1 Introduction

The El Niño-Southern Oscillation (ENSO) is the most prominent interannual mode in the tropics (Philander 1990; Trenberth et al. 1998; Wallace et al. 1998; Latif et al. 1998; Alexander et al. 2002). As the positive and negative aspects of ENSO events, El Niño exhibits a positive sea surface temperature anomaly (SSTA) in the equatorial central and eastern Pacific (CP and EP), whereas there is a negative SSTA occurred in the La Niña mature winter. The observational fact and model experiment results have shown that La Niña is not a mirror image of El Niño, and asymmetric features have been clearly seen in both amplitude (Burgers and Stephenson 1999; An and Jin 2004; Su et al. 2010; Chen et al. 2015) and evolution (Kessler 2002; Larkin and Harrison 2002; Okumura and Deser 2010; Dommenget et al. 2013; McGregor et al. 2013; Chen et al. 2016; Chen and Li 2018) between El Niño and La Niña.

In our previous study (Chen et al. 2016), we conducted an oceanic mixed-layer heat budget analysis to understand the cause of El Nino and La Nina evolution asymmetry. It was found that different from many previous studies that emphasized only the role of wind stress asymmetry, both the dynamic (i.e., wind stress asymmetry) and thermodynamic (i.e., surface shortwave radiative and latent heat flux asymmetry) processes are important in causing the El Niño-La Niña evolution asymmetry. During El Niño mature winter, positive SSTA in EP induces an anomalous anticyclone over the Western North Pacific (hereafter WNPAC) through local air-sea thermodynamic feedback (Wang et al. 2000) and atmospheric moist enthalpy advection (Wu et al. 2017a, b; Li et al. 2017). Easterly anomalies south of the WNPAC trigger upwelling Kelvin waves propagating eastward, which can help El Niño decay fast. In contrast, anomalous cyclone (WNPC) is weaker and tends to shift west of the Philippine in La Niña mature phase (Wu et al. 2010). Thus, westerly anomalies south of the WNPC are weaker and shift westward, which lead to a slow decay. It is worth noting that, due to the asymmetry of SSTA pattern between El Niño and La Niña, the negative cloud-SST and evaporation-SST feedbacks becomes stronger in El Niño than in La Niña. Due to the distinctive decaying rates between El Nino and La Nina composites, El Niño decays much faster and transitions into a cold anomaly by summer of the second year, whereas La Niña decays slowly and remains a negative SSTA. As both Bjerknes thermocline feedback and zonal adjective feedback are strongest in northern fall (Li 1997), this season-dependent coupled instability leads to the growth of the remaining weak cold SSTA in both El Nino and La Nina composite and develops into a La Nina episode by the end of the second year.

However, EP and CP El Niño were not separate and were summed together as a single El Nino group in Chen et al. (2016). A natural question is what is the difference between EP and CP El Nino? Figure 1 shows the time evolutions of equatorial SSTA among EP El Niño, CP El Niño, and La Niña. The result shows that EP El Niño has a larger SSTA amplitude (more than 2 °C) with a center located in 120°W (Fig. 1a), while both CP El Niño and La Niña have weaker SSTA intensity (about 1 °C and − 1 °C, respectively) and tend to shift westward (around 150°W) in their mature phases (Fig. 1b, c). In the second year, both EP and CP El Niño have fast decaying processes and transform into La Niña episode, but La Niña has a slow decay and remains the negative SSTA throughout the year.

Fig. 1
figure 1

Temporal evolutions of sea surface temperature anomaly (SSTA) for composite a EP El Niño (including 1982, 1997, and 2006 El Niños), b CP El Niño (including 1991, 1994, 2002, 2004, and 2009 El Niños), and c La Niña (including 1983/1984, 1995/1996, 1998/1999, 2007/2008, and 2010/2011 La Niñas) along the equator (5°N–5°S; unit: °C)

Full size image

Thus, an important dynamic question that needs to be addressed is why CP El Niño also has a rapid decay and transforms into La Niña by the end of the second year, even though wind stress and heat flux anomaly responses could be very different between EP and CP El Niño. Given that the location of SSTA center during CP El Nino shifts even further westward compared to La Nina, the heat flux anomaly asymmetry between CP El Niño La Niña should be minor. Therefore, it is likely that the ocean dynamics associated with the wind stress forcing may be important in contributing to the fast decaying of CP El Niño after its peak phase.

