On the Relationship between the Winter Eurasian Teleconnection Pattern and the Following Summer Precipitation over China

Many studies have been conducted regarding the influences of monsoon circulation on the weather and climate of East Asia and China since the 20th century (Guo, 1983; Ding, 1994; Shi and Zhu, 1996; Wang, 2001; Li and Zeng, 2002; Zhang et al., 2003; Zhang and Guo, 2005). However, the weather and climate of China are not only affected by tropical and subtropical systems, but are also closely related to the extratropical atmospheric circulation in the NH (Zhang and Tao, 1998). The East Asian summer monsoon system clearly reveals the locations and intensities of the interactions between the cold air activities from the mid-high latitude systems and the warm and moist air flows brought by the subtropical system (Tao and Chen, 1987), which are the main reasons for the location and intensity variation of the summer rain belt in China.

In fact, in an early study of the mid-high latitude circulations in the NH by (Wallace and Gutzler, 1981), it was suggested that there are five significant teleconnection patterns in the 500 hPa geopotential height field in the NH winter: the Pacific-North American (PNA) pattern, the eastern Atlantic pattern, the western Atlantic pattern, the western Pacific pattern, and the Eurasian (EU) pattern. (Barnston and Livezey, 1987) further confirmed the existence of the EU pattern based on rotated EOF analysis (REOF) of the monthly mean 700 hPa geopotential height field. (Hoskins and Karoly, 1981) demonstrated the great-circle theory to interpret the dynamics mechanism of the teleconnection patterns.

Among these five significant patterns, the EU pattern is an important low-frequency pattern with well-known impacts on the atmospheric circulation and climate anomalies in the Eurasian region (Hsu and Wallace, 1985; Barnston and Livezey, 1987). (Li and Chou, 1990) demonstrated that the EU pattern is a major factor influencing the winter precipitation over the middle and lower reaches of the Yangtze River. In a study on the relationship between the Arctic Oscillation and the East Asian winter monsoon (EAWM) (Gong et al., 2001), it was determined that the EU pattern makes a significant contribution to the EAWM system, and its contribution to the Siberian high was found to be 36%. In addition, it was also pointed out that when the EU index is positive, the East Asian air temperature is lower. (Shi and Zhu, 1996) found that in cases of strong EAWM, China tends to be cold and dry in winter, and the atmospheric circulation is characterized by a strong western Pacific pattern and weak EU pattern. Several studies have pointed out that the daily variation of the EU pattern is responsible for climate anomalies over China and Korea, where abnormal cold/warm events are often dependent on the different phases of the EU pattern (Sung et al., 2009; Liu and Chen, 2012; Zuo and Xiao, 2013; Wang and Zhang, 2015). Positive EU phases are accompanied by strong northerly wind and a sudden descent of temperature in South China and Korea, while the probability distribution of cold/warm events is dependent on the phase of the EU pattern (Sung et al., 2009). (Wang et al., 2010) confirmed that the atmospheric circulation anomalies related to the Ural blocking (UB) are associated with the Eurasian wave train from west to east, and exhibit an enhancing influence on the East Asian winter climate anomalies.

Moreover, the pre-winter EU pattern also has great impacts on the following summer climate anomalies over China. (Sun and He, 2004) used the SVD method to reveal the influences of pre-winter Eurasian circulation anomalies on the following summer precipitation over China. The pre-winter circulation of the Eurasian mid-high latitudes bears a very close coupling relationship with the following summer precipitation over China. It was also determined that the pre-winter circulation anomalies may influence the Eurasian summer circulation, as well as the precipitation anomalies over China, through a half-year rhythm relationship. Recently, when the "Conceptual Prediction Model for the Three Rainfall Patterns" in the summer of eastern China was reconstructed by (Zhao and Feng, 2014), the winter EU index (EUI _V) was defined according to the difference of the 850 hPa v-wind anomalies in several key regions over the Eurasian mid-high latitudes. It was found that EUI _V can be used effectively to judge whether or not the main following summer rain-belt would locate in northern China; namely, a north rain-belt pattern. However, the relationship between the winter EU pattern and the following summer precipitation has not been comprehensively examined. In addition, obvious interdecadal variability took place in the global oceans and atmospheric circulations in the late 1970s with global warming (Wang, 1995; Guilderson and Schrag, 1998; Li et al., 2004; IPCC, 2013), and the relationship between regional climates and their major factors of influence has changed (Wang, 2002; Gao et al., 2006; Wang and He, 2012). So, has the relationship between the winter EU pattern and the following summer precipitation undergone change?

