The relationship between the summer (July–September) North Atlantic Oscillation (SNAO) and simultaneous East Asian air temperature is investigated. The results show that the SNAO is related to middle East Asian summer air temperature, however this linkage varies with time on decadal timescale: a strong connection appears after the late 1970s but a weak connection before the late 1970s. Further analysis indicates that this instable relationship may have resulted from the shift of the SNAO mode around the late 1970s. In the period of 1979–2003, the centers of the SNAO mode are located more eastward. A positive-phase (negative-phase) SNAO produces a strong lower-level divergence (convergence) over the Asian jet entrance region, in turn stimulates a strong upper-level convergence (divergence) via the Ekman pumping. Such a convergence (divergence) then excites a zonally oriented quasi-stationary barotropical Rossby wave train along the Asian upper-level jet. Thus the SNAO signal is transported eastward to East Asia, resulting in an anomalous summer air temperature over middle East Asia. However, in the period of 1951–1975, the centers of the SNAO mode are located more westward. The associated upper-level divergence/convergence is away from the Asian jet entrance region, and the SNAO signal cannot be transported eastward to East Asia. Hence the connection is broken.
 The North Atlantic Oscillation (NAO) is the dominant mode of natural climate variability in the North Atlantic and surrounding continents [e.g., Hurrell et al., 2003; Furevik and Nilsen, 2005]. It features primarily a large-scale seesaw in atmospheric mass between the subtropical high and the polar low. The positive-phase (negative-phase) NAO index corresponds to a stronger (weaker) subtropical high and a deeper (shallower) Icelandic low than usual.
 The NAO is also one of the teleconnection patterns that have a year-round presence although it is most active during winter. It is claimed that the summer NAO (SNAO) explains a large portion of the total variance in the atmospheric circulation over the North Atlantic region [Hurrell et al., 2003]. However, up to now there are few studies focusing on the SNAO and its influences. Hurrell and Folland  examined the variation of southern center of SNAO, studied its climatic influence, and found that the southern center of SNAO shows strong interannual to multidecadal variations and it has impacts on summer climate over the North Atlantic regions. More recently, Linderholm et al.  obtained similar results. These studies indicate that the SNAO plays an important role in the summer climate. Does the SNAO also have an influence on its downstream summer climate over East Asia, like the winter NAO? This is one motivation of our study.
 It has been well known that a notable abrupt change occurred globally in climate during the late 1970s [Trenberth, 1990; Wang, 1995, 2001]. Meanwhile, many climate systems' linkages experienced a shift. For example, the connections between both the Indian and East Asian summer monsoons and ENSO are weakened since the late 1970s [Torrence and Webster, 1999; Chang et al., 2001; Wang, 2002; Gao et al., 2006]. Is the relationship between the SNAO and East Asian summer air temperature stable or not? This is the other motivation of our study.
 Thus the purpose of this study is to examine the relationship between the SNAO and East Asian summer air temperature and to explore the possible mechanisms for such relationship. The paper is divided into six sections. Data sets are introduced in section 2. Section 3 examines the relationship between the SNAO and East Asian summer air temperature. The possible mechanism for the coupling of the SNAO and East Asian summer air temperature is explored in section 4. Section 5 and section 6 give the discussion and conclusion, respectively.
 The atmospheric data set applied is the reanalysis produced by the National Centers for Environmental Prediction and the National Center for Atmospheric Research (NCEP/NCAR) [Kalnay et al., 1996]. The variables analyzed include winds, geopotential heights, sea level pressure (SLP), and air temperature.
 The observed monthly temperature data from 160 stations in China are obtained from the China Meteorological Administration for January 1951 to December 2003. The geographical distribution of the stations is present in Figure 1.
3. Relationship Between the SNAO and East Asian Summer Air Temperature
Figure 2a shows the correlation map between the SNAO and Chinese 160-station summer air temperature in the period 1951–2003. It suggests that there are no large scale significant correlations over China, indicating that the connection between the SNAO and Chinese summer air temperature is weak in this period.