The objective of this study is to reveal key physical processes that give rise to the distinctive evolutions among EP El Niño, CP El Niño, and La Niña. The remaining part of this paper is organized as follows. A description of datasets and oceanic mixed-layer heat budget analysis is given in Sect. 2. Sections 3 and 4 discuss the dynamic and thermodynamic processes in causing distinctive evolutions among EP El Niño, CP El Niño, and La Niña. Asymmetric wind effect between CP El Niño and La Niña is given in Sect. 5. Finally, a conclusion and discussion are given in the last section.

2 Datasets and methods

In this study, the primary ocean reanalysis datasets are derived from Simple Ocean Data Assimilation version 2.1.6 (SODA v2.1.6; Carton and Giese 2008), and NCEP Global Ocean Data Assimilation System (GODAS; Saha et al. 2006). The surface heat flux and atmospheric field datasets are mainly from the WHOI Objectively Analyzed Air-Sea Fluxes (OAFlux; Yu et al. 2008), and the NCEP reanalysis version 2 (NCEP v2; Kanamitsu et al. 2002). The observed SST, precipitation, and OLR datasets are from the Extended Reconstructed Sea Surface Temperature version 3b (ERSST v3b; Smith et al. 2008), the Climate Prediction Center (CPC) Merged Analysis of Precipitation (CMAP; Xie and Arkin 1997), and the National Oceanic and Atmospheric Administration (NOAA; Liebmann and Smith 1996), respectively.

Following Li et al. (2002), we conduct an oceanic mixed-layer heat budget analysis to understand the fundamental processes in causing the distinctive decaying rates among EP El Niño, CP El Niño and La Niña. The mixed-layer temperature anomaly tendency equation may be written as

$$\frac{\partial \mathbf{T}\mathbf{^{\prime}}}{\partial \mathbf{t}}=-\left({\mathbf{u}}^{\mathbf{^{\prime}}}\partial \overline{\mathbf{T}}/\partial \mathbf{x}+\overline{\mathbf{u}}\partial {\mathbf{T}}^{\mathbf{^{\prime}}}/\partial \mathbf{x}+{\mathbf{u}}^{\mathbf{^{\prime}}}\partial {\mathbf{T}}^{\mathbf{^{\prime}}}/\partial \mathbf{x}\right)-\left({\mathbf{v}}^{\mathbf{^{\prime}}}\partial \overline{\mathbf{T}}/\partial \mathbf{y}+\overline{\mathbf{v}}\partial {\mathbf{T}}^{\mathbf{^{\prime}}}/\partial \mathbf{y}+{\mathbf{v}}^{\mathbf{^{\prime}}}\partial {\mathbf{T}}^{\mathbf{^{\prime}}}/\partial \mathbf{y}\right)-({\mathbf{w}}^{\mathbf{^{\prime}}}\partial \overline{\mathbf{T}}/\partial \mathbf{z}+\overline{\mathbf{w}}\partial {\mathbf{T}}^{\mathbf{^{\prime}}}/\partial \mathbf{z}+{\mathbf{w}}^{\mathbf{^{\prime}}}\partial {\mathbf{T}}^{\mathbf{^{\prime}}}/\partial \mathbf{z})+\frac{{{\mathbf{Q}}^{\mathbf{^{\prime}}}}_{\mathbf{n}\mathbf{e}\mathbf{t}}}{{\varvec{\uprho}}{\mathbf{C}}_{\mathbf{P}}\mathbf{H}}+\mathbf{R}$$

The detailed description of tendency equation and parameters can be seen from Chen et al. (2016). Figure 2 shows the heat budget analysis result during the decaying phase (January–June in the second year) of EP El Niño, CP El Niño and La Niña averaged in the equatorial CP and EP region (5°N–5°S, 180°–80°W). It is seen that EP El Niño has the largest decaying rate (– 0.37 °C/month), while CP El Niño has a negative temperature anomaly tendency weaker than EP El Niño (– 0.22 °C/month), and La Niña has the weakest decaying rate in its decaying phase (0.13 °C/month). Both dynamic and thermodynamic processes are attributed to the distinctive decaying rates. In the dynamic terms, the major contributor are both the anomalous zonal advection term and the vertical entrainment term, while in the thermodynamic terms, the major contributors are both shortwave radiation term and latent heat flux term. In the following sections, we will discuss the dynamic and thermodynamic processes, respectively.

