1 Introduction

El Niño-Southern Oscillation (ENSO) is the strongest climate variability in the tropics on interannual timescale. It could affect global climate by teleconnections and induce sea surface temperature (SST) anomalies in other ocean basins (Ropelewski and Halpert 1989; Wallace et al. 1998; Klein et al. 1999; Wang et al. 2000; Xie et al. 2002, 2009; Du et al. 2009; Wu and Zhang 2010; Wu et al. 2012). Many previous studies have pointed out that ENSO could modulate the following East Asian summer monsoon (e.g. Huang and Wu 1989; Wang et al. 2000; Lau and Nath 2003, 2006; Chen 2002; Wu et al. 2003; Wu et al. 2014a, b; Chen et al. 2016). And it is widely stated that there is a pronounced low-level anomalous anticyclone over the western North Pacific during the El Niño decaying summer (e.g. Wang et al. 2003; Wu et al. 2003). Typically, the western North Pacific anomalous anticyclone (WNPAC) peaks during the El Niño mature winter and persists through subsequent spring or even early summer (Wang et al. 2000; Wu et al. 2003). The southwesterly wind anomalies on the northwestern flank of the WNPAC may enhance the East Asian Summer monsoon, thereby increasing the Meiyu rainfall via enhancing the low-level moisture transport from the western tropical Pacific into subtropical frontal region (e.g. Zhang et al. 1996, 1999; Chen 2002).

In general, ENSO is locked to boreal winter. Its related central to eastern Pacific SST anomalies develop in boreal summer, peak in boreal winter, and decay quickly in subsequent boreal spring. So how does ENSO affect East Asian climate in subsequent boreal spring and summer? Wang et al. (2000) showed that the anomalous Philippine Sea anticyclone results from an atmospheric Rossby-wave response to suppressed convective heating, which is induced by the local ocean surface cooling and the subsidence forced remotely by the central Pacific warming. The anomalous anticyclone could persist from El Niño mature winter to subsequent early summer through a local positive air-sea coupling in the presence of mean northeasterly trade winds. Another plausible mechanism is called the Indian Ocean capacitor effect. El Niño induces tropical Indian Ocean (TIO) warming by atmospheric bridge (Klein et al. 1999) and oceanic processes (Xie et al. 2002; Liu and Alexander 2007; Du et al. 2009; Li et al. 2015a), like a battery charging a capacitor. Although El Niño decays and vanishes in summer, the TIO warming can persist through summer and unleash its influence, like a discharging capacitor (Yang et al. 2007). Responding to TIO warming, tropospheric temperature displays a Matsuno–Gill pattern (Matsuno 1966; Gill 1980), with a Kelvin wave-like wedge penetrating into the equatorial western Pacific (Xie et al. 2009). The surface friction drives northeasterly winds into the equatorial low pressure in the baroclinic Kelvin wave, inducing surface convergence on the equator and surface divergence off the equator, thereby triggering suppressed deep convection and anomalous anticyclone over the western North Pacific (Xie et al. 2009).

Recent studies suggested more complex processes working together to modulate the WNPAC (e.g. Lee et al. 2006; Kim et al. 2009; Wu et al. 2010, 2011; Kosaka et al. 2013; Xie et al. 2016). Wu et al. (2010) suggested that the WNPAC is maintained by the combined effects of the local cooling in the western North Pacific and the remote forcing from the Indian Ocean warming (IOW) during the strong El Niño decaying summer. While with the onset of the western North Pacific summer monsoon and the retreat of easterly trade winds, the positive thermodynamic feedback between the WNPAC and the underlying SST turns to be a negative feedback (Chou et al. 2009), reducing the SST anomalies in the WNP and its local forcing effect. However, the IOW plays a crucial role in late summer via Kelvin wave induced anticyclonic shear and boundary layer divergence (Wu et al. 2009, 2010). Through a series of coupled model experiments, Kosaka et al. (2013) demonstrated that the Pacific–Japan (PJ) teleconnection pattern is the atmospheric manifestation of an air-sea coupled mode spanning the Indian Ocean and western north Pacific (PJIO mode). The PJ pattern forces the Indian Ocean by a westward propagating atmospheric Rossby wave, and the Indian Ocean SST reinforces the PJ pattern through a tropospheric Kelvin wave in turn. Xie et al. (2016) further proposed a coupled ocean–atmosphere mode over the Indo-western Pacific that combines the local air-sea interaction (Wang et al. 2000) and the Indian Ocean capacitor effect (Xie et al. 2009), and called it Indo-western Pacific ocean capacitor (IPOC) effect. In spring, the anomalous anticyclone and Northwest Pacific cooling are coupled via wind-evaporation-SST feedback in the presence of northeasterly trade winds, and the Northwest Pacific cooling persists to trigger a summer feedback that arises from the interaction of the anomalous anticyclone and the North Indian Ocean warming in the Asian monsoon season (Xie et al. 2016).

