1. Introduction
Most of the previous studies of the El Niño–Southern Oscillation (ENSO) phenomenon have focused on the El Niño episode in which the horizontal sea surface temperature (SST) gradient along the equatorial Pacific reverses from relatively warm in the western part and cool in the eastern part to the opposite. This is usually labeled as a transition from a cold to a warm state of the phenomenon.
Most of the physical mechanisms proposed to explain such a transition generally involve the low-level equatorial wind field. For example, westerly anomalies over the western equatorial Pacific (WEP) have been suggested as the trigger (Wyrtki 1975; Philander and Pacanowski 1981; Philander 1985; Busalacchi and O’Brien 1981; Rasmusson and Carpenter 1982). Many studies have also linked the Asian–Australian monsoon with these anomalies. Possible mechanisms include cold surges from the east Asian winter monsoon (Lau et al. 1983), enhancement of the Australian summer monsoon (Hackert and Hastenrath 1986), and persistent development of the intraseasonal oscillations (Lau and Chan 1985). However, Barnett (1983, 1984, 1985) suggested that the surface wind and sea level pressure (SLP) anomalies originate in the equatorial Indian Ocean (EIO) and propagate slowly eastward into the Pacific. The eastward propagation of tropospheric zonal wind anomalies from the south Asian monsoon region to the western Pacific observed by Yasunari (1985, 1990) and Gutzler and Harrison (1987) seem to support Barnett’s finding. Meehl (1987) emphasized the impact of the biennial variation of the monsoon circulation. Recently, Xu and Chan (2000, hereafter XC) suggested that the WEP westerly anomalies are jointly affected by the Asian–Australian monsoon. They emphasized that it is only when anomalous northerlies from the east Asian winter monsoon converge with anomalous southerlies from Australia that sufficiently strong westerly anomalies can form over the WEP to initiate the warm state of the ENSO.
Since the 1988 La Niña, which represents a transition from the warm to the cold state, there have been some studies of the mechanism of such a transition. The most comprehensive of these is the so-called delayed-oscillator theory (Suarez and Schopf 1988; Graham and White 1988; Battisti 1988; Cane et al. 1990; Munnich et al. 1991). This theory offers a negative feedback mechanism to describe the evolution of ENSO events from the warm to the cold state and back again. According to this mechanism, the ENSO cycle exists due to a combination of equatorial ocean wave dynamics and amplification by a coupled ocean–atmosphere instability (Philander 1983; Hirst 1986). Such a theory supplements the feedback hypothesis of Bjerknes (1969).
However, the basic implications from this theory that the two types of transitions should alternate and occur at regular intervals are not observed. In other words, some other mechanisms must be at work to alter the periodicities in the cycle. Wang and Feng (1996) considered this apparent irregularity to be a result of a chaotic atmosphere–ocean interaction. While this result suggests conditions under which a chaotic behavior may develop, the reasons for the occurrence of these conditions are not addressed. Recently, Li (1997) proposed that the existence of a stationary SST mode is responsible for the phase transition between the warm and cold phases of the ENSO. However, not much observational evidence was provided to support his hypothesis.
In searching for the possible physical processes responsible for the ENSO cycle, it seems to be quite straightforward to examine both the warm-to-cold and the cold-to-warm phases together. On the other hand, because the transitions between the two phases do not necessarily occur alternately in nature, different mechanisms may be responsible for each of the transitions. Thus, it might be more appropriate to study each transition separately. A recent paper by XC has addressed the mechanisms for the cold-to-warm transition. Based on the analyses of meteorological and oceanic observations associated with 12 previous El Niño events, they found that the Asian and Australian monsoons are critical in determining the season during which such an event will occur. This current research represents a follow-up to their study in an attempt to identify possible mechanisms for the warm-to-cold transition based on observations associated with previous La Niña events.
After the datasets have been described in section 2, the basic features of the transition from the warm to cold state of the ENSO cycle are documented in section 3. Similar to XC it is found that the transition can be divided into two types based on the onset time of the La Niña event. In addition, cases in which no transition took place are also identified. Section 4 gives the time evolution of the basic atmospheric and oceanic parameters in each type. Based on these results, a hypothesis is advanced in section 5 to describe the sequence of events that may lead to the onset of a La Niña event. All the results are then summarized and discussed in section 6.
