1 Introduction

The tropics are generally considered to be the driving force of the global climate system, and the El Niño–Southern Oscillation (ENSO) is seen as the typical representation of the interannual tropical coupled air–sea interaction mode. The ENSO exerts a significant influence on climate around the Pacific region, and even globally, via its regulation of the meridional circulation and excitation of the planetary Rossby wave train. The ENSO-related Rossby wave has the potential to modulate the eddy-driven jet in the mid latitudes and further regulate the dominant atmospheric mode in the mid and high latitudes, over both interannual and long-term timescales (Alexander et al. 2002; Carril and Navarra 2001; Ding et al. 2012; Fogt and Bromwich 2006; Grassi et al. 2005; Kwok and Comiso 2002; Lau 1997; L’Heureux and Thompson 2006; Ribera and Mann 2003; Trenberth et al. 1998; Yuan and Martinson 2000). However, the extratropics can also influence the tropics via the atmospheric bridge and the oceanic tunnel (Liu and Yang 2003). The extratropical climate system and associated thermal forcing have the potential to modulate the tropical atmospheric circulation, such as the intertropical convergence zone (ITCZ; Kang et al. 2008), zonal winds (Gollan and Greatbatch 2015), meridional circulation (Liu et al. 2015, 2016; Zheng et al. 2015, 2018; Wu et al. 2009), and Rossby wave train (Liu et al. 2018, 2020).

Furthermore, observational and modeling studies have correlated the preceding extratropical air–sea signal in the Northern Hemisphere (NH) with the subsequent development of the ENSO, which allows the extratropical influence to propagate into the tropical Pacific Ocean via various air–sea interactions (Anderson 2004; Alexander et al. 2010; Boschat et al. 2013; Chang et al. 2007; Ding et al. 2015a, 2017; Li et al. 2014; Tseng et al., 2017; Vimont et al. 2001, 2003; Wang et al. 2012; Zhang et al. 2009a, b). This influence occurs because atmospheric fluctuations in the extratropical NH can imprint their signals onto the local sea surface temperature (SST) via air–sea interactions, and the corresponding SST footprints can persist for several seasons and regulate the atmospheric circulation. These processes have the potential to modulate zonal wind and SST anomalies through the wind–evaporation–SST feedback mechanism, which propagate extratropical influence into the tropics and further act as the precursor indicators to the subsequent ENSO events (Chiang and Vimont 2004; Chen et al. 2014; Vimont et al. 2001, 2003). Recently, there has been increasing evidence to show that extratropical Pacific climate variability in the Southern Hemisphere (SH) makes an indispensable contribution to the development of ENSO, including via the extratropical SST anomalies (Ding et al. 2015b; Yamazaki and Watanabe 2015; Zhang et al. 2014), the transverse cell in southern Australia (Hong and Jin 2014), and the SH annular mode (SAM; Zheng et al. 2017). The mechanism involved is driven mainly by the extratropical seasonal SST footprints (Zhang et al. 2014; Ding et al. 2015b), like its NH counterpart, or by the response of the meridional circulation to extratropical thermal forcing (Yamazaki and Watanabe 2015; Zheng et al. 2017), both of which have the potential to propagate the extratropical influence into the tropical Pacific and regulate zonal wind and SST anomalies in the same region.

