SEARCH

SEARCH BY CITATION

Keywords:

  • ACW;
  • El Niño;
  • meridional teleconnections;
  • multidecadal change

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methods
  5. 3. Evolutionary Development of El Niño Before and After 1977
  6. 4. Evolutionary Development of the ACW Before and After 1977
  7. 5. Evolutionary Development of the GEW Before and After 1977
  8. 6. Directly Linking the Evolution of the ACW to the Evolution of El Niño
  9. 7. Possible Sources of Multidecadal Change in the ACW Over the Southern Hemisphere
  10. 8. Discussion and Conclusions
  11. Acknowledgments
  12. References

[1] The slow eastward phase propagation of covarying sea surface temperature (SST) and sea level pressure (SLP) anomalies in the Antarctic Circumpolar Wave (ACW) persisted in the midlatitude Southern Ocean from 1950 to 2001. Its northern extent reached into the subtropical South Indian and South Pacific oceans, where it influenced the magnitude and phase of El Niño in the eastern equatorial Pacific Ocean. This is corroborated by the observation that multidecadal changes in the ACW can also explain multidecadal changes in the way El Niño evolved over the last half of the twentieth century. Before 1977, El Niño evolved from the slow eastward phase propagation of a coupled SST/SLP wave across the subtropical South Pacific Ocean to South America, then equatorward along the eastern boundary to the equator. During this epoch, the ACW expanded equatorward into a warmer subtropical South Pacific Ocean, accounting for the subtropical coupled SST/SLP wave that initiated El Niño. After 1977, El Niño evolved from the slow eastward phase propagation of a coupled SST/SLP wave across the tropical Pacific Ocean. During this epoch, the ACW receded from a colder subtropical South Pacific Ocean but expanded into a warmer subtropical South Indian Ocean. There it spawned a coupled SST/SLP wave directed equatorward into the Warm Pool north of Australia, thereafter propagating eastward into the eastern equatorial Pacific Ocean to initiate El Niño. If this pattern of global warming persists into the 21st century, then the ACW will continue to influence El Niño through the South Indian Ocean.

1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methods
  5. 3. Evolutionary Development of El Niño Before and After 1977
  6. 4. Evolutionary Development of the ACW Before and After 1977
  7. 5. Evolutionary Development of the GEW Before and After 1977
  8. 6. Directly Linking the Evolution of the ACW to the Evolution of El Niño
  9. 7. Possible Sources of Multidecadal Change in the ACW Over the Southern Hemisphere
  10. 8. Discussion and Conclusions
  11. Acknowledgments
  12. References

[2] The intense warm sea surface temperature (SST) anomalies in the eastern equatorial Pacific Ocean that characterize El Niño occurred every 3 to 7 years over the twentieth century in association with the evolution of covarying SST and SLP anomalies over the globe, called the El Niño-Southern Oscillation (ENSO) [e.g., Allan, 2000; White and Tourre, 2003]. This global ENSO signal is composed of a global standing mode in covarying SST and SLP anomalies [e.g., Bjerknes, 1966; Tourre and White, 1995; Allan, 2000] and an eastward-propagating global traveling wave in covarying SST and SLP anomalies [Yasunari, 1987; Tourre and White, 1997; White and Cayan, 2000; White et al., 2002]. The global standing mode of ENSO is believed to be driven by El Niño, teleconnected globally by anomalous meridional Hadley cells and zonal Walker cells stemming from the warm (cool) SST anomalies in the eastern (western) equatorial Pacific Ocean [e.g., Bjerknes, 1969; Webster, 1994; Trenberth et al., 2000]. On the other hand, the global traveling wave of ENSO propagates slowly eastward around the tropical global ocean as a coupled SST/SLP wave, composed of global zonal wave numbers 1 and 2, with phase speeds ranging from 45° to 90° of longitude per year, taking 4 to 8 years to circle the globe [White and Cayan, 2000]. The warm SST phase of this global ENSO wave (GEW) coincided with the warm SST phase of El Niño over most of the twentieth century [White and Cayan, 2000], but it remains to be seen which is more fundamental.

[3] Already, covarying SST and SLP anomalies in the extratropical South Pacific Ocean have been shown to lead El Niño in the eastern equatorial Pacific Ocean. van Loon and Shea [1985] observed covarying warm SST and low SLP anomalies in the western and central subtropical South Pacific Ocean during the year prior to El Niño (i.e., in 1951, 1953, 1957, 1972, and 1982), finding them propagating eastward into the eastern tropical Pacific Ocean during the following El Niño year [van Loon and Shea, 1987]. Similar results were obtained by Krishnamurti et al. [1986]. This was subsequently confirmed by Allan et al. [1994], who found the eastward propagation of covarying SST and SLP anomalies across the subtropical South Pacific Ocean operating during the first half of the twentieth century as well. In these earlier studies, the slow eastward propagation of seasonal and interannual SST and SLP anomalies spanned the latitude domain in the South Pacific Ocean from the tropics near 20°S to the sea ice edge around Antarctica near 65°S, presaging the Antarctic Circumpolar Wave (ACW) observed from 40°S to 65°S over the same domain from 1982 to 1995 [White and Peterson, 1996; Jacobs and Mitchell, 1996]. Thus it comes as no surprise to find Simmonds and Jacka [1995] and Yuan and Martinson [2000] observing interannual sea ice extent (SIE) anomalies in the eastern Indian and western Pacific sectors of the Southern Ocean leading El Niño by 1 to 2 years. Along a different track, White and Peterson [1996] followed covarying SST and SLP anomalies in ACW from the Indian sector of the Southern Ocean equatorward into the tropical Warm Pool north of Australia during the year prior to El Niño (i.e., in 1982, 1986, 1991). The latter had already been shown to propagate slowly eastward across the tropical Indian and western and central Pacific oceans, influencing the phase and magnitude of El Niño in the eastern equatorial Pacific Ocean during the following year [Yasunari, 1987; Wang, 1995; Tourre and White, 1997; White and Cayan, 2000].

[4] The ACW occurs on the same 3- to 7-year period–scale as ENSO, propagating slowly eastward around the Southern Ocean as a coupled wave in covarying SST, SLP, and SIE anomalies, composed of global zonal wave number 2, with zonal-average phase speed of ∼45° longitude per year, taking 8 years to circle the globe [White and Peterson, 1996; Jacobs and Mitchell, 1996]. It has been observed spanning the extratropical Southern Ocean from the sea ice edge around Antarctica near 65°S to the subtropical South Atlantic, South Indian and western and central South Pacific oceans near 30°S [Peterson and White, 1998; White et al., 1998, 2002]. The ACW does not propagating zonally around the Southern Ocean. Instead, streamlines of the ACW follow a broad meandering path which is displaced or expanded equatorward to within 30° latitude of the equator in the eastern Atlantic, Indian and western and central Pacific sectors and then displaced or retracted poleward to propagate through Drake Passage [White and Chen, 2002]. The ACW travels eastward slower than the tropical GEW except in the vicinity of Drake Passage, where it is phase locked (via meridional atmospheric teleconnections) to the eastward propagation of the GEW across the Warm Pool north of Australia [White et al., 2002]. Thus, in this portion of the Southern Ocean the ACW is driven by the GEW as the latter propagates eastward across the Warm Pool. However, over the remainder of the Southern Ocean the ACW propagates eastward as a coupled SST/SLP wave, relatively independent from tropical atmospheric teleconnections [Cai and Baines, 2001; White et al., 2002]. Furthermore, the ACW in the Indian sector of the Southern Ocean spawns a coupled SST/SLP wave that has been observed propagating equatorward into the tropical Indian Ocean [Peterson and White, 1998]. The latter oceanic teleconnection provides a positive feedback to the GEW in the Warm Pool north of Australia, producing a resonant interaction between the GEW and ACW that reinforces the amplitudes of both via a global feedback loop in the southern hemisphere that takes ∼4 years to complete [White et al., 2002]. Thus it comes as no surprise that the GEW and the ACW are linked to one another over most of the southern hemisphere [White et al., 2002].

