Connections between the Eurasian Teleconnection and Concurrent Variation of Upper-level Jets over East Asia

Identified by (Wallace and Gutzler, 1981), the Eurasian teleconnection pattern (EU) is a west-east wave train pattern stretching from Western Europe to East Asia with three significant action centers. (Barnston and Livezey, 1987) further confirmed the existence of the EU pattern based on orthogonally rotated principle analysis of the monthly mean 700 hPa geopotential height field. As one of the most active low frequency modes over the Eurasian continent, the EU pattern has intrigued many atmospheric scientists over the years since its discovery. (Ohhashi and Yamazaki, 1999) suggested that phase changes of the EU pattern are related to remarkable decadal shifts in the global atmospheric circulation. (Sung et al., 2009) revealed the features of eastward energy propagation associated with the EU pattern. (Wang and Zhang, 2014) examined the daily evolution of the EU pattern and found that both linear and nonlinear dynamic processes play an

essential role in its life cycle. Moreover, the EU pattern has great impacts on climate anomalies over East Asia, and is also an important factor affecting the variability of the East Asian winter monsoon (Gong et al., 2001) and snow accumulation events in Tokyo (Tachibana et al., 2007). 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; Wang and Zhang, 2014).

Understanding the impact of the EU pattern on climate anomalies over East Asia is important for both accurate weather forecasts and climate change studies. (Gong et al., 2001) revealed the important role of the Siberian High in linking signals and the variability of the East Asian winter monsoon. (Takaya and Nakamura, 2005) suggested that the Siberian High is an important anticyclone system since its variability dominates the winter climate over East Asia. Many previous studies have indicated that the Siberian High is an important link between the EU and climate anomalies over East Asia. (Sung et al., 2009) found that the correlation between EU index and temperature in Korea is still statistically significant even when the impact of the Siberian High on winter climate in East Asia is removed. They suggested that some other systems instead of the Siberian High must exist that can link EU signals with climate variability and change over East Asia. In this study, we analyze the link between East Asian climate anomalies and the EU pattern with a focus on the role of the East Asian jet stream (EAJS).

The upper-tropospheric jet stream is a very important large-scale circulation pattern. A review of existing literature indicates that two jets exist over East Asia: the East Asian Subtropical Jet (SJ) and the Polar Front Jet (PFJ), located on the southern and northern sides of the Tibetan Plateau (TP), respectively. The PFJ is mainly driven by eddies (Williams, 1979; Panetta and Held, 1988; Panetta, 1993; Lee, 1997), while the subtropical jet forms due to the angular momentum transported by the thermally-induced direct Hadley circulation (Held and Hou, 1980).

In this study, the SJ is further classified into two types: the Plateau Subtropical Jet (PSJ) and the Oceanic Subtropical Jet (OSJ). The PSJ is located to the south of the TP over land, while the OSJ lies over the eastern coast of the Asian continent, adjacent to the Northwest Pacific Ocean. We distinguish the PSJ from the OSJ because the OSJ is both thermally- and eddy-driven (Li and Wettstein, 2012), which is quite different from the thermally-driven subtropical jet on the global scale. In addition, some previous studies have also revealed that the variation of the OSJ often lags the variation of the PSJ by about five days (Liao and Zhang, 2013), indicating that the PSJ and OSJ have different characteristics of variation. Therefore, this classification is necessary for our purpose, which is to specifically discuss the variation of the EAJS during EU events.

Numerous studies related to variations of the EAJS have been published, in which it has been noted that diabatic heating and transient eddy activity are the two primary factors determining the seasonal evolution of the westerly jet (Zhang et al., 2006; Ren et al., 2008; Kuang et al., 2014). Several studies have reported a poleward shift of global jet streams during the past few decades (Seidel and Randel, 2007; Solomon et al., 2007; Zhang and Huang, 2011; Pena-Ortiz et al., 2013), whereas the location of the East Asian westerly jet has remained almost unchanged. This fact indicates that the variation of the East Asian westerly jet is unique (Zhang and Huang, 2011). The EAJS is also a dominant factor governing climate anomalies over East Asia and downstream regions (Wang et al., 2002; Yang et al., 2002; Jhun and Lee, 2004; Zhang et al., 2008). Furthermore, many studies have pointed out that the EAJS could carry various teleconnection signals to remote regions, resulting in climate anomalies over these areas. For example, the westerly jet could act as a waveguide and transmit the signal of the North Atlantic Oscillation to East Asian and North Pacific regions (Watanabe, 2004; Yu and Zhou, 2004; Li et al, 2005; Xin et al., 2006; Yu and Zhou, 2007). (Gong and Ho, 2003) revealed that the EAJS plays an important role in the relationship between the Arctic Oscillation in the spring and precipitation over East Asia in the summer. However, the role of the EAJS in linking the EU teleconnection pattern with climate anomalies in winter over East Asia is still not clear. The primary objectives of this study are (1) to examine the features of the EAJS corresponding to the variation of the EU pattern, and (2) to explore the role of the EAJS in linking the EU teleconnection pattern with East Asian climate anomalies.

