ENSO-Independent Contemporaneous Variations of Anomalous Circulations in the Northern and Southern Hemispheres:The Polar-Tropical Seesaw Mode

TANG Weijy(唐卫亚) and GUAN Zhaoyong(管兆勇)

1 Collaborative Innovation Center on Forecast and Evaluation of Meteorological Disasters,Nanjing University of Information Science&Technology,Nanjing 210044

2 State Key Laboratory of Severe Weather,Chinese Academy of Meteorological Sciences,Beijing 100081

3 Key Laboratory of Meteorological Disaster of Ministry of Education,Nanjing University of Information Science&Technology,Nanjing 210044

4 Polar Climate System and Global Change Laboratory,Nanjing University of Information Science& Technology,Nanjing 210044

ENSO-Independent Contemporaneous Variations of Anomalous Circulations in the Northern and Southern Hemispheres:The Polar-Tropical Seesaw Mode

TANG Weijy1,2,3(唐卫亚) and GUAN Zhaoyong1,3,4*(管兆勇)

1 Collaborative Innovation Center on Forecast and Evaluation of Meteorological Disasters,Nanjing University of Information Science&Technology,Nanjing 210044

2 State Key Laboratory of Severe Weather,Chinese Academy of Meteorological Sciences,Beijing 100081

3 Key Laboratory of Meteorological Disaster of Ministry of Education,Nanjing University of Information Science&Technology,Nanjing 210044

4 Polar Climate System and Global Change Laboratory,Nanjing University of Information Science& Technology,Nanjing 210044

Using the NCEP/NCAR reanalysis and the ENSO indices from the Climate Prediction Center over the period 1978–2014,we have investigated the contemporaneous circulation variations in the Northern and Southern Hemispheres by performing the singular value decomposition analysis of sea level pressure anomalies(SLPA)after the ENSO signal is regressed out.It is found that there exists a polar-tropical seesaw mode(PTSM)that characterizes with the out of phase fluctuations of SLPA between the polar and tropical regions in the Northern and Southern Hemispheres in boreal winter.This PTSM explains 47.74% of the total covariance of SLPA and is almost independent of ENSO.It demonstrates a long-term trend and oscillation cycles of 2–3 and 4–6 yr.The long-term trend in PTSM indicates that the sea level pressure gradually decreases in the tropics and increases in the polar region with time.This PTSM looks roughly symmetric about the equator besides the seesaw pattern of SLPA between the tropics and polar region in each hemisphere.The disturbances in the geopotential height field in association with the PTSM shows baroclinic features in the tropics whereas equivalent barotropic features in the mid and high latitudes in the troposphere.The anomalous thermal forcing in the tropical region is possibly one of the factors facilitating the formation of this PTSM.Significant global precipitation and temperature anomalies related to the PTSM are observed.In the positive PTSM phase,precipitation and temperature are higher than normal in southern Europe and the Mediterranean and surrounding areas,but lower than normal in northern Europe and Siberia.Precipitation is higher than normal while temperature is lower than normal in Northeast Asia. Significant temperature and precipitation anomalies possibly occur in the regions of western China,northern India,parts of North America,parts of subtropical Africa,Maritime Continent,and Antarctic.These results are helpful for better understanding of the circulation variations and the mechanisms behind the interactions between the Northern and Southern Hemispheres and the related winter climate anomalies over globe.

Northern and Southern Hemispheres,polar-tropical seesaw mode,climate anomaly,boreal winter

1.Introduction

In recent decades,researchers have paid more attention to the atmospheric circulations on hemispheric and global scales because of increases of global observations and rapid development of global climate mod-

els.Our understandings of the variations of global atmospheric circulation and regional climate along with the mechanisms behind have been hence deepened, based on the research progresses in the global circulation changes(e.g.,Christy et al.,1989;Kusunoki et al.,2006;Xiao et al.,2010;Weng et al.,2011;Bai et al.,2012),climate simulations using various global climate system models(e.g.,Giorgi,1990;Thompson and Pollard,1995;Man and Zhou,2011;Zhang,2012; Chen et al.,2014,Zhou et al.,2014),simulated global circulation variations in the context of Atmospheric Model Intercomparison Project(AMIP),Ocean Model Intercomparison Project(OMIP),and Coupled Model Intercomparison Project(CMIPS)programs(Meehl et al.,2007;Wang et al.,2008;Taylor et al.,2012;Li et al.,2013;Shao et al.,2013;Zhang et al.,2013),and interactions between the Northern Hemisphere(NH) and Southern Hemisphere(SH)(Findlater,1969;Carleton,1992;Guan and Yamagata,2001;Zeng and Li, 2002;Huang et al.,2003).Though significant progresses have been made,it is still far from a fully understanding of the complicated variations of atmospheric circulations and the mechanisms behind.

It is well known that the atmospheric circulations vary differently between the NH and SH.The major patterns of the complicated atmospheric circulation variations in the NH are considered to be independent of those in the SH.This can be attributed to the lateral boundary effect of equator where the Coriolis force is zero,the different land-sea contrasts,the asymmetry of solar radiation except in the spring and autumn equinoxes,and the different ocean-atmosphere interactions between the two hemispheres.This is manifested by some strong variabilities in different hemispheres. These are,for example,the North Atlantic Oscillation(NAO)/Arctic Oscillation(AO)(Thompson and Wallace,1998;Wallace,2000;Watanabe,2004)in the mid and high latitude regions in the NH,the Pacific Decadal Oscillation(PDO;Mantua et al.,1997), Antarctic Oscillation(AAO)in the SH(Gong and Wang,1999),and various other teleconnection patterns in each hemisphere(Wallace and Gutzler,1981; Hoskins et al.,1983;Shi and Zhu,1993;Ding et al., 2014;Tan and Chen,2014),indicating the independent circulation variations in the two hemispheres.

