Abstract

The mature phases of El Niño events show a strong tendency of locking to the end of the calendar year. The roles of seasonal variations of the basic state and the relative contributions of individual components of the basic state in this phase locking are investigated using the Zebiak–Cane model. It is shown that seasonal variations of the mean state from July to November have a positive contribution and those from December to June have a negative contribution to preexisting warm SST anomalies. Among the basic-state parameters, the sea surface temperature (SST) is a major factor for the locking of the El Niño mature phase through the anomalous advection of the mean SST gradient. This result differs from previous studies that attribute the El Niño phase locking mainly to seasonal changes in the mean wind divergence. The present result indicates the importance of a proper simulation of mean SST and its seasonal evolution for the simulation of El Niño phase locking in coupled models. Further experiments show that the phase locking of the El Niño mature phase cannot be explained by a balance between the warming trend due to downwelling Kelvin waves and the cooling trend due to upwelling Rossby waves.

1. Introduction

The mature phases of El Niño and La Niña events tend to occur toward the end of a calendar year (Rasmusson and Carpenter 1982; Chang et al. 1995; Tziperman et al. 1998; An and Wang 2001). This phase locking is one of the most robust features of El Niño–Southern Oscillation (ENSO) cycles. Studies have been made to understand this phase locking, and the interaction of ENSO and the seasonal cycle has been recognized as the fundamental cause for the phase locking of ENSO (Jin et al. 1994; Chang et al. 1994, 1995; Tziperman et al. 1994, 1995, 1997; Xie 1995; Neelin et al. 2000; An and Wang 2001).

Zebiak and Cane (1987) proposed that seasonally varying coupling instability strength is responsible for the locking of El Niño mature phases. During the boreal winter, the coupling strength begins to decrease significantly and becomes insufficient to maintain the large anomalies against dissipation, and thus anomalies tend to decrease. Tziperman et al. (1998) and Galanti and Tziperman (2000) incorporated the seasonally varying coupling instability strength (Zebiak and Cane 1987) into the delayed oscillator theory for ENSO cycles (Suarez and Schopf 1988; Battisti and Hirst 1989; Graham and White 1988; Munnich et al. 1991) to examine this mechanism. They showed that ENSO peaks when the seasonally varying amplification of Rossby and Kelvin waves by the coupled instability reaches its minimum strength, at the end of the calendar year. They suggested that the peak time of the ENSO events is set by the dynamics to allow a balance between the warming trend due to downwelling Kelvin waves and the cooling trend due to upwelling Rossby waves. This point of view is supported by An and Wang (2001).

The coupling strength between the atmosphere and ocean depends on the basic state. Important factors contributing to the seasonally changing coupling strength includes the seasonal movement of the Pacific intertropical convergence zone (ITCZ) and its effects on the atmospheric heating (Philander 1983; Hirst 1986), the seasonal change in the zonal SST gradient, the thermocline depth, zonal surface winds, mean SST (Hirst 1986), and mean upwelling (Battisti 1988). The seasonal changes of all these factors are connected to each other. Each of these factors induces the strongest coupling strength at a different time of the year. Note that the mean thermocline depth in the Zebiak–Cane model is prescribed as a function of longitude only and without seasonal variations. Thus, its effects on seasonal variations of the coupling strength are absent in the standard Zebiak–Cane model. Inclusion of this will improve ENSO phase locking to the annual cycle (An and Wang 2001). Because of the important impacts of the basic state on the coupling strength, a proper simulation of the mean state and its seasonal variation is considered to be essential for models to capture the observed ENSO phase locking (e.g., Xie 1995; Tziperman et al. 1997; Li and Hogan 1999; An and Wang 2001; Galanti et al. 2002; Spencer 2004; Vecchi and Harrison 2006; Vecchi 2006).

Although the seasonally varying coupling strength is crucial for the locking of El Niño mature phases, there are considerable differences among previous studies regarding the seasonal dependence of the coupling strength. For instance, Zebiak and Cane (1987) and Galanti et al. (2002) suggested that the coupling strength is strong during the boreal summer–fall and weak during the boreal spring, while An and Wang (2001) proposed that the coupling strength is strong during the boreal spring and weak during the boreal summer–fall. Tziperman et al. (1997, 1998) showed that the coupling strength is strong during summer and weak during fall. Li (1997) indicated that in the stationary SST mode the coupling instability is large in the later half of the year. These differences in the seasonal variation of the coupling strength indicate that the mechanism for the phase locking of ENSO mature phases is still not clear.

The roles of seasonal changes of the basic state in the ENSO phase locking have been examined in previous studies (Tziperman et al. 1997; Galanti and Tziperman 2000; An and Wang 2001). Tziperman et al. (1997) performed sensitivity experiments with the Zebiak–Cane model (Zebiak and Cane 1987) to identify which mean states are most responsible for the phase locking. In their experiments, all basic states are fixed at the annual mean value except for one variable that retains the annual cycle. They found that the seasonal phase locking of El Niño events is dominated by the basic-state wind divergence associated with the ITCZ. The annual cycles of the background SST and ocean upwelling are of secondary importance to the phase locking. An and Wang (2001) repeated experiments similar to those of Tziperman et al. (1997) except for an increase of 20% in the coupling coefficient. Their results suggest that seasonal variations in the basic-state wind divergence are not critical for the phase locking of El Niño mature phases. The different results in previous studies confirm the sensitivity of the results to the coupling strength and point to the necessity for further investigation of the roles of the seasonal varying basic state. Note that because all of the climatological background fields specified in the Zebiak–Cane model are mutually dependent, it may not be physically reasonable to make some of them vary seasonally and others not (Tziperman et al. 1997).

