1. Introduction
Modeling the annual cycle in the Tropics is the first step toward predicting as well as simulating El Niño–Southern Oscillation (ENSO). The state of the art of these modeling efforts, however, does not seem to be sufficient despite the rapid progress in our understanding of ENSO during the active Tropical Ocean and Global Atmosphere (TOGA) decade. There is at least as much work to be done in the next decade as was achieved in the last (Stockdale et al. 1998). In particular, further improvements are required to obtain a reasonable seasonal cycle in coupled general circulation models (CGCMs; Delecluse et al. 1998).
For this reason, using a high-resolution ocean general circulation model (OGCM), Tozuka et al. (2002) recently revisited the seasonal evolution of the Mindanao Dome (cf. Masumoto and Yamagata 1991) in the western tropical Pacific. They showed that the seasonal evolution of the Mindanao Dome is an outcome of interactions between the local and the basinwide seasonal cycle. As an extension of this study, we here examine the Pacific basinwide seasonal cycle to try to shed a new light on this topic.
Among studies devoted to the understanding of the basinwide seasonal cycle, Meyers (1979) was the first to demonstrate that variations at the depth of 14°C isotherm propagate westward along 6°N all the way from the American coast to 145°E as the long nondispersive baroclinic Rossby wave. The origin of the waves is identified as forcing by the annual variation of Ekman pumping in the eastern Pacific. This process was confirmed by Kessler (1990) using more XBT data, Mitchum and Lukas (1990) using sea level data at seven stations in the western tropical Pacific, Yu and McPhaden (1999) using the Tropical Atmosphere–Ocean array (TAO) buoy data, and Wang et al. (2000) using assimilated data. Also, Lukas and Firing (1985) captured the annual signal due to a vertically propagating, equatorially trapped long Rossby wave in the hydrographic data taken during the Hawaii–Tahiti Shuttle Experiment.
The above studies, however, do not discuss ocean–atmosphere coupled processes related to the basinwide seasonal variations. In this regard, Horel (1982) showed for the first time that the annual cycle in the sea surface temperature (SST) and surface wind convergence undergoes systematic longitudinal changes in phase and amplitude in the eastern Pacific. We believe, however, that several important problems on the seasonal ocean–atmosphere interaction have remained unanswered for the last two decades. The first is how the annual Rossby wave, which propagates across the entire Pacific basin, is related to the seasonal air–sea interaction in the eastern Pacific. The second is how the seasonal ocean–atmosphere coupled phenomena are linked to ENSO events.
The understanding of this annual cycle seems to be crucial to that of the interannual climate variations. In this context, Tozuka et al. (2002) have also shown that the interannual variation of the Mindanao Dome is governed by the interannual modulation of both local and basinwide seasonal cycles. The importance of the annual cycle in understanding ENSO events has long been recognized since the seminal work by Bjerknes (1969; see Neelin et al. 1998 for a review). Although the warm peak of some events does not occur in a preferred season (Neelin et al. 2000), it has been accepted that ENSO is generally phase locked to the annual cycle with the warm peak of El Niño occurring toward the end of the year. This is why Rasmusson and Carpenter (1982) analyzed, using a composite analysis, the seasonal evolution of the wind, SST, and rainfall anomaly fields for warm events derived from six major events between 1950 and 1972. Since then, several studies have revealed important roles played by the seasonal cycle in irregularity, chaotic nature, and phase locking of ENSO (Tziperman et al. 1994; Jin et al. 1994). Tziperman et al. (1997) recently revisited the topic, suggesting that the seasonal cycle should not be neglected even in simplified ENSO models. Thus, we need a comprehensive understanding of the seasonal cycle of the tropical Pacific to appreciate ENSO in full depth.
In this study, we investigate the seasonal cycle of the basinwide air–sea interaction using various data in an effort to understand the importance of the seasonal cycle to ENSO events. The content is organized as follows. After a brief description of the dataset used in this study, mean fields of atmospheric and oceanic variables are presented in the next section. In section 3, the air–sea interaction in the eastern Pacific, which plays a crucial role in triggering the whole seasonal cycle, is described. The generation and westward propagation of both warm and cold anomalies are also discussed. The interannual and decadal variations of the tropical Pacific are interpreted in sections 4 and 5 from a new viewpoint. The final section is for summary and discussions.
2. Data and mean fields
As compared with the previous decade, we now have new high-quality and high-resolution data for both oceanic and atmospheric fields, including satellite, reanalysis, and assimilation data that incorporates various kind of available observations. Here, we describe briefly those we use in the present analysis.