Previous studies of the influence of the extratropical signal on tropical SST variability have focused mainly on the Pacific Ocean region, whether in the NH or SH. For example, the anomalous atmospheric or oceanic signal that originates from the extratropical Pacific Ocean and propagates into the equatorial region to regulate ENSO variability via an air–sea interaction process in the Pacific Basin. Several studies have also discussed the possible impact of the extratropical signal outside the Pacific on the tropical Pacific SST (Terray 2011; Levine et al. 2017). However, the possibility of a signal excited by the SAM outside the Pacific Basin that could affect the tropical Pacific has never been addressed. Given that the SAM is the leading atmospheric mode in the SH extratropics and involves hemispheric atmosphere–ocean–sea-ice system variability in the extratropical SH (Thompson et al. 2011), and the SAM signal can be transmitted into the NH via the Indian Ocean (Nan et al. 2009; Dou et al. 2017, 2020), tropical Pacific Ocean (Liu et al. 2018), and Atlantic Ocean (Wu et al. 2015). Therefore, we hypothesize that the propagation of the SAM signal into the tropics is also not limited only to the Pacific tunnel. The present study aims to identify the possible impact of the April–May (AM SAM) on sea temperature during the subsequent three seasons in the central Pacific Ocean and to investigate the corresponding contribution of air–sea interaction processes originating from the Indian Ocean. The outline of this paper is as follows. Detailed descriptions of the datasets and numerical model used are presented in Sect. 2. In the following section, we reveal the linkage between the preceding SAM and the subsequent sea temperature variability in the central Pacific Ocean. We then consider the corresponding mechanisms responsible for this relationship. A brief discussion and our concluding remarks are given in Sect. 4.

2 Datasets and model

2.1 Datasets

For the present study, we extracted monthly mean atmospheric variables from the NCEP/NCAR reanalysis datasets (National Centers for Environmental Prediction–National Center for Atmospheric Research: https://psl.noaa.gov/data/gridded/data.ncep.reanalysis.html), which have a resolution of 2.5° × 2.5°, including geopotential height (HGT), and the horizontal and vertical components of wind and surface heat flux. More details regarding the NCEP/NCAR data can be found in Kalnay et al. (1996). We also used the European Centre for Medium-Range Weather Forecasts (ECMWF) interim reanalysis (ERA-Interim; Dee et al. 2011: https://apps.ecmwf.int/datasets/data/interim-land) and fifth generation ECMWF (ERA5; Hersbach et al. 2020: https://cds.climate.copernicus.eu/cdsapp#!/dataset/reanalysis-era5-single-levels-monthly-means?tab=form) reanalysis to validate our results. The performance of these three datasets were basically consistent with each other. To represent the variability of convective activity, we used the monthly outgoing longwave radiation (OLR) data from the National Oceanic and Atmospheric Administration (NOAA) Interpolated Outgoing Longwave Radiation dataset (Liebmann and Smith 1996: https://psl.noaa.gov/data/gridded/data.interp_OLR.html) gridded at 2.5° × 2.5°. Monthly SST data were obtained from the NOAA ERSST V5 (Extended Reconstructed SST Version 5) dataset at a resolution of 2° × 2° (Huang et al. 2017: https://psl.noaa.gov/data/gridded/data.noaa.ersst.v5.html). We used subsurface sea temperature and circulation variables from the GODAS (NCEP Global Ocean Data Assimilation System) reanalysis dataset on a 0.33° × 1.0° grid (Behringer et al. 1998: https://psl.noaa.gov/data/gridded/data.godas.html). Anomalies in the present study were identified by detrending the data, which removed the long-term monthly mean. Our study focused on the satellite era (1979–2018) to ensure reliable data quality in the SH.

We used two SAM indexes to represent the variability of the SAM: an index modified by Nan and Li (2003), which we refer to as “SAM_Nan&Li” (see http://ljp.gcess.cn/dct/page-/65609), and an index defined by Marshall (2003), which we refer to as “SAM_Marshall” (see http://www.nerc-bas.ac.uk/public/icd/gjma/newsam.1957.2007.txt). The SAM_Nan&Li index is defined as the difference in the standardized zonal mean sea level pressure (SLP) between 40° S and 70° S calculated using the reanalysis data. The SAM_Marshall index is station-based and uses observational SLP from stations at mid and high latitudes in the SH to calculate substitution of the zonal mean SLP. These two SAM indexes are highly correlated with each other, with a correlation coefficient of 0.9 for AM. We used the Niño 3.4, Niño 3, and Niño 4 indexes obtained from NOAA’s Climate Prediction Center to represent SST variability in the tropical Pacific, and also the Modoki index, calculated according to the definition of Ashok et al. (2007).