However,there exist some contemporaneous variations of circulations in the two hemispheres,and even these variations are highly correlated.This is because of some factors including the earth's rotation and other external forcing such as the solar radiation and the ENSO,as well as the cross-equator flow and its subsequent lateral coupling,which provide the possibility for the atmospheric interaction between the two hemispheres(Wang et al.,2013).From the perspective of global dry atmospheric mass conservation,the inter-hemispheric oscillation(IHO;Guan and Yamagata,2001;Guan et al.,2010)is,to a certain degree, another linkage of the circulation variations between the two hemispheres.

As the strongest interannual variability in the equatorial Pacific region,ENSO can induce anomalous changes of atmospheric circulation not only in the tropics,but also in the mid and high latitudes(e.g., Huang et al.,2003;Ashok et al.,2007;Gushchina and Dewitte,2012).For example,the Pacific North America(PNA)teleconnection is triggered by ENSO events (Wallace and Gutzler,1981)in the NH whereas the Pacific South America(PSA)pattern(Carleton,1992)in the SH.Such contemporaneous variations of anomalous circulations in the two hemispheres are considered to be an important mode in the global domain on the interannual timescales.It is found that the ENSO-related anomalies of the near surface circulation that are roughly symmetric about the equator show an east-west distribution with dipole or triple structure in the low latitudes of Indo-Pacific sector. However,some asymmetric components about equator are also observed.These asymmetric circulation anomalies about equator vary with time,which are affected by,and affect in turn the global weather and climate variations and even the global changes.Nonlinear interactions between the symmetric and antisymmetric components of the atmospheric circulation variations may also play an important role in the symmetry of the atmospheric circulation anomalies.Without any doubt,both the symmetric and asymmetric components of circulation anomalies are very important in shaping our climate in global domain(Guan et

al.,1994;Wang et al.,1994).However,until now the major circulation variation patterns in the two hemispheres that are highly correlated and change contemporaneously,symmetric or not,have not been fully revealed yet.

Since the AO and AAO are two strong signals independent of each other,it seems that,besides IHO, the anomalous circulations respectively in the NH and SH must change contemporaneously as a result of the ENSO-related thermal forcing in tropics.However, contributions of ENSO only accounts for a part of the total covariance of the contemporaneous changes in the two hemispheres(Chen,2002;Yuan et al.,2012, 2014;Chen et al.,2013).Thus,is it possible to find the physical mode of circulation anomalies that is independent of ENSO varies contemporaneously in the NH and SH?To answer this question,we reveal this mode in the present study by performing the singular value decomposition(SVD)for the sea level pressure anomalies(SLPA)after the ENSO signal is filtered out.The results will be beneficial for us to better understand the mechanisms of circulation co-variations between the SH and NH and their possible influences on global climate anomalies.

This paper is organized as follows.A brief description of data sources and methodology employed in this study are given in Section 2 after a brief introduction to study purpose.The polar-tropical seesaw mode(PTSM)is explored in Section 3.In Section 4, the structure of this seesaw mode along with its dynamics is investigated.In Section 5,the possible influences of this PTSM on boreal winter climate including surface temperature and precipitation are examined. Section 6 gives the conclusions and discussion.

2.Data and methodology

The data used in this study include(1)the NCAR/NCEP monthly mean reanalysis with a resolution 2.5°(lon.)×2.5°(lat.)in horizontal and 17 isobaric levels in vertical in global domain(Kalnay et al., 1996).The quantities are surface variables including sea level pressure(SLP),surface pressure,and surface temperature,and isobaric level variables including the geopotential height,specific humidity,winds,and air temperature;(2)NOAA global reconstructed sea surface temperature(SST)with a resolution of 2.0°(lon.) ×2.0°(lat.)on a grid mesh;and(3)CMAP precipitation dataset(Xie and Arkin,1997),AO indices, AAO indices,and SST indices over various Ni˜no regions from NOAA Climate Prediction Center(CPC). All the datasets cover the 37-yr period from 1978 to 2014,except that the CMAP precipitation dataset is for the period 1979–2011.Winter is defined to be December,January,and February(DJF)in the NH.Wintertime average refers to the average over DJF.

Methods used in this study are the SVD analysis, the wavelet analysis,and correlation and regression analysis.

3.Contemporaneous circulation changes independent of ENSO

3.1The polar-tropical seesaw mode(PTSM)

The SLPA is a good indicator that describes the variations of circulations and even the related climate anomalies.The SLP anomalies are employed to get the major features of contemporaneous circulation changes in the two hemispheres independent of ENSO. The ENSO signal is removed from SLPA by regressing SLPA onto the Nino3.4 SSTA(sea surface temperature anomalies)index for 36 winters from 1978/1979 to 2013/2014.Let INino3.4be the normalized timeseries of Nino3.4 SSTA index and PSLabe the SLPA. Then we have

whereµis the regression coefficient,and PSLar(denoted by SLPAr hereafter)is the remainder part of SLPA after having the ENSO-related component being removed.Apparently,SLPAr is simultaneously independent of ENSO in statistical sense.

The SVD is performed of the variations of SLPAr. The SLPAr to the north(south)of 2.5°N in the NH (SH)is taken as the left(right)field for the SVD analysis.Statistics of SVD modes are presented in Table 1. The two leading modes of the SVD account for 47.74% and 15.07%of the total of squares of covariance respectively,and the cumulative covariance proportion is up to 62.81%.They represent the spatial modesthat correspond to the most significant contemporaneous SLPAr changes in the two hemispheres during the wintertime.

