Complex Influences of Tropical Indian and Atlantic Oceans on ENSO Prediction

Complex Influences of Tropical Indian and Atlantic Oceans on ENSO Prediction | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Complex Influences of Tropical Indian and Atlantic Oceans on ENSO Prediction Ke Jiang, Jiye Wu, Jing-Jia Luo This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-5352476/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 07 Jun, 2025 Read the published version in Climate Dynamics → Version 1 posted 5 You are reading this latest preprint version Abstract Tropical basin interactions (TBIs) are widely involved in El Niño-Southern Oscillation (ENSO) evolution. Previous dynamical prediction studies often focused on single basin, single start month, and few events, which normally overlooked the complexity in the TBIs’ impact on ENSO prediction. To address these limitations, we conducted six sets of sensitivity hindcast experiments initializing from different seasons during 1983–2018, in which observed monthly and climatological sea surface temperatures (SSTs) over the tropical Indian (TIO) and Atlantic Ocean (TAO) are separately and synchronously prescribed. Results indicate synergistic but complicated roles of tropical SST anomalies (SSTAs) outside the Pacific in predicting ENSO. The results suggest more prominent contributions from TAO SSTAs, due to the well-captured teleconnections between ENSO and primary precursors in the North tropical and equatorial Atlantic. Conversely, the model exhibits large biases in replicating the relationship between ENSO and the basin-wide and dipole modes in the Indian Ocean, weakening the TIO SSTAs’ contributions. Moreover, SSTAs over the remote basins exert asymmetrical and phase-dependent influences on ENSO predictions; more remarkable contributions are found during La Niña and ENSO transition-development phases, indicating the TBIs’ importance in improving the spring barrier of ENSO prediction. Additionally, the impact of TBIs on ENSO prediction displays an interdecadal change; SSTAs outside the Pacific improve (degrade) El Niño prediction before (after) 2000, which may be associated with rapid warming in the TIO and TAO. Our results suggest high complexity in the TIO and TAO’s influence on ENSO prediction, stimulating future efforts for better understanding and models’ performance. ENSO predictions Tropical basin interactions Complexity Model deficiency Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Figure 10 1. Introduction The El Niño-Southern Oscillation (ENSO) is the most dominant interannual variability in the tropics, and its extensive environmental and climatic impacts resonate worldwide (e.g., Bjerknes, 1969 ; Philander, 1983 ; McPhaden et al., 1998 ). Continuous efforts have been made to improve the understanding and prediction of ENSO. Since the early 1990s, with significant progress in ENSO dynamics, climate modeling, data assimilation, as well as the adoption of deep learning, ENSO prediction has been notably improved (e.g., Palmer et al., 2004; Luo et al., 2008 ; Ham et al., 2019 ; Ren et al., 2020 ; Lyu et al. 2024 ). However, ENSO prediction encountered bottlenecks in past decades (e.g., Barnston et al., 2012 ), with previous studies indicating an obvious decline in predictive skill in the early Twenty-First Century (e.g., Hu et al. 2020 ; Zhao et al. 2016 ). Conventionally, the long-term predictability of ENSO is rooted in the Pacific atmosphere-ocean coupled processes and oceanic waves (e.g., Bjerknes, 1969 ; Suarez and Schopf, 1988 ; Jin, 1997 ), thereby wind patterns, oceanic heat content and/or thermocline variations over the equatorial Pacific serve as key precursors of ENSO prediction (e.g., Meinen and McPhaden, 2000 ; Tseng et al., 2017; Fang and Mu, 2018 ). However, the tropical basins are tightly interconnected (e.g., Cai et al., 2019 ; Wang, 2019 ). Growing attentions have been paid to the precursors in the tropical Indian and Atlantic Oceans for the ENSO evolution (e.g., Luo et al., 2010 and 2017 ; Ding et al., 2012 ; Exarchou et al., 2021 ; Jiang and Li, 2021 ; Jiang et al., 2022 ; Jin et al., 2023 ). The warming in the equatorial Atlantic, often called Atlantic Niño, can modify the Walker Circulation (e.g., Polo et al., 2015 ; Wang et al., 2009 ) and induce anomalous easterly winds, triggering an eastward-propagating Kelvin wave to favor the development of La Niña, and vice versa for El Niño development (e.g., Latif and Grötzner, 2000 ; Keenlyside and Latif, 2007 ; Jiang and Li, 2021 ; Wang et al., 2024 ). Moreover, the low-level cyclone to the north-west of the North Tropical Atlantic (NTA) sea surface temperature (SST) warming, produces a cold SST anomaly (SSTA) through enhanced evaporation and cold advection, which further induces low-level anticyclone in the tropical northwestern Pacific. The corresponding equatorial easterly also favors the development of La Niña (e.g., Ham et al., 2013 ; Ham and Kug, 2015 ; Polo et al., 2015 ; Zhang et al., 2021 ; Jiang and Li, 2021 ). The Indian Ocean SST warming also serves as an important driver of easterly wind anomalies in the western Pacific. During the El Niño peak phase, the Indian Ocean Basin Mode (IOBM) predominantly accelerates the rapid transition from El Niño to La Niña by inducing easterly and generating upwelling Kelvin waves in the equatorial western Pacific. During the El Niño development phase, the Indian Ocean Dipole (IOD) is crucial for the growth of El Niño (e.g., Kug et al., 2006 ; Ohba and Ueda, 2009 ; Luo et al., 2010 ; Ohba and Watanabe, 2012 ; Izumo et al., 2016 ; Ha et al., 2017 ; Fan and Meng, 2023 ). Unraveling the inter-basin interactive processes underpinning the ENSO development provides new insight and opportunity for extending ENSO predictability (e.g., Luo et al., 2010 and 2017 ; Frauen and Dommenget, 2012 ; Doi et al., 2017 ; Alexander et al., 2022 ). Recently, utilizing conceptual and statistical models, the importance of tropical Indian and Atlantic Oceans in predicting ENSO has been emphasized (e.g., Zhao et al., 2024 ; Li et al., 2024 ), especially in overcoming the long-standing spring prediction barrier (e.g., Zhao et al., 2023 ; Jin et al., 2023 ). Additionally, based on the dynamical models, the SST variability in the individual tropical basins (the Indian Ocean or Atlantic) is demonstrated to be crucial (e.g., Frauen and Dommenget, 2012 ; Keenlyside et al., 2013 ; Alexander et al., 2022 ), and studies also demonstrated the collective roles of the two tropical basins (e.g., Luo et al., 2017 ; Fan et al., 2024 ), but with a focus on the analyses of a few specific cases and extreme events. Previous studies have confirmed the impact of inter-basin interactions on ENSO through observational studies, statistical models, and dynamical model experiments. However, the complexity of this issue is quite remarkable. For instance, the active roles of inter-basin interactions on interannual timescale in ENSO development and predictions are rebutted in a few recent studies (e.g., Zhang et al., 2021 ; Jiang et al., 2021 ). In addition, some studies suggest that both the tropical Pacific-Indian Ocean and Pacific-Atlantic teleconnections exhibit interdecadal change over the last decades (e.g., Park et al., 2019; Xue et al., 2022 ; Wang et al. 2024 ). Furthermore, the models' capacity to accurately simulate the observed ocean inter-basin interactions remains contentious, which affects their contributions to ENSO predictions. These raise a sequence of new questions, requiring further comprehensions about the complicated influence of the tropical SSTs outside the Pacific on ENSO predictions. This study aims to identify the roles of the tropical Indian and Atlantic Oceans on ENSO prediction during the last four decades and to explore underlying possible mechanisms. 2. Data and Methods 2.1 Observational Data This study utilizes the following observational and reanalysis datasets: (a) monthly Optimum Interpolation Sea Surface Temperature (OISST) from National Oceanic and Atmospheric Administration (NOAA; Reynolds et al., 2002 ) is utilized for model coupled SST-nudging and hindcast validation; (b) monthly mean 10-m horizontal winds are derived from the National Centers for Environmental Prediction (NCEP)-Department of Energy Atmospheric Model Intercomparison Project II reanalysis (Kanamitsu et al., 2002 ); (c) monthly precipitation data from the Global Precipitation Climatology Project (Adler et al., 2003). The analysis period of this study is 1983–2018. ENSO is represented by the Niño3.4 index (5°S-5°N, 170°-120°W). Additionally, the SSTA averaged over the NTA (5°-25°N, 70°-15°W), equatorial eastern Atlantic (4°S-4°N, 20°W-0°) and Indian Ocean Basin (20°S-20°N, 40–110°E) regions, referred as to NTA, Atl3 and IOBM indices, are used to represent the NTA warming/cooling, Atlantic Niño/Niña, and the Indian Ocean wide basin warming/cooling, respectively. The IOD is represented by the DMI index, defined as the differences in SSTAs between the western (10°S-10°N, 50°E-70°E) and eastern (10°S-0°, 90°-110°E) tropical Indian Ocean (Saji et al., 1999 ). The Warm Water Volume (WWV) refers to the volume of water in the equatorial Pacific (5°S-5°N, 120°E-80°W) that is warmer than 20°C, which is crucial for assessing the heat content of the upper ocean in the Pacific (Meinen and McPhaden, 2000 ). The prediction skills are evaluated by calculating the anomaly correlation coefficient (ACC) and root-mean-square error (RMSE) between the observations and model hindcasts: $$\:ACC\left(\tau\:\right)=\frac{\sum\:_{t=1}^{n}\left[a\left(t\right)b\left({t}_{0},\tau\:\right)\right]}{\sqrt{\sum\:_{t=1}^{n}{a}^{2}\left(t\right)}\sqrt{\sum\:_{t=1}^{n}{b}^{2}\left({t}_{0},\tau\:\right)}}$$ $$\:RMSE\left(\tau\:\right)=\sqrt{\frac{1}{n}\sum\:_{i=1}^{n}{\left[a\left(t\right)-b\left({t}_{0},\tau\:\right)\right]}^{2}}$$ where \(\:\text{a}\left(\text{t}\right)\) represents the observation at month \(\:\text{t}\) , and \(\:\text{b}\left({\text{t}}_{0},{\tau\:}\right)\) represent the corresponding model hindcast at a lead month of \(\:{\tau\:}\) initializing from time \(\:{\text{t}}_{0}\) . For statistical analysis, the significance of regression and correlation coefficients is determined through a two-tailed Student's t-test, and the comparison between two correlation coefficients employs Steiger’s z-test (Meng et al., 1992 ). 2.2 The Coupled Model and Sensitivity Hindcast Experiments To quantitatively explore the possible impacts of tropical inter-basin interactions on ENSO prediction skills, six sets of model sensitivity hindcast experiments using the SINTEX-F (Luo et al., 2005 ) are performed, which are organized into three groups (referred to AO, IO, and AO + IO group, respectively). Each group consists of one sensitivity (Obs) and one control (Cli) experiment. In the AO group, observed climatological and monthly SSTs are specified over the tropical Atlantic (20°S-20°N, 70°W-15°E) in the control and sensitivity experiments, respectively. The IO group follows the same design as the AO group, but with the observed climatological and monthly SSTs being specified over the tropical Indian Ocean (20°S-20°N, 40°-120°E). In the AO + IO group, the observed climatological and monthly SSTs are specified in both the tropical Indian and Atlantic Oceans regions. Each experiment comprises of nine ensemble members produced by three SST-nudging initializations and three coupling schemes following Luo et al. ( 2008 ). The hindcasts are initialized from 1st day of February, May, August, and November of 1983–2018 and then integrated for one year; details are summarized in Table 1 . By comparing the results of control and sensitivity experiments, the influence of SSTAs over the specific tropical basins on the ENSO prediction can be measured. Table 1 Description of the three groups (six sets) of model hindcast experiments. Name Description Ensemble hindcast Cli_AO Monthly climatological mean SSTs of 1983–2012 are specified in the tropical Atlantic (20°S-20°N, 70°W-15°E), and atmosphere-ocean is freely coupled in other areas. Number of ensemble member: 9 Initialized from 1st day of February, May, August and November of 1983–2018 Forecast length: 12 months Obs_AO Observed monthly SSTs are specified in the tropical Atlantic (20°S-20°N, 70°W-15°E), and atmosphere-ocean is freely coupled in other areas. Cli_IO Monthly climatological mean SSTs of 1983–2012 are specified in the tropical Indian Ocean (20°S-20°N, 40°-120°E), and atmosphere-ocean is freely coupled in other areas. Obs_AO Observed monthly SSTs are specified in the tropical Indian Ocean (20°S-20°N, 40°-120°E), and atmosphere-ocean is freely coupled in other areas. Cli_AO + IO Same as the Cli experiments above, but with monthly climatological mean SSTs being specified in both the tropical Indian and Atlantic Oceans. Obs_AO + IO Same as the Obs expereimnts above, but with observed monthly SSTs being specified in both the tropical Indian and Atlantic Oceans. Note : The differences between each set of experiments (Obs and Cli) can be analyzed to assess the individual and collective impacts of SSTs in the tropical Atlantic and Indian Oceans on the climate prediction in the tropical Pacific. 2.3 Multiple Regression Method Allowing for a compatible comparison with the results of our experiments, the Pacific intrinsic dynamics and the effects of the various precursors in the Indian and Atlantic Oceans on ENSO needed to be elucidated in the observation. Herein we build a multi-element linear regression model depicting the ENSO’s responses to four dominant climate modes (i.e., NTA, Atlantic Niño/Niña, IOBM, and IOD) following previous studies (Luo et al., 2017 ; Jiang et al., 2022 ). This equation was given as follows: $$\:DJF\:Niño\:3.4\left(\tau\:\right)=a\times\:MAM\:NTA\left(\tau\:\right)+b\times\:JJA\:Atl3\left(\tau\:\right)+c\times\:MAM\:IOBM\left(\tau\:\right)+d\times\:SON\:DMI\left(\tau\:\right)+e\times\:ND\left(-1\right)J\:WWV\:anomalies\left(\tau\:\right)+f$$ Where DJF, MAM, JJA, SON, and ND(-1)J represent the months during and before ENSO years ( \(\:{\tau\:})\) . The regression coefficients derived from the observational data spanning 1983 to 2018 are \(\:\text{a}\:\) = -1.76, \(\:\text{b}\) = -0.16, \(\:\text{c}\) = 1.46, \(\:\text{d}\) = 0.85, and \(\:\text{e}\) = -0.08, with the constant \(\:\text{f}\) = -0.04. To mitigate the influence of ENSO autocorrelation, we reconstruct the equation to reflect the impacts of major tropical SST variabilities outside the Pacific: $$\:reconstructed\:DJF\:Niño\:3.4\left(\tau\:\right)=a\times\:MAM\:NTA\left(\tau\:\right)+b\times\:JJA\:Atl3\left(\tau\:\right)+c\times\:MAM\:IOBM\left(\tau\:\right)+d\times\:SON\:DMI\left(\tau\:\right)$$ 3. Overall impacts of tropical Indian and Atlantic Oceans SSTs on ENSO prediction To generally understand the impacts of tropical Indian and Atlantic Oceans SSTs on ENSO prediction, we evaluate both the ACC and RMSE skill of the Niño 3.4 index prediction in all sets of experiments. Figure 1 a presents the all-season ACC skills for the Niño3.4 index from 1983 to 2018 across various experiments, with the lead time ranging from 1 to 12 months. At the early prediction stage, there are no notable differences between the Obs and Cli experiments in each group, since the initial conditions used for all the hindcast experiments are the same. However, as the lead time increases, the ACC differences among experiments become pronounced, with the Obs experiments consistently demonstrating higher ACC. Simultaneously, the lower RMSEs (Fig. 1 b) further confirm the better skill of the Obs experiments in long-term ENSO forecasts compared to the Cli experiment. Our results not only align with previous findings that highlight the importance of tropical Indian and Atlantic Oceans SSTs in the evolution and prediction of ENSO (e.g., Jansen et al., 2009 ; Luo et al., 2010 and 2017 ; Frauen and Dommenget, 2012 ; Exarchou et al., 2021 ), but also reveal the complex impact of the individual basin. Notably, in the IO group, the ACC improvement of the Obs experiment compared to the Cli experiment is smaller than that of the other two groups at almost all lead times. In particular, at lead times beyond 11 months, the ACC of the Obs_IO experiment even falls below that of the Cli_IO experiment. In contrast, the ACC differences between the Obs and Cli experiment in the AO and AO + IO groups are uniformly positive and getting large with increasing lead time. In general, the Obs_AO exhibits the highest ACC skill and the Obs_AO + IO exhibits the lowest RMSE. The results may suggest that SSTAs in the tropical Atlantic display a larger influence on ENSO prediction than the Indian Ocean does, while the joint influence of SSTAs from the two basins contributes to more stable and reliable predictions at longer lead times. The detailed spatial distribution of ACC differences (Fig. 2 ) shows that the prediction of tropical Pacific SSTs is generally improved at long lead times (beyond 6 months) by prescribing observed SSTAs in the tropical Indian and Atlantic Oceans. The significant skill improvement emerges in the central equatorial Pacific at 3-month lead, and the improvement becomes more apparent and extends to the far eastern Pacific as the lead time increases in the AO + IO group (Fig. S1 m-r, Fig. 2 g-i). At the lead time of 12 