Strong Extratropical Impact on Observed ENSO

Tianying Liu aDepartment of Atmospheric and Oceanic Sciences, Peking University, Beijing, China
bLaoshan Laboratory, Qingdao, China

Search for other papers by Tianying Liu in
Current site
Google Scholar
PubMed
Close
https://orcid.org/0000-0002-3780-0015
,
Zhengyu Liu cAtmospheric Science Program, Department of Geography, The Ohio State University, Columbus, Ohio

Search for other papers by Zhengyu Liu in
Current site
Google Scholar
PubMed
Close
,
Yuchu Zhao cAtmospheric Science Program, Department of Geography, The Ohio State University, Columbus, Ohio

Search for other papers by Yuchu Zhao in
Current site
Google Scholar
PubMed
Close
, and
Shaoqing Zhang bLaoshan Laboratory, Qingdao, China
dKey Laboratory of Physical Oceanography, College of Oceanic and Atmospheric Sciences, Ocean University of China, Qingdao, China

Search for other papers by Shaoqing Zhang in
Current site
Google Scholar
PubMed
Close
Open access

Abstract

Previous studies have indicated that the extratropics can influence ENSO via specific processes. However, it is still unclear to what extent ENSO is influenced by the extratropics in observation. Now we assess this issue by applying the regional data assimilation (RDA) approach in an advanced model, the GFDL CM2.1. Our study confirms a strong extratropical impact on observed ENSO. Quantitatively, the extratropical atmospheric variability poleward of 20° explains 56% of the observed variance of ENSO and greatly influences ∼67% of observed El Niño events during 1969–2008. This extratropical impact is still significant even as far as poleward of 30°. Furthermore, the impact from the southern extratropics is slightly stronger than that from the northern extratropics, partly caused by the Pacific ITCZ location north of the equator and different mixed-layer depth along the northern Pacific meridional mode (NPMM) and the southern Pacific meridional mode (SPMM). Our study further shows that all of three super El Niño events, those in 1972/73, 1982/83, and 1997/98, are influenced greatly by both hemispheric extratropics, with NPMM and SPMM interfering constructively, while most weak and moderate El Niño events are triggered by only one hemispheric extratropics, with NPMM and SPMM interfering destructively. Besides the extratropical Pacific influence on ENSO via NPMM/SPMM, the extratropics also has a potential impact on ENSO by influencing other tropical oceans and then by interbasin interactions.

© 2024 American Meteorological Society. This published article is licensed under the terms of the default AMS reuse license. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding authors: Tianying Liu, liuty109@pku.edu.cn; Zhengyu Liu, liu.7022@osu.edu

Abstract

Previous studies have indicated that the extratropics can influence ENSO via specific processes. However, it is still unclear to what extent ENSO is influenced by the extratropics in observation. Now we assess this issue by applying the regional data assimilation (RDA) approach in an advanced model, the GFDL CM2.1. Our study confirms a strong extratropical impact on observed ENSO. Quantitatively, the extratropical atmospheric variability poleward of 20° explains 56% of the observed variance of ENSO and greatly influences ∼67% of observed El Niño events during 1969–2008. This extratropical impact is still significant even as far as poleward of 30°. Furthermore, the impact from the southern extratropics is slightly stronger than that from the northern extratropics, partly caused by the Pacific ITCZ location north of the equator and different mixed-layer depth along the northern Pacific meridional mode (NPMM) and the southern Pacific meridional mode (SPMM). Our study further shows that all of three super El Niño events, those in 1972/73, 1982/83, and 1997/98, are influenced greatly by both hemispheric extratropics, with NPMM and SPMM interfering constructively, while most weak and moderate El Niño events are triggered by only one hemispheric extratropics, with NPMM and SPMM interfering destructively. Besides the extratropical Pacific influence on ENSO via NPMM/SPMM, the extratropics also has a potential impact on ENSO by influencing other tropical oceans and then by interbasin interactions.

© 2024 American Meteorological Society. This published article is licensed under the terms of the default AMS reuse license. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding authors: Tianying Liu, liuty109@pku.edu.cn; Zhengyu Liu, liu.7022@osu.edu

1. Introduction

El Niño–Southern Oscillation (ENSO) is the dominant interannual climate variability mode and exerts a significant influence over the globe via atmospheric teleconnections (Alexander et al. 2002; Liu and Alexander 2007). Earlier studies suggest that ENSO originates from tropical intrinsic coupled ocean–atmosphere processes (Cane and Zebiak 1985; Zebiak and Cane 1987), either as a self-sustained mode or as a damping mode forced by stochastic forcing (Penland and Sardeshmukh 1995; Jin et al. 2007; Newman et al. 2011). Critical in both theories are the Bjerknes positive ocean–atmosphere feedback (Bjerknes 1969) for the growth of ENSO and a delayed negative feedback for the phase reversal such as in the delayed oscillator model (Schopf and Suarez 1988; Battisti and Hirst 1989) and the recharge–discharge oscillator (RDO) model (Jin 1997a,b). Consistent with the RDO theory, the warm water volume (WWV), or the upper-ocean heat content, in the equatorial Pacific has been observed to lead the Niño-3.4 index by one-quarter of ENSO cycle, therefore serving as a precursor for ENSO (Meinen and McPhaden 2000; McPhaden 2003, 2012). However, the stochastic forcing, such as westerly/easterly wind bursts (WWB/EWB) (Fedorov 2002; Fedorov et al. 2015; Hu and Fedorov 2016) and the Madden–Julian oscillation (MJO) (McPhaden et al. 2006; Tang and Yu 2008), especially their low-frequency components (Capotondi et al. 2018), can excite oceanic equatorial Kelvin waves, thus triggering or disrupting the charging of WWV and thus the development of ENSO, explaining the irregular onset of ENSO events.

Later studies suggest that ENSO can also be triggered or disrupted by external forcing from the extratropics, notably via the North Pacific meridional mode (NPMM) (Vimont et al. 2003a,b; Chiang and Vimont 2004; Chang et al. 2007; Vimont et al. 2009). The NPMM is suggested to be excited by the southern node of the North Pacific Oscillation (NPO) (Rogers 1981; Linkin and Nigam 2008) in boreal winter and evolves southwestward from the subtropical northeastern Pacific to the equator in boreal spring via ocean–atmosphere coupling. The extratropical Pacific influence on ENSO via NPMM contains four pathways (Amaya 2019): the wind–evaporation–SST (WES) feedback that propagates the SST and surface wind anomalies to the equator in boreal spring (Liu and Xie 1994; Xie and Philander 1994; Vimont et al. 2009; Amaya et al. 2017), the trade wind charging (TWC) that adjusts the WWV by the NPMM-induced wind stress curl (Anderson et al. 2013; Anderson and Perez 2015), the summer deep convection (SDC) that causes a Gill-type response (Gill 1980) associated with ITCZ shifts in boreal summer (Vimont et al. 2001, 2003a; Amaya et al. 2019), and the Rossby wave reflection (RWR) that triggers ocean Rossby waves reflected as the equatorial Kelvin waves (Solomon et al. 2008; Alexander et al. 2010). Using a statistical analysis, Chang et al. (2007) suggests that over 70% of the observed ENSO events during 1958–2000 are preceded by positive NPMM in boreal spring, indicating NPMM’s potential role as a good ENSO predictor. However, ENSO forecast skill remains poor if NPMM is the sole predictor (Larson and Kirtman 2014, 2015), implying other precursors from both tropics and southern extratropics may work together with the NPMM to influence ENSO (Anderson 2007; Su et al. 2014; Min et al. 2015; Ding et al. 2017). Recently, a similar precursor parallel to NPMM has been identified in the southern extratropics. This so-called South Pacific meridional mode (SPMM) (Zhang et al. 2014; You and Furtado 2017, 2018) is suggested to be excited by the northern node of the South Pacific Oscillation (SPO) (You and Furtado 2017), and then propagates northwestward from subtropical southeastern Pacific to the equator via WES feedback. So far, however, this SPMM has not been studied extensively, and therefore its seasonality and triggering mechanisms for ENSO have remained less understood relative to NPMM. For example, it is still unclear whether the SPMM triggers ENSO by wind-driven ocean Kelvin waves as NPMM does or the SPMM just modulates the ENSO amplitude by thermal forcing (Larson et al. 2018). Besides, the relative contribution of the two PMMs to ENSO is unclear and is being debated (Zhang et al. 2014; Liguori and Di Lorenzo 2019). Some studies also suggest that NPMM and SPMM will contribute to different ENSO flavors, with NPMM and SPMM favoring the central Pacific (CP) and eastern Pacific (EP) ENSO (Yu and Kim 2011; Vimont et al. 2014; Zhang et al. 2014; Min et al. 2017; You and Furtado 2017), respectively. Other studies argue that there is little evidence of the preference of each PMM for ENSO flavor (Di Lorenzo et al. 2015; Ding et al. 2015).

Previous work leads to one robust but qualitative conclusion: ENSO can be influenced by extratropical processes. The next question is, quantitatively, to what extent the extratropics affects ENSO, especially in observation. This question is challenging because tropical and extratropical processes are closely coupled and therefore difficult to separate. So far, two approaches have been used in most previous studies, both attempting to remove ENSO impact on extratropics and therefore separating the extratropical impact on ENSO. The first is a statistical approach that can be applied to both observations and coupled model simulations, with a linear regression to remove the contemporaneous ENSO impact (Chiang and Vimont 2004; Chang et al. 2007; Larson and Kirtman 2014; You and Furtado 2018). However, ENSO impact is difficult to filter out cleanly due to its complex spatiotemporal and potential nonlinear characteristics. The second is a dynamic approach that can only be applied to model simulations, with ENSO variability completely filtered out by employing a slab-ocean model (Vimont et al. 2009; Zhang et al. 2014). However, the full extratropical influence on ENSO cannot be represented because of the complete suppression of ocean dynamics. Besides, this dynamic approach is based on highly idealized climate models and is therefore not proper for assessing extratropical impact on real-world ENSO events.