2.2 Model

We used an atmospheric general circulation model (AGCM) to investigate the atmospheric response to SAM-related SST variability. Control and sensitive experiments were performed based on the Community Atmosphere Model version 5.3 (CAM5; Hurrell et al. 2013), which is the atmospheric component of the Community Earth System Model (CESM). The horizontal resolution of CAM5 is approximately 2.8° × 2.8° (T42), and the model has 30 vertical layers. We used the F component set to formulate the model. This model is widely used to explore the response of the atmospheric circulation to extratropical SST forcing (Zheng et al. 2015; Liu et al. 2015, 2018). The control experiment was forced by the climatological SST. CAM5 is able to simulate a climatic circulation that is broadly consistent with that seen in the observations. In the southeastern Indian Ocean (SEIO: 30–45° S, 70–115° E), positive and negative SAM-related SST anomaly patterns based on the AM SEIO SST index > 1, and SEIO SST index < − 1, were imposed onto the AM climatological SST, and we refer to these as the positive and negative sensitivity experiments, respectively. For both experiments, the model integration was run for 30 years, and our analysis was based on data from the last 20 years.

We used the Gill model (Gill 1980; Xing et al. 2014a) to investigate how the atmosphere responds to forcing by convective heating. The Gill model is based on a linear shallow water equation, and has been widely used to explain the response of the atmospheric circulation to the equatorial heating forcing (Wu and Zhou 2013; Xing et al. 2014b, 2016; Zhang et al. 2018; Li et al. 2019). Corresponding to the equatorial symmetric warm forcing, there is an easterly (westerly) wind to the east (west) of the heating center, and ascending motions occurs above the heating center (Gill 1980; Xing et al. 2014a).

3 Results

3.1 Observational evidence

Given that ENSO shows the largest seasonal variability during the December–January–February (DJF) period, we used various ENSO indexes to express the variability in DJF SST in the tropical Pacific Ocean. The observed lead–lag correlations between the DJF ENSO indexes and the 2-month running mean of the SAM index (SAMI) are shown in Fig. 1a. The most notable feature is that the preceding AM SAMI shows a significant negative correlation with the DJF Modoki index (red line), with a correlation coefficient (r) of − 0.36 that is significant at the 95% confidence level (Fig. 1a). In addition, there is also a significant relationship between the DJF Niño 4 index (blue line) and the AM SAMI in Fig. 1a. Our correlation analysis of the SAMI and the various ENSO indexes indicate that the AM SAM is significantly related to the subsequent SST variability, and this connection occurs mainly in the tropical central Pacific Ocean rather than the eastern Pacific Ocean. The lead–lag correlation between the DJF Modoki index and the different SAM indexes (Fig. 1b) also confirms the above mentioned negative relationship between the AM SAM and the subsequent DJF Modoki, and further indicates that this relationship is independent of the choice of SAMI and is robust.

Fig. 1
figure 1

a Lead–lag correlation between the SAM_ Nan&Li index and the DJF ENSO index. The red, blue, purple, and green lines indicate the Modoki, Niño4, Niño 3, and Niño 3.4 indexes, respectively. b Lead–lag correlation between the different SAM indexes and DJF Modoki index. The red and blue lines indicate the SAM_Nan&Li and SAM_Marshall indexes, respectively. The dashed black lines indicate the 90%, 95%, and 99% confidence levels for Student’s t test. DJ is defined as the December of this the present year and January of the next following year. “DJ (−1)” indicates the boreal winter of preceding year, and the “DJ (0)” indicates the boreal winter of present year

Full size image

We note that there is also a significant negative correlation for the case where the DJF Modoki leads the AM SAM (r = − 0.40, significant at the 99% confidence level), which indicates that the AM SAM and DJF Modoki may statistically influence each other (such as: preceding DJF Modoki leads to AM SAM leads to following DJF Modoki), and suggests that the relationship between the AM SAM and the following DJF Modoki may contain the influence of the preceding DJF Modoki. To investigate this, we used partial correlation analysis to remove the impact of the preceding DJF Modoki signal. The results showed that AM SAMI was still significantly negatively correlated with the following DJF Modoki index (Fig. 2a). In addition, running correlation analysis further indicated that this relationship remained stable and strengthened after the 1970s (Fig. 2b).