months, the difference reaches 0.3 over the equatorial central-eastern Pacific (Fig. 2 i). Similar skill difference is displayed in the experiments of the AO group (Fig. 2 a-c, Fig. S1 a-f), of which is slightly smaller than AO + IO group at the same lead time, but exhibits a broader area of skill improvement, encompassing the entire tropical Pacific except for the Northwest Pacific. In comparison, the skill difference in the IO group exhibits relatively weak magnitude and complex spatial patterns (Fig. 1 d-f). Besides a slight skill improvement in the central-western equatorial Pacific, a surprising skill decrease originates over the eastern Pacific and extends westward with the increase in lead time (Fig. 2 d-f, Fig. S2g-l). Based on the detailed ACC differences, prescribing observed SSTAs in the tropical Indian and Atlantic Oceans significantly enhances the prediction skill of tropical Pacific SSTs, especially over the central equatorial Pacific. Considering the possible seasonal dependence of the ENSO’s response to the tropical Indian and Atlantic Oceans SSTAs, we further analyzed the prediction skills of the Niño 3.4 indices across different start months and target seasons. Similar to the all-season prediction skill assessment (Fig. 1 ), the contribution from the Atlantic SSTA surpasses that of the Indian Ocean and remains consistently influential throughout the ENSO evolution (Fig. 3 ). Interestingly, the ACC differences are manifested 3 months after initializing from February, 9 months after initializing from August, and 6 months after initializing from November across all three groups of the hindcast experiments (Fig. 3 a-c). The target seasons of these predictions correspond to the transition phase of ENSO, when the atmosphere-ocean coupling in the Pacific is relatively weak, suggesting the critical role of the tropical Indian and Atlantic Oceans in contributing to the predictability of ENSO during its transition and developing phases. In addition, the influence of the Indian and Atlantic Oceans SSTAs on the prediction skill of the ENSO varies across different target seasons (Fig. 3 d-f). During the boreal spring (MAM) and summer (JJA) when the ENSO typically onsets or decays, including the forcing from SSTs outside the Pacific significantly enhances the prediction. Notably, the prediction skill targeted in the summer season is severely poor in the Cli experiments, which is markedly improved in the Obs experiments, especially for the AO and AO + IO groups. This suggests a deteriorated spring prediction barrier (SPB) when excluding the influence of the tropical Indian and Atlantic Oceans SSTs in the ENSO prediction. By contrast, when ENSO typically reaches maturity during boreal autumn (SON) and winter (DJF), the contribution of SSTs from the external oceans to the ENSO prediction is somewhat less pronounced. This underscores that the importance of inter-basin connections in predicting ENSO varies across different phases of the ENSO cycle. 4. Complex and Diverse Impacts of Inter-Basin Interactions on ENSO Prediction 4.1 Asymmetrical impacts on El Niño and La Niña prediction During the period 1983–2018, 11 El Niño and 13 La Niña events are identified (summarized in Table 2 ). The definition follows the criteria of the NOAA Climate Prediction Center, i.e., El Niño (La Niña) events are defined when the three-month running mean of Niño 3.4 index is above 0.5°C (below − 0.5°C) for five consecutive months. Given that prediction skill differences between the two sets of hindcast experiments of each group increase more rapidly from February onward (recall Fig. 3 a-c), the following analysis focuses on composite results from predictions initialized on 1st February. Table 2 Identified El Niño (La Niña) events during the period of 1983–2018. El Niño events La Niña events 1986/87; 1987/88; 1991/92; 1994/95; 1997/98; 2002/03; 2006/07; 2009/10; 2014/15; 2015/16; 2018/19 (11 events in total) 1983/84; 1984/85; 1988/89; 1995/96; 1998/99; 1999/2000; 2000/01; 2005/06; 2007/08; 2008/09;2010/11; 2011/12; 2017/18 (13 events in total) As shown in Fig. 4 , the tropical Indian Ocean and Atlantic Ocean SSTs contribute positively to the prediction of La Niña events, particularly in the equatorial central Pacific during its onset and development phases (recall Fig. 3 c). In contrast, for El Niño events, the prediction skill is not significantly enhanced during the early stage (Fig. 4 f-i). Quite unexpectedly, taking into account the SSTs from these two oceans acts to reduce rather than increase the prediction skill in many regions at mid-long lead times (after July). Particularly, the skill degradation over the east-central Pacific is the most apparent (Fig. 4 g). To better understand how the tropical Indian and Atlantic Oceans SSTs act on the prediction of ENSO evolution, we further explore the difference of tropical SST, precipitation and 10m winds between the Obs and Cli experiments based on the composite of ENSO events. In MAM of the La Niña developing year, a remarkable positive SST anomaly is observed over the NTA region, which induces enhanced convection and thus gives rise to a low-level cyclonic flow over the subtropical northeastern Pacific as a Matsuno-Gill type Rossby-wave response (Fig. 5 a and S2a). Additionally, the NTA warming generates a Kelvin wave response over the tropical Indian Ocean and propagates eastward to the western Pacific (Fig. 5 a-c, Fig. S2a-c). During JJA, Atlantic Niño developed, resulting in an anomalous ascending motion over the Atlantic and anomalous subsidence over the central Pacific (Fig. 5 b-c, Fig. S2b-c). The induced easterly wind anomalies over the central and western equatorial Pacific can stimulate oceanic upwelling Kelvin waves and thus favor the development of La Niña (Fig. 5 d and S2d). The effects of the two Atlantic modes in predicting the development of El Niño in our model accord with previous observational studies (e.g., Ham et al., 2013 ; Jiang and Li, 2021 ; Jiang et al., 2022 ). Apart from the signals in the Atlantic, the weak warm SSTA in the eastern Indian Ocean during the SON promotes and sustains anomalous easterly winds in the western Pacific, thus also contributing to La Niña development (Fig. 5 c). In comparison, the experimental results based on the composite of El Niño events (Fig. 5 e-h) are intricate. As demonstrated in previous studies (e.g., Annamalai et al., 2005 ; Hameed, 2018 ), the positive IOD co-develops with El Niño (Fig. 5 e-h, Fig. S3e-h). However, in contrary to these studies, the positive IOD provokes easterly wind anomalies in the western Pacific, possibly owing to the strong SST warming in the western Indian Ocean, thus hindering the development of El Niño in our model (Fig. S3f-g). In addition, in preceding MAM, the tropical Indian Ocean basin-wide warming also induces an easterly wind response in the western Pacific (Fig. S3e). These unexpected forcings from the Indian Ocean jointly deteriorate the prediction of El Niño (Fig. 4 ). By contrast, the signals over the tropical Atlantic, primarily featuring the negative NTA in MAM and Atlantic Niña in JJA, lead to westerly wind responses in the western Pacific (Fig. S2e-f). This in turn favors the warm SSTA in the equatorial central-eastern Pacific (Fig. S2e-h) and improves the prediction of El Niño (Fig. 4 ). Interestingly, the competing contributions from the SSTAs in the tropical Indian and Atlantic Oceans ultimately lead to a negligible difference between the Obs and Cli experiments of AO + IO group (Fig. 5 e-h). In conclusion, the influence of the tropical Indian and Atlantic Oceans on the occurrence and evolution of ENSO events is asymmetric. SSTAs in the tropical Atlantic significantly contribute to the onset and evolution of ENSO, while SSTAs in the Indian Ocean play a crucial role in the rapid transitions between El Niño and La Niña. 4.2 Influence of Tropical Indo-Pacific and Atlantic-Pacific teleconnection on ENSO Predictions In general, the IOBM, IOD, NTA, and Atlantic Niño/Niña, are the four primary precursors in the tropical Indian and Atlantic Oceans for ENSO prediction. In the observation, the reconstructed DJF Niño 3.4 index based on the indices of the four precursors using the multiple regression model (see Methods) can well reproduce the observed Niño 3.4 value (correlation reachs 0.75). This confirms the critical role of remote forcings from these tropical basins outside the Pacific in involving ENSO evolutions. Note that removing the linear trend of these indices has almost no impact on the results. Based on the observational data, the CCs between the NTA index in MAM, Atl3 index in JJA and reconstructed Niño 3.4 index in DJF are − 0.72 and − 0.27, respectively (Fig. 6 a-b), implying considerable contributions to El Niño/La Niña from preceding SST anomalies over the northern tropical and equatorial Atlantic. These teleconnections are well represented by the model; the corresponding CCs based on the difference between the Obs_AO + IO and Cli_AO + IO experiments are − 0.61, and − 0.24, respectively (Fig. 6 f-g), which is close to the observational counterparts albeit with weaker values. Correspondingly, the observational pathways from the forcing of NTA warming/cooling (Fig. 7 a, e) and Atlantic Niño/Niña (Fig. 7 i, m) to the equatorial Pacific are generally reproduced, with the excitation of easterly/westerly wind anomalies and resulted SSTA cooling/warming in the equatorial Pacific Ocean. This success in reproducing the Atlantic-Pacific connections may accounts for the improved ENSO prediction by incorporating the observed tropical Atlantic SSTA (recall Figs. 1 and 2 ). However, it is worth noting that the responses of the surface winds and SST over the equatorial Pacific in the model experiments are weaker than those in the observation, which is responsible for the slight underestimation of the relationship between ENSO and the Atlantic precursors (recall Fig. 6 a, b, f, g). Regarding the Indo-Pacific connection, the reconstructed Niño 3.4 index in DJF is weakly correlated with IOBM index in MAM (CC = -0.21; Fig. 4 d), while strongly correlated with DMI in SON (CC = 0.73; Fig. 6 e) in the observation. However, as revealed by the difference between Obs_AO + IO and Cli_AO + IO experiments, the predicted relationship between the spring IOBM and the following winter ENSO is overestimated, where the CC between DJF Niño 3.4 and MAM IOBM index is -0.64. In the observation, the IOBM-induced convection excites moderate wind response over the tropical Pacific only during MAM (Fig. 8 a), which contributes to the ENSO transition in the spring (e.g., Kug and Kang, 2006). While in the model, obvious convection anomaly in the Maritime Continent (MC) and its wind response in the western Pacific persists across the spring to summer, and thus largely influence the ENSO development (Fig. 8 e-f). In stark contrast, the prescribed IOD in autumn shows a negligible correlation with the predicted ENSO in the subsequent winter (CC = 0.06). As shown in the previous study (e.g., Ashok et al., 2001 ), the observed positive IOD is coupled with the strong updrafts over the western Indian Ocean and descending branch over the MC (Fig. 8 i), which induces anomalous westerly winds in the western Pacific. The observed westerly wind responses are not replicated by the model experiments (Fig. 8 m). This failure in reproducing the Indo-Pacific interactions may be responsible for the absence of skill improvement by incorporating the observed tropical Indian Ocean SSTAs (recall Figs. 1 and 2 ). It is worth noting that it remains controversial that whether the Indian Ocean actively force the atmosphere or passively respond to the atmosphere. Further studies are warranted to better comprehend the intricate air-sea interactions in the tropical Indian Ocean. To verify the representation of three-ocean interactions in different model predictions, we also selected hindcast data from six dynamical models (i.e., CanCM4i, CMC1-CanCM3, CMC2-CanCM4, GEM-NEMO, GFDL-SPEAR, NCAR-CESM1) participating in the North American Multi-Model Ensemble (NMME) project (Becker et al., 2014 ). In addition, we also employ the hindcast data of the NUIST-CFS1.0, which is a real-time climate prediction system configured with the same coupled model for our experiment (i.e., the SINTEX-F model). Detailed specifications of these hindcast datasets are provided in Table S1 . As in the aforementioned analysis for the reconstructed observation and experimental results, the CCs of the four indices (i.e. NTA, Atl3, DMI, and IOBM) with the Niño 3.4 index are calculated (Fig. 9 ). The uniformly negative correlations between the NTA and Niño 3.4 indices are displayed across all the selected models, although some individual models such as the CMC1_CanCM3 overestimate the correlation while other models such as the GFDL_SPEAR and GEM_NEMO severely underestimate the correlations (Fig. 9 a). Notably, several models poorly capture the impact of Atlantic Niño/Niña on ENSO, particularly the GFDL_SPEAR model exhibits an incorrect positive correlation (Fig. 9 b). Additionally, consistent with the results revealed by our experiments, the impact of the IOBM on ENSO is generally overestimated across all the selected models except for the GFDL-SPEAR and GEM-NEMO. The latter even displays a positive CC between the IOBM and Niño 3.4 indices, which is opposed to the observation (Fig. 9 d). Likewise, similar to the NUIST-CFS1.0, about a half of the selected NMME models endure shortcomings in underestimating the positive correlation between the DMI and Niño 3.4 indices, in particular the CC in the NCAR-CESM1 is near zero (Fig. 9 c). In summary, there are still great challenges and large uncertainties in accurately capturing the tropical inter-basin interactions in the dynamical model prediction systems. From an actual prediction perspective, this limitation will hamper the potential of incorporating tropical SST variations outside the Pacific to enhance ENSO prediction accuracy. 4.3 Interdecadal Change in the Inter-Basin Interactions’ Influence on ENSO Prediction Generally speaking, the aforementioned complexity arises from the asymmetry and interannual cycle of ENSO, the distinction of SST variability in each basin, and the model deficiencies. Apart from them, previous studies have indicated obvious interdecadal change in the ENSO prediction skill and tropical inter-basin interactions. Typically, the ENSO prediction skill has degraded since 2000 (e.g., Hu et al. 2020 ). Surprisingly, compared to the contributions before 2000 (pre-2000), the skill improvement by incorporating the observational SSTAs in the tropical Indian and Atlantic Oceans also becomes less substantial after 2000 (post-2000; Fig. S4). Here we further compare the impact of SSTA over the tropical Indian and Atlantic Oceans on the prediction of each ENSO event during the past four decades. The negative difference of the predicted Niño 3.4 index between the Obs_AO + IO and Cli_AO + IO experiments is found for all of La Niña years except the 1984 and 2008 (Fig. 10 b), implying steadily positive contributions of tropical SSTAs outside the Pacific to La Niña prediction without a clear interdecadal change. In contrast, the contribution from tropical Atlantic and Indian Ocean SSTAs to El Niño prediction exhibits obvious multi-decadal change (Fig. 10 a). Notably, SSTAs over tropical Atlantic and Indian Ocean positively contribute to predicting all El Niños in pre-2000, while negatively contributes to predicting most of El Niños after 2000. The negative contributions may be largely responsible for the skill decline after prescribing the observational SST over the tropical Indian and Atlantic Oceans (recall Figs. 4 a, 5 f-h). To further investigate the possible causes of the interdecadal change in the impact of tropical Atlantic and Indian Ocean SSTAs on El Niño prediction, we conducted a synthetic analysis focusing on the two decadal periods (i.e., pre-2000 and post-2000). During El Niño developing years in pre-2000, the strong NTA cooling in MAM (Fig. 10 c, Fig. S5a) and Atlantic Niña in JJA (Fig. 10 d, Fig. S5b) are observed. A significant onset, development, and peak of positive IOD occur in spring, summer, and autumn, respectively (Fig. 10 c-e, Fig. S6a-c). These precursors jointly induce significant westerly wind anomalies over the equatorial western Pacific and thus facilitate the El Niño growth (Fig. 10 c-e). However, these tropical precursors outside the Pacific are muted in post-2000, and their contributions to El Niño development shift to a negative influence (Fig. 10 g-i), which is possibly owing to the rapid warming over the tropical Indian and Atlantic Oceans over the last decades (e.g., Luo et al., 2012; McGregor et al., 2014 ) as well as the changed relationship among three oceans. It is worth noting that either the cold or warm SSTAs over the tropical Indian Ocean in spring and summer seems to inhibit the development of El Niño during the periods of pre-2000 and post-2000 in our experiments (Fig. S6a-b, e-f), which may be associated with the model deficiencies as was discussed above. The results show that, in autumn, the strong positive IOD in pre-2000 provokes a westerly wind anomaly to improve the El Niño prediction (Fig. S6c). In contrast, the weak eastern pole cooling and the strengthened warming over