To quantitatively assess the extratropical impact on observed ENSO, Lu and Liu (2018) (hereinafter LU18) used the approach of Regional Data Assimilation (RDA) in a coupled general circulation model (CGCM). The RDA approach assimilates observational information of atmosphere and/or ocean in the extratropics, leaving the tropics fully coupled freely to generate ENSO under the condition of observed extratropical climate variability, thus leading to a quantitative assessment of the extratropical influence on observed ENSO. LU18 found a significant extratropical impact on observed ENSO, consistent with a test in the perfect-model scenario (Lu et al. 2017). However, the credibility of this RDA approach for real-world ENSO study relies heavily on the fidelity of the CGCM. This pilot study of LU18 is based on a CGCM (FOAM) with very coarse resolution (7.5° × 4.5° in the atmosphere; 2.8° × 1.4° in the ocean), leaving it of uncertain relevance to the real world. It is therefore highly desirable to further study the extratropical influence on observed ENSO by applying the RDA approach to a much more advanced model, the GFDL CM2.1. Similar to LU18, a pacemaker experiment has been performed in the GFDL CM2.1 (Amaya et al. 2019), only assimilating the observed northern Pacific SST poleward of 15°. However, their assimilation boundary has intruded into the tropics. Besides, their experiment does not include observed extratropical atmospheric variability, leaving the full extratropical impact unclear. Actually, LU18 shows similar results between experiments of the assimilation performed in the extratropical atmosphere and SST and those of the assimilation only performed in the extratropical atmosphere, indicating the important role of the extratropical atmospheric variability in ENSO. Therefore, we will further study the extratropical impact on observed ENSO by applying the RDA approach to GFDL CM2.1, with the focus on the role of extratropical atmospheric variability. Our study confirms the conclusion of LU18 of a great influence of the extratropical atmospheric variability on observed ENSO.

The paper is organized as follows. Section 2 introduces the experimental design and observed datasets. Section 3 quantitatively assesses the extratropical influence on observed ENSO variability and El Niño events. Mechanisms of the extratropical influence on observed ENSO are further analyzed in section 4. A summary and discussion are given in section 5.

2. Data and methods

a. Model description

The GFDL CM2.1 is a state-of-art model used for IPCC AR4 (Delworth et al. 2006). Its atmospheric component (AM2.1) has a horizontal resolution of 2.5° (longitude) × 2° (latitude) and 24 vertical levels with a hybrid coordinate. The ocean component (MOM4) has a refined resolution in the tropics, with the resolution being 1° × 1° in the extratropics and linear telescoping to 1° × ⅓° near the equator. This model can generally simulate a realistic tropical Pacific climate, with observed climatologic SST, surface winds, precipitation, and subsurface thermal structure well captured (Wittenberg et al. 2006). Most important, GFDL CM2.1 is one of the best five modes among 19 climate models of IPCC AR4 in ENSO simulation (van Oldenborgh et al. 2005), with observed ENSO period, skewness, air–sea feedback processes, and thermocline processes during the development of ENSO well performed. These factors greatly increase the reliability of our assessment work.

b. RDA experimental design

The regional data assimilation (RDA) approach is based on the ensemble adjustment Kalman filter (EAKF) scheme (Anderson 2001, 2003; Zhang et al. 2007) and more details can be found in Liu et al. (2022). Now we use this approach in the GFDL CM2.1 model to assess the impact of the extratropical atmospheric variability on observed ENSO. In this approach, the extratropical atmospheric state (T, U, V) is assimilated every 6 h with 4-times daily NCEP–NCAR Reanalysis 1 datasets (Kalnay et al. 1996), leaving the tropics coupled freely to generate ENSO variability under the observed extratropical atmospheric condition. Here the observed extratropical atmospheric condition includes both the climatology and variabilities.

We set up a control experiment (CTRL) and a series of RDA experiments. All experiments are forced by the historical external forcing and run from 1969 to 2008 for 40 years. Each experiment has 12 ensemble members, with the ensemble-mean experiment analyzed. As a reference, the CTRL is an ensemble of historical runs without assimilation and thus the internal variability in the ensemble mean has been largely removed. Similar to LU18, to assess the different latitudinal impacts on observed ENSO, we set up four groups of experiments by gradually moving the assimilation boundary poleward, with atmospheric data assimilation (ADA) activated over the globe (ADAALL), poleward of 10° (ADA10), 20° (ADA20) and 30° (ADA30), respectively. ADAALL serves as a benchmark of the best realizable scenario. Beyond LU18, to assess the respective contribution from the northern and southern extratropics, we perform four additional groups of RDA experiments, with ADA activated north of 20°N (ADA20N) and 30°N (ADA30N), and south of 20°S (ADA20S) and 30°S (ADA30S), respectively.

Therefore, in the ensemble-mean RDA experiments, the tropical variability is only forced by observed extratropical atmospheric variability, with tropical internal variability largely filtered out, revealing the extratropical influence on ENSO in observation.

c. Datasets and indices

The assimilation datasets from NCEP–NCAR Reanalysis 1 (Kalnay et al. 1996) are regarded as the observation (OBS) in our study. The observed SST is from the Hadley Centre Sea Ice and Sea Surface Temperature (HadISST) (Rayner et al. 2003). The observed ocean temperature is from the Met Office Hadley Centre EN series (EN4.2.1) (Good et al. 2013). All datasets are linearly detrended before analysis and the monthly-mean anomalies are derived by subtracting the climatologic (1969–2008) seasonal cycle.

Several indices are used for analysis. The Niño-3.4 index is the average SST anomaly (SSTA) between 5°S and 5°N and between 170° and 120°W. The WWV index is defined as the average upper-ocean (0–300 m) temperature anomaly between 5°S and 5°N and between 120°E and 80°W. The NPMM index is defined as the 3-month running mean area-averaged surface wind speed anomaly over 5°–20°N, 180°–140°W with a negative sign, such that the weakening of surface wind speed represents the positive phase of NPMM/SPMM, which tends to be correlated with a warm SSTA. Also, the SPMM index is defined as the 3-month running mean area-averaged surface wind speed anomaly over 15°–25°S, 110°–80°W with a negative sign, such that the strengthening of surface wind speed represents the negative phase of NPMM/SPMM, which tends to be correlated with a cold SSTA. The NPMM and SPMM indices defined in our study follow some previous studies (Zhang et al. 2014; Lu et al. 2017; Lu and Liu 2018) and are similar to those defined using maximum covariance analysis (MCA) in other previous studies (Chiang and Vimont 2004; You and Furtado 2018), that means using different definition of PMM indices makes no difference to our conclusion.

Besides the indices of Pacific, the indices of other tropical oceans are also used to preliminarily study a potential extratropical influence on ENSO via these two tropical oceans. In the tropical Indian Ocean, the Indian Ocean dipole (IOD) index is defined as the area-averaged SSTA difference between 10°S–10°N, 50°–70°E and 10°S–0°, 90°–110°E (Saji et al. 1999) and the Indian Ocean basin mode (IOBM) index is defined as the area-averaged SSTA over 20°S–20°N, 40°–110°E (Klein et al. 1999). In the tropical Atlantic, the tropical northern Atlantic (TNA) index is defined as the average SSTA over 55°–15°W, 5°–25°N (Enfield et al. 1999) and the Atlantic Niño/Niña is represented by ATL3 index defined as the average SSTA over ATL3 region (20°W–0°, 3°S–3°N).

3. Extratropical impact on observed ENSO

a. Extratropical impact on observed ENSO variability

Our RDA experiments in GFDL CM2.1 confirm a strong extratropical influence on observed ENSO variability, consistent with the study of LU18 in FOAM. This can be seen clearly from a comparison of Niño-3.4 (Fig. 1) and WWV indices (Fig. 2) between RDA experiments (black solid line) and OBS (red solid line). A higher correlation and/or a lower root-mean-square error (RMSE) indicate a greater contribution of specific extratropical region to ENSO. For a better comparison, the RMSE in our study is normalized as the ratio of the original RMSE to the standard deviation (STD) of OBS. Additionally, since ENSO has a strong signature in subsurface ocean dynamics, including the recharge/discharge of the equatorial heat content (WWV) leading ENSO by a quarter of ENSO cycle (Meinen and McPhaden 2000), the extratropical influence on the WWV will be also analyzed, which has not been done previously (LU18; Lu et al. 2017; Amaya et al. 2019).

Fig. 1.
Fig. 1.

The time series of Niño-3.4 index (black solid line) in (a) CTRL and in the RDA experiments, including (b) ADAALL, (c) ADA10, (d) ADA20, (e) ADA20N, (f) ADA20S, (g) ADA30, (h) ADA30N, and (i) ADA30S. The red solid line is the same in all panels and represents the Niño-3.4 index in OBS. The correlation and RMSE between each experiment and OBS are shown in the title of each panel, with significant correlations over 95% confidence level marked with two asterisks. The black dashed lines are the same in all panels and show the STD (0.43°C) of the Niño-3.4 index in CTRL. The observed El Niño events are marked with red circles in each panel.

Citation: Journal of Climate 37, 3; 10.1175/JCLI-D-23-0023.1

Fig. 2.
Fig. 2.

As in Fig. 1, but for the time series of WWV index. The black dashed lines are the same in all panels and show the STD (0.14°C) of WWV index in CTRL.

Citation: Journal of Climate 37, 3; 10.1175/JCLI-D-23-0023.1

We first analyze two benchmark experiments, CTRL and ADAALL, for the “worst” and “best” scenarios, respectively. Without assimilation, the variability of CTRL (Figs. 1a and 2a) has been suppressed severely due to the ensemble mean of 12 independent members. As such, CTRL shows little correlation with OBS and a large RMSE comparable with the STD of OBS, in both the Niño-3.4 index (R = 0.00 and RMSE = 1.12; Fig. 1a) and the WWV index (R = 0.16 and RMSE = 1.01; Fig. 2a). ADAALL can nearly reproduce the observed Niño-3.4 index (Fig. 1b) with a very high correlation (R = 0.92) and a small RMSE (0.40). The WWV index is also simulated well in comparison with OBS (Fig. 2b), as seen in a high correlation (R = 0.70) and a small RMSE (0.73). Relatively, the WWV is less well simulated than the Niño-3.4 index in ADAALL. This is reasonable because SST is forced more effectively by observed surface atmospheric temperature, while the WWV depends more on the oceanic dynamics driven by observed wind stress.

