Analysis of Atlantic SST Variability Factoring Interbasin Links and the Secular Trend: Clarified Structure of the Atlantic Multidecadal Oscillation

Bin Guan Department of Atmospheric and Oceanic Science, University of Maryland, College Park, College Park, Maryland

Search for other papers by Bin Guan in
Current site
Google Scholar
PubMed
Close
and
Sumant Nigam Department of Atmospheric and Oceanic Science, University of Maryland, College Park, College Park, Maryland

Search for other papers by Sumant Nigam in
Current site
Google Scholar
PubMed
Close
Full access

Abstract

Atlantic SST variability in the twentieth century is analyzed factoring the influence of natural SST variability in the Pacific basin and the secular change in global SSTs. The tropical and northern extratropical basins are analyzed together using the extended EOF technique, which permits extraction of the interannual and multidecadal modes in the pan-Atlantic basin in a single step.

The leading mode of Pacific-uninfluenced SST variability is a multidecadal oscillation focused in the extratropical basin, with a period of ∼70 yr. The mode differs from the conventional Atlantic multidecadal oscillation (AMO) in the near quiescence of the tropical–subtropical basin, highlighting the significant influence of the Pacific basin on this region in conventional analysis; as much as 45% of the regional variance resulting from the conventional AMO is due to this influence.

The second and third modes capture the growth (east-to-west development) and decay (near-simultaneous loss of amplitudes) of interannual SST variability in the eastern tropical Atlantic. A nominal 4-yr evolution cycle is identified, but phase transitions are irregular.

The fourth mode describes a north–south tripole with the mature-phase structure resembling the North Atlantic Oscillation’s (NAO’s) SST footprint in winter. The mode lags the NAO by two seasons. Modal evolution involves eastward extension of the main lobe (centered near the separation of the Gulf Stream) along with shrinkage of the oppositely signed two side lobes.

Corresponding author address: Bin Guan, University of Maryland, College Park, 3417 Computer and Space Sciences Bldg., College Park, MD 20742. Email: bguan@atmos.umd.edu

Abstract

Atlantic SST variability in the twentieth century is analyzed factoring the influence of natural SST variability in the Pacific basin and the secular change in global SSTs. The tropical and northern extratropical basins are analyzed together using the extended EOF technique, which permits extraction of the interannual and multidecadal modes in the pan-Atlantic basin in a single step.

The leading mode of Pacific-uninfluenced SST variability is a multidecadal oscillation focused in the extratropical basin, with a period of ∼70 yr. The mode differs from the conventional Atlantic multidecadal oscillation (AMO) in the near quiescence of the tropical–subtropical basin, highlighting the significant influence of the Pacific basin on this region in conventional analysis; as much as 45% of the regional variance resulting from the conventional AMO is due to this influence.

The second and third modes capture the growth (east-to-west development) and decay (near-simultaneous loss of amplitudes) of interannual SST variability in the eastern tropical Atlantic. A nominal 4-yr evolution cycle is identified, but phase transitions are irregular.

The fourth mode describes a north–south tripole with the mature-phase structure resembling the North Atlantic Oscillation’s (NAO’s) SST footprint in winter. The mode lags the NAO by two seasons. Modal evolution involves eastward extension of the main lobe (centered near the separation of the Gulf Stream) along with shrinkage of the oppositely signed two side lobes.

Corresponding author address: Bin Guan, University of Maryland, College Park, 3417 Computer and Space Sciences Bldg., College Park, MD 20742. Email: bguan@atmos.umd.edu

1. Introduction

Coherent, large-scale sea surface temperature (SST) variations are observed in the Atlantic Ocean on interannual to multidecadal time scales. The SST variations impact both local and remote weather and climate, including drought development over North America. A well-known, basinwide variation is the Atlantic multidecadal oscillation (AMO; Enfield et al. 2001), marked by alternation of warm and cold SST anomalies in the North Atlantic every few decades. The structure of interannual variability in the tropical Atlantic is complex, with no single dominant pattern such as El Niño–Southern Oscillation (ENSO) in the Pacific. Two modes have been identified: one is structurally analogous to El Niño, with SST anomalies in the eastern tropical Atlantic maximizing along the African coast (but, unlike El Niño, without oppositely signed anomalies in the extratropics). This mode, associated with southward shift of convection over the eastern Atlantic, has been referred as the Atlantic Niño (Carton et al. 1996; Ruiz-Barradas et al. 2000). The second mode is marked by a north–south SST gradient across the mean location of the Atlantic intertropical convergence zone (e.g., Nobre and Shukla 1996; Tourre et al. 1999; Ruiz-Barradas et al. 2000; Chiang and Vimont 2004) and is referred as the meridional mode (also the gradient mode or the interhemispheric mode) and sometimes as the dipole mode. The physicality of the dipole expression has, however, been questioned (e.g., Enfield et al. 1999; Dommenget and Latif 2000).

A robust characterization of Atlantic SST variability has, however, been stymied by the presence of the strong influence of the Pacific Ocean, especially in the western tropical and subtropical basin where ENSO’s impact is profound and consequential. Analysis of observations has revealed lead–lag links between tropical Pacific and North Atlantic SSTs on various time scales, with the former leading by a few months (Lanzante 1996; Enfield and Mayer 1997; Ruiz-Barradas et al. 2000) to several decades (Latif 2001). Modeling experiments indicate that the tropical Indo-Pacific heating anomalies are important in forcing recent changes in the North Atlantic Oscillation (NAO; Hoerling et al. 2001; Hurrell et al. 2004), which may in turn produce changes in the underlying ocean surface (Eden and Jung 2001; Eden and Willebrand 2001). Also, ENSO-induced freshwater fluxes (Schmittner et al. 2000) can affect the strength of the Atlantic thermohaline circulation and related SST anomalies, as suggested by some modeling studies (Latif et al. 2000; Latif 2001). A recent analysis shows pan-Pacific decadal variability (marked by a horseshoe-like SST structure with the closed end skirting the North American coast) to be linked with AMO-like SST variability in the tropical North Atlantic; with the former leading the AMO index by five seasons (Guan and Nigam 2008, hereafter referred to as GN2008). (The pan-Pacific decadal mode in GN2008 is not ENSO-like.) Taken together, these observational and modeling studies suggest that a considerable portion of Atlantic SST variability (both tropical and extratropical) arises from Pacific basin links. The Pacific influence must thus be factored in when characterizing SST variability in the Atlantic basin.