most tropical Indian Ocean basin lead to stronger easterly anomalies over the Pacific in SON of post-2000, which worsens the prediction of El Niño development (Fig. S6g-h). In the tropical Atlantic, particularly in the equatorial eastern Atlantic, the composited SSTA in El Niño years are much different in pre-2000 and post-2000 (Fig. 10 ; Fig. S5). The weakened Atlantic Niña and the significant West African coastal warming provide unfavorable precondition of El Niño over the equatorial Pacific during post-2000. Overall, the warming trend in the tropical Indian and Atlantic Oceans, and the weakened tropical Indian Ocean-Pacific relationship (e.g., Xue et al., 2022 ) and Atlantic-Pacific relationship (e.g., Zhang et al., 2023) during pot-2000 may collectively reduce the positive contributions of the SSTAs in the tropical Indian and Atlantic Oceans to the El Niño development, further affecting the predictability of ENSO. 5. Conclusion and Discussion As a prominent topic in the community of climate science, pantropical interactions gain extensive attention in observational and modeling studies (e.g., Ham et al., 2013 ; Izumo et al., 2016 ; Jin et al., 2023 ; Fan et al., 2024 ), which is also an issue of debate (e.g., Zhang et al., 2021 ). Several studies over the last decade have suggested the possible contributions of SST anomalies in the tropical Indian Ocean and/or Atlantic on the ENSO predictability (e.g., Luo et al. 2010 , 2017 ; Frauen and Dommenget, 2012 ; Keenlyside et al., 2013 ; Alexander et al., 2022 ). In this study, based on the SINTEX-F coupled model, six sets of sensitivity hindcast experiments are conducted and compared to understand to what extent and how tropical Indian and Atlantic Oceans SSTs jointly and separately impact ENSO prediction since the 1980s. As an extension of earlier studies, this study intergrates ENSO cases in recent decades based on more predictions initialized from four seasons, focusing on the impacts from different tropical basins. Consistent with Keenlyside et al. ( 2013 ) with the study period being only up to 2005, our results confirm that incorporating the realistic Atlantic SST into the model leads to significant skill increases across the entire tropical Pacific, particularly at longer lead times. Furthermore, when contemporaneously specifying the model’s SST in both the Indian and Atlantic Oceans to the observed, larger improvement is exhibited in the equatorial central Pacific, emphasizing the compounded role of the two tropical basins outside the Pacific in enhancing ENSO prediction. If the SST in the tropical Indian and Atlantic Oceans is perfectly predicted, the useful skill in predicting Niño 3.4 can be extended to one year when initializing from every season. Notably, if the influence of the tropical Indian and Atlantic Oceans are excluded, the skills are significantly decreased for ENSO prediction in target seasons from boreal spring to summer, during which ENSO undergoes a rapid phase transition and exhibits low inherent predictability. Hence, incorporating accurate inter-basin interactions into dynamical models may help alleviate the spring predictability barrier, which was also noted in previous studies based on statistical models (e.g., Ren et al., 2019 ; Jin et al., 2023 ). Apart from the importance of tropical basins outside the Pacific in ENSO prediction, our results also suggest the complexities in understanding and predicting the inter-basin connection. For instance, the impacts of SSTAs in the tropical Indian and Atlantic Oceans on the predictive skills for El Niño and La Niña events are asymmetrical. It appears that the La Niña prediction benefits more than that the El Niño prediction does. This however conflicts with a few recent studies, which demonstrated the dominance of the tropical Indian and Atlantic Oceans in boosting super El Niño (e.g., Wang and Wang, 2021 ) based on observations and perfect model hindcast experiments (e.g., Fan et al., 2024 ). This inconsistence may arise from large discrepancies of three-ocean connections between the observation and model prediction, since Fan et al. ( 2024 ) mainly focused on the difference among their experiments rather than a comparison between the observation and model results. In addition, our analysis associates this asymmetry to the interdecadal change in the impact of tropical SSTAs outside the Pacific on El Niño prediction. The influence of the tropical Indian and Atlantic Oceans on El Niño prediciton has weakened since 2000, and their corresponding contributions to El Niño prediction changes from positive to negative values. This may arise from the rapid warming in tropical Indian and Atlantic Oceans and the interdecadal variations in the pantropical interactions. Additionally, the impacts of the Indian Ocean are yet hard to be thoroughly understood due to the existence of model imperfections. Based on our experimental results, the observed SSTA in the Indian Ocean has limited contributions to ENSO prediction in general, even worsening the predictive skills in the eastern Pacific, which contradicts with previous studies (e.g., Luo et al., 2010 and 2017 ; Jin et al., 2023 ). However, this may result from the inadequate representations of the Indo-Pacific connections. The relationships between the two dominant precursors in the Indian Ocean (i.e., spring IOBM and autumn IOD) and wintertime ENSO in the model experiment apparently deviate from the observed. Note that this challenge is commonly suffered by many dynamical models in the NMME (Fig. 9 ) and the CMIP6 models (figure not shown). This underscores the need for further efforts to improve the model’s capability in simulating the tropical inter-basin interactions accurately. Declarations Conflict of interests. The authors declare no conflicts of interest or competing interests. Ethical Approval. This declaration is “not applicable”. Acknowledgments. This work is supported by National Natural Science Foundation of China (Grant No. 42030605 and 42088101). The numerical calculations in this study were conducted in the High Performance Computing Center of Nanjing University of Information Science & Technology. Data Availability Statement. All observational, reanalysis, and model hindcast datasets are publicly available. OISST data at https://psl.noaa.gov/data/gridded/data.noaa.oisst.v2.html. Global Precipitation Climatology Project precipitation data at https://psl.noaa.gov/data/gridded/data.gpcp.html. NCEP2 reanalysis data at https://psl.noaa.gov/data/gridded/data.ncep.reanalysis2.html. NMME hindcast data at http://iridl.ldeo.columbia.edu/SOURCES/.Models/.NMME/. References Adler RF, and Coauthors (2003) The Version-2 Global Precipitation Climatology Project (GPCP) Monthly Precipitation Analysis (1979-Present). J Hydrometeorology, 4(6), 1147–1167. Alexander MA, Shin S-I, and Battisti DS (2022) The Influence of the Trend, Basin Interactions and Ocean Dynamics on Tropical Ocean Prediction. Geophys Res Lett, 49, e2021GL096120. Annamalai H, Xie SP, Mccreary JP, and Murtugudde R (2005) Impact of Indian Ocean Sea Surface Temperature on Developing El Niño. J Clim, 18, 302–319. Ashok K, Guan Z, and Yamagata T (2001) Impact of the Indian Ocean dipole on the relationship between the Indian monsoon rainfall and ENSO. Geophys Res Lett, 28, 4499–4502. Barnston AG, Tippett MK, L'heureux ML, Li S, and Dewitt DG (2012) Skill of Real-Time Seasonal ENSO Model Predictions during 2002-11: Is Our Capability Increasing? Bull Amer Meteor Soc, 93, 631–651. Becker E, den Dool H, Zhang Q (2014) Predictability and forecast skill in NMME. J Clim, 27(15): 5891–5906. Bjerknes J (1969) Atmospheric teleconnections from the equatorial Pacific. Mon Wea Rev, 97, 163–172. Cai W, Wu L, Lengaigne M, and Chang P (2019) Pantropical climate interactions. Science, 363, eaav4236. Ding H, Keenlyside NS, and Latif M (2012) Impact of the Equatorial Atlantic on the El Niño Southern Oscillation. Clim Dyn, 38, 1965–1972. Doi T, Storto A, Behera SK, Navarra A, and Yamagata T (2017) Improved Prediction of the Indian Ocean Dipole Mode by Use of Subsurface Ocean Observations. J Clim, 30, 7953–7970. Exarchou E, Ortega P, Rodríguez-Fonseca B, Losada T, Polo I, and Prodhomme C (2021) Impact of equatorial Atlantic variability on ENSO predictive skill. Nat Commun, 12, 1612. Fan L, and Meng X (2023) The Asymmetric Predictive Power of Indian Ocean Dipole for Subsequent Year's ENSO: Role of Atlantic Ocean as an Intermediary. Geophys Res Lett, 50, e2023GL105525. Fan HJ et al (2024) Coupling is key for the tropical Indian and Atlantic oceans to boost super El Niño. Sci Adv, 10, eadp2281. Fang X-H, and Mu M (2018) Both air-sea components are crucial for El Niño forecast from boreal spring. Sci Rep, 8, 10501. Frauen C, and Dommenget D (2012) Influences of the tropical Indian and Atlantic Oceans on the predictability of ENSO. Geophys Res Lett, 39. Ha K-J, Chu J-E, Lee J-Y, and Yun K-S (2017) Inter-basin coupling between the tropical Indian and Pacific Ocean on interannual timescale: observation and CMIP5 reproduction. Clim Dyn, 48, 459–475. Ham Y-G, Kug J-S, Park J-Y, and Jin F-F (2013) Sea surface temperature in the north tropical Atlantic as a trigger for El Niño/Southern Oscillation events. Nat Geosci, 6, 112–116. Ham Y-G, and Kug J-S (2015) Role of north tropical atlantic SST on the ENSO simulated using CMIP3 and CMIP5 models. Clim Dyn, 45, 3103–3117. Ham Y-G, Kim J-H, and Luo J-J (2019) Deep learning for multi-year ENSO forecasts. Nature, 573, 568–572. Hameed S (2018) The Indian Ocean Dipole. Oxon Res Encycl Climate Sci. Hu Z-Z, Kumar A, Huang B, Zhu J, L’Heureux M, McPhaden MJ, and Yu J-Y (2020) The Interdecadal Shift of ENSO Properties in 1999/2000: A Review. J Clim, 33, 4441-4462. Izumo T, Vialard J, Dayan H, Lengaigne M, and Suresh I (2016) A simple estimation of equatorial Pacific response from windstress to untangle Indian Ocean Dipole and Basin influences on El Niño. Clim Dyn, 46, 2247–2268. Jansen MF, Dommenget D, and Keenlyside N (2009) Tropical Atmosphere-Ocean Interactions in a Conceptual Framework. J Clim, 22, 550–567. Jiang F, Zhang W, Jin F-F, Stuecker MF, and Allan R (2021) El Niño Pacing Orchestrates Inter-Basin Pacific-Indian Ocean Interannual Connections. Geophys Res Lett, 48, e2021GL095242. Jiang L, and Li T (2021) Impacts of Tropical North Atlantic and Equatorial Atlantic SST Anomalies on ENSO. J Clim, 34, 5635–5655. Jiang L, Li T, and Ham Y-G (2022) Critical Role of Tropical North Atlantic SSTA in Boreal Summer in Affecting Subsequent ENSO Evolution. Geophys Res Lett, 49, e2021GL097606. Jin F-F (1997) An Equatorial Ocean Recharge Paradigm for ENSO. Part I: Conceptual Model. J Atmos Sci, 54, 811–829. Jin Y, Meng X, Zhang L, Zhao Y, Cai W, and Wu L (2023) The Indian Ocean Weakens the ENSO Spring Predictability Barrier: Role of the Indian Ocean Basin and Dipole Modes. J Clim, 36, 8331–8345. Kanamitsu M, Ebisuzaki W, Woollen J, Yang S-K, Hnilo JJ, Fiorino M, and Potter GL (2002) NCEP-DOE AMIP-II Reanalysis (R-2). Bull Amer Meteor Soc, 83, 1631–1644. Keenlyside NS, and Latif M (2007) Understanding Equatorial Atlantic Interannual Variability. J Clim, 20, 131-142. Keenlyside NS, Ding H, and Latif M (2013) Potential of equatorial Atlantic variability to enhance El Niño prediction. Geophys Res Lett, 40, 2278–2283. Kug J, Kirtman BP, and Kang I (2006) Interactive Feedback between ENSO and the Indian Ocean in an Interactive Ensemble Coupled Model. J Clim, 19, 6371–6381. Latif M, and Grötzner A (2000) The equatorial Atlantic oscillation and its response to ENSO. Clim Dyn, 16, 213–218. Li T, Tang Y, Lian T, and Hu A (2024) Quantifying the Relative Contributions of the Global Oceans to ENSO Predictability With Deep Learning. Geophys Res Lett, 51, e2023GL106584. Luo J-J, Masson S, Behera S, Shingu S, and Yamagata T (2005) Seasonal Climate Predictability in a Coupled OAGCM Using a Different Approach for Ensemble Forecasts. J Clim, 18, 4474–4497. Luo J-J, Masson S, Behera SK, and Yamagata T (2008) Extended ENSO Predictions Using a Fully Coupled Ocean-Atmosphere Model. J Clim, 21, 84–93. Luo J-J, Zhang R, Behera SK, and Yamagata T (2010) Interaction between El Niño and Extreme Indian Ocean Dipole. J Clim, 23, 726–742. Luo J-J, Yuan C, Sasaki W, and Masson S (2016) Current status of intraseasonal-seasonal-to-interannual prediction of the Indo-Pacific climate. Indo-Pacific Climate Variability and Predictability, pp.63–107. Luo J-J, Liu G, Hendon H, Alves O and Yamagata T (2017) Inter-basin sources for two-year predictability of the multi-year La Niña event in 2010-2012. Sci Rep, 7, 2276. Lyu P, Tang T, and Ling F (2024) Robust and Explainable ENSO Forecasts with Hybrid Convolution and Transformer Networks. Adv Atmos Sci, 41, 1289–1298. McGregor S, Timmermann A, Stuecker MF, England MH, Merrifield M, Jin FF, and Chikamoto Y (2014) Recent Walker circulation strengthening and Pacific cooling amplified by Atlantic warming. Nat Clim Change, 4(10), 888–892. Mcphaden MJ, Busalacchi AJ, Cheney R, and Takeuchi K (1998) The Tropical Ocean-Global Atmosphere observing system: A decade of progress. J Geophys Res: Oceans, 103, 14169–14240. Meinen CS, and Mcphaden MJ (2000) Observations of Warm Water Volume Changes in the Equatorial Pacific and Their Relationship to El Niño and La Niña. J Clim, 13, 3551–3559. Meng X-L, Rosenthal R, and Rubin DB (1992) Comparing correlated correlation coefficients. Psychol Bull, 111, 172–175. Merryfield WJ, and Coauthors (2020) Current and Emerging Developments in Subseasonal to Decadal Prediction. Bull Amer Meteor Soc, 101, E869–E896. Ohba M, and Ueda H (2009) Seasonally Different Response of the Indian Ocean to the Remote Forcing of El Niño: Linking the Dynamics and Thermodynamics. SOLA, 5, 176–179. Ohba M, and Watanabe M (2012) Role of the Indo-Pacific Inter-basin Coupling in Predicting Asymmetric ENSO Transition and Duration. J Clim, 25, 3321–3335. Palmer TN, and Coauthors (2004) Development Of A European Multimodel Ensemble System of Seasonal-to-Interannual Prediction (Demeter). Bull Amer Meteor Soc, 85, 853–872. Park JH, Li T (2019) Interdecadal modulation of El Niño-tropical North Atlantic teleconnection by the Atlantic multi-decadal oscillation. Clim Dyn, 52, 5345–5360. Philander S (1983) El Niño Southern Oscillation phenomena. Nature, 302, 295–301. Polo I, Martin-Rey M, Rodriguez-Fonseca B, Kucharski F, and Mechoso CR (2015) Processes in the Pacific La Niña onset triggered by the Atlantic Niño. Clim Dyn, 44, 115–131. Ren HL, Zuo J, and Deng Y (2019) Statistical predictability of Niño indices for two types of ENSO. Clim Dyn, 52, 5361–5382. Ren HL, Zheng F, Luo J-J, and Zhou G (2020) A Review of Research on Tropical Air-Sea Interaction, ENSO Dynamics and ENSO Prediction in China. J Meteorol Res, 34, 43–62. Reynolds RW, Rayner NA, Smith TM, Stokes DC, and Wang W (2002) An Improved In Situ and Satellite SST Analysis for Climate. J Clim, 15, 1609–1625. Saji N, Goswami B, Vinayachandran P, and Yamagata T (1999) A dipole mode in the tropical Indian Ocean. Nature, 401, 360–363. Suarez MJ, and Schopf PS (1988) A Delayed Action Oscillator for ENSO. J Atmos Sci, 45, 3283–3287. Tang Y, Zhang R-H, Liu T, and Mu M (2018) Progress in ENSO prediction and predictability study. Natl Sci Rev, 5, 826–839. Timmermann A, An SI, Kug JS, and Zhang X (2018) El Niño-Southern Oscillation complexity. Nature, 559, 535–545. Tseng Yh, Hu ZZ, Ding R, and Chen HC (2017) An ENSO prediction approach based on ocean conditions and ocean–atmosphere coupling. Clim Dyn, 48, 2025–2044. Wang CZ, Kucharski F, Barimalala R, and Bracco A (2009) Teleconnections of the tropical Atlantic to the tropical Indian and Pacific Oceans: A review of recent findings. Meteorologische Zeitschrift, 18, 445–454. Wang CZ (2019) Three-ocean interactions and climate variability: a review and perspective. Clim Dyn, 53, 5119–5136. Wang JZ, Wang C (2021) Joint boost to super El Niño from the Indian and Atlantic Oceans. J Clim, 34(12): 4937–4954. Wang R, He J, Luo J-J, and Chen L (2024) Atlantic warming enhances the influence of Atlantic Niño on ENSO. Geophys Res Lett, 51, e2023GL108013. Xue J, Luo J-J, Zhang W, and Yamagata T (2022) ENSO–IOD inter-basin connection is controlled by the Atlantic Multidecadal Oscillation. Geophys Res Lett, 49, e2022GL101571. Zhang L, and Coauthors (2023) Emergence of the Central Atlantic Niño. Sci Adv, 9,eadi5507. Zhang W, Jiang F, Stuecker MF, Jin F-F, and Timmermann A (2021) Spurious North Tropical Atlantic precursors to El Niño. Nat Commun, 12, 3096. Zhao M, Hendon HH, Alves O, Liu G, and Wang G (2016) Weakened Eastern Pacific El Niño Predictability in the Early Twenty-First Century. J Clim, 29, 6805-6822. Zhao S, Jin F-F, Stuecker MF, and Cai W (2024) Explainable El Niño predictability from climate mode interactions. Nature, 630, 891–898. Zhao Y, Jin Y, Capotondi A, Li J, and Sun D (2023) The Role of Tropical Atlantic in ENSO Predictability Barrier. Geophys Res Lett, 50, e2022GL101853. Supplementary Files JiangKeCDSupportinginformation.docx Cite Share Download PDF Status: Published Journal Publication published 07 Jun, 2025 Read the published version in Climate Dynamics → Version 1 posted Editorial decision: Major Revision 27 Jan, 2025 Reviewers agreed at journal 12 Nov, 2024 Reviewers invited by journal 12 Nov, 2024 Editor assigned by journal 06 Nov, 2024 First submitted to journal 29 Oct, 2024 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-5352476","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":377374710,"identity":"59bbdf06-5ba4-4ca1-a6c1-c40ba2e3a712","order_by":0,"name":"Ke Jiang","email":"","orcid":"","institution":"Nanjing University of Information Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"Ke","middleName":"","lastName":"Jiang","suffix":""},{"id":377374711,"identity":"9d9601de-ab91-40aa-9acd-a06d334d087e","order_by":1,"name":"Jiye Wu","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAvklEQVRIiWNgGAWjYHCChAMMDDZy/MzMBx+QoiXNWLKdLdmAFJsOJxqc5zETIEqtwY2Eh4d5Kg4nGB9mMGNgqLGJJkZLwmGeM+l5ZocZ0h4wHEvLbSCkxQykhbfNuhio5bgBY8NhYrX8Y07c3MzYJkGClgbnxA3MzGzEabE/8yDh4JxjacYSh9mYDRKI8Ytke07yhzc1wKjsP//xwYcaG8JaGBh4Eph4YOwEwspBgP0A4w/iVI6CUTAKRsFIBQDnskP8dw7BmQAAAABJRU5ErkJggg==","orcid":"https://orcid.org/0009-0008-6851-9382","institution":"Nanjing University of Information Science and Technology School of Atmospheric Sciences","correspondingAuthor":true,"prefix":"","firstName":"Jiye","middleName":"","lastName":"Wu","suffix":""},{"id":377374712,"identity":"a7e4a444-8d96-48c1-ba5c-c2bf25713f98","order_by":2,"name":"Jing-Jia Luo","email":"","orcid":"","institution":"Nanjing University of Information Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"Jing-Jia","middleName":"","lastName":"Luo","suffix":""}],"badges":[],"createdAt":"2024-10-29 08:29:27","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-5352476/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-5352476/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1007/s00382-025-07721-9","type":"published","date":"2025-06-07T15:57:42+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":70215646,"identity":"53dbdce2-4673-4df0-9c8a-1b899e6a1143","added_by":"auto","created_at":"2024-11-29 15:28:08","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":217530,"visible":true,"origin":"","legend":"\u003cp\u003e(a) ACC skill in predicting SSTAs over Niño 3.4 region (5°S-5°N, 120°-170°W) at lead time of 1-12 months. The green (yellow, magenta) solid lines denote the Obs experiments, the dashed lines represent the Cli experiments. (b) As in (a), but for the RMSE.