Using the NCEP–NCAR reanalysis and other observational datasets, the authors have investigated the relationship of summer rainfall variations between the Hetao region of northern China and the middle and lower reaches of Yangtze River (MLRYR). The results have demonstrated that rainfall in Hetao varies out of phase with that in MLRYR on the interannual time scales. This phenomenon is referred to as the Hetao–Yangtze rainfall seesaw (HYRS). An HYRS index is defined to reveal both spatial and temporal features of HYRS. It is found that the North Atlantic Oscillation (NAO) affects the HYRS. In years when the NAO is in its positive phase, anomalous divergences in the lower troposphere and anomalous convergences in the upper troposphere are observed in regions of the Mediterranean and eastern Europe. The anomalous convergences in the upper troposphere occur as the positive Rossby wave source excites a circumglobal teleconnection (CGT) in the midlatitudes, exhibiting the eastward propagation of Rossby wave energy along the Asian jet. Meanwhile, the Eurasian–Pacific (EUP) teleconnection also affects the HYRS. Influenced mainly by the CGT pattern, the circulations over Hetao and MLRYR are consequently perturbed. The atmosphere over Hetao converges anomalously in the lower troposphere and diverges anomalously in the upper troposphere, facilitating more than normal rainfall there. At the same time, the atmosphere over MLRYR diverges anomalously in the lower troposphere and converges anomalously in the upper troposphere, resulting in more than normal summer rainfall in MLRYR. In this way, the north–south rainfall seesaw is formed. This NAO-induced rainfall seesaw is potentially useful for summer rainfall predictions in both MLRYR and the Hetao region of northern China.
With a vast territory and complex climatic fluctuations, China is affected by extreme weather and climatic events on different spatial and temporal scales (Zhai et al. 2005; Hou and Guan 2013; Jin et al. 2015; Li et al. 2016). As is well known, eastern China is the most economically developed and densely populated area of China, with a geographical concentration of large cities. Within this large area, the Hetao area, an important livestock and heavy industry area, is situated in the northern part of eastern China, and the middle and lower reaches of the Yangtze River (MLRYR), a main crop production area, is located in the southern part of the eastern China. Summer flooding and drought have significant impacts on the crop and livestock production, as well as the daily lives of inhabitants in these areas. Therefore, to explore the features and underlying mechanisms of summer precipitation variability in eastern China, especially in the Hetao region and MLRYR, is of great societal and economic importance.
Summer precipitation in eastern China exhibits distinct local characteristics. Wei and Zhang (1988) claimed that there are three typical precipitation patterns: pattern I has a major rainbelt located in the Yellow River basin and the area to its north; the abundant precipitation region of pattern II is situated between the Yellow River and the Yangtze River; and the excessive precipitation band of pattern III is found along the Yangtze River and the area to its south. With above-normal rainfall in these regions, there is below-normal precipitation in other areas. However, the regional features of anomalous rainfall distributions are far more complicated than the three types of anomalous precipitation patterns.
The rainfall in different region varies differently and is influenced by different factors (e.g., Jin et al. 2015). In the Hetao region, summer precipitation is possibly influenced by many factors, such as the Indian summer monsoon (Zhang et al. 1999; Feng and Hu 2004) and El Niño–Southern Oscillation (ENSO) (Huang and Wu 1989; Zhang et al. 1999; Lau and Weng 2001; Wu and Wang 2002; Feng and Hu 2004). Intensification of the high pressure ridge in the middle troposphere over northeast Asia may strengthen the north wind in eastern China, inducing water vapor transport to Hetao to be suppressed, and thus leading to precipitation deficits and drought there (Lei and Duan 2011). In southern China, the activity of the East Asian summer monsoon is found to play a dominant role in the formation and variations of the rain belt there (Tao and Chen 1987; Wang and LinHo 2002; Ding 2004). Summer precipitation in the MLRYR is closely associated with the western Pacific subtropical high (WPSH) (He et al. 2001). Convective activity around the Philippines influences the occurrences of droughts and floods in the Jianghuai basin (Huang and Li 1987; Nitta 1987). More than this, summer precipitation in MLRYR is also closely related to some remote forcings such as ENSO (Huang and Wu 1989; Zhang et al. 1999; Lau and Weng 2001; Wang and Gu 2016), the Arctic Oscillation (AO) (Thompson and Wallace 1998) in May (Gong and Ho 2003), and the Antarctic Oscillation (AAO; Gong and Wang 1999) in April–May (Nan and Li 2003).