Most of the aforementioned studies paid little attention on the impact of negative phase of ENSO (i.e., La Niña). It is expected that the influence of La Niña may be similar but with opposite sign to that of El Niño. One may take it for granted that the SST anomalies in the central to eastern Pacific during the La Niña decaying summer may be very weak and its remote forcing effect is negligible. Figure 1 shows the evolution of Niño3.4 SST index anomalies from previous winter to subsequent fall for ten El Niño events (top panel) and ten La Niña events (bottom panel) during the period 1979–2010. It is clear that the evolution of central to eastern Pacific SST anomalies after the mature phase of strong El Niño events (e.g. 1982/1983, 1991/1992, 1994/1995, 1997/1998) tend to decay to normal. For all El Niño events except for 1986/1987, there are no pronounced positive SST anomalies in the central to eastern Pacific in summer (JJA). Different features are observed when considering the evolution of central to eastern Pacific SST anomalies after the mature phase of La Niña events. In general, the negative SST anomalies in the central to eastern Pacific decay from La Niña mature winter to subsequent summer. Some of the La Niña events transit into El Niño events (e.g. 2008/2009). However, most of the La Niña events decay slowly and the cooling in the central to eastern Pacific is still obvious (e.g. 1984/1985, 1988/1989, 1999/2000) during the decaying summer. It is conceivable that the central to eastern Pacific cooling during strong La Niña decaying years may exert remote influence on the East Asian Climate. Some studies have noticed the importance of central to eastern Pacific remote forcing on the Asian Climate (e.g. Fan et al. 2013; Hu et al. 2014; He and Wu 2014; Chen et al. 2016). For example, Hu et al. (2014) suggested that the negative eastern Pacific SST anomalies could induce anomalous Walker circulation, which contribute to anomalous convection over the South China Sea and the surrounding regions during spring-to-summer transition. Fan et al. (2013) argued that the summer cooling in the central tropical Pacific (i.e. Niño4 region) may also play a role in the maintenance or development of the western north Pacific anomalous anticyclone. Chen et al. (2016) have demonstrated that there is a low-level anomalous anticyclone over the western North Pacific during El Niño to La Niña transition years, and the central to eastern Pacific cooling plays an important role in maintaining the anomalous anticyclone in summer. As the cooling in the central to eastern Pacific develops from spring to summer, the Pacific cooling forcing effect, which could stimulate a Rossby wave response to its northwest and amplify the WNPAC, appears to strengthen (Chen et al. 2016). Some previous studies (e.g. McPhaden and Zhang 2009; Ohba and Ueda 2009; Okumura and Deser 2010) noted that La Niña is not a simple mirror image of El Niño. After the mature phase, El Niño tends to decay rapidly by next summer, but many La Niña events persist through the following year and often re-intensify in the subsequent winter (Okumura and Deser 2010). The asymmetry between strong El Niño and strong La Niña decaying summer motivate us to re-explore the roles of tropical SST anomalies in modulating the anomalous circulation over the Indo-western Pacific during the strong La Niña decaying years. Does the central to eastern Pacific cooling can remotely modulate atmospheric circulation anomalies over the Indo-western Pacific during strong La Niña decaying years? And what is the relative contribution of tropical SST anomalies in different regions to the circulation anomalies over the Indo-western Pacific?

Fig. 1
figure 1

The evolution of Niño3.4 SST index from previous winter to subsequent fall for ten El Niño events (top panel) and ten La Niña events (bottom panel) during the period 1979–2010. The Niño3.4 SST index is calculated as the SST anomalies average in Niño3.4 region (i.e. 5°S-5°N, 170°W–120°W)

Full size image

The rest of the manuscript is organized as follows. Section 2 describes the datasets and methods used in this study. Section 3 presents the observed anomalies (including SST, precipitation and atmospheric circulation) from strong La Niña decaying spring to summer. Section 4 shows the results of numerical modeling and analyzes the contribution of tropical SST anomalies in modulating the circulation over Indo-western Pacific. Discussions and concluding remarks are given in Sect. 5.

2 Data, methodology and model description

2.1 Data and method

In this study, we use monthly-mean data of the European Centre for Medium-Range Weather Forecasts (ECMWF) interim reanalysis (ERA-Interim, Dee et al. 2011). We also use monthly precipitation from the Global Precipitation Climatology Project (GPCP, Adler et al. 2003). The ERA-Interim reanalysis (GPCP data) is provided with 1.5° × 1.5° (2.5° × 2.5°) resolution. In addition, monthly SST data on a 1° × 1° grid from the Met Office Hadley Centre (HadISST, Rayner et al. 2003) is also utilized. Those three datasets are available over a 32-year period from 1979 to 2010. The seasonal mean anomalies analyzed in this study are departures from their climatological mean for the period 1979–2010. The statistical significance of composite analyses is assessed by Student’s t test.