[5] The thermodynamics of the ACW are only partially understood. Initially, the observations of White and Peterson [1996], Jacobs and Mitchell [1996], and Peterson and White [1998], together with modeling efforts by Qiu and Jin [1997], Goodman and Marshall [1999], and Baines and Cai [2000], indicated that the ACW relied on the zonal advection of SST anomalies by the Antarctic circumpolar current (ACC) for its eastward phase propagation. However, White et al. [1998] realized that the ACW does not follow the main core of the ACC, nor does it propagate with similar speed. Furthermore, they constructed an ocean-atmosphere coupled model of the ACW that had the same phase relationships observed among the principal oceanic and atmospheric variables while yielding eastward phase propagation from ocean-atmosphere coupling alone. Subsequently, White and Chen [2002] and White et al. [2004] diagnosed the troposphere thermal and vorticity budgets involved in the atmospheric response to SST anomalies in the ACW all across the Southern Ocean. They found ocean-atmosphere coupling confined principally to the subtropical front in the eastern Atlantic, Indian, and western and central Pacific sectors of the Southern Ocean near 35°S and to the autumn/winter/spring sea ice edge around Antarctica near 65°S. They also found a feedback from the atmosphere to the ocean that could explain the eastward phase propagation of the ACW. These results indicated that the ACW is a coupled SST/SLP wave that follows two main paths around the Southern Ocean along which the coupling is facilitated by strong meridional gradients in the background thermal state (i.e., the subtropical front and the sea ice edge). Anomalous Ferrell cells driven in both latitude domains by anomalous SST-driven deep convection appear to link covarying SST and SLP anomalies in the ACW throughout the Southern Ocean [White et al., 2004].

[6] A number of studies have now found the amplitude and phase of the ACW fluctuating on decadal, interdecadal, and multidecadal period scales. During the modern era of satellite observations (i.e., >1979), Carril and Navarra [2001] found the ACW robust from 1983 to 1992 and from 1996 to 2000, but relatively weak in between from 1992 to 1996, coinciding with the extended El Niño of 1991 to 1995 [Allan and D'Arrigo, 1999]. Also, Connolley [2002] found the eastward phase propagation of the ACW superposed on a weaker standing mode of global zonal wave number 3 in the Southern Ocean, observed earlier by Cai and Baines [2001]. Before the modern era, relying on SLP anomalies inferred from satellite cloud photographs [Kalnay et al., 1996], a number of studies [Carril and Navarra, 2001; Simmonds, 2003; Connolley, 2002] demonstrated that the ACW disappeared in the high latitude Southern Ocean during the 1970s, with Simmonds [2003] finding the ACW returning to these latitudes in the 1960s. On the other hand, the eastward phase propagation of covarying SST and SLP anomalies observed by van Loon and Shea [1985, 1987] and Allan et al. [1994] in the midlatitude Southern Ocean indicated that the ACW was observed farther north between 30°S and 50°S over most of the earlier epoch.

[7] Since the ACW owes much of its existence as a coupled SST/SLP wave to thermal gradients in the subtropical front near 35°S and the sea ice edge around Antarctic near 65°S, the decadal, interdecadal, and multidecadal changes in the ACW observed during the last half of the twentieth century may have occurred in response to similar changes in the location and intensity of these thermal gradients. Specifically, the multidecadal change during the last half of the twentieth century brought warmer SST's to the central tropical Pacific and Indian oceans after 1977, associated with cooler SST's in the western and central subtropical South Pacific Ocean and to the midlatitude South Indian Ocean [Allan, 2000, pp. 22 and 23]. The latter appears to have weakened the subtropical front near 35°S in the western South Pacific Ocean whilst strengthening it near 40°S in the South Indian Ocean. This may have disrupted the ACW in the subtropical South Pacific Ocean whilst enhancing it in the South Indian Ocean. In the present study, we demonstrate this was indeed the case. Furthermore, we show that this had a profound effect on the evolutionary development of El Niño.

[8] The evolutionary development of covarying SST and SLP anomalies in the Pacific Ocean leading to El Niño has also been observed to change dramatically before and after 1977. Before 1977, Hickey [1975] and others found the evolution of the 1958, 1969, and 1973 El Niños initiated by the collapse of the southeast trade winds in the South Pacific Ocean, associated with northward propagation of warm ST anomalies along the west coast of South America to the equator. This was confirmed by Wang [1995], who found El Niños prior to 1977 deriving from the eastward phase propagation of SST/SLP waves propagating eastward across the subtropical South Pacific Ocean, leading the equatorward propagation along the west coast of South America, as observed earlier by van Loon and Shea [1985, 1987] and Allan et al. [1994]. On the other hand, after 1977, Wang [1995] found the evolution of the 1983, 1987, 1992 El Niños initiated by the slow eastward phase propagation of a coupled SST/SLP wave propagating eastward along the tropical Pacific Ocean from the Warm Pool north of Australia, observed earlier by Yasunari [1987] and subsequently confirmed by Tourre and White [1997] and White and Cayan [2000].

[9] In the present study, we find the ACW persisting in the midlatitude Southern Ocean from 30°S to 50°S over the last half of the twentieth century. Moreover, we find it displaced or expanded equatorward into a warmer subtropical South Pacific Ocean prior to 1977 but displaced or retracted poleward of a cooler subtropics after 1977. In the South Indian Ocean, we find the ACW displaced or expanded equatorward toward a warmer subtropical South Indian Ocean after 1977 but displaced or retracted poleward from a cooler subtropical South Indian Ocean prior to 1977. During both epochs, we find the ACW spawning equatorward propagating coupled SST/SLP waves that influence the magnitude and phase of El Niño from either the subtropical South Pacific Ocean (prior to 1977) and from the tropical South Indian Ocean (after 1977), consistent with multi-decadal changes in El Niño evolution observed by Wang [1995]. Moreover, we can form an educated guess as to how the ACW will influence El Niño in the 21st century, provided the present pattern of global warming persists into the next century.

2. Data and Methods

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methods
  5. 3. Evolutionary Development of El Niño Before and After 1977
  6. 4. Evolutionary Development of the ACW Before and After 1977
  7. 5. Evolutionary Development of the GEW Before and After 1977
  8. 6. Directly Linking the Evolution of the ACW to the Evolution of El Niño
  9. 7. Possible Sources of Multidecadal Change in the ACW Over the Southern Hemisphere
  10. 8. Discussion and Conclusions
  11. Acknowledgments
  12. References

[10] We analyze three variables from the National Centers for Environmental Prediction and National Center for Atmospheric Research (NCEP/NCAR) reanalysis [Kalnay et al., 1996]. We analyze monthly ST, SLP, and zonal surface wind (ZSW) on a 2.5° latitude-longitude grid over the globe for 52 years from 1950 through 2001. The ST is sea surface temperature (SST) over the ocean and surface land temperature (SAT) over the continents. Over the ocean the NCEP/NCAR reanalysis incorporates the COADS surface marine weather observations [Slutz et al., 1985], the Reynolds' sea surface temperature (SST) analysis [Reynolds and Marsico, 1993], and atmospheric soundings from weather ships and satellites, with SLP estimates in the Southern Ocean inferred from satellite cloud photographs [Kalnay et al., 1996].