It seems that summer precipitation anomalies in the Hetao region and MLRYR vary differently in association with different circulation patterns. However, in some decades, precipitation is found to be below normal in the Hetao region (Feng and Hu 2004; Zhai et al. 2005) but above normal in MLRYR (Zhai et al. 2005). This is in agreement with the aforementioned three precipitation patterns proposed by Wei and Zhang (1988). Furthermore, in some other studies (e.g., Hsu and Lin 2007; Jin et al. 2015), the variations of summer precipitation anomalies in Hetao and MLRYR are also suggested to be linked by some dynamical mechanisms. In fact, there may exist an out-of-phase relation of anomalous summer precipitation between southern and northern China on interannual time scales (Huang 2004; Ding and Wang 2005; Zhai et al. 2005; Hsu and Lin 2007; Huang et al. 2012; Jin et al. 2015). Also, this antiphase rainfall relation between Hetao and MLRYR may exist on interdecadal time scales (Jin et al. 2016).
There are several physical mechanisms that possibly explain the out-of-phase relationship of rainfall anomalies in eastern China. First, Yu et al. (2004) suggested influence of the upper-tropospheric cooling over East Asia in summer, causing the westerly jet to move southward and weakening the East Asian summer monsoon, and leading to floods in the south and drought in the north. Second, Huang (2004) argued that the East Asia–Pacific (EAP) teleconnection contributed to the southern floods and northern droughts. Third, the circumglobal teleconnection (CGT) may play an important role in inducing the north–south antiphase relations in rainfall anomalies. The CGT is a zonal atmospheric teleconnection (Branstator 2002; Ding and Wang 2005) that presents in the middle latitudes in both boreal winter and summer. This CGT is formed mainly due to Rossby wave propagations along the westerly jet that acts as a waveguide (Hoskins and Ambrizzi 1993). Ding and Wang (2005) claimed that during positive CGT (Branstator 2002) index years, positive precipitation anomalies and negative precipitation anomalies occurred in northern China and the Yangtze River basin, respectively. Opposite situations occurred in negative CGT index years. This demonstrates that the CGT has different possible effects on summer precipitation in different regions of eastern China. Enomoto et al. (2003) found that decent in the Mediterranean region excited the Silk Road teleconnection pattern, which can influence the Bonin high and hence affect the East Asian climate. The Silk Road teleconnection can be partly considered as the CGT in Eurasia (Ding and Wang 2005). Fourthly, Watanabe (2004) found that during positive NAO periods in boreal winter, the anomalous divergence located over the Mediterranean triggers the quasi-stationary wave train propagating downstream along the Asian jet, which exerts influences on East Asian climate. Sun et al. (2008) claimed that the summer North Atlantic Oscillation (NAO; Walker and Bliss 1932; Hurrell 1995) anomaly causes divergence and convergence anomalies in the Asian jet entrance, which stimulate a zonal Rossby wave train that propagates eastward and affects the summer temperature anomaly in East Asia.
The aforementioned studies suggest a possible out-of-phase relationship in the interannual variations of summer precipitation between MLRYR and the Hetao region. If the out-of-phase relationship of anomalous rainfall indeed exits between Hetao and MLRYR, what will features of the phenomenon look like on interannual time scales? What are the mechanisms behind this phenomenon? Obviously, answering these questions would enable better understanding of the interannual variations of summer rainfall in China and the East Asian summer monsoon activities. Therefore, we investigate the out-of-phase relationship of summer rainfall variations between Hetao and the middle and lower reaches of Yangtze River and its possible mechanisms.
The present paper is organized as follows. After a discussion of the motivations for this study, brief descriptions of the data we used are presented in section 2. In section 3, we reveal the phenomenon of the north–south out-of-phase relationship of summer precipitation between the Hetao region and MLRYR. In section 4, the mechanisms behind this antiphase relation are investigated, and finally a summary is presented.
