pacific, the CT and tropical PC series have domi-nantly interannual timescales (Fig. 2). The GR and, even more so, the North Pacific PC series capture variations that are not correlated with CT variations, are dominantly decadal (although neither has been limited to that timescale; Fig. 2), and are similar to SST patterns in the world oceans associated with ENSO. How the in-terannual and decadal climate indices can be related to such similar SST patterns, while having such different (and, in the case of the PCs, entirely uncorrelated) temporal variations, is something of a mystery, but these similarities will be echoed in the atmospheric expressions of the time series as well as in their hydroclimat-ic expressions over the Americas.

Miller and Schneider (1998) recently listed six cate gories of mechanisms that may contribute to the decadal timescales of SST variations in the North Pacific:

• Stochastic atmospheric forcing with a low-frequency SST response (e.g., Barsugli and Battisti 1998)

• Decadal tropical forcing of midlatitude SSTs (e.g., Trenberth, 1990; Graham, 1994)

• Decadal midlatitude ocean-atmosphere interactions of ocean gyre strength and wind stress curls (e.g., Latif and Barnett, 1994; and as a North Atlantic analog, Deser and Blackmon, 1993)

• Tropical-extratropical interactions through subduction of midlatitude (atmospherically forced) SST anomalies into the subsurface ocean to provide source waters of upwelling in the El Niño region of the tropical Pacific (e.g., Gu and Philander, 1997)

• Slow oceanic wave teleconnections from the tropics to the extratropical oceans (e.g., Jacobs et al., 1994)

• Intrinsic decadal vacillation of midlatitude ocean currents (e.g., Jiang et al., 1995)

To this list, we would add the slow vestiges of irregular interannual ocean atmosphere climate variations. Aperiodic variations on interannual scales will yield

ENSO-like Climate Variations on the Americas decadal components upon averaging to decadal time-scales, and those decadal components will necessarily be similar in appearance to the interannual climate processes. Some combination of these mechanisms, then, presumably generates decadal Pacific SST variations that, in turn, influence and even drive parts of the global climate system with clear, if irregular, decadal timescales.

Notably, the present analysis demonstrates remarkable interhemispheric symmetries in expressions of the decadal climate variations associated with the North Pacific modes. This interhemispheric symmetry may provide clues as to which of the processes listed previously is most likely to be dominant, e.g., those that would be expected to yield strong symmetries about the equator. Resolution of the question of which mechanisms dominate is, however, beyond the scope of the present analysis and will require a combination of ocean-atmosphere modeling with careful analyses of the details of global atmospheric and oceanic observations over the last 50 years.

0 0

Post a comment