1 Introduction

Summer drought has important socioeconomic effects in northeast Asia (NEA). It is imperative, therefore, to understand the factors and underlying physical mechanisms for summer drought in this region. Previous studies have shown that summer drought over NEA is closely related to a weak East Asian summer monsoon (EASM) and can be caused by summer sea surface temperature (SST) anomalies in the tropical Indo-Pacific region (e.g., Huang and Wu 1989; Wang 2001, 2002; Saji and Yamagata 2003; Wang and Fan 2005; Xie et al. 2009; Zhu et al. 2011; Kosaka et al. 2012; Wu et al. 2012b; Xu and Fan 2012; Li et al. 2014, 2015). Summer SST anomalies in the tropical Indo-Pacific region can first modulate the Indian summer monsoon (ISM) and western North Pacific (WNP) summer monsoon (WNPSM), and then influence the EASM via the excitation of teleconnection patterns (Huang and Li 1987; Nitta 1987; Wang et al. 2001; Lu et al. 2002; Wu 2002; Lu 2004; Ding and Wang 2005; Sun et al. 2010).

The connection between the summer rainfall anomalies over India and NEA has been observed in many previous studies (e.g., Guo 1992; Kripalani and Kulkarni 2001; Zhang 2001; Lu et al. 2002; Wu 2002; Sun et al. 2008, 2010; Wei et al. 2014). The heating released by the ISM rainfall anomalies can influence the EASM via exciting the circumglobal teleconnection (CGT) pattern (Lu et al. 2002; Enomoto et al. 2003; Ding and Wang 2005; Ding et al. 2011). Wu (2002) revealed a dominant pattern, called the mid-latitude Asian summer (MAS) pattern, for the interannual variation of upper-level wind over mid-latitude Asia in summer. The MAS pattern includes a zonally uniform variation and a zonal wave-type variation. The zonal wave component of MAS (Rossby wave) can be triggered by the mid-latitude westerly perturbation over West Asia induced by the anomalous rainfall in India (Wu 2002). The positive MAS pattern exhibits anomalous ridges over the northwest of India and NEA. The MAS pattern plays an important role in connecting the ISM and EASM (Wu 2002; Wei et al. 2014). Developing ENSO events can directly modulate the ISM rainfall via anomalous Walker circulation (e.g., Webster and Yang 1992; Ju and Slingo 1995; Kumar et al. 1999; Wu et al. 2012a). Furthermore, the anomalous Walker circulation leads to anomalous rainfall over the western Pacific. The corresponding condensational heating anomalies over this region excite the Gill pattern modulating the ISM rainfall (Lau and Nath 2000). Summer SST anomalies in the tropical Indian Ocean are closely related to the ISM rainfall anomalies (e.g., Ashok et al. 2001; Li et al. 2001; Guan and Yamagata 2003; Saji and Yamagata 2003; Vimal et al. 2012; Wu et al. 2012b; Shukla and Huang 2015). Developing Indian Ocean diploe (IOD) events can influence the ISM rainfall via an anomalous meridional circulation and water vapor transport anomalies (Ashok et al. 2001; Guan and Yamagata 2003; Saji and Yamagata 2003). Vimal et al. (2012) revealed a dipole mode of Indian summer rainfall anomalies, which is associated with simultaneous SST anomalies in the northern Indian Ocean.

A strong WNPSM leads to increased rainfall over the Philippine Sea (Wang et al. 2001). Enhanced convective heating over the Philippines Sea excites the positive East Asia/Pacific teleconnection (EAP) pattern, with anomalous cyclones over the Indo-China Peninsula to the Philippine Sea and Okhotsk Sea, and an anomalous anticyclone over North China to Japan, at the 500 hPa level (Huang and Li 1987; Nitta 1987; Kosaka and Nakamura 2006). The positive EAP pattern results in northward movement of the WNP subtropical high (WNPSH) and hence reduced rainfall over the Yangtze–Huaihe River valley, Korean Peninsula, and Japan (Nitta 1987; Huang and Sun 1994). Summer SST anomalies in the tropical Indo-Pacific region (e.g., the equatorial central and eastern Pacific, tropical western Pacific, and tropical Indian Ocean) can modulate the WNPSM and rainfall anomalies over the Philippine Sea, and hence excite the EAP pattern (e.g., Huang and Li 1987; Nitta 1987; Wang and Zhang 2002; Chou et al. 2003; Li and Wang 2005; Yang et al. 2007; Li et al. 2008; Kim et al. 2009; Wu et al. 2009, 2010; Xie et al. 2009; Li and Zhou 2012; Lu and Lu 2014; He 2015). Although the negative correlation between the SSTs and rainfall over the WNP in summer indicates that the atmospheric variations lead to SST anomalies, local SST anomalies may directly influence the WNPSM (Nitta 1987; Huang and Sun 1994; Lu and Lu 2014; Zhou 2014). Warming in the equatorial central and eastern Pacific during developing El Niño events enhances local convection and then excites an anomalous cyclone over the WNP according to the Gill pattern (Gill 1980), intensifying the WNPSM (Wang and Zhang 2002; Wu et al. 2009). Basin-wide warming in the tropical Indian Ocean induces a baroclinic Kelvin wave and hence triggers an anomalous anticyclone over the WNP, favoring a weak WNPSM (Yang et al. 2007; Xie et al. 2009). Warming in the tropical southeast Indian Ocean can also result in a weak WNPSM via an anomalous meridional circulation (Wu et al. 2012b).