 However, along with the globally abrupt climate change in the late 1970s, the SNAO also has a change at the same time. As shown in Figure 3, the SNAO centers are located farther eastward in 1979–2003 (Figure 3b) compared to 1951–1975 (Figure 3a). In particular, after this shift of SNAO mode, a stronger positive anomaly is over the region from the western North Atlantic to western Europe, which means that the Azores high becomes stronger and shifts spatially more eastward in the latter period relative to the former period. Similar spatial shift of the winter NAO mode has been revealed by some previous studies. Jung et al.  suggested that the eastward shift of winter NAO mode extends the winter NAO influence eastward. Does the eastward shift of SNAO pattern revealed here also extend the SNAO influence eastward and consequently impact the East Asian summer air temperature? To answer this question, the correlations between the SNAO and Chinese summer air temperature before and after the late 1970s are calculated. It is found that the relationship between the SNAO and Chinese summer air temperature varies with time. In the former period (1951–1975), the SNAO and Chinese summer air temperature is weakly correlated (Figure 2b). Except for the eastern part of Northeast China and two small regions over middle China, most parts of China have weak correlations. However, in the latter period (1979–2003), the correlation is strengthened (Figure 2c). A large significant negative correlation is over North China. Of the 72 stations over North China (north of 35°N), 70 stations have negative correlation coefficients among which 41 stations are significant above the 95% confidence level. The maximum coefficient reaches −0.67. It implies that a cold (warm) summer over North China corresponds to a positive-phase (negative-phase) SNAO in the latter period.
 To get a more overall picture, the distribution of the correlation coefficients between the SNAO and East Asian summer surface air temperature from the NCEP/NCAR reanalysis is presented in Figure 4. In the whole period (1951–2003) and the former subperiod (1951–1975), East Asia is covered by weak correlations. However, the situation is changed in the latter subperiod (1979–2003). A significant negative correlation belt covers middle East Asia including North China. The similar results from different data confirm a stronger connection between the SNAO and middle East Asia summer air temperature after the late 1970s.
Figure 5 shows the leading modes of NCEP/NCAR summer surface air temperature over East Asia during three periods of 1951–2003, 1951–1975, and 1979–2003. From it, we can find that, in the last 53 years, the leading mode of East Asian summer air temperature does not exhibit a significant change, although there are some visual differences in the amplitude and distribution of values. The leading mode shows a large value center over middle East Asia, which is similar to the correlation distribution in Figure 4c, thus indicating that the SNAO is an important forcing for the variation of middle East Asian summer air temperature after the late 1970s.
 A similar result is found in an additional index analysis. The middle East Asian summer air temperature is measured by an index defined as the mean temperature of stations over North China (north of 35°N). The correlation between this index and NCEP/NCAR summer surface air temperature exhibits a strong positive correlation over middle East Asia (figure not shown), which is similar to the leading EOF mode in Figure 5. It implies that this observational index can be used to represent the variability of middle East Asian summer air temperature. Figure 6 shows the running correlation between the SNAO index and middle East Asian summer air temperature index with a 25-year window width for the period of 1951–2003. It displays that the correlation between the two becomes more negative with the time, and the significant correlations only exist in the periods after 1977–2001. This confirms that the relationship between the SNAO and middle East Asian summer air temperature varies with time.
4. SNAO-Related East Asian Atmospheric Circulation
 Fluctuations and changes in atmospheric circulation are the main forcing factor for regional climate variability [Trenberth, 1990; Xu, 1993]. Thus the East Asian large-scale circulation anomalies associated with the SNAO are analyzed in this section. In particular, the SNAO-related atmospheric circulation is investigated in two periods of 1951–1975 and 1979–2003. Here, the composite analysis is used. Based on the criterion that the SNAO index is larger than 0.5 (less than −0.5) standard deviation, the positive-phase SNAO years in the period of 1951–1975 (1951, 1953, 1954, 1961, 1967, 1969, 1974, and 1975) and in the period of 1979–2003 (1979, 1981, 1982, 1985, 1988, 1989, 1990, 1991, 1992, and 1997) as well as the negative-phase SNAO years in the period of 1951–1975 (1952, 1956, 1966, 1968, and 1972) and in the period of 1979–2003 (1986, 1987, 1994, 1995, 1998, 1999, 2000, 2001, and 2002) are tagged, respectively.
Figure 7 shows the composite differences of summer geopotential heights at three different levels between positive-phase and negative-phase SNAO years in the former period. It suggests that there are no significant large-scale atmospheric circulation anomalies over East Asia at this time, which backs up the finding that the correlation between the SNAO and middle East Asia summer air temperature is low.