2. Data and methodology
The datasets we used are the NCEP–NCAR reanalysis with a horizontal resolution of 2.5° × 2.5° (Kalnay et al. 1996); the monthly sea surface temperature from the NOAA Extended Reconstructed Sea Surface Temperature (ERSSTv3) with horizontal resolution of 2° × 2° (Smith et al. 2008); the daily precipitation data at 25 stations in the Hetao region and 31 stations in MLRYR out of 753 stations in China, which are retrieved from the National Climate Center of China Meteorological Administration; and NOAA’s Precipitation Reconstruction (PREC) data with horizontal resolution of 2.5° × 2.5° (Chen et al. 2002). All these datasets cover the period of 1961–2015.
Summer months in the present work are June–August (JJA). The Hetao region (35°–42°N, 105°–115°E) and MLRYR region (25°–35°N, 110°–122.5°E) are two key study regions in the present work, which are identified based on the results reported in Jin et al. (2015).
The NAO index is derived from the NOAA Climate Prediction Center (http://www.esrl.noaa.gov/psd/data/correlation/nao.data). Because we are focusing mainly on the interannual variability, the data are detrended and variations longer than 11 years have been removed.
3. The Hetao–MLRYR rainfall seesaw
Climatologically, summer precipitation varies nonuniformly in spatial extent from MLRYR to the Hetao region (Jin et al. 2015). It is seen from Fig. 1a that summer precipitation is lower than 320 mm at all stations in the Hetao region, with a minimum lower than 120 mm, but is higher than 360 mm at all MLRYR stations, with the maximum higher than 600 mm. This indicates that the Hetao region is relatively dry in summer while MLRYR is wet (Zhai et al. 2005; Jin et al. 2015).
Year-to-year changes of summer precipitation in the Hetao region are large but not as large as in MLRYR. In Fig. 1b, it is seen that the standard deviations of JJA rainfall are distributed nonuniformly in spatial extent, with values smaller than 120 mm at all stations in the Hetao region. The minimum value is smaller than 40 mm in this region. However, the standard deviation is bigger than 100 mm at all MLRYR stations with the maximum value above 260 mm. This suggests that summer precipitation varies with larger interannual variability in MLRYR than in the Hetao region.
Previous studies showed that the rainfall variability between MLRYR and the Hetao region has some connections on decadal time scales (Hu 1997; Zhai et al. 2005; Ding et al. 2008; Zhu et al. 2011; Jin et al. 2016). Recently, Jin et al. (2015) performed an empirical orthogonal function (EOF) analysis of summer precipitation anomalies in China, demonstrating that the significant negative correlations between northern China and MLRYR are found in summer precipitation variations. Based on Jin et al. (2015), in situ observational precipitation data at 25 stations in the Hetao region and 31 stations in MLRYR (Fig. 1) are selected for further analysis.
There exist, indeed, high negative correlations between JJA precipitation in Hetao and MLRYR. Here we calculate the correlation coefficients between anomalous rainfall at each of the 25 stations in Hetao region and at each of the 31 stations in MLRYR and find it to be −0.65, which indicates a possible close connection of precipitation between Hetao and MLRYR.
To examine the antiphase relation between Hetao and MLRYR more easily, an interannual rainfall seesaw index is defined as follows. Let be the normalized times series of preprocessed anomalous rainfall averaged over 25 (31) stations in the Hetao region (MLRYR). Then the index of rainfall seesaw between Hetao region and MLRYR is written as
Obviously, when the index is positive, the Hetao region tends to receive more than normal rainfall, or not to be as dry as in MLRYR. When is negative, the situation is opposite. The time series of index and its correlations with station rainfall are shown in Fig. 2.
The anomalous rainfall in the Hetao region varies apparently in antiphase to that in MLRYR; the correlation coefficient of with is found to be −0.65. In this case, is highly correlated with with a correlation coefficient of 0.90 (−0.90). This indicates that well captures the summer precipitation covariation in the Hetao region and MLRYR. Furthermore, because index explains 81% and 81% of total variance of summer precipitation in the Hetao region and in MLRYR respectively, it is useful for describing the relative changes in rainfall anomalies in these two regions. The good out-of-phase relationship between and can also be observed in Fig. 2a. The power spectra of show some periodicities in this out-of-phase relationships between and . The most prominent periods are 2–3 yr (Fig. 2b).
