China is located in the Asian-Australian monsoonregion. Its weather and climate are greatly affectedby monsoon systems, especially during boreal summer(June, July, and August). In this season, both theprecipitation amount and the main rainfall belt arecontrolled by the summer monsoon systems. Amongthe monsoon circulations, the western Pacific subtropical high(WPSH), the East Asian summer monsoon(EASM), and the blocking high in higher latitudes arethought to be the key influencing factors.

Early in the 1950s, Chinese meteorologists beganto explore the relationship between weather in China and the EASM systems(Huang, 1955; Chen, 1957; Tao and Chen, 1957; Ye and Zhu, 1958; Ye et al., 1959; Huang and Yu, 1961; Tao and Xu, 1962; Tao et al., 1962; Wang, 1962). Their results have been widelyapplied and referenced in both operational forecasting and scientific research. Since then, increasing attention has been paid to the EASM structure, its multitimescale variations, its influencing elements, and itscontribution to weather and climate of China, EastAsia, and even the world(Chen et al., 1991; Zhao, 1999; Wu et al., 2002).

Zhou et al. (1964)analyzed the relationship between the WPSH and the beginning/ending dates ofMeiyu along the Y angtze River valley. Based on the5880-gpm contour at 500 hPa, Zhao(1999)defined fiveWPSH indices for objective analysis of the WPSH. i. e., the area index, the intensity index, the ridge index, the northern boundary index, and the western boundary index. Based on these indices, it is revealed thatboth the position and intensity of the WPSH bear asignificant relation to the Europe-Asia circulation pattern(Zhao, 1999). A more southward WPSH alwayscorresponds to a positive-negative-positive pattern ofthe 500-hPa geopotential height anomaly from lowerto higher latitudes in East Asia(Huang, 1992; Huang and Y an, 1999). Under this pattern, more summerprecipitation will occur along the Y angtze River. Onthe contrary, if the WPSH is located further northward, drought will appear along the Y angtze River valley while floods occur in northern and southern China(Zhang, 1999; Zhao, 1999; Chen and Zhao, 2000; He et al., 2001; Wu et al., 2002; Zhang et al., 2003b). Arecent study found that the WPSH has a close relationship with summer rainfall in Japan as well(Nagio and Takahashi, 2012).

Research on the EASM and its impacts on summer rainfall has a long history in China. Zhu(1934)innovatively analyzed the southeasterly monsoon and its influence on rainfall. Similar studies can be foundin Tu and Huang(1944), Tao and Chen(1957), and Ye and Zhu(1958). Among all the circulation systems in-fluencing heavy rainfall in Huaihe River and its south, the monsoon surge plays the most important role(Tao and Wei, 2007). The beginning dates of the rainy season in East Asian subtropical monsoon regions(China and Japan)are earlier than the dates in East Asiantropical monsoon regions(South China Sea and western Pacific)(Zhu et al., 1986; He et al., 2008). The interaction between the two monsoon systems causes theflood or drought in different areas(He et al., 2008). T obetter underst and the interannual variation of EASM and its correspondence to summer precipitation, manyEASM indices have been defined with focuses on different circulation systems(Guo, 1983; Shi et al., 1996;Zhao and Zhang, 1996; Sun et al., 2002; Zhang et al., 2003a). The results indicate that the summer rainfall from South to North China is most directly influenced by the EASM. A stronger EASM will result inmore rainfall in southern and northern China, whilea weaker monsoon will bring more rainfall along theY angtze and Huaihe River valleys.

Besides the tropical and subtropical circulationsystems, the blocking in higher latitudes also plays animportant role in occurrences of drought and flood inChina(Zhao, 1999; Chen and Zhao, 2000), as bothits frequency and intensity directly affect the strength and location of cold air intrusion. Persistent blockingsituations cause decreased rainfall in North China and more rainfall along the Y angtze River(Zhao, 1999;Chen and Zhao, 2000).

In summary, the EASM circulations largely influence the summer rainfall in China. Their close relationships have been widely applied in operational prediction of short-term climate in China(Zhao, 1999;Chen and Zhao, 2000). In traditional statistical forecast models, a La Niña event(i. e., lower SSTs in theequatorial eastern Pacific)is assumed to cause a weakened and more northward WPSH and a stronger summer monsoon, and finally brings floods to northernChina and drought along the Y angtze River; on theother h and , a stronger and more southward WPSH, weaker monsoon systems, and more active blockinghighs will together induce floods along the Y angtzeRiver.