[11] Monthly anomalies of ST, SLP, and ZSW were computed about long-term monthly means defining the mean annual cycle over the 52-year record. Zonal wave number–frequency spectra of monthly ST and SLP anomalies along the equator in the Indo-Pacific Ocean [White and Cayan, 2000] and along 56°S in the Southern Ocean [White and Peterson, 1996; White and Chen, 2002] yield peak spectral energy density for interannual signals of 3- to 7-year period, significant at the 95% confidence level. Recently, this was refined with the multitaper-method singular-value-decomposition (MTM-SVD) analysis applied to global ST and SLP data sets extending from 40°S to 60°N over the twentieth century [White and Tourre, 2003]. This study demonstrated that the relatively broad spectral peak from 3- to 7-year period observed in power spectra is composed of three interannual signals near 2.9-, 3.5-, and 5.5-year period, together with a quasi-biennial signal near 2.2-year period and a quasi-decadal signal near 11-year period, each significant at the 99% confidence level. Thus, in the present study, we isolate the 3.5- and 5.5-year period interannual signals from higher and lower- frequency signals by band-pass filtering using a period admittance window with half power points at 3- and 7-year period [Kaylor, 1977]. Prior to band-pass filtering, we applied maximum entropy spectral analysis [Andersen, 1974] to extend the time sequences by an amount equal to half the filter width on both ends of the record. This allowed the half-power point criterion to be preserved in the frequency response function of the filter, and allowed over half the variance of the signal at the end points to be faithfully represented [White, 2000]. Since the filter response function of Kaylor [1977] is flat, with steep sides and negligible side lobes, the interannual signals near 3.5- and 5.5-year period are effectively isolated from the 2.2-, 2.9-, and 11-year period signals [White and Tourre, 2003].

[12] For the 52-year record, this band-pass filtering yields ∼10 cycles of interannual variability, yielding a conservative estimate of ∼20 effective temporal degrees of freedom, with ∼2 effective degrees of freedom allowed for each cycle [Snedecor and Cochran, 1980]. The band-pass filtering procedure reduces standard errors of the ST and SLP anomalies (i.e., 0.2°C and 0.5 hPa) by a factor of 6 or so; that is, by the square root of 36 nearly independent monthly estimates in the low-pass portion of the band-pass filter determined from autocorrelation analysis of monthly anomalies. This means that standard errors for interannual anomalies are approximately ±0.03° for ST and ±0.08 hPa for SLP. This is well below the contour intervals of 0.1°C and 0.2 hPa used to resolve the eastward phase propagation of the ACW in animation sequences of ST and SLP anomalies (e.g., Figures 1a and 1b).

image

Figure 1a. Animation sequences of maps displaying the evolution of interannual ST and SLP anomalies over the Indo-Pacific Ocean for the 2.5 years leading to January 1958 and January 1973 during El Niño years. Each map in the animation sequence extends zonally around the Indo-Pacific Ocean from 30°E to 60°W in the latitude band 50°N to 50°S. Individual maps are 6 months apart. Positive (negative) ST and SLP anomalies are yellow-to-red (blue), yielding phase and relative magnitude. Color intervals are 0.1°C and 0.2 hPa for ST and SLP anomalies, respectively. The sense of propagation leading to El Niño comes from following the warm covarying ST anomalies (yellow-to-red) and low SLP anomalies (blue) eastward across the subtropical South Pacific Ocean to South America, then equatorward along the eastern boundary to the equator from one map to the next in each animation sequence.

Download figure to PowerPoint

image

Figure 1b. Same as Figure 1a, but for the 2.5 years leading to January 1987 and January 1998 during El Niño years. The sense of propagation leading to El Niño comes from following covarying warm ST anomalies (yellow-to-red) and low SLP anomalies (blue) eastward across the tropical Pacific Ocean from the Warm Pool north of Australia to the eastern equatorial Pacific Ocean next to South America from one map to the next in each animation sequence.

Download figure to PowerPoint

[13] The space-time evolution of the interannual ST, SLP, and ZSW anomalies is represented by dominant modes in the complex empirical orthogonal function (CEOF) analysis [Preisendorfer and Mobley, 1988]. These CEOF modes yield real (IT) and imaginary (IS) amplitude time sequences modifying real (RS) and imaginary (IS) spatial patterns, respectively. These spatial patterns yield CEOF phase sequences (that is, (RS)cos(θ) + (IS)sin(θ)) that characterize the climatological-average evolution of ST, SLP, and ZSW anomalies as a function of phase (θ) over different epochs of the 52-year record. The CEOF analysis also produces a complex product (i.e., (RT)(RS) + (IT)(IS)) that yields the contribution of the dominant CEOF mode to the original time sequence of mapped anomalies. A full discussion of the theory and application of the CEOF analysis is given by von Storch and Zwiers [1999].

3. Evolutionary Development of El Niño Before and After 1977

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methods
  5. 3. Evolutionary Development of El Niño Before and After 1977
  6. 4. Evolutionary Development of the ACW Before and After 1977
  7. 5. Evolutionary Development of the GEW Before and After 1977
  8. 6. Directly Linking the Evolution of the ACW to the Evolution of El Niño
  9. 7. Possible Sources of Multidecadal Change in the ACW Over the Southern Hemisphere
  10. 8. Discussion and Conclusions
  11. Acknowledgments
  12. References

[14] El Niño occurred every 3 to 7 years over the 52-year record from 1950 through 2001, occurring in 1953, 1958, 1963, 1966, 1969, 1973, 1977, 1983, 1987, 1992, 1995, and 1998 [e.g., Allan, 2000; White and Tourre, 2003]. The evolution of covarying ST and SLP anomalies for the 2 to 3 years prior to each El Niño was led by the slow eastward and northward phase propagation of covarying warm ST and low SLP anomalies across the subtropical South Pacific Ocean before 1977, and by the slow eastward phase propagation of covarying warm ST and low SLP across the equatorial Indo-Pacific Ocean after 1977 [Wang, 1995]. Examples of these two types of El Niño evolution are displayed in the animation sequences of covarying ST and SLP anomalies for the 2.5 years leading up to the 1958 and 1973 El Niños (in the epoch before 1977) and the 1987 and 1998 El Niños (in the epoch after 1977).

[15] During the 2.5 years prior to the 1958 and 1973 El Niños (Figure 1a), warm ST anomalies in the eastern equatorial Pacific Ocean during El Niño began to appear 1 year earlier along the west coast of South America, propagating slowly equatorward, and during the 1 year before that propagating slowly eastward into the coast of South America across the subtropical South Pacific Ocean. This slow eastward propagation of warm ST anomalies across the subtropical South Pacific Ocean near 30°S was accompanied by a similar propagation in low SLP anomalies, the latter displaced ∼90° of phase to the west of the former, yielding poleward surface wind anomalies over warm ST anomalies as observed in the ACW [White and Peterson, 1996; White and Chen, 2002]. The eastward propagation in covarying warm ST and low SLP anomalies across the equatorial Pacific Ocean, observed by White and Cayan [2000], was weak during the evolution of these El Niños.

[16] During the 2.5 years leading to the 1987 and 1998 El Niños (Figure 1b), warm ST anomalies in the eastern equatorial Pacific Ocean during El Niño began to appear 1.5 year earlier in the western and central tropical Pacific Ocean, propagating slowly eastward, and during the 1 year before that propagating slowly eastward from the eastern tropical Indian Ocean. This slow eastward propagation of warm ST anomalies across the tropical Indo-Pacific Ocean was accompanied by colocated low SLP anomalies, with westerly ZSW anomalies displaced ∼90° of phase to the west of warm ST anomalies, as observed in the GEW [White and Cayan, 2000]. No equatorward propagation in covarying warm ST and low SLP anomalies occurred prior to the 1987 El Niño along the west coast of South America, but it did occur prior to the 1998 El Niño. Even so, the evolutionary development of the 1998 El Niño was dominated by the slow eastward propagation of warm ST and low SLP anomalies across the tropical Pacific Ocean.

