Introduction

As the most dominant mode of interannual variability in the Pacific, El Niño–Southern Oscillation (ENSO) exerts significant global impacts on weather and climate through atmospheric and oceanic teleconnections1,2,3,4,5,6,7. Considerable research has been dedicated to elucidating the physical mechanisms responsible for ENSO formation2,8,9,10,11,12,13. In the tropical Pacific, atmospheric-oceanic processes, such as sustained westerly wind bursts14,15,16,17, zonal thermocline depth anomalies18, variations in trade wind intensity19,20,21, and zonal sea surface temperature (SST) gradients22, are recognized as key triggers of ENSO events. Beyond the tropical Pacific, remote climate systems, including the Pacific Meridional Mode23,24,25, Indian Ocean and North Atlantic SST anomalies (SSTAs)26,27,28, and the Tibetan Plateau thermal forcing29, also play significant roles in driving ENSO variability.

Recently, three-ocean interactions have attracted growing research interest30. Preceding investigations on the impacts of North Atlantic SSTAs on ENSO primarily examined either tropical North Atlantic (TNA) SSTAs or uniform extratropical North Atlantic SSTAs31,32,33,34,35,36,37. The TNA SSTAs modulate tropical atmospheric circulations via Gill-type Rossby and Kelvin wave responses27,28,38,39,40,41. For example, Ham et al.27 demonstrated that spring TNA warming can induce an anomalous lower-level cyclone on its western side and cold SSTAs in the subtropical eastern Pacific through a Gill-type Rossby wave response. These cold SSTAs trigger easterly wind anomalies over the equatorial central-western Pacific, favoring the development of a La Niña event. Additionally, the TNA warming generates easterly wind anomalies on its eastern side via a Gill-type Kelvin wave response. Such wind anomalies interact with climatological westerly winds, promoting warm SSTAs and positive heating anomalies over the northern Indian Ocean and Maritime Continent, reinforcing easterly wind anomalies over the equatorial Pacific and further providing a conducive condition for a La Niña event42,43,44.

The extratropical North Atlantic SSTAs can influence ENSO by modulating mid-latitude atmospheric and oceanic systems through triggering downstream-propagating Rossby wave trains. For example, a uniform mode of summer negative SSTAs in the extratropical North Atlantic can strengthen the East Asian winter monsoon by triggering Rossby wave trains34,35. These anomalies are further linked to equatorial westerly wind anomalies in the western Pacific during winter and subsequent spring, favoring the development of an El Niño event.

On the interannual timescale, the dominant mode of North Atlantic SSTAs displays as a tripole pattern, which is characterized by a negative-positive-negative (positive phase) or positive-negative-positive (negative phase) SSTAs structure over the TNA, mid-latitude North Atlantic (MNA), and high-latitude North Atlantic (HNA)45,46. Previous studies have demonstrated that there exists significant positive feedback between the North Atlantic Oscillation and North Atlantic tripole SSTAs mode47,48,49. Characterized by strong seasonal persistence, the North Atlantic tripole SSTAs mode plays an important role in regulating downstream weather and climate systems through the generation of mid-latitude Rossby wave trains50,51,52.

However, compared to the influence of TNA SSTAs or uniform extratropical North Atlantic SSTAs, the role of the North Atlantic tripole SSTAs mode in ENSO development has received limited attention, and the associated physical mechanisms remain inadequately understood. Furthermore, the quantitative contributions and underlying physical processes of the TNA, MNA, and HNA SSTAs associated with the North Atlantic tripole SSTAs mode in affecting ENSO events have not been well documented. Addressing these issues is crucial for advancing our understandings of the climate effects of the North Atlantic tripole SSTAs mode as well as improving predictive skills for ENSO events.