Severe drought occurred over NEA in summer 2014, and external factors such as Arctic sea ice anomalies and tropical western Pacific warming played an important role (Wang and He 2015). The severe drought was accompanied by the weak EASM. The values of the EASM index defined by Wang (2002) in June, July, August, and summer mean of 2014 were −1.51, −1.57, −0.61 and −1.62, respectively. Teleconnections such as the Eurasian, EAP, and CGT patterns were prominent (Wang and He 2015). Analysis in the present study found that centers of the severe drought moved northward remarkably during summer 2014 (Fig. 1b–d). Less attention is paid to the northward movement of rainfall anomalies during summer over East Asia and its mechanism. There were remarkable SST anomalies in the whole tropical Indo-Pacific region during summer 2014, for instance, a negative IOD in August 2014, and warming in the equatorial eastern Pacific in each month of summer 2014 (Fig. 2). The value of the standardized IOD index proposed by Saji et al. (1999) in August 2014 was −0.81. The values of Niño3 index were 0.50, 0.46, and 0.15 °C for the months of summer 2014, respectively. Rainfall anomalies over India and the tropical WNP were remarkable (Fig. 3). While SST anomalies were positive in the tropical WNP during summer 2014 (Fig. 2), rainfall anomalies over the region were positive in July 2014 and negative in June and August 2014 (Fig. 3). This means that SST anomalies led to atmospheric anomalies over the tropical WNP in July 2014, but atmospheric anomalies led to SST anomalies in June and August 2014 (Wang et al. 2005; Wu and Kirtman 2007; Lu and Lu 2014). These results indicate that the SST anomalies in the tropical Indo-Pacific region responsible for each month’s drought in summer 2014 may be different. Therefore, the main purpose of this study was to investigate the contribution of SST anomalies in the tropical Indo-Pacific region to the severe drought over NEA in summer 2014 and the mechanisms involved from the month-to-month perspective. The data and methods used are described in Sect. 2. Section 3 presents the month-to-month evolution of the severe drought. Section 4 addresses the role of the SST anomalies in the tropical Indo-Pacific region in each month’s drought. Section 5 examines the possible causes of the northward movement of the severe drought during summer. A summary and further discussion are provided in Sect. 6.

Fig. 1
figure 1

Rainfall anomalies (mm day−1) over NEA in 2014: a June–July–August (JJA); b June; c July; and d August. NEA refers to the continent marked by the red rectangle in (a). The middle and lower reaches of the Yangtze River, the Huaihe River valley, North China and northeast China are also marked in (a). The rectangles in (b)–(d) mark the drought regions where the water vapor budgets are calculated. The vertically integrated water vapor transport anomalies (107 kg s−1) for each boundary are showed. Red (blue) rows and numbers represent northward (southward) or eastward (westward) water vapor transport

Full size image
Fig. 2
figure 2

SST anomalies (°C) in 2014: a June; b July; c August. Rectangles indicate the regions where SST anomalies are used to carry out the sensitive experiments

Full size image
Fig. 3
figure 3

Rainfall anomalies (mm day−1) over the Asia–Pacific region in 2014: a June; b July; and c August

Full size image

2 Data and methods

The datasets used in this study included: (1) the Global Precipitation Climatology Project monthly precipitation, with a horizontal resolution of 2.5° × 2.5° (Adler et al. 2003); (2) the National Centers for Environmental Prediction/National Center for Atmospheric Research monthly reanalysis data, with a horizontal resolution of 2.5° × 2.5° (Kalnay et al. 1996); (3) the National Oceanic and Atmospheric Administration Extended Reconstructed SST, version 3b, with a horizontal resolution of 2.0° × 2.0° (Smith et al. 2008).

In this study, NEA was defined as the continental region of (28°–54°N, 110°–146°E) (see Fig. 1a). This region covers the middle and lower reaches of the Yangtze River, the Huaihe River valley, North China, northeast China, the Korean Peninsula, Japan, and the Stanovoy Range. The summer season was taken to include the months of June, July and August. The average of 1981–2010 was selected as the climatology to calculate the anomalies of variables in 2014. Statistical methods such as simple linear regression and multivariate Empirical Orthogonal Function (MV-EOF) analysis were employed. The simple linear regression was calculated for the period of 1981–2013, to exclude the influence of anomalies in summer 2014 on regression coefficients. The MV-EOF analysis was performed to reveal the coupled modes between atmospheric circulation anomalies and rainfall anomalies. The two-tailed Student’s t test was used to test the statistical significance of regression coefficients. The criteria proposed by North et al. (1982) was used to test whether the modes of the MV-EOF analysis were isolated.

Numerical experiments were performed to study the role of SST anomalies in the tropical Indo-Pacific region in the severe drought over NEA. The ECHAM5 atmospheric general circulation model, developed by the Max Planck Institute for Meteorology, was employed. ECHAM5 is a global spectral model that provides several horizontal and vertical resolution options. We selected a horizontal resolution of T63 and a vertical resolution with 19 levels. The experiments were performed as follows: First, a control experiment was run, in which AMIP II climatological mid-month SST was prescribed, the simulation was integrated for 50 years, and the last 30 years’ results were used as the samples. Then, the SST anomaly sensitivity experiments were run, for which observational SST anomalies added to the climatological SSTs were prescribed, and the simulations were integrated from 31 December to 31 August of the following year with the initial conditions taken from the control experiment. Each sensitivity experiment included 30 samples. See the experiment designs in Table 1.

Table 1 Experiment designs
Full size table

3 Severe drought over NEA during summer 2014

The spatio-temporal features of the severe drought over NEA during summer 2014 were addressed first. The summer mean drought was located in the Huaihe River valley, North China, most of northeast China, and the Korean Peninsula (Fig. 1a). There were two drought centers: one in the Huaihe River valley and the other in the central Korean Peninsula. In June 2014, the drought belt was south of the summer mean drought belt (cf. Fig. 1a, b), with two drought centers in the Yangtze–Huaihe River valley and southern Korean Peninsula to southern Japan, respectively (Fig. 1b). The spatial pattern of the severe drought in July 2014 was similar to that of the summer mean, except that Japan also experienced drought (cf. Fig. 1a, c). The drought belt in August 2014 moved northward compared to that of the summer mean (cf. Fig. 1a, d). It covered North China, northeast China, the northern Korean Peninsula, and the Stanovoy Range (Fig. 1d). Two drought centers were situated in southern parts of northeast China to the northern Korean Peninsula and the Stanovoy Range, respectively. The maximum decreased rainfall in June, July and August 2014, which were all located around the Korean Peninsula, were as high as approximately 4.8, 5.9, and 5.8 mm day−1, respectively. The corresponding decreased percent for the maximum decreased rainfall are 50.2, 59.8, and 74.0%, respectively.

Overall, the summer mean drought in 2014 occurred in the Huaihe River valley, North China, most of northeast China, and the Korean Peninsula. However, the severe drought during summer 2014 experienced obvious month-to-month evolution. It moved northward from June to August 2014, which was partly related to the monthly northward movement of the East Asian subtropical westerly jet during summer 2014, as revealed later in Sect. 5. The evolutions of normalized rainfall anomalies and 1-month standardized precipitation index over NEA during summer 2014 were also analyzed. The results are similar to those described above.

4 Mechanisms for each month’s drought over NEA during summer 2014

As mentioned in Sect. 1, the SST anomalies in the tropical Indo-Pacific region may contribute to the severe drought over NEA. Accordingly, in this section, the related atmospheric circulation anomalies and the underling physical processes induced by the SST anomalies in the tropical Indo-Pacific region are demonstrated for each month’s drought in summer 2014 by observation and model results.