Fig. 2
figure 2

Composite mixed-layer temperature anomaly (MLTA) heat budget analysis during the decaying phase (January–June in the second year) of a EP El Niño, b CP El Niño and c La Niña averaged in the equatorial eastern Pacific (EP; 180°–80°W, 5°S–5°N; °C/month). The values are based on the ensemble average of two ocean reanalysis datasets (SODA v2.1.6 and GODAS), and two surface heat flux products (OAFlux and NCEP R2)

Full size image

3 Dynamic processes in causing distinctive evolutions among EP El Niño, CP El Niño, and La Niña

The oceanic mixed-layer heat budget analysis result illustrates that, in the dynamic processes, it is the propagation of equatorial waves triggered by the wind field that induced the distinctive decaying rates. Thus, the distinctive wind responses over the WNP to ENSO forcing may exert a key effect on the zonal oceanic current difference among EP El Niño, CP El Niño and La Niña in their decaying phases. Figure 3 shows the composite 850 hPa wind, SST and precipitation anomalies in the mature winter (DJF). It can be seen that there are twin cyclonic couplets associated with warm SSTA (positive convective heating anomalies) in the equatorial EP, and a large-scale anomalous WNPAC, with its eastern edge can be extended to the dateline (Fig. 3a, b). Easterly wind anomalies south of the anticyclone can stimulate oceanic upwelling Kelvin waves propagating eastward, thus lead to a fast decay of warm SSTA in the equatorial EP. In the composite CP El Niño mature winter, an off-equatorial anomalous anticyclone also occurred west of 150°E, which is only weaker than that in the composite EP El Niño (Fig. 3c, d). To the south of the anticyclone, there are still anomalous easterly winds along the equatorial WP, which can lead to a fast decaying phase in the following decaying spring. Due to the anticyclonic circulation in the composite CP El Niño has a smaller size and weaker intensity than that in the composite EP El Niño, the anomalous zonal advection is a bit weaker in the decaying phase of CP El Niño (Fig. 2a, b). In contrast, an anomalous cyclone associated with La Niña forcing tends to shift more westward compared to the anticyclone in the composite El Niño, which has a center located in the South China Sea (Fig. 3e, f). It implies that there are easterly wind anomalies, rather than westerly wind anomalies, occurred along the equatorial WP. Due to the lack of the anomalous westerly wind forcing, a much weaker damping rate appeared in the decaying phase of composite La Niña events (Fig. 2c). Furthermore, the distinctive features of zonal wind along the equatorial WP are still existed in the following decaying spring (figure not shown), thus the distinctive damping rates will be maintained in the decaying phase of EP El Niño, CP El Niño and La Niña composite.

Fig. 3
figure 3

Composite (left panel) SSTA (shading; unit: °C) and 850 hPa wind anomalies (vectors; unit: m/s), and (right panel) precipitation (shading; unit: mm/day) and stream function anomalies (contours; unit: 106 m2/s) for a, b EP El Niño, c, d CP El Niño, and e, f La Niña in the mature winter (DJF)

Full size image

For the different wind forcing over the WNP, the thermocline depth anomalies exhibited different evolution features between the EP El Niño, CP El Niño and La Niña in the second year (Fig. 4). For the EP El Niño composite (left panel of Fig. 4), there is a large negative thermocline depth anomaly in the tropical western and central Pacific, and its eastern edge has reached 130°W in mature winter (Fig. 4a). In the following decaying spring (Fig. 4d, g), the strong negative signal continues strengthening as its center propagates eastward. As a result, positive thermocline depth anomaly in the equatorial EP is replaced by negative one in the second half of the second year (Fig. 4j, m). For the CP El Niño composite (middle panel of Fig. 4), although the negative thermocline depth anomaly is weaker than that in the EP El Niño, it can penetrate into the equatorial EP in later of the second year as well (Fig. 4n). Therefore, CP El Niño translates in to a La Niña episode. In contrast, due to the lack of the westerly anomaly forcing, the positive thermocline anomaly remains in the equatorial WP (west of 180°) in the La Niña composite (right panel of Fig. 4), and the negative one cannot change the sign in the equatorial EP throughout the year + 1. As a result, La Niña has a much slower decaying process than both EP and CP El Niño.