When atmospheric assimilation is activated poleward of 10° in ADA10, both the Niño-3.4 (Fig. 1c) and WWV indices are still well simulated (Fig. 2c), with the correlations (RMSEs) reduced (increased) slightly relative to ADAALL. This occurs because 10° is still within the tropics and the response in ADA10 is still forced strongly by tropical observations. When the assimilation boundary moves farther poleward to 20°, which is comparable with the deformation radius of the tropical atmosphere, the tropical variability in ADA20 can be considered as the response to the observed extratropical forcing. This observed extratropical forcing poleward of 20° can still force a strong ENSO response, with high correlations in both the Niño-3.4 index (R = 0.75; Fig. 1d) and WWV index (R = 0.59; Fig. 2d) in ADA20, explaining ∼56% and 35% of observed variance (as the square of R), respectively. Besides, both RMSEs decrease substantially in comparison with CTRL, from 1.12 to 0.75 in the Niño-3.4 index and from 1.01 to 0.86 in the WWV index. This experiment implies a strong influence of the extratropics poleward of 20° on the observed ENSO variability, consistent with LU18 in a coarse-resolution model. Relatively, a moderate decrease (increase) of correlation (RMSE) from ADAALL to ADA20 suggests a relatively small contribution of tropical internal variability. As the assimilation region further shrinks to poleward of 30°, surprisingly, there still remains a significant extratropical influence. The correlations of Niño-3.4 index (R = 0.48; Fig. 1g) and WWV index (R = 0.36; Fig. 2g) between ADA30 and OBS are still statistically significant at the 95% level, with explained variances of ∼23% and 13%, respectively. However, the deteriorated correlation, and especially the RMSE, when compared with those in ADA20, implies an important role of subtropical atmospheric variability.

The extratropical influences are significant from both hemispheres. The correlations of Niño-3.4 index in ADA20N, ADA20S, ADA30N, and ADA30S (Figs. 1e,f,h,i) are 0.57, 0.64, 0.25, and 0.41, respectively. Meanwhile, the corresponding correlations of WWV indices are 0.50, 0.53, 0.18, and 0.35 (Figs. 2e,f,h,i), respectively. All of these correlations are significant except for that of the WWV index in ADA30N. It is worth noting that the impact from the southern extratropics is slightly larger than that from the northern extratropics, since the correlations of both Niño-3.4 index and WWV index in ADA20S/ADA30S are higher than those in ADA20N/ADA30N and the RMSEs of both Niño-3.4 index and WWV index in ADA20S/ADA30S are smaller than those in ADA20N/ADA30N.

The contributions of different extratropical regions to observed ENSO variability are summarized in Fig. 3, which shows the changes of correlations and RMSEs of Niño-3.4 index (Figs. 3a,b) and WWV index (Figs. 3c,d) with assimilation boundary. Generally, the correlations (RMSEs) increase (decrease) gradually as the assimilation boundary moves equatorward (from CTRL to ADAALL). Relatively, however, the correlation and RMSE respectively increases and decreases more rapidly from ADA30 to ADA20, in both the Niño-3.4 and WWV indices. It suggests a greater contribution of the subtropics (20°–30°) to observed ENSO variability. Here the consistent changes between Niño-3.4 index and WWV index reveal the coupled nature of ENSO variability. Furthermore, a greater influence from the southern extratropics than from the northern extratropics can be seen more clearly with the correlations of green squares all higher and the RMSEs all smaller than those of the orange triangles, implying that the signal from the southern extratropics more easily propagates to the equator, causing the high signal-to-noise ratio. One interesting thing is that the influences of the northern and southern extratropics appear largely independent in the high latitudes (>30°), but highly correlated in the subtropics (20°–30°). This can be inferred from the explained variance. The explained variances of Niño-3.4 index (23%) and WWV index (13%) in ADA30 nearly equal the sum of those in ADA30N (6%, 3%) and ADA30S (17%, 12%), whereas the explained variances in ADA20 (56%, 35%) are much smaller than the sum of those in ADA20N (33%, 25%) and ADA20S (41%, 29%), implying a tight connection between the northern and southern subtropics, likely via the rapid and strong tropical impact. The tight connection of the subtropical influences and the strong independence of the high-latitude extratropical influences can be further examined in scatterplot of Niño-3.4 index between ADA20N and ADA20S, and between ADA30N and ADA30S (not shown), with a significant correlation (R = 0.32) of Niño-3.4 index between ADA20N and ADA20S and little correlation (R = −0.01) between ADA30N and ADA30S.

Fig. 3.
Fig. 3.

The changes of (left) correlation (with OBS) and (right) RMSE of the (a),(b) Niño-3.4 index and (c),(d) WWV index with assimilation boundary. The black-outlined star represents CTRL. The blue circles represent the RDA experiments with ADA activated in both hemispheres (ADA30, ADA20, and ADA10). The orange triangles and green squares represent the RDA experiments with ADA activated only in Northern Hemisphere (ADA30N and ADA20N) and Southern Hemisphere (ADA30S and ADA20S), respectively. The filled symbols in (a) and (c) indicate the significant correlation over 95% confidence level based on Student’s t test.

Citation: Journal of Climate 37, 3; 10.1175/JCLI-D-23-0023.1

b. Extratropical impact on observed El Niño events

The extratropical influence on observed El Niño events is further studied, with a focus on two questions. The first is how many historical El Niño events are influenced by the extratropics and the second is to what degree the extratropics contribute to these El Niño events. Figure 4 shows the peak values (y axis) and peak times (colors of circles) of observed El Niño events in RDA experiments. A value of 0.5°C is sufficiently beyond the sampling error of CTRL and can be a criterion for identifying ENSO events in RDA experiments. Therefore, a successfully reproduced El Niño event in RDA experiments is defined to be the same as the OBS when the 3-month running-mean Niño-3.4 index exceeds 0.5°C for at least five consecutive months and is marked with a colored circle in Fig. 4. In contrast, the observed El Niño events not reproduced in RDA experiments are marked with gray circles.

Fig. 4.
Fig. 4.

Scatters of peak values of the observed El Niño events in OBS (x axis) vs RDA experiments (y axis), including (a) ADAALL, (b) ADA10, (c) ADA20, (d) ADA20N, (e) ADA20S, (f) ADA30, (g) ADA30N, and (h) ADA30S. The peak value is defined as the maximal Niño-3.4 index from September of observed El Niño year to March of the next year. The gray and colored circles in each panel represent unsuccessfully and successfully reproduced El Niño events in RDA experiments, respectively. The color of colored circles further indicates the peak time of the El Niño event in RDA experiments relative to OBS. The bluer or redder colors respectively mean the peak times in RDA experiments are earlier or later than or synchronous (in the case of green) with those in OBS. The numerator and denominator in the title represent the number of observed El Niño events in each RDA experiment and OBS, respectively.

Citation: Journal of Climate 37, 3; 10.1175/JCLI-D-23-0023.1

In OBS, there are in total 12 El Niño events during 1969–2008: 1972/73, 1976/77, 1977/78, 1982/83, 1986/87, 1987/88, 1991/92, 1994/95, 1997/98, 2002/03, 2004/05 and 2006/07 events (e.g., 1972/73 is denoted as 72 in Fig. 4). As a benchmark, ADAALL reproduces all historical El Niño events except for a weak 1977/78 event (Fig. 4a), with their peak values nearly the same as OBS and their peak time very close to OBS, revealing a strong atmospheric control via both thermal and dynamic forcing. ADA10 can reproduce 10 of the 12 observed El Niño events (Fig. 4b), with most peak values nearly the same as OBS and some even higher than OBS, which may be related to the model deficiency or a damping effect of deep tropical internal variability.

The great extratropical contribution to observed El Niño events can be seen in ADA20, which is able to reproduce 8 of the 12 (∼67%) observed El Niño events, with most peak values similar with OBS (Fig. 4c). Especially, all of three super El Niño events (those in 1972/73, 1982/83, and 1997/98) are successfully reproduced, implying a great extratropical influence on super El Niño events. However, the peak values of super El Niño events in 1972/73 and 1997/98 are much smaller than OBS, similar to the result of LU18 in FOAM, suggesting a great contribution of tropical processes to these events. As pointed out by previous studies (Boulanger et al. 2004; Lengaigne et al. 2004; Menkes et al. 2014), the super 1997/98 El Niño event is influenced greatly by WWB. The important information is that the peak times (colors of circles) of the eight reproduced El Niño events in ADA20 do not show a systematic delay relative to OBS, with some synchronous with OBS and others leading or lagging behind OBS. No systematic delay of peak times in ADA20 to some extent supports that the observed extratropical variability triggering El Niño events in ADA20 is not a response to the developing El Niño event in observation. Otherwise, this extratropical signal associated with the developing El Niño event in observation needs to take extra time to first propagate back to the equator in ADA20, causing a systematic delay of peak times relative to OBS.

As for respective contributions of the northern and southern extratropics, ADA20N (Fig. 4d) and ADA20S (Fig. 4e) can reproduce 6 and 7 El Niño events, respectively, with their sum much larger than the 8 events in ADA20 (Fig. 4c). It is because 4 El Niño events in ADA20, including 3 super El Niño events (1972/73, 1982/83 and 1997/98), are reproduced in both ADA20N and ADA30S, suggesting a joint contribution from both hemispheric extratropics. The other 4 El Niño events in ADA20 are triggered by only one hemispheric extratropics, with the CP events of 2002/03 and 2004/05 reproduced in ADA20N, and the EP events of 1976/77 and 2006/07 reproduced in ADA20S.

Interestingly, even from poleward of 30°, the extratropics can still contribute significantly to observed El Niño events. ADA30 can reproduce 7 of the 12 (∼58%) historical El Niño events (Fig. 4f), with ADA30N (Fig. 4g) and ADA30S (Fig. 4h) reproducing 3 and 4 events, respectively. However, ADA30 also generates some spurious El Niño events, such as the one in 1979/80 (Fig. 1g), due to a more regular oscillation of Niño-3.4 index in ADA30. This interesting phenomenon needs our further study in the future. In contrast to a joint contribution of ADA20N and ADA20S, there is almost no common El Niño events reproduced in both ADA30N and ADA30S, further suggesting independent influences of the high-latitude northern and southern extratropics on ENSO as discussed in Fig. 3.

4. Mechanisms of extratropical influence on ENSO

a. Mechanisms of extratropical Pacific influence on ENSO

The extratropical Pacific influences ENSO mainly through coupled ocean–atmosphere processes. This can be seen from the correlation maps between RDA experiments and OBS. Since the high correlation in the unassimilated tropics in RDA experiments must be caused by the observed extratropical forcing, the high-correlation pathways from the assimilated extratropics into the tropics can indicate the specific propagation/transport pathways of extratropical influence on the tropics.