The spatial structure of the north–south seesaw in anomalous summer rainfall between Hetao and MLRYR can be explored by the correlation pattern of with anomalous summer precipitation from the 596 stations over China. It is seen from Fig. 2c that large significant positive correlations are observed in northern China while large negative correlations occur in MLRYR, indicating oppositely signed summer rainfall anomalies between these two regions. This scenario can also be found in Fig. 2d, which shows similar patterns of correlation of with anomalous summer precipitation from NOAA PREC data for verification. Based on the results above, we refer to the out-of-phase relationship of anomalous summer rainfall between the Hetao region and MLRYR as the Hetao–Yangtze rainfall seesaw (HYRS) during boreal summer.
It should be noted that the HYRS is, to a certain extent, related to type III, which is also known as the type-II rainfall pattern as discussed in Wei and Zhang (1988), who perform empirical orthogonal function (EOF) analysis. However, the HYRS is largely different from the type-II rainfall pattern because the variance explained by the HYRS index is explicitly larger than that explained by the triple pattern (Wei and Zhang 1988) over Hetao and MLRYR, and the correlations of IHY with rainfall anomalies over southern China are much weaker than those of type-II rainfall pattern–related principal mode with rainfall anomalies over the same region.
4. Circulation anomalies associated with HYRS
To explore mechanism behind HYRS, composite analysis is performed of circulation anomalies. To do this, years when the absolute value of the normalized index is larger than 1 are selected (listed in Table 1). The years when are designated as type-A years, whereas for type-B years . It is seen from Table 1 that, out of 55 years, we have 11 type-A years and 10 type-B years. In type-A years, there occur events with more precipitation in the north and less in the south, whereas in type-B years the events occur with less precipitation in the north and more in the south.
a. Anomalous water vapor transport
The anomalous water vapor transport in shows different features in the Hetao region and MLRYR in years of different types. Figure 3a shows anomalous vapor transport whereas in Fig. 3b anomalous circulation at 850 hPa is presented for reference as it looks similar to Fig. 3a. In type-A years, it can be seen from Fig. 3a that the most striking feature is the significant anticyclonic circulation of vapor transport centered over the Korean Peninsula. The Hetao region and MLRYR are located just at the northern and southern flanks, respectively, of this anomalous anticyclonic circulation. In the Hetao region, water vapor comes from the region south of the Yellow River Valley northeast and from Lake Baikal. However, things are different in MLRYR. Obviously, an anomalous cyclonic circulation of vapor transport is observed over large area of southern China, the South China Sea, and the northwest Pacific (Fig. 3a). Water vapor is anomalously transported to MLRYR via the Yellow Sea from the western Pacific and is then separated into two portions. One is transported farther westward to the vicinity of 30°N, 100°E and then turns southward to reach the Bay of Bengal through the Indochina Peninsula. The other turns northeastward at the vicinity of 30°N, 110°E and then is transported into the Hetao region. The water vapor fluxes diverge over MLRYR whereas they converge over the Hetao region, leading to abundant precipitation in the Hetao region but deficient rainfall in the Yangtze River basin. Similar patterns can also be observed in sea level pressure anomalies (SLPAs) and circulation anomalies at 850 hPa (Fig. 3b). In type-B years, the scenarios look opposite.
Note that the anomalous cyclonic circulation of vapor fluxes in lower latitudes elongates southwestward from the subtropical western Pacific to southern China, weakening the monsoonal transport of water vapor to MLRYR from the Bay of Bengal and the South China Sea (e.g., Guan and Yamagata 2003). More than this, this anomalous cyclonic circulation along with the anticyclonic circulations respectively in both its southern and northern flanks mimics the wave train structure of the East Asia–Pacific/Pacific–Japan teleconnection pattern (EAP/PJ; Huang and Li 1987; Nitta 1987; Huang 2004). Whether the EAP/PJ occurs or not in the HYRS is still not clear. Therefore, the anomalous anticyclonic circulation over the Yellow Sea and the Korean Peninsula appears to play the key role in HYRS of summer rainfall.