Understanding the impact of the EU pattern on climate anomalies over East Asia is important both for accurate weather forecasts and short-term climate forecasts. As mentioned above, the winter EU pattern can impact upon the concurrent weather and climate in East Asia. Plus, it also has "climate effects" on the subsequent climate over China. However, the relationship between the winter EU pattern and precipitation in the following summer over China is still not clear. The primary objectives of this study, therefore, are to discuss the relationship between the winter EU pattern and the following summer precipitation over China.

The remainder of this paper is organized as follows: Section 2 describes the data and methods used in this study. The calculation, temporal evolution of different defined EUIs, and vertical structure of the EU pattern are shown in section 3. The relationship between the winter EUI _V and the following summer precipitation over China is illustrated in section 4, followed in section 5 by a discussion of the summer atmospheric circulation and SST associated with the winter EUI _V. A summary and conclusions are given in section 6.

The main datasets employed in this study are: (2) monthly average precipitation data of 160 stations from the China Meteorological Administration for the period 1968-2013, and monthly global precipitation data——gridded at a resolution of 2.5^°× 2.5^°——from the GPCP for the period 1979-2013 (Huffman et al., 1997; Adler et al., 2003); (3) monthly mean circulation data, gridded at a resolution of 2.5^°× 2.5^°, from the NCEP-NCAR reanalysis (Kalnay et al., 1996) [note, however, that because the quality of the NCEP-NCAR reanalysis data over Asia may be low prior to 1968 (Yang et al., 2002; Wu et al., 2005), only the information since 1968 is analyzed in this study]; (3) SST data, gridded at a resolution of 2^°× 2^°, from ERSST.v3b (Smith et al., 2008); (4) the Niño3.4 SST index from the CPC.

The time period analyzed in this study is 46 winters from 1967/1968 to 2012/2013. Wintertime means are constructed from the monthly means by averaging the data of December-January-February (DJF). Here, the winter of 1968 refers to the 1967/1968 winter. Springtime means are constructed from the monthly means by averaging the data of March-April-May (MAM), and summertime means are constructed from the monthly means by averaging the data of June-July-August (JJA).

Correlation analysis, composite analysis, and linear regression are used to investigate the relationship between the winter EU pattern and the following summer precipitation over China.

Figure 1.  The winter EUI (bars) and cumulative index (lines) from 1968 to 2013: (a) EUI$_WG$; (b) EUI$_BL$; (c) EUI$_V$.

In previous studies, the definitions and calculation methods of the EUI are different. In brief, they include two types. The first method uses the differences of the 500 hPa geopotential height anomaly field at a few key points (Wallace and Gutzler, 1981; Sung et al., 2009). The second method uses the corresponding time coefficients after the REOF on the Eurasian mainland height field of the NH troposphere (Horel, 1981; Hsu and Wallace, 1985; Barnston and Livezey, 1987). In addition, the difference of low-level (850 hPa) v-winds in several regions in the Eurasian mid-high latitudes is defined as the EU index (EUI _V) by (Zhao and Feng, 2014). In order to compare the research results, the three types of definition methods proposed by (Wallace and Gutzler, 1981), (Barnston and Livezey, 1987) and (Zhao and Feng, 2014) are adopted to calculate the winter EUI from 1968 to 2013, and these calculation methods are shown as follows:

The definition of the EUI introduced by (Wallace and Gutzler, 1981) is shown in Eq. (2), and denoted as EUI _WG: $${EUI}_{WG}=-\dfrac{1}{4}Z^*_{55^{\circ}{N},20^{\circ}{E}}+\dfrac{1}{2}Z^*_{55^{\circ}{N},75^{\circ}{E}}- \dfrac{1}{4}Z^*_{40^{\circ}{N},145^{\circ}{E}} , (1)$$ where Z^* represents the normalized seasonal average 500 hPa geopotential height anomaly.

