The middle and high latitudes of the NorthernHemisphere have a high population density, so changesin weather and climate have been studied extensivelyas they heavily influence quality of life. There exist, particularly in winter, various low-frequency climate variabilities, with time periods on seasonal, interannual, decadal, or longer scales. These anomalieshave characteristic spatial structures, and have beennamed "teleconnection patterns" by Wallace and Guzler(1981). These teleconnection patterns are classi-fied into two groups, according to their spatial structure. The first group has two anomalous centers ofdifferent polarity running north-south, and is called"dipole-like pattern. " Examples of such dipole-liketeleconnection patterns are the North Atlantic Oscillation(NAO; van Loon and Rogers, 1978), the ArcticOscillation(AO; Thompson and Wallace, 1998), and the North Pacific Oscillation(NPO; Rogers, 1981). The other group has many anomalous centers runningeast-west, is called "wave train-like pattern. " Wavetrain-like teleconnection patterns include the EurasianPattern(EU; Wallace and Guzler, 1981), and Pacific/North America Pattern(PNA; Wallace and Guzler, 1981). The growth and developmenTof theseteleconnection patterns are believed to be closely related to local wave-mean flow interaction. These teleconnection patterns influence noTonly the weather and climate in their immediate vicinity, but also theatmospheric systems located thous and s of kilometersaway, such as the El Niñ o-Southern Oscillation, theMadden-Julian Oscillation(MJO), or the stratospheric polar vortex. Recently, the mid-high latitudeteleconnections have been the focus of study for manymeteorologists. This article reviews the progress sofar achieved in this field, with special attention paidto those teleconnection patterns that impact weather and climate in China during winter.

The paper is organized as follows. Section 2 reviews the impacts of the AO/NAO and the NPO onweather and climate over East Asia. Section 3 reviewsthe Eurasian teleconnection patterns and their climateimpacts. The interaction between the AO/NAO and the tropical phenomena, such as ENSO, is discussed inSection 4. Stationary waves and their relation to teleconnection patterns are described in Section 5, and some basic questions related to teleconnection patterns are answered in Section 6. Finally, Section 7summarizes the present state of research in this field, and discusses the questions that remain to be addressed by future studies. 2. Influence of the AO/NAO and NPO on theEast Asian climate2. 1 Relationship between the AO/NAO and climate anomalies over East Asia and the influence of solar cycle

The AO is a dominant pattern of mid-high latitude climate variability in the Northern Hemisphere, presenTon the interannual timescale. Associated withthe strength variation of the AO, there is a "seesaw" inthe pressure and atmospheric mass between the Arcticregion and midlatitudes, manifested as the NAO overthe North Atlantic region. The AO has a significantimpacTon the climate noTonly in the Arctic region, but also in low to mid latitudes(Thompson and Wallace, 1998, 2000). Several studies have reported thatthe East Asian winter monsoon(EA WM)tends to beweak during the positive phase of the AO(Gong et al., 2001; Wu and Wang, 2002; Ju et al., 2004; Chen et al., 2005; Suo et al., 2009). Many studies havedemonstrated the interdecadal variations of the EastAsian climate, and mosTof them attribute this to thesea surface temperature(SST)anomalies or the globalwarming effect(e. g., Zhou et al., 2006; Wang et al., 2008; Wei et al., 2011). Solar radiation is the primaryenergy source for the motion of the atmosphere, and the most important interdecadal timescale is the 11-yrsolar cycle. Several studies have found that the influence of the solar cycle was enhanced through oceanatmosphere coupling in the tropics(van Loon et al., 2007; Meehl et al., 2008; Zhou and Chen, 2012; Zhou et al., 2013). Possible solar influences on the AO orNAO have also been reported, with a confined structure in the Atlantic sector during low solar activity, and a hemispherical structure during high solar activity(Kodera, 2003).

The recent work of Chen and Zhou(2012)furtherinvestigated the modulation by the 11-yr solar cycle ofthe relationship between the AO and the East Asianwinter climate. They found that robust warming appeared in northern Asia in response to a positive AOphase during winters with high solar activity . This corresponded to an enhanced anticyclonic circulation at850 hPa over northeastern Asia and a weakened EastAsian trough at 500 hPa, implying that the cold wavesaffecting East Asia were relatively inactive. However, during winters with low solar activity, both the surface warming and the intensities of the anticyclonicflow and the East Asian trough were much less pronounced in the presence of a positive AO phase. Chen and Zhou(2012)proposed that the possible mechanism for this 11-yr solar cycle modulation may be theindirect influence of solar activity on the structure ofthe AO. Their reasoning was that during high solar activity winters, the sea level pressure(SLP)oscillationbetween the polar region and the midlatitudes associated with the AO became stronger, with the significantinfluence of AO extending to East Asia. Meanwhile, the AO-related zonal-mean zonal winds tended to extend further into the stratosphere during high solaractivity winters, leading to a stronger coupling of thetroposphere to the stratosphere. These trends mighthave led to an enhanced AO phase difference, and thus, the associated East Asian climate anomalies becamelarger and more significant. The situation tended toreverse during low solar activity winters. Chen and Zhou(2012)also revealed that the relationship between the winter AO and surface climate anomalies inthe following spring was also modulated by the 11-yr solar cycle, with significant signals appearing onlyduring high solar activity phases. Thus, solar cyclevariation should be taken into account when the AOis used to predict winter and spring climate anomaliesover East Asia.

Several studies demonstrated the impacTof thespring AO/NAO on the East Asian summer monsoon(Gong and Ho, 2003; Wu et al., 2009). Wu et al. (2009)suggested that the North Atlantic triple SSTanomaly pattern, and associated subpolar teleconnections, played a dominant role during this process. Guet al. (2009)further argued that this relationship between the AO/NAO and the summer rainfall in eastern China was not stable, and had an interdecadalvariability . The recent work of Zhou(2013)revealedthat the 11-yr solar cycle significantly modulated therelationship between the SST and the NAO in spring(May-June average). A typical triple pattern was evident for the correlations between the NAO and theNorth Atlantic SSTanomalies in spring, with a centerof positive values around 40°N, and negative values toboth the northern and southern sides, consistent withprevious study of Wu et al. (2009). Grouping the datainto high and low solar activity phases, this triple feature associated with the SSTanomalies was only evident in the low solar activity phase, and both thedomain and the correlation value became larger. During the high solar activity phase, although there werelarger areas of positive correlations between 30° and 45°N, the negative correlations to both the southern and northern sides of the 30°-45°N zone were nearlyabsent. Hence, the triple SSTanomalies in North Atlantic tended to be much weaker from spring to summer in the high solar activity phase, which may haveaffected the developmenTof subpolar teleconnection innorthern Eurasian continent. In this case, there wasno close relationship between the spring NAO and thesummer climate anomalies in East Asia. In contrast, the spring NAO may have a robust impacTon the EastAsian summer climate in the low solar activity phase(Zhou, 2013). However, the issue of how the 11-yrsolar cycle influences the persistence of NAO-relatedtriple SSTanomalies in North Atlantic is still open, and may benefit from further investigations includingdiagnostic and numerical simulations. 2. 2 Influence of the NPO on the East Asianwinter climate