Results

Impacts of the North Atlantic tripole SSTAs mode on ENSO evolution

The North Atlantic tripole SSTAs mode exhibits remarkable persistence from winter to subsequent spring, which is most noticeable in early spring (March)45,46. As illustrated in Fig. 1a, the first leading empirical orthogonal function (EOF) mode of March North Atlantic SST displays as a distinct tripole pattern, characterized by negative SSTAs in the TNA (marked “A”) and the HNA (marked “C”), and positive SSTAs in the MNA (marked “B”). Thus, the North Atlantic tripole SSTAs index (NATSI = 2 × SSTB - SSTA - SSTC) is defined by simultaneously considering area-averaged SSTAs in the TNA, MNA, and HNA. The March NATSI is highly correlated with the principal component time series of the first leading EOF mode (NA_PC1) of March North Atlantic SST (r = 0.80, p < 0.01, Fig. 1b). Furthermore, the March NATSI shows statistically significant correlations (p < 0.01) with NATSI from January to May, demonstrating the strong temporal persistence of the North Atlantic tripole SSTAs mode. Therefore, the March NATSI is a reasonable proxy representing the winter–spring North Atlantic tripole SSTAs mode, which is consistent with previous studies46.

Fig. 1: Statistical relationship between the North Atlantic tripole SSTAs mode and ENSO in observations.
Fig. 1: Statistical relationship between the North Atlantic tripole SSTAs mode and ENSO in observations.
Full size image

a Spatial pattern of the first leading EOF mode of March North Atlantic SST. Value in the upper right is the explained variance. b Time series of the normalized NA_PC1 (red solid curve) and NATSI (black solid curve). The shading represents ±0.8 standard deviations. Correlations between the March NATSI and the following c autumn and d winter SSTs. e–f Similar to (c–d) but for the partial correlations with the signal of previous winter Niño3.4 index removed. The stippled areas in (c–f) denote the significant values exceeding the 95% confidence level.

The March NATSI is significantly and positively correlated with the subsequent autumn (r = 0.34, p < 0.05, Fig. 1c) and winter (r = 0.41, p < 0.01, Fig. 1d) Niño3.4 indices. Moreover, the partial correlation coefficients in subsequent autumn (r = 0.27, p < 0.1, Fig. 1e) and winter (r = 0.35, p < 0.05, Fig. 1f) remain significant after removing the previous winter Niño3.4 index. These results indicate that the winter–spring North Atlantic tripole SSTAs mode is significantly associated with the development of ENSO events in subsequent autumn–winter seasons.

To elucidate the underlying physical mechanisms responsible for this cross-basin influence, we conduct a series of numerical experiments using the coupled Community Earth System Model (CESM1.2.2). The experimental design (Table 1) and methodology are detailed in the Methods section. To evaluate the robustness of the model performance, we calculate the first leading EOF mode of March North Atlantic SST based on the last 43 years of the 250-yr CTRL experiment (Fig. S1a). It exhibits a distinct tripole pattern and bears large resemblance to the observational counterpart (Fig. 1a). Besides, model result shows that the PC1 of March North Atlantic SST is significantly and positively correlated with the subsequent autumn (r = 0.31, p < 0.05, Fig. S1b) and winter (r = 0.40, p < 0.01, Fig. S1c) Niño3.4 indices. The partial correlation coefficients in subsequent autumn (r = 0.35, p < 0.05, Fig. S1d) and winter (r = 0.39, p < 0.01, Fig. S1e) remain significant after removing the previous winter Niño3.4 index. It indicates that the correlation between the simulated North Atlantic tripole SSTAs index (PC1) and ENSO are comparable to that in observations (Fig. 1). That is, the CESM1.2.2 can well represent the North Atlantic tripole SSTAs variability and capture its observed correlation with ENSO.

Table 1 Numerical experimental design
Full size table

To amplify the model response in the sensitivity experiment, we employ the positive-minus-negative composite SSTA of NATSI as the external forcing. Notably, the impact of positive-minus-negative composite SSTA forcing is approximately twice that of the averaged SSTA forcing (i.e., positive-minus-climatological mean, and negative-minus-climatological mean). To evaluate the robustness of the model result, we compare the model output against observations, both derived from the positive-minus-negative composite. Figure S2a shows the observational September-October-November-December (SOND) composite differences in SST between strong positive and negative years based on the ±0.8 standard deviations of NATSI. The SOND Niño3.4 index from the observations (1.07 °C, Fig. S2a) is close to that from the model’s positive-minus-negative composite (1.01 °C, Fig. 2c). This indicates that the model result is reasonable and credible.