The El Niño phase locking has also been studied using complex models. A robust feature for the termination of El Niño events is the southward shift of the near-date line westerly wind stress anomalies during boreal winter (Harrison and Vecchi 1999; Vecchi and Harrison 2003). Vecchi (2006) showed that this feature is related to the seasonal movement of the warmest SST to south of the equator, which brings the convective anomalies from centered on the equator to centered south of the equator. This is also found in other studies (Spencer 2004; Lengaigne et al. 2006). The wind changes over the equatorial central Pacific can generate eastward-propagating thermocline shallowing that leads to the turnaround of the SST tendency in the eastern equatorial Pacific (Vecchi and Harrison 2006). Vecchi (2006) also found that the eastern equatorial Pacific easterlies are tied primarily to changes in climatological SST. A precursor to the ENSO phase transition is the buildup of heat content in the western North Pacific (Weisberg and Wang 1997; Wang et al. 1999; Guilyardi et al. 2003). This buildup is attributed to western North Pacific wind changes (Wang et al. 1999; Guilyardi et al. 2003). The western North Pacific wind response to equatorial Pacific SST anomalies depends on the mean SST state and is stronger in the northern summer/autumn (Guilyardi et al. 2003).

The present study attempts to explore the issue of El Niño phase locking using carefully designed experiments with the Zebiak–Cane model. It is found that seasonal changes in the mean SST have a dominant role for the phase locking of El Niño mature phases. The remainder of the text is organized as follow. In section 2, we investigate the effects of seasonal variations of the basic state on the evolution of SST anomalies around the El Niño mature phase. The relative importance of individual components of the basic state is explored in section 3. In section 4, we diagnose the contributions of various terms in the thermodynamic equation for the SST tendency. In section 5, we examine the mechanisms proposed by previous studies. Section 6 gives conclusions.

2. Effects of seasonal variations of the basic state

In this section, we examine the effects of seasonal variations of the basic state on the seasonal evolution of SST anomalies around the El Niño mature phase. This is accomplished by experiments using the Zebiak–Cane model. This model is suitable for our purpose because El Niño mature phases simulated by this model tend to occur at the end of the calendar year as in observations, and the basic state (surface winds, wind divergence, ocean surface currents, SST, and upwelling) in this model is specified. The basic state in the Zebiak–Cane model can be easily modified to investigate the relative importance of different components in the phase locking of El Niño mature phases. To understand the roles of seasonal variations of the basic state in the El Niño phase locking, we perform a series of experiments with the Zebiak–Cane model. In these experiments, we modify the basic state and examine the resulted changes in the evolution of SST anomalies.

In the control run, the Zebiak–Cane model is integrated for 1000 yr with a seasonally varying basic state. Forty strong El Niño events are randomly chosen from the control run. Here, an El Niño event is classified as strong when maximum monthly mean SST anomalies in the Niño-3 region (5°S–5°N, 150°–90°W) exceed 3°C. The states corresponding to these El Niño events provide initial conditions for the following experiments.

We perform 13 sets of experiments: TESTi (i = 0, 1, . . . , 12). These sets of experiments have the same initial state but differ in the basic state. In these experiments, the basic state is fixed at either the annual mean (i = 0) or the monthly mean of the ith calendar month (i = 1 for January, . . . , 12 for December). These sets of experiments are integrated for one month, starting from a same initial condition provided by the control run. Let Ti (i = 0, 1, . . . , 12) denotes the monthly mean SST anomaly in the Niño-3 region. The quantity

 
formula

which is the deviation of the SST anomalies when the basic state is fixed at a specific month from those when the basic state is fixed at the annual mean, occurs purely due to the difference in the basic state. This SST deviation is used to measure the contribution of the ith month basic state to the SST anomalies in the present study.

For each set of experiments above, 12 different types of experiments are performed. These experiments differ only in the initial conditions. Each type of these experiments is integrated for one month from the first day of each of the 12 months from the June before to the May after the peak of each of the 40 El Niño events with saved initial conditions from the control run. Because there are 40 El Niño events, there are total of 12 × 40 experiments for each set. For each type of experiments, there are 13 sets of experiments as described in the previous paragraph. For example, starting from 1 June of one specific El Niño event, 13 parallel experiments are made with the basic state set at either annual mean or each of the 12 calendar months. These experiments are denoted as TESTijk, where j (j = 1 for June, . . . , 12 for May) refers to the ordinal number of the initial conditions during an El Niño event, k (k = 1, . . . , 40) refers to the ordinal number of the 40 El Niño events, and i (i = 0, 1, . . . , 12) refers to the number of the basic state. Correspondingly, dTijk gives the SST deviation for the jth initial condition of the kth El Niño event. The composite mean deviation, calculated as

 
formula

measures the averaged effects of the ith month mean state to monthly mean SST anomalies. The accumulated effects of seasonal variations of the basic state from June to the lth month is represented by

 
formula

where l = 1, . . . , 12 corresponds to June, July, . . . , May, respectively. Note that the accumulated effects are calculated starting from June.