The EUI is defined by (Zhao and Feng, 2014) according to the significant area of the anomalies in the 850 hPa v-wind anomaly field in a north-type rain year over China. In Eq. (3), V' represents the 850 hPa v-wind anomaly: \begin{eqnarray} \label{eq2}{EUI}_{V}&=&V'_{50^{\circ}\hbox{-}60^{\circ}{N},45^{\circ}\hbox{-}55^{\circ}{E}} +V'_{55^{\circ}\hbox{-}65^{\circ}{N},130^{\circ}\hbox{-}140^{\circ}{E}}-\nonumber\\ &&V'_{35^{\circ}\hbox{-}45^{\circ}{N},10^{\circ}\hbox{-}20^{\circ}{E}}- V'_{55^{\circ}\hbox{-}70^{\circ}{N},95^{\circ}\hbox{-}110^{\circ}{E}}-\nonumber\\ &&V'_{50^{\circ}\hbox{-}65^{\circ}{N},155^{\circ}\hbox{-}165^{\circ}{E}} . (2)\end{eqnarray}A REOF decomposition was carried out by (Barnston and Livezey, 1987) on the 700 hPa geopotential height anomaly field in the extratropical NH (20^°-90^°N, 0^°-360^°). For unification, the seasonal average 500 hPa geopotential height anomaly fields are used here for the REOF decomposition, and the time coefficient corresponding to the sixth mode is defined as the EUI (denoted as EUI _BL). The variance contribution of the sixth mode was 6.1%, and the accumulative variance contribution of the leading six modes reached 76.3%. The spatial distribution types of the leading five modes are similar to the PNA, North Atlantic Oscillation, and other teleconnection patterns. The mode with the first east-west wave trains is the sixth mode over Eurasia, which is similar to the EU pattern defined by (Wallace and Gutzler, 1981).

Figure 1 shows the temporal evolution of the three winter EUIs during 1968-2013. The interannual variability of the three indices is relatively consistent. The correlation coefficient between EUI _WG and EUI _BL is 0.77, that between EUI _WG and EUI _V is 0.76, and that between EUI _BL and EUI _V is 0.72, all above the 99.9% confidence level. Through power spectrum analysis of the three indices (figure not presented), it is found that the quasi-three-year interannual variation period only exists in EUI _BL, while the interdecadal variation period is not significant in the three indices.

From the spatial distribution of the correlation coefficient between the three winter EUIs and the simultaneous 500 hPa geopotential height field (Fig. 2), it can be seen that all three indices show an obvious zonal teleconnection pattern of wave trains over Eurasia, of which the 500 hPa geopotential height anomaly fields in Western Europe, the Urals, and the coast of East Asia present significant negative-positive-negative correlation areas. The three activity centers of the winter EU pattern defined by (Wallace and Gutzler, 1981) are located at (55^°N, 20^°E), (55^°N, 75^°E) and (40^°N, 145^°E), respectively. The three activity centers are all located in the centers of the three high correlation regions in the EUI _WG's correlation diagram (Fig. 2a). Moreover, there are another two positive correlation centers in eastern North America and the northern North Atlantic. In the EUI _BL correlation diagram (Fig. 2b), there are two larger positive correlation areas in eastern North America and the northern North Atlantic, of which the positive correlation is very significant. However, the negative correlation area is smaller in Western Europe, where the significance of the negative correlation is weaker than the EUI _WG. In the EUI _V correlation diagram (Fig. 2c), there are two larger positive correlation areas in eastern North America and the northern North Atlantic, of which the positive correlation is very significant. The negative correlation area in Western Europe is larger and more significant than the EUI _BL. The spatiotemporal characteristics of the three winter EUIs are highly consistent, and the EUI _V shows the best relationship with the following summer precipitation over China among the three indexes from our calculation and analysis. Therefore, we choose the EUI _V to analyze the relationship between the winter EU pattern and the following summer precipitation over China.