Another issue in the analysis of Atlantic SST variability concerns the secular trend in SST. The trend in global surface temperature [e.g., National Aeronautics and Space Administration Goddard Institute for Space Studies (NASA GISS)] is nonstationary, exhibiting decadal/multidecadal fluctuations, including midcentury cooling. As conventional linear detrending, based on least squares fitting, can alias the nonstationary secular trend into the natural variability modes (and vice versa), Trenberth and Shea (2006) argued for more careful detrending in their analysis of the recent warming of the tropical North Atlantic. They used the global mean SST as the trend marker, which is, clearly, preferable to the linear trend, but there is still no assurance that detrending data in this manner will not filter some natural variability as well. The AMO signal relative to the global mean SST trend is weak in comparison with that in the “raw” data. The abrupt warming of the North Atlantic after the mid-1990s is not attributed to natural variability in the Trenberth–Shea analysis. A more careful estimation of trend by Ting et al. (2008), based on the Intergovernmental Panel on Climate Change Fourth Assessment Report (IPCC AR4) model simulations, also indicates the recent warming of the North Atlantic to be a forced (global warming related) signal.

Given the presence of a nonstationary secular trend in SST and the strong influence of the Pacific basin on Atlantic SSTs, the recurrent spatiotemporal structure of SST variability in the Atlantic basin is analyzed in this study after removing these components from the “raw” SST anomalies. The components have been characterized and removed from the Atlantic SSTs using various methods, including spectral filtering, linear regression, empirical orthogonal function (EOF) filtering, and linear inverse modeling (e.g., Mestas-Nuñez and Enfield 1999; Chiang and Vimont 2004; Penland and Matrosova 2006; Parker et al. 2007). A contextual separation of natural variability and secular trend, however, eludes many of these analyses, leaving room for more refined and consistent estimation of the trend and the Pacific basin’s influence on Atlantic SSTs.

The current analysis is facilitated by the authors’ Pacific basin SST decomposition (GN2008; see the appendix for main results) where biennial, ENSO, and the decadal variability modes (Pacific natural variability) are extracted together with a mode that captures the nonstationary secular trend in SSTs (the trend mode); all in a single step from extended EOF (EEOF) analysis. The physicality of modes is extensively evaluated in GN2008 using observational analog counts and fish recruitment records. The obtained decadal modes are found to be as, if not more, physical than previous identifications.

The influence of Pacific natural SST variability on Atlantic SSTs and the footprint of the nonstationary SST-trend mode in the Atlantic basin are linearly removed from the Atlantic SST anomalies before spatiotemporal analysis. This EOF-based removal of Pacific basin’s influence and secular trend is one defining attribute of the current analysis. Mutual consistency of the secular trend and natural variability modes—provided both are obtained from the same EEOF analysis—ensures that neither is aliased into the other. Another outcome of extraction from a single analysis is temporal orthogonality, which makes filtering and detrending a straightforward linear exercise. For reference, the mature phase of the Pacific natural variability modes and the nonstationary SST trend are shown in Fig. 1 (updated from GN2008 with five more years of data). The Atlantic footprints are obtained from regressions of the corresponding principal components (PCs).

Recurrent patterns of variability in the detrended and Pacific-uninfluenced component of Atlantic SST anomalies are extracted using the EEOF analysis technique (Weare and Nasstrom 1982). The analysis identifies patterns on the basis of both spatial and temporal recurrence, without imposition of periodicity constraints. The primary analysis, reported in section 3, describes the four leading modes of natural SST variability in the Atlantic basin in the twentieth century, while data and analysis procedures are described in section 2. The stability (robustness) of the identified SST modes is ascertained from sensitivity analysis in section 4. Summary and concluding remarks follow in section 5.

2. Data and analysis procedures

The SSTs analyzed in this study come from the Met Office’s (UKMO) Hadley Centre Sea Ice and Sea Surface Temperature dataset (HadISST) version 1.1 (Rayner et al. 2003), which is globally available on a 1° × 1° grid for the 1870-onward period. The same data were analyzed in GN2008 to characterize SST variability and trend in the Pacific basin. The long-term (1901–2006) means of each calendar month are first removed from the monthly data, yielding monthly anomalies. Seasonal anomalies are then constructed by averaging monthly anomalies over standard three-month periods [e.g., December–February (DJF), June–August (JJA)]. Seasonal anomalies are interpolated onto a 5° × 2.5° longitude–latitude grid in the interest of computational efficiency. Residual SST anomalies in the Atlantic basin are then constructed by subtracting the PC regressions of the six Pacific natural variability modes (biennial, ENSO and decadal variability) and the secular trend mode from the seasonal anomalies. The residual seasonal anomalies are then area weighted, but not normalized. The undertaken EOF is thus covariance based. The variance of the raw and residual seasonal SST anomalies is compared in Fig. 2. Differences are pronounced in the western tropical Atlantic.

The residual Atlantic SST anomalies are analyzed using the EEOF technique, discussed in the appendix (and also GN2008). The 20°S–60°N, 80°W–20°E domain (with Pacific grid points masked out) is analyzed. A seven-season-wide sampling window (seven lags at seasonal interval) is used in the primary EEOF analysis, and VARIMAX rotation (Richman 1986) is applied to the four leading modes. The number of modes rotated is determined by a scree test of the unrotated eigenvalues; the eigenvalues are checked for statistical significance beforehand by a “rule N” test (Overland and Preisendorfer 1982), which shows the first 55 EEOFs to be above the noise level (p < 0.01). Sensitivity experiments attest to the stability (robustness) of the identified patterns to the choice of sampling window width and number of rotated modes.

3. Results

The four leading modes of residual SST variability in the Atlantic basin exhibit time scales ranging from interannual to multidecadal, as evident from the PCs and related autocorrelations (Fig. 3). Not surprisingly, none of the four modes exhibit any secular trend. These modes should be uncorrelated with natural SST variability in the Pacific (including the biennial, ENSO, and decadal modes) in view of the analysis of residual Atlantic SSTs, and this is confirmed by the near-zero cross correlations between Pacific and Atlantic PCs. Moreover, the four PCs have nearly insignificant projections on the remnant (i.e., trend and Pacific natural variability linearly removed) Pacific/Indian Ocean SSTs (not shown), indicating that they are associated with variability confined to the Atlantic basin. The mode names/labels and the percentage variance explained (in the residual SST data) are listed in Table 1.

a. Atlantic multidecadal oscillation: Clarified structure

The leading mode of residual Atlantic SST variability is marked by its multidecadal time scale (Fig. 3) and extratropical focus (Figs. 4a–c). SST anomalies are as large as 0.4 K during the mature phase, centered at ∼50°N. The precursor phase (Fig. 4a) shows that SST anomalies originate in the Davis Straits and the Labrador Sea. The anomalies advect eastward in the midlatitudes, then equatorward in the eastern basin (along the Canary Current track), and finally westward in the subtropics, much like the North Atlantic subtropical gyre. The subtropical anomalies are weak but statistically significant, as assessed by the corresponding correlation coefficients. It is noteworthy that equatorial Atlantic, especially the eastern basin, is quiescent during modal evolution. An approximately 70-yr period, estimated from zero crossings of the autocorrelation function (which provide an estimate of the half period), is suggested by Fig. 3 (red curve in the lower panel; corresponding time scale is shown at the top in red). The PC distribution (Fig. 3, upper panel) also indicates the ∼70-yr time scale, given the presence of a complete cycle in the record. The distribution also suggests that recent warming of the North Atlantic is not unique, as one occurred in the 1920s–30s as well. A century-long record is, however, insufficient for robust characterization of such long time-scale variability.