Table 1.Statistics of the two leading SVD modes for the SLPAr in the two hemispheres

The contemporaneous SLPAr changes in the two hemispheres that are independent of ENSO are highly correlated;the correlation coefficient is 0.73(P ＜0.05)for the leading SVD mode 1.The spatial distribution of SLPAr in each hemisphere(Figs.1a and 1b)demonstrates an anti-phase feature in the low and high latitude regions,and the SLPAr in the two hemispheres are roughly symmetric about the equator.This is an interesting phenomenon.This mode as revealed by SVD1 is referred to as the PTSM that is characterized by the contemporaneous SLPAr changes in the NH and SH with the seesaw structure of SLPAr between the low and high latitude regions.

The spatial distribution of SVD1(Figs.1a and 1b)displays negative SLP disturbances in the lower latitudes whereas positive in the mid and high latitudes,indicating a seesaw structure in meridional. Large negative disturbance centers are located in North Africa,South Asia,the Maritime Continent (MC),and the equatorial East Pacific.In the high latitudes,the SLP anomalies are largely positive;the positive SLPA centers are located near the polar circle or in the polar region.For the convenience in the following sections,the positive phase(negative phase) of PTSM is defined as the phase when the time coefficient of SVD1 is positive(negative)along with the positive(negative)SLPAr in the polar region while negative(positive)SLPAr in the tropics.

Despite the overall symmetric distribution of PTSM about equator,there still exists large differences in its spatial distribution between the two hemispheres(Figs.1a and 1b).Comparing the SLPAr between the two hemispheres,it can be found that the SLPAr is more zonally symmetric in the NH.Areas where the larger anomalies of SLPAr appear in the NH are broader than those in the SH;in the SH, variations of SLPAr look to be more localized.Such a difference in the spatial distribution of SLPAr might be related to the different land-sea contrasts between the two hemispheres.It may also be relevant to the difference in seasons,i.e.,the NH winter corresponds to the SH summer.

Note that the spatial distribution of the PTSM revealed in the present study is quite different from the results of previous studies about the IHO(Guan and Yamagata,2001;Lu et al.,2008;Guan et al., 2010;Lu et al.,2013).The IHO is a physical mode that describes the simultaneous variations of circulation anomalies in the NH and SH from the perspective of hemispherical-scale atmospheric disturbance(Guan et al.,2010).The physical basis for the IHO is the redistribution of the atmospheric mass in the two hemispheres under the constraint of global dry atmospheric mass conservation(more in the SH and less in the NH,and vice versa).Without any doubt,this IHO mode is important for us to better understand the circulation interactions between the NH and SH.However,the IHO is actually the third mode out of all the EOF(empirical orthogonal function)modes that we obtained from EOF decomposition of the surface pressure anomalies.Considering the total variance of the atmospheric disturbance,we need to find the dominant mode of the contemporaneous circulation changes in the two hemispheres.The PTSM seems like this dominant mode that accounts for 47.74%of total covariance of SLPAr.Explicitly,the PTSM is definitely different from the IHO that the results from the cross-equatorial atmospheric mass exchange between the two hemispheres.Instead,the PTSM in the present study reveals a component of circulation anomalies that varies contemporaneously in the two hemispheres when the ENSO effects have been regressed out.

The correlation coefficient between the time coefficients of the right and left fields of SVD1 is 0.73, suggesting that the SLPAr changes in the NH and SH

are still synergetic even after the ENSO signal is removed.Figure 1c shows clearly that changes in SLPAr are dominated by interannual variability in both the NH and SH.Wavelet analysis of the time series of SLPAr in both hemispheres displays that the major period has changed from 2–3 to 4–6 yr at the beginning of the 21st century.A long-term trend is found in the time series of the time coefficient(Fig.1c)for the most recent 30 years since 1978,indicating that SLP is decreasing in the tropics while increasing in the polar region during the boreal winter in the past 30 years.This result is clearly consistent with the fact that the global warming is more distinct in the polar region than in other regions(IPCC,2014;Li et al., 2014).

The meridional structure of PTSM looks seesawlike.Figure 1d depicts a clear meridional seesaw between the SLPAr in the polar region and in the tropics,which is distinct in both hemispheres.The seesaw between the polar and near-equator regions in both hemispheres looks clear,but in general,it is stronger in the NH than in the SH.The reversal zone for the SLPAr from positive to negative is located between 50°and 55°N in the NH whereas between 55°and 60°S in the SH.

Fig.1.Heterogeneous correlations of SLPAr for mode SVD1 in(a)the NH(left field)and(b)the SH(right field)along with(c)the corresponding time series of coefficients,and(d)the zonal averages of SVD1.Contour intervals are 0.1 in (a)and(b).The critical value at the 95%level of confidence is found to be 0.32 by using a t-test.

In order to reconfirm the PTSM mode as a physi-

cal entity,here we(1)analyze the pattern correlations, i.e.,the pattern similarity correlation coefficients,between the SVD1 and SLPA to examine whether the SLPA displays a pattern similar to SVD1 in certain years;and(2)obtain the interannual variation of regionally averaged SLPAr to examine whether there exists an anti-phase relationship between the SLPA changes in the lower and higher latitudes.

The spatial pattern correlations between the SVD1 and SLPA in the NH and SH are calculated, respectively. If the correlation coefficient is large (|R|>0.55),then the spatial patterns of SLPA and SVD1 left(right)field are considered to be quite similar.Figure 2a shows the normalized time series of the similarity coefficient between the spatial patterns of SVD1 and SLPA.It is seen from Fig.2a that two time series for the NH and SH respectively are quite similar;the correlation of the time series of similarity coefficients in the NH with that in the SH is found to be 0.57.As what we expected,the correlations of time series of similarity coefficients are highly correlated with the time series of coefficients of SVD1 in both hemispheres;both the correlation coefficients are larger than 0.93(P＜0.05).These results indicate that the SVD1 is indeed a mode describing the contemporaneous changes in SLPA in the NH and SH.Table 2 lists the years when the spatial patterns of SLPA and SVD1 are similar.Further examination verifies that the SLPA distribution in these years(figure omitted) is exactly like that shown in Figs.1a and 1b.Note that those years listed in Table 2 include years when ENSO is in its cold phase,warm phase,and neutral phase,indicating that the PTSM mode is independent of ENSO.