The results above have been used in climate prediction in China for many years. For example, basedon the relations among the WPSH, the EASM, and the blocking activities, the extreme flood event alongthe Y angtze River in 1998 was successfully forecasted(National Climate Center, 1998). Nonetheless, the relationships among these systems have also been foundunstable and nonlinear. In 1999, the WPSH wasweaker than normal, but flood events also occurredalong the Y angtze River, which is not consistent withthe scenario mentioned above. Detailed examinationof the cases in the past two decades indicates that theinconsistency appeared not only in 1999.

It is well known that interdecadal changes occurred in both the ocean and the atmosphere in the1970s. Afterwards, the EASM weakened while the WPSH strengthened(Han and Wang, 2007). Under this background, both the relationship between ElNiño-Southern Oscillation(ENSO) and EASM, and the relationship between ENSO and summer rainfallin China, weakened(Wang, 2001; Gao et al., 2006;Wang and He, 2012). Have the relationships betweenthe EASM/WPSH/blocking and summer rainfall alsochanged? If the changes have indeed occurred, whatare their implications for the traditional statistical climate prediction? In the following sections, these questions will be addressed. 2. Data

The observed monthly rainfall amounts at 743Chinese stations are provided by the China Meteorological Administration. For data quality, only thestations for which the data are complete during 1961-2010 are selected for further study. In this paper, the representative stations for North China and theY angtze River valley are the same as those in Chen and Zhao(2000). The atmospheric variables are obtained from the NCEP/NCAR reanalysis(Kalnay et al., 1996) and include 500-hPa geopotential height and 850-hPa horizontal wind on a 2. 5° × 2. 5°latitudelongitude resolution(http://www. esrl. noaa. gov/psd/data/gridded/data. ncep. reanalysis. html). Accordingto the st and ard of World Meteorological Organization(WMO), the climate normals are the latest threedecade averages of climatological variables, i. e., 1981-2010. Thus, in this paper, all the anomalies are derivedbased on the climate normal of 1981-2010.

Monthly WPSH indices are provided by the National Climate Center of China. The EASM indexis defined as the difference of the area-averaged 850-hPa zonal wind speed between the regions 10°-20°N, 100°-150°E and 25°-35°N, 100°-150°E(Zhang et al., 2003a). Following the definition of Zhao(1999), the 500-hPa geopotential height values averaged over 50°-60°N, 120°-150°E; 50°-60°N, 80°-110°E; and 40°-50°N, 40°-70°E are used as the Okhotsk blocking highintensity(OBHI), the Baikal blocking high intensity(BBHI), and the Ural blocking high intensity(UBHI), respectively . 3. Decadal changes of the relationships between EASM circulations and summer rainfall

Since the 1980s, a number of studies have reported that the leading mode of summer precipitationvariability in China is an anti-phase interannual variation in the middle to lower reaches of the Y angtzeRiver against that in South/North China. Comparedwith other regions, precipitation in South China ismore vulnerable to both tropical cyclones and convection. Therefore, in this paper, only the precipitation along the Y angtze River and the precipitation inNorth China are analyzed so as to explore the decadalchanges of the relationships between monsoon circulations and rainfall. The EASM circulations have remark able interannual and interdecadal variations(Miao and Lau, 1990;Qian, 2005; Ding et al., 2013). Figure 1a shows thetime-longitude cross-section of the 500-hPa geopotential height anomaly along 20°-30°N, i. e., along theWPSH center. Distinct differences between certaintime periods are revealed. During most summers before the 1980s, the anomalies were negative, while theybecame positive after 1980. Similar decadal changesare also found in the blocking highs(figures omitted).

The changes may be attributed partly to the increaseof sea surface temperature(SST)in the tropical regions. As shown in Fig. 1b, the SSTs in the western Pacific show a remark able warming trend since1990. The raised SSTs lead directly to the expansionof the atmospheric column and the increase of geopotential height. The correlation coeffcient between theSST in the western Pacific warm pool area and theaveraged 500-hPa geopotential height over 20°-30°N, 120°-150°E in JJA is over 0. 4, exceeding the 95%confidence level. Thus, the WPSH also displays anenhancing and exp and ing trend. Influences of ENSOevents on the WPSH can also be deduced from Figs. 1a and 1b. For example, during strong El Niño phasessuch as in 1983 and 1998, the WPSH is much strongerthan normal.