The spatial structure of the multidecadal mode differs considerably from its previous characterization (AMO; Enfield et al. 2001, their Fig. 1b). The latter is marked by the presence of same-signed SST anomalies across the entire North Atlantic, including the tropical basin. Our analysis indicates that the tropical–subtropical SST signal in previous AMO characterizations is, in fact, reflective of Pacific’s influence on Atlantic SSTs, rather than an intrinsic component of multidecadal oscillation variability. The Pacific influence was, of course, filtered out to clarify the structure of long time-scale variability in the Atlantic basin.

The mature-phase SST structure of the multidecadal oscillation is compared in three analyses in Fig. 5. The displayed structures are 1) from the leading mode of residual Atlantic SST variability described above (AMO-Atl; Fig. 5b); 2) from Atlantic footprints of Pacific natural variability (AMO-Pac; Fig. 5c); and 3) from the raw but detrended (as in this paper) SST anomalies (AMO-Tot; Fig. 5d). (For reference, the AMO-Tot index is correlated at 0.86 with the index based on least squares detrending.) The nonstationary SST trend is thus not contributing to any of the patterns. Note, SST correlations are shown in all cases. Correlations of PC1 are shown in case of AMO-Atl. In the other two cases, correlations of an SST index generated from areal averaging of the variously processed SST anomalies in the North Atlantic basin (equator–70°N; the AMO index region in Enfield et al. 2001) are shown. PC1 and the two SST indexes are shown in Fig. 5a, after some smoothing. The AMO index generated from SST anomalies based on summation of the contributions of the six Pacific and four Atlantic natural variability modes is also shown in Fig. 5a. Together, the two basin effects account for ∼75% of the variance manifest in the AMO-Tot index, given the 0.87 correlation between the two unsmoothed time series. The Pacific basin influence (AMO-Pac) is the major contributor (∼45%) in view of its 0.66 correlation with the AMO-Tot index.1 There is, nonetheless, a significant portion (∼25%) of the AMO-Tot index variance that is unaccounted for by the leading Pacific and Atlantic natural variability modes. The expansive averaging region used in index definition is, perhaps, one reason, for it makes the index susceptible to regional variability patterns (higher-order modes; noise, in context of multidecadal variability?). The AMO-Tot index is thus often smoothed to emphasize its low-frequency component. Smoothing enhances the previously noted correlations, not surprisingly; the combined basin effects are correlated with the AMO-Tot index at 0.97 after 10 applications of the 1–2–1 smoother on the indexes.

The mature-phase structure of the AMO-Atl and AMO-Pac SSTs is quite different. The related indexes are temporally orthogonal, but the corresponding spatial patterns are not formally constrained to be in quadrature by the EEOF analysis or the following VARIMAX rotation. Even so, the patterns are somewhat in quadrature: the AMO-Atl is focused in the extratropical basin (cf. Figs. 4a–c and 5b) while the AMO-Pac is focused in the northern tropical Atlantic (Fig. 5c). Both components are manifest in AMO-Tot (Fig. 5d). In context of the basin contributions, it is of some interest to examine which basin was more influential in setting the trend of the AMO-Tot index in various subperiods of the twentieth century. For example, AMO-Atl contributes more to the earlier upward trend of the AMO-Tot index, especially during the dust bowl drought period (1931–39). A downward trend from the late 1950s to the 1970s, on the other hand, reflects the AMO-Pac influence. The recent upward trend is supported by both basins, but perhaps more by the AMO-Pac index. The influence of Pacific natural SST variability on AMO, captured by the AMO-Pac index, arises essentially from the decadal variability of SSTs in the pan-Pacific basin (the PDVPP mode in GN2008). The PDVPP mode was shown linked to AMO-like SSTs in northern tropical–subtropical Atlantic in Fig. 11 of GN2008, and it was noted there (and also in the introduction of this note) that this mode leads the AMO index by five seasons. The origin and mechanisms of Pacific decadal SST variability, however, remains to be fully investigated and understood.

Multidecadal variability with large amplitudes in the high-latitude North Atlantic has been identified in previous studies as well (e.g., Kawamura 1994; Mestas-Nuñez and Enfield 1999; Parker et al. 2007). A key difference of the current characterization is, however, its independence from the Pacific and Indian Oceans (not shown): Mestas-Nuñez and Enfield (1999), for example, associate their high-latitude, multidecadal mode with the NAO (and variability in the Gulf of Alaska), in contrast with this analysis, where AMO-Atl and NAO are essentially uncorrelated (r < 0.1). This is not surprising, given their very different dominant time scales.

b. Atlantic Niño

The second and third modes of residual variability in the Atlantic basin (Figs. 4d–i) describe growth and decay of interannual SST variations in the eastern tropical basin. Anomalies are asymmetrically distributed about the equator but tilted northwestward from the African coast to the central equatorial Atlantic. The modal evolution features an east-to-west development during the growing phase (AtlNiño; Figs. 4g–i), and an opposite retreat during the decay phase (AtlNiño+; Figs. 4d–f). The phase transitions are, however, irregular (more so than ENSO), as indicated by the lack of zero crossings in the autocorrelation functions (Fig. 3, lower panel). The mature-phase structure of this mode compares well with the Atlantic Niño mode in Ruiz-Barradas et al. (2000), especially the asymmetry about the equator; the somewhat reduced amplitudes are attributable to the removal of the Pacific influence. The relationship of the Atlantic Niño PCs and the Atl3 index (i.e., averaged SST anomalies in the 3°S–3°N, 20°W–0° region; computed with residual Atlantic SST anomalies in the interest of consistency) are shown in Fig. 6. The Atl3 index, defined by Zebiak (1993), is a marker of the coupled equatorial Atlantic mode that is dynamically akin to ENSO. The AtlNiño PC leads the Atl3 index by two seasons (open dots), while AtlNiño+ lags Atl3 by two seasons (filled dots). The two PCs themselves are shifted by four seasons (squares), indicating a 4-yr period of the underlying “oscillation.” The lead–lag correlation of ∼0.8 (i.e., ∼64% explained variance) between AtlNiño PCs and the Atl3 index suggests that SST variability in the equatorial Atlantic is composed of more than just the Atlantic Niño mode. A full accounting of tropical Atlantic variability (including the meridional mode) was, however, not the objective of this study.

c. The SST tripole: NAO related

Mode 4 is marked by decadal variations of a north–south SST tripole in the North Atlantic, with the main lobe centered near the Gulf Stream separation region. SST anomalies in the main lobe develop eastward along 40°N, while the oppositely signed side lobes weaken (Figs. 4j–l). The mature-phase structure bears close resemblance to NAO’s impact on winter SSTs (e.g., Visbeck et al. 2001; Nigam 2003). The related PC lags the NAO index by two seasons (not shown). This is also evident in Fig. 7, which shows PC regressions on the 500-hPa height field. The meridional dipole structure in the Atlantic, marking NAO variability, is most well developed in the two-season lead regressions (left panel). It weakens in two seasons (middle panel), coincident with the peaking of the SST tripole, and then quickly weakens (right panel). The two-season lag of the SST tripole PC vis-à-vis the NAO index indicates a forcing role for the atmosphere. Some authors (Czaja and Frankignoul 2002; Rodwell and Folland 2002) have argued for positive feedback between the NAO and the SST tripole pattern, but their analysis is beyond the scope of this note.