2. Data
The present study utilizes data from four sources. Monthly mean global SST data on a 2° lat × 2° long grid for the period of 1950–97 are based on the analyses of Reynolds and Smith’s (1994) Comprehensive Ocean Atmosphere Data set. Monthly mean global zonal and meridional wind components for 1958–97 and monthly outgoing longwave radiation (OLR) for June 1974–February 1978 and January 1979–January 1998 with a resolution of 2.5° lat × 2.5° long square are from the National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) 40-year Reanalysis datasets. Values of the monthly SST anomalies (SSTA) over the Niño-3.4 region (5°S–5°N, 170°–120°W) for the period of 1950–97 are extracted from the Web site of the Climate Prediction Center (CPC). Ocean temperature anomalies at standard levels (0–400 m) and mixed layer depth (MLD) data for each month from January 1955 through December 1998 are obtained from the Scripps Institute of Oceanography.
3. Basic features associated with the warm-to-cold transition
Similar to the study of XC, the Niño-3.4 SSTA is chosen as the reference in defining the time of the transition. The time when the SSTA first drops below +0.5°C (−0.5°C) after the peak of an El Niño episode is defined as the termination (beginning) of the warm (cold) state of the ENSO cycle. This threshold of 0.5°C is also the level used by the CPC in their definition of the occurrence of ENSO events.
Twelve ENSO episodes (1951, 1953, 1957, 1963, 1965, 1969, 1972, 1976, 1982, 1987, 1992, and 1994) occurred during 1950–97. The 1997 warm event is not included because the La Niña conditions that developed in 1998 had not ended. Two major features are apparent from the time evolution of the Niño-3.4 SSTA during each of the ENSO episodes (Fig. 1). Most of the warm events tend to terminate in the northern winter or early spring (January–April), with a subsequent occurrence of a La Niña (cold) event only in seven cases. In these cases, the cold event generally begins in spring or later summer (see the exact months in Table 1). Based on the starting time of the La Niña episode (see Table 1), three types of transition can be defined:
Spring occurrence (SP) type—1963–65, 1972–74, 1987–89;
Summer occurrence (SU) type—1953–55, 1969–71, 1982–84, 1994–96; and
Nonoccurence (NON) type—1951–53, 1957–59, 1965–67, 1976–78, 1991–93.
The composite SSTA (over Niño-3.4) time series of the three transition types (Fig. 2) show that the SP type has larger SSTA amplitude with the mean negative SSTA dropping below −1.5°C during the La Niña event. The transition from the El Niño to the La Niña is only two months (from the SSTA >0.5°C to SSTA <−0.5°C;see also Table 1). The SU type, in contrast, seems to be weaker, with only a −1.0°C maximum anomaly, and has a slower transition speed (∼4 months from the warm to the cold state). In addition, the negative anomalies extend into year 1,1 with another peak occurring during the boreal winter. Both types of La Niña events show a common feature: the negative maximum anomaly occurs between November 0 and January 1. This phenomenon, which is also found in an El Niño event, suggests a strong phase lock with the seasonal cycle. For the NON type, the SSTA persists between −0.5°C and +0.5°C after the termination of the El Niño. Note also from Fig. 2 that the maximum SSTA during year −1 is about the same for all three types.
These results suggest that it is important to classify the warm-to-cold transition into different categories. Not only can the transition occur at different times but the warm state can maintain itself and never transition into the cold state. From Figs. 1 and 2, the times at which the SSTA reaches the maximum during the warm episode for all the three types do not differ significantly (see also Table 1). And yet vastly different transitions can occur. This apparent predictability problem needs to be addressed. Further, if the transition takes place in different seasons, different seasonal atmospheric and oceanographic conditions must be responsible for triggering the transition.
In the following sections, the oceanic and atmospheric conditions will first be examined to provide some circumstantial evidence of what might be the triggering mechanism. A hypothesis is then advanced to explain the possible reasons why the transition can occur in different seasons and how the warm state (in the NON type) can be maintained.
4. Time evolution of near-equatorial oceanic and atmospheric conditions
a. Oceanic circulation anomalies
1) Sea surface temperature anomalies
The time evolution of the composite SSTA over the equatorial Pacific (5°S–5°N) for the SP type (Fig. 3a) shows that the transition from positive to negative SSTA occurs in February over almost the entire eastern equatorial Pacific (EEP). The negative SSTA propagates slowly westward after the La Niña event has begun and reaches a maximum value of −1.5°C over the central equatorial Pacific (CEP) and EEP around October 0.
For the SU type (Fig. 3b), the time of transition from a positive to a negative SSTA does not occur until around April 0 over the EEP. The westward propagation of the negative SSTA appears to be slightly faster than that in the SP type. The time at which these anomalies reach a negative maximum is again around October 0. Notice also the persistence of these anomalies into year 1, especially in the EEP.
For the NON type, the time evolution of the SSTA is very different from that in the other two types. The SSTA over the EEP remains positive after the termination of the El Niño, and that over the WEP is always negative during the three years.