Based on the spatial patterns shown in Figs.1a and 1b,three regions,i.e.,55°–90°N,30°S–30°N,and 60°–90°S,are roughly selected,over which the areal weighted means of SLPAr are computed.The results are shown in Fig.2b.It is found that when the positive SLPA occurs in the low latitudes,the negative SLPA will simultaneously appear in high latitudes of both hemispheres.That is to say,the SLP change in low-latitudes is highly negatively correlated with the SLPA in high-latitude regions of both hemispheres.In

fact,it is found that the correlation coefficients are found to be–0.77 and–0.41(P＜0.05)in the NH and SH,respectively. These results again suggest that the SLP in high-latitudes will adjust correspondingly when it changes in low-latitude region.Moreover,these suggest that the variations of SLPA in lowlatitudes work as a physical linkage that links the variations of the anomalous circulations in high latitudes between the two hemispheres.

Fig.2.Normalized time series of(a)the pattern correlations between mode SVD1 and SLPAr with red(blue)lines for the NH(SH),and(b)DJF mean SLPAr averaged over zonal regions of 55°–90°N(red),30°S–30°N(black),and 60°–90°S(blue),respectively.

Table 2.The years when the spatial patterns of PTSM and SLPA are similar to each other as identified by pattern correlation coefficient with its absolute value larger than 0.55

3.2Independence of PTSM to ENSO

In order to reconfirm the independence of PTSM to ENSO,we analyze first the contribution of the PTSM and ENSO to SLPA.Figure 3 shows the contributions of SVD1-related component of SLPA and those of the Nino3.4 index-related component of the SLPA to the total variance of the SLPA.It is seen from Fig.3 that the large contributions of PTSM to the variance of SLPA are found in high-latitude regions of the two hemispheres,while the large contributions of ENSO-related variability are found in the tropics. Zonally averaged contributions of SLPA component in association with the PTSM and ENSO to the total SLPA variance are listed in Table 3.Again,large contributions of the PTSM to total variance of SLPA appear in high latitudes of the two hemispheres whereas the contributions of ENSO are mainly observed in the tropics. Specifically in the NH,the PTSM-related SLPA variability accounts for up to 70%of the total SLPA variance in regions around the Barents Sea and Novaya Zemlya in high latitudes and 60%in the western Greenland.It also accounts for up to 70%of the total SLPA variance in North Africa,the Mediterranean region,and southern Europe(Fig.3a).However,only about 30%of the total variance of SLPA can be explained by PTSM-related SLPA variability

in East Asia.In the SH,the large contributions of PTSM-related SLPA are found in high latitudes to the south of 60°S(Fig.3b).

Fig.3.Percentages of the total SLPA variance explained by the Nino3.4 index-related variability(shadings)and those by PTSM-related variability as obtained by regressing SLPA onto the time series of coefficients of SVD1(contours)for (a)the NH and(b)the SH.

The above shows that the PTSM can explain 40%–70%of the total variance of SLPA in high latitudes,but only 10%–40%of the total variance in low latitudes.Apparently,in addition to ENSO,the PTSM is one important mode that describes the SLP variations;it reveals that the tropical SLPA oscillation inherently links the circulation variations in mid and high latitudes of the two hemispheres via redistribution of the atmospheric mass between lower and higher latitudes,the meridional vertical circulations,and the propagation of Rossby waves excited by anomalous forcing in tropical region(Sardeshmukh and Hoskins, 1988).In this way,the close relationship between variations of circulation anomalies in high latitudes of the two hemispheres is hence established.In other words, the PTSM actually explores the seesaw-like anti-phase variations of SLP between tropical low air pressure region and polar higher air pressure region.

The anomalous thermal forcing in tropical region may play a crucial role in the formation of PSTM. It is found that the correlation of time coefficient of the SVD1 with the time series of surface air temperature(surface air pressure)anomalies averaged over 20°S–20°N is 0.46(–0.86),suggesting that the air column anomalously expands and air pressure correspondingly decreases in the tropics.However,in the midlatitudes,anomalous surface air temperature is simultaneously anti-correlated with the time coefficient of the SVD1 with the correlation coefficient of–0.44. More than this,the anti-correlation gradually intensifies poleward.

Based on the aforementioned,the physical meaning of the PTSM is that,when the low air pressure in the tropics changes due to abnormal adiabatic forcing of ocean-atmosphere interaction and/or cloud radiative effects,pressure in the mid and high latitudes will adjust correspondingly for at least the changes in the atmospheric mass will remain balanced on the hemispheric scale.Of course,the reversed situation may also happen,i.e.,air pressure changes in the mid and high latitudes induced by some anomalous inter-nal/external forcing such surface albedo anomalies may result in changes of air pressure in low latitudes.

Table 3.Zonal means of the percentage contribution as shown in Fig.3

It is known that the atmosphere interacts with the oceans in both tropics and higher latitudes.When the significant SSTA occurs in middle and east equatorial Pacific,the atmospheric circulation will respond to this SSTA forcing with a time lag by at least one month.Particularly,the tropical Indian Ocean is usually believed to be a slave to the Pacific,whose SSTA changes are a response to the SSTA changes in east equatorial Pacific with a lag of about six months(e.g., Ashok et al.,2003).In this way,the variations of atmospheric circulation anomalies may have a significant lag correlation with Nino3.4 index.Then,is the PTSM still independent of ENSO?To answer this question, here we calculate the correlation coefficients between various ENSO indices and SVD1 time series of coefficients(Table 4).It shows in Table 4 that no matter whether the PTSM advances ENSO by 1–3 seasons,or is of the same period of ENSO,or lags ENSO by 1–3 seasons,its correlation coefficients with all the ENSO indices are very small.Since ENSO has a cycle of 3–7 yr,we also examine the correlations of wintertime PTSM with wintertime ENSO indices that are either

leading or lagging 1–3 yr(Table 5).No large correlations are found in Table 5.Thereby,it is reasonable to deduce that the PTSM and ENSO are independent of each other from the perspective of statistics.In addition,the correlation between the PTSM and IHO is also very low(Table 4).These are in agreement with what we have discussed previously in the present study,reconfirming the PTSM independent of ENSO.