b. Relation with the NAO and teleconnections in midlatitudes
Some studies have shown that the NAO may modulate climate anomalies in East Asia (Sun et al. 2008; Wu et al. 2009; Wu et al. 2011; Guan and Jin 2013; Jin et al. 2013; Wang et al. 2017), as may the circumglobal teleconnection (Ding and Wang 2005). These suggest that the HYRS may be influenced by the NAO and CGT. To clarify this, we calculate the correlations of index with other indices including the NAO, CGT, and AO (Table 2). It is clearly seen that the correlation coefficient between and the NAO index is as large as 0.45 (Table 2), suggesting that during positive NAO phase years summer precipitation occurs more in northern China and less in MLRYR and vice versa. As the NAO is considered to be another paradigm of the AO in the North Atlantic (Wallace 2000), despite some differences (Ambaum et al. 2001; Wallace and Thompson 2002; Watanabe 2004), correlations of with the AO index (Table 2) are also significant at 95% level of confidence. At the same time, we also find significant correlation between the indexes of HYRS and CGT, implying a relation of HYRS with circumglobal teleconnection.
1) Circulation anomalies related to the NAO
The relations of the HYRS with the NAO can be further examined by examining the -related spatial patterns of sea level pressure anomalies and sea surface temperature anomalies (SSTAs). Figure 4 shows the simultaneous correlations of with SLPAs and with SSTAs over the North Atlantic. For comparison, the correlations of the NAO index with SLPAs and /or SSTAs are also presented. It is clearly seen from Fig. 4a that in the North Atlantic, correlation is positive to the south of 60°N and the large value area is centered at the eastern edge of the North American continent, with the highest coefficient greater than 0.50. In contrast, the correlation is negative to the north of 60°N and the large value areas are located in Baffin Island and Greenland, with negative coefficients stronger than −0.30. This correlation pattern looks very similar to the NAO structure (Fig. 4b) although the latter is stronger; the pattern correlation between Figs. 4a and 4b for the North Atlantic is found to be 0.94, indicating a close relationship between the HYRS and the NAO.
The connections of HYRS can be further verified by the correlations between and sea surface temperature anomalies in the North Atlantic. It is observed in Fig. 4c that there is a south-to-north tripole spatial distribution with signs of coefficients as in a negative–positive–negative pattern from the east edge of the North American continent to Greenland. Similarly, this tripole pattern can also been found in correlations of the NAO index with the North Atlantic SSTA (Fig. 4d). The pattern correlation between Figs. 4c and 4d is 0.82. Although the magnitudes of correlations are more or less variable, the signatures of the NAO-related SSTAs resemble each other in Figs. 4c and 4d. Since the tripole SSTA is induced by the NAO (Pan 2005), the variations of summer rainfall in Hetao and MLRYR must be linked to the NAO based on what we obtain from Table 2 and Fig. 4.
2) Relationship with teleconnections in midlatitudes
As mentioned above, the HYRS is possibly and remotely linked to the NAO, and may be attributable to the CGT. First, the CGT index (ICGT), which is defined by Ding and Wang (2005), has a significant correlation with HYRS index with a correlation coefficient of 0.43 (Table 2). Second, it is known that the CGT is a wave train propagating eastward along the midlatitude westerly jet (Ding and Wang 2005), stimulated by the NAO and stably existing in summer and winter (Watanabe 2004; Sun and Wang 2012).
To examine the role of CGT in the HYRS, composite differences in anomalous circulations are performed, yielding Fig. 5. From Fig. 5a, we can see that there are two positive anomalous geopotential height centers in the upper troposphere over the Pamir (37.5°N, 70°E) and northeastern China as well as the Korean Peninsula, in agreement with Hong and Lu (2016). Meanwhile, a negative center is observed near the greater Caucasus Mountains to the northwest of the Caspian Sea. These anomaly centers display the wave train structure of the CGT in midlatitude Eurasia (Fig. 5a). At 500 hPa, the wave train structure is also correspondingly seen in geopotential height anomalies (Fig. 5b) except for weaker intensities. It is believed that the positive height anomaly center over the Pamir is stimulated by the Indian summer monsoon and contributes to the downstream propagation of the CGT wave train. Hence the Pamir is considered as a key area of the circumglobal teleconnection (Ding and Wang 2005; Behera et al. 2013).