4.1 June

In June 2014, there were anomalous cyclones over the South China Sea and south of Japan, and an anomalous anticyclone around Lake Baikal, at the 850 hPa level (Fig. 4a). Influenced by such anomalous circulations, anomalous northeasterlies dominated over East Asia, which offset the climatological southwesterlies. This meant that the lower-level summer monsoon flow weakened and less water vapor was transported to NEA, favoring the occurrence of drought. The anomalous divergence of water vapor transport for the two drought centers was mainly caused by the weakened input at the southern boundary (Fig. 1b). At the 200 hPa level (Fig. 4c), the drought belt in June 2014 was influenced by an anomalous anticyclone over the middle reach of the Yangtze River, and an anomalous cyclone around southern Japan. As a result, anomalous northwesterlies and northeasterlies converged over the drought belt at the 200 hPa level (Fig. 4c). Over the drought belt in June 2014, anomalous convergence at the 200 hPa level (shaded, Fig. 4c) combined with anomalous divergence at the 850 hPa level (shaded, Fig. 4a) led to anomalous descent (shaded, Fig. 4b). Therefore, drought occurred.

Fig. 4
figure 4

Atmospheric circulation anomalies in June 2014: a 850 hPa wind (vectors; m s−1) and horizontal divergence (shaded; 10−6 s−1); b 500 hPa wind and 700 hPa vertical velocity (shaded; 10−2 Pa s−1); c 200 hPa wind and horizontal divergence. For horizontal divergence and vertical velocity anomalies, only the region (25°–57.5°N, 105°–150°E) is plotted The drought belt is marked in (b) by the red dashed rectangle

Full size image

Rainfall over India and the Philippine Sea also decreased in June 2014 (Fig. 3a). This implies that SST anomalies in the tropical Indo-Pacific region may contribute to the severe drought over NEA in June 2014 via the modulation of rainfall over the two regions. We applied MV-EOF analysis to the 200 hPa wind and rainfall anomalies to reveal the coupled modes between the atmospheric circulation anomalies and rainfall anomalies over Asia in June (Fig. 5). The first MV-EOF mode (MV-EOF1) accounts for approximately 16.9% of the total variance and is distinguishable from other modes. The spatial pattern of the wind anomalies of the MV-EOF1 (Fig. 5a) is the positive MAS pattern revealed by Wu (2002), with anomalous anticyclones over the northwest of India and Korean Peninsula and an anomalous cyclone over southwest China. The rainfall anomalies are in-phase between northeast India and the region of 30°–42°N over NEA (Fig. 5a). The wind and rainfall anomalies of the MV-EOF1 support the fact that the MAS pattern is an important bridge between the ISM and EASM in June. The value of the standardized time coefficients of the MV-EOF1 in June 2014 was approximately −0.90 (Fig. 5b). This indicates that the negative MAS pattern over Asia and the related rainfall anomalies were notable in June 2014.

Fig. 5
figure 5

a Spatial patterns of the MV-EOF1 of June rainfall (shaded) and 200 hPa wind (vectors) anomalies over Asia for 1981–2014, and b corresponding standardized time coefficients. MV-EOF1 accounts for approximately 16.9% of the total variance

Full size image

The regressions against the reversed standardized time coefficients (multiplied by −1.0) of the MV-EOF1 are presented in Fig. 6, to show the rainfall and atmospheric circulation anomalies associated with the negative MAS pattern in June. The quasi-barotropic anomalous cyclone around southern Japan in June 2014 was mainly caused by the negative MAS pattern (cf. Figs. 4, 6b–d). At the 500 and 200 hPa levels, the anomalous anticyclone over the middle reach of the Yangtze River in June 2014 was partly associated with the negative MAS pattern (cf. Figs. 4b, c, 6c, d). Therefore, the negative MAS pattern contributed to the severe drought over NEA in June 2014, especially over the southern Korean Peninsula (cf. Figs. 1b, 6a). It should be noted that a negative MAS pattern exhibited an anomalous cyclonic circulation over the northwest of India at the 200 hPa level (Fig. 6d), but an anomalous anticyclonic circulation appeared in June 2014, with an anomalous cyclonic circulation over northwest China (Fig. 4c). This was possibly because other mechanisms, for instance, upstream Rossby waves, also contributed to the atmospheric circulation anomalies over the northwest of India and the negative MAS pattern (Wu 2002).

Fig. 6
figure 6

Regressed June (a) rainfall (mm day−1) and (b–d) wind (vectors; m s−1) for 1981–2013 against the standardized time coefficients corresponding to the MV-EOF1 of June rainfall and 200 hPa wind anomalies over Asia. The standardized time coefficients are multiplied by −1.0. Black dots in (a) indicate the regression coefficients are significant at the 0.05 significance level. Light and dark shading in (b–d) indicate the regression coefficients of zonal or meridional wind are significant at the 0.10 and 0.05 significance level, respectively

Full size image

By comparing Fig. 3a with Fig. 6a, it is apparent that the rainfall anomalies over northeast India were responsible for the negative MAS pattern in June 2014. But what caused the rainfall anomalies over northeast India? Warmer SSTs appeared over the Arabian Sea (around 10°N, 60°E; Fig. 2a), which were accompanied by local increased rainfall (Fig. 3a). Such an SST–rainfall relationship indicates that the warmer SSTs contributed to rainfall and circulation anomalies over that region. The anomalous cyclonic circulation at the 850 hPa level over the Arabian Sea (Fig. 4a), caused by local warmer SSTs, offset the lower-level westerlies and led to reduced rainfall over the west coast of India (Fig. 3a). This reduced rainfall further induced an anticyclone at the 850 hPa level over India and weakened the northward transportation of water vapor over northeast India (Fig. 4a), subsequently leading to reduced rainfall over northeast India (Fig. 3a). The EXP-Jun-AS supports that the configuration of SST anomalies in the Arabian Sea contributed to the rainfall and wind anomalies at the 850 hPa level over the Arabian Sea and India in June 2014 (cf. Figs. 3a, 4a, 7a, c). However, the decreased rainfall occurs only over some parts of India in EXP-Jun-AS and is much weaker than that in the observation (cf. Figs. 3a, 7a). Two reasons may be responsible for this. One is that the anomalous anticyclone in EXP-Jun-AS moves southeastward compared to that in the observation and is located in the northern Bay of Bengal (cf. Figs. 4a, 7c). The other is that besides warmer SSTs in the Arabian Sea, other processes also contributed to the decreased rainfall in India. Because the simulated decreased rainfall over northeast India is weak (Fig. 7a), the EXP-Jun-AS failed to reproduce the negative MAS pattern and the corresponding drought in June 2014. The converse relationship between summer warmer SSTs in the Arabian Sea and rainfall anomalies over northeast India has been revealed in previous studies (Vimal et al. 2012; Shukla and Huang 2015). Therefore, warmer SSTs in the Arabian Sea were an external factor, probably not an important factor, responsible for the severe drought over NEA in June 2014.