The remainder of this paper is organized as follows. Section 2 describes the data and methods used in this study. The variation of the EAJS during positive and negative phases of the EU pattern is presented in section 3, followed in section 4 by a discussion of the potential mechanisms involved in the variations of the EAJS. Various configurations of the PFJ and OSJ and their behaviors in linking the EU pattern and climate anomalies in China are investigated in section 5. A summary and conclusions are given in section 6.

The National Centers for Environmental Prediction-National Center for Atmospheric Research (NCEP-NCAR) daily reanalysis data are used in this study. The horizontal wind (u,v), geopotential height (z), and air temperature (T) are extracted from the daily reanalysis dataset, which has a horizontal resolution of 2.5^° in both longitude and latitude. The study period covers 60 years of boreal winter months (November to March, NDJFM) from 1950/51 to 2009/10. Daily averaged temperature and precipitation observations from 756 weather stations in China are also used. These observations cover the period from 1 January 1950 to 31 July 2010.

The EU index (EUI) used in this study is calculated based on the definition introduced by (Wallace and Gutzler, 1981): $$ { EUI}=-\dfrac{1}{4}Z^\ast_{( 55N,20E)}+\dfrac{1}{2}Z^\ast_{( 55N,75E)}-\dfrac{1}{4}Z^\ast_{( 40N,145E)} , $$ where Z^* represents the normalized 500 hPa geopotential height anomaly. Persistent EU episodes are defined similar to those in (Horel, 1985), (Mo, 1986), Feldstein (2002, 2003), (Walter and Graf, 2006), and (Feldstein and Dayan, 2008). A persistent episode is identified if the EUI exceeds one standard deviation and lasts for at least five days. If the EUI is positive (negative) during the life span of a persistent episode, this episode will be regarded as an episode in positive (negative) EU phase. Furthermore, if two episodes occur within an interval less than eight days and share the same EU phase, the second episode will be removed. Based on the above criteria, totally 87 positive EU episodes and 85 negative ones are identified in the research period.

To examine variations of the EAJS under different EU patterns, the occurrence frequency of the jet core at each grid point is calculated using the same method described by (Ren et al., 2010). A jet core is identified and the corresponding latitude and longitude are recorded if (1) the wind speed is equal to or greater than 30 m s^-1, and (2) the wind speed is the local maximum of the surrounding 24 grid points. This method is applied to 300 hPa winter daily wind data over the region (20^°-70^°N, 60^°-160^°E).

To investigate the transient eddy activity during positive and negative EU phases, we examine the atmospheric baroclinicity in the lower troposphere, since it is an ideal indicator of the activity of transient eddies (Charney, 1947; Eady, 1949). An accurate measure of the baroclinicity is the Eady growth rate (Eady, 1949; Hoskins and Valdes, 1990), which is defined as $$ \sigma =0.31\left(\dfrac{f}{N}\right)\left(\dfrac{dV}{dz}\right) , $$ where f denotes the Coriolis parameter and N is the Brunt-Väisälä frequency, which describes the buoyancy oscillation (in s^-1). It takes the form of \(N=\rho^\frac12[g(1/\theta)(\partial\theta/\partial p)]^\frac12\), where ρ is the density, g is gravitational acceleration, and θ is the potential temperature in the pressure coordinates. V is the wind velocity, and z is the geopotential height. We calculate σ on the pressure levels of 700 and 850 hPa, since the baroclinic development largely occurs in the lower troposphere (Lunkeit et al., 1998).