Note that there appear to be a strong positive anomaly center of geopotential height in northern Europe, an anomalously negative center around Lake Baikal, and a positive anomaly center around the Korean Peninsula (Figs. 5a,b). These alternatively signed anomalous centers from northern Europe to eastern China indicate a possible route of the wave energy propagation, which looks like the Eurasian–Pacific (EUP) teleconnection pattern (Wallace and Gutzler 1981). The EUP pattern was originally discovered in Northern Hemispheric winter at 500 hPa. For boreal summer, this Eurasian–Pacific teleconnection was also discussed later (e.g., Yang 1992; Zhu and Shi 1993) but the definitions of the EUP teleconnection indices are a little different from that in the study of Wallace and Gutzler (1981). Here we use both the EUP index proposed by Yang (1992) and that by Zhu and Shi (1993) to investigate the relations among summertime NAO, EUP, and HYRS. Let Z* denote the normalized time series of geopotential height anomalies at 500 hPa and let EUPY stand for the index proposed Yang (1992) and EUPZS for the index proposed by Zhu and Shi (1993). Then these indices are respectively expressed as
It is found that the NAO is highly correlated with the EUPY with a correlation coefficient of 0.60 whereas it correlates with EUPZS with a correlation of 0.40. On the other hand, EUPY and EUPZS are respectively correlated with HYRS index with correlation coefficients of 0.37 and 0.25. These correlations suggest that the NAO may affect the HYRS via the EUP teleconnection.
To better understand the role of CGT in connecting NAO to HYRS, it is necessary to examine the quasi-stationary Rossby waves in the extratropics. As is known, the upper-tropospheric meridional wind is a good quantity to describe the zonal teleconnections (Lu et al. 2002; Watanabe 2004). Figure 6 shows the -regressed coefficients of anomalous meridional winds.
The HYRS-related Rossby waves propagate clearly in zonal around 40°N in both the middle and upper troposphere (Fig. 6). As shown in Fig. 6a, there is a negative meridional wind anomaly center at the Atlantic jet exit and a positive meridional wind anomaly zone at the Asian jet entrance, with its centers locating over the Azores and Caspian Sea. They form a negative–positive–negative wave train structure over the Tarim Basin, northern China, and the Japanese archipelago. Almost all these anomaly centers distribute sparsely along the jet stream (Fig. 6a). This is, to a certain extent, similar to the teleconnection pattern propagating along the jet stream from North Africa to East Asia described by Lu et al. (2002), which is also considered as a manifestation of the CGT in Eurasia (Ding and Wang 2005).
The HYRS index is found to be significantly related to the NAO. The CGT seems to operate as a linkage between NAO and HYRS. The index of the CGT is significantly correlated with the NAO index with a correlation coefficient of 0.27. More than this, Fig. 6 suggests that this CGT in association with HYRS is excited by the NAO at the entrance of the Asian jet. This linkage is needed to be further examined physically in two aspects: 1) the wave activity fluxes that display the wave propagations and 2) the Rossby wave sources induced by anomalous divergence.
(i) Wave activity fluxes
Because the direction of the group velocity of the quasi-stationary Rossby waves parallels that of the wave activity fluxes (WAFs; Takaya and Nakamura 2001), the WAFs are diagnosed to clarify the wave energy propagations, and hence to further explore the relationships between the CGT and HYRS, and between the NAO and CGT (Fig. 7).
In the significant anomalous rainfall years, the quasi-stationary Rossby waves disperse eastward along two ways, which can be clearly identified by the WAFs as seen in Fig. 7a. The first is from northeastern Europe southeastward to East Asia; the other is from the vicinity of the Mediterranean eastward into eastern China. The former appears to be related to the EUP teleconnection and the latter to the CGT. Interestingly, Rossby wave energy propagates eastward along both two paths and eventually converges into area around the Korean Peninsula, facilitating the anomalous circulation to maintain there. Clearly related to the later path of wave energy dispersion, the Asian jet acts as the waveguide (Hoskins and Ambrizzi 1993) and manifests the CGT pattern (Ding and Wang 2005). Along the Asian jet, the WAFs in zonal–vertical section also show strong propagations of Rossby wave energy from the jet entrance all the way eastward into East Asia. It is noticed that some WAFs propagate upward and converge into anomalous negative vorticity regions in the middle and upper troposphere (Fig. 7b) although the WAFs near the Tibetan Plateau are artificially amplified due to possible data error there. These energy inputs into upper tropospheric jet stream are very important to maintain the stationary waves; otherwise, they eventually decay. Because of the propagations of quasi-stationary waves, the downstream climate conditions are affected as discussed elsewhere (e.g., Enomoto et al. 2003; Guan and Yamagata 2003).