Two features common to all the three categories are worth noting. First, the SSTAs between October −1 and January 0 over the CEP and EEP ate all very similar, and yet the subsequent evolutions are drastically different. In other words, it is not possible to predict whether a La Niña event is likely to occur using the time series of SSTA alone. Other factors must be considered, which will be addressed in subsequent sections of this paper. Second, the region between 150°E and 180° appears as a gap that separates the Pacific into two relatively independent regions (the EEP and the WEP), each with a different pattern of SSTA change. However, possible reasons for this feature are beyond the scope of this study.
2) Seawater temperature anomalies
In addition to SSTA, ENSO events are also associated with temperature anomalies below the ocean surface. To study the vertical distribution of water temperature, the subsurface seawater temperature (SWT) is examined. During the period of March 0–May 0 in the SP episode (its occurrence phase), the entire CEP and EEP from the surface to 400-m depth is dominated by strong negative anomalies (Fig. 4a). This means that the water is colder than normal throughout a deep layer for a large part of the equatorial Pacific (east of ∼150°E). The two negative maxima along the thermocline reflect a deeper (shallower) mixed layer in the WEP (EEP). In the SU episodes, the negative SWT anomalies are located in the WEP and CEP, with a similar thermocline pattern to that in the SP type (Fig. 4b). However, positive anomalies dominate the EEP from the surface to ∼80 m, which means that the equatorial Pacific is still in the warm state of the ENSO cycle. The NON type shows features very different from the other two types, with above-normal water temperatures covering a large portion of the upper equatorial Pacific except for the extreme WEP (Fig. 4c). The negative anomalies below the surface indicate a deeper mixed layer in the EEP, which is exactly opposite to the situation in the other two types.
Since the onset time of the two types of cold episodes is different, it is useful to examine the time evolution of the subsurface seawater temperature anomaly (SWTA) as well. Because the strongest anomalies occur above 200 m (see Fig. 4), the SWTAs between the surface and 200 m are averaged. This also provides an estimate of the heat content anomalies. The negative SWTAs in the SP episodes are found to propagate eastward with a speed of ∼0.2–0.3 m s−1 (Fig. 5a), which is similar to that of the slow coupled air–sea mode discussed by Hirst (1986) but is much smaller than that of a Kelvin wave. The two negative maxima correspond to the onset (March 0–April 0) and mature phase (January 1) of the La Niña. While cold water dominates the CEP and the EEP, positive anomalies exist over the WEP, as expected.
In the SU episode, a similar eastward propagation of the negative SWTAs is also observed (Fig. 5b). The speed of propagation is even slower. Notice that the decrease in the thermocline depth in the EEP does not occur until around July 0. In the NON type, negative SWTAs are confined in the WEP with no apparent eastward propagation (Fig. 5c). Warm water dominates the entire CEP and EEP throughout most of the 3-yr period.
It should be pointed out that the negative SWTAs present over the WEP in all three categories during year −1 simply reflect the conditions associated with the mature stage of a warm event. What should be the focus is when and if such negative anomalies can propagate eastward. Such propagation will be examined further in section 5.
b. Atmospheric anomalies
1) Zonal wind anomalies
Two common features can be found in the near-surface (1000 hPa) zonal wind anomalies in both La Niña types (Figs. 6a and 6b). Easterly anomalies apparently propagate eastward from the EIO to the Pacific Ocean starting from around October −1. These anomalies reach a maximum at around the date line between October 0 and December 0. Notice that the region over the WEP between 120° and 150°E appears to be a “gap,” or a boundary, between easterly and westerly anomalies.
Some important differences can also be identified between the two types. Over the CEP (EIO), the time of transition of the zonal wind anomalies is around January 0 (March 0) in the SP type but around April 0 (July 0) in the SU type. The easterly anomalies in year 0 in the SP type also appear to be stronger and have a larger spatial extent compared with those in the SU type. In addition, the westerly anomalies in the SU type around the CEP in year −1 extend farther eastward during the El Niño year (and into year 0) while in the SP type, easterly anomalies are found east of 120°E. The low-level zonal wind anomalies in the NON type exhibit completely different characteristics (Fig. 6c). The westerly anomalies over the CEP and EEP simply weaken with the decay of the El Niño event and remain weakly positive afterward. Similarly, easterly anomalies prevail over the EIO from year −1 to year 0.
At the upper levels, the changes in the zonal wind anomalies in the three types of transition episodes are almost opposite to their counterpart at the low levels (not shown). This result reflects the reversal of the Walker circulation.
2) Convection
Since OLR data are only available after 1974, no composites are made. Instead, one individual case from each type (1987–89 for SP, 1982–84 for SU, and 1991–93 for NON) is used to illustrate the type of convective anomalies present.