It is worth noting that,despite the different spatial distributions of extreme values of SLPAr and the different reversal latitudes where the SLPAr changes from positive to negative,the PTSM is still highly correlated with AO and AAO and the correlation is relatively high(Table 4).This implies that there exist certain relationships between the PTSM and AO/AAO. However,if both ENSO and AO/AAO signals are filtered out by using the regression analysis,the PTSM mode can still be obtained by using the SVD analysis of abnormal SLP,except that the anomalies of SLP in the SH are much weaker than those in the NH.

4.Spatial structure of the PTSM

4.1 Anomalous horizontal circulations

The horizontal pattern of the ENSO-independent PTSM in SLPA as displayed in Fig.1 reveals the major mode of contemporaneous changes of circulation anomalies in the NH and SH.However,the whole features of this PTSM at different isobaric levels have not exhibited till now.As the SLPAs vary as a result of both the variations of anomalous atmospheric circulations above the earth surface and the abnormal surface thermal forcing,the structure of PTSM above sea level may show some interesting features.The spatial features of PTSM at different heights can be examined by using the regression analysis.In this way,we have obtained the anomalous circulations at levels 850,500, and 200 hPa(Fig.4).

Similar to the aforementioned,the abnormal circulation patterns at various levels of the troposphere look also roughly symmetric about the equator,particularly in the mid and high latitudes.This can be seen from Figs.4a and 4c in places including the area east of the Meridian,the area nearby 60°E,the Asian-Australian region,the western Pacific,area east of the dateline,and the American region.Over the Asian-Australian region(90°–120°E),the anomalous anticyclone,cyclone,and anticyclone are found from the equator to the polar region in both hemispheres,and they are roughly symmetric about the equator.These anomalous circulations intensify with height with an equivalent barotropic structure.In the region nearby

150°W to the east of the dateline,the anomalous cyclone,anticyclone,and cyclone are distributed alternatively in the meridional direction and also roughly symmetric about the equator in the NH and SH.Similarly,these systems are also equivalent barotropic in the vertical and intensify with height.

Table 4.Lag correlations of the time series of coefficients of SVD1 with various indices of ENSO,AO,AAO,and IHO

Table 5.Lag correlations of the time series of coefficients of SVD1 with various wintertime ENSO indices that are either leading or lagging 1–3 yr

Fig.4.Anomalous winds and geopotential heights as obtained by regressing these quantities onto the time series of coefficients of SVD1.(a)Rotational(streamlines)and divergent(vectors;m s−1)wind components at 850 hPa with shades in blue for anomalous convergence.(b)Anomalous wind(vectors;m s−1)and geopotential height(contour;gpm) at 500 hPa.Bold arrows and grey shaded areas are for values at/above the 95%confidence level by using an F-test.(c) Anomalous rotational(streamlines)and divergent(vectors;m s−1)wind components at 200 hPa with shadings in red for divergence anomalies.

Relatively large differences are observed in the anomalous circulations related to the PTSM between the SH and NH although the overall appearance of anomalous circulations is symmetric about the equator.The wintertime abnormal circulation in the mid and high latitudes is stronger in the NH than that in the SH.And disturbances in the mid and high latitudes in the NH are more zonal than those in the SH. In the high latitudes in the SH,zonal disturbances are relatively weak.

The anomalous circulations over the midlatitude Eurasia,Pacific,and Atlantic are probably associated with some teleconnection patterns in the NH.For example,the disturbances occurring in West Europe, the Ural Mountains,and the coastal region in East Asia are probably associated with the Eurasian(EU) teleconnection pattern;disturbances occurring in the North Atlantic including its subtropical region exhibit an East Atlantic(EA)teleconnection pattern.It is interesting to note that even after the Nino3.4 signals are filtered out,a horizontal wave train similar to the PNA teleconnection pattern still exists from the central-eastern Pacific to North Pacific and then to North America and finally to the Gulf of Mexico (Wallace and Gutzler,1981).In the SH,a wave train with the structure similar to the PSA pattern also occurs over the southern Pacific and Atlantic(Carleton, 1992).

Furthermore,when the PTSM is in its positive phase,the anomalous easterly winds occur in the mid and lower troposphere over the equatorial central Pacific(Figs.4a and 4b)whereas anomalous westerly winds occur at 200 hPa in the upper troposphere(Fig. 4c).Meanwhile,over the Indian Ocean,westerly wind anomalies are dominant in the mid and lower troposphere while easterly anomalies occur in the upper troposphere.These anomalous winds induce convergence in the mid and lower troposphere and divergence in the upper troposphere above the MC,which is actually the ascending branch of the abnormal Walker circulation. Anomalous convective activities in this MC region act as an“atmospheric bridge”when the equatorial Pacific interacts with the tropical Indian Ocean(Wu and Meng,1998;Alexander et al.,2002).The descending branch of the anomalous Walker circulation is located in the equatorial central-eastern Pacific and its center is around 150°W nearby the equator.This means that the climatological mean Walker circulation is intensified when the PTSM is in its positive phase.Note that the situations are opposite when PTSM in its negative phase.

4.2Anomalous vertical circulations

In order to further reveal the vertical structure of the PTSM,we have performed regressions of geopotential height anomaly in the troposphere and vertical velocity anomaly onto the time series of coefficients of SVD1.The results are shown in Fig.5.