4. Sensitivity analysis

Robustness of the above described modes of residual Atlantic SST variability is assessed by perturbing the primary analysis in various ways, as summarized in Table 2. Each sensitivity test focuses on one aspect of the primary analysis (T0), including the analysis domain (T1), analysis period (T2), number of rotated modes (T3), width of the sampling window (T4–T6), prefiltering of Pacific natural variability (T7), and the use of rotation itself (T8). In T1–T3 and T8, the primary analysis is perturbed only in the indicated aspect. Tests T4–T6 are designed to assess if the seven-season sampling window used in T0 is of sufficient width to robustly sample multidecadal variability (i.e., the AMO). To do this, the residual SSTs used in T0 are further filtered by removing the Atlantic Niño and the tripole modes. The resulting data are then subjected to EEOF analysis with wider sampling windows (14, 22, and 28 seasons) but without rotation since only the first mode (i.e., AMO) is of interest. T7 tests whether the modes identified in residual SSTs are also present in the raw ones (but without the secular trend); in particular, the influence of Pacific natural variability on Atlantic SSTs is not filtered prior to analysis in this test. The test should address concerns about the potential impact of such filtering on AMO structure. (Five modes are rotated in T7, allowing for the additional expression of the Pacific influence.)

The results of the sensitivity tests are given in Table 3, where each mode in T0 is compared to their counterpart in the test cases by tabulating the mode order, percentage variance explained, and PC correlation. The results suggest that the following:

  • The four leading modes of residual Atlantic SST variability are robust to reasonable variations of sampling domain and period (T1 and T2) and to the number of rotated modes (T3).

  • The seven-season sampling window used in T0, while relatively short, is sufficient for a robust characterization of AMO’s spatiotemporal structure (T4–T6).

  • The four leading modes can also be found in the analysis of raw, but detrended, Atlantic SSTs (modes 2–5 in T7), that is, in the presence of the influence of Pacific natural variability, indicating robustness to the filtering process. The similarity of the AMO-Atl structure in T0 and T7 (not shown) attests to the stability of this mode. The near orthogonality of this mode (as characterized in T7) and Pacific natural variability (PDVPP in particular; PC correlation is 0.1) justifies linear filtering of the latter in the primary analysis (T0). The leading mode in T7, consistently, captures the Pacific’s influence on the Atlantic basin; it is correlated with PDVPP at a 2–3-season lag and bears close resemblance to the AMO-Pac (Fig. 5c).

  • The AMO and the tripole mode are insensitive to EOF rotation, but the Atlantic Niño modes (T8) are not.

5. Summary and concluding remarks

The twentieth-century Atlantic SST variability is analyzed in context of secular changes and interbasin connectivity. These influences have, hitherto, not been accounted for when identifying recurrent modes of Atlantic basin variability; potentially aliasing the modal structures. The present analysis accounts for these influences by factoring out Pacific natural variability and the nonstationary secular trend from the raw SST anomalies, prior to objective identification of recurrent patterns with the extended EOF analysis technique. Modal stability is assessed from an extensive suite of sensitivity tests. The four leading modes of residual Atlantic SST variability are

  • Atlantic Multidecadal Oscillation (AMO-Atl): this leading mode is characterized by alternation of warm and cold anomalies centered in the North Atlantic basin, with a cycle of ∼70 yr. The mode differs from the conventional AMO (AMO-Tot here; Fig. 5d) in the tropics and subtropics where its footprint is muted because of the exclusion of the Pacific influence. Both the phase and trend of the conventional AMO index are found to be impacted by the Pacific influence (cf. Fig. 5a). The AMO mode (AMO-Atl; Fig. 5b) and the Pacific influence on the Atlantic (AMO-Pac; Fig. 5c) together explain about 75% of the variance represented by the conventional AMO (AMO-Tot), with about 45% coming from the Pacific influence;

  • Atlantic Niño: the growth (east-to-west development) and decay (near-simultaneous loss of amplitudes) of interannual SST anomalies in the eastern tropical Atlantic are captured by two modes, each describing one phase of the evolution. Their quadrature phase relationship at four-season lag suggests a 4-yr time scale, but phase transitions are indicated to be irregular as autocorrelations of the related PCs do not change sign (although they do approach zero at ±1 yr;

  • The SST tripole: a north–south tripole resembling NAO’s winter SST footprint is identified. Its evolution features an eastward extension of the main lobe (in the Gulf Stream separation region) and concurrent shrinking of the oppositely signed two side lobes. The mode lags the atmospheric NAO index by two seasons.

The present analysis provides a refined description of SST variability in the Atlantic basin, with the refinement in AMO structure particularly striking. The clarified description, especially the lack of amplitude in the tropical–subtropical North Atlantic, finds some resonance in recent AMO modeling studies that fail to produce a large SST signal in this region (T. L. Delworth 2008, personal communication). The clarified description should facilitate understanding and modeling of multidecadal variability in the Atlantic basin and of the interbasin connections. The recent warming of the Atlantic has elicited great interest in its origin, and our analysis suggests a significant role of Pacific natural variability in it.

This discriminating analysis of Atlantic SST variability also provides well-defined targets for climate simulations of the twentieth century. Of particular interest would be the changes in modal structure in the IPCC simulations of climate change. The consistency between Pacific natural variability, the nonstationary secular trend, and the residual Atlantic modes also facilitates analysis of regional hydroclimate variability and change; a study of the causes of long-term droughts over North America is currently underway.

Finally, one needs to acknowledge the possibility that Pacific’s influence on the Atlantic basin is nonnormal (Penland and Matrosova 2006) and/or nonlinear. The influence captured in the AMO-Pac reflects only the linear association. It is, however, noteworthy that the clarified AMO structure (AMO-Atl) is not dependent on the linearity assumption, as the same mode is recovered from an analysis conducted without advance filtering of the Pacific influence.