The observational analysis in this study is based on five La Niña cases (i.e. 1983/1984, 1984/1985, 1988/1989, 1998/1999, 1999/2000). These selected cases are corresponding to a strong La Niña state in preceding winter. Thus the composite years are denoted as strong La Niña decaying years.

2.2 Model description

The atmospheric generation circulation model (AGCM) used in this study is Community Atmosphere Model Version 4 (CAM4), which was developed with significant community collaboration at the National Center for Atmospheric Research (NCAR). It is the atmosphere component of Community Climate System Model Version 4 (CCSM4) that was released in 2010 (Gent et al. 2011). Due to the limitation of computational resources, the CAM4 model is run at a horizontal resolution roughly equivalent to 1.9 latitude × 2.5 longitude, with 26 vertical levels in a hybrid sigma-pressure coordinate system that extends from the surface to approximately 3.5-hPa. A general overview of CAM4 mean climate simulation is provided by Neale et al. (2013). More comprehensive and detailed description about CAM4 can be found in its technical report (Neale et al. 2010).

To identify the roles of tropical SST anomalies in modulating the circulation over western Pacific, we conduct one control run (referred to as CTL run) and four sensitivity runs. Each run is a 35-year simulation and only the last 30-year results are analyzed in this study. The CTL run is forced by repeating climatological mean annual cycles of SST, while the sensitivity runs are designed with the SST anomalies shown in Fig. 2 added to the climatological mean annual cycle of SST in tropical central to eastern Pacific only (referred to as TCEP run), tropical Indian Ocean only (referred to as IO run), western North Pacific only (referred to as WNP run), or in all three regions (referred to as ALL run). A Student’s t test is used to examine the statistical significance of the difference between sensitivity run and control run.

Fig. 2
figure 2

SST anomalies used in the CAM4 sensitivity experiments. The stippled areas denote that the anomalies are significant at 90 % confidence level according to student’s t test

Full size image

3 Observational analysis

In this section, we present the observed SST, precipitation and atmospheric circulation anomalies from strong La Niña decaying spring to summer. From spring to summer, there are cold SST anomalies in the central to eastern Pacific (Fig. 3). The maximum negative SST anomalies appear in the central Pacific in spring. With the decay of SST anomalies in tropical central to eastern Pacific, the maximum negative SST anomalies split into two parts. One is still near the dateline, and the other is in eastern Pacific (Fig. 3). To the northwest of tropical Pacific cold SST anomalies, there is a southwest to northeast band of positive SST anomalies. From March–April-May (MAM) to April–May-June (AMJ), the positive SST anomalies extend to southeast of Philippine Islands, while the warm SST anomalies near the Philippine Islands nearly vanish in JJA (Fig. 3d). Besides, it is obvious that there is a robust anomalous signal over tropical Indian Ocean, exhibiting a basin-wide cooling. In spring, maximum SST anomalies appear in the South Indian Ocean (Fig. 3a, b). And then the cold SST anomalies in the North Indian Ocean develop from spring to summer (Fig. 3c, d). This finding is consistent with Du et al. (2009) who found that the North Indian Ocean displays a peculiar double-peak warming (cooling) with the second peak larger in magnitude and persisting through the El Niño (La Niña) subsequent summer.

Fig. 3
figure 3

Composite SST (shading, unit:°C) and precipitation anomalies (contour, unit:mm/day) from subsequent spring to summer for five strong La Niña events (i.e. 1983/1984, 1984/1985, 1988/1989, 1998/1999 and 1999/2000). The stippled areas denote that the SST anomalies are significant at 90 % confidence level according to student’s t test. The contour levels for precipitation anomalies are −1.2 (red), −0.6 (red), 0.6 (green), and 1.2 (green), respectively

Full size image

Corresponding to the central to eastern Pacific cooling, below-normal precipitation appears over the equatorial central to eastern Pacific. Meanwhile, the eastern part of the Maritime Continent suffers more rainfall (Fig. 3). From spring to summer, the negative (positive) precipitation anomalies persist over the central Pacific (the eastern part of the Maritime Continent). These results indicate that there is an anomalous Walker circulation with ascending branch over the eastern part of the Maritime Continent and descending branch over central to eastern Pacific (Fig. 4). At the same time, the convection over southwest Indian Ocean is suppressed. In JJA, most of the North Indian Ocean exhibit significant negative rainfall anomalies (Fig. 3d). It is surprising that the convection to east of Philippine Islands seems to enhance from spring to summer, and a positive Pacific–Japan pattern is established in summer with enhanced convection over subtropical western Pacific and suppressed convection over Japan (Fig. 3d).