Fig. 1 Time-longitude cross-sections of(a)summer 500-hPa geopotential height anomaly along 20°-30°N and (b)summer SST anomaly along 10°S-10°-N. Positive values are shaded.

Figure options

In boreal summer, the climatological center ofWPSH lies within 20°-30°N(figure omitted). Figure 2 shows the 21-yr running correlation coeffcients(CCs)between the 500-hPa geopotential height averaged over 20°-30°N and the summer precipitationalong the Y angtze River and in North China. Thedecadal variations of their relationship are clearly revealed. Before the 1990s, precipitation along theY angtze River had a significant positive correlationwith the intensity of the WPSH(Fig. 2a), especially over 110°-140°E. This relationship is consistentwith previous findings and has been applied in operational statistical climate prediction models(Zhao, 1999; Chen and Zhao, 2000). However, after the 1990s, the positive correlation weakened rapidly, especially inthe 2010s when the CCs decreased to only about 0. 1. This means that the WPSH had less direct impacts onthe summer rainfall along the Y angtze River valley inthis period, and this relationship became less important for forecasting. The weakened relationship canbe attributed to the shift of the main rainfall belt inthis period. It moves northward to the Huaihe Rivervalley and to northern China, and the Y angtze Rivergoes into its dry episode, whereas the WPSH exhibitsa decadal strengthening trend(Zhao et al., 2008; Si et al., 2009, 2012).

Fig. 2. 21-yr running correlation coeffcients between the 500-hPa geopotential height averaged over 20°-30°N and the summer precipitation(a)along the Yangtze River and (b)in North China. The number on the ordinate gives thecentral year of the 21-yr running period(e. g., 1981 means the period 1971-1991). The number on the abscissa gives thelongitude. Shaded areas imply the results significant at the 95% confidence level.

Figure options

On the other h and , in the recent two decades(2000s and 2010s), the relationship between theWPSH intensity and summer rainfall in North Chinashows a significant strengthening trend(Fig. 2b). Before the 1990s, the CCs were negative and the mostsignificant negative center was located over the IndianOcean. This is consistent with previous studies(Hao et al., 2011). In this period, negative CCs also appeared over western Pacific but did not exceed the95% confidence level for Student's t-test(Zhang et al., 2008). After the 1990s, the negative correlation overthe Indian Ocean decayed clearly, and the weak negative CCs over the western Pacific became significantlypositive. As mentioned before, since the 1990s, abovenormal rainfall in North China always corresponds toa more westward and stronger WPSH. This correspondence is inconsistent with that assumed in traditionalseasonal forecast models, but the results in Fig. 2 haveconfirmed existence of the inconsistency .

Similar to Fig. 2, Fig. 3 shows the 21-yr running CCs between the summer precipitation averagedat different latitudes in eastern China(east of 110°E) and the area index(Fig. 3a) and western boundaryindex(Fig. 3b)of the WPSH. These two indices represent the strength and position of the WPSH. Decadalchanges can also be seen in Fig. 3. Before the 1990s, the area index had a significant positive relationshipwith the precipitation along the Y angtze River valley, but a weak and negative relationship with the precipitation in North China. The western boundary index also had a significant negative correlation withthe precipitation along the Y angtze River valley, buta weak positive correlation with the rainfall in NorthChina. The results indicate that before the 1990s, amore westward and stronger WPSH always led to morerainfall along the Y angtze River and less in South and North China, and vice versa.

Fig. 3. 21-yr running correlation coeffcients between the precipitation at each latitude in eastern China and the(a)area index and (b)western boundary index of the WPSH. The number on the abscissa gives the central year of the 21-yrrunning period. The number on the ordinate gives the latitude. Shaded areas imply the results significant at the 95%confidence level.

Figure options

However, decadal changes occurred in the recenttwo decades. The significant positive correlation between the WPSH area index and rainfall shifted fromthe Y angtze River to North China. Similarly, the significant negative correlation between precipitation and the western boundary index also moved northward. Note that in North China and northeastern China, thepositive CCs before the 1990s became strongly negative. Though negative CCs still lay along the Y angtzeRiver, their values decreased remark ably. Therefore, it can be concluded from Figs. 2 and 3 that the relationship between the WPSH and summer rainfall hasundoubtedly changed.