Acknowledgments

Bin Guan was supported, in part, by the Ann G. Wylie Dissertation Fellowship from the University of Maryland Graduate School. This work constitutes part of the doctoral thesis of the first author. Both authors acknowledge NOAA and NSF support through CPPA-NA17EC1483 and ATM-0649666 grants.

REFERENCES

  • Carton, J. A., X. Cao, B. S. Giese, and A. M. Da Silva, 1996: Decadal and interannual SST variability in the tropical Atlantic Ocean. J. Phys. Oceanogr., 26 , 11651175.

    • Search Google Scholar
    • Export Citation
  • Chiang, J. C. H., and D. J. Vimont, 2004: Analogous Pacific and Atlantic meridional modes of tropical atmosphere–ocean variability. J. Climate, 17 , 41434158.

    • Search Google Scholar
    • Export Citation
  • Czaja, A., and C. Frankignoul, 2002: Observed impact of Atlantic SST anomalies on the North Atlantic Oscillation. J. Climate, 15 , 606623.

    • Search Google Scholar
    • Export Citation
  • Dommenget, D., and M. Latif, 2000: Interannual to decadal variability in the tropical Atlantic. J. Climate, 13 , 777792.

  • Eden, C., and T. Jung, 2001: North Atlantic interdecadal variability: Oceanic response to the North Atlantic Oscillation (1865–1997). J. Climate, 14 , 676691.

    • Search Google Scholar
    • Export Citation
  • Eden, C., and J. Willebrand, 2001: Mechanism of interannual to decadal variability of the North Atlantic circulation. J. Climate, 14 , 22662280.

    • Search Google Scholar
    • Export Citation
  • Enfield, D. B., and D. A. Mayer, 1997: Tropical Atlantic SST variability and its relation to El Niño–Southern Oscillation. J. Geophys. Res., 102 , 929945.

    • Search Google Scholar
    • Export Citation
  • Enfield, D. B., A. M. Mestas-Nuñez, D. A. Mayer, and L. Cid-Serrano, 1999: How ubiquitous is the dipole relationship in tropical Atlantic sea surface temperatures. J. Geophys. Res., 104 , 78417848.

    • Search Google Scholar
    • Export Citation
  • Enfield, D. B., A. M. Mestas-Nuñez, and P. J. Trimble, 2001: The Atlantic multidecadal oscillation and its relation to rainfall and river flows in the continental U.S. Geophys. Res. Lett., 28 , 20772080.

    • Search Google Scholar
    • Export Citation
  • Guan, B., and S. Nigam, 2008: Pacific sea surface temperatures in the twentieth century: An evolution-centric analysis of variability and trend. J. Climate, 21 , 27902809.

    • Search Google Scholar
    • Export Citation
  • Hare, S. R., and N. J. Mantua, 2000: Empirical evidence for North Pacific regime shifts in 1977 and 1989. Prog. Oceanogr., 47 , 103145.

    • Search Google Scholar
    • Export Citation
  • Hoerling, M. P., J. W. Hurrell, and T. Xu, 2001: Tropical origins for recent North Atlantic climate change. Science, 292 , 9092.

  • Hurrell, J. W., M. P. Hoerling, A. Phillips, and T. Xu, 2004: Twentieth-century North Atlantic climate change. Part I: Assessing determinism. Climate Dyn., 23 , 371389.

    • Search Google Scholar
    • Export Citation
  • Kawamura, R., 1994: A rotated EOF analysis of global sea surface temperature variability with interannual and interdecadal scales. J. Phys. Oceanogr., 24 , 707715.

    • Search Google Scholar
    • Export Citation
  • Lanzante, J. R., 1996: Lag relationships involving tropical sea surface temperatures. J. Climate, 9 , 25682578.

  • Latif, M., 2001: Tropical Pacific/Atlantic Ocean interactions at multi-decadal time scales. Geophys. Res. Lett., 28 , 539542.

  • Latif, M., E. Roeckner, U. Mikolajewicz, and R. Voss, 2000: Tropical stabilization of the thermohaline circulation in a greenhouse warming simulation. J. Climate, 13 , 18091813.

    • Search Google Scholar
    • Export Citation
  • Mantua, N. J., S. R. Hare, Y. Zhang, J. M. Wallace, and R. C. Francis, 1997: A Pacific interdecadal climate oscillation with impacts on salmon production. Bull. Amer. Meteor. Soc., 78 , 10691079.

    • Search Google Scholar
    • Export Citation
  • Mestas-Nuñez, A. M., and D. B. Enfield, 1999: Rotated global modes of non-ENSO sea surface temperature variability. J. Climate, 12 , 27342746.

    • Search Google Scholar
    • Export Citation
  • Nigam, S., 2003: Teleconnections. Encyclopedia of Atmospheric Sciences, J. R. Holton et al., Eds., Academic Press, 2243–2269.

  • Nobre, P., and J. Shukla, 1996: Variations of sea surface temperature, wind stress, and rainfall over the tropical Atlantic and South America. J. Climate, 9 , 24642479.

    • Search Google Scholar
    • Export Citation
  • Overland, J. E., and R. W. Preisendorfer, 1982: A significance test for principal components applied to a cyclone climatology. Mon. Wea. Rev., 110 , 14.

    • Search Google Scholar
    • Export Citation
  • Parker, D., C. Folland, A. Scaife, J. Knight, A. Colman, P. Baines, and B. Dong, 2007: Decadal to multidecadal variability and the climate change background. J. Geophys. Res., 112 , D18115. doi:10.1029/2007JD008411.

    • Search Google Scholar
    • Export Citation
  • Penland, C., and L. Matrosova, 2006: Studies of El Niño and interdecadal variability in tropical sea surface temperatures using a nonnormal filter. J. Climate, 19 , 57965815.

    • Search Google Scholar
    • Export Citation
  • Rayner, N. A., D. E. Parker, E. B. Horton, C. K. Folland, L. V. Alexander, D. P. Rowell, E. C. Kent, and A. Kaplan, 2003: Global analyses of sea surface temperature, sea ice, and night marine air temperature since the late nineteenth century. J. Geophys. Res., 108 , 4407. doi:10.1029/2002JD002670.

    • Search Google Scholar
    • Export Citation
  • Richman, M. B., 1986: Rotation of principal components. Int. J. Climatol., 6 , 293335.

  • Rodwell, M. J., and C. K. Folland, 2002: Atlantic air–sea interaction and seasonal predictability. Quart. J. Roy. Meteor. Soc., 128 , 14131443.

    • Search Google Scholar
    • Export Citation
  • Ruiz-Barradas, A., J. A. Carton, and S. Nigam, 2000: Structure of interannual-to-decadal climate variability in the tropical Atlantic sector. J. Climate, 13 , 32853297.