[17] The evolution of covarying ST and SLP anomalies in these four examples can be extended to each of the El Niños before and after 1977 by computing the complex empirical orthogonal function (CEOF) analysis of the two variables over the two epochs (see section 2 on data and methods). The dominant CEOF for the earlier epoch from 1950 to 1977 has an amplitude time sequence with peaks during the El Niños of 1953, 1958, 1966, 1969, 1973, and 1977 (Figure 2a, top). The corresponding phase sequences (Figure 2a, bottom) display eastward propagation across the subtropical South Pacific Ocean of warm ST and low SLP weights near 30°S, with the center of low SLP weights displaced to the west of the center of warm ST weights by ∼90° of phase, with little or no eastward phase propagation indicated in either of the two variables across the tropical Pacific Ocean.

image

Figure 2a. (top) Time sequences of amplitudes associated with the dominant CEOF modes of interannual ST and SLP anomalies over the Indo-Pacific Ocean for the 27 years from 1950 through 1977. Amplitudes are displayed in units of standard deviation. These dominant CEOF modes explain 36% and 43% of the total interannual ST and SLP variance, respectively, over the 27-year record. Amplitude time sequences for ST and SLP modes cross-correlate at 0.99, significant at the 95% confidence level for ∼12 effective degrees of freedom [Snedecor and Cochran, 1980]. (bottom) Phase sequences of maps associated with the dominant CEOF modes of interannual ST and SLP anomalies, with each map extending over the Indo-Pacific Ocean from 30°E to 60°W in the latitude band 50°N to 50°S. Each phase sequence extends from −180° of phase to 0° phase, or approximately one-half cycle of variability, leading to the eastern equatorial warming of El Niño during 0° phase at the bottom of the page. Positive (negative) ST and SLP weights are unshaded (shaded), yielding phase and relative magnitude. Contour intervals for spatial weights of both variables are given by 0.06 of one standard deviation. The sense of propagation leading to El Niño comes from following positive ST weights (unshaded) and negative SLP weights (shaded) eastward across the subtropical Pacific Ocean to South America, then equatorward along the eastern boundary to the equator from one map to the next in each phase sequence.

Download figure to PowerPoint

[18] The dominant CEOF mode for the later epoch has an amplitude time sequence with peaks during the El Niños of 1983, 1987, and 1998 (Figure 2b, top). This CEOF mode has little to say of what happened during the extended El Niño of 1990 to 1995 [Allan and D'Arrigo, 1999]. The corresponding phase sequences (Figure 2b, bottom) display eastward propagation across the tropical Pacific Ocean of covarying warm ST and low SLP weights, with weak eastward propagation of warm ST weights indicated in the subtropical South Pacific Ocean. Comparing the eastward propagation of SLP weights across the South Pacific Ocean before and after 1977 (Figure 2a, bottom, and Figure 2b, bottom), and focusing on the transition from La Niña to El Niño during lag −90°, finds the SLP weights significantly stronger in the Southern Hemisphere before 1977 than afterward.

image

Figure 2b. Same as Figure 2a, but for the 25 years from 1977 through 2001. These dominant CEOF modes explain 51% of the total interannual ST and SLP variance over the 25-year record. Amplitude time sequences for ST and SLP modes cross-correlate at 0.97, significant at the 95% confidence level for ∼8 effective degrees of freedom [Snedecor and Cochran, 1980]. The sense of propagation leading to El Niño comes from following positive ST weights (unshaded) and negative SLP weights (shaded) eastward across the tropical Pacific Ocean from the Warm Pool north of Australia to the eastern equatorial Pacific Ocean next to South America from one map to the next in each phase sequence.

Download figure to PowerPoint

[19] Examining these CEOF phase sequences more closely finds the warm ST weights in the eastern equatorial Pacific Ocean during El Niños before 1977 (Figure 2a) developing ∼6 months earlier than during El Niños after 1977 (Figure 2b). During the earlier epoch, the slow eastward propagation of covarying warm ST and low SLP weights of the GEW arrived in the central equatorial Pacific Ocean after warming had already begun in the eastern equatorial Pacific Ocean in response to equatorward propagation along the west coast of South America. This produced a double peak in warm ST weights along the equator during El Niño (lag 0°, Figure 2a, bottom) resulting from westerly ZSW weights forming near 150°W, west of and in response to the warm ST anomalies in the eastern equatorial ocean (see Figure 5 below). This double peak was absent during El Niño in the epoch after 1977 (lag 0°, Figure 2b, bottom), when the GEW penetrated eastward all the way to the west coast of South America.

4. Evolutionary Development of the ACW Before and After 1977

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methods
  5. 3. Evolutionary Development of El Niño Before and After 1977
  6. 4. Evolutionary Development of the ACW Before and After 1977
  7. 5. Evolutionary Development of the GEW Before and After 1977
  8. 6. Directly Linking the Evolution of the ACW to the Evolution of El Niño
  9. 7. Possible Sources of Multidecadal Change in the ACW Over the Southern Hemisphere
  10. 8. Discussion and Conclusions
  11. Acknowledgments
  12. References

[20] The evolution of the ACW was first observed in covarying SST, SLP, and SEI anomalies propagating slowly eastward across the Southern Ocean from 40°S to 65°S [White and Peterson, 1996; Peterson and White, 1998; Gloersen and White, 2001]. It has also been observed in sea level height anomalies over the latitude domain 40°S to 65°S [Jacobs and Mitchell, 1996]. Also, it has been observed in cyclone density anomalies [Simmonds, 2003], with low (high) SLP anomalies in the ACW coinciding with larger (smaller) cyclone density and cyclone intensity anomalies. This multivariate representation of the ACW was achieved in the modern satellite era (beginning in the late 1970s), with infrared radiometry for SST, microwave soundings for temperature and humidity profiles in the troposphere, microwave imaging for sea ice concentration, and surface altimetry for sea level height. Before the late 1970s, only SLP data were available in the Southern Ocean, inferred from satellite cloud photographs [Kalnay et al., 1996]. More recently, the ACW has been observed extending farther to the north than previously realized [White and Chen, 2002], with a main branch following the subtropical front near 35°S across the eastern Atlantic, Indian, and western and central Pacific sectors of the Southern Ocean, consistent with van Loon and Shea [1985, 1987] and Allan et al. [1994]. This realization allows SLP observations from volunteer observing ships (VOS) in the midlatitude Southern Ocean from 30°S to 50°S to be used in detecting the eastward phase propagation of the ACW extending back in time to the 1950's [Slutz et al., 1985].