The SST data is the monthly dataset edited by Rayner et al. (2002, manuscript submitted to J. Geophys. Res.). This is based on marine surface observations and the satellite Advanced Very High Resolution Radiometer (AVHRR) data and covers the period from 1871 to 1999. The monthly rainfall data of 2.5° resolution from 1979 to 2000 is derived from the Climate Prediction Center (CPC) Merged Analysis of Precipitation (CMAP), which is composed of gauge data, several satellite estimates, and model output (Xie and Arkin 1996). We also use sea level pressure (SLP) data, outgoing longwave radiation (OLR) data, and wind stress data from the National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) reanalysis project (Kalnay et al. 1996). In addition, we analyze the Simple Ocean Data Assimilation (SODA) output for sea surface height (SSH; Carton et al. 2000). For sections 3b and 3c, we use the satellite data for wind stress (Kubota et al. 2002) and SSH. The monthly SSH field from the Ocean Topography Experiment (TOPEX)/Poseidon is created and maintained by the National Oceanic and Atmospheric Administration (NOAA)/Laboratory for Satellite Altimetry.
From these data, we first calculate monthly anomaly fields by subtracting annual mean fields from the corresponding monthly mean climatologies. As seen in Fig. 1, the annual mean SST field is characterized by a cold tongue in the eastern equatorial Pacific and a warm pool in the western Pacific. The SSH pattern shows ridges and troughs associated with North Equatorial Current and North Equatorial Countercurrent. The SLP is generally lower in the western Pacific than in the eastern Pacific, which is associated with the prevailing Walker Circulation (Bjerknes 1969) and consistent with the underlying SST field (Lindzen and Nigam 1987). The northeast (southeast) trade winds blow in the North (South) Pacific and the wind in the equatorial band is generally easterly. The OLR and precipitation fields are consistent; wherever OLR is low, convective activity is high and thus precipitation rate is high.
3. Annual ENSO
a. Air–sea interaction in the eastern Pacific
It is known from the time of the conquistadors that SST warms off the coast of Peru each year around the Christmas season (Fig. 2a), and this is the origin of the classical term El Niño (Wyrtki 1975). Subsequently, anomalous winds converge on the western flank of the warm sea surface temperature anomaly (SSTA), resulting in a westerly wind stress anomaly along the equator. It suppresses equatorial upwelling and latent heat loss and, hence, further enhances warm SSTA along the equator. By April, the warm SSTA expands farther westward along 3°S (Fig. 2c), and the anomalous wind converges over the western flank of the newly formed warm SSTA, leading to the further westward expansion of the warm SSTA (Fig. 2b). To show these basinwide interesting behaviors more clearly, we have prepared Fig. 3a, in which the westward propagation of the warm SSTA coupled with the westerly anomaly and negative outgoing longwave radiation anomaly (OLRA) is captured well in the time–longitude diagrams along the equator. Thus, Fig. 3a, per se, suggests that a positive feedback takes place in the eastern equatorial Pacific, as partly discussed in the pioneering work of Horel (1982).
As is well known, the warm SSTA influences the sea level pressure anomaly (SLPA) pattern in the Tropics; here the SLPA in the eastern equatorial Pacific is lowest in spring when the SSTA is positive there, as expected from the study of Lindzen and Nigam (1987). In order to see the east–west seesaw pattern in atmospheric pressure anomalies associated with the above air–sea interaction, we here first calculated the seasonal Southern Oscillation index (SOI), which is defined as the SLPA difference between Tahiti and Darwin. However, it has turned out that the seasonal SOI is not a good indicator of the annual air–sea interaction. This is because the variation in SLPA at Darwin is dominated by the heating of the Australian continent during austral summer and cooling during winter (Fig. 4a). Consequently, the seasonal SOI has a minimum value in July and a maximum value in January (Fig. 4b). In addition, as shown recently by Behera and Yamagata (2003), the SLP at Darwin is not only influenced by the variations in the Pacific, but is also influenced by the Indian Ocean climate variability. Therefore, it cannot be used for the present purpose of describing the seasonal strength of the Walker circulation in the Pacific.
An alternative index is defined as the SLPA difference between the value averaged over the eastern equatorial Pacific (2.5°S–2.5°N, 110°–100°W), where the positive SSTA takes its largest value in April, and that averaged over western equatorial Pacific (2.5°S, 150°–160°E), where the rising branch of the Walker circulation is normally located (Fig. 4c). The SLPA difference takes negative values during the boreal spring, indicating a weakened Walker circulation (a westerly wind anomaly over the equatorial Pacific). This corresponds to the annual El Niño–like condition and is consistent with the weakened zonal winds along the equator (Fig. 2).
The positive feedback terminates as a result of intrusion of a cold anomaly that first appears off the coast of South America, owing to the strengthened southerly winds that favor coastal upwelling (Mitchell and Wallace 1992; Nigam 1997). The air–sea interaction of the cold episode, which takes place in the latter half of the year, is almost a mirror image of the warm episode in the first half of the year.
b. Propagation of a cold anomaly
Here and in the following subsection, we discuss the SSH variations that reflect subsurface temperature fields associated with the seasonal air–sea interaction. The SSH depression anomaly first appears in the off-equatorial region of the eastern Pacific from February to March (Fig. 5). To investigate the mechanism for this depression associated with the cold anomaly quantitatively, we adopt the method developed by Gill and Clarke (1974), which is described in the appendix. Briefly, the method solves the linear equations of a particular baroclinic mode for the SSHA response forced by an equatorial zonal wind stress anomaly.