    • Search Google Scholar
    • Export Citation
  • Schmittner, A., C. Appenzeller, and T. F. Stocker, 2000: Enhanced Atlantic freshwater export during El Niño. Geophys. Res. Lett., 27 , 11631166.

    • Search Google Scholar
    • Export Citation
  • Ting, M., Y. Kushnir, R. Seager, and C. Li, 2008: Forced and internal 20th century SST trends in the North Atlantic. J. Climate, 22 , 14691481.

    • Search Google Scholar
    • Export Citation
  • Tourre, Y. M., B. Rajagopalan, and Y. Kushnir, 1999: Dominant patterns of climate variability in the Atlantic Ocean during the last 136 years. J. Climate, 12 , 22852299.

    • Search Google Scholar
    • Export Citation
  • Trenberth, K. E., and D. J. Shea, 2006: Atlantic hurricanes and natural variability in 2005. Geophys. Res. Lett., 33 , L12704. doi:10.1029/2006GL026894.

    • Search Google Scholar
    • Export Citation
  • Visbeck, M. H., J. W. Hurrell, L. Polvani, and H. M. Cullen, 2001: The North Atlantic Oscillation: Past, present, and future. Proc. Natl. Acad. Sci. USA, 98 , 1287612877.

    • Search Google Scholar
    • Export Citation
  • Weare, B. C., and J. S. Nasstrom, 1982: Examples of extended empirical orthogonal function analyses. Mon. Wea. Rev., 110 , 481485.

  • Zebiak, S. E., 1993: Air–sea interaction in the equatorial Atlantic region. J. Climate, 6 , 15671586.

APPENDIX

Pacific SST EEOF

EEOF formulation

Seasonal anomalies (3-month means) are formed and appropriately weighted, denoted by
i1520-0442-22-15-4228-ea1
where N is the number of seasons and M is the number of grid cells. In EEOF analysis with a (2L + 1)Δt temporal sampling window, an extended anomaly matrix AE is constructed as
i1520-0442-22-15-4228-ea2
The parameters L and Δt are chosen so that the sampling window [(2L + 1)Δt] coves a significant portion of variability evolution; L = 2 in GN2008, and 3 in the current analysis;2 Δt = 3 months. The EOF analysis is based on singular value decomposition (SVD) of AE.

Pacific natural SST variability and secular trend (summary of GN2008)

In the primary analysis, seven modes are extracted from rotated EEOF of seasonally resolved, unfiltered SST in the pan-Pacific domain (20°S–60°N, 120°E–60°W) during 1900–2007. Canonical ENSO variability is encapsulated in two modes that depict the growth and decay phases. Another interannual mode, energetic in recent decades, is shown linked to the west-to-east SST development seen in post-climate-shift ENSOs: the noncanonical ESNO mode. Pacific decadal variability (PDV) is characterized by two modes: the pan-Pacific mode has a horseshoe structure with the closed end skirting the North American coast and a quiescent eastern equatorial Pacific. The second decadal mode—the North Pacific mode—captures the 1976/77 climate shift and is closer to the Pacific decadal oscillation (Mantua et al. 1997). Implicit accommodation of natural variability leads to a nonstationary SST trend, including midcentury cooling.

Robustness of these modes is ascertained from perturbation of the primary analysis. Eight sensitivity tests are conducted, each focusing on an aspect of the primary analysis, including analysis domain, climatology base period, analysis period, number of rotated modes, and sampling window width. That EOF rotation yields more physical modes of variability is evaluated from the observational analog count. It is found that rotation increases the number of analogs by ∼20%. The physicality of the two decadal modes (whose long time scales potentially permit marine population adjustments) is assessed from correlations with biological time series in the North Pacific (Hare and Mantua 2000). The reported extractions evidently fare at least as well as the PDO.

Fig. 1.
Fig. 1.

Pacific natural SST variability and the secular trend: EEOF modes (mature-phase patterns) of Pacific (20°S–60°N, 120°E–60°W) SSTs during the twentieth century (1900–2007); see GN2008 for details, and the appendix for a brief description. Atlantic footprints are obtained from regressions of the PCs. Solid (dashed) contours denote positive (negative) values and the zero contour is suppressed. Contour interval is 0.1 K.

Citation: Journal of Climate 22, 15; 10.1175/2009JCLI2921.1

Fig. 2.
Fig. 2.

Ratio (%) of the variance of the residual and raw SST anomalies. Residual SSTs are obtained by linearly removing Pacific natural variability and the secular trend (see Fig. 1) from the raw data.

Citation: Journal of Climate 22, 15; 10.1175/2009JCLI2921.1

Fig. 3.
Fig. 3.

(top) PCs of residual Atlantic SST variability in the twentieth century (1901–2006). Tick marks on the vertical axis are drawn every three units. The original PCs are shaded, while heavily smoothed versions (from 50 applications of a 1–2–1 smoother) are shown using solid black lines. (bottom) Autocorrelations of the PCs. Note that the AMO autocorrelation is shown using the top scale.

Citation: Journal of Climate 22, 15; 10.1175/2009JCLI2921.1

Fig. 4.
Fig. 4.

Time evolution of the leading four modes of residual Atlantic SST variability. (a)–(c) AMO-Atl. (d)–(f) Decaying phase of Atlantic Niño. (g)–(i) Growing phase of Atlantic Niño. (j)–(l) Tripole. Note that the AMO evolution is displayed at four-season intervals, the Atlantic Niño modes at two-season intervals, and the tripole mode at three-season intervals. Maps are obtained by regressing lead–lagged residual SST anomalies onto the PCs shown in Fig. 2 (top), with the label t denoting simultaneous regressions. Solid (dashed) contours denote positive (negative) values and the zero contour is suppressed. Contour interval is 0.1 K.

Citation: Journal of Climate 22, 15; 10.1175/2009JCLI2921.1

Fig. 5.
Fig. 5.

(a) AMO indexes based on reconstructed and raw (but detrended) SST data. Reconstructions are based on individual EEOF mode or mode combinations, indicated by the curve key; the numbers show the correlation between the reconstructed and the raw AMO (i.e., AMO-Tot). The AMO indexes are smoothed by 10 applications of a 1–2–1 smoother before being displayed. (b)–(d) Correlations between raw Atlantic SST anomalies and the AMO index based on (b) residual Atlantic mode 1 (the AMO-Atl mode), (c) Pacific natural variability, and (d) raw SSTs without the secular trend (see Fig. 1). Correlation calculations in (a)–(d) are all based on unsmoothed indexes. Contour interval is 0.1. Contour/shading threshold is 0.2.

Citation: Journal of Climate 22, 15; 10.1175/2009JCLI2921.1

Fig. 6.
Fig. 6.