Fig. 2: Seasonal evolution of SST and atmospheric circulation anomalies between the NATM and the NACLM experiments.
Fig. 2: Seasonal evolution of SST and atmospheric circulation anomalies between the NATM and the NACLM experiments.
Full size image

Composite differences in SST (shading, °C) and 850-hPa wind (vectors, m s−1) between the NATM and the NACLM experiments in a MAM, b JJA, and c SOND. d–f Similar to (a–c) but for the precipitation (mm day−1). The stippled areas in (a–f) and black vectors in (a–c) denote the significant values exceeding the 95% confidence level.

According to the difference between the NATM and the NACLM experiments, the SSTAs over the equatorial central-eastern Pacific are weak and insignificant in March-April-May (MAM), but become positive and significant in June-July-August (JJA), and develop to a strong El Niño event in SOND (Fig. 2a–c). That is, the positive phase of the winter–spring North Atlantic tripole SSTAs mode can stimulate a SOND El Niño event. Specifically, significant lower-level westerly wind anomalies appear over the equatorial western Pacific in MAM (Fig. 2a), which can trigger the eastward and downward propagating oceanic Kelvin wave, deepening the equatorial central-eastern Pacific thermocline (Fig. 3d–f). Besides, the westerly wind anomalies can also transport warm water from the equatorial western Pacific to the central-eastern Pacific (Fig. 2a–c). Therefore, significantly positive SSTAs appear over the equatorial central-eastern Pacific from JJA to SOND, which induce ascending (descending) motions over the equatorial central-eastern (western) Pacific and enhances the lower-level westerly wind anomalies in the western Pacific (Fig. 3a–c). In summary, the MAM lower-level westerly wind anomalies over the equatorial western Pacific play a crucial role in triggering a SOND El Niño event by the above Bjerknes positive feedback2. Next, we investigate how the North Atlantic tripole SSTAs mode induces the MAM westerly wind anomalies over the equatorial western Pacific.

Fig. 3: Seasonal evolution of atmospheric circulation and ocean potential temperature anomalies between the NATM and the NACLM experiments.
Fig. 3: Seasonal evolution of atmospheric circulation and ocean potential temperature anomalies between the NATM and the NACLM experiments.
Full size image

Composite differences in vertical cross-section of zonal circulation (vectors, m s−1; vertical speed is scaled by 100) and vertical velocity (shading, 0.01 Pa s−1) averaged along 5°S–5°N between the NATM and the NACLM experiments in a MAM, b JJA, and c SOND. d–f Similar to (a–c) but for the vertical cross-section of ocean potential temperature (shading, °C) averaged along 5°S–5°N. The green and black solid lines represent the thermocline depth in the NATM and the NACLM experiments, respectively. The shading and black vectors in (a–c) and stippled areas in (d–f) denote the significant values exceeding the 95% confidence level.

As a component of the North Atlantic tripole SSTAs mode, negative TNA SSTAs can cause negative precipitation and heating anomalies over the tropical western Atlantic in MAM (Fig. 2d), which can trigger an anomalous lower-level anticyclone on its western side (centered around 20°N/100°W) through a Gill-type Rossby wave response. On the western flank of the anomalous anticyclone, southerly wind anomalies weaken the climatological northeasterly trade winds, reducing evaporation and upward surface latent heat flux over the subtropical eastern Pacific. Through the wind-evaporation-SST feedback53, these conditions favor the development of positive SSTAs (Fig. 2a). The positive SSTAs, together with the atmospheric warm and wet advection, induce more convection in situ and an anomalous lower-level cyclone on its western side (centered around 25°N/170°W) by a Gill-type Rossby wave response. On the southern flank of the anomalous cyclone, the MAM westerly wind anomalies appear over the equatorial western Pacific (Fig. 2a).