A depression anomaly lower than –2 cm is first generated by the westerly wind stress anomaly along the equator in the eastern Pacific in late February. Then, it propagates westward at the phase speed of about 0.6 m s–1 (Fig. 5). As the westerly anomaly shifts westward owing to the westward migration of the air–sea interaction, it further forces negative sea surface height anomaly (SSHA). The amplitude of the SSHA induced by the wind-forced Rossby wave (R1ewf) is larger than observed in the inviscid model because of the semiresonant condition. If we assume a damping timescale of 180 days in the linear model as suggested by Yu and McPhaden (1999), the depression anomaly generated by the westerly wind anomaly is quantitatively in agreement with the observation. The importance of the annual wind stress in forcing the first meridional mode Rossby wave is discussed by Kessler and McCreary (1993) using the linear, continuously stratified model to explain hydrographic observations. Here, we note the importance of the semiresonant condition between the westward-propagating component of winds and the phase speed of the first mode Rossby wave (cf. Yamagata 1987).
c. Propagation of a warm anomaly
The westerly wind anomaly also triggers the annual “warm” equatorial Kelvin wave. This downwelling Kelvin wave generates the warm anomaly along 5°N in the eastern Pacific after it reflects from the eastern boundary as Rossby waves during late boreal spring. However, the reflected Rossby waves (R1rfl) are weak in comparison with the wind-forced part (R1ewf); this result is in agreement with Minobe and Takeuchi (1995) and Yu and McPhaden (1999).
While propagating westward, the warm anomaly amplifies owing to the easterly anomaly associated with the opposite phase of the air–sea interaction described above (Fig. 2). However, its amplitude is smaller than that of the “cold” Rossby wave since the easterly wind stress anomaly during the cold phase of the seasonal air–sea interaction in the latter half of the year is weaker and less effective in triggering the Rossby wave, as indicated by Fig. A1 in the appendix. Despite that, the warm anomaly is further forced owing to the easterly component of the northeast trade winds as the trades strengthen from the central to western Pacific during the boreal winter (Fig. 2). We note that the seasonal intrusion of this warm anomaly eventually destroys the Mindanao Dome in the western tropical Pacific in the boreal spring of the following year (Masumoto and Yamagata 1991; Tozuka et al. 2002).
4. Interannual variations
The seasonal evolution of the anomalous wind and SST field is very similar to the seasonal evolution of the interannual El Niño shown by Rasmusson and Carpenter (1982). Although this aspect of El Niño was partly discussed during the pre-TOGA period, it is necessary to shed a new light based on the new viewpoint presented in the previous section. Here, the interannual modulation of the annual cycle and the relationship between the annual cycle and ENSO during 1980–99 are investigated. After subtracting the annual mean field as in Rasmusson and Carpenter (1982), we apply complex EOF (CEOF) and composite methods to various fields of interest.
In order to gain an insight into the role played by each CEOF mode in the interannual variations, we first construct time series of Niño-3 SSTA (5°S–5°N, 90°–150°W) as the Niño-3 index is frequently used to define ENSO (Fig. 6). The second CEOF mode, whose temporal phase reflects the interannual variation associated with ENSO (Fig. 6b) and may be called an “interannual ENSO” mode, explains 14% of the total variance. Although the variance contribution of the second mode is much smaller than the first mode, which explains 74% of total variance, this does not necessarily mean that the second mode is always secondary; it may appear as the major mode in some years (cf. Yamagata et al. 2002). In the present case, it is found that the second mode plays an important role in ENSO events. Actually, the SSTA associated with this mode is positive with peaks ranging from 1.5° to 2.5°C during typical El Niño years, while it is negative with peaks of about –1.5°C during La Niña years.
The first CEOF mode, or an “annual ENSO” mode, has a period of 1 yr and is in phase with the mean seasonal cycle revealed in the last section. However, the amplitude of the seasonal cycle undergoes an interannual modulation (Fig. 6b). Moreover, the correlation coefficient between the first CEOF mode minus the mean seasonal cycle and the second mode is 0.42, which is significant at 95% level and suggests some nonorthogonality between the interannual modulation of the seasonal cycle and the interannual ENSO mode time series. During typical El Niño years (1982–83, and 1997–98), positive peaks in boreal spring are warmer by 0.5° to 1.0°C, while the negative peak in boreal autumn is warmer. On the other hand, minimum SSTA achieved by the first mode is colder by 0.5°–1.0°C during La Niña years (1984–85, 1988–89), but interestingly also during the prolonged El Niño of the early 1990s. This suggests that the modulation of the annual ENSO provides a key to understanding the peculiar behavior of the prolonged El Niño.
Another important characteristic of ENSO is the zonal swing in the atmospheric pressure in the Pacific. A pattern for the SO is obtained by correlating the time series of Fig. 6b with that of the global SLPA (Fig. 7). We used the partial correlation method instead of the simple correlation because these two modes are not orthogonal in time as indicated by the fact that the correlation coefficient between the first mode minus the mean seasonal cycle and the second mode is 0.42 (cf. Yamagata et al. 2003). Both show well-defined zonal seesaw patterns in the tropical Pacific; the first mode minus the mean seasonal cycle shows a correlation confined in the tropical Pacific, whereas the second mode shows a significant correlation, even in the Indian Ocean.