The obtained composite mean deviation for the Niño-3 SST (ΔTi, i = 1 for January, . . . , 12 for December) is shown in Fig. 1a and the accumulated deviation (ATl, l = 1 for June, . . . , 12 for May) is shown in Fig. 1b, respectively. From Fig. 1a, the composite deviation has a clear seasonal variation, negative from December to June and positive from July to November. This indicates that the July–November mean state contributes positively and the December–June mean state contributes negatively to preexisting positive SST anomalies. The most favorable month is August and the most unfavorable month is April. As such, the positive SST anomalies increase from June, peak in November, and turn to decrease after December (Fig. 1b).

Fig. 1.

(a) Composite mean deviation of Niño-3 SST anomalies (°C) due to perturbation of mean state at individual months from the annual mean and (b) accumulation of the composite mean deviation calculated starting from June.

Fig. 1.

(a) Composite mean deviation of Niño-3 SST anomalies (°C) due to perturbation of mean state at individual months from the annual mean and (b) accumulation of the composite mean deviation calculated starting from June.

Figure 2 compares the composite Niño-3 SST anomalies for the 40 El Niño events in the control run and those derived from 40 experiments in which the model is integrated for one year starting from the first day of June of the 40 El Niño events with the basic state fixed at the annual mean. With the fixed annual mean basic state, the maximum Niño-3 SST anomalies are smaller and the peak time is delayed by 3 months. When the accumulated effects of seasonal variations of the basic state are added, the maximum Niño-3 SST anomalies are close to that in the control run. Figure 2 also suggests that the SST anomalies are nearly the linear superposition of inherent anomalies (i.e., when the basic state is fixed at the annual mean) and those deviations due to seasonal changes in the basic state. From Fig. 1a and Fig. 1b, the accumulated deviation due to seasonal changes in the basic state reaches 0.7°C in November and −1.9°C in May, respectively, with a range of 2.6°C. The range of inherent SST anomalies is about 1.5°C (Fig. 2). Thus, seasonal variations of the basic state have a relatively larger contribution to the fluctuation of SST anomalies.

Fig. 2.

Composite mean Niño-3 SST anomalies (°C) for 40 El Niño events. The solid line represents the case in the control run. The dotted line represents the case in which the basic state is fixed at the annual mean and the model is integrated for one year, starting from 1 June with the initial value taken from the control run (called the inherent anomalies). The dashed line represents the sum of the inherent anomalies (the dotted line) and the accumulated deviation due to seasonal variations in the basic state (Fig. 1b).

Fig. 2.

Composite mean Niño-3 SST anomalies (°C) for 40 El Niño events. The solid line represents the case in the control run. The dotted line represents the case in which the basic state is fixed at the annual mean and the model is integrated for one year, starting from 1 June with the initial value taken from the control run (called the inherent anomalies). The dashed line represents the sum of the inherent anomalies (the dotted line) and the accumulated deviation due to seasonal variations in the basic state (Fig. 1b).

3. Relative importance of individual components of the basic state

As reviewed in the introduction, there are inconsistent results in previous studies regarding which mean state is most important for the phase locking of El Niño mature phases. Here, we reexamine this issue with similar experiments as described in the previous section. As seen from the previous section, our experiments are designed to examine how the model SST anomalies evolve under different basic states and whether a preexisting warm event can still peak at the end of the calendar year when the mean state is modified. The use of 40 El Niño events ensures that the results are robust.

To determine the relative importance of seasonal variations of individual components of the basic state in the phase locking of El Niño mature phases, we perform experiments with only one component of the basic state changing with the season and with the rest of the components fixed at their annual means. We calculate the composite mean deviation of Niño-3 SST anomalies in the same way as in the previous section. The results are shown in Figs. 3a–c. The sum of the contributions due to individual components is displayed in Fig. 3d. For comparison, we also include in Fig. 3d the curve from Fig. 1a, that is, the composite deviation when all the components of the basic state change with season.

Fig. 3.

Composite mean deviations of Niño-3 SST anomalies (°C) for the experiments in which only one component of the basic state is set to a specific month and the rest of the components are fixed at their annual means: (a) mean SST is seasonal; (b) mean ocean currents (line with long dashes), surface winds (solid line), and mean upwelling (line with short dashes) are seasonal, respectively; (c) mean wind divergence is seasonal; and (d) the sum of the composite deviation due to changes in each component of the basic state (dashed line) and the composite deviation due to changes in all the components of the basic state (solid line).

Fig. 3.

Composite mean deviations of Niño-3 SST anomalies (°C) for the experiments in which only one component of the basic state is set to a specific month and the rest of the components are fixed at their annual means: (a) mean SST is seasonal; (b) mean ocean currents (line with long dashes), surface winds (solid line), and mean upwelling (line with short dashes) are seasonal, respectively; (c) mean wind divergence is seasonal; and (d) the sum of the composite deviation due to changes in each component of the basic state (dashed line) and the composite deviation due to changes in all the components of the basic state (solid line).

It is apparent that the background SST change has a dominant contribution to the phase locking of El Niño mature phases. Secondary contribution comes from the ocean current change. The contributions from seasonal changes in other components are negligible. From Fig. 3d, the effects of seasonal variations of the basic state are nearly the linear superposition of the effects of seasonal variations of individual components.