Above the equator(Fig.5a),the sign-reverse of geopotential height anomalies from positive to negative are found at around 700 hPa over 90°W–45°E,and at about 500 hPa over 175°E–90°W.Interestingly,the sign-reverse height at around 90°E is at about 450 hPa, higher than in other regions over the equator.The largest negative anomalies of geopotential height over equatorial region are observed in layer from the earth surface to 850 hPa while the largest positive anomalies at about 200 hPa.

In the mid and high latitudes of the NH,it is seen from Figs.5b–g that the disturbances of the geopotential height anomaly in troposphere are the equivalent barotropic although the axis of maximum anomalies of geopotential height in the mid and lower troposphere tilts slightly poleward with height.In the SH,disturbances of anomalous geopotential heights largely demonstrate an equivalent barotropic structure.

Fig.5.Regressions of anomalous vertical circulation(streamlines)and geopotential height(shaded contours;gpm) onto time-series of coefficients of the SVD1 with that(a)for EQ,(b)0°,(c)30°E,(d)100°E,(e)150°E,(f)150°W,and (g)60°W,respectively.

The ascending branch of the anomalous Walker circulation(Fig.5a)is located above the MC,while the two descending branches are found over the central-western Indian Ocean and the central-eastern Pacific to the east of the dateline,respectively.A couple of vertical circulations can also be found from the

equatorial eastern Pacific to the equatorial Atlantic and the equatorial Africa.The vertical extension of these two circulations is lower than that of the Walker circulation over Indo-Pacific sector.During positive phase of PTSM,these anomalous vertical circulations are favorable for the formation and maintenance of negative SLPAr over the MC region,and possibly resulting in negative SLPAr in eastern Pacific due to eastward propagation of Kelvin waves in equatorial region.

It is noticed that the geopotential height anomalies in the NH and SH distribute in a way of roughly symmetric about the equator(Figs.5b–g).However, the anomalies in the NH are distinctly larger than those in the SH.Moreover,the abnormal meridional circulations look,to a certain extent,different in different meridional planes.

In general,the PTSM-related circulation change shows a seesaw feature between the polar region and the tropics.Its spatial structure is roughly symmetric about the equator.The circulation anomaly shown in the PTSM mode is stronger and with more distinct zonal disturbances in the NH than in the SH.The related disturbances of anomalous geopotential heights intensify with height with equivalent barotropic structure in mid and high latitudes in both hemispheres.

5.Relationship between PTSM and the global climate anomaly

The correlations of PTSM with global precipitation and temperature anomalies are calculated by using time series of coefficients of SVD1 to investigate the linkage between the PTSM mode and climate anomalies at various regions of the world.It is found that anomalous global precipitation and temperature changes are closely related to the PTSM.The precipitation and temperature anomalies during boreal winter are highly correlated with PTSM not only in mid and high latitudes but also in tropical regions in both hemispheres,suggesting that the PTSM has possible impacts on surface climate in wide areas of the globe(Fig.6).These impacts on climate conditions in two hemispheres are contemporaneously.Note that correlations in the tropics are also significant though relatively weaker than those in the extra-tropics.

5.1 Precipitation anomalies related to PTSM

In the mid and high latitudes of the NH,when the PTSM is in its positive phase,precipitation is less than normal from Baffin Island in North America to the eastern Siberia;these dry climate conditions are specifically observed in the area from North Atlantic to central-northern Europe and Siberia,in region west of the Alaska of US,and in southwestern part of China. However,several regions in the extra-tropics receive more than normal precipitation.These regions are the northeastern part of North America,Azores in central Atlantic,southern Europe,the Mediterranean and its surrounding areas,Mongolia,the northern part of China,easternmost of Russia,the northern part of Japan,and part of Northeast Pacific.

In the tropical regions,significant positive correlations are sparsely found in the Bay of Bengal,the MC region,the eastern equatorial Pacific,the Amazon region in South America,and eastern equatorial Africa along with the western equatorial Atlantic,indicating that more precipitation during boreal winter is received in these regions when PTSM is in its positive phase.Nevertheless,significant correlations are observed in the equatorial Indian Ocean,implying the rainfall is less than normal there.

In the extra-tropics of the SH,high positive correlations are found in the northern Australia,the west coast of Australia,and the west coasts of both South America and South Africa in midlatitudes.Strong positive correlations are also found in both South Pacific and South Atlantic south of 45°S,indicating that there will be more than normal rainfall in these regions when PSTM is in its positive phase(Fig.6a). Simultaneously,negative correlations are found in the Antarctica and its vicinity,the scattered regions in subtropical part of the southern Pacific,the southern Atlantic,and the southern Indian Ocean,demonstrating that less rainfall will possibly occur in these regions.The situations will be opposite when the PTSM is in its negative phase.

These precipitation anomalies that possibly occur in global domain are intrinsically related to anomalous circulation patterns(Fig.4)and atmospheric mois-

ture transport(Fig.6a).It is found from Fig.6a that where there significantly northward(southward) transports of water vapor in the NH(SH),or the water vapor transported from oceanic region into the land area,there will possibly occur more than normal rainfall.On the other hand,less than normal rainfall tends to occur in regions where the water vapor fluxes point southward(northward)in the NH(SH),and in regions where the vapor fluxes explicitly diverge.

5.2 Temperature anomalies related to PTSM

The surface air temperatures in boreal winter are possibly influenced by the PTSM.In the mid and high latitudes of the NH,significantly positive correlations are found over regions around Greenland and Aleutian islands,indicating that in these regions the surface air temperature is higher when the PTSM is in positive phase.However,it is seen in Fig.6b that significantly

negative correlations are found in a geographical band from the southeastern part of the United States northeastward to Eurasian continents and then to Japan islands,suggesting that the climate conditions are significantly colder than normal.