To investigate the common features of the warm events, composite diagrams are constructed by simply averaging three warm events (1982–83, 1986–87, and 1997–98) excluding the prolonged El Niño in the early 1990s. The interannual ENSO mode is captured by the second CEOF mode for SSTA and OLRA, and, interestingly by the first CEOF mode for SSHA (Fig. 8). In an early stage of El Niño, a positive SSHA originates from the western equatorial Pacific owing to the westerly wind stress anomaly [positive zonal wind stress anomaly (ZWSA)] as described by Wyrtki (1975) and propagates eastward (Philander et al. 1984; Yamagata 1985). This leads to a deepening in the thermocline depth and the SST warming in the central and eastern Pacific, which peaks toward the end of the year. Since large El Niño events (1982–83 and 1997–98) peak toward the end of the year, monthly climatology of the second CEOF mode of SSTA shows positive anomaly for this time of the year (figures not shown). Associated with this warmer SST is the enhanced convective activity (negative OLRA) and the weaker Walker circulation in the central and eastern Pacific as suggested from Fig. 7.
At the same time, a negative SSHA (shallower thermocline depth) is achieved in the off-equatorial western Pacific, resulting in a SST cooling and a relatively higher off-equatorial SLPA (Wang and Zhang 2002). Then, the equatorial easterly wind stress anomaly in the far western Pacific induced by the pair of atmospheric anticyclones eventually releases cold Kelvin waves from the western Pacific. As the cold Kelvin waves propagate eastward, they cause the thermocline to shoal and terminate the SST warming in the eastern Pacific. Composite diagrams for La Niña years show almost a mirror image of the above description (figures not shown). Since the above air–sea interaction needs almost two years to complete the cycle, it may be closely related to the tropical biennial oscillation (TBO) discussed by Meehl et al. (2003) and Rao et al. (2002).
The annual ENSO mode, on the other hand, is captured by the first CEOF mode for SSTA and OLRA, and again interestingly, the second CEOF mode for SSHA (Fig. 9). The evolution of the annual ENSO mode seems to take a normal course during the first half of year 0, but the annual reestablishment of the cold tongue is weaker during the latter half of year 0; the minimum SSTA of the cold tongue in October (+0) is about 1°C warmer than the climatology. Because of the warmer SST, the convective activity is not much suppressed (lower OLRA). Furthermore, the annual warming in the eastern tropical Pacific is significantly prominent in the next spring, with the maximum SSTA in April (+1) higher than 1.5°C. This results in more convective activity and lower OLRA over the eastern Pacific.
The weak seasonal reestablishment of the cold tongue during El Niño years can be explained by changes in the background fields caused by the interannual ENSO mode. As shown in Fig. 8, the interannual ENSO mode leads to the deeper than normal thermocline depth, resulting in less effective upwelling of the subsurface water. Similarly, the seasonal warming in the eastern Pacific during boreal spring is stronger owing to deeper thermocline. Thus, El Niño may be interpreted as the interaction between the annual ENSO mode and the interannual ENSO mode.
One might suspect that the above interannual modulation of the seasonal cycle is a statistical artifact. The EOF analysis may lead to degenerated modes in a situation where two different modes explain near-identical variances (North et al. 1982; Behera et al. 2003). However, all CEOF modes discussed in this paper have statistically significant separations from other modes. We also note that the possibility of interannual modulation of the seasonal cycle was proposed by Gu and Philander (1995) and Kim and Chung (2001) using different methods, although no specific mechanisms were given.
5. Decadal variations
One interesting difference between El Niño before and after the late 1970s is the propagation characteristic (Wallace et al. 1998; Wang and An 2001), as revealed in Fig. 3. The SST deviation from the mean seasonal cycle propagates westward in the 1972–73 El Niño, while it propagates eastward in the 1982–83 El Niño. However, we now shed a new light on this property from our viewpoint. Since the Pacific decadal climate shift occurred in the late 1970s and changed the background field of the equatorial Pacific (e.g., Luo and Yamagata 2001), it is reasonable to construct anomaly fields (deviation from the annual mean) separately for a preshift (1958–75) and a postshift (1980–99) period. A remarkable difference in the relative variance contribution of the annual ENSO and the interannual ENSO is found for the two periods. During the preshift period, the ratio of the annual ENSO mode to the interannual ENSO mode is 6.9, while it decreases to 5.1 for the postshift period. This increase in the amplitude of the interannual ENSO mode in the postshift period is also observed in the SODA SSTA and other oceanic and atmospheric fields such as SSH and OLR. To confirm this further, we applied wavelet analysis (Torrence and Compo 1998) to Niño 3 SSTA (Fig. 10). For both periods, two spectral peaks exist in the global wavelet spectrum as expected. However, the peak associated with the interannual ENSO mode only in the preshift period is not significant at 95% confidence level. This seems reasonable considering the fact that the two largest warm events of the last century occurred in the postshift period. The increase in the amplitude of the interannual ENSO mode in the postshift period favors the appearance of eastward propagation.