Running correlations between the EASM index(Zhang et al., 2003a) and precipitation at each latitudein eastern China shows analogous changes. Figure 4shows the weakening trend of the negative CCs between the EASM index and rainfall along the Y angtzeRiver. The significant negative CCs before the 1990svalidated the scenario used in traditional seasonal for ecast models, namely, more(less)rainfall along theY angtze River corresponds to a weaker(stronger)EASM. However, after the 1990s, the strong negative correlation weakened sharply. Another changeoccurred in northern China, especially in the regionsnorth of 35°N, where the CCs changed from positiveto negative. From Figs. 3 and 4, it can be concludedthat the changed relationship between rainfall and theWPSH is consistent with the changed relationship between rainfall and the EASM.

Fig. 4. As in Fig. 3, but for the CCs between rainfall and EASM index.

Figure options

In addition to the above-mentioned changed relationships, the summer rainfall in eastern China alsoexperienced a changed relationship with the circulations in higher latitudes. Figure 5 shows the 21-yrrunning CCs between summer precipitation and theUBHI, BBHI, and OBHI, respectively. For the rainfallalong the Y angtze River valley(red curves in Fig. 5), the UBHI and BBHI have more important impacts, especially the Baikal high. Their close relationships withsummer rainfall were stable until around 1990. Butduring the 1990s and 2010s, the relationships weakened rapidly. For the precipitation in North China(black curves), the close negative relationship with theUral high during 1960-1990 also weakened and becameweak positive. This result supports the same conclusion on the changed relationship between rainfall and WPSH/EASM.

Fig. 5. 21-yr running CCs between summer precipitation and the UBHI(solid curves), BBHI(thin dashed curves), and OBHI(dotted curves), respectively . Red lines showthe precipitation along the Yangtze River and black linesshow the precipitation in North China. The number onthe abscissa means the central year of the 21-yr runningperiod. Thick dashed lines mean the 95% confidence level.

Figure options

Among the systems influencing the summer rainfall in China, the WPSH undoubtedly plays the mostcritical role. Its meridional position determines location of the interaction between warm and cold airmasses, and its western boundary defines the transport path of the southwesterly monsoon moisture fromocean to continent. Due to global warming, SSTs haveincreased in most parts of the ocean, particularly inthe western Pacific(figures omitted). The warm SSTslead directly to the expansion of the atmospheric column and increase of the geopotential height, with aCC between the two fields above 0. 4 as mentioned inSection 3. Figure 6 shows the 500-hPa geopotentialheight averaged between 20° and 30°N in each summer from 1961 to 2010. Obviously, the WPSH enhanced significantly and exp and ed westward since the1980s, especially over the most recent two decades. Insome years such as 1998 and 2010, the western boundary of the WPSH even reached the eastern part of theEast Asian continent.

Fig. 6. 500-hPa geopotential height averaged between20° and 30°N in each summer from 1961 to 2010. Valuesgreater than 5880 gpm are shaded. The open and filled circles mean the five weaker and stronger WPSH cases studiedin this paper.

Figure options

As mentioned earlier, both the shape and intensity of the WPSH can cause differences in the moisturetransport path and strength, and finally alter the location of the Meiyu rainfall belt. For summer rainfallin eastern China, two moisture transport channels exist. The first channel is via the southwesterly fromthe Arabian Sea and the Bay of Bengal, and the second is via the southeasterly from the western Pacific. Both channels relate closely to the WPSH. For example, a stronger southwest monsoon can propel thesubtropical high to be further eastward and weaken itsstrength, and thus alter the moisture transport path and intensity of the southeast monsoon, and vice versa(figures omitted).