Fig. 2
figure 2

a Lead–lag correlation between the different SAM indexes and the DJF Modoki index without the preceding DJF Modoki signal. b Running correlation between the different SAM indexes and the DJF Modoki index using a 31-year moving window. The red and blue lines are the SAM_Nan&Li and SAM_Marshall indexes, respectively. The dashed black lines indicate the 90%, 95%, and 99% confidence levels for Student’s t test

Full size image

Furthermore, the preceding AM zonal averaged HGT in the SH is also significantly correlated with the DJF Modoki index (Fig. 3a), and the distinctive feature is a negative SAM pattern that involves a significant seesaw HGT structure between the mid and high latitudes in the SH. This negative SAM pattern is also significant after removing the preceding DJF Modoki signal (Fig. 3b). This provides further evidence that the positive (negative) preceding AM SAM frequently favors the following anomalous cold (warm) water over the tropical central Pacific Ocean during DJF, and that this relationship is robust. In the next section, we investigate the possible physical mechanism that drives this relationship.

Fig. 3
figure 3

Regression maps between the DJF Modoki index and AM zonal mean geopotential height in the SH a before and b after removing the preceding DJF Modoki signal in NCEP reanalysis. The black and green stippled areas indicate regression coefficients significant at the 90% and 95% confidence levels, respectively

Full size image

3.2 A possible physical mechanism

3.2.1 SST anomalies corresponding to the AM SAM and DJF Modoki

Here, we are interested in how the preceding SAM persists into the following DJF season and exerts an influence on the tropical central Pacific Ocean. As the persistence of the atmospheric signal itself is of short duration, we take into consideration the underlying ocean, which has the potential to be imprinted by the SAM-related surface wind via the modulation of the air–sea heat exchange and inducing the oceanic meridional Ekman transport (Thompson and Wallace 2000a, b; Trenberth et al. 2005; Sen Gupta and England 2006; Sallée et al. 2010; Yeo and Kim 2015; Dou and Wu 2018). Figure 4 presents maps of the regression coefficients of the AM SST anomalies based on the following DJF Modoki index (Fig. 4a) and the AM SAMI (Fig. 4b). Corresponding to the positive DJF Modoki case, there are notable cold SST anomalies in the SEIO (green box in Fig. 4) during AM (Fig. 4a). By contrast, there are significant positive SST anomalies correlated with the positive AM SAM phase in the same region (Fig. 4b). These anomalous SST patterns are also significant even after removing the preceding DJF Modoki signal (Fig. 4c, d). Unless otherwise specified, in the following analysis, all results exclude the influence of the preceding DJF Modoki signal. The out-of-phase nature of the anomalous SST patterns related to the DJF Modoki and the AM SAM also accords with the negative relationship between the DJF Modoki and the AM SAM. We therefore infer that anomalous SSTs in the SEIO help to establish the relationship between the AM SAM and the following DJF SST variability of the tropical central Pacific Ocean.

Fig. 4
figure 4

Regression maps of AM SST anomalies with the a following DJF Modoki index, and b AM SAM_Nan&Li index. c, d Correspond to a and b, but without the preceding DJF Modoki signal. The black and green stippled areas indicate regression coefficients significant at the 90% and 95% confidence levels, respectively. The green boxes indicate the southeastern Indian Ocean

Full size image

Moreover, the SST in the SEIO has also been interpreted as an “ocean bridge” that propagates the influence of the SAM into the NH, such as the East Asian monsoon region (Nan et al. 2009; Dou et al. 2020) and Indian monsoon region (Dou et al. 2017). The positive AM SAM is typically accompanied by the eastern zonal wind around 30–40° S over the southern Indian Ocean (Fig. 5a), which can not only induce the southward Ekman transport of warm water (Fig. 5b), but can also regulate weaker local wind speeds that influence the rate of air–sea heat release. Corresponding to the positive AM SAM, there is significant positive downward sensible heat flux in the extratropical SEIO, and negative sensible and latent heat flux in the extratropical southwestern Indian Ocean (SWIO). The cooperation between the Ekman transport and sensible heat flux variability contributes to the significantly warmer SST anomalies in the SEIO region. The significant positive Ekman transport and negative sensible and latent heat fluxes cancel each other out in the SWIO. Therefore, the significant SST anomalies exist only in the SEIO region. When the SAM is in its negative phase, the situation is reversed. Thus, the above two SAM-related dynamic and thermodynamic processes can regulate SST variability in the SEIO.