While the familiar east–west dipole structure in the OLR anomalies appears in all three types, some differences can be found among them (Fig. 7). Whereas the transition in these anomalies occurs around January 0 in the SP type, positive and negative convective anomalies continue to propagate eastward and into year 0 for both the SU and NON types. The switch does not occur until around July 0 for the SU type. No transition occurs in the NON type except perhaps for a brief period over the WEP in the summer and fall of year 0. Another important feature is that the convection over the EEP in the SP type (at least in this example) was below normal beginning around October −1 (i.e., positive OLR anomalies).
5. A possible mechanism
In the previous section, two important precursors related to the occurrence of a cold event have been identified: near-equatorial anomalies in the zonal winds near the oceansurface and those in the ocean temperature beneath the ocean surface. The findings of Wang et al. (1999) suggest that the changes in the heat content anomalies observed in Fig. 5, which are a reflection of the thermocline depth, probably result from those in the near-surface wind fields. In other words, the atmosphere is likely responsible for initiating a cold event through variations in the wind fields. Possible mechanisms for such variations are explored, with the objective of identifying a physical mechanism.
a. Sea level pressure
Since changes in the zonal wind anomalies over the CEP first occur during the winter from year −1 to year 0 (see Fig. 6), the sea level pressure anomalies (SLPA) for the months November −1–February 0 are averaged to examine possible differences among the three types. In the Northern Hemisphere (NH), a typical Pacific–North America (PNA) pattern is observed in all types (Fig. 8). However, an area of positive SLPA exists in the region (20°–40°N, 120°–160°W) in the SP type. The negative anomalies in the NON type off the western coast of North America (the eastern North Pacific) are the largest. These patterns resemble the different phases of the North Pacific Oscillation (e.g., Gershunov and Barnett 1998).
A more striking difference between the SP and the other two types lies in the Southern Hemisphere (SH) within the longitude strip of 120°W to the date line. Between 20° and 40°S, the maximum SLPA is >2 hPa in the SP type while that in the other two types is ∼0.5 hPa. To the south (40°–60°S), the negative SLPA in the SP type is the smallest (∼−2 hPa versus ∼−4 to −5 hPa in the other two). The interpretation of these patterns is that the SH subtropical high is much stronger than normal in the SP type. This should then cause a strengthening of the southeast trades off the equatorial Pacific.
Because of this enormous difference in the SLPA patterns, the time evolutions of the SLPA within the strip 120°W to the date line are examined (Fig. 9). In the NH midlatitudes, the SLPA change from negative to positive at around January 0 in both La Niña types, although for the SP type, the change seems to occur earlier in the subtropics between 20° and 30°N. This corresponds to the positive area found in Fig. 8a.
Again, the most dramatic difference is found in the subtropical region of the SH (∼25°–45°S). During the period of November −1–April 0, the SLPA is positive (negative) in the SP (SU) type. This suggests that the SH subtropical high in the SP type begins to strengthen at around November −1, with the maximum anomaly occurring around January 0. For the SU type, no such strengthening occurs. In fact, the SH subtropical high has the largest negative anomaly at around January 0 and only strengthens slightly beginning around April 0. For the NON type, the SLPA in the SH subtropics is quite uniform throughout year 0 and year 1.
Thus, a major difference between the two types of La Niña events is that for the SP type, the subtropical high in the eastern Pacific (in both the NH and the SH) strengthens in late autumn of year −1 while those for the SU type remain weaker than normal. This result suggests that the subtropical highs in both hemispheres must play an important role in determining when and if a La Niña event will occur.
b. The hypothesis
Based on the time evolutions of the various parameters examined so far, it is possible to develop a hypothesis to describe the sequence of events that determines when a La Niña event is likely to occur.
In the SP episodes, starting at around October −1, the subtropical highs in both the NH and SH in the eastern and central Pacific begin to strengthen (Figs. 10a and 10b). This enhances the northeast and southeast trades off the equatorial Pacific east of the date line (Figs. 10a and 10b). The enhanced easterlies then cause surface water to drift poleward due to Ekman forcing, which reduces the thermocline depth. The upwelling sets up westward-propagating Rossby waves in the EEP and CEP in both hemispheres, as evidenced by the time–longitude cross sections of the MLD anomalies (Figs. 10c and 10d). These waves reach the western edge of the Pacific by January 0 and are reflected eastward so that the near-equatorial thermocline becomes shallower with time (Fig. 10e). The reduction in the thermocline depth along the equator implies a cooling of the ocean and a concomitant establishment of easterly anomalies that propagate eastward (see Fig. 6a). By April 0, the ocean cooling causes the SSTA over the Niño-3.4 to decrease below −0.5°C and La Niña occurs. This sequence of events is schematically depicted in Fig. 11.