The transient anomalies are obtained by removing the seasonal cycle from the original data at each grid point. The seasonal cycle is obtained by taking the calendar mean and applying a 31-day running average. We further divide the transient anomalies into two parts: $$ a=a_{ LFE}+a_{ HFE} , $$ where a represents anomalies, a _LFE are calculated using a 10-day cut off low-pass Lanczos filter, and a _HFE represents a 10-day high-pass Lanczos filter. Hereafter, a _LFE and a _HFE are referred to as low-frequency eddies (LFE) and high-frequency eddies (HFE), respectively. (Hoskins et al., 1983) pointed out that eddy fluxes on either side of this cut-off frequency tend to yield different structural properties. Besides, a five-day running average is applied to transient eddies to reduce the sampling fluctuation before performing the composite average.

In order to describe the zonal flow acceleration induced by low- and high-frequency eddy forcing, the following diagnostic variables are calculated. We use the divergence of low-frequency vector E to evaluate the zonal flow acceleration induced by the LFE. The local horizontal Eliassen-Palm vector (E) is defined as \[ E=(\overline v'²-u'²,-\overline u'v') , \] where u' and v' indicate the result from the 10-day cut off low-pass filter. The divergence (convergence) of E corresponds to a forcing on the large-scale horizontal circulation via increasing (decreasing) the mean westerly flow (Hoskins et al., 1983).

The zonal wind tendency due to the HFE momentum forcing is expressed as $$ \left(\dfrac{\partial {u}}{\partial t}\right)_{ HFE}\equiv-\dfrac{\partial}{\partial y}\nabla^{-2} (-\nabla\cdot\overline {{V}_{ HFE}\zeta_{ HFE}}) , $$ where V _HFE and ζ _HFE denote the horizontal velocity vector and the relative vorticity resulting from the high-pass filter, respectively. The overbar denotes the 10-day cut off low-frequency Lanczos filter (Holopainen et al., 1982; Nakamura, 1992).

We first examine the occurrence frequency of the jet core at 300 hPa, which is calculated using the method described in section 2. Figure 1 shows that in both positive and negative EU phases, the jet cores are mainly concentrated in two belts located on the southern and northern sides of the TP, corresponding to the PFJ and PSJ respectively. Two jet core belts merge in the coastal region of East Asia indicating the OSJ. In positive EU phases (Fig. 1a), a high occurrence frequency of the jet core appears to the north of 60^°N with a northwest-southeast orientation, which is more northward than the normal position of the PFJ. There is a broad area with low occurrence frequency of the jet core between the PFJ and PSJ, forming a clear separation zone between the two jets. In negative EU phases, the jet core of the PFJ appears in an area between 45^°N and 55^°N with a west-east orientation, which is almost parallel to the PSJ and located more southward than its climatological position (e.g., Zhang and Xiao, 2013, Fig. 1). In addition, the magnitude of the jet core occurrence frequency for the PFJ is much higher in negative than positive EU phases.

In the PSJ region, there are three centers with high occurrence frequencies of jet cores. In positive EU phases, the

Figure 1.  Spatial distribution of the occurrence frequency of jet cores at 300 hPa during (a) positive and (b) negative EU phases.

Figure 2.  Time series of lagged correlations (a) between the EUI and PFJI; (b) between the EUI and OSJI; and (c) between the EUI and PSJI for the period 1951-2010. Solid lines indicate positive correlation and dashed lines indicates negative correlation. Positive (negative) values on the vertical axis indicate the number of days the EUI leads (lags) the PFJI, OSJI and PSJI. Areas exceeding the 95% and 99% confidence levels are shaded with light and dark blue and red, respectively. (d) The average of the 60-yr correlation coefficient. Positive (negative) values on the horizontal axis indicates the number of days the EUI leads (lags) the PFJI, OSJI, and PSJI. The black dashed lines in (d) denote the 99% confidence level.

jet core occurrence frequency at these three centers is lower than that in negative phases, except for the leftmost jet core center, which shows no significant changes between positive and negative EU phases. The OSJ region has two centers with large occurrence frequencies of jet cores in positive EU phases, located to the south and east of Japan. An obvious decrease of jet core occurrence frequency appears in negative EU phases, leading to an absence of the jet core center south of Japan. Meanwhile, the position of jet core occurrence frequency in the OSJ region during both EU phases shows little disparity.