Fig. 4
figure 4

Composite 500-hPa omega (shading, unit:Pa/s, multiplied by 102), 200-hPa velocity potential (contour, unit:s−1, scaled by 106) and divergent wind anomalies (vector, unit:m/s) from subsequent spring to summer for five strong La Niña events (i.e. 1983/1984, 1984/1985, 1988/1989, 1998/1999 and 1999/2000). The stippled areas denote the omega anomalies are significant at 90 % confidence level according to student’s t test. The contour levels (scaled by 106) for velocity potential anomalies are −2.5, −2.0, −1.5, −1.0, −0.5, 0.5, 1.0, 1.5, 2.0 and 2.5, respectively. The black vectors denote that the divergent wind anomalies are significant at 90 % confidence level according to student’s t test

Full size image

There are low-level easterly wind anomalies over equatorial western to central Pacific and westerly wind anomalies over the Indian Ocean from MAM to JJA (Fig. 5). The low-level easterly and westerly wind anomalies meet over the Maritime Continent, which correspond to anomalous high-level divergence and low-level convergence, resulting in enhanced convection over the Maritime Continent (Figs. 4, 5). Over subtropical western Pacific, pronounced cyclonic circulation anomalies persist from spring to summer. In the meanwhile, anomalous anticyclonic wind anomalies start to form over Japan (Fig. 5c, d). This significant wind anomaly pattern over East Asia in JJA corresponds to aforementioned positive Pacific–Japan Pattern with enhanced convection over subtropical western Pacific and suppressed convection over Japan. The cyclonic wind anomaly over the Indo-western Pacific seems to cover a larger subtropical area in spring, when compared to summer (Fig. 5a, d). The dipole pattern of SST anomalies in the tropical Pacific induces an anomalous Walker circulation, with a downdraft over the central to eastern Pacific and an updraft over the western Pacific. The updraft over the western Pacific and the local warm SST anomalies flavor the development of anomalous low-level cyclone over western North Pacific (WNPC). The northeasterly wind anomalies on the western flank of WNPC decrease the moisture transport to South China. Besides, anomalous local Hadley circulation develops between western Pacific and East Asia, with rising motion in the tropics and sinking motion in the subtropics. These processes provide a favorable environment for the decreased rainfall over Japan.

Fig. 5
figure 5

Composite precipitation (shading, unit:mm/day) and low-level wind anomalies (vector, unit:m/s) from subsequent spring to summer for five strong La Niña events (i.e. 1983/1984, 1984/1985, 1988/1989, 1998/1999 and 1999/2000). The stippled areas denote that the precipitation anomalies are significant at 90 % confidence level according to student’s t test. The black vectors denote the wind anomalies are significant at 90 % confidence level according to student’s t test

Full size image

In spring, corresponding to south Indian Ocean cooling (Fig. 3a, b), there are high-level convergence and low-level divergence in situ (Fig. 4a, b). Anomalous vertical downdraft suppresses the deep convection (Fig. 4a, b) and results in less rainfall (Fig. 5a, b). Besides, there are southeasterly wind anomalies over tropical southwestern Indian Ocean. The southeasterly wind anomalies turn to westerly wind anomalies near the equator. The southwest Indian Ocean cooling seems to maintain an antisymmetric circulation pattern over tropical western Indian Ocean with southeasterly south of the equator and southwesterly wind anomalies north of the equator, a feature of that of spring asymmetric mode over the tropical Indian Ocean (Wu et al. 2008). As the mean winds turn into southwesterly in May over the North Indian Ocean, wind speed and surface evaporation are enhanced. Negative SST anomalies appear in the North Indian Ocean, likely a response of ocean to the atmosphere via increased westerly wind speed and enhanced surface heat flux and oceanic mixing (Wu et al. 2008; Du et al. 2009, 2013; He et al. 2015).

Actually, the north Indian Ocean cooling and subtropical western north Pacific warming may be coupled with WNPC. In the south flank of WNPC, the westerly wind anomaly may enhance the winds over the North Indian Ocean and the South China Sea during summer, thereby the ocean loses more heat to the atmosphere, leading to cooling of the in situ SST (second cooling of the North Indian Ocean SST). In addition, the westerly extends to the east of the Philippine Islands where the easterly trade winds prevail in spring to early summer, thus reducing the surface wind speed and upward latent heat flux, maintaining the subtropical western north Pacific warming. The less rainfall over tropical Indian Ocean and more rainfall over subtropical western North Pacific indicate that the Indian Ocean SST cooling or WNP local warming is not just a slave to atmospheric circulation change, they may have feedback to the atmosphere.