\u003c/p\u003e","description":"","filename":"Fig1.png","url":"https://assets-eu.researchsquare.com/files/rs-5352476/v1/def30b5a17f4738a808bf5b3.png"},{"id":70215643,"identity":"e7839afc-b9cc-47d1-b6e4-e1f23d78c3e0","added_by":"auto","created_at":"2024-11-29 15:28:08","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":1242198,"visible":true,"origin":"","legend":"\u003cp\u003eACC skill differences between the Obs and Cli experiment of the Atlantic (AO; left column), Indian Ocean (IO; middle column), and Indian and Atlantic Oceans (AO+IO; right column) groups in predicting SSTAs at lead time of 6 (1\u003csup\u003est\u003c/sup\u003e row), 9 (2\u003csup\u003end\u003c/sup\u003e row) and 12 (3\u003csup\u003erd\u003c/sup\u003e row) months. Stippled areas in (a-i) represent that the differences in ACCs are significant at the 95% confidence level.\u003c/p\u003e","description":"","filename":"Fig2.png","url":"https://assets-eu.researchsquare.com/files/rs-5352476/v1/a8444b0d12162aa9dd02d153.png"},{"id":70216788,"identity":"f855bf79-0c7c-49d9-80c3-e2bed0440a01","added_by":"auto","created_at":"2024-11-29 15:52:08","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":664910,"visible":true,"origin":"","legend":"\u003cp\u003eImpact of tropical Indian and Atlantic Oceans SSTAs on the prediction skills of Niño 3.4 index (5°S-5°N, 120°-170°W) for different start months (left column) and target seasons (right column). (a-c) ACC skill based on nine-member ensemble mean predictions of the Cli (dashed lines) and Obs (solid lines) hindcast experiments starting on the 1st day of February (green), May (orange), August (blue), and November (grey) at lead time of 1-12 months. (d-f) ACC skill of the Cli (box bar) and Obs (shaded bar) hindcast experiments targeted on MAM, JJA, SON, and DJF at 1 (green bar), 4 (red bar), 7 (yellow bar), and 10 (blue bar) months lead. The slash filling denotes the ACC differences between the Obs and Cli hindcast experiments are statistically significant at the 95% confidence level.\u003c/p\u003e","description":"","filename":"Fig3.png","url":"https://assets-eu.researchsquare.com/files/rs-5352476/v1/2f2114e003a680c2bdcccc2f.png"},{"id":70215841,"identity":"39fdb80d-4947-42bb-b053-6b29fdea19fb","added_by":"auto","created_at":"2024-11-29 15:36:08","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":2651607,"visible":true,"origin":"","legend":"\u003cp\u003eImpact of tropical Indian and Atlantic Oceans SSTAs (AO+IO group) on the prediction skills for El Niño and La Niña events when initialized on 1st February. (a) ACC skill in predicting Niño 3.4 indices at lead time of 1-12 months. (b-i) The ACC skill differences between the Obs and Cli experiments in predicting SST anomalies in June, September, December, and next January for La Niña (1\u003csup\u003est\u003c/sup\u003e row) and El Niño events (2\u003csup\u003end\u003c/sup\u003e row). Stippled areas in (b-i) represent that the differences in ACCs are significant at the 95% confidence level.\u003c/p\u003e","description":"","filename":"Fig4.png","url":"https://assets-eu.researchsquare.com/files/rs-5352476/v1/c1342d152652278e786da7fa.png"},{"id":70215641,"identity":"92d29326-a7b0-4f18-9d20-1cb043b6c073","added_by":"auto","created_at":"2024-11-29 15:28:08","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":1306996,"visible":true,"origin":"","legend":"\u003cp\u003eThe composited differences in SST (shading, °C) and 10m winds (vectors, m·s\u003csup\u003e-1\u003c/sup\u003e) anomalies between the prediction of Obs_AO+IO and Cli AO+IO experiment initialized on 1st Feb during the developing years of La Niña (left column) and El Niño (right column). Stippled areas and bold vectors indicate that the differences of SST and winds exceeding the 95% confidence level, respectively. Hollow circles denote the difference of precipitation (mm·day\u003csup\u003e-1\u003c/sup\u003e) exceeding the 95% confidence level.\u003c/p\u003e","description":"","filename":"Fig5.png","url":"https://assets-eu.researchsquare.com/files/rs-5352476/v1/f48eefbd7bce3cefcc645b43.png"},{"id":70215648,"identity":"e58458b1-bc35-4e3d-b27c-d68115151513","added_by":"auto","created_at":"2024-11-29 15:28:08","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":565763,"visible":true,"origin":"","legend":"\u003cp\u003eComparison of the correlation between the precursors in the tropical Indian and Atlantic Oceans and ENSO in the observations and model hindcast experiments. (a) The observed DJF Niño3.4 (red dashed line) and reconstructed DJF Niño3.4 (black solid line) based on the indices of the NTA, Atl3, IOBM, and DMI (see Methods). (b-i) Scatterplots between MAM NTA, JJA Atl3, MAM IOBM, SON DMI, and DJF Niño 3.4 indices based on (b-e) reconstructed observations and (f-i) the differences between Obs and Cli experiments of AO+IO group. Blue, red and black circles represent La Niña, El Niño and moderate events, respectively. The gray lines indicate the regression line. The numbers in the upper right corner denotes the correlation coefficients (CCs).\u003c/p\u003e","description":"","filename":"Fig6.png","url":"https://assets-eu.researchsquare.com/files/rs-5352476/v1/2582a5bb52bddaf15fa5494c.png"},{"id":70215838,"identity":"e3c817b6-9719-4130-b633-a640464c5586","added_by":"auto","created_at":"2024-11-29 15:36:08","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":2117896,"visible":true,"origin":"","legend":"\u003cp\u003e(a-d) Lagged partial-regressed anomalies of (a) MAM(0), (b) JJA(0), (c) SON(0), (d) D(0)J(1)F(1) of SST (shading, °C), 10m winds (vectors, m·s\u003csup\u003e-1\u003c/sup\u003e) onto MAM(0) NTA index during 1983-2018. The anomalies represent the response to one standard deviation (STD) of MAM(0) NTA index. (e-h) Lagged partial-regressed anomalies onto MAM(0) NTA index based on the differences between the Obs and Cli experiments of AO group. (i-l) As in a-d, but for the partial-regressed anomalies of (i) JJA(0), (j) SON(0), (k) D(0)J(1)F(1), (l) MAM(1) of SST (shading, °C), 10m winds (vectors, m·s\u003csup\u003e-1\u003c/sup\u003e), and precipitation (dots, mm·day\u003csup\u003e-1\u003c/sup\u003e) onto JJA(0) Atl3 index during 1983-2018. (m-p) As in e-h, but for the lagged partial-regressed anomalies onto JJA(0) Atl3 index. Shaded areas and bold vectors indicate that the regressed SST and winds are significant at the 95% confidence level, respectively. Hollow circles denote the regressed precipitation anomalies exceeding the 95% confidence level.\u003c/p\u003e","description":"","filename":"Fig7.png","url":"https://assets-eu.researchsquare.com/files/rs-5352476/v1/c7bbfed75609709c9a67be29.png"},{"id":70215650,"identity":"07d22533-6833-4b7f-bf2c-d815fbadeff6","added_by":"auto","created_at":"2024-11-29 15:28:08","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":2047642,"visible":true,"origin":"","legend":"\u003cp\u003e(a-d) Lagged partial-regressed anomalies of (a) MAM(0), (b) JJA(0), (c) SON(0), (d) D(0)J(1)F(1) of SST (shading, °C), 10m winds (vectors, m·s-1) onto MAM(0) IOBM index during 1983-2018. The anomalies represent the response to one STD of MAM(0) IOBM index. (e-h) Lagged partial-regressed anomalies onto MAM(0) IOBM index based on the differences between the Obs and Cli experiments of IO group. (i-l) As in a-d, but for the partial-regressed anomalies of (i) SON(0), (j) D(0)J(1)F(1), (k) MAM(1), (l) JJA(1) of SST (shading, °C), 10m winds (vectors, m·s\u003csup\u003e-1\u003c/sup\u003e), and precipitation (dots, mm·day\u003csup\u003e-1\u003c/sup\u003e) onto SON(0) DMI index during 1983-2018. (m-p) As in e-h, but for the lagged partial-regressed anomalies onto SON(0) DMI index. Shaded areas and bold vectors indicate that the regressed SST and winds are significant at the 95% confidence level, respectively. Hollow circles denote the regressed precipitation (mm·day\u003csup\u003e-1\u003c/sup\u003e) exceeding the 95% confidence level.\u003c/p\u003e","description":"","filename":"Fig8.png","url":"https://assets-eu.researchsquare.com/files/rs-5352476/v1/acdd6535f5be2437c1f5af23.png"},{"id":70216524,"identity":"7d50bb16-ebb4-43c9-82c7-64a256b86dc1","added_by":"auto","created_at":"2024-11-29 15:44:08","extension":"png","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":237310,"visible":true,"origin":"","legend":"\u003cp\u003eComparison of the correlations of NTA (1\u003csup\u003est\u003c/sup\u003e column), Atl3 (2\u003csup\u003end\u003c/sup\u003e column), DMI (3\u003csup\u003erd\u003c/sup\u003e column), and IOBM (4\u003csup\u003eth\u003c/sup\u003e column) indices with Niño 3.4 index based on the hindcasts of NUIST-CFS1.0 and six NMME models, as well as the observations. The red line represents the CCs calculated using the reconstructed Niño 3.4 index (same as the results shown in Fig. 6b-e), and the blue line represents the CCs based on the differences between the Obs and Cli experiments (same as the results shown in Fig. 6f-i).\u003c/p\u003e","description":"","filename":"Fig9.png","url":"https://assets-eu.researchsquare.com/files/rs-5352476/v1/5d943a95c8e43a3aaab20f31.png"},{"id":70215837,"identity":"2e5d17f0-a9c1-4dee-9952-eebcafb61c9d","added_by":"auto","created_at":"2024-11-29 15:36:08","extension":"png","order_by":10,"title":"Figure 10","display":"","copyAsset":false,"role":"figure","size":2349868,"visible":true,"origin":"","legend":"\u003cp\u003eDifference in SSTA between the Obs_AO+IO and Cli_AO+IO experiments in predicting (a) individual El Niño (left column) and (b) La Niña (right column) events initialized on 1st Feb. (c-j) The composited differences in SST (shading, °C) and 10m winds (vectors, m·s\u003csup\u003e-1\u003c/sup\u003e) anomalies between the prediction of Obs_AO+IO and Cli AO+IO experiment initialized on 1st Feb of the developing years of El Niño events during 1983-2000 (left column) and 2001-2018 (right column). Stippled areas and bold vectors indicate that the differences of SST and winds exceeding the 95% confidence level, respectively. Hollow circles denote the difference of precipitation (mm·day\u003csup\u003e-1\u003c/sup\u003e) exceeding the 95% confidence level.\u003c/p\u003e","description":"","filename":"Fig10.png","url":"https://assets-eu.researchsquare.com/files/rs-5352476/v1/e700fc53909786707e692cb9.png"},{"id":84242734,"identity":"c6461e9e-899d-4e86-8021-7142c3321202","added_by":"auto","created_at":"2025-06-09 16:11:56","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":11770581,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5352476/v1/ab6eeefb-6c17-4f75-971c-688b2e63560c.pdf"},{"id":70215652,"identity":"fb2bb7c8-1a91-4e70-b793-6cb2aa3727be","added_by":"auto","created_at":"2024-11-29 15:28:10","extension":"docx","order_by":14,"title":"","display":"","copyAsset":false,"role":"supplement","size":45026356,"visible":true,"origin":"","legend":"","description":"","filename":"JiangKeCDSupportinginformation.docx","url":"https://assets-eu.researchsquare.com/files/rs-5352476/v1/b8569bf944a7a751d305a0ee.docx"}],"financialInterests":"","formattedTitle":"Complex Influences of Tropical Indian and Atlantic Oceans on ENSO Prediction","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eThe El Ni\u0026ntilde;o-Southern Oscillation (ENSO) is the most dominant interannual variability in the tropics, and its extensive environmental and climatic impacts resonate worldwide (e.g., Bjerknes, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e1969\u003c/span\u003e; Philander, \u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e1983\u003c/span\u003e; McPhaden et al., \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e1998\u003c/span\u003e). Continuous efforts have been made to improve the understanding and prediction of ENSO. Since the early 1990s, with significant progress in ENSO dynamics, climate modeling, data assimilation, as well as the adoption of deep learning, ENSO prediction has been notably improved (e.g., Palmer et al., 2004; Luo et al., \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e2008\u003c/span\u003e; Ham et al., \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Ren et al., \u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Lyu et al. \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). However, ENSO prediction encountered bottlenecks in past decades (e.g., Barnston et al., \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2012\u003c/span\u003e), with previous studies indicating an obvious decline in predictive skill in the early Twenty-First Century (e.g., Hu et al. \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Zhao et al. \u003cspan citationid=\"CR67\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). Conventionally, the long-term predictability of ENSO is rooted in the Pacific atmosphere-ocean coupled processes and oceanic waves (e.g., Bjerknes, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e1969\u003c/span\u003e; Suarez and Schopf, \u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e1988\u003c/span\u003e; Jin, \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e1997\u003c/span\u003e), thereby wind patterns, oceanic heat content and/or thermocline variations over the equatorial Pacific serve as key precursors of ENSO prediction (e.g., Meinen and McPhaden, \u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e2000\u003c/span\u003e; Tseng et al., 2017; Fang and Mu, \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2018\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eHowever, the tropical basins are tightly interconnected (e.g., Cai et al., \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Wang, \u003cspan citationid=\"CR61\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). Growing attentions have been paid to the precursors in the tropical Indian and Atlantic Oceans for the ENSO evolution (e.g., Luo et al., \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2010\u003c/span\u003e and \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Ding et al., \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e2012\u003c/span\u003e; Exarchou et al., \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Jiang and Li, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Jiang et al., \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Jin et al., \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). The warming in the equatorial Atlantic, often called Atlantic Ni\u0026ntilde;o, can modify the Walker Circulation (e.g., Polo et al., \u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e2015\u003c/span\u003e; Wang et al., \u003cspan citationid=\"CR60\" class=\"CitationRef\"\u003e2009\u003c/span\u003e) and induce anomalous easterly winds, triggering an eastward-propagating Kelvin wave to favor the development of La Ni\u0026ntilde;a, and \u003cem\u003evice versa\u003c/em\u003e for El Ni\u0026ntilde;o development (e.g., Latif and Gr\u0026ouml;tzner, \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2000\u003c/span\u003e; Keenlyside and Latif, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2007\u003c/span\u003e; Jiang and Li, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Wang et al., \u003cspan citationid=\"CR63\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Moreover, the low-level cyclone to the north-west