Fig.6.Correlations of time series of coefficients of SVD1 with precipitation anomalies(shaded contours)along with superimposed anomalous water vapor fluxes(vectors;kg m−1s−1)as obtained by regressing the anomalous vapor fluxes integrated vertically from the earth surface up to 300 hPa onto the time series of coefficients of SVD1.(a)Bold arrows indicate the fluxes are significant at/above the 95%confidence level by using an F-test.(b)Correlations of temperature anomalies with time series of coefficients of SVD1.The critical value at the 95%confidence level is found to be 0.32 by using a t-test.

In the tropical and subtropical regions,significant positive correlations of time series of coefficients of SVD1 with temperature anomalies are observed over tropical Atlantic,and regions around the Sahara desert and Saudi Arabia Peninsula,Maldives Island, the southwestern part of China,the northern part of Indian subcontinent,the MC,the western part of Australia,and the tropical western Pacific.This indicates that the higher than normal surface air temperatures possibly occur in these tropical and subtropical regions.The warmer surface conditions can in return drive the atmosphere to ascend anomalously in tropical and subtropical regions.

In the mid and high latitudes of the SH,the large positive correlations exist in Antarctic whereas significant negative correlations are found over oceans south of South America and south of Australia(Fig.6b). This means that when the PTSM in positive phase,the warmer than normal condition will occur over Antarctic and colder than normal temperatures will possibly appear over Southern Ocean.

6.Conclusions and discussion

By performing the SVD analysis on SLPA,we have investigated the leading SVD mode(SVD1)and its related circulation variations as well as the climate anomalies in both the NH and SH in boreal winter. Major conclusions are as follows.

There indeed exists the polar-tropical seesaw mode(PTSM)that describes the contemporaneous circulation variations in the NH and SH.This mode can be identified by using the SVD analysis on SLPA from which the ENSO signal is regressed out.This PTSM can explain 47.74%of the total SLPA covariance of circulation anomalies in the two hemispheres. It is found that the PTSM in the NH is in overall stronger than that in the SH.This mode varies mainly on interannual timescale with periodicities of 2–3 and 4–6 yr,and has a long-term trend with decreases in SLP in the tropics and increases in SLP in the polar region in the recent 30 years.

The PTSM looks roughly symmetric about the equator,and has a horizontal structure with lower SLPA in tropical region while higher SLPA in mid and high latitudes in both hemispheres when PTSM is in its positive phase.The PTSM-related anomalous circulations over Eurasia,Pacific,and Atlantic in the NH are associated with atmospheric teleconnections including EU and PNA patterns(Wallace and Gutzler,1981).In the SH,a wave train similar to the PSA teleconnection pattern is found over the southern Pacific and Atlantic.Moreover,in the tropical region, disturbances of geopotential height in the lower troposphere are in opposite phase to those in the upper troposphere,showing that the anomalous circulations are baroclinic correspondingly.In the mid and high latitudes,the anomalous circulations are equivalent barotropic.

Significant wintertime temperature and precipitation anomalies correlated with the PTSM can be found globally. In the positive phase of PTSM,positive temperature and precipitation anomalies occur in the Azores Islands in the central Atlantic,southern Europe,and the Mediterranean and nearby areas while negative precipitation and temperature anomalies occur in North Atlantic,central and northern Europe, and Siberian region.Precipitation is higher than normal but temperature is lower than normal in Northeast Asia.In western China,precipitation is lower than normal but temperature is higher than normal. Over the northeastern part of North America,both precipitation and temperature are higher than normal. Negative temperature anomaly is found in southeast America.Meanwhile,precipitation is higher than normal in southwest Africa and northern Australia while positive temperature anomaly can be found in centraleastern South America,North and Southeast Africa, and southern Australia.

It is worth noting that AO is correlated with AAO with a value of 0.33(P＜0.05),indicating that circulation changes in the mid and high latitudes of the two hemispheres are not highly correlated.Both AO

and AAO show an anti-phase SLPA oscillation between the mid–high latitudes and the polar region, and the corresponding SLPA changes are basically symmetric in zonal(Gong and Wang,1999;Thompson and Wallace,2000).The correlation of the time coefficient of SVD1 with AO is found to be–0.86 and that with AAO is–0.42(P＜ 0.05)(Table 4), but the spatial structure of PTSM is different from those of both AO and AAO.This difference can be found at least in two aspects:(1)the zonal mean of SVD1 mode(Fig.1d)shows larger anomalies in both the tropics and the polar region with opposite signs whereas both the AO and AAO show their strong anomalies in midlatitudes and polar regions,and(2) the latitudes where signs of anomalies of SLPA reverse from positive to negative are also different from those of both AO and AAO.These suggest that the tropical variability in PTSM may be more important than that in AO/AAO,being possibly a result of response of the SLPA in mid and high latitudes to SLPA changes in the tropics.Certainly,a reversed situation might exist if SLPA change in low latitudes as a response to SLPA changes in mid and high latitudes.In this sense,despite the high correlation between the AO and PTSM,AO actually represents the SLP oscillation between the polar region and the midlatitudes while the PTSM depicts a seesaw like variation between the polar region and the tropics.

In addition,the NH annular mode is physically consistent with AO,which to a certain degree is a result of diabatic surface heating in polar or mid latitudes.The internal dynamic processes in the atmosphere may also play an important role in forming this annular mode.However,it is well known that the tropical region works as a source region for atmospheric energy supply with a very strong forcing of diabatic heating.The PTSM is most likely a result of atmospheric response to anomalous tropical thermal forcing over both oceans and continents(Figs.5 and 6b).Therefore,despite the high correlation between AO and PTSM,they are different due to possible different causes.Anyway,in-depth study is necessary to further explore the mechanism for the PTSM formation and the dynamic and thermo-dynamic processes involved in the interaction between mid–high latitudes and low latitudes,and hence to clarify whether the PTSM is or not AO.