For composite analysis, five stronger and fiveweaker WPSH cases during 1991-2010 are chosen asshown in Fig. 6. The stronger WPSH cases occurredin 1991, 1995, 1998, 2003, and 2010; while the fiveweaker WPSH cases occurred in 1992, 1997, 1999, 2000, and 2002. The composite 850-hPa wind and 500-hPa geopotential height are displayed in Fig. 7. A remark able expansion of the WPSH appeared inthe recent two decades. In comparison to the climatology of 1981-2010, the western boundary of the5880-gpm contour in the stronger WPSH years extended to 120°E. The southerly components of boththe southwesterly monsoon on its western side and thesoutheasterly monsoon on its southern side are muchstronger in the stronger WPSH cases than their counterparts in the weaker WPSH cases. Therefore, instronger WPSH years, the moisture over the SouthChina Sea and the western Pacific could be transported more directly to the continent. The southerlywind speed maximum is located along the Y angtzeRiver, so moisture at lower levels converges in the regions north of the river. This explains why a strongerWPSH brings more rainfall to northern China insteadof to the Y angtze River in the recent two decades. It also confirms the conclusion by Zhang(1999)thatthe water vapor condition in North China is closelyrelated to the intensity of the southerly monsoon inEast Asia.

Fig. 7. Composite of the 850-hPa wind(arrows) and 5880-gpm contours of the 500-hPa geopotential height inthe stronger(black) and weaker(red)WPSH cases. Thethick dashed contour is the climatology . Shaded areas indicate that the mean differences of wind directions betweenthe two cases are greater than 60°.

Figure options

However, the circulations in the weaker WPSHcases display an asymmetric feature owing to thedecadal enhancement of the WPSH. Composite resultsindicate that the WPSH is not more eastward thanits climatology. Under this pattern, abundant moisture cannot be easily brought to the Y angtze River. Distinct differences in moisture transport between thestronger and weaker WPSH cases occurred over thetropical western Pacific, i. e., east of 120°E. In theseregions, the zonal wind directions differ almost 60 degrees in the two WPSH cases(stronger vs. weaker; SeeFig. 7). However, over the regions west of 120°E, thewind directions are almost the same although theirspeeds are different. This indicates that the WPSHplays a more important role in changing the relationships between the WPSH/EASM and rainfall in recentdecades. This result also agrees with the conclusion by Liu and Ding(2011)that the moisture is mainly transported from the west and south sides of North Chinato cause flood events in the region. 5. Conclusions and discussion

The EASM circulations exert large influences onthe summer rainfall in China. Their close relationships have been widely used in operational predictionof short-term climate in China. In traditional statistical forecast models, a La Niña event(i. e., lowerSSTs in the equatorial eastern Pacific)is assumedto cause a weakened and more northward WPSH and stronger summer monsoon, and finally brings floods tonorthern China and drought along the Y angtze River;on the other h and , a stronger and more southwardWPSH, weaker monsoon, and more active blockinghighs will together induce floods along the Y angtzeRiver.

However, obvious changes occurred over the recent two decades(after the 1990s). The relationshipsbetween EASM circulations and rainfall in easternChina have all changed remark ably. For example, before the 1990s, precipitation along the Y angtze Riverhad a significant positive correlation with the intensityof the WPSH, but after that the positive correlationweakened rapidly. In the same period, the CCs between summer rainfall in North China and the WPSHalso changed from weakly negative to significantlypositive. These changes present a big challenge to theapplication of the traditional seasonal forecast model.

The asymmetric features of the monsoon circulations in stronger and weaker WPSH cases during1991-2010 may cause this change. In this period, owing to global warming, the mean western boundary of the WPSH in stronger years can extend quitefar westward to 120°E. The southwesterly monsoonon the western side of WPSH is much stronger, especially the meridional component. More moistureover the South China Sea and the western Pacificis transported directly to the regions north of theY angtze River valley, i. e., North China, to form aregion of convergence. More summer rainfall is thenfound in northern China instead of along the Y angtzeRiver. Meanwhile, in weaker WPSH cases, the westernboundary of the WPSH is near its climatology around140°E. Above normal rainfall cannot be easily broughtto the Y angtze River valley. It could be concluded thatthe WPSH plays a more important role in changingthe relationships between WPSH/EASM and rainfallin eastern China in the recent two decades.

Previous studies considered that an El Niño eventwould lead to a stronger WPSH, such as in 1998, whilea La Niña event would result in a weaker WPSH. Wealso conducted a preliminary analysis to explore thechanged relationship between ENSO and the WPSHin the recent two decades(figures omitted). Theincreased frequency of the central Pacific-type ENSOmay possibly alter the traditional relationship. Therefore, it is also necessary to analyze the roles of ENSOin changing the relationship between monsoon circulations and summer rainfall. This will be conductedpossibly in a future study .

Acknowledgments: The authors thank thesupport of the National Innovation T eam of ClimatePrediction of the China Meteorological Administration.