Concerning the climatological distribution of the occurrence frequency of the jet core in winter (Ren et al., 2010; Zhang and Xiao, 2013; Liao and Zhang, 2013), we select key regions corresponding to the three jet streams: a key PFJ region over (45^°-60^°N, 70^°-100^°E); a key PSJ region over (22.5^°-32.5^°N, 70^°-100^°E); and a key OSJ region over (27.5^°-37.5^°N, 130^°-160^°E). The spatial average zonal wind over the three selected key regions at 300 hPa is used to define the PFJ index (PFJI), PSJ index (PSJI), and OSJ index (OSJI), respectively.

We further investigate the correlation between the EU pattern and the EAJS. Figure 2 displays the lagged correlation features between EUI and PFJI, and PSJI and OSJI, separately, in winters from 1951 to 2010. The correlation coefficients in each winter are calculated first (Figs. 2a-c). An average of the correlation coefficient is then applied (Fig. 2d) to the 60-yr correlation data. As shown in Fig. 2a, the correlation coefficient between EUI and PFJI is significantly negative during 1951-2010, indicating that positive EU phases tend to accompany a weak PFJ. The maximum correlation coefficient appears at lag(0). The OSJI is also significantly correlated with the EUI, but different from the PFJI; the maximum correlation coefficient between the EUI and OSJI appears when the EU pattern leads the OSJ by about five days. However, there is no significant lag correlation between the EUI and PSJI, as shown in Fig. 2c. The average of the 60-year lag correlation coefficients (Fig. 2d) between the EUI and the three jet indexes shows a similar feature. The EUI and PFJI are simultaneously correlated with a value of -0.55, while the EUI and OSJI are significantly correlated with a value of 0.39 at lag(+5) days (i.e., the EU pattern leads the OSJ by about five days). However, the correlation between the EUI and PSJI is not significant at any lag time.

Figure 3.  Latitude-height cross sections of zonal wind speed (a) averaged over 70^°-100^°E and in positive EU phase, (b) averaged over 70^°-100^°E and in negative EU phase, (c) averaged over 130^°-160^°E and in positive EU phase, and (d) averaged over 130^°-160^°E and in negative EU phase. The interval is 15 m s^-1.

Figure 4.  Time evolution of maximum wind in the (a, b, c) PFJ and (d, e, f) OSJ key region 15 days before and after the EU peak day (day 0) during the positive (red line) and negative (blue line) EU phases: (a, d) meridional direction; (b, e) zonal direction; (c, f) wind speed. The units of the vertical axis in (a, d), (b, e) and (c, f) are degrees north, degrees east and m s^-1, respectively. The solid and dashed black lines respectively represent the positive and negative EU index from lag(-15) to lag(15) days.

Figure 3 shows the latitude-height cross sections of zonal wind for positive and negative phases of the EU pattern. Zonally averaged zonal wind over 70^°-100^°E denotes the PFJ and PSJ, while the zonal average over 130^°-160^°E represents the OSJ. The PSJ shows no evident difference in intensity and location between positive and negative EU phases (Figs. 3a and b), indicating that the EU pattern has no significant influence on the PSJ. In positive EU phases, the PFJ and PSJ are well separated at all levels because the PFJ is located more poleward than its normal position. In negative EU phases, however, the PFJ merges with the PSJ and there is no distinct border between them. The maximum zonal wind of the PFJ lies near 45^°N at the 250 hPa level, and is stronger than it is in positive phase. The wind speed in the OSJ is strong in positive EU phases, with a maximum wind speed of 65 m s-1.

The above analyses indicate that the EU pattern does not exert great influence on the PSJ. The EU pattern propagates along the northern side of the TP, and so cannot exert much influence on the jet, which lies on the southern side of the TP. Therefore, our subsequent analyses will focus mainly on the PFJ and OSJ. Figure 4 shows the variations of location and intensity of the jet centers in the PFJ and OSJ key regions with the evolution of the EU pattern. The jet center is defined as where the local maximum wind speed occurs in the key jet region. Since positive and negative EU phases show similar evolutionary features despite their opposite signs, we only discuss the situation in positive EU phases. The lag(0) day denotes the peak day when the EU pattern attains its local maximum amplitude [hereafter negative lag days represent the days prior to lag(0) and positive lag days indicate the days after lag(0)]. For the PFJ region, before the EU pattern experiences evident growth, the jet center shows a slight southward shift from lag(-15) days to lag(-6) days in positive EU phases (Fig. 4a). In the developing period of the EU pattern, the PFJ center moves quickly northward and attains its northernmost position when the EU pattern reaches its peak. In the decaying stage of the EU pattern, the PFJ center moves rapidly southward until reaching its normal position. Accompanied by the southward shift of the PFJ before lag(-5) days, the jet center also experiences a slight eastward shift in the PFJ region (Fig. 4b). An obvious westward shift of the jet center occurs simultaneously with the northward shift of the PFJ. With the development of the EU pattern, the intensity of the PFJ weakens (Fig. 4c). Until lag(0), the PFJ center wind speed attains the lowest value.