Our results disagree with Tziperman et al. (1997) regarding which component of the basic state has a dominant contribution to the phase locking of El Niño mature phase in the Zebiak–Cane model. This disagreement is related to the different experimental design. In Tziperman et al. (1997), the model is integrated continuously with the basic state of one component varying and that of other components fixed. In reality, all the components in the basic state are related together. When only one component is allowed to vary with season and the other components are fixed, the coupled strength will deviate largely from the reality. In the Zebiak–Cane model, the evolution of SST anomalies depends critically on the coupling strength. This sensitivity has been pointed out in previous studies. For example, by increasing the coupling coefficient by 20% in the Zebiak–Cane model, An and Wang (2001) obtained results different from Tziperman et al. (1997) though their experimental design is similar to that of Tziperman et al. (1997). This indicates the importance of a properly varying coupling strength, and the results based on a distorted seasonal varying coupling strength may need verification. Our experimental design is focused on how the model response changes under different basic states given the same initial condition. Because the model is integrated for only one month, the distortion of the coupling strength is minimized in the present study.

In our experiments, the model is integrated for only one month. This allows fast oceanic processes, such as horizontal temperature advections, to impact SST anomalies but does not allow impacts of slow thermocline processes associated with ocean waves induced by wind changes, as they do not have enough time to affect SST anomalies. This differs from the experiments performed by Tziperman et al. (1997) that allows both fast and slow oceanic processes. One may infer that this difference is one reason for the different results between the present study and the previous studies (Tziperman et al. 1997, 1998). This, however, seems not to be the case according to Fig. 2, which compares the composite Niño-3 SST anomalies from the control run (solid line) and the accumulated deviations (dashed line) due to seasonal changes of the basic state derived from experiments. In these experiments, ocean waves in association with SST and wind changes do not have enough time to affect thermocline changes in the equatorial eastern Pacific and in turn impact SST anomalies there since the model is integrated for only one month. As such, the accumulated deviations do not include most of the effects of ocean waves. Nevertheless, the solid and dashed lines in Fig. 2 are close to each other, though there is one-month difference in the peak time. This suggests that ocean wave processes are not critical for the timing of El Niño peak phases and thus are not a main reason for the difference of our results from previous studies.

Our results are consistent with previous complex model studies (Spencer 2004; Vecchi 2006; Lengaigne et al. 2006) regarding the importance of mean SST changes in the El Niño phase transition. These studies showed that the response of atmospheric winds and convection to El Niño depends on the mean SST distribution. The warmest SST is located north of equator during July–October and moves to south of equator toward the end of the calendar year. This movement of the warmest SST induces a dramatic change in the wind response over the equatorial central Pacific and in turn in the thermocline anomalies in the eastern equatorial Pacific. There are, however, differences in the manner by which the mean SST affects the SST evolution. Our study suggests that the impacts of mean SST change are via the anomalous advection of the mean SST gradient, whereas in the previous studies the SST influence is through eastward-propagating equatorial Pacific thermocline shallowing and the contribution of advection is not discussed. In the next section, we will discuss previous diagnostic studies that show the contribution of anomalous zonal advection of the mean SST gradient to the El Niño phase locking.

Two extra sets of experiments have been performed in order to confirm our results. In these experiments, the model is integrated for one year, starting from 1 June with initial conditions from each of the 40 El Niño events in the control run. In the first set, only one component of the basic state varies with season and the other components are fixed at annual mean. The results of this set of experiments are shown in Fig. 4a. In the second set, one component of the basic state is fixed at annual mean and the other components vary with season. The results of this set of experiments are shown in Fig. 4b. For comparison, we include in Fig. 4 the composite SST anomalies for the 40 El Niño events in the control run. From Fig. 4a, the composite SST anomalies in the experiments with only one component of the basic state varying with season show notable differences from those in the control run. In comparison, the discrepancy is the least when only SST changes with season. The SST anomalies when only the wind divergence changes with season are close to those when all the components are fixed at annual mean. From Fig. 4b, the largest deviation from the control run is seen when SST is fixed at annual mean and the smallest when the wind divergence is fixed at annual mean. The results of these extra experiments support those from primary experiments.

Fig. 4.

(a) Composite mean Niño-3 SST anomalies (°C) for 40 El Niño events. The solid line is for the control run. The other lines are for experiments in which the model is integrated for one year, starting from 1 June with the initial values from each of the 40 El Niño events in the control run. In these experiments, only one component varies with season and the other components are fixed at annual mean, or all the components are fixed at annual mean. (b) Similar to (a), but for experiments in which only one component is fixed at annual mean and the other components vary with season.

Fig. 4.

(a) Composite mean Niño-3 SST anomalies (°C) for 40 El Niño events. The solid line is for the control run. The other lines are for experiments in which the model is integrated for one year, starting from 1 June with the initial values from each of the 40 El Niño events in the control run. In these experiments, only one component varies with season and the other components are fixed at annual mean, or all the components are fixed at annual mean. (b) Similar to (a), but for experiments in which only one component is fixed at annual mean and the other components vary with season.

The current coupled models have difficulty in simulating correctly the mean SST gradient distribution in the tropical Pacific and its seasonal evolution (Mechoso et al. 1995). These models also show problems in reproducing the observed phase locking of ENSO (Latif et al. 2001). The present result indicates the importance of a proper simulation of mean SST and its seasonal evolution for capturing the El Niño phase locking in the model, consistent with previous studies (e.g., Spencer 2004; Vecchi 2006).

4. Diagnosis of the SST equation

The results of the previous section show that seasonal variations in the mean SST play a dominant role for the locking of El Niño mature phases to the end of the calendar year. In this section, we perform diagnosis of the SST equation to verify the effects of seasonal SST change.