As mentioned above, the NAO may modulate the downstream climate through the CGT along the axis of the Asian jet. This connection of the NAO to the CGT is observed from WAFs in the region around the North Atlantic. It is seen that the zonal components of the WAFs in Fig. 7b demonstrate that the wave energy propagates into the entrance area of the Asian jet stream, indicating that the Asian jet at 200 hPa is perturbed by the NAO. More than this, NAO-related perturbations also disperse energy westerly in the higher latitudes (Fig. 7a), leading to energy propagation over northern Europe where the EUP (Wallace and Gutzler 1981) wave train begins. These connections between the CGT and NAO show why the HYRS is influenced by NAO.
Note that there is also meridional propagation of Rossby wave energy in East Asia. The WAFs display a southward dispersion of energy from northern China and the Korean Peninsula (Fig. 7a) to southern China. It is known that the EAP/PJ pattern is usually observed in East Asia in the lower troposphere during boreal summer (Huang 2004; Kosaka and Nakamura 2006). This EAP pattern also appears to be found in Fig. 3a. For the EAP/PJ wave train, it has a signature of the northward propagation of energy once this perturbation is excited by the anomalous tropical forcing. However, the southward propagations of Rossby wave energy at 200 hPa as denoted by WAFs (Fig. 7a) are explicitly contradictory to the energy propagation direction of EAP/PJ. Therefore, whether or not there exists an EAP/PJ pattern in East Asia during the HYRS events deserves further investigation.
(ii) Rossby wave sources
The Rossby waves are generated by the wave sources. Out of different kinds of Rossby wave sources (RWS) the divergent flow is the most important component in a linear system. In years when the HYRS index is in its positive phase, the Azores high pressure, especially its western part, is intensified as displayed in Fig. 4a. This intensified high pressure extends all the way to the Mediterranean. Correspondingly, an anomalous decent of air is induced due to Ekman pumping in this anomalous high pressure zone (Watanabe 2004; Sun et al. 2008). Therefore, divergent flows occur in the lower troposphere as observed in Fig. 8a, with the divergence zone covering the North Atlantic, Europe, and the Mediterranean. To compensate for the air mass outflow in the lower troposphere, a corresponding convergence zone is present in the upper troposphere at 200 hPa, with three convergence centers (Fig. 8b) located over the Mediterranean, eastern Europe, and the North Atlantic, respectively. These anomalous convergent flows at 200 hPa may excite the anomalous cyclonic circulations (Rodwell and Hoskins 1996) that appear in regions northwest of the convergent centers (Sardeshmukh and Hoskins 1988). When the HYRS index is in its negative phase, the scenarios look the opposite.
Note that, during positive phase of , an apparent convergent center is over the Maritime Continent (MC) region (Ramage 1968) at 850 hPa and a divergent center over 200 hPa, indicating that there exists relationships between climate anomalies in eastern China and the MC region. These scenarios will be oppositely signed when in the negative phase of .
To depict RWS more clearly, we calculate the RWS term in the potential vorticity (PV) equation (Sardeshmukh and Hoskins 1988), which is written as
The significant anomalous Rossby wave sources and sinks exist mainly along the Asian jet stream. The convergence anomaly in the upper troposphere around the Mediterranean (Fig. 8b) generates a positive Rossby wave source anomaly there (Fig. 9) and excites Rossby waves propagated eastward along the waveguide (Sardeshmukh and Hoskins 1988; Watanabe 2004; Wu et al. 2009; Jin et al. 2013). This eastern European region with a positive RWS anomaly indicates that the region is a key area for the connection between the NAO and CGT. Of course, during the eastward propagations of Rossby waves, the Rossby wave sources and sinks are sure to modulate the intensities of anomalous cyclonic and anticyclonic circulations all the way from the entrance of the Asian jet to the region around Hetao and the Yangtze River.
The relationships of summer precipitation between the Hetao region and MLRYR, as well as its relationship with the NAO, have been investigated based on observational and reanalysis data. The main results are summarized as follows.