For the OSJ region, a northward shift of the jet center occurs at lag(-4) days, which is a few days later than the shift of the PFJ (Fig. 4d). The jet center attains its northernmost position five days after the EU peak time and remains at its northernmost position in the following days. After lag(+5) days, the OSJ center moves gradually southward. When the OSJ center moves northward, an obvious eastward shift of the OSJ center is also observed (Fig. 4e), accompanied by an increasing of the jet center wind speed (Fig. 4f). The intensity of the jet center wind speed attains its strongest intensity at lag(+5) days——the same time as the jet center reaches its northernmost and easternmost position. The evolutionary features of the jet centers in the PFJ and OSJ are consistent with the previous analysis of the relationship between the EU pattern and the EAJS (Figs. 2 and 3).

In order to reveal the variation of baroclinicity in response to the variation of the EU pattern, Fig. 5 presents the spatial distributions of the zonal wind and Eady growth rate σ at 300 hPa over East Asia for positive and negative EU phases. The PFJ is weaker and located more poleward, but the OSJ is stronger in positive EU phases (Fig. 5a) than in negative phases. There are two branches of σ over the East Asian landmass in both EU phases, and they merge over the western Pacific. In positive EU phases, the northern branch is located at high latitudes north of 60^°N (Fig. 5a), which is consistent with the poleward shift of the PFJ shown in Figs. 1, 3 and 4. In negative EU phases, the northern branch is located in the midlatitudes and gets stronger (Fig. 5b). The southern branch of σ over the East Asian landmass shows little difference between positive and negative EU phases. The maximum of σ is located over the western Pacific region in positive phases (Fig. 5a), accompanied by a strong OSJ; whereas in negative phases, the maximum of σ is located over the high-latitude region, corresponding to a fairly strong PFJ (Fig. 5b). The distribution of σ matches well with the spatial pattern of the EAJS. As σ is a good indicator for the activity of transient eddies, the above results imply that transient eddy forcing may play a critical role in the variation of the EAJS. Next, we investigate the impacts of high- and low-frequency eddy forcing on the coupled system of the EU pattern and jets.

In section 3, we revealed that the EUI is significantly and negatively correlated with the simultaneous PFJI. It is also well correlated with the OSJI when the EUI leads the OSJI by five days. To explore the physical mechanisms underpinning

Figure 5.  Spatial distributions of zonal wind speed (shaded) and Eady growth rate (contours) during (a) positive and (b) negative EU phases. The interval of zonal wind speed is 15 m s^-1 and the interval of Eady growth rate σ is 0.2 d^-1.

these relationships, the influence of high- and low-frequency eddies are examined in this section.

Since the variation of the EU pattern occurs simultaneously or prior to the variation of the jets, the lagged regression of zonal wind tendency anomalies against the EUI for lead times ranging from 0 to +5 days are calculated. The zonal wind tendency anomalies are attributed to HFE forcing. The results are shown in Fig. 6. At the beginning of the EU pattern (i.e., lag time of 0 days), negative anomalies of wind tendency are located at 40^°-60^°N over the East Asian landmass, which is the key region of the PFJ, and the center of positive anomalies lies to the north of 60^°N. This result suggests that HFE forcing drives the PFJ poleward by decelerating the westerly in the PFJ key region and accelerating the westerly in regions north of 60^°N in positive EU phases. The above spatial pattern of westerly changes is consistent with the distribution of zonal wind speed and jet core frequency shown in the previous section. In the following days, the wind tendency caused by HFE in the PFJ region and on its north side gradually weakens. Apparently, HFE contributes to the simultaneous negative relationship between the EUI and PFJI and is responsible for the northward shift of the PFJ.