Fig. 7
figure 7

Rainfall (mm day−1) and 850 hPa wind (m s−1) anomalies for EXP-Jun-AS (left panel) and EXP-Jun-SCS (right panel). Black dots in (a) and (b) indicate rainfall anomalies are significant at the 0.05 significance level. Light and dark shading in (c) and (d) indicate the anomalies of zonal or meridional wind are significant at the 0.10 and 0.05 significance level, respectively

Full size image

Analogously, warmer SSTs in the South China Sea (Fig. 2a) induced local increased rainfall (Fig. 3a) and a baroclinic structure of the atmospheric circulation anomalies over the South China Sea, with anomalous cyclones at the 850 and 500 hPa levels (Fig. 4a, b) and an anomalous anticyclone at the 200 hPa level (Fig. 4c). The related anomalous northeasterlies at the 850 hPa level counteracted the climatological lower-level summer monsoon flow over South China and contributed to the severe drought over NEA in June 2014. The EXP-Jun-SCS demonstrates that warmer SSTs in the South China Sea induced locally enhanced rainfall and an anomalous cyclone at the 850 hPa level, and hence led to anomalous northeasterlies over South China and decreased rainfall in the middle and lower reach of the Yangtze River (cf. Figs. 3a, 4a, 7b, d).

There were warmer SSTs in the Philippine Sea (Fig. 2a), but the rainfall decreased (Fig. 3a). The out-of-phase between the SST anomalies and rainfall anomalies means that the atmospheric circulation anomalies led to SST anomalies over the Philippine Sea. Therefore, the warming ocean in the Philippine Sea made no contribution to the severe drought over NEA in June 2014. The reduced rainfall over the Philippine Sea was related to the anomalous meridional and zonal circulation cells (Fig. 8a) induced by the warming in east of New Guinea and off the equator in the central Pacific [around (10°N, 180°E)] (Fig. 2a), respectively. Based on the reduced rainfall region over the Philippine Sea (Fig. 3a), we defined an area-averaged rainfall index over (6.25°–21.25°N, 126.25°–163.75°E) from 1981 to 2014 and calculated regressions against it to demonstrate whether it was responsible for the severe drought over NEA in June 2014. Reduced rainfall over the Philippine Sea in June excites the negative EAP pattern, with a quasi-barotropic anomalous cyclone over central Japan. The anomalous center is located at around 40°N, which is north of that in June 2014. In addition, the reduced rainfall over the Philippine Sea is unable to induce a statistically significant rainfall reduction over the Korean Peninsula and southern Japan. Therefore, the reduced rainfall over the Philippine Sea and the warming in east of New Guinea and off the equator in the central Pacific were not important to the severe drought over NEA in June 2014. The rainfall anomalies over the Philippines Sea were remarkable in June 2014, but the positive EAP was unremarkable possibly due to weak easterly shear over the Philippines Sea in June (Lu 2004). The horizontal divergent wind anomalies induced by the warming in the equatorial eastern pacific mainly confined to east of 150°W (Fig. 8a), and hence had little effect on the rainfall over northeast India and the tropical WNP. Therefore, the warming in the equatorial eastern pacific may be unimportant to this month severe over NEA.

Fig. 8
figure 8

Velocity potential anomalies (contours; 105 m2 s−1) and corresponding divergent wind (vectors; m s−1) at the 200 hPa level in 2014: a June; b July; c August

Full size image

4.2 July

In July 2014, anomalous northeasterlies prevailed over East Asia south of 40°N at the 850 hPa level (Fig. 9a). This led to weak lower-level summer monsoon flow and less water vapor transported to NEA, favoring the occurrence of drought. The anomalous divergence of water vapor transport was mainly caused by the weakened input at the southern boundary for the drought in the Huaihe River valley. Weakened input at the western boundary and strengthened output at the eastern boundary were both important to the drought center in the Central Korean Peninsula (Fig. 1c). The anomalous northeasterlies were caused by the anomalous cyclone over the South China Sea to the Philippine Sea and the anomalous anticyclone over eastern China to Japan (Fig. 9a). At the 200 hPa level, there were anomalous cyclones over the Yangtze–Yellow River valley and Stanovoy Range, and anomalous anticyclones over northwest China and the subtropical WNP (Fig. 9c). This led to anomalous divergence at the 200 hPa level over the drought belt in July 2014 (Fig. 9c). Here, anomalous convergence at the 200 hPa level (Fig. 9c, shaded) and anomalous divergence at the 850 hPa level (Fig. 9a, shaded) induced anomalous descent (Fig. 9b, shaded) and, subsequently, the severe drought (Fig. 1c).

Fig. 9
figure 9

Atmospheric circulation anomalies in July 2014: a 850 hPa wind (vectors; m s−1) and horizontal divergence (shaded; 10−6 s−1); b 500 hPa wind and vertical velocity (shaded; 10−2 Pa s−1); c 200 hPa wind and horizontal divergence. For horizontal divergence and vertical velocity anomalies, only the area (25°–57.5°N, 105°–150°E) is plotted. The drought belt is marked in (b) by the red dashed rectangle

Full size image

The MAS pattern exhibits a quasi-barotropic anomalous circulation center around the Korean Peninsula, which is critical for rainfall variation over NEA (Wu 2002). However, this was not the case for the atmospheric circulation anomalies in July 2014 (Fig. 9). Therefore, the contribution of the MAS pattern to the severe drought in July 2014 was negligible.