The NPO manifests as the second eigenvectorof the SLP field over North Pacific on interannualtimescales, with a teleconnection mode of oppositepressure variations over the Aleutian Isl and s and Hawaii(Wallace and Gutzler, 1981). When the NPO isin positive phase, the north-south dipole pattern in theSLP shows an anomalous cyclone around the BeringStraiT and an anomalous anticyclone in the whole ofthe subtropical Pacific, with a barotropic structure extending from the lower troposphere to the tropopauseregion(Wang et al., 2011). It is generally recognizedthat in winter, the NPO influences the climate over the"downstream" region of North America(Linkin and Nigam, 2008) and also the strength of the EA WM. During boreal winter, the subtropical Pacific anticyclone associated with the NPO can extend westwardto the coasTof East Asia and weaken the northwesterlies of the EA WM. Consequently, significant warmingis observed over East Asia and more precipitation appears in the south of the Huai River area(Wang et al., 2011).

In the mid 1970s, the atmospheric circulation underwent a significant change over North Pacific, as theAleutian low(AL)both deepened and shifted eastward(Trenberth and Hurrell, 1994). Wang et al. (2007)alsofound a corresponding change of the typical period ofthe NPO in the mid 1970s, and a related interdecadalchange in the relationship between the NPO and EastAsian winter climate. Before 1976, the influences ofthe NPO on the air temperature over East Asia weresignificant along and close to the coast, but very weaknorth of 40°N inl and . However, this situation reversedafter 1976. This interdecadal change was suggestedto be related to the weakening of the remote forcingfrom the tropical eastern Pacific SSTanomaly, associated with the ENSO to the NPO(Wang et al., 2007). The ENSO had a dominant influence on the NPO viathe stationary wave propagation from the equatorialcentral-eastern Pacific to North Pacific before 1976. At the same time, the NPO mainly influenced thewinter air temperature over southern East Asia, indicating a close link to the southern path of EA WM. After 1976, the atmospheric circulation related to theNPO was quitedifferent, and exhibited a circumglobalwave train pattern over the extratropical regions in theNorthern Hemisphere. There were enhanced northward wave propagations, particularly over the extratropics of East Asia. Hence, the influence of the equatorial central-eastern Pacific SSTon the NPO was veryweak after 1976. The NPO mainly influenced the winter air temperature over northern East Asia, thus indicating a close link to the northern path of the EA WM(Wang et al., 2009a).

The change around 1976/77 in North Pacificalsoappears in the ocean itself, referred to as the PacificDecadal Oscillation(PDO; Mantua et al., 1997). Inthe warm phase of the PDO, the SSTanomalies arecold in central North Pacific, and warm in the inshorewaters along the west coasTof the Americas. The situation is roughly opposite in the cold phase of PDO. The PDO is a strong signal of climate variability on theinterdecadal timescale. On the one h and, the PDO isa disturbance superimposed on the long-term climatetrend, which can directly lead to interdecadal changesof climate in the Pacific and its surrounding area. Onthe other h and, the PDO is an important backgroundfor the interannual variability, significantly modulating the interannual variation, such as the ENSO and its impacts(Yang et al., 2004; Zhu et al., 2008; Feng et al., 2014). The typical influence of ENSO on the NorthAmerican climate was found to be modulated by thePDO(Gershunov and Barnett, 1998). The ENSOAustralian precipitation relationship was found to berobusTonly during the low phase of the PDO(Power et al., 1999). Several recent studies also indicated thatthe relationship of East Asian winter climate to thetropical and extratropical systems underwent a significant change around 1976(Wang et al., 2007, 2008, 2009b, 2010). They found that the impacts of the atmospheric internal signals over mid-high latitudes suchas the Ural blocking, the NPO, etc., on the East Asianwinter climate were enhanced, whereas the impactfrom ENSO weakened. The climate over East Asiawas warmer during El Niñ o winters, while the EA WMwas usually weaker(Zhang et al., 1996; Chen et al., 2000). However, this relationship was also sub ject tointerdecadal change, particularly related to the phaseof the PDO. The close relationship between the ENSO and the winter climate over East Asia was robusTonlyin the low phase of the PDO. This modulation maybe accounted for by the difference in the strength and location of the anticyclone over the Philippine Sea during the different phases of the PDO(Wang et al., 2008; Chen et al., 2013a). Since the PDO was in its highphase from 1976 to the beginning of 21st century, theimpacTof ENSO on the East Asian winter climate wasweaker than that before 1976.

The surface air temperature(SAT)over East Asiacan be significantly influenced by the second singularvalue decomposition(SVD2)mode, a metric obtainedby applying an SVD analysis on 500-hPa geopotential height in the Northern Hemisphere and SST inthe tropical Pacific Ocean, with both observ ations and outpuTof seasonal forecasts from atmospheric generalcirculation models(Jia and Lin, 2011). The atmospheric componenTof SVD2 has been found to sharemany spatial similarities with the AO. The time variation of SVD2, however, was found to be more closelycorrelated to the variation of SATover China thanthe AO. When the SVD2 was in its positive phase, the SATover China tended to be warmer than normal. Furthermore, the seasonal forecasts of SAT and precipitation over China were post-processed bYusinga statistical approach(Jia et al., 2010, 2014), based onthe relationship mentioned above. The results presenta significant improvement. The mechanisms of the in-fluence of this mode on East Asian climate, and therelationship of this mode to the tropical SST and NPOrequire further investigation. 3. T eleconnection patterns over the EurasiancontinenT and their climate impacts3. 1 Ural blocking and the East Asian winterclimate

During winter, the Ural Mountains region has oneof the highest blocking frequencies in the NorthernHemisphere. This blocking is of great importance forthe weather and climate of downstream regions suchas East Asia(Ding and Krishnamurti, 1987; Takaya and Nakamura, 2005). Most previous studies have focused on cases of blocking based on the definition ofblocking by Liu(1994). They first suggested that thepersistence of a blocking pattern could be expressedin terms of the duration of the corresponding anomalypatterns, and defined a blocking index to measure theresemblance of a particular circulation pattern to theblocking regime. Based on this index, Barriopedro etal. (2006)developed a modified version to capture thewinter mean regional anomaly signature of the Uralblocking regime. The blocking frequency in the region was defined as high when the Ural blocking index(UBI)was positive, and vice versa. Using the UBI, Wang et al. (2010)investigated the interannual variation of winter Ural blocking activity and its impacTonthe EA WM. They suggested that the Ural blocking exerted its impact via a quasi-barotropic wave train overthe Eurasian continent. When the Ural blocking activity was strong, this wave train propagated both upward and eastward, and negative geopotential heightanomalies were induced over East Asia. Hence, theEast Asian trough deepened and the EA WM strengthened. However, deciphering the causes of interannualvariation of the Ural blocking is difficult, since thevariability from internal dynamics is much larger thanthat from external factors. The previous study of Li(2004)suggested that the North Atlantic SSTanomalies could affect the interannual variation oFural blocking in early winter.