For the SU type, the subtropical highs in both hemispheres are not enhanced in the late fall of year −1; in fact, they do not become more intense until around April 0, when the trade winds are also enhanced (Figs. 12a and 12b). The westward-propagating Rossby waves associated with the upwelling therefore only start to propagate westward around that time, especially in the SH. Not much propagation is seen in the NH, though. The slow air–sea coupled mode in the equatorial region can also be seen to propagate eastward beginning around July 0 (Fig. 12e) from the CEP, which then results in the onset of the SU cold event.
Because the subtropical highs in both hemispheres remain weaker than normal throughout year 0 in the NON type (see Fig. 9c), the strengthening of the trades cannot occur. Hence, no La Niña event can take place.
The crucial point in the above scenario is when and if the subtropical highs in both hemispheres will strengthen. The problem then becomes an investigation of the cause(s), such as strengthening. Since the subtropical high in the NH is part of the PNA pattern, perhaps a study can be made on whether the phenomenon is related to changes in this pattern. A similar study can be made on the subtropical high in the SH. A corresponding “Pacific–South America (PSA)” pattern may even be identified, which is related to the warm state and goes through different changes in such a way as to initiate or not initiate a La Niña event. However, these are beyond the scope of the present study.
It should be pointed out that this hypothesis is similar to that proposed by Li (1997) except that the feedback mechanisms are slightly different. Here, the asymmetry between the Northern and Southern Hemispheres appears prominently in the fact that the subtropical highs in the two hemispheres do not contribute equally to the feedback process. The subtropical high in the SH plays the pivotal role in determining whether and when a La Niña event will take place. In addition, the possible phase lock with the annual cycle is specifically addressed. Note that the mechanism of the intensification of the subtropical highs is not explicitly discussed because it is considered to be outside the objective of the present study.
6. Summary and discussion
a. Summary
This study attempts to identify the physical processes responsible for the initiation of a La Niña event through the analyses of oceanic and atmospheric parameters associated with seven such events in the past (during the period of 1951–96). Since this event can be considered as the cold phase of the entire El Niño–Southern Oscillation (ENSO) cycle, the conditions in the warm (El Niño) phase are first examined. It is found that of the 12 El Niño events studied, three initiated the cold phase two months after the termination of the El Niño. In four other cases, the time lag from the termination to the onset is around 5 months. In the other five cases, no La Niña occurred.
These results lead to the categorization of the La Niña events into three types: the spring occurrence (SP) type, in which the onset is in the boreal spring; the summer occurrence (SU) type, in which the onset does not occur until the boreal summer; and the nonoccurrence (NON) type, in which no La Niña event occurred. Comparisons of the sea surface temperature (SST) among the three types suggest that the SST anomalies are quite similar prior to the termination of the El Niño. However, differences among the three categories exist in the subsurface ocean structure. The thermocline in the SP type becomes shallow over the EEP much sooner than that in the SU type while the ocean temperature throughout a deep layer remains above normal in the NON type. The decrease in the heat content anomalies, which can be used as a proxy for those in the mixed-layer depth, also appears to propagate eastward in both types of cold events. The near-equatorial zonal winds in both the western and central equatorial Pacific also change sign at different times between these two types.
A further examination of the strength of the subtropical highs in both hemispheres in the eastern Pacific suggests that an enhancement of these highs can lead to a strengthening of the trades and a subsequent decrease in the mixed-layer depth. Rossby waves generated as a result of this upwelling then propagate westward and get reflected by the Asian coast. This results in a rising of the equatorial thermocline and a concomitant switch of near-equatorial zonal winds, and hence, the onset of the SP-type La Niña event. The reason for the delay in the onset of the SU type is because such an enhancement does not occur at the right time, especially for the subtropical high in the SH, and therefore cannot be anchored with the annual cycle. For the cases in which no La Niña event occurred, the subtropical highs never strengthened.
b. Statistical significance
The small sample size in each of the three categories of events (3, 4, and 5) raises the issue of statistical significance of the results. Therefore, differences in the anomalies between two categories are compared using the two-sample t test. Rather than testing all the variables analyzed in this study, the two most crucial parameters are examined: the SSTA, based on which the classification of the three categories is made, and the SLPA, based on which the hypothesis for the initiation of the cold event is proposed.
As might be expected from Fig. 2, the most significant difference in SSTA between the SP and SU episodes prior to the transition is in the Niño-3.4 region, where the t values are very large (Fig. 13a). These values are significant above the 99% level. A similarly significant difference in SSTA is also found between the SP and NON events for the same period (Fig. 13b) and between the SU and NON events around the onset time of the former (Fig. 13c). Therefore, these three categories of events can be considered to be independent, which is also evident from the results presented in the paper.