Interannual variations of summer precipitation in the eastern part of China demonstrate a striking out-of-phase relationship between the Hetao region and MLRYR; the rainfall anomalies averaged over Hetao are highly and negatively correlated with those over MLRYR, with a correlation coefficient of −0.63 over the period 1961–2015. This out-of-phase relation in summer rainfall anomalies is referred to here as the HYRS (Hetao–Yangtze rainfall seesaw). An HYRS index is defined to better describe this phenomenon and to quantify the intensity of this seesaw in summer rainfall anomalies.
The anomalous circulation and water vapor transports over East Asia facilitate the formation of the HYRS. Also, the anomalous anticyclonic circulation in the lower troposphere between the Yellow and Yangtze Rivers plays a crucial role in formation of the HYRS. When the HYRS index is in its positive phase, the water vapor converges toward the Hetao region from the western Pacific through MLRYR whereas it diverges over MLRYR, leading to greater than normal summer precipitation in the north and less than normal summer precipitation in the south. The opposite occurs when the HYRS index is in its negative phase.
The HYRS is affected by both the CGT and waves from northeast Europe in association with the EUP. With respect to the CGT, it is found that the quasi-stationary Rossby waves propagate along the Asian jet stream from the entrance of the jet eastward to eastern China. The Asian jet works as a waveguide. The wave sources also play a role in maintaining the anomalous circulations over East Asia. With respect to the EUP, the wave activity fluxes also show wave energy dispersing from northeast Europe along the ray path southeastward to the Hetao and MLRYR regions, leading to and maintaining the anomalous circulation in eastern China.
The HYRS is found to be closely related to the NAO; the correlation of the HYRS index with the NAO index is 0.45. On the other hand, the composite of geopotential height in the middle to upper levels between positive and negative HYRS index years reveals that the HTRS is also related to the EUP teleconnection. This suggests that the NAO may influence the HYRS via both CGT and EUP teleconnections.
In positive HYRS index years, the NAO is relatively stronger. The westerly jet over North Atlantic is intensified. The wave energy may propagate eastward into the entrance region of the Asian jet and northeastern Europe. When the NAO is strong, the Azores high over the North Atlantic is intensified, inducing stronger than normal divergence in the lower troposphere. The corresponding convergent flows in the upper troposphere induce the Rossby wave sources near the entrance of Asian jet stream, inducing the Rossby wave to be generated. The generated Rossby waves propagate along the Asian jet all the way into eastern China, inducing the rainfall anomalies in both the Hetao and MLRYR regions, resulting in the HYRS.
Previous studies have shown that spring NAO has an impact on the East Asian summer monsoon (EASM), thereby affecting the East Asian climate (Wu et al. 2009). In the present study we demonstrate that the NAO has a simultaneous relation with the CGT via perturbations over eastern Europe and the Mediterranean, thereby affecting the HYRS. The CGT seems to serve as a bridge in the relationship between the NAO and HYRS. Since the CGT also varies on intraseasonal time scales (Branstator 2002; Watanabe 2004), can we predict the HYRS by investigating the low-frequency oscillation of the CGT? This problem remains to be explored.
The NAO can excite a Rossby wave train propagating eastward along the jet that acts as a waveguide (Watanabe 2004). Ding and Wang (2005) proposed two possible mechanisms for the excitation of the CGT, one involving an anomalous disturbance around the exit of the North Atlantic jet and the other an abnormal Indian summer monsoon. A large number of studies have demonstrated that the Indian summer monsoon is closely related to summer precipitation in northern China (Zhang et al. 1999; Feng and Hu 2004). The Indian summer monsoon not only serves as a bridge in the influence of ENSO on summer precipitation in northern China (Feng and Hu 2004) but also impacts the relationship between the ENSO and CGT (Ding and Wang 2005). Determining the role that the Indian summer monsoon and CGT play between ENSO and the out-of-phase relationship of summer precipitation between the north and south requires further investigation.
The authors are very grateful to the anonymous reviewers for their helpful comments. This research is jointly supported by the National Natural Science Foundation of China (Grant 41330425), the Natural Science Foundation of Jiangsu Province of China (BK20160956), the Natural Science Foundation of China (41575081), the Specialized Research Fund for the Doctoral Program of Higher Education (Grant 2103322811001), and the PAPD project of Jiangsu Province. Dachao Jin is also supported by the Startup Foundation for Introducing Talent of NUIST (Grants 2014r006 and 2015r035).