However, this influence becameweak in mid winter. In mid winter, the NAO hada close relationship with the blocking over North Atlantic, whereas its relation to the Ural blocking wasweak(Huang J. P . et al., 2006). There are increasingpieces of evidence suggesting that the stratosphericcirculation anomaly can influence the weather and climate in the troposphere(Baldwin and Dunkerton, 2001; Chen and Kang, 2006; Huang et al., 2007; Chen et al., 2013c). Recently, Wang et al. (2010)showedthat the stratospheric polar jet had a close relationship with the UBI, especially when the stratosphereled the troposphere by one month. Further analysisindicated that the Ural blocking in winter had becomeweak in recent decades, leading to a decreased blocking frequency . However, the relationship between theUral blocking activity and the East Asian trough and Siberian high-both may represent the EA WM-becamemuch closer. Particularly after 1976/1977, strengthened stratospheric polar night jeT and polar vortexprevented the planetary waves from propagating intothe stratosphere. Therefore, the Ural blocking signalwas found to propagate further eastward in the troposphere to East Asia, thus exerting more influenceson the East Asian winter climate. Evidently, furtherstudy on how the stratospheric polar night jet influences the Ural blocking is needed. 3.2 The Eurasian tele connection(EU)patterns and their impacTon climate

China has a wide meridional coverage, so climatelinked disasters such as drought, flood, and winterfreezing rain/snow storms are related to both the tropical SST and the mid-high latitude atmospheric circulation. For example, Huang Ronghui et al. (2006)found an obvious interdecadal change around 1976, with persistent drought in North China and increasedrainfall in the Yangtze and Huai River valleys. Theyfurther suggested that this change was associated withan EU anomaly . Several further studies also suggestedthat a persistent positive EU-like anomaly played adominant role in the record-breaking, long-lasting coldepisode over southern China in January 2008(Wen et al., 2009; Zhou et al., 2009). However, there were threeEU patterns identified in previous studies by Wallace and Gutzler(1981) and Barnston and Livezey(1987), so it was not clear which EU pattern played the mostimportant role. Hence, Liu et al. (2014)conducteda comprehensive side-by-side comparison among thethree EU patterns, including their temporal variability, three-dimensional structure, and possible mechanisms.

The conventional EU pattern was first identifiedby Wallace and Gutzler(1981). Later, Barnston and Livezey(1987)identified two different EU-like patterns, using the rotated empirical orthogonal function(REOF)method to decompose the anomalies of monthly geopotential height at 700 hPa inthe subtropics of the Northern Hemisphere. Theywere named the EU1 and EU2 by Barnston and Livezey(1987), and were later called the Sc and inavia(SCAND) and the East Atlantic/W est Russia(EA TL/WRUS)patterns by the US Climate Prediction Center/National Oceanic and AtmosphericAdministration(CPC/NOAA). Liu(2013)identifiedthe teleconnection patterns by applying the REOFmethod to the monthly mean 500-hPa geopotentialheighTanomalies in the extratropics of the Northern Hemisphere. Figure 1 presents the obtainedREOF spatial distributions and their time series. TheREOF5 had four centers of action, with two negativecenters around the Sc and inavian Peninsula(50°-70°N, 0°-20°E) and East Asia(40°-50°N, 110°-130°E) and two positive centers over the Atlantic Ocean to thewesTof Europe(40°-55°N, 30°-10°W) and Siberia(55°-75°N, 60°-90°E)(Fig. 1a). This mode somewhat resembled the conventional EU pattern(Wallace and Gutzler, 1981). Liu(2013)calculated themonthly EU index according to the definition of Wallace and Gutzler(1981)from January 1948 to December 2009. The temporal and spatial correlation coe°cients between the REOF5 and the conventionalEU pattern were 0. 64 and 0. 88, respectively . Hence, the REOF5 was identified as the conventional EU pattern. The mode shown in Fig. 1c had three centersof action, with two negative centers over western Europe(40°-55°N, 30°W-10°E) and Siberia(45°-60°N, 80°-105°E) and a positive one around the Scand inavian Peninsula(60°-70°N, 20°-40°E), resembling theEU1(Barnston and Livezey, 1987)or SCAND pattern. The correlation coe°cient between the time series of REOF3 and the monthly mean SCAND indexprovided by CPC/NOAA was 0. 76. Therefore, theREOF3 was the SCAND pattern identified in previous studies. The REOF10 had three centers of action, with a positive center over western Europe(55°-60°N, 15°W-0°E) and two negative centers north ofthe Black and Caspian seas(45°-60°N, 30°-45°E) and North Atlantic(40°-50°N, 60°-40°W)(Fig. 1e), resembling the EU2(Barnston and Livezey, 1987)orEA TL/WRUS pattern. The correlation coefficient between the REOF10 time series and the monthly meanEA TL/WRUS index provided by the CPC/NOAAwas 0. 62. Therefore, the REOF10 reflected theEA TL/WRUS pattern identified in previous studies.

Fig. 1. (a)The fifth REOF mode(REOF5)of the monthly mean 500-hPa geopotential height field for 1948-2009 and (b)the corresponding normalized PC time series for REOF5. (c) and (d), and (e) and (f)are as in(a) and (b), respectively, except for the third and tenth REOF modes. The percentages at the top right indicate the variance explained by eachREOF mode. The contour interv al in(a), (c), and (e)is 0. 01 m. The zero contour lines are omitted.

Figure options

Figure 2 shows the monthly evolution of thevariance of the EU, SCAND, and EA TL/WRUS indices. The variances of the three teleconnection patterns were larger during January-March and October-December, but smaller during May-August. Amongthe distributions of the three index variances therewere also differences. The EU had a maximum inDecember, its secondary extreme value in February, and was low during May-August with the minimumin July . The SCAND had a maximum in November, its secondary extreme value in January, and waslow during May-August with the minimum in July . The EA TL/WRUS had a maximum in January, itssecondary extreme value in February, and its minimum in July . Generally, the three teleconnectionpatterns had larger variances during autumn(October and November) and winter(December-February) and smaller variances during summer(June and July). Their different climate impacts were compared in theNorthern Hemisphere winter(Liu and Chen, 2012; Liu, 2013). The research on the winter EU indicated that in winter, during a positive EU phase, coldanomalies appeared in mosTof China, with signifi-cant cooling in the south to the Yangtze River, and the region from Inner Mongolia to southern NortheastChina(Liu and Chen, 2012). In a winter with positiveSCAND phase, significant cooling anomalies occurredin mosTof eastern China and Xinjiang Region, and significant warming anomalies occurred to the easTof theTibetan Plateau. The influences of the EA TL/WRUSpattern on the climate in China were weak, with theSATanomalies mainly to the north of 40°N. In addition, Liu and Wang(2014)found that the SCANDpattern experienced a clear and abrupt change in 1979. Further analysis indicated that the centers of both theSCAND patterns over Europe and Siberia at 500 hPaextended further southeastward after 1979. The center over the Sc and inavian Peninsula, however, showedlittle change. Accompanied with the changes in spatial pattern after 1979, the influences of the SCANDpattern on the wintertime SAT in the Northern Hemisphere were also enhanced. In particular, there wereenlarged regional temperature anomalies in the polarzone associated with the positive(negative)phase ofthe SCAND phase after 1979. In the meantime, negative(positive)SATanomalies associated with the positive(negative)phase of the SCAND pattern over thenorthern Eurasian continent extended further southeastward, even reaching the Yangtze River valley and Japan.