The SST anomaly equation in the TESTi (i = 0, 1, . . . , 12) is

 
formula

where M(x) is x if x > 0 and 0 if x ≤ 0. The equation for the SST deviation (TiT0, i = 1, . . . , 12) can be written in the following form:

 
formula

where

 
formula
 
formula
 
formula

The terms A, B, and C represent the contributions due to changes in anomalous current advection of mean SST, mean current advection of anomalous SST, and upwelling. Figures 5a,b,c show contributions of the above three terms to the Niño-3 SST deviations. Included for comparison are the composite Niño-3 SST deviations derived from experiments with the corresponding component of the basic state changing with season.

Fig. 5.

Contributions to Niño-3 SST deviation (dashed lines) from (a) changes in the anomalous current advection of mean SST, (b) the mean current advection of anomalous SST, and (c) the upwelling. Included for comparison are the composite deviations derived from experiments with the corresponding component of the basic state changing with season (solid lines).

Fig. 5.

Contributions to Niño-3 SST deviation (dashed lines) from (a) changes in the anomalous current advection of mean SST, (b) the mean current advection of anomalous SST, and (c) the upwelling. Included for comparison are the composite deviations derived from experiments with the corresponding component of the basic state changing with season (solid lines).

In Fig. 5, the composite deviations due to changes in the anomalous advection of the mean SST gradient are the largest, followed by those due to changes in the advection of anomalous SST by mean currents. This confirms the results of the previous section. Note that the inconsistency seen in Fig. 5a around March–April is due to the restriction on the maximum SST in the model. The model only allows total SSTs less than 30°C. When the SST exceeds 30°C, it is adjusted back to 30°C. This occurs in March and April when the mean SST in the central and eastern equatorial Pacific can reach 27.5°–28°C (Fig. 6a). This restriction exaggerates the contribution of seasonal SST changes derived from the experiment when only SST is allowed to vary with season. This is confirmed by performing extra experiments with the above restriction removed. The obtained composite SST deviation when only SST is allowed to vary with season is close to the contribution from the anomalous advection of mean SST based on the diagnosis, that is, the dashed curve in Fig. 5a. Nevertheless, the anomalous advection of mean SST is still the largest term. We also performed experiments for weak El Niño events (maximum Niño-3 SST anomalies are below 2°C). Again, the anomalous advection of the mean SST gradient still has the leading contribution to the SST deviations though its magnitude is reduced compared to that for strong events.

Fig. 6.

Month–longitude plot of (a) climatological mean SST (°C) and (b) zonal gradient of monthly mean SST deviation from annual mean [°C (1000 km)−1] along the equator (5°S–5°N).

Fig. 6.

Month–longitude plot of (a) climatological mean SST (°C) and (b) zonal gradient of monthly mean SST deviation from annual mean [°C (1000 km)−1] along the equator (5°S–5°N).

To understand how the SST basic state affects the El Niño phase locking, we show in Fig. 6 the climatological mean SST and the zonal gradient of monthly mean SST deviation from the annual mean along the equator. It is obvious that the zonal SST gradient is larger during July–November and smaller during December–June compared to the annual mean. Given same eastward surface current anomalies in the equatorial central Pacific, the anomalous advection of the mean SST gradient would contribute positively to the warming during July–November but negatively during December–June. Thus, the SST basic-state change is favorable for a turnaround of the SST tendency around November–December.

The role of anomalous advection of the mean SST to El Niño’s phase transitions is consistent with previous diagnostic studies (Kang et al. 2001; An et al. 1999). Kang et al. (2001) performed an SST budget analysis based on the ocean assimilation data produced by the National Centers for Environmental Prediction (NCEP; Ji et al. 1995). Their results show that the term of anomalous zonal advection of the mean SST gradient is consistent with the SST tendency (see their Fig. 2a and Fig. 1d). In particular, the zonal advection term changes sign at the peak of SST anomalies. The mean upwelling term contributes largely to the SST warming in the eastern equatorial Pacific (their Fig. 2e). However, the mean upwelling term switches its sign at a time different from the SST tendency. Diagnostics of SST equation in the Zebiak–Cane model also shows that the anomalous zonal advection of the mean SST gradient contributes to the turnaround of the SST tendency (An et al. 1999, see their Fig. 1a). Our results are not against Jin and An (1999) and An et al. (1999), who pointed out that the zonal advection feedback (related to the zonal mean of zonal current anomalies) and the thermocline feedback (related to the zonal contrast of thermocline anomalies) are responsible for the phase transition of the ENSO in an almost equally manner.

Many previous studies emphasized the role of the thermocline feedback associated with the mean upwelling in the ENSO development and phase transition (Suarez and Schopf 1988; Battisti and Hirst 1989; Jin 1996, 1997a, b; Li 1997). Our study suggests that the mean upwelling term is small. There seems to be an apparent disagreement between the present study and previous results. This is, however, not the case. The present study considers the mean upwelling acting on the vertical gradient of temperature anomalies, whereas the thermocline feedback in these previous studies refers to the mean upwelling associated with only subsurface temperature anomalies. In the Zebiak–Cane model, the mean upwelling associated with subsurface temperature anomalies (thermocline depth anomalies) does contribute to the ENSO evolution (Zebiak and Cane 1987; An et al. 1999). However, only the zonal mean part of thermocline depth anomalies tends to be in phase with the SST tendency (Jin and An 1999; An et al. 1999). The mean upwelling associated with surface temperature anomalies has a damping effect in the Zebiak–Cane model. This damping effect cancels most of the thermocline feedback so that the total mean upwelling term is small in the present study.