4 Numerical experiments

It is difficult to isolate the effects of tropical SST anomalies over different regions in modulating the circulation over Indo-western Pacific by observational analyses. To reveal the relative contribution of central to eastern Pacific, tropical Indian Ocean and western north Pacific SST anomalies, five idealized CAM4 numerical experiments (i.e. CTL run, TCEP run, IO run, WNP run, and ALL run) are conducted respectively, which may help to isolate the impacts of SST anomalies in the Indian and Pacific Oceans. The ensemble mean difference between TCEP run (IO run) and CTL run indicates the remote forcing effect of central to eastern Pacific (Indian Ocean) cooling. The local forcing effect of western north Pacific SST anomalies may be evaluated through analyzing the difference between WNP run and CTL run. And the ensemble mean difference between ALL run and CTL run provides an evaluation on the combined effect of tropical SST anomalies.

Figure 6 presents the simulated precipitation and 850-hPa wind differences between ALL run and CTL run. The spatial correlation coefficient between ALL run simulated precipitation differences and observed precipitation anomalies is high (about 0.68) in spring and relatively low (about 0.32) in summer. The observed low-level zonal wind anomaly pattern seems to be realistically reproduced by ALL run. The spatial correlation coefficient between ALL run simulated zonal wind differences and observations reaches about 0.79 in MAM and is still high in JJA (about 0.64). Some key characteristics, such as easterly (westerly) wind response over tropical central to eastern Pacific (Indian Ocean) and cyclonic circulation response over the western North Pacific resemble the observations. The reasonable low-level zonal wind response of ALL run gives us confidence to further examine the individual contribution of SST anomalies in different regions of tropical Pacific and Indian Ocean in modulating the circulation over the Indo-western North Pacific.

Fig. 6
figure 6

The simulated precipitation (shading, unit:mm/day) and low-level wind (vector, unit:m/s) difference between ALL run and CTL run. Only the precipitation anomalies that are significant at 90 % confidence level are shaded. And the black vectors denote that the low-level wind anomalies are significant at 90 % confidence level according to student’s t test

Full size image

The simulated precipitation and 850-hPa wind differences between TCEP run and CTL run are shown in Fig. 7. In response to negative SST anomalies in the central to eastern Pacific, the convection over most of the equatorial Pacific is suppressed from spring to summer. This suppression appears to be strong in spring and relatively weak in summer. As the cold tongue over southeast Pacific develops and extends westward from spring to summer, it is too cold to trigger convection over southeast Pacific in summer. Thus, although there are still obvious negative SST anomalies in tropical central to eastern Pacific in JJA, there is no significant precipitation signal over tropical southeast Pacific. In central Pacific, three are negative rainfall differences persisting from spring to summer. To the northeast of the negative precipitation differences, there are anticyclonic wind differences, which are atmospheric Rossby wave response to the central Pacific cooling. The anticyclonic wind response over the southeast of the Philippine island seems to slightly decay from spring to summer, which are in accordance with the slight decay of negative SST anomalies in the central Pacific. Besides, the central to eastern Pacific cooling contributes to the enhanced deep convection over the western North Pacific. There is anomalous downward vertical motion over equatorial central to eastern Pacific (Fig. 8), which corresponds to in situ anomalous high-level convergence and low-level divergence. At the same time, anomalous upward vertical motion appears near the Philippines, in accompany with high-level divergence and low-level convergence. This anomalous Walker circulation links the western North Pacific climate to the equatorial central and eastern Pacific SST anomalies. The convective heating over the western North Pacific in turn forces atmospheric Rossby wave to its west, which may amplify the anomalous cyclone over the Indo-western Pacific (Fig. 7). The TCEP run confirms that the central to eastern Pacific SST anomalies from strong La Niña decaying spring to summer could play a role in modulating the circulation over the Indo-western Pacific.