Apart from the above tropical pathway, this study also reveals that the positive phase of winter–spring North Atlantic tripole SSTAs mode can contribute to an El Niño event through the mid-latitude pathway. Specifically, the positive phase of North Atlantic tripole SSTAs mode can trigger a mid-latitude Rossby wave train, which propagates from the North Atlantic to Asia and the North Pacific, resulting in anomalous upper-level negative geopotential height and cyclone over South Asia (centered around 15°N/90°E) and positive geopotential height and anticyclone over the subtropical North Pacific (centered around 20°N/170°E) in April-May-June (AMJ) (Fig. 4a). Notably, the mid-latitude Rossby wave train response in MAM is similar to that in AMJ, with the strength is a little bit weak (figure not shown). Thus, the upper-level westerly wind anomalies over the tropical Indian Ocean and easterly wind anomalies over the tropical western Pacific converge over the Maritime Continent, inducing descending motions and lower-level divergence (Fig. 4b–c). Finally, the late spring lower-level westerly wind anomalies appear over the equatorial western Pacific (Fig. 4c). The above process is referred to as the gearing between the Indian and Pacific Oceans (GIP)54. The GIP process describes the coupled zonal-vertical circulations between the tropical Indian Ocean and the Pacific, characterized by a clockwise circulation in one region and an anticlockwise circulation in the other.

Fig. 4: Atmospheric circulation anomalies in AMJ between the NATM and the NACLM experiments.
Fig. 4: Atmospheric circulation anomalies in AMJ between the NATM and the NACLM experiments.
Full size image

Composite differences in AMJ between the NATM and the NACLM experiments. a 200-hPa geopotential height (shading, gpm) and stationary wave activity flux (vectors, m2 s−2). b 200-hPa geopotential height (shading, gpm) and wind (vectors, m s−1). c 500-hPa vertical velocity (shading, 0.01 Pa s−1) and 850-hPa wind (vectors, m s−1). The stippled areas and black vectors denote the significant values exceeding the 95% confidence level.

Additionally, the mid-latitude teleconnection Rossby wave train induced by the positive phase of North Atlantic tripole SSTAs mode can result in anomalous equivalent barotropic negative geopotential height and cyclone over the central North Pacific (centered around 40°N/170°W), inducing an anomalous lower-level cyclone (Figs. 4a and 2a). The southwesterly wind anomalies on the southeast flank of the lower-level cyclone weaken the climatological northeasterly trade winds over the subtropical North Pacific, inducing positive SSTAs by the wind-evaporation-SST feedback (Fig. 2a). The positive SSTAs over the subtropical North Pacific can modulate atmospheric circulations, leading to the westerly wind anomalies over the equatorial central-western Pacific. This process is referred to as the seasonal footprinting mechanism (SFM)55,56. It is noticed that the positive SSTAs over the subtropical North Pacific and the anomalous lower-level cyclone over the central North Pacific are induced by both tropical Gill-type Rossby wave and the mid-latitude SFM process.

In summary, the positive phase of winter–spring North Atlantic tripole SSTAs mode can contribute to the following autumn–winter El Niño event through both tropical and mid-latitude pathways. The former is associated with the Gill-type Rossby wave response, while the latter involves the GIP and SFM processes.

Differential impacts of TNA, MNA, and HNA SSTAs on ENSO evolution

In this section, we explore the relative contributions and underlying physical mechanisms of TNA, MNA, and HNA SSTAs associated with the North Atlantic tripole SSTAs mode in affecting ENSO events. The composite differences in SST and 850-hPa winds between the TNA and the NACLM experiments in MAM, JJA, and SOND are shown in Fig. 5a–c. We find that, associated with the winter–spring negative TNA SSTAs, the equatorial western Pacific is prevailed by significant lower-level westerly wind anomalies in MAM, JJA, and SOND, which can induce positive SSTAs in the equatorial central-eastern Pacific. These positive SSTAs are insignificant in MAM and gradually become significant in JJA and SOND due to the Bjerknes positive feedback. As a consequence, a strong autumn–winter El Niño event occurs with the SOND Niño 3.4 index equaling 0.80 °C. In the differences between the MNA and the NACLM experiments (Fig. 5d–f), and between the HNA and the NACLM experiments (Fig. 5g–i), the lower-level westerly wind anomalies also appear over the equatorial western Pacific from MAM to SOND, but with smaller strengths relative to those in the difference between the TNA and the NACLM experiments. Therefore, the positive SOND SSTAs over the equatorial central-eastern Pacific and SOND Niño 3.4 indices (0.67 °C and 0.48 °C) induced by the MNA and HNA SSTAs are also weaker than those induced by the TNA SSTAs. Through comparing the SOND Niño 3.4 indices obtained from these three sets of experiments, the TNA, MNA, and HNA SSTAs of the North Atlantic tripole SSTAs mode contribute about 43%, 34%, and 23% to the subsequent SOND El Niño events, respectively.