Fig. 4
figure 4

Thermocline depth anomalies (shading; unit: m) for composite (left panel) EP El Niño, (middle panel) CP El Niño, and (right panel) La Niña in the second year

Full size image

Figure 5 illustrates the composite equatorial zonal ocean currents and temperature anomalies for EP El Niño, CP El Niño, and La Niña in the year + 1. It can be clearly seen that, both EP and CP El Niño have stronger westward geostrophic currents (negative thermocline depth anomaly induced) at the upper ocean in the earlier half of year + 1. The westward currents anomalies advect climatological mean SST, which lead to an anomalous cold zonal advection. As a result, the cold subsurface water can easily spread into the equatorial EP in the end. In contrast, for the La Niña composite (right panel of Fig. 5), the eastward geostrophic currents are weaker due to lack of westerly anomalies forcing over the equatorial WP. Therefore, the anomalous warm subsurface water is locked in the WP, while La Niña decays slowly and keeps its sign in the end of the second year.

Fig. 5
figure 5

Composite equatorial (2°S–2°N) ocean anomalous zonal currents (vectors; unit: m/s) and temperature anomalies (shading; unit: °C) for (left panel) EP El Niño, (middle panel) CP El Niño, and (right panel) La Niña in the second year

Full size image

4 Thermodynamic processes in causing distinctive evolutions among EP El Niño, CP El Niño, and La Niña

In additional to the wind effect, the local heat flux forcing also has an effect on the SSTA change in the equatorial EP. Previous study illustrated that the asymmetric thermodynamic forcing is as important as the asymmetric dynamic forcing during ENSO decaying phase, and the major contributors are shortwave radiation and latent heat flux (Chen et al. 2016).

In warm oceans, a modest SSTA may induce deep convection in situ. The increased clouds tend to reduce the downward shortwave radiation and thus decrease the SST, which lead to a negative cloud-radiation-SST feedback between the atmosphere and ocean (Li et al. 2003). The composite OLR anomalies during the decaying phases of EP El Niño, CP El Niño, and La Niña are shown in the left panel of Fig. 6. In the EP El Niño composite (Fig. 6a), negative OLR anomaly occurs in the equatorial central and eastern Pacific, with a center located in 140°W. In contrast, due to the westward shift of SSTA in both CP El Niño and La Niña, the OLR anomalies tend to shift westward with centers around 180° (Fig. 6c, e). Note that the amplitude of SSTA in EP El Niño is two times as large as that in CP El Niño and La Niña (Fig. 1), thus a larger OLR anomaly occurred during EP El Niño decaying phase (Fig. 6a). For the EP El Niño composite, the negative shortwave radiation (or OLR) forcing can affect the equatorial EP SSTA in situ (180°–80°W), whereas it tends to shift more westward (150°E–150°W) for CP El Niño and La Niña composite. As a result, in the heat budget area (180°–80°W, 5°N–5°S), there is a large damping term of shortwave radiation in EP El Niño (– 0.14 °C/month), but much weaker in CP El Niño and La Niña (– 0.04 and 0.04 °C/month, respectively).

Fig. 6
figure 6

Composite patterns of anomalous (left panel) OLR (shading; unit: W/m2), sea surface temperature (contours; unit: °C) and (right panel) sea-air specific humidity difference (unit: g/kg) fields during the decaying phase (Jan–Jun in year + 1) of a, b EP El Niño, c, d CP El Niño and e, f La Niña

Full size image

The asymmetry of the anomalous latent heat fluxes also has an impact on the distinctive decaying rates of EP El Niño, CP El Niño, and La Niña. And it is mainly attributed to the anomalous sea-air specific humidity difference (qs–qa; Chen et al. 2016). As the qs–qa field is proportional to local SST (Li and Wang 1994), a larger positive qs–qa anomaly is located in the equatorial EP during the decaying phase of EP El Niño. Meanwhile, the weaker qs–qa anomalies tend to shift westward with centers around 180° in both CP El Niño and La Niña, in a way similar to the cloud-radiation-SST feedback (right panel of Fig. 6). As a result, the surface latent heat term is larger in EP El Niño (– 0.2 °C/month), but much weaker in CP El Niño and La Niña (– 0.08 and 0.06 °C/month, respectively).