Cross correlations of AtlNiño and AtlNiño+ PCs, and of each with the Atl3 index (SST anomalies averaged over 3°S–3°N, 20°W–0°; see text for details), at various seasonal lead–lags. The curve key is in the upper-left corner, with the following plotting convention: when r(A,B) > 0 for t < 0, B leads A; if r > 0 for t > 0, B lags A. Cross correlations show that AtlNiño (AtlNiño+) PC leads (lags) the Atl3 index by two seasons, and consistently, AtlNiño+ lags AtlNiño by four seasons.

Citation: Journal of Climate 22, 15; 10.1175/2009JCLI2921.1

Fig. 7.
Fig. 7.

(left) Two-season lead, (middle) simultaneous, and (right) two-season lag regressions of 500-hPa height on the tripole mode PC. Contour interval is 3 m. The 30° and 60°N latitude circles are indicated by the dotted lines.

Citation: Journal of Climate 22, 15; 10.1175/2009JCLI2921.1

Table 1.

Leading modes of seasonal SST variability in the Atlantic identified by rotated EEOF analysis after linearly removing the seven leading Pacific modes. Variability in the pan-Atlantic domain (20°S–60°N, 80°W–20°E) is analyzed during 1901–2006.

Table 1.
Table 2.

Sensitivity tests: T0 is the primary analysis. The Atlantic analysis domain is 20°S–60°N, 80°W–20°E, while the Atlantic–Indian analysis domain extends farther eastward to 120°E. For each test, only the attributes different from T0 are indicated.

Table 2.
Table 3.

Sensitivity test results: Column 1 shows the leading modes identified in the primary analysis (T0); numbers following the name indicate the percentage variance explained by the mode and its rank, respectively. Columns 2–9 list attributes of the leading modes in the eight sensitivity tests (T1–T8), with the three slash-delimited numbers indicating correlation between the test case and primary analysis PCs, the percentage variance explained by that mode, and its rank in the test analysis.

Table 3.

1

In GN2008, the Pacific analysis domain included some grid points in the Atlantic basin (westward of 60°W). The Pacific analysis was repeated with the Atlantic basin points masked out to assess their contribution. The new analysis confirms that the large amplitude of the AMO-Pac in the western tropical North Atlantic is not due to the inadvertent inclusion of a few Atlantic grid points in our Pacific basin analysis.

2

The value of L is chosen in view of the presence of interannual SST variability. Sensitivity tests show that separate analysis with a larger L is not necessary to characterize the spatiotemporal structure of the lower-frequency modes.

Save
  • Carton, J. A., X. Cao, B. S. Giese, and A. M. Da Silva, 1996: Decadal and interannual SST variability in the tropical Atlantic Ocean. J. Phys. Oceanogr., 26 , 11651175.

    • Search Google Scholar
    • Export Citation
  • Chiang, J. C. H., and D. J. Vimont, 2004: Analogous Pacific and Atlantic meridional modes of tropical atmosphere–ocean variability. J. Climate, 17 , 41434158.

    • Search Google Scholar
    • Export Citation
  • Czaja, A., and C. Frankignoul, 2002: Observed impact of Atlantic SST anomalies on the North Atlantic Oscillation. J. Climate, 15 , 606623.

    • Search Google Scholar
    • Export Citation
  • Dommenget, D., and M. Latif, 2000: Interannual to decadal variability in the tropical Atlantic. J. Climate, 13 , 777792.

  • Eden, C., and T. Jung, 2001: North Atlantic interdecadal variability: Oceanic response to the North Atlantic Oscillation (1865–1997). J. Climate, 14 , 676691.

    • Search Google Scholar
    • Export Citation
  • Eden, C., and J. Willebrand, 2001: Mechanism of interannual to decadal variability of the North Atlantic circulation. J. Climate, 14 , 22662280.

    • Search Google Scholar
    • Export Citation
  • Enfield, D. B., and D. A. Mayer, 1997: Tropical Atlantic SST variability and its relation to El Niño–Southern Oscillation. J. Geophys. Res., 102 , 929945.

    • Search Google Scholar
    • Export Citation
  • Enfield, D. B., A. M. Mestas-Nuñez, D. A. Mayer, and L. Cid-Serrano, 1999: How ubiquitous is the dipole relationship in tropical Atlantic sea surface temperatures. J. Geophys. Res., 104 , 78417848.

    • Search Google Scholar
    • Export Citation
  • Enfield, D. B., A. M. Mestas-Nuñez, and P. J. Trimble, 2001: The Atlantic multidecadal oscillation and its relation to rainfall and river flows in the continental U.S. Geophys. Res. Lett., 28 , 20772080.

    • Search Google Scholar
    • Export Citation
  • Guan, B., and S. Nigam, 2008: Pacific sea surface temperatures in the twentieth century: An evolution-centric analysis of variability and trend. J. Climate, 21 , 27902809.

    • Search Google Scholar
    • Export Citation
  • Hare, S. R., and N. J. Mantua, 2000: Empirical evidence for North Pacific regime shifts in 1977 and 1989. Prog. Oceanogr., 47 , 103145.

    • Search Google Scholar
    • Export Citation
  • Hoerling, M. P., J. W. Hurrell, and T. Xu, 2001: Tropical origins for recent North Atlantic climate change. Science, 292 , 9092.

  • Hurrell, J. W., M. P. Hoerling, A. Phillips, and T. Xu, 2004: Twentieth-century North Atlantic climate change. Part I: Assessing determinism. Climate Dyn., 23 , 371389.

    • Search Google Scholar
    • Export Citation
  • Kawamura, R., 1994: A rotated EOF analysis of global sea surface temperature variability with interannual and interdecadal scales. J. Phys. Oceanogr., 24 , 707715.

    • Search Google Scholar
    • Export Citation
  • Lanzante, J. R., 1996: Lag relationships involving tropical sea surface temperatures. J. Climate, 9 , 25682578.

  • Latif, M., 2001: Tropical Pacific/Atlantic Ocean interactions at multi-decadal time scales. Geophys. Res. Lett., 28 , 539542.

  • Latif, M., E. Roeckner, U. Mikolajewicz, and R. Voss, 2000: Tropical stabilization of the thermohaline circulation in a greenhouse warming simulation. J. Climate, 13 , 18091813.

    • Search Google Scholar
    • Export Citation
  • Mantua, N. J., S. R. Hare, Y. Zhang, J. M. Wallace, and R. C. Francis, 1997: A Pacific interdecadal climate oscillation with impacts on salmon production. Bull. Amer. Meteor. Soc., 78 , 10691079.

    • Search Google Scholar
    • Export Citation
  • Mestas-Nuñez, A. M., and D. B. Enfield, 1999: Rotated global modes of non-ENSO sea surface temperature variability. J. Climate, 12 , 27342746.

    • Search Google Scholar
    • Export Citation
  • Nigam, S., 2003: Teleconnections. Encyclopedia of Atmospheric Sciences, J. R. Holton et al., Eds., Academic Press, 2243–2269.