Fig. 7
figure 7

Same as Fig. 6, but for the difference between TCEP run and CTL run

Full size image
Fig. 8
figure 8

The simulated 500-hPa omega (shading), 200-hPa velocity potential (contour, unit:s−1, scaled by 106) and divergent wind (vector, unit:m/s) difference between TCEP run and CTL run. Only the omega and divergent wind anomalies that are significant at 90 % confidence level are plotted. The contour levels (scaled by 106) for velocity potential anomalies are −2.5, −2.0, −1.5, −1.0, −0.5, 0.5, 1.0, 1.5, 2.0 and 2.5, respectively

Full size image

The remote forcing effect of the Indian Ocean cooling is inferred by analyzing the ensemble mean difference between IO run and CTL run. Figure 9 shows the simulated precipitation and low-level wind differences between IO run and CTL run. The convection over southwest Indian Ocean is suppressed in spring in response to tropical southwest Indian Ocean cooling (Fig. 9a, b). Besides, the southwest Indian Ocean cooling seems to induce anomalous downward flow and enhance the Somali cross-equator flow from spring to summer (Figs. 9a, b, 10a, b). From air-sea coupling perspective, the westerly wind anomalies over tropical North Indian Ocean may enhance the wind speed in situ and induce excessive evaporation and entrainment cooling, cooling the underlying SST, which can explain the observed North Indian Ocean cooling.

Fig. 9
figure 9

Same as Fig. 6, but for the difference between IO run and CTL run

Full size image
Fig. 10
figure 10

Same as Fig. 8, but for the difference between IO run and CTL run

Full size image

In response to Indian Ocean basin wide cooling, there are anomalous low-level westerly wind over tropical eastern Indian Ocean and western Pacific. The maximum zonal wind response appear near the equator and cyclonic wind shear appears over the Bay of Bengal to subtropical western Pacific (Fig. 9b–d). This finding agrees with former studies that Indian Ocean SST anomalies could modulate East Asian climate via eastward propagating atmospheric Kelvin wave (e.g., Lee et al. 2006; Yang et al. 2007; Wu et al. 2009; Xie et al. 2009; Du et al. 2011). In spring, there is no obvious rainfall response near the Philippine Islands. The convection near the Philippines appears to be enhanced from spring to summer, which corresponds to the intensification of anomalous cyclonic wind shear over western North Pacific. The numerical results of IO run indicate that the Indian Ocean Cooling effect may be weak in spring. With the increase of the Indian negative SST anomalies from spring to summer, the Indian Ocean remote forcing effect seems to be strong in summer.

Figure 11 shows the simulated precipitation and low-level wind differences between WNP run and CTL run. It is no surprise that the convection over the Philippine Sea is enhanced corresponding to in situ positive SST anomalies, consistent with Wu et al. (2014a, b). The positive precipitation response gradually decrease and nearly vanish in summer. The positive SST anomaly in WNP also plays a role for the maintenance of anomalous cyclone in spring via atmospheric Rossby wave response. In MAM, There are significant low level cyclonic wind response over the South China Sea and the Philippine Sea. However, with the onset of the East Asian summer monsoon, the positive thermodynamic feedback between atmospheric Rossby wave response and the underlying SST anomalies (Wang et al. 2000) turns to be a negative feedback (Chou et al. 2009). Thus, the positive SST anomaly near Philippine Sea gradually decays in observation, so does its local forcing effect.

Fig. 11
figure 11

Same as Fig. 6, but for the difference between WNP run and CTL run

Full size image

The above analyses show that tropical SST anomalies in central to eastern Pacific, western North Pacific, and Indian Ocean may all contribute to cyclonic wind anomalies over the Indo-western Pacific by different physical processes. For a better comparison, a wind shear index (WSI) is computed and analyzed in this study. The WSI is defined as the difference of zonal wind between a southern region (100°–130°E, 5°–15°N) and northern region (110°–140°E, 25°–35°N). The WSI is identical to the western North Pacific monsoon index (WNPMI) defined by Wang et al. (2001) except that the norhtern region shifts northward compared to Wang’ s definition. A positive (negative) WSI anomaly is associated with a cyclonic (anticyclonic) wind anomaly over the Indo-western Pacific.

We have computed the WSI from 1979 to 2010 using the ERA-Interim reanalysis dataset. The observational evolution of WSI anomalies during strong La Niña decaying years is shown in Fig. 12. The WSI anomalies from MAM to JJA for five selected years are positive, which suggest that there is an anomalous cyclone over the Indo-western Pacific from spring to summer. From MAM to JJA, the composite mean WSI anomalies remain to be large positive.

Fig. 12
figure 12

The evolution of low-level wind shear index for (a) observational results, and (b) experimental results. The wind shear index is defined as the difference of zonal wind anomalies between a southern region (100°-130°E, 5°–15°N) and northern region (110°–140°, 25°–35°N)

Full size image

Since the WSI is defined as wind shear over Indo-western Pacific, the evolution of this index may indicate the maintenance of the anomalous cyclone in situ. We also compute the WSI from MAM to JJA for five idealized experiments. The ensemble mean differences between respective sensitivity run and CTL run are presented in Fig. 12b. For the WNP run, the simulated WSI ensemble mean differences evolve from a relatively strong positive value in spring to a slightly positive value in summer. For the IOW run, the simulated WSI ensemble mean differences appear to be weak in spring and strong in summer. These results suggest that as the positive SST anomalies over WNP decay, the local forcing effect decays as well. However, the Indian Ocean remote forcing effect seems to play an important role in modulating the East Asia climate during boreal summer via cold Kelvin wave induced cyclonic shear and boundary layer convergence.