The meridional pathway of extratropical Pacific influence on ENSO can be first studied in the zonal-mean correlation (with OBS) of monthly-mean atmospheric and upper-oceanic temperature anomalies in the Pacific (Fig. 5). As benchmarks, the correlation is nearly 0 everywhere in CTRL (Fig. 5a), but almost 1 in the entire atmosphere in ADAALL (Fig. 5b), as expected. In ADAALL, moderate correlations also appear in the extratropical oceanic mixed layer, caused by a strong extratropical atmospheric thermal forcing, and deep into the equatorial thermocline, due to the thermal as well as dynamic forcing from the equatorial atmosphere. In RDA experiments, the correlation is almost 1 in the assimilated extratropical atmosphere and is also moderate in the extratropical oceanic mixed layer (Figs. 5c–i). From the assimilated extratropics into the freely coupled tropics, there are in general two high-correlation pathways (Figs. 5d–i). One penetrates equatorward in the free atmosphere especially in the upper troposphere (upper pathway). This upper pathway can be caused by the equatorward transport via the upper branch of summer Hadley cell (HC; Liu et al. 2023) or meridional propagation of Rossby waves via the westerly duct (WD; Webster and Holton 1982). The other penetrates the boundary layer atmosphere as well as the oceanic mixed layer (surface pathway). These two pathways imply that the extratropics may influence ENSO by either atmospheric processes or coupled ocean–atmosphere processes.

Fig. 5.
Fig. 5.

Zonal-mean correlations of monthly-mean atmospheric and upper-oceanic temperature anomalies in the Pacific (110°E–70°W) between OBS and (a) CTRL, (b) ADAALL, (c) ADA10, (d) ADA20, (e) ADA20N, (f) ADA20S, (g) ADA30, (h) ADA30N, and (i) ADA30S. Only significant correlations over 95% confidence level based on Student’s t test are drawn. The dashed lines indicate the assimilation boundaries.

Citation: Journal of Climate 37, 3; 10.1175/JCLI-D-23-0023.1

A further careful examination shows that a stronger impact from southern extratropics than from northern extratropics occurs in surface pathway, as seen in a higher-correlation surface pathway in ADA20S/30S than in ADA20N/30N (Figs. 5f,i vs Figs. 5e,h). In contrast, the correlation of upper pathway in ADA20N is higher than that in ADA20S, which is mainly caused by the seasonal asymmetry of ITCZ position (see Fig. 6 in Liu et al. 2023). Since the ITCZ position or the ascending branch of HC in boreal summer is more poleward than that in boreal winter, the northern extratropical information poleward of 20°N can be transported into the deep tropics more easily via the upper branch of summer HC, causing a higher-correlation upper pathway in ADA20N. Here the higher-correlation surface pathway in ADA20S/30S (Figs. 5f,i vs Figs. 5e,h) is consistent with the higher correlations of Niño-3.4 and WWV indices in ADA20S/30S (Figs. 3a,c), indicating the important role of coupled ocean–atmosphere processes in extratropical influence on ENSO. Therefore, the stronger coupled ocean–atmosphere processes from the southern extratropical Pacific contribute greatly to a greater southern extratropical influence on ENSO. Furthermore, a higher correlation in the equatorial upper-ocean (0–300 m) heat content than in the equatorial boundary layer air temperature (Figs. 5d,e) suggests that ENSO is forced more by the wind-driven thermocline process than by the downward atmospheric heat flux, consistent with the dynamics of ENSO in general.

The coupled ocean–atmosphere processes of extratropical influence on ENSO can be further studied in the correlation of monthly-mean Pacific SSTA (Fig. 6) and surface zonal wind anomaly (Fig. 7) between RDA experiments and OBS. In ADA20 and ADA20N, there is a high-correlation tongue in both SSTA (Figs. 6a,c) and surface zonal wind anomaly (Figs. 7a,c) that is initiated from the subtropical northeastern Pacific and extends southwestward, consistent with the path of NPMM (Chiang and Vimont 2004; Amaya et al. 2019), implying the equatorward propagation of observed coupled SST–zonal wind anomalies via WES feedback. The NPMM-like pathway also appears in ADA30/ADA30N (Figs. 6b,d and 7b,d), but with a lower correlation than in ADA20/ADA20N, confirming the great contribution of subtropics to ENSO. Further examination shows that it is the rapid drop of correlation between 20° and 30°N in ADA30N (Fig. 6d) that greatly contributes to a lower correlation of NPMM and, further, a lower correlation of Niño-3.4 index in ADA30N than in ADA20N. This rapid-drop location is consistent with the boundary of HC and also the dividing line of trade wind and westerly wind. Since the equatorward propagation via WES feedback can only happen in trade-wind region (Liu and Xie 1994), the high-latitude (>30°) extratropical information must be first transported equatorward by transient eddies (Held 2001; Kang et al. 2008, 2009) before entering the trade-wind region, causing a rapid drop of correlation between 20° and 30°N and, further, a lower correlation of NPMM and Niño-3.4 index in ADA30N.

Fig. 6.
Fig. 6.

Correlation of monthly-mean Pacific SSTA between OBS and (a) ADA20, (b) ADA30, (c) ADA20N, (d) ADA30N, (e) ADA20S, and (f) ADA30S. Only the significant correlation over 95% confidence level based on Student’s t test is drawn. The climatologic ITCZ position is represented by annual-mean precipitation (green contours; contour interval is 3 mm day−1). The blue line and black line with circles represent the path of NPMM and SPMM, respectively, with NPMM from the point of 25°N, 120°W to the point of 0°, 180° and SPMM from the point of 25°S, 80°W to the point of 0°, 160°W. The dashed lines indicate the assimilation boundaries.

Citation: Journal of Climate 37, 3; 10.1175/JCLI-D-23-0023.1

Fig. 7.
Fig. 7.

As in Fig. 6, but for the correlation of monthly-mean surface zonal wind anomalies.

Citation: Journal of Climate 37, 3; 10.1175/JCLI-D-23-0023.1

Similarly, in ADA20/ADA20S and ADA30/ADA30S, there is also a high-correlation tongue in both SSTA (Figs. 6a,b,e,f) and surface zonal wind anomaly (Figs. 7a,b,e,f), starting from the subtropical southeastern Pacific and extending northwestward into the equator, reminiscent of the SPMM (Zhang et al. 2014; You and Furtado 2018). Besides, the correlation of SST along the SPMM in ADA20S/ADA30S (Figs. 6e,f) is higher than that along the NPMM in ADA20N/ADA30N (Figs. 6c,d), which is consistent with a higher-correlation surface pathway in zonal mean in ADA20S/ADA30S (Figs. 5f,i vs Figs. 5e,h), implying that a stronger WES feedback of SPMM than of NPMM contributes to a greater southern extratropical influence on ENSO. This stronger WES feedback of SPMM may be caused by two factors. First, it may be related to the position of Pacific ITCZ. According to both the theory obtained from a simple model (Liu and Xie 1994) and the results of sensitivity experiments (Zhang et al. 2014), the Pacific ITCZ north of the equator will to some degree prevent the northern extratropical signal from reaching the equator by hampering either the equatorward propagation via the WES feedback or the equatorward advection via trade wind (Kang et al. 2018; Shin et al. 2021), thus weakening the influence of the northern extratropics. Here the high correlation of SST along the NPMM in ADA20N/30N (Figs. 6c,d) drops sharply at the location of Pacific ITCZ (∼10°N; green contours in Figs. 6c,d), which is in sharp contrast to the relatively uniform high correlation along the SPMM in ADA20S/30S (Figs. 6e,f), further supporting the hypothesis of previous studies. Second, it may be related to a shallower mixed-layer depth (MLD) along the SPMM. Many previous studies indicate that the equatorward evolutions of NPMM and SPMM via WES feedback both are most active in boreal spring (You and Furtado 2018; Amaya 2019). Here the MLD in boreal spring [March–May (MAM)] along the SPMM (black line in Figs. 8a,b,e,f) is much shallower than that along the NPMM (blue line in Figs. 8a–d). This shallower MLD will contribute to a more sensitive SSTA response to wind speed anomaly along SPMM, finally causing a stronger WES feedback of SPMM. It is also interesting that the highest correlation of SST in the equator is located in the eastern Pacific in ADA20S/30S (Figs. 6e,f) but in the central Pacific in ADA20N/30N (Figs. 6c,d). This difference of correlation pattern seems to support some studies that the NPMM is more favorable for CP-ENSO while the SPMM is more favorable for EP-ENSO (Yu and Kim 2011; Vimont et al. 2014; Zhang et al. 2014; Min et al. 2017; You and Furtado 2017).

Fig. 8.
Fig. 8.

Climatologic MAM-mean mixed-layer depth in the Pacific. The blue line and black line with circles represent the path of NPMM and SPMM, respectively. The dashed lines indicate the assimilation boundaries.

Citation: Journal of Climate 37, 3; 10.1175/JCLI-D-23-0023.1

We further investigate the extratropical influence on ENSO via different coupled ocean–atmosphere processes of PMM mentioned in Amaya (2019). Since the seasonality of PMM indices in our RDA experiments is highly consistent with the observation and also the previous studies (not shown), and these previous studies have indicated that the equatorward evolutions of both NPMM and SPMM are most active in boreal spring. we follow up these studies to do the lag regression of Pacific surface variables, including SSTA, SLP anomaly (SLPA), surface wind and latent heat anomalies, and also the oceanic variables, including ocean heat content and vertically integrated meridional mass transport onto the normalized MAM NPMM (Fig. 9) and SPMM (Fig. 10) indices in RDA experiments.

Fig. 9.
Fig. 9.

Lag regression of SSTA (shading), SLPA (red and blue contours represent positive and negative values, respectively; contour interval is 0.2 hPa), surface wind anomaly (vectors; m s−1), and latent heat flux anomaly (green contour represents downward anomaly; contour interval is 3 W m−2) onto the standardized MAM NPMM index in (a) MAM, (b) JJA, (c) SON, and (d) DJF in ADA20N (SON, not defined in the text, indicates September–November). Only significant correlations over 95% confidence level based on Student’s t test are drawn. (e)–(h) As in (a)–(d), but for the vertically integrated (0–300 m) meridional mass transport anomaly (shading) and ocean heat content (vertical-average ocean temperature over 0–300 m) anomaly (magenta contours indicate positive values, contour interval is 0.1°C). (i)–(p) As in (a)–(h), but in ADA30N.