5 Asymmetric wind effect between CP El Niño and La Niña

During the mature phases of ENSO events, EP El Niño has stronger amplitude than CP El Niño due to farther eastward SSTA. As a result, there is a large-scale WNPAC which leads to a fast decay of EP El Niño. Meanwhile, farther eastward SSTA can induce stronger local negative thermodynamic feedbacks (shortwave radiation and latent heat flux), which has a great impact on the fast decay of EP El Niño, either. On the contrary, both CP El Niño and La Niña have SSTA center located in the equatorial CP (around 150°W; Fig. 1b, c), thus thermodynamic feedbacks are comparable (Fig. 6c–f) and have little influence on the decaying process (Fig. 2b, c). As a result, the asymmetric evolution between CP El Niño and La Niña is primarily attributed to the dynamic processes. In the dynamic processes, the propagation of equatorial waves triggered by the wind field is the key factor. In the CP El Niño, the WNP is dominated by anticyclone, with its eastern edge located in 150°E. The easterly anomalies south of the anticyclone induce eastward upwelling Kelvin waves in equatorial WP, and the negative thermocline anomalies can penetrate into the EP rapidly. The eastward anomalous cold advection in the upper ocean can cool the original warm SSTA in the CP El Niño (middle panel of Figs. 2, 3, 4, 5). Otherwise, the anomalous WNPC in La Niña has a westward shift to the South China Sea, with its eastern edge located in 130°E, which has a westward shift of 20 longitude degrees compared to CP El Niño. As a result, it is dominated by anomalous anticyclone over the WNP. Due to lack of the anomalous westerly forcing, the negative thermocline anomalies in western Pacific is hard to extend eastward. There is a weaker warm ocean advection, and La Niña decays slowly (bottom panel of Figs. 2, 3; right panel of Figs. 4, 5).

In order to investigate the cause of atmospheric circulation asymmetry between CP El Niño and La Niña, we calculate the symmetric and asymmetric components of the 850 hPa wind field, SSTA and precipitation anomalies (Fig. 7). Considering that SSTA in CP El Niño is slightly stronger than that of La Niña, the normalized components calculated according to Niño-3.4 SSTA amplitude are given. It can be seen from the normalized symmetric component (Fig. 7a, b) that, during the mature winter of CP El Niño (La Niña), the positive (negative) SSTA center is located at 150°W. As result, the positive (negative) precipitation anomaly has a center west of the dateline. In response to the convection, there are twin cyclone (anticyclone) couplets in the off-equatorial CP region. The anomalous descending (ascending) motion region is over the Maritime Continent and WNP, where anomalous WNPAC (WNPC) can be clearly seen with its east edge reaches to 140°E. For the normalized asymmetric component (Fig. 7c, d), both CP El Niño and La Niña have zonal dipole patterns of SSTA along the equatorial WP and CP. Correspondingly, the anomalous convective heating also has an east–west dipole mode. Positive precipitation anomaly is located in Maritime Continent, while the negative one occurs around 160°E. The WNP region is dominated by anomalous anticyclone, and it can affect whole WNP, ranging from 120° E to 180°. It seems that the anomalous anticyclone is induced by the equatorial dipole heating pattern. We use anomaly GCM to further verify the heating effect as follows.

Fig. 7
figure 7

Composite a, b normalized symmetric and c, d asymmetric components of (left panel) SSTA (shading; unit: °C) and 850 hPa wind anomalies (vectors; unit: m/s), and (right panel) precipitation (shading; unit: mm/day) and stream function anomalies (contours; unit: 0.5 × 106 m2/s) between CP El Niño and La Niña in the mature winter (DJF). The symmetric component is obtained by half of difference of them, while the asymmetric component estimated by half of sum of them. The normalized components are calculated according to Niño-3.4 SSTA amplitude

Full size image

To study how the atmospheric wind responds to the symmetric and asymmetric precipitation anomalies under ENSO mature winter, an anomaly atmospheric GCM was used. This model was developed based on the Geophysical Fluid Dynamics Laboratory (GFDL) global spectral dry AGCM (Held and Suarez 1994). The detailed description of anomaly GCM can be referred to Jiang and Li (2005) and Li et al. (2006).