Enhanced rainfall appeared over the Philippine Sea in July 2014 (Fig. 3b). This meant that a strong WNPSM may play an important role via the EAP pattern. We applied MV-EOF analysis to the 850 hPa wind and rainfall anomalies over the East Asia–WNP region to reveal the major coupled modes between the rainfall anomalies and lower-level circulation anomalies (Fig. 10). MV-EOF1 and MV-EOF2 account for approximately 20.8 and 14.0% of the total variance respectively, and are both distinguishable from neighboring modes. The spatial patterns of MV-EOF1 exhibit increased rainfall over the Philippine Sea to South China across the South China Sea (Fig. 10a, shaded). At the 850 hPa level, an anomalous cyclonic circulation dominates, corresponding to the rainfall anomalies (Fig. 10a, vectors). Therefore, MV-EOF1 represents the variation of the WNPSM and related rainfall. The value of the standardized time coefficients of the MV-EOF1 in July 2014 was 1.01, which also indicates that the WNPSM was stronger than normal in July 2014. A strong WNPSM in July led to reduced rainfall over most of NEA (Fig. 11a) through inducing the positive EAP pattern at the 850 and 500 hPa levels (Fig. 11c, e) and a wave-like pattern at the 200 hPa level over the Asia–WNP region (Fig. 11g). In July 2014, the strong WNPSM contributed to the anomalous cyclone over the South China Sea to the Philippine Sea and the anomalous anticyclone over eastern China to Japan at the 850 hPa level, and hence led to the weak lower-level summer monsoon flow over East Asia south of 40°N (cf. Figs. 9a, 11c). At the 200 hPa level, the anomalous cyclone over the Yangtze–Yellow River valley and the anomalous anticyclone over northwest China in July 2014 were also closely related to the strong WNPSM (cf. Figs. 9c, 11g). Therefore, the strong WNPSM played an important role in the severe drought over NEA in July 2014.

Fig. 10
figure 10

Spatial patterns of the a MV-EOF1 and b MV-EOF2 of July rainfall (shaded) and 850 hPa wind (vectors) anomalies over the East Asia–Northwest Pacific region for 1981–2014 and their (c, d) corresponding standardized time coefficients. MV-EOF1 and MV-EOF2 account for approximately 20.8 and 14.0% of the total variance, respectively

Full size image
Fig. 11
figure 11

Regressed July (a) rainfall (mm day−1) and (c, e, g) wind (m s−1) for 1981–2013 against the standardized time coefficients corresponding to the MV-EOF1 of July rainfall and 850 hPa wind anomalies over the East Asia–Northwest Pacific region. b, d, f, h As in (a, c, e, g), but for MV-EOF2. Black dots in (a, b) indicate that regression coefficients are significant at the 0.05 significance level. Light and dark shading in (c–h) indicate the regression coefficients of zonal or meridional wind are significant at the 0.10 and 0.05 significance level, respectively

Full size image

Corresponding to the strong WNPSM and increased rainfall over the Philippine Sea (Fig. 3b), SSTs were warmer than normal in the tropical WNP (Fig. 2b). Such an SST–rainfall relationship indicates that warmer SSTs led to the strong WNPSM and increased rainfall over the Philippine Sea. EXP-Jul-TWNP basically reproduced the rainfall and the atmospheric circulation anomalies over the tropical Indian Ocean and WNP in July 2014 (cf. Figs. 3b, 9a, 8b, 12). The positive EAP pattern was also reproduced well (cf. Figs. 9a, 12b), in spite of failing to reproduce the corresponding drought (cf. Figs. 3b, 11a, 12a). Therefore, warmer SSTs in the tropical WNP were an important external factor for the severe drought over NEA in July 2014.

Fig. 12
figure 12

a Rainfall (mm day−1), b 850 hPa wind (m s−1), and c 200 hPa velocity potential (105 m2 s−2) and divergent wind (m s−1) anomalies for EXP-Jul-TWNP. Black dots in (a) indicate rainfall anomalies are significant at the 0.05 significance level. Light and dark shading in (b) indicate the anomalies of zonal or meridional wind are significant at the 0.10 and 0.05 significance level, respectively

Full size image

The spatial patterns of the MV-EOF2 show a horseshoe pattern for the rainfall anomalies, with reduced rainfall over the subtropical WNP and increased rainfall around it (Fig. 10b, shaded). Corresponding to the rainfall anomalies, there is an anomalous anticyclonic circulation over the subtropical WNP and an anomalous cyclonic circulation over the South China Sea to the tropical WNP at the 850 hPa level (Fig. 10b, vectors). The value of the standardized time coefficients of MV-EOF2 in July 2014 was 1.39 (Fig. 10d). This indicates that MV-EOF2 was notable. MV-EOF2 triggers an EAP-like pattern over the East Asia–WNP region (Fig. 11d, f, h), which also appeared in the circulation anomalies at the 850 and 500 hPa levels in July 2014 (Fig. 9a, b). The EAP-like pattern made a contribution to the anomalous cyclonic circulation over the Stanovoy Range in July 2014 (cf. Figs. 9, 11d, f, h), and subsequently to the severe drought over the lower and middle reaches of the Yellow River (cf. Figs. 1c, 11b).

As revealed by previous studies (Wang and Zhang 2002; Wu et al. 2009), summer warming in the equatorial central and eastern Pacific during developing El Niño events can lead to local increased rainfall and enhanced condensational heating. A pair of anomalous cyclones over the western Pacific are excited by the enhanced heating according to the Gill pattern. This leads to increased rainfall over the off-equatorial WNP [close to (10°N, 150°E)], which may then trigger the EAP-like pattern (Kosaka et al. 2012). In July 2014, warmer SSTs appeared over the equatorial eastern Pacific (Fig. 2b). This contributed to increased rainfall over the equatorial central and eastern Pacific (Fig. 3b), inducing an anomalous cyclone over the off-equatorial WNP [close to (10°N, 150°E); Fig. 9a]. The anomalous cyclone led to increased rainfall locally (Fig. 3b), triggering the EAP-like pattern (Fig. 9a, b). It should be noted that local warming (Fig. 3b) was another reason for the increased rainfall over the off-equatorial WNP. The correlation coefficient between the time coefficients of MV-EOF2 and Niño-3 index in July during 1981–2013 was 0.55, which was statistically significant at the 0.01 significance level. The regressed anomalous circulations and rainfall against Niño-3 index in July over the East Asia–WNP were similar to those of MV-EOF2 (Fig. 11b, d, f, g). Therefore, remarkable warming in the equatorial eastern Pacific (Fig. 2b) was another important external factor for the severe drought over NEA in July 2014. However, The EXP-Jul-EEP failed to reproduce the physical processes described above due to failing to reproduce the enhanced rainfall over the equatorial central Pacific.