Fig. 5: Seasonal evolution of SST and atmospheric circulation anomalies in the TNA, MNA, and HNA experiments compared with the NACLM experiment.
Fig. 5: Seasonal evolution of SST and atmospheric circulation anomalies in the TNA, MNA, and HNA experiments compared with the NACLM experiment.
Full size image

Composite differences in SST (shading, °C) and 850-hPa wind (vectors, m s−1) between the TNA and the NACLM experiments in a MAM, b JJA, and c SOND. d–f Similar to (a–c) but for the composite differences between the MNA and the NACLM experiments. g–i Similar to (a–c) but for the composite differences between the HNA and the NACLM experiments. The stippled areas and black vectors denote the significant values exceeding the 95% confidence level.

According to the difference between the TNA and the NACLM experiments, we find that the winter–spring negative TNA SSTAs can contribute to a subsequent autumn–winter El Niño event through both tropical and mid-latitude pathways. The tropical pathway is analogous to that in the difference between the NAT and the NACLM experiments. Specifically, the negative TNA SSTAs trigger the southerly wind anomalies over the subtropical eastern Pacific by a Gill-type Rossby wave response, which induces the positive SSTAs by the wind-evaporation-SST feedback (Fig. 5a). These positive SSTAs can lead to an anomalous lower-level cyclone on its western side. On the southern flank of the anomalous cyclone, MAM westerly wind anomalies appear over the equatorial central-western Pacific, stimulating a subsequent El Niño event by the Bjerknes positive feedback (Fig. 5a–c).

Additionally, accompanied with the winter–spring negative TNA SSTAs, upper-level negative (positive) geopotential height anomalies appear over the southern (northern) North Atlantic in AMJ, and two mid-latitude Rossby trains are initiated (Fig. 6a). One is originated in the southern North Atlantic, and propagates along West Africa, and South Asia, resulting in upper-level negative geopotential height anomalies over South Asia. The other Rossby wave train propagates along the southern North Atlantic, the northern North Atlantic, Europe, East Asia, the central North Pacific, and the subtropical North Pacific, inducing upper-level positive (negative) geopotential height anomalies over the subtropical North Pacific (central North Pacific). On the southern flank of negative (positive) geopotential height anomalies over South Asia (the subtropical North Pacific), the upper-level tropical Indian Ocean (western Pacific) is prevailed by westerly (easterly) wind anomalies, which converge over the Maritime Continent (Fig. 6b). Consequently, descending motions and lower-level divergence appear over the Maritime Continent, further inducing lower-level westerly wind anomalies over the equatorial western Pacific by the GIP process (Fig. 6c).

Fig. 6: Atmospheric circulation anomalies in AMJ between the TNA and the NACLM experiments.
Fig. 6: Atmospheric circulation anomalies in AMJ between the TNA and the NACLM experiments.
Full size image

Similar to Fig. 4 but for the composite differences between the TNA and the NACLM experiments.

On the other hand, negative geopotential height anomalies over the central North Pacific induced by the negative TNA SSTAs are equivalent barotropic (Figs. 6a and 5a). Thus, the lower-level central North Pacific is occupied with an anomalous cyclone, which weakens the climatological northeasterly trade winds and further results in positive SSTAs over the subtropical North Pacific by the wind-evaporation-SST feedback (Fig. 5a). The positive SSTAs over the subtropical North Pacific can lead to westerly wind anomalies over the equatorial central-western Pacific by the SFM process.