[21] In the present study, we chose the anomalous zonal surface wind (ZSW) as the principal variable for detecting the ACW in the midlatitude South Indian and South Pacific oceans throughout the 52-year record [Kalnay et al., 1996]. This variable is colocated with SST anomalies in the ACW, with negative ZSW anomalies coinciding with warm SST anomalies [White et al., 1998]. We begin by displaying animation sequences of ZSW anomalies for the 3 years leading to the 1958, 1973, 1987, and 1998 El Niño (Figure 3). During each of these El Niños, we find negative (blue) ZSW anomalies occupying the central and eastern midlatitude South Pacific Ocean and positive (yellow-to-red) ZSW anomalies occupying the western midlatitude South Pacific Ocean. Both negative and positive ZSW anomalies display the characteristic tilt of the ACW from southwest-to-northeast direction observed in a rectangular projection [White and Peterson, 1996], which translates into a planetary spiraling of the ACW around the southern hemisphere in polar stereographic projection [White et al., 1998]. Prior to each of these El Niños, the positive and negative ZSW anomalies propagated slowly eastward over the latitude domain from 30°S to 50°S during both epochs before and after 1977. In each case, the phase speed was ∼45° longitude per year, as observed in the ACW [White and Peterson, 1996]. Moreover, the path of the ZSW anomalies across the midlatitude South Indian and South Pacific oceans hugged the south coast of Australia, propagating through New Zealand into the central midlatitude South Pacific Ocean before propagating southward toward Drake Passage, as observed in the ACW by White and Chen [2002]. So, here we find ZSW anomalies of the ACW reaching to within 25° latitude of the equator in the South Indian and South Pacific oceans.

image

Figure 3. Animation sequences of maps display the evolution of interannual ZSW anomalies over the extratropical South Indian and South Pacific oceans for the 3 years leading to January 1958, 1973, 1983, and 1998 in El Niño years. Each map in the animation sequence extends zonally from 30°E to 60°W in the latitude band 20°S to 60°S. Individual maps are 6 months apart. The sense of propagation revealing the ACW comes from following the dashed line tracing the eastward progress of negative ZSW anomalies from one map to the next in each animation sequence. The dashed line is in the same location in each animation sequence. Yellow-to-red (blue) colors indicate positive (negative) anomalies, with color intervals of 0.2 m s−1.

Download figure to PowerPoint

[22] The phase locking between the ACW and El Niño, which is evident in Figure 3, could already have been inferred from the positive feedback between the GEW in the tropics and the ACW in the Southern Ocean [White et al., 2002] and the association between the GEW and El Niño [White and Cayan, 2000]. Here we observe it directly; moreover, we find the slow eastward phase propagation of the ACW across the midlatitude South Indian and South Pacific oceans occurring in both epochs (i.e., before and after 1977). However, careful examination of the animation sequences during the year prior to each El Niño (Figure 3) finds the negative ZSW anomalies penetrating father north into the subtropical South Pacific Ocean (to within 25° latitude of the equator) before 1977. Examining the ACW leading to the other El Niño years on record finds similar patterns and propagation in the ZSW anomalies leading up to the stronger 1953, 1969, 1977, 1983, and 1992 El Niños but not leading up to the weaker 1963, 1966, and 1995 El Niño (not shown).

[23] These results are supported by comparing the dominant CEOF modes of interannual ZSW anomalies from 20°S to 60°S over the two epochs before and after 1977 (Figure 4). The amplitude time sequence for each (Figure 4a) displays peaks during the El Niños of each epoch (i.e., 1953, 1958, 1969, 1973, 1977 and 1983, 1987, 1992, 1998, respectively). The corresponding phase sequences (Figure 4b) display similar eastward phase propagation for negative ZSW weights from the region west of New Zealand during −180° phase to the eastern midlatitude South Pacific Ocean during 0° phase. The eastward phase propagation takes ∼2 years to transit this distance. These phase sequences display a combination of the Pacific-South America (PSA) standing mode and the ACW traveling wave [Cai and Baines, 2001; White et al., 2002]. In these CEOF phase sequences, the standing mode during both epochs had a larger amplitude than the ACW by 2 to 1; i.e., with the ACW exhibiting an amplitude of 2 contours southeast of New Zealand during −90° phase, and the sum of ACW and PSA exhibiting an amplitude of 6 contours upstream from Drake Passage during 0° phase. This is the opposite of that observed during the decade from 1983 to 1992 [White et al., 2002] when the amplitude of the ACW exceeded that of the PSA by 2 to 1. Even so, in both epochs, the ACW displayed similar propagation characteristics.

image

Figure 4. (a) Time sequences of amplitudes associated with the dominant CEOF modes of interannual ZSW anomalies for the 27-year record from 1950 through 1977 and for the 25-year record from 1977 through 2001. This allows the ACW to be examined over both portions of the 52-year record before and after 1977. Amplitudes are displayed in units of standard deviation. These dominant CEOF modes explain 57% and 54% of the total interannual ZSW variance, respectively, over the earlier and later portions of the 52-year record. (b) Phase sequences of maps associated with the dominant CEOF modes of interannual ZSW anomalies, with each map extending over the extratropical South Indian and South Pacific oceans from 30°E to 60°W in the latitude band 20°S to 60°S. Each phase sequence extends from −180° of phase to 0° phase, or approximately one-half cycle of variability, phase locked with El Niño during 0° phase. Positive (negative) ZSW weights are unshaded (shaded), yielding phase and relative magnitude. Contour intervals for spatial weights of both variables are given by 0.06 of one standard deviation. The sense of propagation revealing the ACW comes from following positive and negative ZSW weights eastward across the subtropical South Indian and South Pacific oceans from one map to the next in each phase sequence.

Download figure to PowerPoint

[24] Despite these similarities, subtle but profound differences occur in these two CEOF phase sequences (Figure 4b). During −90° phase, negative ZSW weights in the central South Pacific Ocean reached to within 25° latitude of the equator before 1977, but only to within 35° latitude of the equator after 1977. On the other hand, in the western South Indian Ocean, during 90° phase (Figure 4b), negative ZSW weights reached to within 25° latitude of the equator after 1977, but only to within 35° latitude of the equator before 1977.

5. Evolutionary Development of the GEW Before and After 1977

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methods
  5. 3. Evolutionary Development of El Niño Before and After 1977
  6. 4. Evolutionary Development of the ACW Before and After 1977
  7. 5. Evolutionary Development of the GEW Before and After 1977
  8. 6. Directly Linking the Evolution of the ACW to the Evolution of El Niño
  9. 7. Possible Sources of Multidecadal Change in the ACW Over the Southern Hemisphere
  10. 8. Discussion and Conclusions
  11. Acknowledgments
  12. References

[25] White and Cayan [2000] observed the slow eastward phase propagation of covarying warm ST and low SLP anomalies in the GEW across the tropical Indo-Pacific Ocean leading all the stronger El Niños of the twentieth century, including the 1953, 1958, 1966, 1969, 1973, 1983, 1987, and 1998 El Niños. They also found the GEW absent or replaced by westward propagating signals leading the weaker 1963 and 1995 El Niños. This result appears to contradict Wang [1995], who found the stronger 1958, 1966, and 1973 El Niños led by warm ST and low SLP anomalies propagating equatorward along the west coast of South America and, earlier, eastward across the subtropical South Pacific Ocean.

[26] We resolve this apparent contradiction by examining the dominant CEOF modes of ZSW anomalies over the tropical Indo-Pacific Ocean from 20°S to 20°N over the two epochs before and after 1977 (Figure 5). The amplitude time sequence for the dominant CEOF computed for the epoch before 1977 displays peaks for El Niños in 1953, 1958, 1969, 1973, and 1977, while that after 1977 displays peaks for El Niños in 1983, 1987, and 1998 (Figure 5a). These amplitude time sequences modulate corresponding phase sequences (Figure 5b), which display broadly similar eastward phase propagation during both epochs, with positive ZSW weights in the eastern tropical Indian Ocean during phase −180° propagating eastward to the central tropical Pacific Ocean during phase 0°. These phase sequences display a combination of global standing mode (i.e., the ENSO standing mode) and global traveling wave (i.e., the GEW), both separated from one another by White et al. [2002]. In these CEOF phase sequences, the ENSO standing mode in both epochs had a larger amplitude than the GEW by about a factor of 2, with the GEW exhibiting an amplitude of 2 contours traveling across the Warm Pool north of Australia during −90° phase, and the sum of the GEW and the ENSO standing mode exhibiting an amplitude of 6 contours in the eastern Indian Ocean during 0° phase. This is similar to that observed [White and Cayan, 2000] during the decade from 1980 to 1989 when the GEW was particularly robust. Thus, in both epochs before and after 1977, the GEW was present and displayed similar propagation characteristics.

image

Figure 5. Same as Figure 4, but for ZSW anomalies in the tropical Indo-Pacific Ocean over the latitude band 20°S to 20°N. These dominant CEOF modes explain 52% and 66% of the total interannual ZSW variance, respectively, over the earlier and later portions of the 52-year record. The sense of propagation revealing the GEW comes from following positive and negative ZSW weights eastward across the tropical Indo-Pacific Ocean from one map to the next in each phase sequence.