Fig. 5
figure 5

Partial regression maps of AM a surface zonal wind, b Ekman heat flux, c sensible heat flux and d latent heat flux based on the AM SAMI after removing the previous DJF Modoki signal in NCEP reanalysis. The pink and green stippled areas indicate regression coefficients significant at the 90% and 95% confidence levels, respectively. The green boxes indicate the region of southeastern Indian Ocean

Full size image

Next, we defined the standardized regional mean SST anomalies over the SEIO as the SEIO SST index (SEIOSI) to quantify SST variability in this region. The SEIOSI is significantly positively correlated with the AM SAMI (r = 0.51, significant at the 99% confidence level), suggesting that the positive AM SAM may usually be associated with the simultaneous anomalous warm SST over the SEIO. In addition, the AM SEIOSI exhibits a negative relationship with the DJF Modoki index, with a correlation coefficient of − 0.38 (significant at the 95% confidence level). The inverse correlation of the AM SEIOSI with AM SAMI and DJF Modoki further indicates that SST in the SEIO may become a potential bridge to connect the relationship between the AM SAM and the DJF Modoki.

3.2.2 Atmospheric response to anomalous SST in the SEIO

Here, we consider how SST anomalies associated with AM SAM in the SEIO can affect the subsequent tropical central Pacific variability. The SST anomalies in the SEIO can change the meridional gradient of SST in the Indian Ocean, which have the potential to regulate the meridional circulation (Marshall and Connolley 2006; Chen et al. 2010; Zheng et al. 2015, 2018). Firstly, we investigated the corresponding response of the atmospheric meridional circulation. When the SEIO SST anomalies are in a positive phase, there is significant anomalous downward motion to the south of 45° S and upward movement around 20° S and 0°, which correspond to the adjustments in the meridional circulation (Fig. 6a). The response of the meridional circulation to the anomalous SST in the SEIO was also basically reproduced in our sensitivity experiments using the CAM5 model. The composite differences in the meridional circulation between the positive and negative experiments are presented in Fig. 6b. Corresponding to the warmer AM SST anomalies forcing in the SEIO, there is a significant anomalous sinking movement around 40° S, and ascending motion around 20° S and above 500 hPa around 0°. This scenario is broadly consistent with the observed anomalous vertical movement response. However, the model overestimates the downward motion near 10° S and also the upward motion south of 30° S, which are insignificant in the observations. In addition, the significant upward motion in the middle and lower troposphere around 0° are underestimated in the model. A possible explanation for this is that the model overestimates (underestimates) the variation of vertical movement in the 10° S and 30° S (0°) in the control run when compared with the observations. Overall, the model simulation approximately captures the response of the observed meridional circulation and the associated anomalous upward motion in the tropical region during AM.

Fig. 6
figure 6

a Partial regression map of the AM vertical velocity and meridional wind averaged between 90–120° E based on the AM SEIOSI after removing the preceding DJF Modoki signal in NCEP reanalysis. b The composite difference of the simulated response of AM vertical velocity and meridional wind averaged between 90–120° E between the positive and negative sensitivity experiment in CAM5. The pink and green stippled areas indicate vertical velocity significant at the 90% and 95% confidence levels, respectively. The black vectors indicate significance at the 90% confidence level