Then, what happened during the period of prolonged El Niño in the early 1990s (Trenberth and Hoar 1996; Goddard and Graham 1997; Rajagopalan et al. 1997) from the above perspective? Throughout this period, the interannual ENSO mode continues to be slightly positive (Fig. 6b). The annual ENSO mode shows positive peaks in the earlier part of the year, but those are only above the mean seasonal cycle by less than 0.5°C. On the other hand, the negative peaks in the latter half of the year are comparable to some of La Niña years (Fig. 6a). Thus, despite the limitation of a linear viewpoint, the present decomposition into two modes provides us a meaningful key to understand the long-term variation of the ENSO phenomenon.
6. Summary and discussions
Using various observational data, we have investigated two distinct modes of air–sea coupled phenomena in the tropical Pacific to present an alternative view of ENSO. The first mode we have identified is the basinwide seasonal cycle, which may be called annual ENSO and is summarized in Fig. 11. The positive SSTA off the Peru coast triggers a series of the air–sea interaction processes in this mode. At the west of the positive SSTA, westerly anomaly strengthens and causes convergence of warm anomaly at the equator. This causes all of the positive SSTA, positive ZWSA, positive precipitation anomaly, and negative OLRA to propagate westward and initiate the air–sea interaction in the eastern Pacific. This important positive feedback, which initiates the basin-scale seasonal variation, is originally described by Horel (1982). We have found here that the same westerly wind stress anomaly generates the cold anomaly in the off-equatorial region and this propagates westward as the cold Rossby wave associated with the surface depression. It is found that the semiresonant condition between the westward propagating component of westerly wind and the Rossby wave plays an important role in amplifying the cold anomaly. The positive feedback is terminated by the southerly wind anomaly in the late boreal spring (Mitchell and Wallace 1992) and the cold phase begins. This cold phase, just like a mirror image of the warm phase, evolves in the latter half of the year as a result of air–sea interaction. The oceanic part of the present result is further supported by the OGCM experiment of Kagimoto (1999), although the model amplitude of the seasonal cycle of SST in the eastern tropical Pacific is smaller than observations by about 20%.
The interannual ENSO mode, which is the air–sea coupled phenomenon initiated by the westerly wind stress anomaly (e.g., Yamagata and Masumoto 1989), leads to the deepening in the thermocline depth and the SST warming in the central and eastern Pacific. This, in turn, leads to the weaker cold season and warmer warm season owing to less effective upwelling of the cold subsurface water. Thus, the positive SSTA of both the annual ENSO mode and the interannual ENSO modes contributes to the SST warming during El Niño. The seminal work of Rasmusson and Carpenter (1982) has been criticized occasionally since their ENSO composite diagrams include the seasonal signal. Our study, however, clearly shows that they captured two important modes contributing to ENSO events. Also, as Meyers (1982) suggested using a single time series of sea level at Truk Island, we need to be careful in calculating interannual anomalies; just subtracting a mean climatological seasonal cycle is misleading since both the amplitude and phase of the seasonal cycle may change interannually.
The decadal modulation of ENSO may be interpreted by superposing the two leading modes discussed here. ENSO is often considered as a nonlinear oscillator forced by the seasonal cycle (Tziperman et al. 1994; Jin et al. 1994), and the seasonal cycle is suggested to play an important role in generating decadal variations of the ENSO. Our perspective developed here, despite the limitation of the linear analysis, supports those ideas and strengthens the importance of the air–sea interaction in the seasonal cycle itself. Wang and An (2001) have recently discussed the interdecadal change in ENSO propagation characteristics; this may be explained by changes in the relative phase and amplitude of these two modes. The larger amplitude of the eastward propagating interannual ENSO mode in recent decades favors the eastward propagation. In the early 1990s, the interannual ENSO mode is in the positive phase, while the annual ENSO mode is in the negative phase. This may explain the prolonged El Niño without any La Niña.
To resolve the annual ENSO may also provide a benchmark test for the performance of CGCMs. As summarized by Mechoso et al. (1995) and Delecluse et al. (1998), and more recently by Latif et al. (2001), no CGCMs to date have succeeded in reproducing the seasonal cycle of the eastern equatorial Pacific. If ENSO can be an outcome of interaction between the interannual modulation of the annual ENSO mode and the interannual ENSO mode as discussed here, this must be a serious defect of the current status of ENSO modeling (cf. Meehl et al. 2001). Further studies using CGCMs to confirm that the annual ENSO is an air–sea coupled phenomenon and that ENSO is the result of interaction between two distinct air–sea coupled modes are under way. As stated by Rasmusson et al. (1999), “the description and understanding of the annual cycle are, therefore fundamental to the description and understanding of non-seasonal variability, and no less a scientific challenge.”