Li (1997) proposed a stationary SST mode to understand physical mechanisms for the ENSO phase transition. The stationary SST mode tends to be destabilized under the July–November mean state in July–November and stabilized under the December–June mean state (see his Fig. 12). Our results are consistent with Li (1997) about the timing of favorable mean states. However, there is a difference regarding which process contributes to the seasonal change in the instability. Li (1997) emphasized the effect of mean vertical upwelling associated with anomalous thermocline depth changes. Our study indicates that the anomalous advection of the mean SST gradient has the leading contribution to the seasonal dependence. This difference has following reasons. First, in the stationary SST mode, the anomalous advection term was omitted according to a scale analysis based on given parameter values. Second, Li (1997) separated the mean upwelling term associated with SST and subsurface temperature anomalies and focused on the mean upwelling of the subsurface temperature anomalies associated with thermocline depth anomalies, whereas the present study considers the mean upwelling of the vertical gradient of temperature anomalies. As pointed out in the previous paragraph, the mean upwelling associated with surface temperature anomalies has a damping effect and tends to cancel that associated with thermocline depth anomalies. Third, the thermocline effects depend on the parameterization of subsurface temperature anomalies. Li (1997) parameterized subsurface temperature anomalies using thermocline depth anomalies under a constant mean thermocline depth of 50 m. In the Zebiak–Cane model, the parameterization depends on the longitudinal varying mean thermocline depth. In comparison, the mean upwelling term in the Zebiak–Cane model is less sensitive to thermocline depth anomalies compared to the stationary SST mode.

5. Test of mechanisms of previous studies

Tziperman et al. (1998) and Galanti and Tziperman (2000) suggested that the peak time of warm SST anomalies is set by a balance between the warming trend due to downwelling Kelvin waves and cooling trends due to upwelling Rossby waves. Here, we test this mechanism using experiments with the Zebiak–Cane model. We perform 40 experiments corresponding to the 40 El Niño events. In these experiments, the reflection condition at the western boundary is removed and the rest of the conditions are the same as those in the control run. The model is integrated for one year, starting from the first day of June, as chosen in section 2. Because the Rossby waves excited in the central Pacific and amplified by the strong air–sea coupling during the summertime have not reached the western boundary before 1 June, the balance between the warming and cooling trends does not happen under this case. According to this mechanism, the El Niño mature phase shall not occur at the end of the calendar year. However, the composite results for the 40 experiments show that the El Niño events still peak around the end of the calendar year (Fig. 7). The results suggest that this mechanism is not responsible for the locking of El Niño mature phases to the end of the calendar year. The difference in the amplitude of SST anomalies around the El Niño mature phase in Fig. 7 does suggest the negative feedback of reflected upwelling Rossby waves.

Fig. 7.

Composite Niño-3 SST anomalies (°C) for the 40 El Niño events. The solid line is for the control run and the dashed line is for the experiments in which the reflection condition at the western boundary is removed and the rest of the conditions are the same as those in the control run.

Fig. 7.

Composite Niño-3 SST anomalies (°C) for the 40 El Niño events. The solid line is for the control run and the dashed line is for the experiments in which the reflection condition at the western boundary is removed and the rest of the conditions are the same as those in the control run.

One may argue that the initial values of the ocean model already have enough memory of the information of the previous ENSO cycle to affect the phase locking. The results from the experiments in which the model is integrated for one year starting from 40 different initial states at 1 June and the basic state is fixed at the annual mean show that the peak of composite Niño-3 SST anomalies does not occur at the end of the year (Fig. 2, dotted curve). This indicates that the ocean memory of the information of the previous ENSO cycle cannot account for the phase locking of warm events.

6. Conclusions

The importance of seasonal variations in the basic state for the locking of El Niño mature phases to the end of the calendar year is explored through carefully designed experiments with the Zebiak–Cane model. The results show that the evolution of SST anomalies around the El Niño mature phase is a linear superposition of the inherent SST anomalies (the basic state fixed at the annual mean) and that due to seasonal variations in the basic state. Seasonal variations in the basic state contribute to increases of Niño-3 SST anomalies from July to November and decreases of Niño-3 SST anomalies from December to June. It is inferred that the phase locking of El Niño mature phases is closely related to seasonal effects of the basic state.

Further experiments demonstrate that the seasonal change in the mean SST is a predominating factor leading to the locking of El Niño mature phases to the end of the calendar year. This is confirmed by a diagnosis analysis of the SST tendency equation, which shows that the dominant contribution to the evolution of Niño-3 SST deviation around the El Niño mature phase is the anomalous advection of the mean SST gradient. The present result differs from previous studies that attribute the phase locking of El Niño mature phases to seasonal changes in the mean wind divergence. Our results point to the importance of a proper simulation of mean SST and its seasonal variation in the coupled model for capturing the El Niño phase locking.