Based on the difference between the MNA and the NACLM experiments, an El Niño event occurs in SOND, which is mainly generated by the mid-latitude pathway. Specifically, associated with the winter–spring positive MNA SSTAs, significantly positive geopotential height anomalies appear over the upper-level central North Atlantic in AMJ (Fig. 7a). A Rossby wave train originated in the central North Atlantic propagates downstream to Asia and the Pacific, resulting in upper-level negative geopotential height anomalies over South Asia and the central North Pacific and positive geopotential height anomalies over the subtropical North Pacific (Fig. 7a). The GIP and SFM processes are also obvious in inducing the lower-level westerly wind anomalies over the equatorial western Pacific in the difference between the MNA and the NACLM experiments (Fig. 7a–c and 5d). The upper-level negative geopotential height anomalies over South Asia and descending motions over the Maritime continent in the difference between the MNA and the NACLM experiments are slightly smaller than those in the difference between the TNA and the NACLM experiments (Figs. 6 and 7), suggesting that the GIP process resulted from the positive MNA SSTAs are less significant than that induced by the negative TNA SSTAs. Besides, the upper-level negative geopotential height anomalies over the central North Pacific in the difference between the MNA and the NACLM experiments are stronger than those in the difference between the TNA and the NACLM experiments (Figs. 6 and 7), implying that the SFM process associated with the positive MNA SSTAs is stronger than that induced by the negative TNA SSTAs. Particularly, the SOND Niño3.4 index in the difference between the MNA and the NACLM experiments is smaller than that in the difference between the TNA and the NACLM experiments (Fig. 5c and f), indicating that the strength of El Niño events would weaken without the tropical pathway.

Fig. 7: Atmospheric circulation anomalies in AMJ between the MNA and the NACLM experiments.
Fig. 7: Atmospheric circulation anomalies in AMJ between the MNA and the NACLM experiments.
Full size image

Similar to Fig. 4 but for the composite differences between the MNA and the NACLM experiments.

The winter–spring negative HNA SSTAs can stimulate a SOND El Niño event through the mid-latitude pathway based on the result obtained from the difference between the HNA and the NACLM experiments. Specifically, associated with the negative HNA SSTAs, negative geopotential height anomalies appear over the upper-level northeastern North Atlantic in AMJ (Fig. 8a). A Rossby wave train originated in the northeastern North Atlantic propagates downstream to Asia and the Pacific, resulting in upper-level negative geopotential height anomalies over South Asia and the central North Pacific and positive geopotential height anomalies over the subtropical North Pacific (Fig. 8a). The GIP and SFM processes are also significant in inducing the lower-level westerly wind anomalies over the equatorial western Pacific in the difference between the HNA and the NACLM experiments (Figs. 8a–c and 5g). Based on the upper-level geopotential height anomalies over the tropical Indo-Pacific and 500-hPa descending motions over the Maritime continent, the GIP process initiated by the negative HNA SSTAs are less significant than that induced by the positive MNA SSTAs (Figs. 7 and 8). The SFM process induced by the negative HNA SSTAs are also less significant than that induced by the positive MNA SSTAs, because the upper-level negative geopotential height anomalies over the central North Pacific associated with negative HNA SSTAs are weaker than those associated with the positive MNA SSTAs (Figs. 7 and 8). The SOND Niño3.4 index in the difference between the HNA and the NACLM experiments is the smallest. Thus, the strength of El Niño events induced by the HNA SSTAs is the smallest as compared to that induced by the TNA and MNA SSTAs (Fig. 5).

Fig. 8: Atmospheric circulation anomalies in AMJ between the HNA and the NACLM experiments.
Fig. 8: Atmospheric circulation anomalies in AMJ between the HNA and the NACLM experiments.
Full size image

Similar to Fig. 4 but for the composite differences between the HNA and the NACLM experiments.

Discussion

Substantial efforts have been devoted to addressing the impacts of North Atlantic SSTAs on ENSO, predominantly focusing on either TNA SSTAs or uniform extratropical North Atlantic SSTAs27,34,35,42,43,44. However, as the dominant mode of the North Atlantic SSTAs on the interannual timescale, how the North Atlantic tripole SSTAs mode affects ENSO remains poorly understood. In addition, the relative contribution of three individual components (i.e., TNA, MNA, and HNA SSTAs) of the North Atlantic tripole SSTAs mode is even rarely considered. Addressing these issues can enrich the physical understanding of ENSO evolution and provide valuable insights for the North Atlantic tripole SSTAs mode and its climate impacts.