For TCEP run, the simulated WSI differences in MAM and AMJ are stronger than those in MJJ and JJA. Besides, the WSI differences in summer are still strong. This indicate that the equatorial central to eastern Pacific cooling could modulate the anomalous cyclone over Indo-western Pacific and its remote forcing effect can persist into summer. As we know, during El Niño decaying summer, it is hardly to find obvious positive SST anomalies in equatorial central to eastern Pacific. While during the La Niña decaying phase, there are some cases in which the negative SST anomalies in equatorial central to eastern Pacific can maintain and persist into summer. We hypothesize that such cooling could also contribute to anomalous cyclonic wind anomaly over the Indo-western Pacific. And the TCEP run confirms our speculation and indicates that the remote forcing effect of equatorial central to eastern Pacific cooling also play an important role in modulating the East Asian climate from spring to summer.

From spring to early summer, more realistic WSI are simulated by ALL run. This result further indicates that the anomalous Indo-western Pacific cyclone is maintained from spring to summer by a combined effect of tropical Indian Ocean, WNP and central to eastern Pacific SST anomalies.

5 Discussions and concluding remarks

In this study, we select the cases corresponding to strong La Niña events in the preceding winter and investigate the role of tropical SST anomalies in modulating the East Asian climate from subsequent spring to summer. For the selected cases, although the La Niña related SST anomalies over equatorial central to eastern Pacific gradually decay from preceding winter to subsequent summer, significant negative SST anomalies still appear in equatorial central to eastern Pacific in summer. The cooling in tropical central to eastern Pacific induces downward vertical motion and suppresses the in situ convection, thus strengthens the Walker circulation. Maritime Continent and the Philippines suffer more rainfall from spring to summer. Local convection near the Philippines seems to be enhanced from spring to summer. Below-normal precipitation persists from spring to summer in tropical southwest Indian Ocean, which is in correspondence with the cold SST anomalies in situ. The southwest Indian Ocean cooling in spring induces anomalous cross-equatorial flows. Asymmetric zonal wind anomalies around the equator are found in western Indian Ocean from spring to early summer. Negative SST anomalies from late spring to summer in the North Indian Ocean are likely a response to atmosphere change as westerly wind anomalies increase surface wind speed and thus enhance surface evaporation and oceanic mixing in Asian monsoon season. From strong La Niña decaying spring to summer, there are cyclonic wind anomalies over subtropical WNP. In view of the limited number of cases for the composite, we extend the analysis period to 1948–2010 based on the HadISST (Rayner et al. 2003) and ECMWF ERA-20C reanalysis (Poli et al. 2013) to obtain more strong La Nina events. The composite SST and wind anomaly patterns are quite similar (figures not shown).

We explore the relative importance of tropical SST anomalies in maintaining the anomalous cyclone over Indo-western Pacific from strong La Niña decaying spring to summer by numerical modeling. A series of sensitivity experiments are conducted in which SST anomalies in the tropical central to equatorial Pacific, Indian Ocean and western north Pacific are specified individually or in combination. The modeling results show that the tropical SST anomalies work together to sustain the anomalous cyclone over Indo-western Pacific from spring to summer. The WNP SST anomalies may play a role in modulating the Indo-western Pacific climate via a Rossby wave-type response. The contribution of local positive SST anomalies in western North Pacific sustains from spring to early summer, but its local forcing effect weakens from spring to summer as the in situ SST anomalies decay and related rainfall anomalies vanish. Anomalous Walker circulation with ascending branch over the Maritime Continent to the Philippines and descending branch over equatorial Pacific is induced by central to eastern Pacific cooling. As the cooling in equatorial central to eastern Pacific slightly decays from spring to summer, the impact of equatorial Pacific cooling decreases in summer.

Wu et al. (2010) selected five strong El Niño events (i.e. 1982/1983, 1991/1992, 1994/1995, 1997/1998, 2004/2005) and studied East Asian climate anomalies in the El Niño decaying summer. Typically, El Niño events peak in winter, decay quickly in early spring, and vanish in summer. Thus, their composite result shows that SST anomalies in the equatorial central to eastern Pacific are weak in summer. In this study, five strong La Niña events are selected. In these cases, the SST anomalies in the equatorial central to eastern Pacific decay slowly, and significant negative SST anomalies are present in equatorial Pacific in summer, which is different from strong El Niño scenario. It is conceivable that the cold SST anomalies may exert remote forcing effect on the East Asian climate. Besides, this study supports former findings (e.g. Lee et al. 2006; Yang et al. 2007; Xie et al. 2009; Wu et al. 2010) that the impact of the Indian Ocean SST anomaly increases in late summer. The enhanced influence of tropical Indian Ocean SST anomalies in late summer may be due to dependence of atmospheric response on climatological monsoon trough formation (Wu et al. 2010) and the increase of cold SST anomalies in situ.