Significant negative zonal wind speed tendency anomalies also appear to the north of 30^°N, while a weak positive tendency center is located to the south for a lag time of 0 days. This result suggests that a deceleration of zonal wind appears in the center of the OSJ region and an acceleration of zonal wind occurs in the southern part of the OSJ. In the following days, the negative anomalies of the zonal wind speed tendency in the OSJ region become weaker. For a lag time of +3 days (i.e., using the EUI from three days prior), a positive zone of wind tendency anomalies appears to the south of Japan, indicating that HFE tends to increase the zonal wind speed in the OSJ region three days after the EU pattern starts. The positive tendency gradually intensifies in the following days and reaches its maximum five days after the EU pattern starts. This is probably the reason why the variation of the OSJ lags the variation of the EU pattern by five days.

We also examine the role of LFE. Lagged regression of local EP flux (E) divergence due to LFE against the EUI is shown in Fig. 7. At the beginning of the EU pattern (i.e., lag time of 0 days), a large positive anomaly center lies over the Baikal vicinity, accompanied by a negative anomaly center to its west. These EP flux divergence anomaly centers could induce a strong westerly wind speed increase in the eastern PFJ region and a wind speed decrease in the western PFJ region. As a result, the PFJ will shift eastwards. It is obvious that changes in LFE forcing could induce eastward-westward shifts of the PFJ. The maximum of the wind tendency anomalies due to LFE over the PFJ region appears at the beginning of the EU pattern (lag time of zero days), indicating a simultaneous variation between the EU pattern and the PFJ.

In addition, note that a positive EP flux anomaly center is located to the south of Japan, which will accelerate the westerly wind in the OSJ region. This result is in agreement with the previous result shown in Fig. 6, i.e. that a positive relationship exists between the EU pattern and the OSJ. However, the divergence of vector E in the OSJ region reaches its peak value at the beginning of the EU pattern (i.e., lag time of 0 days), implying that LFE does not contribute to the lag correlation between the EU pattern and the OSJ.

As mentioned above, HFE forcing decelerates (accelerates) the zonal wind in the PFJ during positive (negative) EU phases, and thus is responsible for the simultaneous negative correlation between the EU pattern and the PFJ. This decrease (increase) of wind may be induced by the high-frequency eddy transportation of momentum. HFE forcing is also the reason why the variation of the OSJ lags the variation of the EU pattern.

Figure 6.  Lagged regression of zonal wind tendency due to HFE forcing at the 300 hPa level for (a) lag(0), (b) lag(+1), (c) lag(+2), (d) lag(+3), (e) lag(+4), and (f) lag(+5). Positive lag days indicate the number of days the EU pattern leads the wind tendency. The contour interval is 0.2× 10^-6 m s^-1 d^-1, with the contour between -0.1×10^-6 and 0.1× 10^-6 omitted. Light and heavily shaded areas respectively indicate values exceeding the 95% and 99% confidence levels.

Figure 7.  Lagged regression of divergence of local EP flux (E) due to low-frequency eddies at the 300 hPa level for (a) lag(0), (b) lag(+1), (c) lag(+2), (d) lag(+3), (e) lag(+4), and (f) lag(+5). Positive lag days indicate the number of days the EU pattern leads the wind tendency. The contour interval is 1× 10^-5 m s^-2, with the contour between -0.5× 10^-5 and 0.5× 10^-5 omitted. Light and heavily shaded areas respectively indicate values exceeding the 95% and 99% confidence level.

Figure 8.  (a) Lagged correlation between the EUI and TI of North China for the period 1951-2010. Positive (negative) values on the vertical axis indicate the EUI leads (lags) the TI. (b) 60-yr average of the correlation coefficient in (a). The values on the horizontal axis carry the same meaning as those on the vertical axis in (a). Panels (c, d) are the same as (a, b), except for the lagged correlation between the EUI and PI of East China. Light and heavily shaded areas in (a, c) respectively indicate values exceeding the 90% and 95% confidence level. The dashed lines in (b, d) denote the 90% and 95% confidence levels.