During the time (November −1–February 0) when the subtropical highs associated with SP events begin to strengthen (see Fig. 9a), the t-test result shows that the SLPAs in the subtropical areas of both hemispheres are significantly different between the SP and SU episodes (Fig. 14a). Two interesting results are also found. The areas of significant differences also include the equatorial region. In addition, all the significant regions appear to be some type of wave structure. However, these results will not be investigated further in this study. The importance of the SH subtropical high is further highlighted in the t-test result between the SP and the NON event (Fig. 14b). The subtropical region in the SH is the area with the highest t values that are significant above the 90% level.
The tests presented here clearly demonstrate that despite the small sample size, cold events of the ENSO cycle can indeed be divided into two general categories based on the onset time. Further, the physical hypothesis advanced in this study should be largely valid.
c. Discussion
A simple hypothesis is advanced in this study to describe the sequence of events that determine when and if a La Niña event will occur after the termination of an El Niño event. This hypothesis provides an explanation on the irregularity of the occurrence of a La Niña event following the termination of an El Niño episode, which cannot be accounted for in the delayed-oscillator theory. While the present observational results support to some extent the proposed mechanism advanced by Li (1997) based on modeling, they emphasize the difference in importance between the subtropical highs in the two hemispheres, with the subtropical high in the SH being the key determinant. It is of interest to note that XC also emphasized the role of the austral summer monsoon in the initiation of the warm phase of the ENSO cycle. It appears, therefore, that more attention should be focused on the contribution of the SH circulation to the overall ENSO problem.
The reason for the strengthening of the Pacific subtropical highs is not addressed in the present study. Since this occurs during the El Niño year, their intensity is likely to be related to the conditions associated with the El Niño event. It might, therefore, be possible to establish a feedback mechanism between the warm and cold phases of the ENSO cycle. That is, different conditions in an El Niño event (other than SSTA) may be responsible for the intensity of these two subtropical highs that subsequently determines when and if a La Niña event will occur in the following year. Li (1997) has also suggested that the Hadley circulation may also act as a negative feedback factor in this process, as originally put forth by Bjerknes (1969). The results of the present study have further suggested that the Hadley cell is asymmetric about the equator. Therefore, the feedback process must be studied with such an asymmetry in mind.
While this study has categorized the cold event into two categories so that the main characteristics of each type can be better identified, each individual event may occur at slightly different times. In fact, the specific time at which a cold event occurs will depend very much on the evolution of the two subtropical highs and to what extent they can anchor themselves with the annual cycle. Although the results presented here are based on a relatively small sample size, the statistical tests do indicate that this categorization is largely valid. Since it will take another few decades to significantly increase the sample size, it might be useful to verify the proposed hypothesis through numerical simulation using climate models. This may be an approach in future studies.
Acknowledgments
The authors would like to thank the various agencies for providing the different datasets used in this study. Much of the work of the first author was completed while he was a Croucher Foundation Senior Research Fellow and on sabbatical at both the Naval Postgraduate School and the University of Hawaii. The author would like to thank Profs. C. P. Chang, R. L. Elsberry, and B. Wang for supporting his visit to their institutions. Comments from Profs. R. Haney and C. P. Chang improved an earlier version of the manuscript. The suggestion by one of the reviewers, Dr. T. Li, to examine the subsurface ocean structure greatly strengthened the proposed hypothesis.
This research was sponsored by the City University of Hong Kong Grant 9360017. Work of the second author was also partly sponsored by the National Natural Science Foundation of China Grant 49705062.
REFERENCES
Barnett, T. P., 1983: Interaction of the monsoon and Pacific trade wind systems at interannual time scales. Part I: The equatorial zone. Mon. Wea. Rev.,111, 756–773.
——, 1984: Interaction of the monsoon and Pacific trade wind system at interannual time scales. Part III: A partial anatomy of the Southern Oscillation. Mon. Wea. Rev.,112, 2388–2400.
——, 1985: Variations in near global sea level pressure. J. Atmos. Sci.,42, 478–501.
Battisti, D. S., 1988: The dynamics and thermodynamics of a warming event in a coupled tropical atmosphere/ocean model. J. Atmos. Sci.,45, 2889–2919.
Bjerknes, J., 1969: Atmospheric teleconnections from the equatorial Pacific. Mon. Wea. Rev.,97, 163–172.
Busalacchi, A., and J. J. O’Brien, 1981: Interannual variability of equatorial Pacific in the 1960’s. J. Geophys. Res.,86, 10 901–10 907.