As revealed by previous studies, El Niño and La Niña exhibit significant differences in their spatial structure and seasonal evolution. El Niño tends to decay rapidly by next summer after its mature phase, while La Niña tends to persist through the following year and often re-intensify in subsequent winter (Okumura and Deser 2010). Compared to El Niño composite, the tropospheric temperature structure during La Niña decaying summer is not just that with the sign reversed (Fig. 13). The strong tropospheric temperature signal over the center Pacific indicates that the central Pacific cooling may play an essential role in modulating WNPC during strong La Niña decaying summer. In the equatorial western Pacific, the low level westerly wind anomaly induced by the atmospheric Kelvin wave response to tropical cooling over the Indian Ocean may compete with the low level easterly wind anomaly forced by precipitation anomalies over the central Pacific during La Niña. The associated deep convection anomalies are displaced westward during La Niña compared to El Niño because of nonlinear dependence of atmospheric deep convection on absolute SST. This enables the low level winds in the western equatorial Pacific to be less affected by remote forcing from the Indian Ocean. Thus, there are anticyclonic wind anomalies and cyclonic wind anomalies in western Pacific (0°N–30°N) during La Niña decaying summer, whereas only anticyclonic wind anomalies are observed during the El Niño decaying summer.

Fig. 13
figure 13

Composite map of tropospheric temperature (850–250 hPa, shading, unit:°C) and surface wind (vector, unit:m/s) anomalies based on ERA-Interim reanalysis for (a) El Niño and (b) La Niña. The black vectors denote the wind anomalies are significant at 90 % confidence level according to student’s t test. Only the tropospheric temperature anomalies that are significant at 90 % confidence level are plotted. The cases used in El Niño composites are 1982/1983, 1991/1992, 1997/1998, 2002/2003, 2009/2010

Full size image

The ALL run can well reproduce the cyclonic wind anomalies over the Indo-western Pacific from spring to early summer. The location of anomalous cyclone in summertime is slightly eastward compared to TCEP run or observations. Thus, although there is obvious anomalous cyclone near the Philippines, the simulated WSI in JJA is not strong. This appears to be a common problem for many AGCMs in simulating the Asian monsoon trough and subtropical frontal rain band (e.g., Lau et al. 1996; Kang et al. 2002; Lin 2007; Li and Xie 2014) and may be partly due to the neglect of air-sea coupling processes (e.g. Lee et al. 2006; Wu et al. 2006; Wu and Kirtman 2007; Zhou et al. 2009; Song and Zhou 2014a, b; Li et al. 2015b).

This study defines a wind shear index (WSI). The positive WSI anomalies indicate the existence of anomalous cyclone over the Indo-western Pacific. However, the WSI anomalies seem to have strong fluctuation from MAM to JJA. One possible explanation is that the relative importance of different factors for the WNPC may vary with time. An alternative explanation is the effect of regional air-sea interaction. Wu and Hu (2015) and Hu and Wu (2016) identified a clear signal of local air-sea interaction over the South China Sea during the spring to summer transition. The negative feedback of the atmospheric changes may induce local SST anomalies that in turn contribute negatively to atmospheric circulation anomalies. Besides, the WSI is defined as the difference of low-level wind anomaly between two fixed regions. Slight movement of WNPC during spring to summer may change the values of WSI. In this study, we mainly focus the roles of tropical SST anomalies in modulating the western north Pacific anomalous cyclone during strong La Niña events. We infer that the roles of tropical SST anomalies during weak or moderate La Niña decaying years may be different from those in strong La Niña decaying years due to the differences in the evolution of SST anomalies in the tropical Indian Ocean and central to eastern Pacific. Besides, there are no easterly wind anomalies over the tropical western Pacific and the negative SST anomalies in the central Pacific and associated suppressed convection in situ vanish in weak or moderate La Niña decaying summer (figures not shown).

One interesting question is whether there is a decadal change in the role of tropical SST anomalies in modulating the western North Pacific anomalous cyclone during strong La Niña decaying years, for example, following the phase change of the interdecadal Pacific Oscillation (IPO). The La Niña cases in this study all occurred before 2000, corresponding to the positive phase of IPO. The above issue will be addressed in our future work.