Citation: Journal of Climate 37, 3; 10.1175/JCLI-D-23-0023.1

Fig. 10.
Fig. 10.

As in Fig. 9, but in ADA20S and ADA30S.

Citation: Journal of Climate 37, 3; 10.1175/JCLI-D-23-0023.1

First, PMMs can trigger ENSO via WES feedback. In ADA20N/30N, during MAM, the warm SSTA (shading in Figs. 9a,i) and westerly wind anomaly (vectors in Figs. 9a,i) over the northeastern Pacific are excited by the southern node of NPO-like SLPA (blue contours in Figs. 9a,i), and then propagate southwestward into the equator, accompanied by the decreased latent heat flux (green contours in Figs. 9a,i), confirming the effect of the WES feedback. Then this westerly anomaly reaching the western equator will excite the downwelling ocean Kelvin waves, adjusting the equatorial thermocline depth (contour in Figs. 9f,n) and finally causing the growth of SSTA in the central and eastern equatorial Pacific (shading in Figs. 9c,d,k,i). Similarly, in ADA20S/30S, the warm SSTA (shading in Figs. 10a,i) and the weakened trade wind (vectors in Figs. 10a,i) over the southeastern Pacific are excited by the north node of SPO-like SLPA (blue contours in Figs. 10a,i), and then propagate northwestward in MAM via WES feedback, accompanied by the decreased latent heat loss (green contours in Figs. 10a,i). However, different from the wind-driven thermocline processes excited by NPMM in ADA20N/ADA30N, the warm SSTA and wind anomaly reaching the eastern equator in ADA20S/30S do not cause a significant adjustment of ocean Kelvin waves and, further, the adjustment of thermocline depth (Figs. 10f,n). Instead, they just induce a warming of the eastern equatorial SST and further trigger the Bjerknes feedback to develop ENSO (Figs. 10c,d,g,h). Here the seasonal evolution of SPMM in our RDA experiments supports the arguments of Larson et al. (2018) that SPMM is more likely a thermally driven source of ENSO. It should be noticed that a doubt occurred recently as to the feasibility of equatorward propagation via WES feedback due to the effect of ocean dynamics (Hu et al. 2023; Shu et al. 2023). Therefore, the NPMM influences ENSO via WES feedback may need to be further investigated in the future.

Second, NPMM can also trigger ENSO via the TWC mechanism. During MAM, an NPMM-induced westerly wind anomaly north of the equator (vectors in Figs. 9e,m) will form a negative wind stress curl and further drive an anomalous equatorward Sverdrup transport north of the equator (shading in Figs. 9e,m), thus causing a mass convergence in the western and central equatorial Pacific, finally charging the equatorial ocean heat content (contours in Figs. 9e,m). In contrast, SPMM-induced westerly wind anomaly (vectors in Figs. 10e,m) does not promote a charging of the ocean heat content either in the eastern equator or in the western equator (Figs. 10e,f,m,n), indicating the inapplicability of TWC mechanism for SPMM. Besides, this result is unchanged even if we use the boreal summer [June–August (JJA)] SPMM index, implying the robustness of our conclusion.

Third, NPMM can trigger ENSO via the SDC mechanism. Since this mechanism takes effect in boreal later summer when the Pacific ITCZ reaches its northernmost position and anchors with NPMM-induced warm SSTA (Amaya et al. 2019), we analyzed it by regressing the July–September (JAS) surface wind anomaly and precipitation anomaly onto the JAS NPMM index in ADA20N/30N (Figs. 11a,b). During JAS, the NPMM-induced warm SSTA (shading in Figs. 11a,b) will promote the northward shift of Pacific ITCZ (green and brown contours in Figs. 11a,b), strengthening the convection over the warm SSTA region and decreasing the precipitation over its south. This meridional asymmetric convective heating will cause a Gill-like atmospheric response, including the westerly wind anomalies west of the heating source that can project onto the equator (vectors in Figs. 11a,b), consistent with the asymmetric forcing case in Gill (1980). These westerly wind anomalies both over the north of the equator and over the western equator in later summer can also contribute to the development of ENSO via TWC mechanism and via triggering ocean Kelvin waves, respectively. It should be pointed out that the surface wind anomaly in JAS is not only the response to the anomalous convective heating, but also contains the response to SSTA. A more rigorous study should be done in the future by performing a sensitivity experiment with SST prescribed to climatologic values. Besides, it is more interesting to note that this deep convection mechanism can also apply to SPMM. However, this mechanism for SPMM is most active in boreal spring (MAM) instead of in boreal summer (Fig. 11d). Since only in boreal spring will the Pacific ITCZ split into two branches, with the slight one over south of the equator, the warm SST anomaly associated with SPMM can anchor with this southern branch of ITCZ, leading to the southeastward extension of ITCZ. Our finding here is consistent with Min et al. (2017) and Zhang et al. (2021).

Fig. 11.
Fig. 11.

Regression of JAS SSTA (shading), SLPA (red and blue contours represent positive and negative values, respectively; contour interval is 0.2 hPa), surface wind anomaly (vectors; m s−1), and precipitation anomaly (green and brown contours represent positive and negative values; contour interval is 0.4 mm day−1) onto the standardized JAS NPMM index in (a) ADA20N and (b) ADA30N. (c),(d) As in (a) and (b), but for the regression of MAM anomalies fields onto the standardized MAM SPMM index in ADA20S and ADA30S, respectively. Only significant correlations over 95% confidence level based on Student’s t test are drawn.

Citation: Journal of Climate 37, 3; 10.1175/JCLI-D-23-0023.1

b. Combined effects of ENSO precursors in the Pacific

We first study the contributions of the two extratropical Pacific precursors (NPMM and SPMM) to observed ENSO. Figure 12a shows the peak values of observed El Niño events (y axis) and magnitudes of observed MAM NPMM (colors of bottom bars) and SPMM (colors of top bars). These events are sorted in ascending order by their peak values and into four groups of weak (2004/05, 1976/77, 2006/07, and 1977/78), moderate (1994/95, 1986/87, and 2002/03), strong (1987/88 and 1991/92), and super (1972/73, 1982/83, and 1997/98) El Niño events. Figures 12b–d show the peak-value ratio (y axis) of El Niño events in ADA20/N/S to those in OBS and magnitudes of MAM NPMM (colors of bottom bars in Fig. 12b) and SPMM (colors of top bars in Fig. 12b) in these experiments. Since there is only NPMM in ADA20N and SPMM in ADA20S, the colors of bars in Figs. 12c and 12d represent the magnitudes of MAM NPMM and SPMM, respectively. The successfully reproduced El Niño events in ADA20/N/S (Figs. 12b–d) are marked with stars.

Fig. 12.
Fig. 12.

The peak-value ratio of ADA20/N/S to OBS in observed El Niño events and the magnitude of NPMM and SPMM index in boreal spring (MAM): (a) Peak values (y axis) of El Niño events in OBS, along with peak value ratio (y axis) of (b) ADA20, (c) ADA20N, and (d) ADA20S to OBS in observed El Niño events. Colors of the bottom and top of bars in (a) and (b) represent the magnitude of MAM NPMM or SPMM index, respectively, in OBS [in (a)] and ADA20 [in (b)]. Colors of bars in (c) and (d) represent the magnitude of MAM NPMM index in ADA20N [in (c)] and MAM SPMM index in ADA20S [in (d)]. The El Niño events in (a)–(d) are sorted in ascending order by the peak values in OBS and into four groups. The successfully reproduced El Niño events in (b)–(d) are marked with stars.

Citation: Journal of Climate 37, 3; 10.1175/JCLI-D-23-0023.1

First, all of the three super El Niño events (1972/73, 1982/83, and 1997/98) are influenced greatly by both hemispheric extratropics, with a joint contribution of positive NPMM and SPMM in boreal spring (Figs. 12b–d). Despite the strong extratropical contributions to super El Niño events, the peak values of the 1972/73 and 1997/98 events in ADA20 (Fig. 12b) are still far below the OBS, implying an important role of the tropical processes as mentioned before. Second, most weak and moderate El Niño events are triggered by only one hemispheric extratropics (Fig. 12c vs Fig. 12d), with an opposite contribution of NPMM and SPMM (Fig. 12b). Besides, the peak values of these weak and moderate events in ADA20 (Fig. 12b) are usually comparable to or even greater than OBS, suggesting an unimportant or even a damping role of the tropical processes. Our results here are highly consistent with the statistical analysis based on observation in You and Furtado (2018). Third, a strong El Niño event of 1991/92 appears in ADAALL and ADA10 but not in ADA20/N/S, implying little extratropical contribution and a strong tropical contribution to this event. It should be noted that the NPMM/SPMM indices sometimes are slightly different among RDA experiments and OBS, which is a natural characteristic of RDA approach and can also be seen in LU18. Since the regions in which the NPMM and SPMM indices are defined contain the unassimilated tropics, and the responses of the unassimilated region will be different with the changes of assimilated regions, these two indices will be slightly different among RDA experiments. Nevertheless, these slight differences do not influence our major conclusion.

We further systematically study the combined effects of the tropical Pacific precursor (WWV) and the two extratropical Pacific precursors (PMMs) on observed ENSO. Figure 13 shows the relation between each precursor in boreal spring (MAM) and Niño-3.4 index in the following winter (DJF) in OBS and RDA experiments. Before that, we first check the relation among these precursors. There is a significant correlation between NPMM and WWV in MAM (numbers in the titles of Figs. 13a–e), consistent with the TWC mechanism we discussed before, implying that NPMM is one way for extratropics to regulate WWV. In contrast, the low correlation between SPMM and WWV in MAM (numbers in the titles of Figs. 13f–j) further indicates the inapplicability of TWC mechanism for SPMM, consistent with our discussion before and also the result of You and Furtado (2018).

Fig. 13.
Fig. 13.

(a)–(e) Scatterplot of the normalized MAM NPMM index and DJF Niño-3.4 index in OBS and RDA experiments when MAM NPMM has the same (red dots and red regression line) and opposite (blue dots and blue regression line) sign with MAM WWV, with the corresponding correlation coefficients marked in same colors. The correlation of all dots is marked in black color, and the black line denotes regression line. The significant correlation over 95% confidence level based on Student’s t test is marked with two asterisks. (f)–(j) As in (a)–(e), but for the scatterplots of MAM SPMM index and DJF Niño-3.4 index when MAM SPMM has the same (red dots) or opposite (blue dots) sign as MAM WWV. (k)–(o) As in (a)–(e), but for scatterplots of MAM WWV index and DJF Niño-3.4 index when MAM WWV has the same (red dots) or opposite (blue dots) sign as MAM NPMM.