 In the latter period, the situation is changed (Figure 8). The composite difference of the atmospheric circulations shows a negative anomaly belt covering middle East Asia at all levels. This negative anomaly and the extent of significant areas are larger at the middle-to-upper levels and decrease toward lower levels. The significant negative areas nearly disappear at 1000 hPa. It indicates that the SNAO is related mainly to the East Asian atmospheric circulations at middle-to-upper levels. Here we select the averaged 250 hPa geopotential height over the region 45°N-55°N, 105°E-125°E (selected area shown in Figure 8a) to represent the East Asian atmospheric circulation variability. Similar to Figure 6, the close connection between the SNAO and East Asian summer atmospheric circulation is also found only after the late 1970s (Figure 9).
 Some previous studies have showed that the geopotential heights at middle to upper levels determine mainly surface air temperature anomalies [Namias, 1948; Xoplaki et al., 2003]. A warm (cool) air temperature is related more to a high (low)-pressure pattern [Kozuchowski et al., 1992; Xoplaki et al., 2003]. In order to compare and contrast the atmospheric circulation anomalies associated with middle East Asian summer air temperature in the two periods, we present here the linear regression patterns of 250 hPa and 500 hPa geopotential heights based on the middle East Asian summer air temperature index in 1951–1975 and 1979–2003. From Figure 10, it is first found that the middle East Asian summer air temperature related atmospheric circulations are generally consistent in these two periods. Corresponding to a warm summer, middle East Asia is covered by positive geopotential height anomalies. Such a relationship is well reflected in the index analysis. As shown in Figure 11, the middle East Asian summer air temperature and atmospheric circulation indices show a similar variability. Their correlation coefficient is 0.61 in 1951–2003.
 On the other hand, there are some visual differences between these two periods' atmospheric circulations. In the former period, the large positive geopotential height values overlie mainly eastern middle East Asia, while in the latter period the large positive geopotential height values are distributed more zonally and cover most of middle East Asian. Such changes in the circulation anomalies are coherent with those in the summer air temperature anomalies. As displayed in Figure 12, in the former period the warm center is located over Northeast China, while in the latter period the warm center is presented more zonally and covers most of North China.
 Comparing the SNAO and middle East Asian summer air temperature related atmospheric circulations, we can see that the SNAO-related atmospheric circulations over East Asia during the latter period are in agreement with those associated with the middle East Asian summer air temperature, which implies that the middle East Asian geopotential height anomaly at the middle-to-high levels may serve as the bridge linking the SNAO and middle East Asian summer air temperature. When the SNAO is closely related to the geopotential height anomaly in the latter period, there is a close relationship between the SNAO and middle East Asian summer air temperature. However, when the connection between the SNAO and the geopotential height anomaly is broken in the former period, the relationship between the SNAO and middle East Asian summer air temperature becomes weak.
Hoskins and Ambrizzi , Branstator , and Ding and Wang  showed that the midlatitude upper lever jet acts as a waveguide of the Rossby wave dispersion. Watanabe  further pointed out that the variability of the winter NAO is tied to the East Asian climate variability via exciting quasi-stationary wave train trapped on the Asian jet. In section 4, it is shown that the strongest connection between the SNAO and East Asian summer atmospheric circulation mainly emerges at the upper level. Thus the upper atmospheric process may account for such far remote connection between the SNAO and middle East Asian summer atmospheric circulation and air temperature after the late 1970s.
 The upper-tropospheric meridional wind can depict zonally oriented teleconnections quite well [e.g., Watanabe, 2004]. Figure 13a presents the composite difference of 250 hPa meridional wind between positive-phase and negative-phase SNAO in 1979–2003. It illustrates alternating anomalous northerly southerly winds along the Asian jet emanating to East Asia and the North Pacific. This wave train can exert an influence on East Asian summer circulation. The anomalous northerly and southerly winds over middle East Asia indicate a negative departure of geopotential height, consistent with Figure 8. Thus the zonally oriented wave train along the Asian jet transports the signal of the SNAO eastward to East Asia.
 However the situation is changed in 1951–1975.The meridional wind anomalies exhibit a difference between the two periods over the coastal North Atlantic, which suggests that the SNAO-related local circulation at the upper level also exists a decadal variation (Figure 13). In addition, although the Asian jet structure does not have a significant change in 1951–1975 relative to 1979–2003, there is no obvious wave train pattern trapped in the Asian jet (Figure 13b). Why do the atmospheric circulations have such significant difference in these two periods?