Fig. 2. Monthly evolution of the variance of the(a)conventional EU index, (b)SCAND index, and (c)EA TL/WRUSindex.

Figure options

Cyclones and anticyclones are main weather systems influencing mid-high latitudes, and are very active during the EA WM, so an investigation of wintertime cyclones and anticyclones is helpful for underst and ing the variation of the EA WM. 3. 3. 1 Climatological featuresMost past cyclone and anticyclone studies overChina have been based on poor-quality datasets, spanning short time periods, or sub jective methods using weather charts(e. g., Chen et al., 1991; Zhu et al., 2000). At the same time, objective methodsfor detecting and tracking cyclones and anticyclonesbased on high-quality reanalysis datasets encompassing longer time periods and automatic numerical algorithms have been rapidly developing, and appliedto cyclone and anticyclone studies(Murry and Simmonds, 1991; Hodges, 1994; Simmonds and Keay, 2000). Wang X. et al. (2009)studied cyclones overEast Asia based on the objective detecting and tracking method and the ECMWF reanalysis dataset. Theyfound that the highest cyclone frequency was in spring, and the lowest in winter. Moreover, winter cyclone frequency was low from 1958 to the mid 1970s, increaseduntil the mid 1980s, and subsequently decreased again. However, the annual mean cyclone intensity(definedas the minimum pressure in each cyclone center)had adecreasing trend for the time period they considered. Based on the NCEP/NCAR reanalysis data, Chen L. et al. (2014)studied the wintertime cyclone and anticyclone activities over China for the time period1948-2007. They found that among the cyclones and anticyclones affecting China, half were generated outof China(mainly in Inner Mongolia) and half withinChina. The cyclones and anticyclones assumed significant asymmetry in their source, path, and lysis. Forthe cyclones, their primary sources were, in order, Inner Mongolia, Northeast China, and the Jianghuai region. A secondary source was the zone from Xinjiangto North China to the northeasTof Southwest China. For the anticyclones, Inner Mongolia and its surrounding, particularly northeastern Inner Mongolia, was aprimary source region. A secondary source region wasthe zone from eastern Northwest China across ShanxiHenan-Hebei to the Jianghuai region. The isl and ofNew Siberia was also a main source of anticyclones.

Away from their source regions, the cyclonestended to move eastward or northeastward across EastAsia. The entire North Pacific, and the NortheastChina-Jeju Isl and and further eastward, were a lysis region. The anticyclones moved largely southwestward, and died out most frequently over East Asia and its neighboring West Pacific. The asymmetrical cyclone and anticyclone path may result from the steering by the stationary East Asian trough. Typically, cyclones(anticyclones)were located east(west)of theupper wave trough axis. Upper waves fully developed and became stationary when they moved upstream tothe climatological location of the East Asian troughdue to l and -sea thermal forcing. Under the steeringof the East Asian trough, the low-level anticyclonesmoved along the northwesterlies of the East Asiantrough and diminished at lower latitudes. In contrast, the low-level cyclones were steered along the southwesterlies of the East Asian trough and travelled longdistances, occasionally reaching even Northeast Pacific(Chen L. et al., 2014). 3. 3. 2 Relation to upper-level jets

In the past decades, significant changes in theNorthern Hemisphere wintertime cyclone and anticyclone activities have been detected. The cyclone frequency increased in high latitudes but decreased atmidlatitudes during 1957-1997, while the cyclone intensity increased for both latitude bands(McCabe et al., 2001). Such changes were most significant duringthe mid 1970s. The North Atlantic showed the sametrend as the whole Northern Hemisphere. Over NorthPacific, the cyclone path was more southward, and its intensity increased at the same time(Chang and Fu, 2002; Nakamura and Izumi, 2002; Nie et al., 2008; Zhang et al., 2012).

Since the mid 1980s, the EA WM has weakenedsignificantly(Xu et al., 2006; Gao, 2007; Wang et al., 2009b; Wang and Chen, 2010). This was associated with the increasing frequency of the cyclones and anticyclones, i. e., being lower(higher)before(after)the mid 1980s(Chen L. et al., 2014). Responding to this change, the East Asian polar front jet wasweaker(stronger)before(after)the mid 1980s. Thismeans that the increase in strength of the polar-frontjet was likely partially responsible for the increase inthe cyclonic and anticyclonic activities(Chen L. et al., 2014). 4. Interaction between the AO/NAO and tropical atmosphere4. 1 Influence of the spring AO on ENSO

ENSO is the strongest interannual variabilitymode of the climate system, with signals in the tropical Pacific SST, and associated SLP and atmospheric and oceanic circulations. ENSO may directly induceweather and climate anomalies in the tropical Pacific, but it also indirectly exerts substantial influenceson the global weather and climate via teleconnection(e. g., Huang and Wu, 1989; Zhang et al., 1996; Wang et al., 2000; Huang et al., 2004). Recently, severalstudies have indicated that the AO had a significantimpact noTonly on the northern extratropical weather and climate, but also on the tropical climate variability(e. g., Zhou and Miller, 2005; Choi et al., 2012). In addition, the spring AO influenced the followingENSO occurrence. Nakamura et al. (2006, 2007)demonstrated that the anomalous westerly winds inthe tropical western Pacific associated with a positive AO phase in spring tended to propagate eastward during the following summer and autumn, triggering an ENSO event. They suggested that dry and cold surges occurred more frequently during the springwith positive AO phase. The inflow of surge associated cold advections would strengthen the convective activity in the tropical western Pacific, leadingto the Gill type response in the atmosphere(Gill, 1980). A pair of anomalous cyclonic cells appearedoff-equator on both sides of the equatorial western Pacific. Correspondingly, the anomalous westerly windswere seen in the equatorial western Pacific. However, this interpretation contradicts the results of Jeong and Ho(2005), who found that the frequency of the wintertime cold surge occurrence was higher in negativethan in positive and neutral AO phases. Chen et al. (2014a)further explored the physical process of theinfluence of AO in spring on the ENSO in the following winter. They suggested that the formation ofthe westerly wind burst in the tropical western Pacific was not caused by cold surge activity, but viathe interaction between synoptic-scale eddies and themean flow over North Pacific. The equatorial westerly anomalies excited downwelling equatorial Kelvinwaves, which propagated subsequently to the equatorial central-eastern Pacific, leading to SST warming there in summer through to fall. The tropicalSST, atmospheric heating, and atmospheric circulation anomalies were sustained and developed throughthe Bjerknes positive feedback mechanism, which resulted in an El Niñ o event in the tropical eastern Pacific in fall and winter. Chen et al. (2014b)also foundthat the AO-ENSO relationship experienced a pronounced interdecadal change at the beginning of the1970s. The spring AO influence on the subsequentENSO was weak before 1970. In contrast, the springAO exerted a substantial influence on the followingwinter ENSO after 1970.