Citation: Journal of Climate 37, 3; 10.1175/JCLI-D-23-0023.1

Each precursor alone seems to be a good predictor for ENSO, because of the significant correlation (black numbers at the upper left corner of each panel) between each precursor in MAM and the DJF Niño-3.4 index in OBS as well as RDA experiments. The correlations between two extratropical precursors (NPMM and SPMM) and ENSO are comparable, with ranges of 0.48–0.65 (Figs. 13a–e) and 0.45–0.57 (Figs. 13f–j), respectively. While the WWV has a higher correlation with ENSO, ranging from 0.61 to 0.78 (Figs. 13k–o), likely indicating a more important role of WWV in ENSO. However, the correlation between each precursor and ENSO is heavily dependent on the relation among precursors, which means any precursor alone is insufficient to trigger ENSO events. When NPMM/SPMM has the same sign as the WWV (red dots in Fig. 13), the correlation between each precursor and ENSO is much higher (red numbers at the upper left corner of each panel), as expected. When NPMM/SPMM have the opposite sign as WWV (blue dots in Fig. 13), the correlation between each precursor and ENSO will decrease greatly (blue numbers at the upper left corner of each panel). Especially, there will be little or even a negative correlation between NPMM/SPMM and ENSO (blue numbers in Figs. 13a–j), but there remains a positive correlation between WWV and ENSO (blue numbers in Figs. 13k–o), with a range of 0.21–0.55. These results are consistent with You and Furtado (2018) and may imply the dominant role of WWV in ENSO. It should be pointed out that, different from the statistical analysis based on observation of You and Furtado (2018), where the tropical precursor of WWV is forced by both tropical internal variability and extratropical variability, here the WWV in ADA20/N/S and ADA30/N/S is forced predominantly by the extratropics due to the suppression of tropical internal variability. Therefore, the dominant role of WWV in ENSO in our RDA experiments further highlights the importance of extratropical forcing.

Besides, the combination of these precursors can regulate the magnitudes of ENSO events. If NPMM/SPMM and WWV in MAM are opposite (blue dots in Fig. 13), then the amplitude of DJF Niño-3.4 index (the corresponding y axis of blue dots) is usually small, indicating no ENSO events or weak ENSO events. In contrast, when the amplitude of DJF Niño-3.4 index is very large, implying the occurrence of strong or super ENSO events, the NPMM/SPMM and WWV in the preceding spring usually have the same sign (the extreme values of red dots).

c. Potential extratropical influences on ENSO via other tropical oceans

Since our RDA experiments assimilate the whole extratropical atmosphere, besides the extratropical influence on ENSO via NPMM/SPMM, the extratropical impact on ENSO by influencing other tropical oceans and then by interactions among tropical oceans should also be discussed. Since the three tropical oceans are fully interactive mainly via atmospheric bridge (Cai et al. 2019; Wang 2019), a detailed investigation of the other two tropical oceans’ impacts on ENSO is hard to implement in the current RDA experiments and is worthy to be further studied by designing the special RDA experiments. Here we only provide a potential extratropical influence on ENSO via other tropical oceans.

First, the extratropical influences on the other two tropical oceans are examined. Since the IOB and IOD are the first two climate variability modes of the tropical Indian Ocean, and the TNA and Atlantic Niño/Niña are the major climate variability mode of the tropical Atlantic, the correlations of these indices between RDA experiments and OBS are analyzed (Fig. 14). From ADAALL to ADA20, the correlations of indices in the tropical Indian Ocean and tropical Atlantic are still at a high level, implying that the extratropics may also have a strong influence on these two tropical oceans. Certainly, the high correlations in these two tropical oceans in ADA20 can also be caused partly by the ENSO teleconnections. If we further examine the respective influences of the northern and southern extratropics, we can find that the extratropical influence on IOD is dominated by that from the southern extratropics. This finding is interesting since the IOD is historically regarded as a forced mode of the developing ENSO. Here the moderate correlations of IOD index in ADA20S/30S imply that the IOD is influenced greatly by the southern extratropics. In contrast, the extratropical influence on TNA is dominated by northern extratropical influence, which is natural since TNA represents the SST anomaly of the northern tropical Atlantic. As compared with the dominant role of only one hemispheric extratropics in IOD and TNA, the northern and southern extratropical influences on IOB and ATL3 are relatively comparable.

Fig. 14.
Fig. 14.

The correlations of the time series of IOB, IOD, TNA, and ATL3 indices between OBS and RDA experiments, including ADAALL, ADA20, ADA30, ADA20N, ADA30N, ADA20S, and ADA30S. Correlations that do not pass 95% confidence level of Student’s t test are marked with an open star.

Citation: Journal of Climate 37, 3; 10.1175/JCLI-D-23-0023.1

The high correlations of the indices in the tropical Indian Ocean and tropical Atlantic in ADA20 have indicated a potential extratropical influence on these two tropical oceans. Here we will further study potential influences of these two tropical oceans on ENSO by showing the lead–lag correlations between the indices of these two tropical oceans and Niño-3.4 index in both OBS and RDA experiments (Fig. 15).

Fig. 15.
Fig. 15.

The lead–lag correlations between the Niño-3.4 index and the (a) IOB index, (b) IOD index, (c) TNA index, and (d) ATL3 index in the OBS and RDA experiments. Only significant correlations over 95% confidence level based on Student’s t test are marked with dots.

Citation: Journal of Climate 37, 3; 10.1175/JCLI-D-23-0023.1

First, high positive correlations occur when Niño-3.4 index leads IOB and TNA indices by 3–6 months in both OBS and RDA experiments (red shading in Figs. 15a,c), consistent with the previous studies that the El Niño peaking in boreal winter will cause a warm IOB and TNA in the following spring [see Fig. 2 in Wang (2019)]. It suggests a strong impact of ENSO on other tropical oceans, further confirming the high correlations of the indices in the Indian Ocean and tropical Atlantic are caused partly by the ENSO teleconnections.

Second, significant negative correlations also occur when IOB and TNA indices lead Niño-3.4 index by about one year in both OBS and RDA experiments (blue shading in Figs. 15a,c), suggesting that a warm IOB or TNA that usually peaks in boreal spring will contribute to a La Niña event in boreal winter. According to previous studies, a warm IOB in boreal spring can cause a Gill response with atmospheric Kelvin waves propagating eastward and therefore forms the easterly wind anomaly in the western equatorial Pacific (Xie et al. 2009, 2016). Then this easterly wind anomaly will excite the upwelling ocean Kelvin waves and further cause the development of La Niña. Meanwhile, a warm TNA in boreal spring can also induce a Gill response with a low-level cyclone over the subtropical northeastern Pacific, thus causing the northeasterly wind anomaly in the west of this cyclone. This strengthened trade wind will produce a cold SSTA [see Fig. 2 in Cai et al. (2019)] and further a low-level anticyclone to the west of this cold SSTA, finally producing the easterly wind anomaly in the western equatorial Pacific and triggering the La Niña event (Ham et al. 2013; Wang et al. 2017).

Third, high positive correlations with IOD leading Niño-3.4 index by ∼3 months and moderate negative correlations with IOD leading Niño-3.4 index by ∼15 months (Fig. 15b) suggest that a positive IOD that often peaks in boreal fall will contribute to an El Niño event in the following winter and a La Niña event in winter of the next year. Here a positive IOD in boreal fall may reinforce the El Niño event by producing the westerly wind anomaly in the western equatorial Pacific. Meanwhile, a positive IOD with a cold southeastern Indian Ocean will increases the strength of the Indonesian Throughflow (ITF) and further discharge the WWV in the western Pacific, finally causing a La Niña event in boreal winter of the next year (Yuan et al. 2013).

At last, moderate negative correlations occur when ATL3 index leads Niño-3.4 index by ∼6 months in both OBS and RDA experiments (Fig. 15d), consistent with the results of previous studies showing that the Atlantic Niño always peaks in boreal summer and contributes to La Niña events in the following winter by changing the Atlantic Walker circulation (Wang et al. 2006; Polo et al. 2015).

5. Summary and discussion

Our study confirms a strong extratropical impact on observed ENSO in a much more advanced climate model, GFDL CM2.1. Specifically, extratropical atmospheric variability poleward of 20° can contribute to ∼56% of variance of observed ENSO variability and can trigger 8 of the 12 observed El Niño events. Furthermore, even the high-latitude extratropics poleward of 30° can still exert a statistically significant impact on observed ENSO.

Beyond LU18, we also study the respective contributions from the northern and southern extratropics. Our experiments show an overall greater impact on ENSO from the southern extratropics than the from northern extratropics, which may be partly caused by the location of Pacific ITCZ north of the equator and a shallower mixed-layer depth along the SPMM. A further examination of the extratropical impact on El Niño events indicates that all super El Niño events are influenced greatly by both hemispheric extratropics, with NPMM and SPMM interfering constructively with each other. In contrast, most weak and moderate El Niño events are triggered by only one hemisphere’s extratropics, with NPMM and SPMM usually interfering destructively with each other. Our study also suggests a potential extratropical impact on ENSO by influencing the other two tropical oceans and then by interbasin interactions. Since the three tropical oceans are fully interactive, a more rigorous study of extratropical influence on ENSO via other tropical oceans is required in the future by designing the special RDA experiments.

Note, however, that in our RDA experiments the observational extratropical atmospheric state may have already contained the observed tropical information due to the strong tropical impact on the extratropics via atmospheric bridge. This fully interactive nature of tropical–extratropical climate in the real world makes it difficult to definitely attribute the origin of the observed ENSO variability to remote and local processes. Although the RDA approach does have its limitations, until now it has been the best way for us to assess the extratropical influence on ENSO in observation. Especially, our RDA experiments reveal that the extratropical atmospheric variability does have a great influence on ENSO via various physical processes.