There were reduced rainfall over the equatorial central Indian Ocean (Fig. 3b), where the SSTs were warmer than normal (Fig. 2b). The out-of-phase between rainfall anomalies and SST anomalies over that region reflects that the SST anomalies were caused by the atmospheric circulation anomalies, which was induced by the enhanced convection over the tropical WNP (Figs. 8b, 12c). Warming in southwest of Sumatra led to increased rainfall in situ (cf. Figs. 2b, 3b). However, the increased rainfall over there was unable to induce anomalous meridional circulation cell to influence the rainfall over the tropical WNP (Fig. 8b). Therefore, the SST anomalies in the tropical Indian Ocean may make no significant contribution to the severe drought in the NEA in July 2014.

4.3 August

At the 850 and 500 hPa levels, wind anomalies exhibited a negative EAP pattern in August 2014 (Fig. 13a, b). Meanwhile, a negative MAS pattern dominated at the 200 hPa level (Fig. 13c). Influenced by the negative EAP and MAS patterns, anomalous convergence at the 200 hPa level and anomalous divergence at the 850 hPa level (Fig. 13a, c, shaded) over the drought belt in August 2014 led to anomalous descent (Fig. 13b, shaded) and the severe drought. The anomalous northeasterlies prevailed over the drought belt in August 2014, offsetting the lower-level summer monsoon flow (Fig. 13a) and favoring the severe drought in that month. The anomalous divergence of water vapor transport for the southern drought center was mainly caused by the weakened input at the southern boundary (Fig. 1d). The weakened input at the southern and western boundary were both important to the anomalous divergence of water vapor transport of the northern drought center (Fig. 1d).

Fig. 13
figure 13

Atmospheric circulation anomalies in August 2014: a 850 hPa wind (m s−1) and horizontal divergence (10−6 s−1); b 500 hPa wind and vertical velocity (10−2 Pa s−1); c 200 hPa wind and horizontal divergence. For horizontal divergence and vertical velocity anomalies, only the area (25°–57.5°N, 105°–150°E) is plotted. The drought belt is marked in (b) by the red dashed rectangle

Full size image

Rainfall decreased remarkably over India and the Philippine Sea in August 2014 (Fig. 3c). This means that the ISM and WNPSM were both weaker than normal, lending further support to the fact that the negative MAS and EAP patterns were two important mechanisms for the severe drought in August 2014. Analogously, the MV-EOF1 of 200 hPa wind and rainfall anomalies over Asia in August represents the coupling between the MAS pattern and rainfall anomalies. MV-EOF1 accounts for 17.4% of the total variance, and is distinguishable from other modes. The value of the standardized time coefficients of the MV-EOF1 in August 2014 was −1.40, indicating a notable negative MAS pattern. Figure 14 displays the circulation and rainfall anomalies associated with the negative MAS pattern in August. The negative MAS pattern made an important contribution to the quasi-barotropic anomalous cyclone around the Sea of Japan (cf. Figs. 13, 14b–d) and then led to drought over the east parts of North China, southern parts of northeast China, the northern Korean Peninsula, and the Stanovoy Range in August 2014 (cf. Figs. 1d, 14a). The negative MAS pattern in August 2014 may be excited by the reduced rainfall belt at around 20°N from India to the southeast of Taiwan (cf. Figs. 3c, 14a), similar to the findings of Guan and Yamagata (2003). The variation of the WNPSM and related rainfall in August can be depicted by the MV-EOF1 of 850 hPa wind and rainfall anomalies over the East Asia–WNP region, which accounts for 24.0% of the total variance and is distinguishable from other modes. In August 2014, the anomalous anticyclonic circulations over the Stanovoy Range were partly caused by the negative EAP pattern triggered by the weak WNPSM (cf. Figs. 13, 15b–d). Therefore, the weak WNPSM played a role in the drought over the Stanovoy Range in August 2014 (cf. Figs. 1d, 15a).

Fig. 14
figure 14

Regressed August a rainfall (mm day−1) and b–d wind (m s−1) for 1981–2013 against the standardized time coefficients corresponding to the MV-EOF1 of August rainfall and 200 hPa wind anomalies over Asia. The standardized time coefficients are multiplied by −1.0. Black dots in (a) indicate the regression coefficients are significant at the 0.05 significance level. Light and dark shading in (b–d) indicate the regression coefficients of zonal or meridional wind are significant at the 0.10 and 0.05 significance level respectively

Full size image
Fig. 15
figure 15

Regressed August a rainfall (mm day−1) and b–d wind (m s−1) for 1981–2013 against the standardized time coefficients corresponding to the MV-EOF1 of August rainfall and 850 hPa wind anomalies over the East Asia–Northwest Pacific region. The standardized time coefficients are multiplied by −1.0. Black dots in (a) indicate the regression coefficients are significant at the 0.05 significance level. Light and dark shading in (b–d) indicate the regression coefficients of zonal or meridional wind are significant at the 0.10 and 0.05 significance level respectively

Full size image

The out-of-phase rainfall-SST relationship over the tropical WNP (cf. Figs. 2c, 3c) indicates that the atmospheric circulation anomalies led to SST anomalies over the regions. Therefore, the sea surface warming in the tropical WNP made no contribution to the severe drought over NEA in August 2014. Corresponding to the reduced rainfall over India to the tropical WNP (Fig. 3c), anomalous convergence dominated over that region at the 200 hPa level (Fig. 8c). It was connected to the anomalous divergence over the equatorial eastern Pacific via an anomalous zonal circulation, and to the anomalous divergence over the tropical southeast Indian Ocean and Indonesia via an anomalous meridional circulation (Fig. 8c). The anomalous zonal circulation was caused by the increased rainfall over the equatorial eastern Pacific (Fig. 3c), which was partly induced by the warmer SSTs in that region (Fig. 2c). Therefore, warmer SSTs in the equatorial eastern Pacific favored a weak WPNSM and reduced rainfall over the Philippine Sea via an anomalous zonal circulation, then triggered a negative EAP pattern, and finally contributed to the severe drought over NEA in August 2014. The EXP-Aug-EEP basically reproduced the enhanced rainfall and its induced Gill pattern over the equatorial eastern Pacific in August 2014. The enhanced rainfall in the equatorial eastern Pacific in the EXP-Aug-EEP triggered an anomalous zonal circulation cell, but the decent branch of the cell is located in the around (10°N, 165°W), which is 30° of longitude east than that in observation (Fig. 8c). Therefore, the EXP-Aug-EEP failed to reproduce the decreased rainfall in the tropical WNP, and then the negative EAP pattern and corresponding drought over NEA.