Figure 2.  Correlation between the EUI and 500 hPa height field (20$^\circ$-90$^\circ$N, 0$^\circ$-360$^\circ$) in winter (DJF-averaged) from 1968 to 2013: (a) EUI$_WG$; (b)EUI$_BL$; (c) EUI$_V$. The shading from light to dark exceed the 95%, 99% and 99.9% confidence level, respectively. The contour interval is 0.2. The black dots are the three activity centers of the EU teleconnection pattern defined by Wallace and Gutzler (1981).

Figure 3.  Correlation between the winter EUI$_V$ and the following summer (JJA-averaged) precipitation over China during (a) 1968-2013 and (b) 1981-2013. The black dots indicate the 95% confidence level.

Figure 4.  21-year moving correlation between the EUI$_V$ and (a) $R_NC$ and (b) $R_JH$. The solid (dotted) line in (b) indicates the 95% (90%) confidence level.

Figure 5.  The winter EUI$_V$ and "Three Rainfall Patterns" in the summer of eastern China from 1981 to 2013. The black (green) dotted line indicates EUI$_V$ greater than 2 (lesser than $-2$).

Figure 3 shows the distributions of the correlation coefficient between the EUI _V and the following summer precipitation over China. There is a positive correlation in North China and a negative correlation in the Yangtze River-Huaihe River basins during 1968-2013 (Fig. 3a). Plus, the areas and stations above the 95% confidence level increase significantly during 1981-2013 (Fig. 3b) compared with during 1968-2013. According to Fig. 3b, two areas with a high density of stations above the 95% confidence level are selected. They are: eastern Xinjiang-western North China [hereinafter referred to simply as "North China"; (37^°-47^°N, 85^°-110^°E); 14 stations], and the Yangtze River-Huaihe River basins [(30^°-34^°N, 110^°-125^°E); 15 stations]. The area-averaged summer precipitation of the above two regions is represented by R _NC and R _YH, respectively. EUI _V has a weak positive correlation (correlation coefficient of 0.18) with R _NC, and a weak negative correlation (-0.10) with R _YH during 1968-81. Whereas, EUI _V has a significant positive correlation (0.42, exceeding the 99% confidence level) with R _NC, and a significant negative correlation (-0.56, exceeding the 99.9% confidence level) with R _YH during 1981-2013. To further confirm that the relationship between EUI _V and summer precipitation has changed, Fig. 4 shows the 21-year moving correlation between the R _NC, R _YH, and EUI _V. The EUI _V and R _NC show positive correlation, with the correlation slowly weakening after the mid-1980s, and strengthening recently (Fig. 4a). The correlation coefficient between the EUI _V and R _YH is relatively weak before the early 1980s, but it increases significantly after the mid-1980s (Fig. 4b).

In the early 1980s, (Liao et al., 1981) classified the summer rain-belt of eastern China into three patterns ("Three Rainfall Patterns"), which are respectively described as follows: Pattern I is the northern pattern, of which the main rain-belt is located in the Yellow River basin and the region to the north; Pattern II is the central pattern, of which the main rain-belt is located between the Yellow River and Yangtze River; and Pattern III is the southern pattern, of which the main rain-belt is located in the Yangtze River basin or regions south of the Yangtze River. Figure 5 shows the relationship between the winter EUI _V and the "Three Rainfall Patterns" in the summer of eastern China during 1981-2013. There are nine years in which the EUI _V is greater than 2. With the exception of 2000 and 2002, all of the other seven years belong to Pattern I. Also, there are 17 years with an EUI _V less than 0, and only one year belongs to Pattern I (1994). There are nine years for which the EUI _V is less than -2, all of which are Pattern II or III, and none is Pattern I. Generally speaking, the EUI _V could be used to effectively predict Pattern I years. The years for which the EUI _V is greater than 2 (1981, 1985, 1988, 1992, 1995, 2000, 2002, 2004, 2012) and less than -2 (1982, 1989, 1991, 1996, 1997, 1998, 2003, 2007, 2008) are selected for composite analysis.