It is possible that some of our results depend on the model itself. However, we believe the major conclusion of a strong extratropical impact on ENSO is likely to be robust. First, this result is reproduced in two independent CGCMs, especially now the GFDL CM2.1 model, which performs especially well in the tropics. Second, our model study is consistent with many previous statistical analyses (e.g., Vimont et al. 2003b; Chang et al. 2007; You and Furtado 2018), even though most of them only focus on the impact of the northern extratropics. If this extratropical impact is true, it is surprising to see, quantitatively, such a strong impact, even from as far as 30°. This strong extratropical impact poses both opportunities and challenges to ENSO prediction. On the one hand, the prediction of some ENSO events may be improved if more attention can be paid to the earlier atmospheric variability in the extratropics. On the other hand, given the dominance of stochastic atmospheric variability in the extratropics, the extratropical atmospheric variability also provides a strong noise on tropics that may alter the nature of the ENSO development there. Further studies are needed to further understand the extratropical impact and the potential to improve climate prediction.

Acknowledgments.

This work is supported by U.S. NSF AGS-1656907. The authors declare that there is no conflict of interests regarding the publication of this article.

Data availability statement.

NCEP Reanalysis data are provided by the NOAA/OAR/ESRL PSL from their websites at https://psl.noaa.gov/data/gridded/data.ncep.reanalysis.derived.pressure.html. Sea surface temperature from HadISST and ocean temperature from EN4 (EN4.2.1) are provided by the Met Office Hadley Centre from their website https://www.metoffice.gov.uk/hadobs/index.html. Reasonable requests for data of RDA experiments can be made to the corresponding authors.

REFERENCES

  • Alexander, M. A., I. Bladé, M. Newman, J. R. Lanzante, N.-C. Lau, and J. D. Scott, 2002: The atmospheric bridge: The influence of ENSO teleconnections on air–sea interaction over the global oceans. J. Climate, 15, 22052231, https://doi.org/10.1175/1520-0442(2002)015<2205:TABTIO>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Alexander, M. A., D. J. Vimont, P. Chang, and J. D. Scott, 2010: The impact of extratropical atmospheric variability on ENSO: Testing the seasonal footprinting mechanism using coupled model experiments. J. Climate, 23, 28852901, https://doi.org/10.1175/2010JCLI3205.1.

    • Search Google Scholar
    • Export Citation
  • Amaya, D. J., 2019: The Pacific meridional mode and ENSO: A review. Curr. Climate Change Rep., 5, 296307, https://doi.org/10.1007/s40641-019-00142-x.

    • Search Google Scholar
    • Export Citation
  • Amaya, D. J., M. J. DeFlorio, A. J. Miller, and S.-P. Xie, 2017: WES feedback and the Atlantic meridional mode: Observations and CMIP5 comparisons. Climate Dyn., 49, 16651679, https://doi.org/10.1007/s00382-016-3411-1.

    • Search Google Scholar
    • Export Citation
  • Amaya, D. J., Y. Kosaka, W. Zhou, Y. Zhang, S.-P. Xie, and A. J. Miller, 2019: The North Pacific pacemaker effect on historical ENSO and its mechanisms. J. Climate, 32, 76437661, https://doi.org/10.1175/JCLI-D-19-0040.1.

    • Search Google Scholar
    • Export Citation
  • Anderson, B. T., 2007: On the joint role of subtropical atmospheric variability and equatorial subsurface heat content anomalies in initiating the onset of ENSO events. J. Climate, 20, 15931599, https://doi.org/10.1175/JCLI4075.1.

    • Search Google Scholar
    • Export Citation
  • Anderson, B. T., and R. C. Perez, 2015: ENSO and non-ENSO induced charging and discharging of the equatorial Pacific. Climate Dyn., 45, 23092327, https://doi.org/10.1007/s00382-015-2472-x.

    • Search Google Scholar
    • Export Citation
  • Anderson, B. T., R. C. Perez, and A. Karspeck, 2013: Triggering of El Niño onset through trade wind–induced charging of the equatorial Pacific. Geophys. Res. Lett., 40, 12121216, https://doi.org/10.1002/grl.50200.

    • Search Google Scholar
    • Export Citation
  • Anderson, J. L., 2001: An ensemble adjustment Kalman filter for data assimilation. Mon. Wea. Rev., 129, 28842903, https://doi.org/10.1175/1520-0493(2001)129<2884:AEAKFF>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Anderson, J. L., 2003: A local least squares framework for ensemble filtering. Mon. Wea. Rev., 131, 634642, https://doi.org/10.1175/1520-0493(2003)131<0634:ALLSFF>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • 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, 16871712, https://doi.org/10.1175/1520-0469(1989)046<1687:IVIATA>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Bjerknes, J., 1969: Atmospheric teleconnections from the equatorial Pacific. Mon. Wea. Rev., 97, 163172, https://doi.org/10.1175/1520-0493(1969)097<0163:ATFTEP>2.3.CO;2.

    • Search Google Scholar
    • Export Citation
  • Boulanger, J.-P., C. Menkes, and M. Lengaigne, 2004: Role of high- and low-frequency winds and wave reflection in the onset, growth and termination of the 1997–1998 El Niño. Climate Dyn., 22, 267280, https://doi.org/10.1007/s00382-003-0383-8.

    • Search Google Scholar
    • Export Citation
  • Cai, W., and Coauthors, 2019: Pantropical climate interactions. Science, 363, eaav4236, https://doi.org/10.1126/science.aav4236.

  • Cane, M. A., and S. E. Zebiak, 1985: A theory for El Niño and the Southern Oscillation. Science, 228, 10851087, https://doi.org/10.1126/science.228.4703.1085.

    • Search Google Scholar
    • Export Citation
  • Capotondi, A., P. D. Sardeshmukh, and L. Ricciardulli, 2018: The nature of the stochastic wind forcing of ENSO. J. Climate, 31, 80818099, https://doi.org/10.1175/JCLI-D-17-0842.1.

    • Search Google Scholar
    • Export Citation
  • Chang, P., L. Zhang, R. Saravanan, D. J. Vimont, J. C. H. Chiang, L. Ji, H. Seidel, and M. K. Tippett, 2007: Pacific meridional mode and El Niño–Southern Oscillation. Geophys. Res. Lett., 34, L16608, https://doi.org/10.1029/2007GL030302.

    • Search Google Scholar
    • Export Citation
  • Chiang, J. C. H., and D. J. Vimont, 2004: Analogous Pacific and Atlantic meridional modes of tropical atmosphere–ocean variability. J. Climate, 17, 41434158, https://doi.org/10.1175/JCLI4953.1.

    • Search Google Scholar
    • Export Citation
  • Delworth, T. L., and Coauthors, 2006: GFDL’s CM2 global coupled climate models. Part I: Formulation and simulation characteristics. J. Climate, 19, 643674, https://doi.org/10.1175/JCLI3629.1.

    • Search Google Scholar
    • Export Citation
  • Di Lorenzo, E., G. Liguori, N. Schneider, J. C. Furtado, B. T. Anderson, and M. A. Alexander, 2015: ENSO and meridional modes: A null hypothesis for Pacific climate variability. Geophys. Res. Lett., 42, 94409448, https://doi.org/10.1002/2015GL066281.

    • Search Google Scholar
    • Export Citation
  • Ding, R., J. Li, Y.-h. Tseng, C. Sun, and Y. Guo, 2015: The Victoria mode in the North Pacific linking extratropical sea level pressure variations to ENSO. J. Geophys. Res. Atmos., 120, 2745, https://doi.org/10.1002/2014JD022221.

    • Search Google Scholar
    • Export Citation
  • Ding, R., J. Li, Y.-h. Tseng, C. Sun, and F. Xie, 2017: Joint impact of North and South Pacific extratropical atmospheric variability on the onset of ENSO events. J. Geophys. Res. Atmos., 122, 279298, https://doi.org/10.1002/2016JD025502.

    • Search Google Scholar
    • Export Citation
  • Enfield, D. B., A. M. Mestas-Nuñez, D. A. Mayer, and L. Cid-Serrano, 1999: How ubiquitous is the dipole relationship in tropical Atlantic sea surface temperatures? J. Geophys. Res., 104, 78417848, https://doi.org/10.1029/1998JC900109.

    • Search Google Scholar
    • Export Citation
  • Fedorov, A. V., 2002: The response of the coupled tropical ocean–atmosphere to westerly wind bursts. Quart. J. Roy. Meteor. Soc., 128 (579), 123, https://doi.org/10.1002/qj.200212857901.

    • Search Google Scholar
    • Export Citation
  • Fedorov, A. V., S. Hu, M. Lengaigne, and E. Guilyardi, 2015: The impact of westerly wind bursts and ocean initial state on the development, and diversity of El Niño events. Climate Dyn., 44, 13811401, https://doi.org/10.1007/s00382-014-2126-4.

    • Search Google Scholar
    • Export Citation
  • Gill, A. E., 1980: Some simple solutions for heat-induced tropical circulation. Quart. J. Roy. Meteor. Soc., 106, 447462, https://doi.org/10.1002/qj.49710644905.

    • Search Google Scholar
    • Export Citation
  • Good, S. A., M. J. Martin, and N. A. Rayner, 2013: EN4: Quality controlled ocean temperature and salinity profiles and monthly objective analyses with uncertainty estimates. J. Geophys. Res. Oceans, 118, 67046716, https://doi.org/10.1002/2013JC009067.

    • Search Google Scholar
    • Export Citation
  • Ham, Y.-G., J.-S. Kug, J.-Y. Park, and F.-F. Jin, 2013: Sea surface temperature in the north tropical Atlantic as a trigger for El Niño/Southern Oscillation events. Nat. Geosci., 6, 112116, https://doi.org/10.1038/ngeo1686.

    • Search Google Scholar
    • Export Citation
  • Held, I. M., 2001: The partitioning of the poleward energy transport between the tropical ocean and atmosphere. J. Atmos. Sci., 58, 943948, https://doi.org/10.1175/1520-0469(2001)058<0943:TPOTPE>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Hu, R., T. Lian, J. Feng, and D. Chen, 2023: Pacific meridional mode does not induce strong positive SST anomalies in the central equatorial Pacific. J. Climate, 36, 41134131, https://doi.org/10.1175/JCLI-D-22-0503.1.

    • Search Google Scholar
    • Export Citation
  • Hu, S., and A. V. Fedorov, 2016: Exceptionally strong easterly wind burst stalling El Niño of 2014. Proc. Natl. Acad. Sci. USA, 113, 20052010, https://doi.org/10.1073/pnas.1514182113.