Acknowledgments
We are very grateful to J. P. McCreary and R. Lukas for valuable comments on the original manuscript. This study benefited from discussions with S. Behera, Z. Liu, Y. Masumoto, G. Meyers, and S. Minobe. The wavelet software was provided by C. Torrence and G. Compo and is available at http://paos.colorado.edu/research/wavelets/. We are also indebted to T. Kagimoto for providing the OGCM output and J.-J. Luo for providing the CEOF code. Useful comments provided by E. Firing and two anonymous reviewers helped us to improve the earlier manuscripts. This work is supported by the grant from the Mitsubishi Foundation.
REFERENCES
Behera, S. K., and T. Yamagata, 2003: Influence of the Indian Ocean dipole on the Southern Oscillation. J. Meteor. Soc. Japan, 81 , 169–177.
Behera, S. K., S. A. Rao, H. N. Saji, and T. Yamagata, 2003: Comments on “A cautionary note on the interpretation of EOFs.”. J. Climate, 16 , 1087–1093.
Bjerknes, J., 1969: Atmospheric teleconnections from the equatorial Pacific. Mon. Wea. Rev., 97 , 820–829.
Carton, J. A., G. Chepurin, X. Cao, and B. Giese, 2000: A Simple Ocean Data Assimilation analysis of the global upper ocean 1950–95. Part I: Methodology. J. Phys. Oceanogr., 30 , 294–309.
Delecluse, P., M. K. Davey, Y. Kitamura, S. G. H. Philander, M. Suarez, and L. Bengtsson, 1998: Coupled general circulation modeling of the tropical Pacific. J. Geophys. Res., 103 , 14357–14373.
Gill, A. E., and A. J. Clarke, 1974: Wind-induced upwelling, coastal currents and sea-level changes. Deep-Sea Res., 21 , 325–345.
Goddard, L., and N. E. Graham, 1997: El Niño in the 1990s. J. Geophys. Res., 102 , 10423–10436.
Gu, D., and S. G. H. Philander, 1995: Secular changes of annual and interannual variability in the Tropics during the past century. J. Climate, 8 , 864–876.
Horel, J. D., 1982: On the annual cycle of the tropical Pacific atmosphere and ocean. Mon. Wea. Rev., 110 , 1863–1878.
Jin, F-F., J. D. Neelin, and M. Ghil, 1994: El Niño on the devil's staircase. Science, 264 , 70–72.
Kagimoto, T., 1999: Numerical study on transport variations of the Kuroshio. Ph.D. dissertation, University of Tokyo, 75 pp.
Kalnay, E., and Coauthors. 1996: The NCEP/NCAR 40-year Reanalysis Project. Bull. Amer. Meteor. Soc., 77 , 437–471.
Kessler, W. S., 1990: Observations of long Rossby waves in the northern tropical Pacific. J. Geophys. Res., 95 , 5183–5217.
Kessler, W. S., and J. P. McCreary, 1993: The annual wind-driven Rossby wave in the subthermocline equatorial Pacific. J. Phys. Oceanogr., 23 , 1192–1207.
Kim, K-Y., and C. Chung, 2001: On the evolution of the annual cycle in the tropical Pacific. J. Climate, 14 , 991–994.
Kubota, M., N. Iwasaka, S. Kizu, M. Konda, and K. Kutsuwada, 2002: Japanese ocean flux data sets with use of remote sensing observations (J-OFURO). J. Oceanogr., 58 , 213–225.
Latif, M., and Coauthors. 2001: ENSIP: The El Niño simulation intercomparison project. Climate Dyn., 18 , 255–276.
Lindzen, R. S., and S. Nigam, 1987: On the role of sea surface temperature gradients in forcing low-level winds and convergence in the tropics. J. Atmos. Sci., 44 , 2418–2436.
Lukas, R., and E. Firing, 1985: The annual Rossby wave in the central equatorial Pacific Ocean. J. Phys. Oceanogr., 15 , 55–67.
Luo, J-J., and T. Yamagata, 2001: Long-term El Niño-Southern Oscillation (ENSO)-like variation with special emphasis on the South Pacific. J. Geophys. Res., 106 , 22211–22227.
Masumoto, Y., and T. Yamagata, 1991: Response of the western tropical pacific to the Asian winter monsoon: The generation of the Mindanao Dome. J. Phys. Oceanogr., 21 , 1386–1398.
Mechoso, C. R., and Coauthors. 1995: The seasonal cycle over the tropical Pacific in coupled ocean–atmosphere general circulation models. Mon. Wea. Rev., 123 , 2825–2838.
Meehl, G. A., R. Lukas, G. N. Kiladis, K. M. Weickmann, A. J. Matthews, and M. Wheeler, 2001: A conceptual framework for time and space scale interactions in the climate system. Climate Dyn., 17 , 753–775.
Meehl, G. A., J. M. Arblaster, and J. Loschnigg, 2003: Coupled ocean–atmosphere dynamical processes in the tropical Indian and Pacific Oceans and the TBO. J. Climate,, 16 , 2138–2158.
Meyers, G., 1979: On the annual Rossby wave in the tropical North Pacific Ocean. J. Phys. Oceanogr., 9 , 663–674.
Meyers, G., 1982: Annual and interannual variation in sea level near Turk Island. J. Phys. Oceanogr., 12 , 1161–1168.