 This different behavior of SNAO between 1951–1975 and 1979–2003 may result from the shift of the SNAO mode around the late 1970s. Corresponding to a positive-phase SNAO in 1979–2003, there is a strong positive SLP departure over the Asian jet entrance region as shown in Figure 14a. Consequently, there is a strong divergent wind at the lower level of the region (Figure 15b). Because the horizontal flow is divergent from the high, mass continuity demands a compensating vertical inflow. Thus the Ekman pumping produces descending motion above the high. At the upper level, to compensate the downward outflow, a mass convergence is needed. Thus in Figure 15a, an upper level convergence overlies the lower level divergence over the Asian jet entrance region. Based on the Rossby wave source theory [Sardeshmukh and Hoskins, 1988], such an upper level convergence can lead to anomalous Rossby wave source over the region, and thus should excite the Rossby wave along the Asian jet. This point is conformed by the simulated result. Using a linear barotropic model, Watanabe  investigated the atmospheric response to the pointwise vorticity forcing placed over the Asian jet entrance region and obtained a similar zonally oriented wave train along the Asian jet. Thus via this zonally oriented wave train, the signal of the SNAO is transported to East Asia, and consequently the SNAO has influences on East Asian summer circulation and air temperature. However, corresponding to a positive-phase SNAO in 1951–1975, the strong positive SLP departure is over the middle North Atlantic (Figure 14b). Thus the SNAO-related SLP departure and lower level divergence over the Asian jet entrance region are weaker compared to the period 1979–2003 (Figure 14b and Figure 16b). The magnitude of the anomalous velocity potential over the region in 1951–1975 is only about one third of that in 1979–2003. Correspondingly, the anomalous convergence at the upper-level over the Asian jet entrance region is too weak to excite the wave train pattern.
 Along with change of the SNAO mode, the standard deviation of the southern center of the SNAO is increased in 1979–2003 (Figure 17). In particular, the eastward shift of the SNAO southern center activates the atmosphere over the Mediterranean region in 1979–2003. As shown in Figure 17, the averaged standard deviation over the Mediterranean region is increased by more than 30% and the maximum of increase reaches 60%. Such an increase in the variability of atmosphere favors occurrence of stronger divergence/convergence over the Asian jet entrance region, and consequently stimulate eastward-propagating Rossby wave train along the Asian jet as described in last section to influence the East Asian summer climate.
 This study explores the relationship between the SNAO and simultaneous middle East Asian summer air temperature. It is found that this relationship is different in the periods before and after the late 1970s when the globally abrupt climate change happened. In the former period, there is no signal of the SNAO in the variability of the middle East Asian summer air temperature. In contrast, in the latter period, the situation is changed. The variability of the SNAO is related to middle East Asian summer air temperature. A positive-phase (negative-phase) SNAO favors a cool (warm) summer over the region.
 The different relationship between the SNAO and middle East Asian summer air temperature in these two periods may result from the shift of the NAO mode around the late 1970s. Before the late 1970s, the NAO centers are located more westward. But after the late 1970s, they shift eastward. Jung et al.  pointed out that accompanying the eastward shift of the winter NAO mode, its impact also extends eastward. Also, by the land surface feedbacks of Eurasian Continent, the winter NAO also has a delay impact on summer climate [Ogi et al., 2004]. This study shows that along with the eastward shift of SNAO centers, especially the shift of the southern center, the influence of the SNAO is also extended eastward. In the latter period, a positive-phase SNAO stimulates a strong divergence at the lower level over the Asian jet entrance region. Consequently, a strong convergence is produced at the upper level over the same region via the Ekman pumping [Feldstein, 2003; Watanabe, 2004]. The strong convergence then acts as a Rossby wave source and excites a zonally oriented quasi-stationary wave train along the Asian jet to influence the East Asian summer atmospheric circulation. Finally, the changed circulation results in anomalous summer air temperature over middle East Asia.
 The authors are grateful to three anonymous reviewers for their valuable comments and helpful advices. This research was jointly supported by the 973 Program (grant 2006CB403600) and the National Natural Science Foundation of China (grant 40631005 and 40620130113).