of the North Tropical Atlantic (NTA) sea surface temperature (SST) warming, produces a cold SST anomaly (SSTA) through enhanced evaporation and cold advection, which further induces low-level anticyclone in the tropical northwestern Pacific. The corresponding equatorial easterly also favors the development of La Ni\u0026ntilde;a (e.g., Ham et al., \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2013\u003c/span\u003e; Ham and Kug, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2015\u003c/span\u003e; Polo et al., \u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e2015\u003c/span\u003e; Zhang et al., \u003cspan citationid=\"CR66\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Jiang and Li, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). The Indian Ocean SST warming also serves as an important driver of easterly wind anomalies in the western Pacific. During the El Ni\u0026ntilde;o peak phase, the Indian Ocean Basin Mode (IOBM) predominantly accelerates the rapid transition from El Ni\u0026ntilde;o to La Ni\u0026ntilde;a by inducing easterly and generating upwelling Kelvin waves in the equatorial western Pacific. During the El Ni\u0026ntilde;o development phase, the Indian Ocean Dipole (IOD) is crucial for the growth of El Ni\u0026ntilde;o (e.g., Kug et al., \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2006\u003c/span\u003e; Ohba and Ueda, \u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e2009\u003c/span\u003e; Luo et al., \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2010\u003c/span\u003e; Ohba and Watanabe, \u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e2012\u003c/span\u003e; Izumo et al., \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2016\u003c/span\u003e; Ha et al., \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Fan and Meng, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2023\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eUnraveling the inter-basin interactive processes underpinning the ENSO development provides new insight and opportunity for extending ENSO predictability (e.g., Luo et al., \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2010\u003c/span\u003e and \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Frauen and Dommenget, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2012\u003c/span\u003e; Doi et al., \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Alexander et al., \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Recently, utilizing conceptual and statistical models, the importance of tropical Indian and Atlantic Oceans in predicting ENSO has been emphasized (e.g., Zhao et al., \u003cspan citationid=\"CR68\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Li et al., \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2024\u003c/span\u003e), especially in overcoming the long-standing spring prediction barrier (e.g., Zhao et al., \u003cspan citationid=\"CR69\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Jin et al., \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Additionally, based on the dynamical models, the SST variability in the individual tropical basins (the Indian Ocean or Atlantic) is demonstrated to be crucial (e.g., Frauen and Dommenget, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2012\u003c/span\u003e; Keenlyside et al., \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2013\u003c/span\u003e; Alexander et al., \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2022\u003c/span\u003e), and studies also demonstrated the collective roles of the two tropical basins (e.g., Luo et al., \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Fan et al., \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2024\u003c/span\u003e), but with a focus on the analyses of a few specific cases and extreme events. Previous studies have confirmed the impact of inter-basin interactions on ENSO through observational studies, statistical models, and dynamical model experiments. However, the complexity of this issue is quite remarkable. For instance, the active roles of inter-basin interactions on interannual timescale in ENSO development and predictions are rebutted in a few recent studies (e.g., Zhang et al., \u003cspan citationid=\"CR66\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Jiang et al., \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). In addition, some studies suggest that both the tropical Pacific-Indian Ocean and Pacific-Atlantic teleconnections exhibit interdecadal change over the last decades (e.g., Park et al., 2019; Xue et al., \u003cspan citationid=\"CR64\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Wang et al. \u003cspan citationid=\"CR63\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Furthermore, the models' capacity to accurately simulate the observed ocean inter-basin interactions remains contentious, which affects their contributions to ENSO predictions. These raise a sequence of new questions, requiring further comprehensions about the complicated influence of the tropical SSTs outside the Pacific on ENSO predictions. This study aims to identify the roles of the tropical Indian and Atlantic Oceans on ENSO prediction during the last four decades and to explore underlying possible mechanisms.\u003c/p\u003e"},{"header":"2. Data and Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e2.1 Observational Data\u003c/h2\u003e \u003cp\u003eThis study utilizes the following observational and reanalysis datasets: (a) monthly Optimum Interpolation Sea Surface Temperature (OISST) from National Oceanic and Atmospheric Administration (NOAA; Reynolds et al., \u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e2002\u003c/span\u003e) is utilized for model coupled SST-nudging and hindcast validation; (b) monthly mean 10-m horizontal winds are derived from the National Centers for Environmental Prediction (NCEP)-Department of Energy Atmospheric Model Intercomparison Project II reanalysis (Kanamitsu et al., \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2002\u003c/span\u003e); (c) monthly precipitation data from the Global Precipitation Climatology Project (Adler et al., 2003). The analysis period of this study is 1983\u0026ndash;2018. ENSO is represented by the Ni\u0026ntilde;o3.4 index (5\u0026deg;S-5\u0026deg;N, 170\u0026deg;-120\u0026deg;W). Additionally, the SSTA averaged over the NTA (5\u0026deg;-25\u0026deg;N, 70\u0026deg;-15\u0026deg;W), equatorial eastern Atlantic (4\u0026deg;S-4\u0026deg;N, 20\u0026deg;W-0\u0026deg;) and Indian Ocean Basin (20\u0026deg;S-20\u0026deg;N, 40\u0026ndash;110\u0026deg;E) regions, referred as to NTA, Atl3 and IOBM indices, are used to represent the NTA warming/cooling, Atlantic Ni\u0026ntilde;o/Ni\u0026ntilde;a, and the Indian Ocean wide basin warming/cooling, respectively. The IOD is represented by the DMI index, defined as the differences in SSTAs between the western (10\u0026deg;S-10\u0026deg;N, 50\u0026deg;E-70\u0026deg;E) and eastern (10\u0026deg;S-0\u0026deg;, 90\u0026deg;-110\u0026deg;E) tropical Indian Ocean (Saji et al., \u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e1999\u003c/span\u003e). The Warm Water Volume (WWV) refers to the volume of water in the equatorial Pacific (5\u0026deg;S-5\u0026deg;N, 120\u0026deg;E-80\u0026deg;W) that is warmer than 20\u0026deg;C, which is crucial for assessing the heat content of the upper ocean in the Pacific (Meinen and McPhaden, \u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e2000\u003c/span\u003e). The prediction skills are evaluated by calculating the anomaly correlation coefficient (ACC) and root-mean-square error (RMSE) between the observations and model hindcasts:\u003cdiv id=\"Equa\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equa\" name=\"EquationSource\"\u003e\n$$\\:ACC\\left(\\tau\\:\\right)=\\frac{\\sum\\:_{t=1}^{n}\\left[a\\left(t\\right)b\\left({t}_{0},\\tau\\:\\right)\\right]}{\\sqrt{\\sum\\:_{t=1}^{n}{a}^{2}\\left(t\\right)}\\sqrt{\\sum\\:_{t=1}^{n}{b}^{2}\\left({t}_{0},\\tau\\:\\right)}}$$\u003c/div\u003e\u003c/div\u003e\u003cdiv id=\"Equb\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equb\" name=\"EquationSource\"\u003e\n$$\\:RMSE\\left(\\tau\\:\\right)=\\sqrt{\\frac{1}{n}\\sum\\:_{i=1}^{n}{\\left[a\\left(t\\right)-b\\left({t}_{0},\\tau\\:\\right)\\right]}^{2}}$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003ewhere \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\text{a}\\left(\\text{t}\\right)\\)\u003c/span\u003e\u003c/span\u003e represents the observation at month \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\text{t}\\)\u003c/span\u003e\u003c/span\u003e, and \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\text{b}\\left({\\text{t}}_{0},{\\tau\\:}\\right)\\)\u003c/span\u003e\u003c/span\u003e represent the corresponding model hindcast at a lead month of \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\tau\\:}\\)\u003c/span\u003e\u003c/span\u003e initializing from time \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\text{t}}_{0}\\)\u003c/span\u003e\u003c/span\u003e. For statistical analysis, the significance of regression and correlation coefficients is determined through a two-tailed Student's t-test, and the comparison between two correlation coefficients employs Steiger\u0026rsquo;s z-test (Meng et al., \u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e1992\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e2.2 The Coupled Model and Sensitivity Hindcast Experiments\u003c/h2\u003e \u003cp\u003eTo quantitatively explore the possible impacts of tropical inter-basin interactions on ENSO prediction skills, six sets of model sensitivity hindcast experiments using the SINTEX-F (Luo et al., \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e2005\u003c/span\u003e) are performed, which are organized into three groups (referred to AO, IO, and AO\u0026thinsp;+\u0026thinsp;IO group, respectively). Each group consists of one sensitivity (Obs) and one control (Cli) experiment. In the AO group, observed climatological and monthly SSTs are specified over the tropical Atlantic (20\u0026deg;S-20\u0026deg;N, 70\u0026deg;W-15\u0026deg;E) in the control and sensitivity experiments, respectively. The IO group follows the same design as the AO group, but with the observed climatological and monthly SSTs being specified over the tropical Indian Ocean (20\u0026deg;S-20\u0026deg;N, 40\u0026deg;-120\u0026deg;E). In the AO\u0026thinsp;+\u0026thinsp;IO group, the observed climatological and monthly SSTs are specified in both the tropical Indian and Atlantic Oceans regions. Each experiment comprises of nine ensemble members produced by three SST-nudging initializations and three coupling schemes following Luo et al. (\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e2008\u003c/span\u003e). The hindcasts are initialized from 1st day of February, May, August, and November of 1983\u0026ndash;2018 and then integrated for one year; details are summarized in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. By comparing the results of control and sensitivity experiments, the influence of SSTAs over the specific tropical basins on the ENSO prediction can be measured.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eDescription of the three groups (six sets) of model hindcast experiments.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eName\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eDescription\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eEnsemble hindcast\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCli_AO\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMonthly climatological mean SSTs of 1983\u0026ndash;2012 are specified in the tropical Atlantic (20\u0026deg;S-20\u0026deg;N, 70\u0026deg;W-15\u0026deg;E), and atmosphere-ocean is freely coupled in other areas.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\" morerows=\"5\" rowspan=\"6\"\u003e \u003cp\u003eNumber of ensemble member: 9\u003c/p\u003e \u003cp\u003eInitialized from 1st day of February, May, August and November of 1983\u0026ndash;2018\u003c/p\u003e \u003cp\u003eForecast length: 12 months\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eObs_AO\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eObserved monthly SSTs are specified in the tropical Atlantic (20\u0026deg;S-20\u0026deg;N, 70\u0026deg;W-15\u0026deg;E), and atmosphere-ocean is freely coupled in other areas.\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCli_IO\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMonthly climatological mean SSTs of 1983\u0026ndash;2012 are specified in the tropical Indian Ocean (20\u0026deg;S-20\u0026deg;N, 40\u0026deg;-120\u0026deg;E), and atmosphere-ocean is freely coupled in other areas.\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eObs_AO\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eObserved monthly SSTs are specified in the tropical Indian Ocean (20\u0026deg;S-20\u0026deg;N, 40\u0026deg;-120\u0026deg;E), and atmosphere-ocean is freely coupled in other areas.\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCli_AO\u0026thinsp;+\u0026thinsp;IO\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSame as the Cli experiments above, but with monthly climatological mean SSTs being specified in both the tropical Indian and Atlantic Oceans.\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eObs_AO\u0026thinsp;+\u0026thinsp;IO\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSame as the Obs expereimnts above, but with observed monthly SSTs being specified in both the tropical Indian and Atlantic Oceans.\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"3\"\u003e\u003cb\u003eNote\u003c/b\u003e: The differences between each set of experiments (Obs and Cli) can be analyzed to assess the individual and collective impacts of SSTs in the tropical Atlantic and Indian Oceans on the climate prediction in the tropical Pacific.\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003e2.3 Multiple Regression Method\u003c/h2\u003e \u003cp\u003eAllowing for a compatible comparison with the results of our experiments, the Pacific intrinsic dynamics and the effects of the various precursors in the Indian and Atlantic Oceans on ENSO needed to be elucidated in the observation. Herein we build a multi-element linear regression model depicting the ENSO\u0026rsquo;s responses to four dominant climate modes (i.e., NTA, Atlantic Ni\u0026ntilde;o/Ni\u0026ntilde;a, IOBM, and IOD) following previous studies (Luo et al., \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Jiang et al., \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). This equation was given as follows:\u003cdiv id=\"Equc\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equc\" name=\"EquationSource\"\u003e\n$$\\:DJF\\:Ni\u0026ntilde;o\\:3.4\\left(\\tau\\:\\right)=a\\times\\:MAM\\:NTA\\left(\\tau\\:\\right)+b\\times\\:JJA\\:Atl3\\left(\\tau\\:\\right)+c\\times\\:MAM\\:IOBM\\left(\\tau\\:\\right)+d\\times\\:SON\\:DMI\\left(\\tau\\:\\right)+e\\times\\:ND\\left(-1\\right)J\\:WWV\\:anomalies\\left(\\tau\\:\\right)+f$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eWhere DJF, MAM, JJA, SON, and ND(-1)J represent the months during and before ENSO years (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\tau\\:})\\)\u003c/span\u003e\u003c/span\u003e. The regression coefficients derived from the observational data spanning 1983 to 2018 are \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\text{a}\\:\\)\u003c/span\u003e\u003c/span\u003e= -1.76, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\text{b}\\)\u003c/span\u003e\u003c/span\u003e = -0.16, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\text{c}\\)\u003c/span\u003e\u003c/span\u003e = 1.46, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\text{d}\\)\u003c/span\u003e\u003c/span\u003e = 0.85, and \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\text{e}\\)\u003c/span\u003e\u003c/span\u003e = -0.08, with the constant \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\text{f}\\)\u003c/span\u003e\u003c/span\u003e = -0.04. To mitigate the influence of ENSO autocorrelation, we reconstruct the equation to reflect the impacts of major tropical SST variabilities outside the Pacific:\u003cdiv id=\"Equd\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equd\" name=\"EquationSource\"\u003e\n$$\\:reconstructed\\:DJF\\:Ni\u0026ntilde;o\\:3.4\\left(\\tau\\:\\right)=a\\times\\:MAM\\:NTA\\left(\\tau\\:\\right)+b\\times\\:JJA\\:Atl3\\left(\\tau\\:\\right)+c\\times\\:MAM\\:IOBM\\left(\\tau\\:\\right)+d\\times\\:SON\\:DMI\\left(\\tau\\:\\right)$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003c/div\u003e"},{"header":"3. Overall impacts of tropical Indian and Atlantic Oceans SSTs on ENSO prediction","content":"\u003cp\u003eTo generally understand the impacts of tropical Indian and Atlantic Oceans SSTs on ENSO prediction, we evaluate both the ACC and RMSE skill of the Ni\u0026ntilde;o 3.4 index prediction in all sets of experiments. Figure\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ea presents the all-season ACC skills for the Ni\u0026ntilde;o3.4 index from 1983 to 2018 across various experiments, with the lead time ranging from 1 to 12 months. At the early prediction stage, there are no notable differences between the Obs and Cli experiments in each group, since the initial conditions used for all the hindcast experiments are the same. However, as the lead time increases, the ACC differences among experiments become pronounced, with the Obs experiments consistently demonstrating higher ACC. Simultaneously, the lower RMSEs (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eb) further confirm the better skill of the Obs experiments in long-term ENSO forecasts compared to the Cli experiment. Our results not only align with previous findings that highlight the importance of tropical Indian and Atlantic Oceans SSTs in the evolution and prediction of ENSO (e.g., Jansen et al., \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2009\u003c/span\u003e; Luo et al., \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2010\u003c/span\u003e and \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Frauen and Dommenget, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2012\u003c/span\u003e; Exarchou et al., \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2021\u003c/span\u003e), but also reveal the complex impact of the individual basin. Notably, in the IO group, the ACC improvement of the Obs experiment compared to the Cli experiment is smaller than that of the other two groups at almost all lead times. In particular, at lead times beyond 11 months, the ACC of the Obs_IO experiment even falls below that of the Cli_IO experiment. In contrast, the ACC differences between the Obs and Cli experiment in the AO and AO\u0026thinsp;+\u0026thinsp;IO groups are uniformly positive and getting large with increasing lead time. In general, the Obs_AO exhibits the highest ACC skill and the Obs_AO\u0026thinsp;+\u0026thinsp;IO exhibits the lowest RMSE. The results may suggest that SSTAs in the tropical Atlantic display a larger influence on ENSO prediction than the Indian Ocean does, while the joint influence of SSTAs from the two basins contributes to more stable and reliable predictions at longer lead times.