Download figure to PowerPoint

[27] Again, subtle but profound differences occur in these two CEOF phase sequences (Figure 5b), as occurred in the two CEOF modes revealing the ACW (Figure 4b). The negative ZSW weights in the GEW propagated eastward along the equator during −90° phase, extending to ∼140°W in the earlier epoch, but to ∼100°W in the later epoch (Figure 5b). Furthermore, the positive ZSW weights during −90° phase propagated eastward to 120°E in the earlier epoch, but only to 150°E in the later epoch.

[28] So, Wang [1995] found the 1958, 1966, and 1973 El Niños initiated by warm SST and low SLP anomalies propagating equatorward along the west coast of South America during an epoch when the GEW propagated eastward no farther than the central equatorial Pacific Ocean. While the GEW did not initiate El Niño during this epoch, it appears to have intensified it in the central equatorial Pacific Ocean, leading to the double peak in SST anomaly (Figure 2a, bottom).

6. Directly Linking the Evolution of the ACW to the Evolution of El Niño

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methods
  5. 3. Evolutionary Development of El Niño Before and After 1977
  6. 4. Evolutionary Development of the ACW Before and After 1977
  7. 5. Evolutionary Development of the GEW Before and After 1977
  8. 6. Directly Linking the Evolution of the ACW to the Evolution of El Niño
  9. 7. Possible Sources of Multidecadal Change in the ACW Over the Southern Hemisphere
  10. 8. Discussion and Conclusions
  11. Acknowledgments
  12. References

[29] There are two main paths across the southern hemisphere oceans along which covarying warm SST and low SLP anomalies in the ACW influence the phase and magnitude of El Niño (Figures 1 and 2). One path dominated the evolution of El Niño before 1977 with covarying warm ST and low SLP anomalies propagating equatorward onto the eastern equatorial Pacific Ocean from the south along the west coast of South America and, earlier, from their eastward propagation across the subtropical South Pacific Ocean [van Loon and Shea, 1987; Allan et al., 1994; Wang, 1995]. The latter occurred in association with the eastward propagation of the ACW across the subtropical South Pacific Ocean, reaching northward to within 25° latitude of the equator (Figures 3 and 4). Another path dominated the evolution of El Niño after 1977, with covarying warm SST and low SLP anomalies of the GEW propagating eastward across the tropical Indo-Pacific Ocean [Yasunari, 1987; Wang, 1995; Tourre and White, 1997; White and Cayan, 2000]. The latter was joined in the tropical Warm Pool north of Australia by a north branch of the ACW propagating equatorward through the South Indian Ocean [Peterson and White, 1998; White et al., 2002]. To verify these influences of the ACW on El Niño, we construct time-distance diagrams of SST and SLP anomalies along these two paths (Figures 6a and 6b).

image

Figure 6a. (top) A path in the subtropical South Pacific Ocean along which covarying SST and SLP anomalies propagated from north of New Zealand eastward across the subtropical South Pacific Ocean to South America, then equatorward along the eastern boundary to the equator, leading to El Niño during the epoch before 1977. (bottom) Time-distance diagrams of interannual SST and SLP anomalies along the path displayed in Figure 6a (top), for the 52 years from 1950 through 2001. Negative (positive) anomalies are shaded (unshaded), with contours of 0.1°C and 0.2 hPa for SST and SLP anomalies, respectively. The sloping dark lines, the same in each time-distance diagram, allow the alignment of covarying SST and SLP anomalies to be visualized, and to provide a slope from which the mean phase speed along the path can be estimated (see text for further details).

Download figure to PowerPoint

image

Figure 6b. Same as Figure 6a, but along a path in the Indo-Pacific Ocean along which interannual SST and SLP anomalies propagate eastward from south of Africa to the west coast of Australia, then equatorward into the tropical Warm Pool north of Australia, and then eastward across the tropical Pacific Ocean, leading to El Niño during the epoch after 1977.

Download figure to PowerPoint

[30] The eastward propagation of the ACW across the subtropical South Pacific Ocean influenced El Niño evolution before 1977 along a path that began north of New Zealand, extended eastward to the coast of South America, and then northward to the equator (Figure 6a, top). The time-distance diagrams of interannual SST and SLP anomalies along this path (Figure 6a, bottom) display eastward propagating signals in both variables over most of the earlier epoch from 1950 to 1977. Thus covarying warm SST and low SLP anomalies along this path led the 1953, 1958, 1966, 1969, 1973, and 1977 El Niño, but not the 1963 El Niño. In the subtropics, low SLP anomalies lag warm SST anomalies by ∼90° of spatial phase, but in the tropics they are co-located with warm ST anomalies. These anomalies took ∼2 years to transit this path from north of New Zealand to the eastern equatorial Pacific Ocean at an average phase speed of 50° to 60° longitude per year. After 1977, only the 1998 El Niño was led by covarying warm ST and low SLP anomalies along this path.

[31] The ACW influenced El Niño evolution along a path extending from the midlatitude South Indian Ocean to the eastern equatorial Pacific Ocean after 1977. This path began south of Africa, extended eastward to the west coast of Australia, then northward into the tropical Warm Pool north of Australia, where it pushed eastward into the eastern equatorial Pacific Ocean (Figure 6b, top). The time-distance diagrams of SST and SLP anomalies along this path (Figure 6b, bottom) display eastward propagating signals in both variables over most of the later epoch from 1977 through 2001. Thus covarying warm SST and low SLP anomalies led the 1983, 1987, 1992, and 1998 El Niño, but not the 1995 El Niño. In the extratropics, low SLP anomalies lag warm ST anomalies by ∼90° of spatial phase, but in the tropics they are colocated with warm ST anomalies. These anomalies took 4 to 5 years to transit this path from south of Africa to the eastern equatorial Pacific Ocean at an average phase speed of 50° to 60° of longitude per year. Prior to 1977, none of the earlier El Niños were led by covarying warm ST and SLP anomalies along this path.

[32] Now we display the Indo-Pacific Ocean distribution of phase velocity of interannual SLP anomalies computed for the decades of 1968 to 1977 and 1980 to 1989, representing the two epochs before and after 1977 (Figure 7). Zonal and meridional phase speeds were estimated at each 4° latitude by 8° longitude grid point from time-and space-lag covariance matrices of interannual SLP anomalies [White et al., 1985]. Then, streamlines were drawn connecting the distribution of phase velocities using the NCAR graphics package [Middleton-Link et al., 1995] (see online NCAR graphics documentation). These streamlines yield the path along which SLP anomalies propagate across the Indo-Pacific Ocean.

image

Figure 7. Distributions of streamlines of the average phase velocity of interannual SLP anomalies propagating eastward across the Indo-Pacific Ocean (30°E to 50°W) during decades when the ACW, GEW, and El Niño were robust before and after 1977; that is, (a) from 1968 through 1977 and (b) from 1980 through 1989. Streamlines were drawn using the NCAR graphic package [Middleton-Link et al., 1995], based on a uniform grid (4° latitude by 8° longitude) of average phase velocity computed for each decade. Phase speeds are inversely proportional to streamline width, with magnitude along the main paths given by slopes in Figures 6a and 6b. Thick gray bands in each map indicate meandering paths along which SLP anomalies in the ACW and the GEW propagated eastward in the tropics and extratropics, respectively, and a north branch (i.e., NACW) along which SLP anomalies in the ACW influence the GEW and El Niño.