Moreover, several studies have demonstrated thatthe wintertime NPO was able to affect the occurrenceof ENSO in the following winter via the seasonal footprinting mechanism(SFM; Vimont et al., 2001, 2003). Figures 3a-c present the first leading SVD mode ofthe wintertime(November-March average)SLP overNorth Pacific and the SST in the following winter(October-February average)over the tropical Pacific, and its time series. The SVD1 accounted for 62% ofthe total covariance. The anomalous SLP pattern inFig. 3a bears close resemblance to the NPO structure obtained from the Empirical Orthogonal Function(EOF)analysis(Linkin and Nigam, 2008). Theanomalous SST pattern in Fig. 3b illustrates a typicalEl Niñ o peak phase during the following winter. Theregression patterns of the SLP in winter and the SSTduring the following winter upon the normalized SLPPC1 are further displayed in Figs. 3d and 3e to con-firm that the wintertime extratropical NPO was significantly linked to the ENSO in the tropical Pacificin the following winter. However, studies also foundthat NPO-like atmospheric forcing from midlatitudesdid not always trigger an ENSO(Alexander et al., 2010; Park et al., 2013). The spring AO can influence the following winter SST in the equatorial centraleastern Pacific. Further, the springtime SSTanomalies over the subtropical North Pacific associated withthe spring AO bore close resemblance to those associated with the preceding wintertime NPO. Motivated by these qualitatively similar structures of SSTanomalies, Chen S. et al. (2013)hypothesized that thespring AO related SSTanomalies could strengthen orweaken the SST "footprint" in the upper subtropicalNorth Pacific generated by the preceding winter NPO. This would then influence the formation of the westerly wind anomalies generated by the air-sea interaction in the following summer and finally influenced theSFM-driven SSTanomalies in the equatorial centraleastern Pacific. With diagnostics of the data, ChenS. et al. (2013)presented evidence that the springAO exerted a robust modulation on the connectionbetween the NPO and ENSO. Only when the springAO was in positive phase, could a positive(negative)phase of wintertime NPO lead to significant El Niñ olike warming(La Niñ a-like cold)anomalies in the tropical central-eastern Pacific during the following wintervia the SFM. However, the connection between theNPO and ENSO was weak when the spring AO was inits negative phase.

Fig. 3. The first leading SVD homogeneous modes of(a)wintertime(i. e., November through next March)SLP overNorth Pacific(20°{80°N, 120°E{90°W) and (b)SST in the following winter(i. e., Ocotober through next February)overthe tropical Pacific(25°S{25°N, 120°E{80°W), and (c)the time series of the first leading SVD mode. Anomalies of(d)SLP as for(a), and (e)SST as for(b), regressed onto the normalized time series for SLP in(c). The shadings in(d) and (e)indicate the anomalies different from zero at the 5% significance level. Contour interv als are 0. 2 hPa in(d) and 0. 2‰ in(e). The zero contour lines are omitted and negative values are dashed.

Figure options

The NAO has been shown to exhibit variability on a wide range of timescales: from its intrinsictimescale of about 10 days to decades and even centuries(Hurrell and Loon, 1997; Feldstein, 2000; Semenov et al., 2008). The NAO variability may becaused by the internal dynamics of the extratropics(Robinson, 1993; Yu and Hartmann, 1993; Lee and Feldstein, 1996)or by tropical heating. The NAO mayin turn affect, through exciting a wave train in the areadownstream, even the tropical Indian Ocean and WestPacific(Palmer, 1988). On the decadal timescale, Hoerling et al. (2001)found by numerical modeling thata warming of the Indian and Pacific Ocean surface wasable to force a circulation pattern for later half of the20th century, very similar to the positive phase of theNAO. The same result was obtained by later numericalstudies(Hoerling et al., 2004; Selten et al., 2004; Bader and Latif, 2005; Zhou and Miller, 2005). These findings contrast with those of Cohen and Barlow(2005), who argued that the SST trend in the Indo-westernPacific warm pool region was unrelated to the NAOtrend.

On the intraseasonal timescale, there was also evidence that tropical convection can modify the wintertime circulation over North Atlantic. For example, phase 3 of the tropical Madden-Julian Oscillation(MJO)index, characterized by enhanced convectionover the Indian Ocean and reduced convection over thewestern Pacific Ocean, was associated with positivepolarity of the NAO. When the MJO was observed tobe in phase 6, which exhibits opposite convection characteristics to thaTof phase 3, the NAO index tended tocoincide with its negative polarity(Zhou and Miller, 2005; L' Heureux and Higgins, 2007; Cassou, 2008; Lin et al., 2009; Lin and Brunet, 2011).

Considering the NAO's downstream influence, Hsu et al. (1990)found that a subtropical Rossby wavetrain excited by the NAO and propagating southeastward from the North Atlantic to the Indian Ocean, wasa trigger for convection over the Indian Ocean. Similarhigh-frequency Rossby wave trains were also observedto propagate southeastward from the Asian-Pacific jet and forced convection over the central tropical Pacific(Matthews and Kiladis, 1999; Palmer, 1988).

Bader and Latif(2005)suggested that tropicalconvection over the Indian Ocean and the NAO may belinked through a circumglobal teleconnection pattern(CTP). Based on the same idea, Yuan et al. (2011)investigated further the relationship between convection over the tropical Indian and West Pacific oceans. They found thaTon the subseasonal scale, negative(positive)NAO events led to enhanced precipitationover the tropical Indian(W est Pacific)Ocean. Theyalso found that convection over the tropical Indian(W est Pacific)Ocean led to a positive(negative)NAO. On the decadal scale, convection over the tropical Indian Ocean, compared with the tropical West Pacific, enhanced significantly from 1958-1979 to 1980-2001, possibly indicating a positive phase trend of NAO during 1980-2001. 5. Stationary waves and teleconnections overNorth Pacific in winter

Stationary waves are mostly active during winter in the Northern Hemisphere. At the surface, stationary waves assume a structure with three semipermanent atmospheric centers of action: the Siberianhigh, AL, and the Icel and ic low(IL). In the mid and upper troposphere, stationary waves manifest inthe height field as a negative anomalous center overNorth Pacific, a positive center over western NorthAmerica, a negative center over eastern North America, and a positive center over eastern North Atlantic and downstream. These stationary waves have apparent variability on seasonal to interannual and decadaltimescales, heavily affecting the circulation and climate over the wintertime Northern Hemisphere.