Figure 9.  Spatial distributions of temperature anomalies in China during (a, b) positive and (c, d) negative EU phases with different configurations of the PFJ and OSJ as in Table 1: (a) type 3; (b) type 4; (c) type 1; (d) type 2. The dots indicate those stations that exceed the 90% confidence level. (units: ^°C)

Figure 10.  The same as Fig. 9, except for precipitation anomalies. (units: mm)

Several previous studies have revealed that the EU pattern has significant influences on the variation of temperature and precipitation in China (Sung et al., 2009; Wang and Zhang, 2014). To further investigate the specific features of the relationship between the EU pattern and surface temperature/precipitation, lagged correlation analysis is performed on a daily time scale. Figures 8a and b depict the lagged correlation between daily EUI and temperature index (TI) in North China for each individual year (Fig. 8a) and the 60-yr average (Fig. 8b). It is obvious that from 1951 to 2010, the EUI and TI are significantly negatively correlated when the EUI leads the TI (or the TI lags the EUI) by a few days. The lagged correlation demonstrates an obvious interannual variation, which is beyond the scope of the current study. The 60-yr average of the correlation coefficient shows that the maximum correlation coefficient appears at the lag time of +2 days (i.e., the EUI leads the TI by two days). The lagged correlation between the EUI and PI in East China shown in Figs. 8c and d indicates that EUI and precipitation anomalies are highly and negatively correlated. The maximum correlation coefficient appears at the lag time of +4 days (i.e., when the EUI leads the PI by about five days).

A few studies have pointed out that concurrent variation between the OSJ and PFJ is important for winter weather and climate in East Asia (Zhang and Xiao, 2013; Liao and Zhang, 2013). To investigate whether the concurrent variation of the PFJ and OSJ is associated with EU features and its importance for winter climate anomalies in China, we further classify the concurrent variation of the PFJ and OSJ into four different types and investigate their relationship with the EU pattern. We define the jet as a strong jet when the jet index is equal or greater than 0.5. In contrast, the jet is regarded as a weak one if the jet index is equal or less than -0.5. Four configuration types are determined: a strong PFJ corresponding to a weak OSJ (PFJI ≥0.5 and OSJI ≤-0.5, type 1); a strong PFJ corresponding to a strong OSJ (PFJI ≥0.5 and OSJI ≥0.5, type2); a weak PFJ corresponding to a strong OSJ (PFJI ≤-0.5 and OSJI ≥0.5, type 3); and a weak PFJ corresponding to a weak OSJ (PFJI ≤-0.5 and OSJI ≤-0.5, type 4). The numbers of these four jet configuration types in positive and negative EU persistent episodes are shown in Table 1.

The spatial distribution of winter temperature anomalies in China with different configurations of the PFJ and OSJ in positive and negative EU phases is shown in Fig. 9. In positive EU phases, significant negative temperature anomalies exist all over China under type 1 (Fig. 9a), whereas obvious positive temperature anomalies appear over China in negative EU phases (Fig. 9c). This result is similar to that of (Wang and Zhang, 2014) and is consistent with the correlation analyses result, which shows the EUI and TI are negatively correlated. If the configuration of the PFJ and OSJ has no influence on the climate effect of the EU pattern, then the temperature anomaly should have the same sign in the same EU phase, regardless of the intensity of the PFJ and OSJ. However, the temperature anomalies are of completely opposite signs under types 3 and 4, although both types appear in positive EU phase. The same situation is also found under type 1 and 2; both occur in the negative EU phase but the temperature anomalies are of opposite signs under the two types. These results imply that the negative correlation between the EU and winter temperature anomalies exists only when one jet is strong and the other is weak. The relationship will be broken when the PFJ and OSJ are both strong.

A similar feature can be found in the distribution of precipitation anomalies. Figure 10a shows clearly that negative precipitation anomalies appear in eastern China when the PFJ is weak and the OSJ is strong in positive EU phases, while the opposite is true when the PFJ is strong and the OSJ is weak in negative EU phases (Fig. 10c). This result is consistent with our previous finding that the EUI is negatively correlated with the PI in eastern China when the PFJ and OSJ are out of phase. However, even in the same positive EU phase, precipitation anomalies under type 4 are quite different from those under type 3; and a similar difference in precipitation anomalies is also found between type 2 and 1, although both are in negative EU phase. Positive (negative) precipitation anomalies appear in southwestern China under type 4 (type 2), while under type 3 (type 1), positive (negative) anomalies are found in eastern China. The precipitation anomalies in eastern China are not obvious in type 2 and 4 when the two jets have a similar intensity. The negative correlation between the EU pattern and precipitation anomalies in eastern China no longer exists if the two jets are of the same/similar intensity.