A schematic diagram of how the positive phase of North Atlantic tripole SSTAs mode influence El Niño events in late spring is shown in Fig. 9. Observational evidences and model results indicate that the positive phase of winter–spring North Atlantic tripole SSTAs mode can result in the MAM westerly wind anomalies over the equatorial western Pacific by triggering the tropical westward Gill-type Rossby wave response and mid-latitude eastward Rossby wave trains, subsequently contributing to an autumn–winter El Niño event by the Bjerknes positive feedback. Specifically, the tropical westward Gill-type Rossby wave response is mainly related to negative TNA SSTAs, which can result in a MAM anomalous lower-level anticyclone on its western side. This anomalous anticyclone results in positive SSTAs over the subtropical eastern Pacific by the wind-evaporation-SST feedback, further inducing westerly wind anomalies over the equatorial western Pacific.

Fig. 9: Schematic diagram of how the positive phase of North Atlantic tripole SSTAs mode influence El Niño events in late spring.
Fig. 9: Schematic diagram of how the positive phase of North Atlantic tripole SSTAs mode influence El Niño events in late spring.
Full size image

a Upper-level atmospheric circulation anomalies. b SST and lower-level atmospheric circulation anomalies.

The mid-latitude eastward Rossby wave trains are generated by the North Atlantic tripole SSTAs mode, which can influence ENSO events by the combined effects of the GIP and SFM processes. In the GIP process, the positive phase of North Atlantic tripole SSTAs mode can stimulate a downstream-propagating Rossby wave train to South Asia and the subtropical North Pacific, resulting in upper-level convergence and descending motions over the Maritime Continent in late spring. As a consequence, lower-level westerly wind anomalies occur over the equatorial western Pacific. Furthermore, the positive phase of North Atlantic tripole SSTAs mode can also stimulate a downstream-propagating Rossby wave train, resulting in equivalent barotropic negative geopotential height anomalies and anomalous cyclone over the central North Pacific. The lower-level anomalous cyclone leads to positive SSTAs over the subtropical North Pacific and consequently westerly wind anomalies in the equatorial central-western Pacific in late spring by the SFM process. Ultimately, the MAM westerly wind anomalies over the equatorial central-western Pacific induced by the above tropical and mid-latitude pathways can contribute to a subsequent autumn–winter El Niño event by the Bjerknes positive feedback.

Based on a series of numerical experiments, we have quantified the relative contributions of the TNA, MNA, and HNA SSTAs associated with the North Atlantic tripole SSTAs mode. The results signify that the TNA SSTAs whose contribution is about 43%, can generate an El Niño event through both tropical and mid-latitude pathways. In contrast, the MNA and HNA SSTAs can contribute to an El Niño event only through the mid-latitude pathway, accounting for 34% and 23% proportions, respectively.

According to a Gill-type Kelvin wave response, negative TNA SSTAs would theoretically induce the lower-level westerly wind anomalies over the tropical North Indian Ocean57. However, lower-level easterly wind anomalies occur in this region (Fig. 4c), implying that the Kelvin wave response associated with the negative TNA SSTAs is relatively weak in the current study. The weakened Kelvin wave response may be explained by two aspects. One is the GIP process, which results in easterly wind anomalies over the tropical North Indian Ocean and consequently weaken the effect of the Kelvin wave response. The other is the small area of the TNA, which results in a weaker Gill-type Kelvin wave response compared to the basin-wide tropical Atlantic region. In addition, this study reveals that the North Atlantic tripole SSTAs mode can modulate ENSO variability mainly through the tropical and mid-latitude pathways. Nevertheless, the quantitative assessment of their relative contributions requires a further systematic investigation.

In this study, to amplify the model response, we employ the positive-minus-negative composite SSTA as the external forcing. Although this composite primarily represents the positive phase, it inherently incorporates signals from both positive and negative phases. Furthermore, due to the antisymmetric feature of the SSTA forcing between the two phases, the response to the positive-minus-negative composite (representing the positive phase) is expected to be the opposite of that to the negative-minus-positive composite (representing the negative phase). Therefore, only the model results for the positive-minus-negative composite SSTA are presented in this study. Additionally, to compare the influences of the positive and negative phases of the North Atlantic tripole SSTA mode on ENSO, Fig. S2b and c present the observational SOND composite differences in SST between strong positive/negative years and the climatological mean, respectively. The corresponding SOND Niño3.4 index values are 0.52 °C for the positive phase (Fig. S2b) and -0.55 °C for the negative phase (Fig. S2c). That is, the positive phase result is nearly antisymmetric to the negative phase result.