Full size image

The anomalous upward motion in the tropics (90–120° E) indicates increasing tropical convection with negative OLR anomalies (Fig. 7a) over the Maritime Continent. We used the Gill model (1980) to investigate the atmospheric response to this convective heating. The related heating forcing (Fig. 7b) can result in the horizontal Kelvin response and anomalous vertical clockwise circulation with the anomalous easterlies (westerlies) to the east (west) of the heating source center (Fig. 7c) and the upward (downward) branch above the heating source center (in the central Pacific Ocean, Fig. 7d). Therefore, there is an observational anomalous Walker circulation response as shown in Fig. 8. The observations and the model results both indicate that anomalous SST in the SEIO can induce anomalous upward motion in the tropics, which results in the intense convection over the Maritime Continent and in turn regulates the Walker circulation and induces the Kelvin response to the east of 130° E during AM. The associated easterly wind anomalies have the potential to maintain the development of the following anomalous zonal wind and sea temperature in the equatorial central Pacific region via tropical air–sea interaction. These relationships will be investigated in the next section.

Fig. 7
figure 7

a Partial regression map of AM OLR based on the AM SEIOSI after removing the preceding DJF Modoki signal. The pink and green stippled areas indicate regression coefficients significant at the 90% and 95% confidence levels, respectively. b The ideal heating forcing in the Gill model. c The response of zonal wind to the heating forcing in the Gill model. d The response of vertical circulation to the heating forcing in the Gill model

Full size image
Fig. 8
figure 8

Partial regression map of the AM vertical velocity and zonal wind averaged between 5° S–5° N based on the AM SEIOSI after removing the preceding DJF Modoki signal in NCEP reanalysis. The pink and green stippled areas indicate zonal and vertical wind significant at the 90% and 95% confidence levels, respectively. The black vectors indicate significance at the 90% confidence level

Full size image

3.2.3 Influence of zonal wind on sea temperature in the tropical central Pacific Ocean

Zonal wind anomalies can make a vital contribution to the variability in sea temperature anomalies within equatorial regions (Bjerknes 1969). Figure 9 presents the evolution of the anomalous surface wind, sea temperature, and ocean currents associated with the AM SEIO SST. In particular, there is an anomalous easterly wind at the equator between 130° E and 150° E (Fig. 9a) during AM. This anomalous easterly wind (vector in Fig. 9a) to the east of 140° E enhances the easterly wind speed in the climatic trade winds region (shading in Fig. 9a), which can induce significant upward motion that moves cold water into the upper ocean (Fig. 9f). The corresponding anomalous eastward oceanic current can propagate this cold water into the central Pacific Ocean (160–180° E). These upward and eastward motions result in significant cold water at depths of 100–150 m in the central Pacific Ocean during AM. The upward movement of cold water can reach the surface ocean during June–July (JJ) (Fig. 9g), and the anomalous easterly zonal wind develops further to the east and outside the equatorial region (Fig. 9b). In the August–September (AS) and October–November (ON), the surface anomalous easterly winds and cold water develop further (Fig. 9c, d, h, i). Corresponding to the oceanic ascending motion around 140–160° E, there is downward motion to the east of 180° E, which can lead to the downward and westward transport of the cold water. Therefore, the significant area of cold water exists mainly around 160–200° E. The maximum upward motion anomalies in the central Pacific Ocean, which correspond to one standard deviation of SEIOSI, can reach the same magnitude as the climatic vertical velocity (i.e., 10−6 m/s), and this can generate an average increase of 12.8% in Ekman upwelling between AM and December–January (DJ). Furthermore, the maximum cold sea temperature anomaly can reach − 0.48 K in the central Pacific Ocean.

Fig. 9
figure 9

Partial regression maps of the a–e surface wind (vector) and climatic zonal wind (shading) and f–j sea temperature (shading) and oceanic current (vector) averaged between 10° S–10° N based on the AM SEIOSI from AM to DJ after removing the preceding DJF Modoki signal. The pink and green stippled areas indicate regression coefficients significant at the 90% and 95% confidence levels, respectively. The black vectors indicate significance at the 90% confidence level