Download figure to PowerPoint

[33] From 1968 to 1977, these streamlines delineate the path of the GEW in the tropics and the ACW in the extratropics as depicted in SLP anomaly (Figure 7a). The eastward streamlines of the ACW were directed to the northeast from the central midlatitude South Indian Ocean, hugging the south coast of Australia, pushing through New Zealand, and extending equatorward to within 25° latitude of the equator in the central South Pacific Ocean near 100°W, the latter much farther north than observed previously [e.g., White and Chen, 2002]. Moreover, an equatorward branch took the ACW signal from the central and eastern subtropical South Pacific Ocean into the eastern equatorial Pacific Ocean across a longitude band from 140°W to 80°W. This is consistent with the time-distance diagram of SLP anomalies in Figure 6a. The path of the GEW extended from the western tropical Indian Ocean into the central equatorial Pacific Ocean, subsequently diverted northward of the eastern equatorial Pacific Ocean, in association with the equatorward displacement or/or expansion of the ACW in the central and eastern subtropical South Pacific Ocean. This is consistent with the phase propagation of the GEW displayed in ZSW CEOF phase sequences (Figure 5). We also find a path of the ACW in the Indian sector of the Southern Ocean directed equatorward, extending northeastward into the tropical Warm Pool north of Australia, joining with the GEW propagating into the same region directly from the west. Both paths subsequently continue farther east into the central equatorial Pacific Ocean, bypassing the eastern equatorial Pacific Ocean to penetrate northeastward into the Caribbean Sea.

[34] From 1980 to 1989, the eastward streamlines of the ACW can be seen angling northeast from the southern tip of Africa, extending northeastward to the subtropical South Indian Ocean near 30°S, 110°E, then continuing southeastward along the south coast of Australia, diving south near New Zealand into the high-latitude central and eastern South Pacific Ocean (Figure 7b). This is similar to the path delineated by White and Chen [2002] from 1983 to 1992. Compared to the earlier decade (Figure 7a), the ACW penetrated farther north into the South Indian Ocean and receded to the middle and high latitude South Pacific Ocean, consistent with results displayed in Figure 4. Furthermore, those streamlines of the ACW penetrating to within 30°S of the equator in the eastern South Indian Ocean, in turn, spawned a north branch directed northeastward into the tropical Warm Pool north of Australia. There, a portion of this northern branch of the ACW joined the GEW propagating into the Warm Pool from the west north of the equator, but most of it turned eastward south of the equator and continued on into the eastern equatorial Pacific Ocean. Thus the warm phase of the ACW in the midlatitude South Indian Ocean took 3 to 4 years to influence the warm phase of El Niño along this path, consistent with that displayed in the time-distance diagrams of SLP anomalies (Figure 6b). Moreover, the path of the GEW extended from the western tropical Indian Ocean into the eastern equatorial Pacific Ocean, no longer diverted there by streamlines penetrating northward across the central and eastern subtropical South Pacific Ocean, as in the earlier decade (Figure 7a). Thus both the GEW and the ACW worked together to initiate El Niño in the eastern equatorial Pacific Ocean during this epoch.

7. Possible Sources of Multidecadal Change in the ACW Over the Southern Hemisphere

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methods
  5. 3. Evolutionary Development of El Niño Before and After 1977
  6. 4. Evolutionary Development of the ACW Before and After 1977
  7. 5. Evolutionary Development of the GEW Before and After 1977
  8. 6. Directly Linking the Evolution of the ACW to the Evolution of El Niño
  9. 7. Possible Sources of Multidecadal Change in the ACW Over the Southern Hemisphere
  10. 8. Discussion and Conclusions
  11. Acknowledgments
  12. References

[35] In section 1, we suggested that multidecadal changes in the SST anomaly pattern over the southern hemisphere might be responsible for multidecadal changes in the path or latitudinal expanse of the ACW (Figure 7), since the coupling thermodynamics of the ACW depend on the intensity of the subtropical front and associated troposphere thermal gradients [White and Chen, 2002]. Here we further develop this hypothesis by plotting the distribution of multidecadal SST anomalies over the southern hemisphere from 1968 to 1977 and from 1980 to 1989 (Figure 8), the same as for the streamlines of the GEW and ACW (Figure 7). Subsequently, we superimpose the ACW streamlines in Figure 7 onto the multidecadal SST anomalies in Figure 8.

image

Figure 8. Distributions of multidecadal SST anomalies for the two decades (a) from 1968 through 1977 and (b) from 1980 through 1989. The multidecadal SST anomalies are computed as the 10-year average for each decade (i.e., 1968–1977 and 1980–1989), subtracted from the mean over the period 1968 to 1989. The resulting warm (cool) SST anomalies are unshaded (shaded), with contour interval of 0.1°C. Superimposed on these SST anomaly distributions are the main path of the ACW and its north branch (NACW), the latter influencing the phase and magnitude of El Niño in the eastern equatorial Pacific Ocean, repeated from Figure 7.

Download figure to PowerPoint

[36] During the earlier epoch (i.e., from 1968 to 1977), we find the South Indian Ocean cooler by ∼0.1°C and the western and central subtropical South Pacific Ocean from 10°S to 40°S warmer by as much as ∼0.3°C. This weakened the subtropical front in the South Indian Ocean near 40°S whilst strengthening it in the western and central South Pacific Ocean north of New Zealand near 35°S. In this latter region, the path of the ACW followed the intensified subtropical front north and east of New Zealand, continuing on to the coast of South America. During the later epoch (i.e., from 1980 to 1989), the situation was reversed, with a stronger subtropical front in the South Indian Ocean and a weaker one north of New Zealand. Over this latter region, the path of the ACW was diverted south of New Zealand, avoiding the cool SST anomalies north of New Zealand (weakening the subtropical front here), whilst penetrating much farther equatorward into the South Indian Ocean, following the warm SST anomalies into the eastern ocean. Here, we find the north branch of the ACW in the South Indian Ocean following warm SST anomalies into the Warm Pool north of Australia and all across the tropical South Pacific Ocean. It remains to test this hypothesis in a global coupled model of the ACW to determine whether these multidecadal SST patterns of variability with magnitude ≤0.3°C can steer the ACW in the manner indicated.

8. Discussion and Conclusions

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methods
  5. 3. Evolutionary Development of El Niño Before and After 1977
  6. 4. Evolutionary Development of the ACW Before and After 1977
  7. 5. Evolutionary Development of the GEW Before and After 1977
  8. 6. Directly Linking the Evolution of the ACW to the Evolution of El Niño
  9. 7. Possible Sources of Multidecadal Change in the ACW Over the Southern Hemisphere
  10. 8. Discussion and Conclusions
  11. Acknowledgments
  12. References

[37] First, we confirmed that the evolution of El Niño was fundamentally different in epochs from 1950 through 1976 and from 1977 through 2001, consistent with Wang [1995]. Before 1977, El Niño was initiated by covarying warm ST and low SLP anomalies taking ∼2 years to propagate eastward across the subtropical South Pacific Ocean to South America near 30°S, then equatorward along the eastern boundary to the equator, and westward across the eastern equatorial Pacific Ocean. After 1977, El Niño was initiated by covarying warm ST and low SLP anomalies in the GEW/ACW taking ∼2 years to propagate eastward across the tropical eastern Indian and western and central Pacific oceans. Both of these different paths coexisted on occasion (e.g., during the 1998 El Niño), but one or the other usually dominated in their respective epochs. Though not specifically addressed in this study, the same multidecadal changes in the evolution of El Niño applies to La Niña as well [e.g., Allan et al., 1994].