Methods

Data and methods

The monthly mean SST is obtained from the Hadley Centre Sea Ice and Sea Surface Temperature Data Set version 1 (HadISST1) with 1° × 1° resolution and for the period of 1980–202258. The monthly mean model output includes both oceanic and atmospheric variables. The oceanic variables comprise SST and potential temperature. The atmospheric variables include geopotential height, temperature, horizontal wind, vertical velocity, and precipitation.

In this study, the linear trends of all monthly mean variables are removed. The 2–9-year bandpass Lanczos filter is employed to emphasize the interannual variability59. The wave activity flux is applied to diagnose the propagation of the Rossby wave train60. The Niño3.4 index, defined as area-averaged SSTAs over 5°S–5°N/170°–120°W, is used to depict the intensity of ENSO events. The statistical significances of the results are assessed based on the two-tailed Student’s t-test.

Numerical experimental design

The coupled general circulation model (CGCM) experiments are designed to explore the physical mechanisms of the North Atlantic tripole SSTAs mode in influencing ENSO events. Furthermore, the relative contributions and physical processes of the TNA, MNA, and HNA SSTAs associated with the North Atlantic tripole SSTAs mode are also investigated. The CGCM adopted in this study is the CESM version 1.2.2, which is developed by the National Center for Atmospheric Research (NCAR)61. The atmospheric component is the Community Atmosphere Model (CAM) version 4 with a horizontal resolution of 1.9° × 2.5°62. The oceanic component is an extension of the Parallel Ocean Program (POP) version 2, which is an irregular grid (approximately 1° × 1°)63.

Six sets of experiments are designed in this study (Table 1). The control run (CTRL) experiment is integrated for 250 years with freely coupled atmospheric and oceanic components. The other five experiments are sensitivity experiments, which are integrated for 30 years each. Their initial fields are taken from 1 January of the last years in the CTRL experiment. In the NACLM sensitivity experiment, the North Atlantic (10°–70°N, 80°–10°W) is prescribed with the observed climatological mean SST annual cycle (January-December) during 1980–2010, whereas the ocean and atmosphere are freely coupled elsewhere. In the NATM sensitivity experiment, the SST from 1 January to 31 May in every year over the North Atlantic (10°–70°N, 80°–10°W) is replaced by observed climatological mean SST plus SSTAs, which are calculated by the composite difference between strong positive and negative years based on the ±0.8 standard deviations of March NATSI. The other is set to the same as the NACLM experiment. The primary reason for adopting the observed climatological mean SST is the presence of systematic bias in the model’s climatological mean SST. Using the model’s climatological mean SST combined with the observed North Atlantic tripole SSTAs would result in a forcing field that deviates from the actual observed SST conditions. As a consequence, the model response driven by such a biased SST forcing may diverge considerably from observations. By using the observed climatological mean SST combined with the observed North Atlantic tripole SSTAs, the simulated response more closely approximates real-world conditions42.

The TNA, MNA, and HNA sensitivity experiments are the same as the NATM experiment, with the SST forcings separately changed over the TNA (10°–40°N, 80°–10°W), MNA (20°–60°N, 80°–10°W), and HNA (40°–70°N, 80°–10°W) regions. To get the negative, positive, and negative SSTAs over the TNA, MNA, and HNA, respectively, the observed composite negative SSTAs over the TNA and HNA are reserved, and the positive SSTAs are set to zero. The observed composite positive SSTAs over the MNA are reserved, and the negative SSTAs are set to zero.

Consequently, the difference between the NATM and the NACLM experiments is analyzed to assess the impact of the North Atlantic tripole SSTAs mode on ENSO events. The differences between the TNA and the NACLM experiments, between the MNA and the NACLM experiments, and between the HNA and the NACLM experiments are used to investigate the impacts of negative TNA, positive MNA, and negative HNA SSTAs on ENSO events, respectively. Through comparing the above experiment results, we can investigate the relative contributions of TNA, MNA, and HNA SSTAs involved in ENSO evolution.