Figure 6 shows the composite anomalies of precipitation in summer for + EUI _V and - EUI _V years, and their difference. For the + EUI _V composite, a "plus-minus-plus-minus-plus" anomaly wave train is apparent from eastern Kazakhstan-western Xinjiang, northeastern Xinjiang-western Mongolia, and North China and the Yangtze River-Huaihe River basins to the Philippine Sea basin, of which the anomaly is significantly positive in North China, but significantly negative in the Yangtze-Huaihe River basin (Fig. 6a). For the - EUI _V composite, a "minus-plus-minus-plus-minus" anomaly wave train is apparent from eastern Kazakhstan-western Xinjiang, western Inner Mongolia, and North China and the Yangtze River-Huaihe River basins to the Philippine Sea basin. Significantly positive anomalies are present in the Yangtze-Huaihe River basin (Fig. 6b). From the difference distribution between + EUI _V and - EUI _V years, the differences among the above five areas are very significant. Therefore, the winter EU pattern has an extra-seasonal connection with the following summer precipitation in China and the surrounding areas.

To explain the above-mentioned relationships between the winter EUI _V and the following summer precipitation anomalies, we first show the anomalies of geopotential height at 500 hPa and winds at 850 hPa in the summer, obtained as regression upon the winter EUI _V (Fig. 7). Significantly negative geopotential height (Fig. 7a) and cyclonic circulation (Fig. 7b) anomalies exist over the Ural Mountains, Okhotsk Sea and the subtropical western North Pacific in the following summer. That is, in positive winter EUI _V years, the UB and Okhotsk blocking (OB) are inactive, zonal circulation prevails in the mid-high latitudes, and the western Pacific subtropical high (WPSH) tends to be weaker and locates to the north of its normal position in the following summer. This leads to above-normal moisture penetrating into the northern part of East China. As a result, there are significant positive (negative) precipitation anomalies over North China (the Yangtze River-Huaihe River basins). In negative winter EUI _V years, the UB and OB are active, meridional circulation prevails in the mid-high latitudes, and the WPSH tends to be stronger and locates to the south of its normal position in the following summer. As a result, there are significant positive precipitation anomalies over the Yangtze River-Huaihe River basins. These results are apparent via composite anomalies of 500 hPa geopotential height and 850 hPa wind in summer for winter + EUI _V and - EUI _V years, and their differences (figure not presented).

Figure 6.  Composite of the following summer (JJA-averaged) precipitation anomaly percentage in East Asia under the (a) positive and (b) negative winter EUI$_V$, and the (c) composite difference between (a) and (b). The black dots indicate the 95% confidence level.

Figure 7.  Anomalies of the following summer (JJA-averaged) (a) 500 hPa geopotential height (gpm) and (b) 850 hPa winds (m s$^-1$) regressed upon the winter EUI$_V$. The dark and light shading in (a) indicates that the anomalies are significantly different from zero at the 5% and 10% level, respectively. The dark and light shading in (b) indicates that the $u$-wind anomalies are significant at the 95% and 90% confidence level, respectively. The contour interval in (a) is 2 gpm.

Figure 8.  Correlation between the DJF EUI$_V$ and (a) DJF SST, (b) MAM SST, and (c) JJA SST during 1981-2013. The dark and light shading indicates the 95% and 90% confidence level, respectively. And the solid (dotted) lines indicate the positive (negative) values. The contour interval is 0.1.

Figure 9.  Anomalies of the following summer 850 hPa winds (m s$^-1$) regressed upon the MAM (a) $-1.0\times$ Niño3.4 and (b) West Pacific SST zonal difference index (WPZDI) during 1981-2013. The dark and light shading indicates that the $u$-wind anomalies are significant at the 95% and 90% confidence level, respectively.