[38] Second, we found the evolution of the ACW to be fundamentally different before and after 1977 as well. Before 1977, we found the path of the ACW across the Southern Ocean displaced to the south in the Indian sector and displaced to the north in the Pacific sector, compared with that observed after 1977. This may explain why the ACW could not been detected along 56°S in the Pacific sector of the Southern Ocean during the 1970's [e.g., Simmonds, 2003]. Before 1977, the path of the ACW was displaced or expanded into the central and eastern subtropical South Pacific Ocean, coming to within 25° latitude of the equator near 100°W, where it spawned a northern branch near the eastern boundary that propagated equatorward into the eastern equatorial Pacific Ocean. After 1977, the path of the ACW in the Pacific sector of the Southern Ocean receded to the middle and high latitudes, whilst that in the Indian sector was displaced or expanded northward into the subtropical South Indian Ocean near 30°S, spawning a north branch that transmitted the ACW signal into the Warm Pool north of Australia. This north branch had been observed previously to reinforce the GEW propagating eastward from the western tropical Indian Ocean [Peterson and White, 1998; White et al., 2002].

[39] Third, we found the GEW influencing El Niño differently before and after 1977 as well. Before 1977, the warm phase of the GEW propagated only as far east as the central equatorial Pacific Ocean, where it was diverted northward, propagating eastward into the Caribbean Sea between 10°N and 20°N, associated with the northward displacement or expansion of the ACW into the subtropical South Pacific Ocean, the latter initiating each El Niño during this epoch. After 1977, when the ACW receded to the middle and high latitude South Pacific Ocean, the warm phase of the GEW penetrated all the way into the eastern equatorial Pacific Ocean, initiating each El Niño during this epoch. These results qualify the observations of White and Cayan [2000], who found the GEW leading each of the significant El Niños over the 52-year record. Here we find the GEW unable to initiate El Niño prior to 1977, but it was able to strengthen El Niño in the central equatorial Pacific Ocean.

[40] Finally, we found evidence that the north branch of the ACW in the South Indian Ocean influenced El Niño independently of the GEW after 1977. As the ACW penetrated to within 25° of the equator in the eastern South Indian Ocean, its north branch propagated into the Warm Pool north of Australia but south of the equator, subsequently propagating eastward into the eastern equatorial Pacific Ocean on the south side of the equator, paralleling but apparently independent of the GEW directed eastward on the north side of the equator. Recently, White et al. [2002] stated that the GEW and the north branch of the ACW merged in the Warm Pool north of Australia. However, here, tracking the movement of SLP anomalies using space-time covariance analysis, we find the eastward propagation of the GEW across the tropical Indian Ocean pushed north of the equator by the north branch of the ACW. Subsequently, both the GEW and the north branch of the ACW converged onto the equator in the eastern equatorial Pacific Ocean, either initiating El Niño or intensifying its magnitude.

[41] Thus the meridional displacement or expansion of the ACW into the subtropical South Pacific and South Indian oceans before and after 1977 affected the path of the GEW propagating eastward across the tropical Indo-Pacific Ocean, displacing it into the northern hemisphere at different longitudes during different epochs. Before 1977, this northward displacement of the GEW and ACW in the central and eastern Pacific Ocean accompanied the initiation of El Niño by the ACW propagating equatorward along the eastern boundary of South America. After 1977, this displacement of the GEW and the ACW in the eastern Indian Ocean accompanied the initiation of El Niño by the north branch of the ACW propagating equatorward along the eastern boundary of Australia and then eastward across the tropical South Pacific Ocean.

[42] These multidecadal changes in the GEW, the ACW, and the evolutionary development of El Niño during the second half of the twentieth century appear to be caused by multidecadal changes in the SST over the Indo-Pacific and Southern Oceans that affect all three phenomena. This is particularly true of the ACW, since its coupling thermodynamics depends on the strong meridional temperature gradient in both the ocean and atmosphere in the vicinity of the subtropical front [White and Chen, 2002]. During the earlier epoch (that is, 1968 to 1977), we found the subtropical front in the South Indian Ocean weaker than normal while that in the western and central South Pacific Ocean was stronger than normal, consistent with the northward displacement or expansion of the ACW into the subtropical South Pacific Ocean. During the later epoch (that is, 1980 to 1989), the situation was nearly reversed from the earlier epoch, with the subtropical front stronger in the South Indian Ocean and weaker in the western and central South Pacific Ocean. Again, this is consistent with the northward displacement or expansion of the ACW into the subtropical South Indian Ocean and its recession into the middle and high latitude South Pacific Ocean.

[43] These results indicate that the phase and magnitude of El Niño are influenced by the eastward phase propagation of the ACW and GEW across the Indo-Pacific Ocean from 20°N to 50°S, the latter two phenomena phase locked to one another and interacting resonantly to maintain their amplitude [White et al., 2002]. Here we find this interaction between the ACW and the GEW accommodating significant multidecadal changes in path and intensity over the second half of the twentieth century. Moreover, this accommodation has a profound effect on the evolutionary development of El Niño. However, this interaction (which is global in extent) cannot explain the source of El Niño, which is an ocean-atmosphere coupled mode of the tropical Pacific Basin [e.g., Zebiak and Cane, 1987]. Even so, the ACW and GEW appear capable of influencing this basin mode, the latter perhaps co-oscillating with the global traveling waves. However, until El Niño is simulated in a numerical coupled model within the context of realistic simulations of the GEW and ACW, such a hypotheses remains untested.

Acknowledgments

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methods
  5. 3. Evolutionary Development of El Niño Before and After 1977
  6. 4. Evolutionary Development of the ACW Before and After 1977
  7. 5. Evolutionary Development of the GEW Before and After 1977
  8. 6. Directly Linking the Evolution of the ACW to the Evolution of El Niño
  9. 7. Possible Sources of Multidecadal Change in the ACW Over the Southern Hemisphere
  10. 8. Discussion and Conclusions
  11. Acknowledgments
  12. References

[44] Appreciation is extended to Arthur (Ted) Walker who was responsible for the streamline analyses conducted in this study. We extend our thanks to Andrea Fincham, who is responsible for drafting the figures. W. B. White and J. Annis are supported by the National Science Foundation (NSF OCE9910730). They are also supported by the National Aeronautics and Space Administration (NASA) under contract NASA-JPL 1205106 and Office of Global Programs of NOAA (NOAA NA17RJ1231) at the Experimental Climate Prediction Center. Warren White is supported by the Scripps Institution of Oceanography of the University of California, San Diego.

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methods
  5. 3. Evolutionary Development of El Niño Before and After 1977
  6. 4. Evolutionary Development of the ACW Before and After 1977
  7. 5. Evolutionary Development of the GEW Before and After 1977
  8. 6. Directly Linking the Evolution of the ACW to the Evolution of El Niño
  9. 7. Possible Sources of Multidecadal Change in the ACW Over the Southern Hemisphere
  10. 8. Discussion and Conclusions
  11. Acknowledgments
  12. References