The three-dimensional propagation of stationarywaves is well described by the wave activity flux(Plumb, 1985). The flux is parallel to the local groupvelocity and thus provides a good indicator of wave energy propagation. In the lower and mid troposphere, there exist two centers of wave activity over East Asia and western North Pacific, and North Atlantic, respectively . There also exists a center over the easternNorth Pacific and North America. The first two centers are called the East Asia/W est Pacific and NorthAtlantic wave trains, respectively, while the last centeris called the eastern North Pacific wave train(EPW), which is weaker than the previous two(Plumb, 1985; Zhou et al., 2012). 5. 1 EPW and PNA

Though it is weaker in magnitude, the variance ofEPW is as large as the other wave trains, which meansthat the EPW has apparent variability that impactsthe circulation and climate over the EPW area, itssurroundings, and even beyond. Zhou et al. (2012)defined an EPW index, in which the vertical wave fluxis chosen, before being averaged over the lower to midtroposphere over the eastern North Pacific for eachwinter. The results showed that the index was in itssecond maximum during 1958-1964, weakest in 1965-1975, and reached its maximum in 1976-1987. After1987, the EPW weakened again(Zhou et al., 2012).

To ascertain the EPW's structure in the geopotential height field for its weak and active phases, thewave index was st and ardized to enable a winter to belabeled as an active(a non-active)winter when thest and ardized index of the winter was larger than 1(less than -1). The composite of geopotential heightshows that at the surface, the AL was very active, withits center beyond the International Date Line(IDL)for the active winters. In contrast, the AL was weak, with its center located wesTof the IDL for weak winters. In the upper troposphere and for active winters, the negative anomalous center over North Pacific wasstrong and extended from the East Asian coast to thesoutheasTof the Aleutian Isl and s; the positive heightanomaly over western North America was also strong. For weak winters, the negative heighTanomaly overNorth Pacific was weak and located over West Pacific to the wesTof the IDL, while the positive heightanomaly over western North America was very weakwith its center shifted westward to the eastern NorthPacific. Very interestingly, the composite height difference of the active minus weak winters assumed aPNA-like pattern over the North Pacific-North America sector. The EPW resembled the PNA noTonlyin spatial structure, but also in time variability . Thecorrelation was as high as 0. 7 for the EPW and PNAindices(Zhou et al., 2012).

Therefore, EPW and PNA were the same phenomenon, but in different physical fields. PNA described the anomalous pattern in the geopotential field and the EPW was a wave train in a wave flux field. Physically, when the atmospheric centers of actionsuch as the AL and the upper-layer anomalous centersover North Pacific and North America varied in theirlocation or strength, low-frequency waves were excited and propagated downstream and vertically from thelower troposphere to the stratosphere. In this way, the low-frequency PNA-like wave train formed. During active(weak)winters, the EPW was strong(weak), and assumed a positive(negative)phase PNA(Zhou et al., 2012). Both the horizontal and vertical propagations of the EPW were of fundamental importanceto atmospheric teleconnections discussed in the nextsection. 5. 2 EPW, AL-IL se esaw, and AO

The AO is a strong signal of climate variabilityin the wintertime extratropics(Thompson and Wallace, 1998, 2000). Since the work of Thompson and Wallace(1998), however, there has been an ongoingdebate over the question of whether the AO patternis a real physical mode or if it is just a consequenceof the EOF methodologYused in its definition(Deser, 2000; Ambaum et al., 2001; Wallace and Thompson, 2002). On the one h and, due to the fact that AOis strongly correlated with NAO, it was believed thatAO is actually the NAO, not a new mode. However, if this is true, the existence of the Pacific center cannot be explained. On the other h and, Thompson and Wallace(2000)proposed that the AO was a fundamental mode of climate variability, describing the seesawof mass between the Arctic and midlatitudes. If thiswere the case, the three centers of action of the AO, i. e., the Arctic center, Atlantic center, and Pacific center, should be closely correlated. Deser(2000), however, found that among the three centers, only the Arctic and Atlantic centers were strongly correlated, exhibiting a negative value, indicative of the well-knownNAO, while the Pacific center was only weakly correlated with the other two centers. Consequently, thedebate on the AO mechanism has lasted for a longtime.

The Arctic-Pacific link is generally weak, thoughsome studies have shown that the AL and IL do notfluctuate independently; rather, the two lows exhibit astrong negative correlation from one winter to the next(Kutzbach, 1970; van Loon and Rogers, 1978; Wallace and Gutzler, 1981; van Loon and Madden, 1983). Thisnegative correlation is called a seesaw oscillation, and its formation and relationship to the AO formation arevery interesting.

Honda et al. (2001)found that the AL-IL seesawshowed a seasonal dependence, with the most significant seesaw observed in late winter(February to midMarch)during 1973-1994. They further demonstratedthat the AL-IL seesaw may be excited by a PNA-likewave train propagating across North Pacific and NorthAmerica, and then to North Atlantic. Castanheira and Graf(2003)presented another interesting findingthat the AL-IL seesaw may be impacted by the stratospheric polar vortex. For a strong(weak)polar vortex, the seesaw was remark ably strong(weak). This meansthat the polar vortex may have a significant impacton the AL-IL seesaw formation by its flexion of theupward propagation of planetary waves from the troposphere.

Recently, Sun and Tan(2013)further investigatedthe seasonal and interannual variation and mechanismof the AL-IL seesaw and proposed that both the horizontal and vertical propagation and the reflection ofthe polar vortex may contribute to the AL-IL seesawformation. They showed that the AL-IL seesaw exhibited apparent seasonal, interannual, and decadal variability . The AL and IL correlation was -0. 33 and -0. 35for December and January, respectively . For Februarythe correlation was -0. 26, smaller than in December and January, while in March the two lows showed nostatistically significant correlation. The correlation forthe entire winter was -0. 26. Interdecadal variability inthe AL-IL correlation was also apparent. In December, the two lows were significantly negatively correlated only for the sub-period 1995-2009(the correlation was -0. 55), while for the other two periods, 1948-1972 and 1973-1994, the AL had no significant relationto the IL. In January, the AL-IL correlation was -0. 4or so for the three sub-periods, however, only for period 1948-1972 was the AL-IL correlation statisticallysignificant; while in February the AL-IL correlationreached its highest value of -0. 73 for 1973-1994, and there was no statistical significance for the other twosub-periods. Therefore, the conclusion of Honda etal. (2001)that the AL-IL seesaw was most significantin late winter was only apparent for the sub-period1973-1994, not for the entire period of 1948-2009. ForMarch, AL and IL still showed no statistically signifi-cant correlation for any of the three sub-periods.