The anomalous meridional circulation over the tropical Indian Ocean to the tropical WNP in August 2014 (Fig. 8c) was closely related to the negative phase of the IOD, with colder SSTs in the tropical western Indian Ocean and warmer SSTs in the tropical southeast Indian Ocean (Fig. 2c). The negative IOD led to increased rainfall over the tropical southeast Indian Ocean and Indonesia (Fig. 3c). The increased rainfall induced anomalous divergence over these regions at the 200 hPa level (Fig. 8c). The anomalous northward divergent flow converged over India to the WNP (Fig. 8c) and resulted in reduced rainfall over that region, which excited the negative MAS and EAP patterns (Fig. 13). The physical processes described above are consistent with previous studies (e.g. Guan and Yamagata 2003). Therefore, the negative IOD in the tropical Indian Ocean was an important external factor that contributed to the severe drought over NEA in August 2014. The EXP-Aug-TIO basically reproduced the rainfall and atmospheric circulation anomalies over the tropical Indian Ocean and western Pacific in August 2014 (cf. Figs. 3c, 13a, 8c, 16), but failed to reproduce the negative MAS pattern and corresponding drought probably due to the weak rainfall anomalies in northwest India of EXP-Aug-TIO and limited skill of ECHAM5.

Fig. 16
figure 16

a Rainfall (mm day−1), b 850 hPa wind (m s−1), and c 200 hPa velocity potential (105 m2 s−2) and divergent wind (m s−1) anomalies for EXP-Aug-TIO. Black dots in (a) indicate rainfall anomalies are significant at the 0.05 significance level. Light and dark shading in (b) indicate the anomalies of zonal or meridional wind are significant at the 0.10 and 0.05 significance level, respectively

Full size image

The severe drought over NEA in summer 2014 was accompanied by weak lower-level summer monsoon flow. SST anomalies in the tropical Indo-Pacific region made an important contribution to the severe drought. The MAS and EAP patterns were the main bridges between the SSTs anomalies in the tropical Indo-Pacific region and the weak lower-level summer monsoon flow and severe drought. In June 2014, warming in the Arabian Sea induced reduced rainfall over northeast India and then excited the negative MAS pattern favoring the severe drought. Warming in the South China Sea was another reason for the severe drought in June 2014. In July 2014, the positive EAP pattern and the EAP-like pattern were responsible for the severe drought. The two teleconnections were excited by the reduced rainfall over the tropical WNP, which were caused by warming in the tropical WNP and equatorial eastern Pacific. In August 2014, the negative IOD and warming in the equatorial eastern Pacific led to weakened ISM and WNPSM rainfall via anomalous meridional and zonal circulations, respectively. The anomalous heating released by the reduced rainfall triggered the negative MAS and EAP patterns, favoring the severe drought in August 2014.

5 Possible reasons for the northward movement of the severe drought

As revealed in Sect. 3, the severe drought over NEA experienced northward movement during summer 2014. Based on the analysis in Sect. 4, two possible reasons are proposed. One is the monthly northward movement of the East Asian subtropical westerly jet during summer 2014. Climatologically, the East Asian subtropical westerly jet axis at the 200 hPa level for June and August is located approximately along 37.5°N and 45°N, respectively (Lu 2004). In June and August, the anomalous circulation center of the MAS pattern over the Korean Peninsula at the 200 hPa level was located almost at the same latitude as the jet axis (Figs. 6d, 14d). Therefore, it can be inferred that the East Asian subtropical westerly jet axis in June being south of that in August may be an important reason for the anomalous circulation center of the MAS pattern over the Korean Peninsula in June being south of that in August (cf. Figs. 6d, 14d). The meridional location of the anomalous circulation centers of the CGT pattern over NEA depends on that of the East Asian subtropical westerly jet, owing to its propagating eastward along westerly jet (Ding and Wang 2005). Therefore, the inference is reasonable considering that the wave-type variation of the MAS pattern is closely related to the CGT pattern over Asia. In fact, both the MAS and CGT patterns can be triggered by the ISM rainfall anomalies. The anomalous circulation centers of the MAS pattern over the northwest of India and around the Korean Peninsula in June and August (Figs. 6d, 14d) were basically consistent with that of the CGT revealed by Ding and Wang (2005). Meanwhile, the rainfall anomalies for the MAS pattern over Asia in June and August (Figs. 6a, 14a) were similar to those for the CGT pattern. As revealed in Sect. 4, the negative MAS pattern played an important role in the severe drought over NEA in June and August 2014. Furthermore, the East Asian subtropical westerly jet axis in June 2014 was south of that in August 2014. As a result, the quasi-barotropic anomalous cyclone around the Korean Peninsula, and the related reduced rainfall, in June 2014 were south of those in August 2014. This may partly explain why the severe drought in June 2014 was south of that in August 2014. In short, the East Asian subtropical westerly jet axis in June 2014 was south of that in August 2014. This led to the quasi-barotropic anomalous cyclone of the negative MAS pattern around the Korean Peninsula in June 2014 being south of that in August 2014, and hence the severe drought over NEA in June 2014 being south of that in August 2014.

The other reason is that the mechanisms responsible for each month’s drought during summer 2014 were different. As revealed in Sect. 4, the severe droughts over NEA in June and August 2014 were closely related to the negative MAS pattern. The meridional teleconnections over the East Asia–WNP region were responsible for the severe drought in July 2014. The reduced rainfall over NEA caused by the negative MAS pattern in June and August was situated at around 35°N (Fig. 6a) and 40°N (Fig. 14a), respectively. The EAP-like teleconnection in July favored reduced rainfall at around 37°N over NEA (Fig. 11b). Therefore, the mechanisms responsible for each month’s drought being different partly led to the northward movement of the severe drought over NEA during summer 2014.

6 Summary and discussion

The month-to-month evolution of the severe drought over NEA during 2014 summer was addressed in this study. The role of the SST anomalies in the tropical Indo-Pacific region in the severe drought was also investigated. And finally, the reasons for the northward movement of the severe drought were explored.

The summer mean drought in summer 2014 occurred in the Huaihe River valley, North China, most of northeast China, and the Korean Peninsula. However, the severe drought experienced obvious northward movement from June to August 2014. The severe drought in July 2014 was similar to that of the summer mean, except that Japan also experienced drought. In June 2014, the drought belt was south of that in July 2014, with two drought centers in the Yangtze–Huaihe River valley and southern Korean Peninsula to southern Japan respectively. The drought belt in August 2014 moved northward relative to that in July 2014. Two drought centers were situated in southern parts of northeast China to the northern Korean Peninsula and Stanovoy Range, respectively. The severe drought for each month of summer 2014 was accompanied by weak lower-level summer monsoon flow. The mechanisms by which the SST anomalies in the tropical Indo-Pacific region contributed to each month’s drought over NEA during summer 2014 are summarized in Fig. 17.