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe detailed spatial distribution of ACC differences (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e) shows that the prediction of tropical Pacific SSTs is generally improved at long lead times (beyond 6 months) by prescribing observed SSTAs in the tropical Indian and Atlantic Oceans. The significant skill improvement emerges in the central equatorial Pacific at 3-month lead, and the improvement becomes more apparent and extends to the far eastern Pacific as the lead time increases in the AO\u0026thinsp;+\u0026thinsp;IO group (Fig. \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003em-r, Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eg-i). At the lead time of 12 months, the difference reaches 0.3 over the equatorial central-eastern Pacific (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ei). Similar skill difference is displayed in the experiments of the AO group (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ea-c, Fig. \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003ea-f), of which is slightly smaller than AO\u0026thinsp;+\u0026thinsp;IO group at the same lead time, but exhibits a broader area of skill improvement, encompassing the entire tropical Pacific except for the Northwest Pacific. In comparison, the skill difference in the IO group exhibits relatively weak magnitude and complex spatial patterns (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ed-f). Besides a slight skill improvement in the central-western equatorial Pacific, a surprising skill decrease originates over the eastern Pacific and extends westward with the increase in lead time (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ed-f, Fig. S2g-l). Based on the detailed ACC differences, prescribing observed SSTAs in the tropical Indian and Atlantic Oceans significantly enhances the prediction skill of tropical Pacific SSTs, especially over the central equatorial Pacific.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eConsidering the possible seasonal dependence of the ENSO\u0026rsquo;s response to the tropical Indian and Atlantic Oceans SSTAs, we further analyzed the prediction skills of the Ni\u0026ntilde;o 3.4 indices across different start months and target seasons. Similar to the all-season prediction skill assessment (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e), the contribution from the Atlantic SSTA surpasses that of the Indian Ocean and remains consistently influential throughout the ENSO evolution (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). Interestingly, the ACC differences are manifested 3 months after initializing from February, 9 months after initializing from August, and 6 months after initializing from November across all three groups of the hindcast experiments (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ea-c). The target seasons of these predictions correspond to the transition phase of ENSO, when the atmosphere-ocean coupling in the Pacific is relatively weak, suggesting the critical role of the tropical Indian and Atlantic Oceans in contributing to the predictability of ENSO during its transition and developing phases. In addition, the influence of the Indian and Atlantic Oceans SSTAs on the prediction skill of the ENSO varies across different target seasons (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ed-f). During the boreal spring (MAM) and summer (JJA) when the ENSO typically onsets or decays, including the forcing from SSTs outside the Pacific significantly enhances the prediction. Notably, the prediction skill targeted in the summer season is severely poor in the Cli experiments, which is markedly improved in the Obs experiments, especially for the AO and AO\u0026thinsp;+\u0026thinsp;IO groups. This suggests a deteriorated spring prediction barrier (SPB) when excluding the influence of the tropical Indian and Atlantic Oceans SSTs in the ENSO prediction. By contrast, when ENSO typically reaches maturity during boreal autumn (SON) and winter (DJF), the contribution of SSTs from the external oceans to the ENSO prediction is somewhat less pronounced. This underscores that the importance of inter-basin connections in predicting ENSO varies across different phases of the ENSO cycle.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e"},{"header":"4. Complex and Diverse Impacts of Inter-Basin Interactions on ENSO Prediction","content":"\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003e4.1 Asymmetrical impacts on El Ni\u0026ntilde;o and La Ni\u0026ntilde;a prediction\u003c/h2\u003e \u003cp\u003eDuring the period 1983\u0026ndash;2018, 11 El Ni\u0026ntilde;o and 13 La Ni\u0026ntilde;a events are identified (summarized in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). The definition follows the criteria of the NOAA Climate Prediction Center, i.e., El Ni\u0026ntilde;o (La Ni\u0026ntilde;a) events are defined when the three-month running mean of Ni\u0026ntilde;o 3.4 index is above 0.5\u0026deg;C (below \u0026minus;\u0026thinsp;0.5\u0026deg;C) for five consecutive months. Given that prediction skill differences between the two sets of hindcast experiments of each group increase more rapidly from February onward (recall Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ea-c), the following analysis focuses on composite results from predictions initialized on 1st February.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eIdentified El Ni\u0026ntilde;o (La Ni\u0026ntilde;a) events during the period of 1983\u0026ndash;2018.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"2\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEl Ni\u0026ntilde;o events\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eLa Ni\u0026ntilde;a events\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e1986/87; 1987/88; 1991/92; 1994/95; 1997/98; 2002/03; 2006/07; 2009/10; 2014/15; 2015/16; 2018/19 (11 events in total)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1983/84; 1984/85; 1988/89; 1995/96; 1998/99; 1999/2000; 2000/01; 2005/06; 2007/08; 2008/09;2010/11; 2011/12; 2017/18 (13 events in total)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eAs shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e, the tropical Indian Ocean and Atlantic Ocean SSTs contribute positively to the prediction of La Ni\u0026ntilde;a events, particularly in the equatorial central Pacific during its onset and development phases (recall Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ec). In contrast, for El Ni\u0026ntilde;o events, the prediction skill is not significantly enhanced during the early stage (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ef-i). Quite unexpectedly, taking into account the SSTs from these two oceans acts to reduce rather than increase the prediction skill in many regions at mid-long lead times (after July). Particularly, the skill degradation over the east-central Pacific is the most apparent (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eg).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eTo better understand how the tropical Indian and Atlantic Oceans SSTs act on the prediction of ENSO evolution, we further explore the difference of tropical SST, precipitation and 10m winds between the Obs and Cli experiments based on the composite of ENSO events. In MAM of the La Ni\u0026ntilde;a developing year, a remarkable positive SST anomaly is observed over the NTA region, which induces enhanced convection and thus gives rise to a low-level cyclonic flow over the subtropical northeastern Pacific as a Matsuno-Gill type Rossby-wave response (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ea and S2a). Additionally, the NTA warming generates a Kelvin wave response over the tropical Indian Ocean and propagates eastward to the western Pacific (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ea-c, Fig. S2a-c). During JJA, Atlantic Ni\u0026ntilde;o developed, resulting in an anomalous ascending motion over the Atlantic and anomalous subsidence over the central Pacific (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eb-c, Fig. S2b-c). The induced easterly wind anomalies over the central and western equatorial Pacific can stimulate oceanic upwelling Kelvin waves and thus favor the development of La Ni\u0026ntilde;a (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ed and S2d). The effects of the two Atlantic modes in predicting the development of El Ni\u0026ntilde;o in our model accord with previous observational studies (e.g., Ham et al., \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2013\u003c/span\u003e; Jiang and Li, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Jiang et al., \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Apart from the signals in the Atlantic, the weak warm SSTA in the eastern Indian Ocean during the SON promotes and sustains anomalous easterly winds in the western Pacific, thus also contributing to La Ni\u0026ntilde;a development (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ec).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eIn comparison, the experimental results based on the composite of El Ni\u0026ntilde;o events (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ee-h) are intricate. As demonstrated in previous studies (e.g., Annamalai et al., \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2005\u003c/span\u003e; Hameed, \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2018\u003c/span\u003e), the positive IOD co-develops with El Ni\u0026ntilde;o (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ee-h, Fig. S3e-h). However, in contrary to these studies, the positive IOD provokes easterly wind anomalies in the western Pacific, possibly owing to the strong SST warming in the western Indian Ocean, thus hindering the development of El Ni\u0026ntilde;o in our model (Fig. S3f-g). In addition, in preceding MAM, the tropical Indian Ocean basin-wide warming also induces an easterly wind response in the western Pacific (Fig. S3e). These unexpected forcings from the Indian Ocean jointly deteriorate the prediction of El Ni\u0026ntilde;o (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e). By contrast, the signals over the tropical Atlantic, primarily featuring the negative NTA in MAM and Atlantic Ni\u0026ntilde;a in JJA, lead to westerly wind responses in the western Pacific (Fig. S2e-f). This in turn favors the warm SSTA in the equatorial central-eastern Pacific (Fig. S2e-h) and improves the prediction of El Ni\u0026ntilde;o (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e). Interestingly, the competing contributions from the SSTAs in the tropical Indian and Atlantic Oceans ultimately lead to a negligible difference between the Obs and Cli experiments of AO\u0026thinsp;+\u0026thinsp;IO group (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ee-h). In conclusion, the influence of the tropical Indian and Atlantic Oceans on the occurrence and evolution of ENSO events is asymmetric. SSTAs in the tropical Atlantic significantly contribute to the onset and evolution of ENSO, while SSTAs in the Indian Ocean play a crucial role in the rapid transitions between El Ni\u0026ntilde;o and La Ni\u0026ntilde;a.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003e4.2 Influence of Tropical Indo-Pacific and Atlantic-Pacific teleconnection on ENSO Predictions\u003c/h2\u003e \u003cp\u003eIn general, the IOBM, IOD, NTA, and Atlantic Ni\u0026ntilde;o/Ni\u0026ntilde;a, are the four primary precursors in the tropical Indian and Atlantic Oceans for ENSO prediction. In the observation, the reconstructed DJF Ni\u0026ntilde;o 3.4 index based on the indices of the four precursors using the multiple regression model (see Methods) can well reproduce the observed Ni\u0026ntilde;o 3.4 value (correlation reachs 0.75). This confirms the critical role of remote forcings from these tropical basins outside the Pacific in involving ENSO evolutions. Note that removing the linear trend of these indices has almost no impact on the results.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eBased on the observational data, the CCs between the NTA index in MAM, Atl3 index in JJA and reconstructed Ni\u0026ntilde;o 3.4 index in DJF are \u0026minus;\u0026thinsp;0.72 and \u0026minus;\u0026thinsp;0.27, respectively (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ea-b), implying considerable contributions to El Ni\u0026ntilde;o/La Ni\u0026ntilde;a from preceding SST anomalies over the northern tropical and equatorial Atlantic. These teleconnections are well represented by the model; the corresponding CCs based on the difference between the Obs_AO\u0026thinsp;+\u0026thinsp;IO and Cli_AO\u0026thinsp;+\u0026thinsp;IO experiments are \u0026minus;\u0026thinsp;0.61, and \u0026minus;\u0026thinsp;0.24, respectively (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ef-g), which is close to the observational counterparts albeit with weaker values. Correspondingly, the observational pathways from the forcing of NTA warming/cooling (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003ea, e) and Atlantic Ni\u0026ntilde;o/Ni\u0026ntilde;a (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003ei, m) to the equatorial Pacific are generally reproduced, with the excitation of easterly/westerly wind anomalies and resulted SSTA cooling/warming in the equatorial Pacific Ocean. This success in reproducing the Atlantic-Pacific connections may accounts for the improved ENSO prediction by incorporating the observed tropical Atlantic SSTA (recall Figs.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e and \u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). However, it is worth noting that the responses of the surface winds and SST over the equatorial Pacific in the model experiments are weaker than those in the observation, which is responsible for the slight underestimation of the relationship between ENSO and the Atlantic precursors (recall Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ea, b, f, g).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eRegarding the Indo-Pacific connection, the reconstructed Ni\u0026ntilde;o 3.4 index in DJF is weakly correlated with IOBM index in MAM (CC = -0.21; Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ed), while strongly correlated with DMI in SON (CC\u0026thinsp;=\u0026thinsp;0.73; Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ee) in the observation. However, as revealed by the difference between Obs_AO\u0026thinsp;+\u0026thinsp;IO and Cli_AO\u0026thinsp;+\u0026thinsp;IO experiments, the predicted relationship between the spring IOBM and the following winter ENSO is overestimated, where the CC between DJF Ni\u0026ntilde;o 3.4 and MAM IOBM index is -0.64. In the observation, the IOBM-induced convection excites moderate wind response over the tropical Pacific only during MAM (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003ea), which contributes to the ENSO transition in the spring (e.g., Kug and Kang, 2006). While in the model, obvious convection anomaly in the Maritime Continent (MC) and its wind response in the western Pacific persists across the spring to summer, and thus largely influence the ENSO development (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003ee-f). In stark contrast, the prescribed IOD in autumn shows a negligible correlation with the predicted ENSO in the subsequent winter (CC\u0026thinsp;=\u0026thinsp;0.06). As shown in the previous study (e.g., Ashok et al., \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2001\u003c/span\u003e), the observed positive IOD is coupled with the strong updrafts over the western Indian Ocean and descending branch over the MC (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003ei), which induces anomalous westerly winds in the western Pacific. The observed westerly wind responses are not replicated by the model experiments (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003em). This failure in reproducing the Indo-Pacific interactions may be responsible for the absence of skill improvement by incorporating the observed tropical Indian Ocean SSTAs (recall Figs.