Cane, M. A., M. Munnich, and S. E. Zebiak, 1990: A study of self-excited oscillations of the tropical ocean–atmosphere system. Part I: Linear analysis. J. Atmos. Sci.,47, 1562–1577.
Gershunov, A., and T. P. Barnett, 1998: Interdecadal modulation of ENSO teleconnections. Bull. Amer. Meteor. Soc.,79, 2715–2725.
Graham, N. E., and W. B. White, 1988:The El Niño cycle: A natural oscillator of the Pacific ocean–atmosphere system. Science,240, 1293–1302.
Gutzler, D. S., and D. E. Harrison, 1987: The structure and evolution of seasonal wind anomalies over the near-equatorial eastern Indian and western Pacific Oceans. Mon. Wea. Rev.,115, 169–192.
Hackert, E. C., and S. Hastenrath, 1986: Mechanisms of Java rainfall anomalies. Mon. Wea. Rev.,114, 745–757.
Hirst, A. C., 1986: Unstable and damped equatorial modes in simple coupled ocean–atmosphere models. J. Atmos. Sci.,43, 606–630.
Lau, K. M., and P. H. Chan, 1985: Aspects of the 40–50 day oscillation during the northern winter as inferred from outgoing longwave radiation. Mon. Wea. Rev.,113, 1889–1909.
——, C.-P. Chang, and P. H. Chan, 1983: Short-term planetary-scale interactions over the Tropics and midlatitudes. Part II: Winter MONEX period. Mon. Wea. Rev.,111, 1372–1388.
Li, T., 1997: Phase transition of the El Niño–Southern Oscillation: A stationary SST mode. J. Atmos. Sci.,54, 2872–2887.
Meehl, G. A., 1987: The annual cycle and interannual variability in the tropical Pacific and Indian Ocean region. Mon. Wea. Rev.,115, 27–50.
Munnich, M., M. A. Cane, and S. E. Zebiak, 1991: A study of self-excited oscillations of the tropical ocean–atmosphere system. Part II: Nonlinear cases. J. Atmos. Sci.,48, 1238–1248.
Philander, S. G., 1983: El Niño–Southern Oscillation phenomena. Nature,302, 295–301.
——, 1985: El Niño and La Niña. J. Atmos. Sci.,42, 2652–2662.
——, and R. C. Pacanowski, 1981: The response of equatorial oceans to periodic forcing. J. Geophys. Res.,86, 1903–1916.
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.
Reynolds, R. W., and T. M. Smith, 1994: Improved global sea surface temperature analyses using optimum interpolation. J. Climate,7, 929–948.
Suarez, M. J., and P. S. Schopf, 1988: A delayed action oscillator for ENSO. J. Atmos. Sci.,45, 3283–3287.
Wang, B., and Z. Feng, 1996: Chaotic oscillations of tropical climate:A dynamic system theory for ENSO. J. Atmos. Sci.,53, 2786–2802.
——, R. Wu, and R. Lukas, 1999: Roles of the western north Pacific wind variation in thermocline adjustment and ENSO phase transition. J. Meteor. Soc. Japan,77, 1–16.
Wyrtki, K., 1975: El Niño—The dynamic response of the equatorial Pacific Ocean to atmospheric forcing. J. Phys. Oceanogr.,5, 572–584.
Xu, J., and J. C. L. Chan, 2000: The role of the Asian/Australian monsoon system in the onset time of El Niño events. J. Climate, in press.
Yasunari, T., 1985: Zonally propagating modes of the global east–west circulation associated with the Southern Oscillation. J. Meteor. Soc. Japan,63, 1013–1029.
——, 1990: Impact of Indian monsoon on the coupled atmosphere/ocean system in the tropical Pacific. Meteor. Atmos. Phys.,44, 29–41.
Time series of the Niño-3.4 SSTA during the transition from an El Niño to a La Niña episode. The 0.5°C (−0.5°C) line is the threshold for identifying the El Niño (La Niña) episodes.
Citation: Journal of Climate 13, 12; 10.1175/1520-0442(2000)013<2056:PMRFTT>2.0.CO;2
Composite time series of the SSTA over Niño-3.4 for the three types of transition episodes. Solid, dashed, and dotted lines indicate the SSTA of SP-type, SU-type, and NON-type episodes, respectively.
Citation: Journal of Climate 13, 12; 10.1175/1520-0442(2000)013<2056:PMRFTT>2.0.CO;2
Longitude–time cross section of SSTA composites over the equatorial area (5°S–5°N) for the (a) SP-type, (b) SU-type, and (c) NON-type episodes. Dark (light) shaded areas indicate SSTA > 0.5°C (<−0.5°C).