Fig. 17
figure 17

Schematic diagram illustrating the mechanisms by which SST anomalies in the tropical Indo-Pacific region contributed to each month’s drought during summer 2014 over NEA: a June; b July; c August

Full size image

In June 2014, warming in the Arabian Sea led to reduced rainfall over northeast India and triggered the negative MAS pattern. The negative MAS pattern contributed to the quasi-barotropic anomalous cyclone over southern Japan and the anomalous anticyclonic circulation over the middle reach of the Yangtze River at the middle and upper levels. Warming SSTs in the South China Sea induced the local anomalous cyclonic circulation at the lower level. At the lower level, the anomalous cyclonic circulations over the south of Japan and the South China Sea reduced the lower-level summer monsoon flow over the drought belt in June 2014 and South China. This led to less water vapor transported to NEA, and thus favored drought over the region. At the upper level, the anomalous cyclonic circulation over southern Japan, combined with the anomalous anticyclonic circulation over the middle reach of the Yangtze River, resulted in anomalous high-level divergence over the drought belt in June 2014. The anomalous high-level divergence induced anomalous descent and the severe drought in June 2014. The physical processes described above are partly confirmed by the numerical experiments.

In July 2014, warming in the tropical WNP induced a strong WNPSM and increased rainfall over the Philippine Sea. Enhanced heating caused by the increased rainfall excited the positive EAP pattern, contributing to the anomalous cyclone over the South China Sea to the Philippine Sea and anomalous anticyclone over eastern China to Japan at the lower level. The anomalous cyclone and anticyclone led to anomalous northeasterlies over East Asia south of 40°N in July 2014. As a result, the lower-level summer monsoon flow weakened, less water vapor was transported to NEA, and hence severe drought occurred over there. These physical processes described above are basically reproduced by the numerical experiment. Meanwhile, warming in the eastern equatorial Pacific contributed to enhanced convection activities over the equatorial central and eastern Pacific. Enhanced heating corresponding to the enhanced convection activities led to increased rainfall over the off-equatorial western Pacific through the Gill pattern. The local warming also contributed to the increased rainfall. The increased rainfall triggered the EAP-like pattern, contributing to the anomalous quasi-barotropic cyclone over the Stanovoy Range. At the upper level, the strong WNPSM excited the anomalous anticyclone over northwest China and anomalous cyclone over the Yangtze–Yellow River valley, and the EAP-like pattern contributed to the anomalous cyclone over the Stanovoy Range. These circulation anomalies led to anomalous divergence at the upper level over the drought belt in July 2014, and then anomalous descent, inducing the severe drought over NEA in July 2014.

In August 2014, the negative IOD induced an anomalous meridional circulation, and warming in the equatorial eastern Pacific induced an anomalous zonal circulation. The anomalous meridional and zonal circulations led to anomalous convergence at the upper level over India to the tropical WNP and, subsequently, anomalous descent. As a result, the ISM and WNPSM weakened and then excited negative MAS and EAP patterns, respectively. The negative MAS pattern led to the quasi-barotropic anomalous cyclone around the Sea of Japan, and the negative EAP pattern contributed to the quasi-barotropic anomalous anticyclone over the Stanovoy Range. Thus, anomalous northeasterlies prevailed at the lower level, favoring the weak summer monsoon flow. Anomalous divergence appeared at the upper level over the drought belt in August 2014. These features then led to the severe drought over the region. The physical processes over the tropical Indian Ocean and western Pacific associated with the negative IOD are basically reproduced by the numerical experiment.

The monthly northward movement of the East Asian subtropical westerly jet during summer 2014 may partly explain the northward movement of the severe drought. The East Asian subtropical westerly jet axis in June 2014 was south of that in August 2014. This may lead to the quasi-barotropic anomalous cyclone of the negative MAS pattern around the Korean Peninsula in June 2014 being south of that in August 2014, and hence the severe drought in June 2014 being south of that in August 2014. The EAP-like pattern favored the severe drought at 37°N over NEA in July 2014, which was located in the middle of the drought belt caused by the negative MAS pattern in June and August 2014. Therefore, the fact that different mechanisms were responsible for the severe drought in each month during summer 2014 may also partly explain the northward movement of the severe drought.

Both extratropical signals and SST anomalies in the tropical Indo-Pacific region were important to the severe drought over NEA in summer 2014 (Wang and He 2015). The contribution of SST anomalies in the tropical Indo-Pacific region to the severe drought was addressed from the month-to-month perspective in this study. The main finding was that the SST anomalies responsible for each month’s drought during summer 2014 were different. This indicates that successfully depicting the month-to-month evolution of the air-sea interaction over the tropical Indo-Pacific region in the model is of great importance to improving the short-term climate prediction skill over the Asia–WNP region.

Remote effects can modulate the air-sea relationship over the WNP in summer (Wang et al. 2005; Wu and Kirtman 2007; Lu and Lu 2014). Lu and Lu (2014) suggested that remote effects play a crucial role in forming the negative SST–rainfall relationship over the WNP, while local SSTs tend to produce a positive SST–rainfall relationship and partially offset the remote effects. The rainfall and SST anomalies were out-of-phase over the tropical WNP in June and August 2014 (decreased rainfall, but warmer SST; cf. Figs. 2a, c, 3a, c), which indicates atmospheric circulation anomalies led to SST anomalies. However, in July 2014, the rainfall and SST anomalies were in-phase over the tropical WNP (increased rainfall and warmer SST; cf. Figs. 2b, 3b), which indicates SST anomalies led to atmospheric circulation anomalies. We propose the possible reason for the unstable SST–rainfall relationship over the tropical WNP during summer 2014 as follow. In June 2014, although the SST anomalies were positive in the tropical WNP, the convection was suppressed by the remote effects (the warming in east of New Guinea and off the equator in the central Pacific; Fig. 2a). The positive SST anomalies grew up due to increased downward shortwave radiative flux. In July 2014, the positive SST anomalies were strong enough to overcome the remote effects and hence induced anomalous ascent (Fig. 3b). Meanwhile, the positive SSA anomalies got weakened over the tropical WNP (Fig. 2b) due to decreased downward shortwave radiative flux. In August 2014, the positive SST anomalies was weak and hence the remote effects (the negative IOD and warming in the equatorial eastern Pacific; Fig. 2c) dominated again over the tropical WNP. Convection was suppressed again (Fig. 3c).