Sun and Tan(2013)considered the effects of theEPW and the polar vortex on the AL-IL seesaw, separately and combined. Without considering thestrength of the EPW, the AL and IL had no significantcorrelation in the case of a weak polar vortex, whilefor the strong polar vortex case, the AL-IL correlation reached -0. 36. Without considering the strengthof the polar vortex, the AL and IL had no relation fora weak EPW(strength below average) and the AL-ILcorrelation was -0. 42 when EPW was strong(strengthabove average). The stronger the EPW, the higher theAL-IL correlation. Therefore, it was clear that bothEPW and polar vortex contributed to the AL-IL seesaw, and were able to work cooperatively . For weakEPW and polar vortex, the AL and IL fluctuated independently . For weak EPW, the AL-IL correlation wasas high as -0. 34 under strong polar vortex conditions, so the polar vortex had a strong impacTon the seesawfor the weak EPW case. When the EPW was strong, the AL-IL correlation was strong, independenTof thepolar vortex condition. This indicates that for a strongEPW case, the AL-IL seesaw was mainly caused byhorizontal propagation of the EPW from East Pacificthrough North America, before turning toward Icel and . Sun and Tan(2013)provided evidence of thehorizontal and vertical propagation of EPW, supporting the above statistical results.

Sun and Tan(2013)argued that the high inversecorrelation between the AL and IL actually depictedthe NAO, while the Pacific center in the AO may hintat the presence of the AL-IL seesaw. To confirm theirargument, Sun and Tan(2013)performed EOF analyses with the months of strong AL-IL seesaw and theremaining months. The first EOF showed a strongerPacific center than the climatology for the former case, and the Pacific center disappeared from the EOF pattern for the latter case. This implied that the AO wasreally an imprint left by both the AO and the AL-ILseesaw. 6. Further consideration of the factors associated with atmospheric teleconnection6. 1 Formation of temperature anomaly associated with AO

In the geopotential height field, the AO has a pattern of three anomalous centers: the Artic center, theNorth Atlantic Center, and the North Pacific center; while in the lower-level temperature field, the AO pattern has three anomalous cooling centers over the Arctic Ocean, Greenl and, and northeastern Canada, respectively, and two warming centers over Eurasia and the United States of America. Thompson and Wallace(1998)showed that the temperature anomalies maybe caused by temperature advection induced by zonalmean anomalous zonal wind. Enhanced westerlies associated with the positive AO phase bring more warmmarine air into the interior of continents, warming theair there, and pushing cold continental air into thewestern oceans, cooling the air there. Their explanation emphasized the role of l and -sea thermal contrast. Some weakness, however, still exists in their explanation. First, theYused the zonal mean anomalouszonal wind, not the anomalous zonal wind itself, tocalculate the anomalous temperature advection, whichcannot reflect adequately the local characteristics ofthe temperature anomalies. Second and more importantly, they did not consider the role of temperatureadvection induced by the anomalous meridional wind and the temperature contrast between the higher and lower latitudes, which is the most dominant temperature gradient. In view of this argument, and based onNCEP/NCAR reanalysis data, Suo et al. (2009)recalculated the temperature anomalies for a positive AO, and obtained a better result. The temperature advection by the anomalous zonal wind north of 40°broughtwarm ocean air over the continents and cold l and airover the oceans, dominantly contributing to the cooling over Greenl and, its surroundings, and the BeringSea. The anomalous meridional wind assumed a wavelike form with wavenumber 3, which brought warm airnorthward and cold air southward, and caused warming in the United States and cooling in southern Europe and Canada. The temperature advection by boththe zonal and meridional wind anomalies caused thewarming in northern Eurasian continent. 6. 2 Mechanism of poleward propagation of theanomalous westerlies over North Atlantic

Wave-mean flow interaction is coherently a mechanism responsible for low-frequency variability on various timescales. The poleward propagation of theanomalous westerlies is a direct consequence of wavemean flow interaction. Examination of this polewardpropagation aids underst and ing of the mechanism ofthe AO and Antarctic oscillation. Evidence of poleward propagation was first shown in the observ ationalstudy of Riehl et al. (1950). James et al. (1994)foundthat similar poleward propagation could also occur inan idealized model. With a barotropic weakly nonlinear model, James and Dodd(1996)suggested that thepoleward propagation originated from the interactionbetween the mean flow and equatorward-propagatingRossby waves from mid and high latitudes. The eddymomentum flux convergence at the poleward flank of apositive zonal wind anomaly caused the positive zonalwind anomaly to shift poleward. Robinson(2000)further argued that a poleward shift in low-level baroclinicity may in fact play an important role in drivingthe poleward propagation. The above two mechanismswere confirmed by the observ ational study of Feldstein(1998).

The modeling study of Lee et al. (2007)showedthat the poleward propagation was initiated in thetropics by the breaking of Rossby waves generated atmidlatitudes. This wave breaking deposits negativezonal momentum and decelerates the ambient zonalwind. This results in a poleward shift in the locationof the critical latitude and a reduction in the meridional potential vorticity gradient. As a result, subsequent midlatitudinal waves break poleward of theprevious critical latitude. In this manner, the zonalmean flow anomalies propagate poleward. Recently, Yuan et al. (2013)further confirmed with the ERA40data the modeling resulTof Lee et al. (2007). 6. 3 Spatial and temporal scales of atmosphericteleconnection

Most past studies on low-frequency climate variability are based on monthly or seasonal mean data(Wallace and Gutzler, 1981; Barnston and Livezey, 1987). This time average is su°cient for underst and ing anomalies that have a timescale longer thantwo months, such as seasonal, interannual, or longertimescales. However, dailYunfiltered data have latelybeen used to study the intraseasonal variability ofthe teleconnections(Feldstein, 2002; Athanasiadis et al., 2010; Johnson and Feldstein, 2010). Feldstein(2002) and Johnson and Feldstein(2010)found someteleconnections, such as the PNA and NAO, havetimescales of around 10 days, in good agreement withthe timescale obtained with other approaches such asnumerical modeling(Cash and Lee, 2001; Feldstein, 2002).

As for the spatial shapes of the teleconnections, the results obtained with different approaches(e. g., point correlation and EOF analysis)do not agree, which means that the shapes of the teleconnectionsmay not be unique. Johnson and Feldstein(2010)applied a cluster analysis and a self-organizing map(SOM)analysis to the study of teleconnections basedon unfiltered daily data. They found that there existeda number of PNA-like patterns in the sea-level pressure field. These PNA-like SOMs possessed two polarities, each corresponding to one phase of the PNA. The timescale of the PNA-like SOMs was also around10 days. Using this method, they tried to relate thePNA to convective heating over the tropics. Similarly, Johnson et al. (2008)provided an explanation on theformation of the decadal variation of NAO, and Lee and Feldstein(2013)detected the relationship betweenozone depletion, forcing of greenhouse gases, and thesouthern annular mode. 6. 4 Spatial relation of jets and storm tracks

Jets and storm tracks are closely related, and always interact with each other. Jets can trigger synoptic waves through barotropic or baroclinic instability, forming storm tracks, while the synoptic wave can, in turn, supply energy to the planetary wave or themean flow for their development. Therefore, to studythe relation of storm tracks with jets benefits our underst and ing of teleconnection formation mechanisms.