The Zebiak–Cane model has several limitations. First, the mean thermocline depth, which is a critical factor for ENSO, does not vary with season in the standard Zebiak–Cane model. In reality, the seasonal displacement of thermocline in the equatorial eastern and central Pacific can reach 10–30 m (Wang et al. 2000). Second, the atmospheric component of the model is not able to reproduce the observed wind anomalies in the western North Pacific (An and Wang 2001), whose seasonality is suggested as a factor in the ENSO phase transition (Harrison and Vecchi 1999; Wang et al. 1999). Third, the parameterization of the entrainment or subsurface temperature has uncertainty and the SST evolution is sensitive to this (Mantua and Battisti 1995). Efforts have been made to improve the representation of the entrainment temperature in ENSO simulations (e.g., Dewitt and Perigaud 1996; Zhang et al. 2003, 2005). Fourth, seasonal changes in tropical Pacific mean winds contribute to seasonal changes in mean thermocline depth (Wang et al. 2000). In the Zebiak–Cane model, because the mean thermocline depth does not vary with season, the impacts of seasonal wind changes may be significantly underestimated based on experiments with the Zebiak–Cane model. In view of these limitations, the results obtained based on experiments with the Zebiak–Cane model may deviate from reality and it is still inconclusive regarding which component of the basic state has the leading contribution to the El Niño phase locking.

Acknowledgments

The authors appreciate greatly the comments of Dr. Shang-Ping Xie (editor) and two anonymous reviewers. This work is supported by the National Key Programme of China for Developing Basic Science (2004CB418300).