Minobe, S., and K. Takeuchi, 1995: Annual period equatorial waves in the Pacific Ocean. J. Geophys. Res., 100 , 18379–18392.
Mitchell, T. P., and J. M. Wallace, 1992: The annual cycle in equatorial convection and sea surface temperature. J. Climate, 5 , 1140–1156.
Mitchum, G. T., and R. Lukas, 1990: Westward propagation of annual sea level and wind signals in the western Pacific Ocean. J. Climate, 3 , 1102–1110.
Neelin, J. D., D. S. Battisti, A. C. Hirst, F. Jin, Y. Wakata, T. Yamagata, and S. Zebiak, 1998: ENSO theory. J. Geophys. Res., 103 , 14261–14290.
Neelin, J. D., F-F. Jin, and H-H. Syu, 2000: Variations in ENSO phase locking. J. Climate, 13 , 2570–2590.
Nigam, S., 1997: The annual warm to cold phase transition in the eastern equatorial Pacific: Diagnostics of the role of stratus cloud-top cooling. J. Climate, 10 , 2447–2467.
North, G. R., T. L. Bell, and R. F. Cahalan, 1982: Sampling errors in the estimation of empirical orthogonal functions. Mon. Wea. Rev., 110 , 699–706.
Philander, S. G. H., T. Yamagata, and R. C. Pacanowski, 1984: Unstable air–sea interaction in the tropics. J. Atmos. Sci., 41 , 604–613.
Rajagopalan, B., U. Lall, and M. A. Cane, 1997: Anomalous ENSO occurrences: An alternative view. J. Climate, 10 , 2351–2357.
Rao, S. A., S. K. Behera, Y. Masumoto, and T. Yamagata, 2002: Interannual subsurface variability in the tropical Indian Ocean with a special emphasis on the Indian Ocean dipole. Deep-Sea Res., 49B , 1549–1572.
Rasmusson, E. M., and T. H. Carpenter, 1982: Variations in tropical sea surface temperature and surface wind fields associated with the Southern Oscillation/El Niño. Mon. Wea. Rev., 110 , 354–384.
Rasmusson, E. M., M. Chelliah, and C. F. Ropelewski, 1999: The observed climate of the 20th century. Modeling the Earth's Climate and Its Variability, W. R. Holland, S. Joussaume, and F. David, Eds., Elsevier Science, 1–138.
Stockdale, T. N., A. J. Busalacchi, D. E. Harrison, and R. Seager, 1998: Ocean modeling for ENSO. J. Geophys. Res., 103 , 14325–14355.
Torrence, C., and G. P. Compo, 1998: A practical guide to wavelet analysis. Bull. Amer. Meteor. Soc., 79 , 61–78.
Tozuka, T., T. Kagimoto, Y. Masumoto, and T. Yamagata, 2002: Simulated multiscale variations in the western tropical Pacific: The Mindanao Dome revisited. J. Phys. Oceanogr., 32 , 1338–1359.
Trenberth, K. E., and T. J. Hoar, 1996: The 1990–1995 El Niño–Southern Oscillation event: Longest on record. Geophys. Res. Lett., 23 , 57–60.
Tziperman, E., L. Stone, M. A. Cane, and H. Jarosh, 1994: El Niño chaos: Overlapping of resonances between the seasonal cycle and the Pacific ocean–atmosphere oscillator. Science, 264 , 72–74.
Tziperman, E., S. E. Zebiak, and M. A. Cane, 1997: Mechanisms of seasonal–ENSO interaction. J. Atmos. Sci., 54 , 61–71.
Wallace, J. M., E. M. Rasmusson, T. P. Mitchell, V. E. Kousky, E. S. Sarachik, and H. von Storch, 1998: On the structure and evolution of ENSO-related climate variability in the tropical Pacific: Lessons from TOGA. J. Geophys. Res., 103 , 14241–14259.
Wang, B., and S-I. An, 2001: Why the properties of El Niño changed during the late 1970s. Geophys. Res. Lett., 28 , 3709–3712.
Wang, B., and Q. Zhang, 2002: Pacific–East Asian teleconnection. Part II: How the Phillippine Sea anomalous anticyclone is established during El Niño development. J. Climate, 15 , 3252–3265.
Wang, B., R. Wu, and R. Lukas, 2000: Annual adjustment of the thermocline in the tropical Pacific Ocean. J. Climate, 13 , 596–616.
Wyrtki, K., 1975: El Niño—The dynamic response of the equatorial Pacific Ocean to atmospheric forcing. J. Phys. Oceanogr., 5 , 572–584.
Xie, P. P., and P. A. Arkin, 1996: Analyses of global monthly precipitation using gauge observations, satellite estimates, and numerical model predictions. J. Climate, 9 , 840–858.
Yamagata, T., 1985: Stability of a simple air–sea coupled model in the tropics. Coupled Ocean–Atmosphere Models, J. C. J. Nihoul, Ed., Elsevier Oceanography Series, Vol. 40, Elsevier Science, 637–657.
Yamagata, T., 1987: A simple moist model relevant to the origin of intraseasonal disturbances in the tropics. J. Meteor. Soc. Japan, 65 , 153–165.