Full size image

In addition, the speed of the anomalous AM easterly wind decreases the wind speed around 130–140° E (the climatic westerly wind region; shading in Fig. 9a), and this corresponds to the weak warm water above 50 m west of 130° E (Fig. 9f). The climatic surface westerly wind expands eastward with time, and the anomalous easterly wind in the west Pacific Ocean further decreases the wind speed (Fig. 9b–e), resulting in the downward movement of warm water and its eastward propagation into the central Pacific Ocean from JJ to DJ (Fig. 9g–j), which finally warms the cold sea temperature in the central Pacific Ocean during DJ. In brief, the anomalous AM easterly wind associated with the AM SST anomalies to the east of 140° E can enhance the zonal wind speed over the tropical central Pacific Ocean, where it is controlled by the prevailing easterly trade winds. The increased wind speed favors oceanic upwelling of cold water and its eastward propagation into the central Pacific Ocean. This cold water can reach the surface during JJ and trigger variability in the overlying atmosphere, maintain the following easterly zonal wind anomalies, and further induce subsurface upward motion and develop the cold water anomalies. The anomalous easterlies to the east of 140° E reduce the wind speed and result in the downward and eastward transport of warm water. Taken together, these coupled air–sea processes result in the development, and then decay, of the cold water in the central Pacific Ocean between AM and DJ.

4 Summary and discussion

The influence of the SH extratropical atmospheric signal on the following tropical SST variability has attracted increasing research attention, which has focused mainly on the air–sea interaction processes in the Pacific Ocean region. In this study, we explored the possible impact of the preceding AM SAM on the following sea temperature variability in the central Pacific Ocean from the Indian Ocean perspective. The positive AM SAM favors cold sea water in the tropical central Pacific Ocean over the period between AM and DJ. This relationship is robust when the influences of the preceding ENSO signal and the long-term trend are excluded.

The possible physical mechanisms underlying this significant negative relationship involve the following coupled air–sea processes. The positive AM SAM-related surface wind favors the simultaneous warm SST in the SEIO via its influence on the oceanic Ekman transport and the exchange of sensible and latent heat fluxes. The anomalously warmer SST changes the meridional circulation in the Indian Ocean and causes ascending motion anomalies in the tropics. This can result in enhanced convection over the Marine Continent. The associated heating forces the Kelvin response and anomalous clockwise zonal–vertical circulation in the tropical region, resulting in downward motion around 145–165° E and easterly wind anomalies east of 130° E during AM. The anomalous easterlies to the east of 140° E tend to enhance the prevailing easterly trade winds, and induce the oceanic upwelling and eastward transport of cold water into the central Pacific Ocean. This cold water in turn amplifies the development of the easterly wind and further maintains the cold water into the boreal winter. This process is summarized in the schematic diagram in Fig. 10. Therefore, the anomalous SST in the SEIO and the associated air–sea interactions function as an indispensable bridge to promote the extratropical atmospheric signal—SAM—into the deep tropical region and regulate tropical sea temperature variability. Note that the relationship described in this study is significant at the 90% level after removing the previous DJF Modoki signal. This indicates that relationship between the AM SAM and DJF Modoki perhaps feeds back from the prior DJF Modoki, and that this helps to strengthen the overall relationship.

Fig. 10
figure 10

Schematic diagram showing the influence of the AM SAM related SST (red box) in SEIO on the following sea temperature (blue boxes) in the central Pacific Ocean from AM to DJ. The green arrows indicate the vertical circulation. The yellow arrows indicate the surface wind. The blue arrows indicate oceanic currents

Full size image

The lead correlation between the AM SAM and the following Modoki implies that the preceding SAM has the potential to affect subsequent tropical variability, and this influence is mainly observed in the central Pacific Ocean. This relationship may provide an additional source of predictability for the tropical central Pacific Ocean sea temperature from the perspective of the extratropical Indian Ocean. Combined with previous studies that have also emphasized the inferred contribution of the extratropical signal within the Pacific Basin to climatic variability in the deep tropical eastern Pacific, we conclude that assimilating the extratropical signal into coupled models should improve the prediction skill for the tropical Pacific climate. Further research will be needed if the various extratropical influences are to be incorporated into future predictions of the tropical Pacific climate.