REFERENCES

REFERENCES
An
,
S-I.
, and
B.
Wang
,
2001
:
Mechanisms of locking of the El Niño and La Niña mature phases to boreal winter.
J. Climate
,
14
,
2164
2176
.
An
,
S-I.
,
F-F.
Jin
, and
I-S.
Kang
,
1999
:
The role of zonal advection feedback in phase transition and growth of ENSO in the Cane-Zebiak model.
J. Meteor. Soc. Japan
,
77
,
1151
1160
.
Battisti
,
D. S.
,
1988
:
Dynamics and thermodynamics of a warming event in a coupled tropical atmosphere–ocean model.
J. Atmos. Sci.
,
45
,
2889
2919
.
Battisti
,
D. S.
, and
A. C.
Hirst
,
1989
:
Interannual variability in a tropical atmosphere–ocean model: Influence of the basic state, ocean geometry, and nonlinearity.
J. Atmos. Sci.
,
46
,
1687
1712
.
Chang
,
P.
,
B.
Wang
,
T.
Li
, and
L.
Ji
,
1994
:
Interaction between the seasonal cycle and the Southern Oscillation: Frequency entrainment and chaos in an intermediate coupled ocean-atmosphere model.
Geophys. Res. Lett.
,
21
,
2817
2820
.
Chang
,
P.
,
L.
Ji
,
B.
Wang
, and
T.
Li
,
1995
:
Interactions between the seasonal cycle and El Niño-Southern Oscillation in an intermediate coupled ocean-atmosphere model.
J. Atmos. Sci.
,
52
,
2353
2372
.
Dewitt
,
B.
, and
C.
Perigaud
,
1996
:
El Niño-La Niña events simulated with Cane and Zebiak’s model and observed with satellite or in situ data. Part II: Model forced with observations.
J. Climate
,
9
,
1188
1207
.
Galanti
,
E.
, and
E.
Tziperman
,
2000
:
ENSO’s phase locking to the seasonal cycle in the fast-SST, fast-wave, and mixed-mode regimes.
J. Atmos. Sci.
,
57
,
2936
2950
.
Galanti
,
E.
,
E.
Tziperman
,
M.
Harrison
,
A.
Rosati
,
R.
Giering
, and
Z.
Sirkes
,
2002
:
The equatorial thermocline outcropping—A seasonal control on the tropical Pacific ocean–atmosphere instability strength.
J. Climate
,
15
,
2721
2739
.
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
.
Guilyardi
,
E.
,
P.
Delecluse
,
S.
Gualdi
, and
A.
Navarra
,
2003
:
Mechanisms for ENSO phase change in a coupled GCM.
J. Climate
,
16
,
1141
1158
.
Harrison
,
D. E.
, and
G. A.
Vecchi
,
1999
:
On the termination of El Niño.
Geophys. Res. Lett.
,
26
,
1593
1596
.
Hirst
,
A. C.
,
1986
:
Unstable and damped equatorial modes in simple coupled ocean-atmosphere models.
J. Atmos. Sci.
,
43
,
606
630
.
Ji
,
M.
,
A.
Leetmaa
, and
J.
Derber
,
1995
:
An ocean analysis system for seasonal to interannual climate studies.
Mon. Wea. Rev.
,
123
,
460
481
.
Jin
,
F-F.
,
1996
:
Tropical ocean-atmosphere interaction, the Pacific cold tongue, and the El Niño-Southern Oscillation.
Science
,
274
,
76
78
.
Jin
,
F-F.
,
1997a
:
An equatorial ocean recharge paradigm for ENSO. Part I: Conceptual model.
J. Atmos. Sci.
,
54
,
811
829
.
Jin
,
F-F.
,
1997b
:
An equatorial ocean recharge paradigm for ENSO. Part II: A stripped-down coupled model.
J. Atmos. Sci.
,
54
,
830
847
.
Jin
,
F-F.
, and
S-I.
An
,
1999
:
Thermocline and zonal advective feedbacks within the equatorial ocean recharge oscillator model for ENSO.
Geophys. Res. Lett.
,
26
,
2989
2992
.
Jin
,
F-F.
,
D. J.
Neelin
, and
M.
Ghil
,
1994
:
El Niño on the devil’s staircase: Annual subharmonic steps to chaos.
Science
,
264
,
70
72
.
Kang
,
I-S.
,
S-I.
An
, and
F-F.
Jin
,
2001
:
A systematic approximation of the SST anomaly equation for ENSO.
J. Meteor. Soc. Japan
,
79
,
1
10
.
Latif
,
M.
, and
Coauthors
,
2001
:
ENSIP: The El Niño simulation intercomparison project.
Climate Dyn.
,
18
,
255
276
.
Lengaigne
,
M.
,
J-P.
Boulanger
,
C.
Menkes
, and
H.
Spencer
,
2006
:
Influence of the seasonal cycle on the termination of El Niño events in a coupled general circulation model.
J. Climate
,
19
,
1850
1868
.
Li
,
T.
,
1997
:
Phase transition of the El Niño-Southern Oscillation: A stationary SST mode.
J. Atmos. Sci.
,
54
,
2872
2887
.
Li
,
T.
, and
T. F.
Hogan
,
1999
:
The role of the annual-mean climate on seasonal and interannual variability of the tropical Pacific in a coupled GCM.
J. Climate
,
12
,
780
792
.
Mantua
,
N. J.
, and
D. S.
Battisti
,
1995
:
Aperiodic variability in the Zebiak-Cane coupled ocean-atmosphere model: Air-sea interactions in the western equatorial Pacific.
J. Climate
,
8
,
2897
2927
.
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
.
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
.
Neelin
,
J. D.
,
F-F.
Jin
, and
H-H.
Syu
,
2000
:
Variations in ENSO phase locking.
J. Climate
,
13
,
2570
2590
.
Philander
,
S. G.
,
1983
:
El Niño-Southern Oscillation phenomena.
Nature
,
302
,
295
302
.
Rasmusson
,
E.
, and
T.
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
.
Spencer
,
H.
,
2004
:
Role of the atmosphere in seasonal phase locking of El Niño.
Geophys. Res. Lett.
,
31
.
L24104, doi:10.1029/2004GL021619
.
Suarez
,
M. J.
, and
P. S.
Schopf
,
1988
:
A delayed action oscillator for ENSO.
J. Atmos. Sci.
,
45
,
3283
3287
.
Tziperman
,
E.
,
L.
Stone
,
M.
Cane
, and
H.
Jarosh
,
1994
:
El Niño chaos: Overlapping of resonance between the seasonal cycle and the Pacific ocean-atmosphere oscillator.
Science
,
264
,
72
74
.
Tziperman
,
E.
,
M. A.
Cane
, and
S. E.
Zebiak
,
1995
:
Irregularity and locking to the seasonal cycle in an ENSO prediction model as explained by the quasi-periodicity route to chaos.
J. Atmos. Sci.
,
52
,
293
306
.
Tziperman
,
E.
,
S. E.
Zebiak
, and
M. A.
Cane
,
1997
:
Mechanisms of seasonal–ENSO interaction.
J. Atmos. Sci.
,
54
,
61
71
.
Tziperman
,
E.
,
M. A.
Cane
,
S. E.
Zebiak
,
Y.
Xue
, and
B.
Blumenthal
,
1998
:
Locking of El Niño’s peak time to the end of the calendar year in the delayed oscillator picture of ENSO.
J. Climate
,
11
,
2191
2199
.
Vecchi
,
G. A.
,
2006
:
The termination of the 1997–98 El Niño. Part II: Mechanisms of atmospheric change.
J. Climate
,
19
,
2647
2664
.
Vecchi
,
G. A.
, and
D. E.
Harrison
,
2003
:
On the termination of the 2002–03 El Niño event.
Geophys. Res. Lett.
,
30
.
1964, doi:10.1029/2003GL017564
.
Vecchi
,
G. A.
, and
D. E.
Harrison
,
2006
:
The termination of the 1997–98 El Niño. Part I: Mechanisms of oceanic change.
J. Climate
,
19
,
2633
2646
.
Wang
,
B.
,
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
.
Wang
,
B.
,
R.
Wu
, and
R.
Lukas
,
2000
:
Annual adjustment of the thermocline in the tropical Pacific Ocean.
J. Climate
,
13
,
596
616
.
Weisberg
,
R.
, and
C.
Wang
,
1997
:
A western Pacific oscillator paradigm for the El Niño-Southern Oscillation.
Geophys. Res. Lett.
,
24
,
779
782
.
Xie
,
S-P.
,
1995
:
Interaction between the annual and interannual variations in the equatorial Pacific.
J. Phys. Oceanogr.
,
25
,
1930
1941
.
Zebiak
,
S. E.
, and
M. A.
Cane
,
1987
:
A model El Niño–Southern Oscillation.
Mon. Wea. Rev.
,
115
,
2262
2278
.
Zhang
,
R-H.
,
S. E.
Zebiak
,
R.
Kleeman
, and
N.
Keenlyside
,
2003
:
A new intermediate coupled model for El Niño simulation and prediction.
Geophys. Res. Lett.
,
30
.
2012, doi:10.1029/2003GL018010
.
Zhang
,
R-H.
,
R.
Kleeman
,
S. E.
Zebiak
,
N.
Keenlyside
, and
S.
Raynaud
,
2005
:
An empirical parameterization of subsurface entrainment temperature for improved SST anomaly simulations in an intermediate ocean model.
J. Climate
,
18
,
350
371
.

Footnotes

Corresponding author address: Bangliang Yan, Institute of Atmospheric Physics, Chinese Academy of Sciences, P.O. Box 2718, Beijing 100080, China. Email: ybl@sgi50s.iap.ac.cn