Yamagata, T., and Y. Masumoto, 1989: A simple ocean–atmosphere coupled model of the origin of a warm ENSO event. Philos. Trans. Roy. Soc. London, 329A , 225–236.
Yamagata, T., S. K. Behera, S. A. Rao, Z. Guan, K. Ashok, and H. N. Saji, 2002: The Indian Ocean dipole: A physical entity. CLIVAR Exch., 24 , 1–6.
Yamagata, T., S. K. Behera, S. A. Rao, Z. Guan, K. Ashok, and H. N. Saji, 2003: Comments on “Dipoles, temperature gradient, and tropical climate anomalies.”. Bull. Amer. Meteor. Soc., in press.
Yu, X., and M. J. McPhaden, 1999: Seasonal variability in the equatorial Pacific. J. Phys. Oceanogr., 29 , 925–947.
APPENDIX
Equatorial Kelvin and Rossby Waves
The method we used to quantitatively discuss the effects of the Kelvin and Rossby waves are based on Gill and Clarke (1974). We summarize the method here.
The annual mean field of (a) SST (°C), (b) SSH (cm), (c) SLP (hPa), (d) wind stress (10–2 N m–2), (e) OLR (W m–2), and (f) precipitation rate (mm day–1)
Citation: Journal of Physical Oceanography 33, 8; 10.1175/2414.1
Annual march of wind stress anomaly fields and SSTA for (a) Jan, (b) Apr, (c) Jul, and (d) Oct. The magnitude of wind stress anomaly is shown by the vector drawn below the figure (N m–2). Contour interval is 0.5°C
Citation: Journal of Physical Oceanography 33, 8; 10.1175/2414.1
The time–longitude diagrams of (a)–(c) SSTA, (d)–(f) ZWSA, and (g)–(i) OLRA along the equator for the annual cycle, 1972–73 El Niño, and 1982–83 El Niño, respectively. Contour intervals are 0.3°C, 4 × 10–3 N m–2, 4 W m–2, 1 mm day–1, for (left) annual cycle, and doubled for (middle) 1982–83 and (right) 1972–73 El Niño figures. Negative anomalies are shaded
Citation: Journal of Physical Oceanography 33, 8; 10.1175/2414.1
(a) Annual march of the SLPA. Contour interval is 1 hPa. (b) The SLPA at Tahiti and Darwin. The seasonal SOI is also shown; units: hPa. (c) The SLPA at the eastern equatorial Pacific (EP; 2.5°S–2.5°N, 110°–100°W) and the SLPA at the western equatorial Pacific (WP; 2.5°–2.5°S, 150°–160°E); the difference of SLPA between EP and WP is also shown (SLPD); units: hPa
Citation: Journal of Physical Oceanography 33, 8; 10.1175/2414.1
Time–longitude diagrams of (a) SSHA, and SSHA owing to wind-forced first mode Rossby wave, (b) without damping, and (c) with damping timescale of 180 days along 5°N. Contour interval is 2 cm. Negative anomalies are shaded
Citation: Journal of Physical Oceanography 33, 8; 10.1175/2414.1
Time series of the Niño-3 SSTA from 1980 to 1999: (a) mean seasonal cycle (solid line) and SSTA reconstructed from the first CEOF mode (dotted line); (b) SSTA reconstructed from the first CEOF mode (dotted line), SSTA reconstructed from the second CEOF mode (thick solid line), and the difference between the first CEOF mode and the mean seasonal cycle (thin solid line)
Citation: Journal of Physical Oceanography 33, 8; 10.1175/2414.1
Partial correlation between the global SLPA (deviation from mean seasonal cycle) and (a) the first mode minus mean seasonal cycle (msc), or (b) the second mode
Citation: Journal of Physical Oceanography 33, 8; 10.1175/2414.1
El Niño composite of reconstructed fields for (a) the second CEOF mode of SSTA, (b) the second mode of OLRA, and (c) the first mode of SSHA. Variance contributions are 14%, 14%, and 44%, respectively. Contour intervals are 0.5°C, 5 W m–2, and 3 cm. Negative anomalies are shaded
Citation: Journal of Physical Oceanography 33, 8; 10.1175/2414.1
As in Fig. 8 but for (a) first CEOF mode of SSTA, (b) first mode of OLRA, and (c) second mode of SSHA. Variance contributions are 74%, 46%, and 15%, respectively
Citation: Journal of Physical Oceanography 33, 8; 10.1175/2414.1
Normalized global wavelet spectrum of Niño-3 SSTA for (a) 1958–75 and (b) 1980–99 (solid line). Also shown is the 95% confidence spectrum (dashed line)
Citation: Journal of Physical Oceanography 33, 8; 10.1175/2414.1
Schematic diagrams of the annual ENSO during both warm and cold phase
Citation: Journal of Physical Oceanography 33, 8; 10.1175/2414.1
Fig. A1. Wind forcing function X0 and P2. Negative anomalies are shaded
Citation: Journal of Physical Oceanography 33, 8; 10.1175/2414.1