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e and \u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). It is worth noting that it remains controversial that whether the Indian Ocean actively force the atmosphere or passively respond to the atmosphere. Further studies are warranted to better comprehend the intricate air-sea interactions in the tropical Indian Ocean.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eTo verify the representation of three-ocean interactions in different model predictions, we also selected hindcast data from six dynamical models (i.e., CanCM4i, CMC1-CanCM3, CMC2-CanCM4, GEM-NEMO, GFDL-SPEAR, NCAR-CESM1) participating in the North American Multi-Model Ensemble (NMME) project (Becker et al., \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2014\u003c/span\u003e). In addition, we also employ the hindcast data of the NUIST-CFS1.0, which is a real-time climate prediction system configured with the same coupled model for our experiment (i.e., the SINTEX-F model). Detailed specifications of these hindcast datasets are provided in Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e. As in the aforementioned analysis for the reconstructed observation and experimental results, the CCs of the four indices (i.e. NTA, Atl3, DMI, and IOBM) with the Ni\u0026ntilde;o 3.4 index are calculated (Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003e). The uniformly negative correlations between the NTA and Ni\u0026ntilde;o 3.4 indices are displayed across all the selected models, although some individual models such as the CMC1_CanCM3 overestimate the correlation while other models such as the GFDL_SPEAR and GEM_NEMO severely underestimate the correlations (Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003ea). Notably, several models poorly capture the impact of Atlantic Ni\u0026ntilde;o/Ni\u0026ntilde;a on ENSO, particularly the GFDL_SPEAR model exhibits an incorrect positive correlation (Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003eb).\u003c/p\u003e \u003cp\u003eAdditionally, consistent with the results revealed by our experiments, the impact of the IOBM on ENSO is generally overestimated across all the selected models except for the GFDL-SPEAR and GEM-NEMO. The latter even displays a positive CC between the IOBM and Ni\u0026ntilde;o 3.4 indices, which is opposed to the observation (Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003ed). Likewise, similar to the NUIST-CFS1.0, about a half of the selected NMME models endure shortcomings in underestimating the positive correlation between the DMI and Ni\u0026ntilde;o 3.4 indices, in particular the CC in the NCAR-CESM1 is near zero (Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003ec). In summary, there are still great challenges and large uncertainties in accurately capturing the tropical inter-basin interactions in the dynamical model prediction systems. From an actual prediction perspective, this limitation will hamper the potential of incorporating tropical SST variations outside the Pacific to enhance ENSO prediction accuracy.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003e4.3 Interdecadal Change in the Inter-Basin Interactions\u0026rsquo; Influence on ENSO Prediction\u003c/h2\u003e \u003cp\u003eGenerally speaking, the aforementioned complexity arises from the asymmetry and interannual cycle of ENSO, the distinction of SST variability in each basin, and the model deficiencies. Apart from them, previous studies have indicated obvious interdecadal change in the ENSO prediction skill and tropical inter-basin interactions. Typically, the ENSO prediction skill has degraded since 2000 (e.g., Hu et al. \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). Surprisingly, compared to the contributions before 2000 (pre-2000), the skill improvement by incorporating the observational SSTAs in the tropical Indian and Atlantic Oceans also becomes less substantial after 2000 (post-2000; Fig. S4). Here we further compare the impact of SSTA over the tropical Indian and Atlantic Oceans on the prediction of each ENSO event during the past four decades. The negative difference of the predicted Ni\u0026ntilde;o 3.4 index between the Obs_AO\u0026thinsp;+\u0026thinsp;IO and Cli_AO\u0026thinsp;+\u0026thinsp;IO experiments is found for all of La Ni\u0026ntilde;a years except the 1984 and 2008 (Fig.\u0026nbsp;\u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e10\u003c/span\u003eb), implying steadily positive contributions of tropical SSTAs outside the Pacific to La Ni\u0026ntilde;a prediction without a clear interdecadal change. In contrast, the contribution from tropical Atlantic and Indian Ocean SSTAs to El Ni\u0026ntilde;o prediction exhibits obvious multi-decadal change (Fig.\u0026nbsp;\u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e10\u003c/span\u003ea). Notably, SSTAs over tropical Atlantic and Indian Ocean positively contribute to predicting all El Ni\u0026ntilde;os in pre-2000, while negatively contributes to predicting most of El Ni\u0026ntilde;os after 2000. The negative contributions may be largely responsible for the skill decline after prescribing the observational SST over the tropical Indian and Atlantic Oceans (recall Figs.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ea, \u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ef-h).\u003c/p\u003e \u003cp\u003eTo further investigate the possible causes of the interdecadal change in the impact of tropical Atlantic and Indian Ocean SSTAs on El Ni\u0026ntilde;o prediction, we conducted a synthetic analysis focusing on the two decadal periods (i.e., pre-2000 and post-2000). During El Ni\u0026ntilde;o developing years in pre-2000, the strong NTA cooling in MAM (Fig.\u0026nbsp;\u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e10\u003c/span\u003ec, Fig. S5a) and Atlantic Ni\u0026ntilde;a in JJA (Fig.\u0026nbsp;\u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e10\u003c/span\u003ed, Fig. S5b) are observed. A significant onset, development, and peak of positive IOD occur in spring, summer, and autumn, respectively (Fig.\u0026nbsp;\u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e10\u003c/span\u003ec-e, Fig. S6a-c). These precursors jointly induce significant westerly wind anomalies over the equatorial western Pacific and thus facilitate the El Ni\u0026ntilde;o growth (Fig.\u0026nbsp;\u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e10\u003c/span\u003ec-e). However, these tropical precursors outside the Pacific are muted in post-2000, and their contributions to El Ni\u0026ntilde;o development shift to a negative influence (Fig.\u0026nbsp;\u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e10\u003c/span\u003eg-i), which is possibly owing to the rapid warming over the tropical Indian and Atlantic Oceans over the last decades (e.g., Luo et al., 2012; McGregor et al., \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e2014\u003c/span\u003e) as well as the changed relationship among three oceans. It is worth noting that either the cold or warm SSTAs over the tropical Indian Ocean in spring and summer seems to inhibit the development of El Ni\u0026ntilde;o during the periods of pre-2000 and post-2000 in our experiments (Fig. S6a-b, e-f), which may be associated with the model deficiencies as was discussed above. The results show that, in autumn, the strong positive IOD in pre-2000 provokes a westerly wind anomaly to improve the El Ni\u0026ntilde;o prediction (Fig. S6c). In contrast, the weak eastern pole cooling and the strengthened warming over most tropical Indian Ocean basin lead to stronger easterly anomalies over the Pacific in SON of post-2000, which worsens the prediction of El Ni\u0026ntilde;o development (Fig. S6g-h). In the tropical Atlantic, particularly in the equatorial eastern Atlantic, the composited SSTA in El Ni\u0026ntilde;o years are much different in pre-2000 and post-2000 (Fig.\u0026nbsp;\u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e10\u003c/span\u003e; Fig. S5). The weakened Atlantic Ni\u0026ntilde;a and the significant West African coastal warming provide unfavorable precondition of El Ni\u0026ntilde;o over the equatorial Pacific during post-2000. Overall, the warming trend in the tropical Indian and Atlantic Oceans, and the weakened tropical Indian Ocean-Pacific relationship (e.g., Xue et al., \u003cspan citationid=\"CR64\" class=\"CitationRef\"\u003e2022\u003c/span\u003e) and Atlantic-Pacific relationship (e.g., Zhang et al., 2023) during pot-2000 may collectively reduce the positive contributions of the SSTAs in the tropical Indian and Atlantic Oceans to the El Ni\u0026ntilde;o development, further affecting the predictability of ENSO.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"5. Conclusion and Discussion","content":"\u003cp\u003eAs a prominent topic in the community of climate science, pantropical interactions gain extensive attention in observational and modeling studies (e.g., Ham et al., \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2013\u003c/span\u003e; Izumo et al., \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2016\u003c/span\u003e; Jin et al., \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Fan et al., \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2024\u003c/span\u003e), which is also an issue of debate (e.g., Zhang et al., \u003cspan citationid=\"CR66\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). Several studies over the last decade have suggested the possible contributions of SST anomalies in the tropical Indian Ocean and/or Atlantic on the ENSO predictability (e.g., Luo et al. \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2010\u003c/span\u003e, \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Frauen and Dommenget, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2012\u003c/span\u003e; Keenlyside et al., \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2013\u003c/span\u003e; Alexander et al., \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). In this study, based on the SINTEX-F coupled model, six sets of sensitivity hindcast experiments are conducted and compared to understand to what extent and how tropical Indian and Atlantic Oceans SSTs jointly and separately impact ENSO prediction since the 1980s.\u003c/p\u003e \u003cp\u003eAs an extension of earlier studies, this study intergrates ENSO cases in recent decades based on more predictions initialized from four seasons, focusing on the impacts from different tropical basins. Consistent with Keenlyside et al. (\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2013\u003c/span\u003e) with the study period being only up to 2005, our results confirm that incorporating the realistic Atlantic SST into the model leads to significant skill increases across the entire tropical Pacific, particularly at longer lead times. Furthermore, when contemporaneously specifying the model\u0026rsquo;s SST in both the Indian and Atlantic Oceans to the observed, larger improvement is exhibited in the equatorial central Pacific, emphasizing the compounded role of the two tropical basins outside the Pacific in enhancing ENSO prediction. If the SST in the tropical Indian and Atlantic Oceans is perfectly predicted, the useful skill in predicting Ni\u0026ntilde;o 3.4 can be extended to one year when initializing from every season. Notably, if the influence of the tropical Indian and Atlantic Oceans are excluded, the skills are significantly decreased for ENSO prediction in target seasons from boreal spring to summer, during which ENSO undergoes a rapid phase transition and exhibits low inherent predictability. Hence, incorporating accurate inter-basin interactions into dynamical models may help alleviate the spring predictability barrier, which was also noted in previous studies based on statistical models (e.g., Ren et al., \u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Jin et al., \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2023\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eApart from the importance of tropical basins outside the Pacific in ENSO prediction, our results also suggest the complexities in understanding and predicting the inter-basin connection. For instance, the impacts of SSTAs in the tropical Indian and Atlantic Oceans on the predictive skills for El Ni\u0026ntilde;o and La Ni\u0026ntilde;a events are asymmetrical. It appears that the La Ni\u0026ntilde;a prediction benefits more than that the El Ni\u0026ntilde;o prediction does. This however conflicts with a few recent studies, which demonstrated the dominance of the tropical Indian and Atlantic Oceans in boosting super El Ni\u0026ntilde;o (e.g., Wang and Wang, \u003cspan citationid=\"CR62\" class=\"CitationRef\"\u003e2021\u003c/span\u003e) based on observations and perfect model hindcast experiments (e.g., Fan et al., \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). This inconsistence may arise from large discrepancies of three-ocean connections between the observation and model prediction, since Fan et al. (\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2024\u003c/span\u003e) mainly focused on the difference among their experiments rather than a comparison between the observation and model results. In addition, our analysis associates this asymmetry to the interdecadal change in the impact of tropical SSTAs outside the Pacific on El Ni\u0026ntilde;o prediction. The influence of the tropical Indian and Atlantic Oceans on El Ni\u0026ntilde;o prediciton has weakened since 2000, and their corresponding contributions to El Ni\u0026ntilde;o prediction changes from positive to negative values. This may arise from the rapid warming in tropical Indian and Atlantic Oceans and the interdecadal variations in the pantropical interactions.\u003c/p\u003e \u003cp\u003eAdditionally, the impacts of the Indian Ocean are yet hard to be thoroughly understood due to the existence of model imperfections. Based on our experimental results, the observed SSTA in the Indian Ocean has limited contributions to ENSO prediction in general, even worsening the predictive skills in the eastern Pacific, which contradicts with previous studies (e.g., Luo et al., \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2010\u003c/span\u003e and \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Jin et al., \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). However, this may result from the inadequate representations of the Indo-Pacific connections. The relationships between the two dominant precursors in the Indian Ocean (i.e., spring IOBM and autumn IOD) and wintertime ENSO in the model experiment apparently deviate from the observed. Note that this challenge is commonly suffered by many dynamical models in the NMME (Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003e) and the CMIP6 models (figure not shown). This underscores the need for further efforts to improve the model\u0026rsquo;s capability in simulating the tropical inter-basin interactions accurately.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eConflict of interests.\u0026nbsp;\u003c/strong\u003eThe authors declare no conflicts of interest or competing interests.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthical Approval.\u0026nbsp;\u003c/strong\u003eThis declaration is \u0026ldquo;not applicable\u0026rdquo;.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgments.\u0026nbsp;\u003c/strong\u003eThis work is supported by National Natural Science Foundation of China (Grant No. 42030605 and 42088101). The numerical calculations in this study were conducted in the High Performance Computing Center of Nanjing University of Information Science \u0026amp; Technology.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData Availability Statement.\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll observational, reanalysis, and model hindcast datasets are publicly available. OISST data at https://psl.noaa.gov/data/gridded/data.noaa.oisst.v2.html. Global Precipitation Climatology Project precipitation data at https://psl.noaa.gov/data/gridded/data.gpcp.html. NCEP2 reanalysis data at https://psl.noaa.gov/data/gridded/data.ncep.reanalysis2.html. NMME hindcast data at http://iridl.ldeo.columbia.edu/SOURCES/.Models/.NMME/.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eAdler RF, and Coauthors (2003) The Version-2 Global Precipitation Climatology Project (GPCP) Monthly Precipitation Analysis (1979-Present). J Hydrometeorology, 4(6), 1147\u0026ndash;1167.\u003c/li\u003e\n\u003cli\u003eAlexander MA, Shin S-I, and Battisti DS (2022) The Influence of the Trend, Basin Interactions and Ocean Dynamics on Tropical Ocean Prediction. Geophys Res Lett, 49, e2021GL096120.