To help explain the summer circulation anomalies in association with the winter EUI _V, the correlations between the winter EUI _V and SST are shown in Fig. 8. There are significant negative (weak positive) correlations in the western North Pacific and subtropical central North Pacific (western Pacific) between the DJF EUI _V and DJF SST (Fig. 8a). Furthermore, the correlation distribution is very much like a La Niña pattern. The correlation between the DJF EUI _V and MAM SST (Fig. 8b) is similar to that of Fig. 8a, but there is a significant positive correlation in the western Pacific warm pool (WPWP) region, and the negative correlations in the western North Pacific become more significant. We define the SST anomaly (SSTA) difference in the MAM WPWP region (5^°-20^°N, 115^°-130^°E) and northwestern North Pacific (45^°-55^°N, 150^°E-165^°W) as the West Pacific SST zonal difference index (WPZDI). The correlation coefficient between the EUI _V and WPZDI is 0.55, exceeding the 99% confidence level, and the correlation between the DJF EUI _V and JJA SST (Fig. 8c) is insignificant.

We further discuss the atmospheric circulation anomalies in association with the MAM SSTA. Figure 9 displays the JJA 850 hPa wind anomalies obtained by regression on the MAM Niño3.4 (multiplied by -1.0) (Fig. 9a) and WPZDI (Fig. 9b). A significantly cyclonic circulation anomaly is observed to control the subtropical western North Pacific, and an anticyclonic circulation anomaly exists over the Japanese islands and surrounding ocean (Fig. 9a). That is, in negative MAM Niño3.4 years (like La Niña years), the WPSH tends to be weaker and locates to the north of its normal position in the following summer. From the MAM WPZDI-related circulation anomalies (Fig. 9b), a cyclonic circulation anomaly is also observed to control the subtropical western North Pacific, and an anticyclonic circulation anomaly exists over the Japanese islands and surrounding ocean. Also, these circulation anomalies are very similar to the EUI _V-related circulation anomalies (Fig. 7b). Therefore, SSTAs over the northwestern Pacific and subtropical central North Pacific may both contribute to the formation of EUI _V-related circulation anomalies over the western North Pacific.

This paper examines the relationship between the winter EU pattern and the following summer precipitation over China using NCEP-NCAR, GPCP, and Chinese 160-station data for the period 1968-2013. The difference of low-level (850 hPa) v-winds in several regions in the Eurasian mid-high latitudes is defined as the EUI _V by (Zhao and Feng, 2014). The results show that there is a significant positive (negative) correlation between the winter EUI _V and the following summer precipitation over North China (the Yangtze River-Huaihe River basins). Meanwhile, an interdecadal variability exists in the interannual relationship, and the correlation has become significantly enhanced since the early 1980s. Thus, the proposed EUI _V may have implications for the prediction of summer precipitation anomalies in the above regions.

In positive winter EUI _V years, the UB and OB are inactive, zonal circulation prevails in the mid-high latitudes, and the WPSH tends to be weaker and locates to the north of its normal position in the following summer. This leads to above-normal moisture penetrating into the northern part of East China. As a result, there are significant positive (negative) precipitation anomalies over North China (the Yangtze River-Huaihe River basins), and vice versa. Our present study shows that the winter EU pattern bears a close association with the following summer precipitation over China via key components of the East Asian summer monsoon system, such as the UB, OB and WPSH. Previous studies have demonstrated that atmospheric internal dynamic processes, including the Pacific-Japan or East Asia-Pacific wave train from the tropics (Nitta, 1987; Huang and Sun, 1992) and the "silk road" wave train from the mid-high latitudes in the NH (Enomoto et al., 2003; Enomoto, 2004), can exert substantial influence the interannual variability of WPSH. Further examination shows that the SSTA over the northwestern Pacific and subtropical central North Pacific may both contribute to the formation of EUI _V-related circulation anomalies over the western North Pacific. Hence, the EUI _V could be used as an effective predictor of summer precipitation anomalies in North China and the Yangtze River-Huaihe River basins. However, the extra-seasonal mechanism of influence of the winter EU on the following summer precipitation over China requires further study.