  • Nobre, P., and J. Shukla, 1996: Variations of sea surface temperature, wind stress, and rainfall over the tropical Atlantic and South America. J. Climate, 9 , 24642479.

    • Search Google Scholar
    • Export Citation
  • Overland, J. E., and R. W. Preisendorfer, 1982: A significance test for principal components applied to a cyclone climatology. Mon. Wea. Rev., 110 , 14.

    • Search Google Scholar
    • Export Citation
  • Parker, D., C. Folland, A. Scaife, J. Knight, A. Colman, P. Baines, and B. Dong, 2007: Decadal to multidecadal variability and the climate change background. J. Geophys. Res., 112 , D18115. doi:10.1029/2007JD008411.

    • Search Google Scholar
    • Export Citation
  • Penland, C., and L. Matrosova, 2006: Studies of El Niño and interdecadal variability in tropical sea surface temperatures using a nonnormal filter. J. Climate, 19 , 57965815.

    • Search Google Scholar
    • Export Citation
  • Rayner, N. A., D. E. Parker, E. B. Horton, C. K. Folland, L. V. Alexander, D. P. Rowell, E. C. Kent, and A. Kaplan, 2003: Global analyses of sea surface temperature, sea ice, and night marine air temperature since the late nineteenth century. J. Geophys. Res., 108 , 4407. doi:10.1029/2002JD002670.

    • Search Google Scholar
    • Export Citation
  • Richman, M. B., 1986: Rotation of principal components. Int. J. Climatol., 6 , 293335.

  • Rodwell, M. J., and C. K. Folland, 2002: Atlantic air–sea interaction and seasonal predictability. Quart. J. Roy. Meteor. Soc., 128 , 14131443.

    • Search Google Scholar
    • Export Citation
  • Ruiz-Barradas, A., J. A. Carton, and S. Nigam, 2000: Structure of interannual-to-decadal climate variability in the tropical Atlantic sector. J. Climate, 13 , 32853297.

    • Search Google Scholar
    • Export Citation
  • Schmittner, A., C. Appenzeller, and T. F. Stocker, 2000: Enhanced Atlantic freshwater export during El Niño. Geophys. Res. Lett., 27 , 11631166.

    • Search Google Scholar
    • Export Citation
  • Ting, M., Y. Kushnir, R. Seager, and C. Li, 2008: Forced and internal 20th century SST trends in the North Atlantic. J. Climate, 22 , 14691481.

    • Search Google Scholar
    • Export Citation
  • Tourre, Y. M., B. Rajagopalan, and Y. Kushnir, 1999: Dominant patterns of climate variability in the Atlantic Ocean during the last 136 years. J. Climate, 12 , 22852299.

    • Search Google Scholar
    • Export Citation
  • Trenberth, K. E., and D. J. Shea, 2006: Atlantic hurricanes and natural variability in 2005. Geophys. Res. Lett., 33 , L12704. doi:10.1029/2006GL026894.

    • Search Google Scholar
    • Export Citation
  • Visbeck, M. H., J. W. Hurrell, L. Polvani, and H. M. Cullen, 2001: The North Atlantic Oscillation: Past, present, and future. Proc. Natl. Acad. Sci. USA, 98 , 1287612877.

    • Search Google Scholar
    • Export Citation
  • Weare, B. C., and J. S. Nasstrom, 1982: Examples of extended empirical orthogonal function analyses. Mon. Wea. Rev., 110 , 481485.

  • Zebiak, S. E., 1993: Air–sea interaction in the equatorial Atlantic region. J. Climate, 6 , 15671586.

  • Fig. 1.

    Pacific natural SST variability and the secular trend: EEOF modes (mature-phase patterns) of Pacific (20°S–60°N, 120°E–60°W) SSTs during the twentieth century (1900–2007); see GN2008 for details, and the appendix for a brief description. Atlantic footprints are obtained from regressions of the PCs. Solid (dashed) contours denote positive (negative) values and the zero contour is suppressed. Contour interval is 0.1 K.

  • Fig. 2.

    Ratio (%) of the variance of the residual and raw SST anomalies. Residual SSTs are obtained by linearly removing Pacific natural variability and the secular trend (see Fig. 1) from the raw data.

  • Fig. 3.

    (top) PCs of residual Atlantic SST variability in the twentieth century (1901–2006). Tick marks on the vertical axis are drawn every three units. The original PCs are shaded, while heavily smoothed versions (from 50 applications of a 1–2–1 smoother) are shown using solid black lines. (bottom) Autocorrelations of the PCs. Note that the AMO autocorrelation is shown using the top scale.

  • Fig. 4.

    Time evolution of the leading four modes of residual Atlantic SST variability. (a)–(c) AMO-Atl. (d)–(f) Decaying phase of Atlantic Niño. (g)–(i) Growing phase of Atlantic Niño. (j)–(l) Tripole. Note that the AMO evolution is displayed at four-season intervals, the Atlantic Niño modes at two-season intervals, and the tripole mode at three-season intervals. Maps are obtained by regressing lead–lagged residual SST anomalies onto the PCs shown in Fig. 2 (top), with the label t denoting simultaneous regressions. Solid (dashed) contours denote positive (negative) values and the zero contour is suppressed. Contour interval is 0.1 K.

  • Fig. 5.

    (a) AMO indexes based on reconstructed and raw (but detrended) SST data. Reconstructions are based on individual EEOF mode or mode combinations, indicated by the curve key; the numbers show the correlation between the reconstructed and the raw AMO (i.e., AMO-Tot). The AMO indexes are smoothed by 10 applications of a 1–2–1 smoother before being displayed. (b)–(d) Correlations between raw Atlantic SST anomalies and the AMO index based on (b) residual Atlantic mode 1 (the AMO-Atl mode), (c) Pacific natural variability, and (d) raw SSTs without the secular trend (see Fig. 1). Correlation calculations in (a)–(d) are all based on unsmoothed indexes. Contour interval is 0.1. Contour/shading threshold is 0.2.

  • Fig. 6.

    Cross correlations of AtlNiño and AtlNiño+ PCs, and of each with the Atl3 index (SST anomalies averaged over 3°S–3°N, 20°W–0°; see text for details), at various seasonal lead–lags. The curve key is in the upper-left corner, with the following plotting convention: when r(A,B) > 0 for t < 0, B leads A; if r > 0 for t > 0, B lags A. Cross correlations show that AtlNiño (AtlNiño+) PC leads (lags) the Atl3 index by two seasons, and consistently, AtlNiño+ lags AtlNiño by four seasons.

  • Fig. 7.

    (left) Two-season lead, (middle) simultaneous, and (right) two-season lag regressions of 500-hPa height on the tripole mode PC. Contour interval is 3 m. The 30° and 60°N latitude circles are indicated by the dotted lines.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 943 525 20
PDF Downloads 388 74 8