It has long been known that in both the Northern and Southern Hemispheres, the storm tracks and their associated jets do not coincide completely, withthe peak of storm tracks shifted poleward about 10°latitude apart. Yang et al. (2007)studied the baroclinic wave-mean flow interactions in an extendedquasi-geostrophic two-layer model in which two kindsof nonlinearity, nonlinear advection of vorticity and vorticity-divergence production, were included. Theyfound that the former kind of nonlinearity led to downstream developmenTof the baroclinic wave packets, while the second kind of nonlinearity was responsiblefor the asymmetry in the north-south flanks of theupstream parTof the wave packets, which caused thepoleward shifTof the center of the storm track. 6. 5 Energetics and variability of storm tracks

Storm tracks vary in position and intensity ontimescales ranging from seasonal(Lau, 1988; Nakamura, 1992; Christophy et al., 1997)to interannual(Trenberth and Hurrell, 1994; Straus and Shukla, 1997; Zhang and Held, 1999) and decadal(Ebisuzaki and Chelliash, 1998; Nakamura and Izumi, 1999; Geng and Sugi, 2001; Chang and Fu, 2002). Chang et al. (2002)found that storm tracks were strongly modulated by ENSO on the interannual timescale. DuringEl Niñ o years, enhanced Hadley cell drove the stormtrack over North Pacific to shift equatorward and eastward. During La Niñ a, the reverse was true. However, the ENSO impacTon the storm track over NorthAtlantic was weak. It is well known that AO isthe strongest signal in the extratropical regions, so itseems reasonable to assume that AO has a significantinfluence on the Northern Hemisphere storm tracks. Nie et al. (2008)studied this assumption and foundthat the North Atlantic storm track shifted northward and eastward during strong positive AO winters, inagreement with McCabe et al. (2001). A similar shiftwas also detected in the zone of baroclinic energy conversion, caused by planetary wave-baroclinic wave interaction. In contrast, the North Atlantic storm trackretreated westward during strong negative AO winters. They also found that the baroclinic waves moved eastward along middle Atlantic, while during strong negative AO winters, the storm track split into northern and southern branches. The impact associated withthis bifurcation on weather and climate over Eurasia and China needs further study . 7. Summary and discussion

This paper has reviewed the progress made inrecent years in the study of the dynamics of atmospheric teleconnection, particularly the teleconnectionpatterns that have had a significant impacTon China. The key points are as follows.

(1)The impacTof solar activity on the East Asianclimate was documented. The solar cycle was shownto lead to changes in the spatial structure of theAO/NAO, via its influence on the temperature and winds in the stratosphere, and induce anomalous EastAsian climate in winter or spring. Moreover, the 11-yr solar cycle also modulated the relationship betweenthe spring AO/NAO and the following East Asiansummer monsoon. The possible mechanism of thismodulation was suggested to be solar activity impacting the air-sea interaction in North Atlantic.

(2)The relationship between the stratospheric polar night jeT and the circulation over East Asia wasestablished. After 1976, the polar night jet strengthened, suppressing vertical propagation of the signal associated with the anomalous circulation over the Uralstoward the stratosphere. More signals propagated inthe troposphere, influencing the circulation over EastAsia. Associated with this change, the winter cyclone and anticyclone activity over East Asia increased(decreased)after(before)the mid 1970s. Particularly after the mid 1980s, the upper polar-front jet was significantly enhanced and the frequency of cyclone and anticyclone activities also reached a peak.

(3)Three Eurasian teleconnections(EU, SCAND, and EA TL/WRUS)were defined, all with a clear seasonal variation. They were weak during summer, butstrong during autumn and winter. These three teleconnections were associated with different externalforcing sources, and had different impacts on the winter climate in China.

(4)The link between the teleconnections in mid and high latitudes and the tropical atmosphere wasdiscussed. On the seasonal to decadal timescales, the tropical convective heating influenced the NAOthrough circumglobal wave trains, and the reverse wasalso true. In addition, anomalous teleconnection patterns in the extratropics, such as the NPO, were shownto influence the occurrence of ENSO in subsequentseasons via the SFM. The spring AO-associated circulation anomalies were supported by the interactionbetween synoptic-scale eddies and mean flows, and theanomalous westerly wind correspondingly formed inthe equatorial western Pacific. These equatorial westerly anomalies excited downwelling equatorial Kelvinwaves, leading to SST warming in the central-easterntropical Pacific in summer. The tropical SSTanomalies sustained and developed through the Bjerknes positive feedback mechanism, and thus, an El Niñ o eventoccurred in fall and winter. On the other h and, thespring AO-related SSTanomalies were shown to eitherincrease or decrease the previous winter NPO-relatedSSTanomalies over the subtropical North Pacific, in-fluencing the westerly wind anomalies over the tropical western Pacific in summer via the air-sea interaction. Hence, the SSTanomalies forced by the SFMin the equatorial central-eastern Pacific were affected. The primary influence was on the occurrence of theENSO.

(5)The relation between the EPW and PNA wasdiscussed, and it was shown that the EPW played animportant role in the AL-IL seesaw through its horizontal and vertical propagation and reflection by thepolar vortex. Both the NAO and PNA contributed tothe AO formation.

(6)The atmospheric teleconnections, such asPNA, WP, and EP, could be resolved into a number of basic structures called SOMs. These SOMswere similar in structure and had two polarities. Eachgroup of the SOMs with the same polarity constitutedone phase of the related teleconnection. The SOMshad an intrinsic timescale of approximately 10 days, and the variability on longer timescales such as seasonal, interannual, or decadal, could be obtained bystatistical averaging of the SOMs' frequency .

Although progress in the study of the dynamics of the teleconnections in extratropical regions hasbeen achieved, most work has focused on descriptionsof the phenomenon. Further studies on mechanismsare still needed, especially those that can address thefollowing questions.

(i)Past studies have focused only on relationships between the ENSO, AO/NAO, and NPO(e. g., Chen et al., 2013b; Chen et al., 2014a). Evidently, further studies on the interaction between differentteleconnections in mid and high latitudes, and theirintegrated impacts on climate, are needed.

(ii)Several studies(e. g., Chen and Li, 2007; Wei et al., 2007; Chen and Wei, 2009)have revealed thelink of the AO, EU, and PNA to the stratosphere. TheENSO in particular can influence the stratosphere circulation, which in turn influences the teleconnectionsin mid-high latitudes in the troposphere. However, the mechanisms are yet to be determined.

(iii)The factors, such as local wave-mean flowinteraction, convective heating in the tropics, changesin polar sea ice, strength of the polar vortex, and activity of solar spots, which determine the frequency ofthe SOMs, are still unclear.

Acknowledgments: We thank the two anonymous reviewers for their constructive suggestions and comments, which lead to a significant improvement in the paper.