Citation: Journal of Climate 13, 12; 10.1175/1520-0442(2000)013<2056:PMRFTT>2.0.CO;2
Depth–longitude cross section of seawater temperature anomalies during the period of Mar 0–May 0 in the equatorial area (5°S–5°N) for the (a) SP-type, (b) SU-type, and (c) NON-type episodes. Shaded areas indicate negative anomalies. Unit: 0.01°C. The dashed line in each figure indicates the position of the thermocline.
Citation: Journal of Climate 13, 12; 10.1175/1520-0442(2000)013<2056:PMRFTT>2.0.CO;2
Longitude–time cross section of averaged SWT anomalies averaged between the surface and 200-m depth in the equatorial area (5°S–5°N) for (a) SP-type, (b) SU-type, and (c) NON-type episodes. Unit: 0.01°C.
Citation: Journal of Climate 13, 12; 10.1175/1520-0442(2000)013<2056:PMRFTT>2.0.CO;2
Longitude–time cross section of composites of the 1000-hPa zonal wind anomalies in the equatorial area (5°S–5°N) for (a) SP-type, (b) SU-type, and (c) NON-type episodes. Shaded areas indicate zonal wind anomalies <−0.5 m s−1.
Citation: Journal of Climate 13, 12; 10.1175/1520-0442(2000)013<2056:PMRFTT>2.0.CO;2
Longitude–time cross section of OLR anomalies over the equatorial area (5°S–5°N) during (a) 1987–89 (SP type), (b) 1982–84 (SU type), and (c) 1991–93 (NON type). Unit: W m−2.
Citation: Journal of Climate 13, 12; 10.1175/1520-0442(2000)013<2056:PMRFTT>2.0.CO;2
Composite SLPA for the period of Nov −1 to Feb 0 in the (a) SP-type, (b) SU-type, and (c) NON-type episodes. Shaded areas indicate SLPA < 0. Unit: 0.01 hPa.
Citation: Journal of Climate 13, 12; 10.1175/1520-0442(2000)013<2056:PMRFTT>2.0.CO;2
Latitude–time cross section of SLPA composites over the eastern Pacific (180°–120°W) for the (a) SP-type, (b) SU-type, and (c) NON-type episodes. Shaded areas indicate SLPA < 0. Unit: 0.01 hPa.
Citation: Journal of Climate 13, 12; 10.1175/1520-0442(2000)013<2056:PMRFTT>2.0.CO;2
Composites of various parameters related to the onset of the SP episodes: anomalies in the (a) SH tropical high (180°–120°W, 15°–35°S; solid, unit: hPa) and easterlies (180°–120°W, 15°–20°S; dashed, unit: m s−1) and (b) NH subtropical high (180°–120°W, 22.5°–32.5°N; solid) and easterlies (180°–120°W, 15°–20°N; dashed) and longitude–time cross section of MLD anomalies (unit: m) over (c) 6°–12°S, (d) 4°–10°N, and (e) 4°N–4°S.
Citation: Journal of Climate 13, 12; 10.1175/1520-0442(2000)013<2056:PMRFTT>2.0.CO;2
Schematic of the processes that lead to the occurrence of the SP-type La Niña event.
Citation: Journal of Climate 13, 12; 10.1175/1520-0442(2000)013<2056:PMRFTT>2.0.CO;2
As in Fig. 10 except for the composites of SU episodes. Because the MLD data are not available for the 1953–55 episode, these composites are for those in 1969–71, 1982–84, and 1994–96.
Citation: Journal of Climate 13, 12; 10.1175/1520-0442(2000)013<2056:PMRFTT>2.0.CO;2
The t values from testing the difference in SSTAs (averaged for the period indicated) between two categories: (a) SP and SU (Feb 0–May 0), (b) SP and NON (Feb 0–May 0), and (c) SU and NON (May 0–Aug 0). Shaded areas indicate significance above the 90% level.
Citation: Journal of Climate 13, 12; 10.1175/1520-0442(2000)013<2056:PMRFTT>2.0.CO;2
As in Fig. 13 except for the SLPA (a) between the SP and SU events and (b) between the SP and NON events for the period of Nov −1 to Feb 0.
Citation: Journal of Climate 13, 12; 10.1175/1520-0442(2000)013<2056:PMRFTT>2.0.CO;2
Termination time of El Niño and the occurrence time of La Niña episodes during 1951–97 based on the Niño-3.4 SSTA. The duration of transition is between an El Niño and a La Niña event.
Hereafter, the year after an El Niño event is designated as year 0. This is also the year during which the SSTA first reaches a minimum (the La Niña year) for both the SP and the SU types. The year before and after are labeled as year −1 and year +1, respectively. The months in each of these years will be labeled the same way.