    • Search Google Scholar
    • Export Citation
  • Jin, F.-F., 1997a: An equatorial ocean recharge paradigm for ENSO. Part I: Conceptual model. J. Atmos. Sci., 54, 811829, https://doi.org/10.1175/1520-0469(1997)054<0811:AEORPF>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Jin, F.-F., 1997b: An equatorial ocean recharge paradigm for ENSO. Part II: A stripped-down coupled model. J. Atmos. Sci., 54, 830847, https://doi.org/10.1175/1520-0469(1997)054<0830:AEORPF>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Jin, F.-F., L. Lin, A. Timmermann, and J. Zhao, 2007: Ensemble-mean dynamics of the ENSO recharge oscillator under state-dependent stochastic forcing. Geophys. Res. Lett., 34, L03807, https://doi.org/10.1029/2006GL027372.

    • Search Google Scholar
    • Export Citation
  • Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77, 437472, https://doi.org/10.1175/1520-0477(1996)077<0437:TNYRP>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Kang, S. M., I. M. Held, D. M. W. Frierson, and M. Zhao, 2008: The response of the ITCZ to extratropical thermal forcing: Idealized slab-ocean experiments with a GCM. J. Climate, 21, 35213532, https://doi.org/10.1175/2007JCLI2146.1.

    • Search Google Scholar
    • Export Citation
  • Kang, S. M., D. M. W. Frierson, and I. M. Held, 2009: The tropical response to extratropical thermal forcing in an idealized GCM: The importance of radiative feedbacks and convective parameterization. J. Atmos. Sci., 66, 28122827, https://doi.org/10.1175/2009JAS2924.1.

    • Search Google Scholar
    • Export Citation
  • Kang, S. M., K. Park, Y.-T. Hwang, and W.-T. Hsiao, 2018: Contrasting tropical climate response pattern to localized thermal forcing over different ocean basins. Geophys. Res. Lett., 45, 12 54412 552, https://doi.org/10.1029/2018GL080697.

    • Search Google Scholar
    • Export Citation
  • Klein, S. A., B. J. Soden, and N.-C. Lau, 1999: Remote sea surface temperature variations during ENSO: Evidence for a tropical atmospheric bridge. J. Climate, 12, 917932, https://doi.org/10.1175/1520-0442(1999)012<0917:RSSTVD>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Larson, S. M., and B. P. Kirtman, 2014: The Pacific meridional mode as an ENSO precursor and predictor in the North American Multimodel Ensemble. J. Climate, 27, 70187032, https://doi.org/10.1175/JCLI-D-14-00055.1.

    • Search Google Scholar
    • Export Citation
  • Larson, S. M., and B. P. Kirtman, 2015: An alternate approach to ensemble ENSO forecast spread: Application to the 2014 forecast. Geophys. Res. Lett., 42, 94119415, https://doi.org/10.1002/2015GL066173.

    • Search Google Scholar
    • Export Citation
  • Larson, S. M., K. V. Pegion, and B. P. Kirtman, 2018: The South Pacific meridional mode as a thermally driven source of ENSO amplitude modulation and uncertainty. J. Climate, 31, 51275145, https://doi.org/10.1175/JCLI-D-17-0722.1.

    • Search Google Scholar
    • Export Citation
  • Lengaigne, M., E. Guilyardi, J.-P. Boulanger, C. Menkes, P. Delecluse, P. Inness, J. Cole, and J. Slingo, 2004: Triggering of El Niño by westerly wind events in a coupled general circulation model. Climate Dyn., 23, 601620, https://doi.org/10.1007/s00382-004-0457-2.

    • Search Google Scholar
    • Export Citation
  • Liguori, G., and E. Di Lorenzo, 2019: Separating the North and South Pacific meridional modes contributions to ENSO and tropical decadal variability. Geophys. Res. Lett., 46, 906915, https://doi.org/10.1029/2018GL080320.

    • Search Google Scholar
    • Export Citation
  • Linkin, M. E., and S. Nigam, 2008: The North Pacific Oscillation–west Pacific teleconnection pattern: Mature-phase structure and winter impacts. J. Climate, 21, 19791997, https://doi.org/10.1175/2007JCLI2048.1.

    • Search Google Scholar
    • Export Citation
  • Liu, T., Z. Liu, Y. Zhao, and S. Zhang, 2022: Investigating extratropical influence on the equatorial Atlantic zonal bias with regional data assimilation. J. Climate, 35, 61016117, https://doi.org/10.1175/JCLI-D-21-0610.1.

    • Search Google Scholar
    • Export Citation
  • Liu, T., Z. Liu, Y. Zhao, and S. Zhang, 2023: Subtropical impact on the tropical double-ITCZ bias in the GFDL CM2.1 model. J. Climate, 36, 38333847, https://doi.org/10.1175/JCLI-D-22-0432.1.

    • Search Google Scholar
    • Export Citation
  • Liu, Z., and S. Xie, 1994: Equatorward propagation of coupled air–sea disturbances with application to the annual cycle of the eastern tropical Pacific. J. Atmos. Sci., 51, 38073822, https://doi.org/10.1175/1520-0469(1994)051<3807:EPOCAD>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Liu, Z., and M. Alexander, 2007: Atmospheric bridge, oceanic tunnel, and global climatic teleconnections. Rev. Geophys., 45, RG2005, https://doi.org/10.1029/2005RG000172.

    • Search Google Scholar
    • Export Citation
  • Lu, F., and Z. Liu, 2018: Assessing extratropical influence on observed El Niño–Southern Oscillation events using regional coupled data assimilation. J. Climate, 31, 89618969, https://doi.org/10.1175/JCLI-D-17-0849.1.

    • Search Google Scholar
    • Export Citation
  • Lu, F., Z. Liu, Y. Liu, S. Zhang, and R. Jacob, 2017: Understanding the control of extratropical atmospheric variability on ENSO using a coupled data assimilation approach. Climate Dyn., 48, 31393160, https://doi.org/10.1007/s00382-016-3256-7.

    • Search Google Scholar
    • Export Citation
  • McPhaden, M. J., 2003: Tropical Pacific Ocean heat content variations and ENSO persistence barriers. Geophys. Res. Lett., 30, 1480, https://doi.org/10.1029/2003GL016872.

    • Search Google Scholar
    • Export Citation
  • McPhaden, M. J., 2012: A 21st century shift in the relationship between ENSO SST and warm water volume anomalies. Geophys. Res. Lett., 39, L09706, https://doi.org/10.1029/2012GL051826.

    • Search Google Scholar
    • Export Citation
  • McPhaden, M. J., X. Zhang, H. H. Hendon, and M. C. Wheeler, 2006: Large scale dynamics and MJO forcing of ENSO variability. Geophys. Res. Lett., 33, L16702, https://doi.org/10.1029/2006GL026786.

    • Search Google Scholar
    • Export Citation
  • Meinen, C. S., and M. J. McPhaden, 2000: Observations of warm water volume changes in the equatorial Pacific and their relationship to El Niño and La Niña. J. Climate, 13, 35513559, https://doi.org/10.1175/1520-0442(2000)013<3551:OOWWVC>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Menkes, C. E., M. Lengaigne, J. Vialard, M. Puy, P. Marchesiello, S. Cravatte, and G. Cambon, 2014: About the role of westerly wind events in the possible development of an El Niño in 2014. Geophys. Res. Lett., 41, 64766483, https://doi.org/10.1002/2014GL061186.

    • Search Google Scholar
    • Export Citation
  • Min, Q., J. Su, R. Zhang, and X. Rong, 2015: What hindered the El Niño pattern in 2014? Geophys. Res. Lett., 42, 67626770, https://doi.org/10.1002/2015GL064899.

    • Search Google Scholar
    • Export Citation
  • Min, Q., J. Su, and R. Zhang, 2017: Impact of the South and North Pacific meridional modes on the El Niño–Southern Oscillation: Observational analysis and comparison. J. Climate, 30, 17051720, https://doi.org/10.1175/JCLI-D-16-0063.1.

    • Search Google Scholar
    • Export Citation
  • Newman, M., S.-I. Shin, and M. A. Alexander, 2011: Natural variation in ENSO flavors. Geophys. Res. Lett., 38, L14705, https://doi.org/10.1029/2011GL047658.

    • Search Google Scholar
    • Export Citation
  • Penland, C., and P. D. Sardeshmukh, 1995: The optimal growth of tropical sea surface temperature anomalies. J. Climate, 8, 19992024, https://doi.org/10.1175/1520-0442(1995)008<1999:TOGOTS>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Polo, I., M. Martin-Rey, B. Rodriguez-Fonseca, F. Kucharski, and C. R. Mechoso, 2015: Processes in the Pacific La Niña onset triggered by the Atlantic Niño. Climate Dyn., 44, 115131, https://doi.org/10.1007/s00382-014-2354-7.

    • Search Google Scholar
    • Export Citation
  • Rayner, N. A., D. E. Parker, E. B. Horton, C. K. Folland, L. V. Alexander, D. P. Rowell, E. C. Kent, and A. Kaplan, 2003: Global analyses of sea surface temperature, sea ice, and night marine air temperature since the late nineteenth century. J. Geophys. Res., 108, 4407, https://doi.org/10.1029/2002JD002670.

    • Search Google Scholar
    • Export Citation
  • Rogers, J. C., 1981: The North Pacific Oscillation. Int. J. Climatol., 1, 3957, https://doi.org/10.1002/joc.3370010106.

  • Saji, N. H., B. N. Goswami, P. N. Vinayachandran, and T. Yamagata, 1999: A dipole mode in the tropical Indian Ocean. Nature, 401, 360363, https://doi.org/10.1038/43854.

    • Search Google Scholar
    • Export Citation
  • Schopf, P. S., and M. J. Suarez, 1988: Vacillations in a coupled ocean–atmosphere model. J. Atmos. Sci., 45, 549566, https://doi.org/10.1175/1520-0469(1988)045<0549:VIACOM>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Shin, Y., S. M. Kang, K. Takahashi, M. F. Stuecker, Y.-T. Hwang, and D. Kim, 2021: Evolution of the tropical response to periodic extratropical thermal forcing. J. Climate, 34, 63356353, https://doi.org/10.1175/JCLI-D-20-0493.1.

    • Search Google Scholar
    • Export Citation
  • Shu, Q., Y. Zhang, D. J. Amaya, S. M. Larson, Y. Kosaka, J.-C. Yang, and X. Lin, 2023: Role of ocean advections during the evolution of the Pacific meridional modes. J. Climate, 36, 43274343, https://doi.org/10.1175/JCLI-D-22-0296.1.

    • Search Google Scholar