\u003c/li\u003e\n\u003cli\u003eAnnamalai H, Xie SP, Mccreary JP, and Murtugudde R (2005) Impact of Indian Ocean Sea Surface Temperature on Developing El Ni\u0026ntilde;o. J Clim, 18, 302\u0026ndash;319.\u003c/li\u003e\n\u003cli\u003eAshok K, Guan Z, and Yamagata T (2001) Impact of the Indian Ocean dipole on the relationship between the Indian monsoon rainfall and ENSO. Geophys Res Lett, 28, 4499\u0026ndash;4502.\u003c/li\u003e\n\u003cli\u003eBarnston AG, Tippett MK, L\u0026apos;heureux ML, Li S, and Dewitt DG (2012) Skill of Real-Time Seasonal ENSO Model Predictions during 2002-11: Is Our Capability Increasing? Bull Amer Meteor Soc, 93, 631\u0026ndash;651.\u003c/li\u003e\n\u003cli\u003eBecker E, den Dool H, Zhang Q (2014) Predictability and forecast skill in NMME. J Clim, 27(15): 5891\u0026ndash;5906.\u003c/li\u003e\n\u003cli\u003eBjerknes J (1969) Atmospheric teleconnections from the equatorial Pacific. Mon Wea Rev, 97, 163\u0026ndash;172.\u003c/li\u003e\n\u003cli\u003eCai W, Wu L, Lengaigne M, and Chang P (2019) Pantropical climate interactions. Science, 363, eaav4236.\u003c/li\u003e\n\u003cli\u003eDing H, Keenlyside NS, and Latif M (2012) Impact of the Equatorial Atlantic on the El Ni\u0026ntilde;o Southern Oscillation. Clim Dyn, 38, 1965\u0026ndash;1972.\u003c/li\u003e\n\u003cli\u003eDoi T, Storto A, Behera SK, Navarra A, and Yamagata T (2017) Improved Prediction of the Indian Ocean Dipole Mode by Use of Subsurface Ocean Observations. J Clim, 30, 7953\u0026ndash;7970.\u003c/li\u003e\n\u003cli\u003eExarchou E, Ortega P, Rodr\u0026iacute;guez-Fonseca B, Losada T, Polo I, and Prodhomme C (2021) Impact of equatorial Atlantic variability on ENSO predictive skill. Nat Commun, 12, 1612.\u003c/li\u003e\n\u003cli\u003eFan L, and Meng X (2023) The Asymmetric Predictive Power of Indian Ocean Dipole for Subsequent Year\u0026apos;s ENSO: Role of Atlantic Ocean as an Intermediary. Geophys Res Lett, 50, e2023GL105525.\u003c/li\u003e\n\u003cli\u003eFan HJ et al (2024) Coupling is key for the tropical Indian and Atlantic oceans to boost super El Ni\u0026ntilde;o. Sci Adv, 10, eadp2281.\u003c/li\u003e\n\u003cli\u003eFang X-H, and Mu M (2018) Both air-sea components are crucial for El Ni\u0026ntilde;o forecast from boreal spring. Sci Rep, 8, 10501.\u003c/li\u003e\n\u003cli\u003eFrauen C, and Dommenget D (2012) Influences of the tropical Indian and Atlantic Oceans on the predictability of ENSO. Geophys Res Lett, 39.\u003c/li\u003e\n\u003cli\u003eHa K-J, Chu J-E, Lee J-Y, and Yun K-S (2017) Inter-basin coupling between the tropical Indian and Pacific Ocean on interannual timescale: observation and CMIP5 reproduction. Clim Dyn, 48, 459\u0026ndash;475.\u003c/li\u003e\n\u003cli\u003eHam Y-G, Kug J-S, Park J-Y, and Jin F-F (2013) Sea surface temperature in the north tropical Atlantic as a trigger for El Ni\u0026ntilde;o/Southern Oscillation events. Nat Geosci, 6, 112\u0026ndash;116.\u003c/li\u003e\n\u003cli\u003eHam Y-G, and Kug J-S (2015) Role of north tropical atlantic SST on the ENSO simulated using CMIP3 and CMIP5 models. Clim Dyn, 45, 3103\u0026ndash;3117.\u003c/li\u003e\n\u003cli\u003eHam Y-G, Kim J-H, and Luo J-J (2019) Deep learning for multi-year ENSO forecasts. Nature, 573, 568\u0026ndash;572.\u003c/li\u003e\n\u003cli\u003eHameed S (2018) The Indian Ocean Dipole. Oxon Res Encycl Climate Sci.\u003c/li\u003e\n\u003cli\u003eHu Z-Z, Kumar A, Huang B, Zhu J, L\u0026rsquo;Heureux M, McPhaden MJ, and Yu J-Y (2020) The Interdecadal Shift of ENSO Properties in 1999/2000: A Review. J Clim, 33, 4441-4462.\u003c/li\u003e\n\u003cli\u003eIzumo T, Vialard J, Dayan H, Lengaigne M, and Suresh I (2016) A simple estimation of equatorial Pacific response from windstress to untangle Indian Ocean Dipole and Basin influences on El Ni\u0026ntilde;o. Clim Dyn, 46, 2247\u0026ndash;2268.\u003c/li\u003e\n\u003cli\u003eJansen MF, Dommenget D, and Keenlyside N (2009) Tropical Atmosphere-Ocean Interactions in a Conceptual Framework. J Clim, 22, 550\u0026ndash;567.\u003c/li\u003e\n\u003cli\u003eJiang F, Zhang W, Jin F-F, Stuecker MF, and Allan R (2021) El Ni\u0026ntilde;o Pacing Orchestrates Inter-Basin Pacific-Indian Ocean Interannual Connections. Geophys Res Lett, 48, e2021GL095242.\u003c/li\u003e\n\u003cli\u003eJiang L, and Li T (2021) Impacts of Tropical North Atlantic and Equatorial Atlantic SST Anomalies on ENSO. J Clim, 34, 5635\u0026ndash;5655.\u003c/li\u003e\n\u003cli\u003eJiang L, Li T, and Ham Y-G (2022) Critical Role of Tropical North Atlantic SSTA in Boreal Summer in Affecting Subsequent ENSO Evolution. Geophys Res Lett, 49, e2021GL097606.\u003c/li\u003e\n\u003cli\u003eJin F-F (1997) An Equatorial Ocean Recharge Paradigm for ENSO. Part I: Conceptual Model. J Atmos Sci, 54, 811\u0026ndash;829.\u003c/li\u003e\n\u003cli\u003eJin Y, Meng X, Zhang L, Zhao Y, Cai W, and Wu L (2023) The Indian Ocean Weakens the ENSO Spring Predictability Barrier: Role of the Indian Ocean Basin and Dipole Modes. J Clim, 36, 8331\u0026ndash;8345.\u003c/li\u003e\n\u003cli\u003eKanamitsu M, Ebisuzaki W, Woollen J, Yang S-K, Hnilo JJ, Fiorino M, and Potter GL (2002) NCEP-DOE AMIP-II Reanalysis (R-2). Bull Amer Meteor Soc, 83, 1631\u0026ndash;1644.\u003c/li\u003e\n\u003cli\u003eKeenlyside NS, and Latif M (2007) Understanding Equatorial Atlantic Interannual Variability. J Clim, 20, 131-142.\u003c/li\u003e\n\u003cli\u003eKeenlyside NS, Ding H, and Latif M (2013) Potential of equatorial Atlantic variability to enhance El Ni\u0026ntilde;o prediction. Geophys Res Lett, 40, 2278\u0026ndash;2283.\u003c/li\u003e\n\u003cli\u003eKug J, Kirtman BP, and Kang I (2006) Interactive Feedback between ENSO and the Indian Ocean in an Interactive Ensemble Coupled Model. J Clim, 19, 6371\u0026ndash;6381.\u003c/li\u003e\n\u003cli\u003eLatif M, and Gr\u0026ouml;tzner A (2000) The equatorial Atlantic oscillation and its response to ENSO. Clim Dyn, 16, 213\u0026ndash;218.\u003c/li\u003e\n\u003cli\u003eLi T, Tang Y, Lian T, and Hu A (2024) Quantifying the Relative Contributions of the Global Oceans to ENSO Predictability With Deep Learning. Geophys Res Lett, 51, e2023GL106584.\u003c/li\u003e\n\u003cli\u003eLuo J-J, Masson S, Behera S, Shingu S, and Yamagata T (2005) Seasonal Climate Predictability in a Coupled OAGCM Using a Different Approach for Ensemble Forecasts. J Clim, 18, 4474\u0026ndash;4497.\u003c/li\u003e\n\u003cli\u003eLuo J-J, Masson S, Behera SK, and Yamagata T (2008) Extended ENSO Predictions Using a Fully Coupled Ocean-Atmosphere Model. J Clim, 21, 84\u0026ndash;93.\u003c/li\u003e\n\u003cli\u003eLuo J-J, Zhang R, Behera SK, and Yamagata T (2010) Interaction between El Ni\u0026ntilde;o and Extreme Indian Ocean Dipole. J Clim, 23, 726\u0026ndash;742.\u003c/li\u003e\n\u003cli\u003eLuo J-J, Yuan C, Sasaki W, and Masson S (2016) Current status of intraseasonal-seasonal-to-interannual prediction of the Indo-Pacific climate. Indo-Pacific Climate Variability and Predictability, pp.63\u0026ndash;107. \u003c/li\u003e\n\u003cli\u003eLuo J-J, Liu G, Hendon H, Alves O and Yamagata T (2017) Inter-basin sources for two-year predictability of the multi-year La Ni\u0026ntilde;a event in 2010-2012. Sci Rep, 7, 2276.\u003c/li\u003e\n\u003cli\u003eLyu P, Tang T, and Ling F (2024) Robust and Explainable ENSO Forecasts with Hybrid Convolution and Transformer Networks. Adv Atmos Sci, 41, 1289\u0026ndash;1298.\u003c/li\u003e\n\u003cli\u003eMcGregor S, Timmermann A, Stuecker MF, England MH, Merrifield M, Jin FF, and Chikamoto Y (2014) Recent Walker circulation strengthening and Pacific cooling amplified by Atlantic warming. Nat Clim Change, 4(10), 888\u0026ndash;892. \u003c/li\u003e\n\u003cli\u003eMcphaden MJ, Busalacchi AJ, Cheney R, and Takeuchi K (1998) The Tropical Ocean-Global Atmosphere observing system: A decade of progress. J Geophys Res: Oceans, 103, 14169\u0026ndash;14240.\u003c/li\u003e\n\u003cli\u003eMeinen CS, and Mcphaden MJ (2000) Observations of Warm Water Volume Changes in the Equatorial Pacific and Their Relationship to El Ni\u0026ntilde;o and La Ni\u0026ntilde;a. J Clim, 13, 3551\u0026ndash;3559.\u003c/li\u003e\n\u003cli\u003eMeng X-L, Rosenthal R, and Rubin DB (1992) Comparing correlated correlation coefficients. Psychol Bull, 111, 172\u0026ndash;175.\u003c/li\u003e\n\u003cli\u003eMerryfield WJ, and Coauthors (2020) Current and Emerging Developments in Subseasonal to Decadal Prediction. Bull Amer Meteor Soc, 101, E869\u0026ndash;E896.\u003c/li\u003e\n\u003cli\u003eOhba M, and Ueda H (2009) Seasonally Different Response of the Indian Ocean to the Remote Forcing of El Ni\u0026ntilde;o: Linking the Dynamics and Thermodynamics. SOLA, 5, 176\u0026ndash;179. \u003c/li\u003e\n\u003cli\u003eOhba M, and Watanabe M (2012) Role of the Indo-Pacific Inter-basin Coupling in Predicting Asymmetric ENSO Transition and Duration. J Clim, 25, 3321\u0026ndash;3335.\u003c/li\u003e\n\u003cli\u003ePalmer TN, and Coauthors (2004) Development Of A European Multimodel Ensemble System of Seasonal-to-Interannual Prediction (Demeter). Bull Amer Meteor Soc, 85, 853\u0026ndash;872.\u003c/li\u003e\n\u003cli\u003ePark JH, Li T (2019) Interdecadal modulation of El Ni\u0026ntilde;o-tropical North Atlantic teleconnection by the Atlantic multi-decadal oscillation. Clim Dyn, 52, 5345\u0026ndash;5360.\u003c/li\u003e\n\u003cli\u003ePhilander S (1983) El Ni\u0026ntilde;o Southern Oscillation phenomena. Nature, 302, 295\u0026ndash;301.\u003c/li\u003e\n\u003cli\u003ePolo I, Martin-Rey M, Rodriguez-Fonseca B, Kucharski F, and Mechoso CR (2015) Processes in the Pacific La Ni\u0026ntilde;a onset triggered by the Atlantic Ni\u0026ntilde;o. Clim Dyn, 44, 115\u0026ndash;131.\u003c/li\u003e\n\u003cli\u003eRen HL, Zuo J, and Deng Y (2019) Statistical predictability of Ni\u0026ntilde;o indices for two types of ENSO. Clim Dyn, 52, 5361\u0026ndash;5382.\u003c/li\u003e\n\u003cli\u003eRen HL, Zheng F, Luo J-J, and Zhou G (2020) A Review of Research on Tropical Air-Sea Interaction, ENSO Dynamics and ENSO Prediction in China. J Meteorol Res, 34, 43\u0026ndash;62.\u003c/li\u003e\n\u003cli\u003eReynolds RW, Rayner NA, Smith TM, Stokes DC, and Wang W (2002) An Improved In Situ and Satellite SST Analysis for Climate. J Clim, 15, 1609\u0026ndash;1625.\u003c/li\u003e\n\u003cli\u003eSaji N, Goswami B, Vinayachandran P, and Yamagata T (1999) A dipole mode in the tropical Indian Ocean. Nature, 401, 360\u0026ndash;363.\u003c/li\u003e\n\u003cli\u003eSuarez MJ, and Schopf PS (1988) A Delayed Action Oscillator for ENSO. J Atmos Sci, 45, 3283\u0026ndash;3287.\u003c/li\u003e\n\u003cli\u003eTang Y, Zhang R-H, Liu T, and Mu M (2018) Progress in ENSO prediction and predictability study. Natl Sci Rev, 5, 826\u0026ndash;839.\u003c/li\u003e\n\u003cli\u003eTimmermann A, An SI, Kug JS, and Zhang X (2018) El Ni\u0026ntilde;o-Southern Oscillation complexity. Nature, 559, 535\u0026ndash;545.\u003c/li\u003e\n\u003cli\u003eTseng Yh, Hu ZZ, Ding R, and Chen HC (2017) An ENSO prediction approach based on ocean conditions and ocean\u0026ndash;atmosphere coupling. Clim Dyn, 48, 2025\u0026ndash;2044.\u003c/li\u003e\n\u003cli\u003eWang CZ, Kucharski F, Barimalala R, and Bracco A (2009) Teleconnections of the tropical Atlantic to the tropical Indian and Pacific Oceans: A review of recent findings. Meteorologische Zeitschrift, 18, 445\u0026ndash;454.\u003c/li\u003e\n\u003cli\u003eWang CZ (2019) Three-ocean interactions and climate variability: a review and perspective. Clim Dyn, 53, 5119\u0026ndash;5136.\u003c/li\u003e\n\u003cli\u003eWang JZ, Wang C (2021) Joint boost to super El Ni\u0026ntilde;o from the Indian and Atlantic Oceans. J Clim, 34(12): 4937\u0026ndash;4954.\u003c/li\u003e\n\u003cli\u003eWang R, He J, Luo J-J, and Chen L (2024) Atlantic warming enhances the influence of Atlantic Ni\u0026ntilde;o on ENSO. Geophys Res Lett, 51, e2023GL108013.\u003c/li\u003e\n\u003cli\u003eXue J, Luo J-J, Zhang W, and Yamagata T (2022) ENSO\u0026ndash;IOD inter-basin connection is controlled by the Atlantic Multidecadal Oscillation. Geophys Res Lett, 49, e2022GL101571.\u003c/li\u003e\n\u003cli\u003eZhang L, and Coauthors (2023) Emergence of the Central Atlantic Ni\u0026ntilde;o. Sci Adv, 9,eadi5507.\u003c/li\u003e\n\u003cli\u003eZhang W, Jiang F, Stuecker MF, Jin F-F, and Timmermann A (2021) Spurious North Tropical Atlantic precursors to El Ni\u0026ntilde;o. Nat Commun, 12, 3096.\u003c/li\u003e\n\u003cli\u003eZhao M, Hendon HH, Alves O, Liu G, and Wang G (2016) Weakened Eastern Pacific El Ni\u0026ntilde;o Predictability in the Early Twenty-First Century. J Clim, 29, 6805-6822.\u003c/li\u003e\n\u003cli\u003eZhao S, Jin F-F, Stuecker MF, and Cai W (2024) Explainable El Ni\u0026ntilde;o predictability from climate mode interactions. Nature, 630, 891\u0026ndash;898.\u003c/li\u003e\n\u003cli\u003eZhao Y, Jin Y, Capotondi A, Li J, and Sun D (2023) The Role of Tropical Atlantic in ENSO Predictability Barrier. Geophys Res Lett, 50, e2022GL101853.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":true,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"climate-dynamics","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"cldy","sideBox":"Learn more about [Climate Dynamics](https://www.springer.com/journal/382)","snPcode":"382","submissionUrl":"https://submission.nature.com/new-submission/382/3","title":"Climate Dynamics","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"ENSO predictions, Tropical basin interactions, Complexity, Model deficiency","lastPublishedDoi":"10.21203/rs.3.rs-5352476/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-5352476/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eTropical basin interactions (TBIs) are widely involved in El Ni\u0026ntilde;o-Southern Oscillation (ENSO) evolution. Previous dynamical prediction studies often focused on single basin, single start month, and few events, which normally overlooked the complexity in the TBIs\u0026rsquo; impact on ENSO prediction. To address these limitations, we conducted six sets of sensitivity hindcast experiments initializing from different seasons during 1983\u0026ndash;2018, in which observed monthly and climatological sea surface temperatures (SSTs) over the tropical Indian (TIO) and Atlantic Ocean (TAO) are separately and synchronously prescribed. Results indicate synergistic but complicated roles of tropical SST anomalies (SSTAs) outside the Pacific in predicting ENSO. The results suggest more prominent contributions from TAO SSTAs, due to the well-captured teleconnections between ENSO and primary precursors in the North tropical and equatorial Atlantic. Conversely, the model exhibits large biases in replicating the relationship between ENSO and the basin-wide and dipole modes in the Indian Ocean, weakening the TIO SSTAs\u0026rsquo; contributions. Moreover, SSTAs over the remote basins exert asymmetrical and phase-dependent influences on ENSO predictions; more remarkable contributions are found during La Ni\u0026ntilde;a and ENSO transition-development phases, indicating the TBIs\u0026rsquo; importance in improving the spring barrier of ENSO prediction. Additionally, the impact of TBIs on ENSO prediction displays an interdecadal change; SSTAs outside the Pacific improve (degrade) El Ni\u0026ntilde;o prediction before (after) 2000, which may be associated with rapid warming in the TIO and TAO. Our results suggest high complexity in the TIO and TAO\u0026rsquo;s influence on ENSO prediction, stimulating future efforts for better understanding and models\u0026rsquo; performance.\u003c/p\u003e","manuscriptTitle":"Complex Influences of Tropical Indian and Atlantic Oceans on ENSO Prediction","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-11-29 15:28:03","doi":"10.21203/rs.3.rs-5352476/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Major Revision","date":"2025-01-27T18:05:13+00:00","index":"","fulltext":""},{"type":"reviewerAgreed","content":"","date":"2024-11-13T00:13:11+00:00","index":0,"fulltext":""},{"type":"reviewersInvited","content":"","date":"2024-11-12T21:58:21+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2024-11-06T10:38:42+00:00","index":"","fulltext":""},{"type":"submitted","content":"Climate Dynamics","date":"2024-10-29T04:27:29+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"climate-dynamics","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"cldy","sideBox":"Learn more about [Climate Dynamics](https://www.springer.com/journal/382)","snPcode":"382","submissionUrl":"https://submission.nature.com/new-submission/382/3","title":"Climate Dynamics","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"dd1e296d-bf6c-462d-9106-c2fdbe7946b1","owner":[],"postedDate":"November 29th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2025-06-09T16:07:30+00:00","versionOfRecord":{"articleIdentity":"rs-5352476","link":"https://doi.org/10.1007/s00382-025-07721-9","journal":{"identity":"climate-dynamics","isVorOnly":false,"title":"Climate Dynamics"},"publishedOn":"2025-06-07 15:57:42","publishedOnDateReadable":"June 7th, 2025"},"versionCreatedAt":"2024-11-29 15:28:03","video":"","vorDoi":"10.1007/s00382-025-07721-9","vorDoiUrl":"https://doi.org/10.1007/s00382-025-07721-9","workflowStages":[]},"version":"v1","identity":"rs-5352476","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-5352476","identity":"rs-5352476","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}