As discussed above, there is a significant negative correlation between EU and temperature/precipitation anomalies in China. The variation of the EU pattern leads the variation of surface temperature and precipitation by about two and four days, respectively. After dividing the concurrent variation of the PFJ and OSJ into four types based on the configuration of the PFJ and OSJ, we can obtain more accurate features of climate anomalies in East Asia. The negative relationship between EU and surface climate anomalies in China only exists under the configuration types of weak PFJ-strong OSJ or strong PFJ-weak OSJ. If the PFJ and OSJ have a similar intensity, the obvious negative relationship will be broken. Therefore, the spatial distribution of temperature and precipitation anomalies in China is not only influenced by the phase of the EU pattern, but is also related to the configuration of the PFJ and OSJ. This result implies that the EAJS is very important in linking the EU signal to climate anomalies in China.

The present study has examined the variation features of the EAJS corresponding to positive and negative EU phases in the Northern Hemisphere winter based on analysis of 60-yr (1951-2010) NCEP-NCAR reanalysis daily data. The eddy forcing effects due to high- and low-frequency eddies on the relationship between the EU pattern and the EASJ have been investigated. The configuration of the East Asian jet streams and their concurrent variations have been revealed to explore the important effects of the EASJ in linking the EU signal to climate anomalies in China. The main results can be summarized as follows.

There are three jets in the East Asian region: the PFJ, which is located over the poleward side of the TP with a winter climatology key region over (45^°-60^°N, 70^°-100^°E); the PSJ, which lies over the southern side of the TP with a winter climatology key region over (22.5^°-32.5^°N, 70^°-100^°E); and the OSJ, which is situated over the western North Pacific Ocean with a winter climatology key region over (27.5^°-37.5^°N, 130^°-160^°E). The EU pattern is negatively correlated with the simultaneous PFJ. A positive correlation was found between the EU pattern and the OSJ, and the variation of the OSJ lags the EU pattern by about five days. The correlation between the EUI and PSJI is not significant at any lag time. The spatial distribution of the occurrence frequency of the jet core and zonal wind shows that the positive EU phase is accompanied by a northward shift and weakened PFJ, but an intensified OSJ. In positive EU phases, the PFJ moves to the north of 60^°N, resulting in a distinctly separated PFJ and PSJ.

The zonal wind tendency anomalies caused by high- and low-frequency eddy forcing is responsible for the variation of the PFJ and OSJ with respect to the EU pattern. The negative westerly tendency anomaly center located over the PFJ key region and the positive anomaly center located north of 60^°N results in a weak and northward shift of the PFJ. Negative anomalies of zonal wind tendencies are also found over the OSJ region in the simultaneous regression. When the EU pattern leads the variation of the OSJ by a few days, acceleration tendencies of zonal wind appear in the OSJ key region, implying that high-frequency eddy forcing is responsible for the lagged variation of the OSJ relative to the EU pattern. The analysis of low-frequency eddy forcing showed a westward-eastward shift of the PFJ accompanied by the variation of the EU pattern. Low-frequency eddy forcing is responsible for the positive relationship between the EU pattern and the OSJ.

The EU pattern is negatively correlated with temperature anomalies in northern China and precipitation anomalies in eastern China. The significant correlation appears when the EU leads the temperature and precipitation anomalies by two days and four days, respectively. Negative temperature (precipitation) anomalies appear during positive EU phases only when the PFJ is weak and the OSJ is strong, while the opposite is true if the PFJ and OSJ are of the same intensity. These results suggest that the configuration of the PFJ and OSJ has great influence in linking the EU signal to climate anomalies in China.

The present study has revealed possible effects and mechanisms of the EU pattern on the variation of East Asian jet streams. Our results clearly indicate that the effects of the EU pattern on the winter climate in East Asia are more accurate when the configuration of the PFJ and OSJ is considered. This is very helpful for the forecasting of temperature and precipitation anomalies in China. However, several questions still remain unanswered and need to be addressed in future work. For example, it is not clear how the EU pattern can influence the high- and low-frequency eddies and why different configurations of the PFJ and OSJ can change the climatic effects of the EU pattern. Answers to these questions will be helpful for understanding the important impact of the EU pattern on East Asian climate anomalies.