In this study, the observed winter (DJF) mean state is specified as the model background state. The symmetric, asymmetric and total heating anomaly calculated based on observed precipitation anomaly field during ENSO mature winter are specified in the model. An idealized vertical profile with a maximum center in middle troposphere is specified for the diabatic heating anomaly.

The 850 hPa wind (vectors) and stream function (contours) anomalies simulated by anomaly AGCM respond to the normalized symmetric components of CP El Niño and La Niña diabatic heating (shading) are given (Fig. 8a, b). It is seen that the anomalous WNPAC and WNPC are symmetric. The eastern edge of WNPAC (WNPC) reaches to 140°E, which is similar to that in the observational fact (Fig. 7b). Figure 8c shows the wind field respond to the asymmetric component of heating, and there is an anomalous anticyclone over the whole WNP region (120°E–180°). The anomalous anticyclonic circulation enhances the intensity and extends the size of the WNPAC, while weakens the WNPC and leads to a westward shift. As a result, in the real forcing experiments (including both symmetric and asymmetric components), there is a stronger anomalous anticyclone over the WNP with its eastern edge extended to 150°E for the CP El Niño (Fig. 8d), while the anomalous cyclone is weaker with its eastern edge has a westward retreat for more than 20 degrees of longitude in the La Niña composite (Fig. 8e). And the simulation results are consistent with the observations (Fig. 3d, f).

Fig. 8
figure 8

The 850 hPa wind (vectors; unit: m/s) and stream function (contours; unit: 105 m2 s−1) anomalies simulated by anomaly AGCM respond to the (top panel) normalized symmetric, (middle panel) asymmetric, and (bottom panel) original precipitation field during ENSO mature winter (DJF). The diabatic heating rates (shading; unit: K day−1) are also shown

Full size image

6 Conclusion

The observational fact illustrated that during the mature winter of ENSO events, EP El Niño has larger SSTA amplitude with a center located in 120°W, while both CP El Niño and La Niña have weaker intensity of SSTA and shifts more westward to 150°W. The ocean mixed-layer temperature anomaly heat budget analysis result shows that EP El Niño has the largest decaying rate, while CP El Niño also has a negative tendency weaker than EP El Niño, and La Niña has the weakest damping rate. For the EP El Niño, both dynamic and thermodynamic processes are contributed to the fast decay. The former is related to wind response (anticyclone) over the WNP, whereas the latter is associated with negative cloud-radiation-SST and evaporation-SST feedbacks. Due to the westward shift of weaker SSTA, the thermodynamic damping effect is weaker in both CP El Niño and La Niña. It is the wind induced (WNPAC) equatorial wave effect (eastward propagating upwelling Kelvin waves) that contributes to the fast decaying process of CP El Niño. For the La Niña composite, due to lack of anomalous westerly forcing south of the WNPC, there is a slow decay and SSTA remains the negative sign in summer of the second year. Because both Bjerknes thermocline feedback and zonal advective feedback are strongest in northern fall, thus the weak SSTA starts to develop and reaches a peak in the following boreal winter. Both EP and CP El Niño transform into La Niña episode, but La Niña re-develops into another cold phase in the end of the second year.

By conducting a series of experiments using dry AGCM, the asymmetric wind effect between CP El Niño and La Niña is investigated. The model simulation result shows that it is the asymmetric component of convective heating that contributes to the wind field asymmetry. In the mature winter, the asymmetric component of SSTA is characterized by a zonal dipole mode, with positive anomaly over the WNP and negative one around the dateline along the equator. As a result, the asymmetric convective heating also represents a zonal dipole mode, which leads to an anomalous anticyclone over the WNP. The anomalous anticyclonic circulation enhances the intensity and extends the size of the WNPAC, while weakens the WNPC and leads to a westward shift. Therefore, the composite CP El Niño decays fast and transforms into a cold phase, but no phase change occurs in the La Niña composite.