Moreover,note that there occurred a severe sea ice change event in region to the west of 144.6°E, 66.56°–67.6°S in late December 2013,making two polar vessels,Akademik Shokalskiy and Xuelong trapped in the Antarctic coastal(Wang et al.,2014).It was found that a cyclone that occurred in edge area of Antactic induced the rapid creation of extreme Antarctic sea ice conditions there,causing this vessel trap event (Wang et al.,2014).As seen in Fig.1,in 2013,the PTSM was in its positive phase.It is speculated that higher than normal temperature anomalies were possibly observed in the Antarctic region(Fig.6b)in 2013.Whether there are some relationships of the PTSM with this kind of extreme sea ice conditions or not deserves more investigations in the future.

Acknowledgments.TheNCEP/NCARreanalysisdata used hereareobtained from the NOAA-CIRES Climate Diagnosis Center accessible at http://www.esrl.noaa.gov.The CMAP rainfall data, oscillation index,and the Nino indices are downloaded from http://www.cpc.ncep.noaa.gov. All graphs in this paper are plotted by using GrADS.

Alexander,M.A.,I.Blad´e,M.Newman,et al.,2002: The atmospheric bridge:The influence of ENSO teleconnections on air-sea interaction over the global oceans.J.Climate,15,2205–2231.

Ashok,K.,Z.Y.Guan,and T.Yamagata,2003:A look at the relationship between the ENSO and the Indian Ocean dipole.J.Meteor.Soc.Japan,81, 41–56.

Ashok,K.,S.K.Behera,S.A.Rao,et al.,2007:El Ni˜no Modoki and its possible teleconnection.J.Geophys. Res.,112,C11007.

Bai Xuxu,Li Chongyin,and Li Lin,2012:Numerical simulation study of the Madden–Julian oscillation influences on spring precipitation in China.Acta Meteor.Sinica,70,986–1003.(in Chinese)

Carleton,A.M.,1992:Synoptic interactions between Antarctica and lower latitudes.Aust.Meteor.Mag., 40,129–147.

Chen Wen,2002:Impacts of El Ni˜no and La Ni˜na on the cycle of the East Asian winter and summer monsoons.Chinese J.Atmos.Sci.,26,595–610.(in Chinese)

Chen Shangfeng,Chen Wen,and Wei Ke,2013:Recent trends in winter temperature extremes in eastern China and their relationship with the Arctic oscillation and ENSO.Adv.Atmos.Sci.,30,1712–1724.

Chen Xiaolong,Zhou Tianjun,and Guo Zhun,2014:Climate sensitivities of two versions of FGOALS model to idealized radiative forcing.Sci.China(Earth Sci.),57,1363–1373.

Christy,J.R.,K.E.Trenberth,and J.R.Anderson,1989: Large-scale redistributions of atmospheric mass.J. Climate,2,137–148.

Ding Yihui,Liu Yanju,Liang Sujie,et al.,2014:Interdecadal variability of the East Asian winter monsoon and its possible links to global climate change.J. Meteor.Res.,28,693–713.

Findlater,J.,1969:A major low-level air current near the Indian Ocean during the northern summer.Quart. J.Roy.Meteor.Soc.,95,362–380.

Giorgi,F.,1990:Simulation of regional climate using a limited area model nested in a general circulation model.J.Climate,3,941–963.

Gong,D.Y.,and S.W.Wang,1999:Definition of Antarctic Oscillation index.Geophys.Res.Lett.,26,459–462.

Guan Zhaoyong,Xu Jianjun,and Wang Panxing,1994: Odd and even symmetry of atmospheric circulation. Part I:Theoretical basis and climatic characteristics. Scientia Meteor.Sinica,14,300–310.(in Chinese)

Guan,Z.Y.,and T.Yamagata,2001:Interhemispheric oscillation in the surface air pressure field.Geophys. Res.Lett.,28,263–266.

Guan Zhaoyong,Lu Chuhan,Mei Shilong,et al.,2010: Seasonality of interannual inter-hemispheric oscillations over the past five decades.Adv.Atmos.Sci., 27,1043–1050.

Gushchina,D.,and B.Dewitte,2012:Intraseasonal tropical atmospheric variability associated with the two flavors of El Ni˜no.Mon.Wea.Rev.,140,3669–3681.

Hoskins,B.J.,I.N.James,and G.H.White,1983: The shape,propagation and mean-flow interaction of large-scale weather systems.J.Atmos.Sci.,40, 1595–1612.

Huang Ronghui,Chen Wen,Ding Yihui,et al.,2003: Studies on the monsoon dynamics and the interaction between monsoon and ENSO cycle.Chinese J.Atmos.Sci.,27,484–502.(in Chinese)

IPCC,2014:Summary for Policymakers.Climate Change 2014:Mitigation of Climate Change.Contribution of Working Group III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change, Edenhofer,O.,R.Pichs-Madruga,Y.Sokona,et al., Eds.,Cambridge University Press,5 pp.

Kalnay,E.,M.Kanamitsu,R.Kistler,et al.,1996: The NCEP-NCAR 40-year reanalysis project.Bull. Amer.Meteor.Soc.,77,437–472.

Kusunoki,S.,J.Yoshimura,H.Yoshimura,et al.,2006: Change of Baiu rain band in global warming projection by an atmospheric general circulation model with a 20-km grid size.J.Meteor.Soc.Japan,84, 581–611.

Li Lijuan,Lin Pengfei,Yu Yongqiang,et al.,2013:The flexible global ocean-atmosphere-land system model: Grid-point version 2:FGOALS-g2.Adv.Atmos. Sci.,30,543–560.

Li Li,Guan Zhaoyong,and Cai Jiaxi,2014:Interannual variation of winter temperature difference between polar and equatorial regions associated with East Asian climate anomalies. Chin. Sci. Bull.,59, 2720–2727.(in Chinese)

Lu Chuhan,Guan Zhaoyong,Mei Shilong,et al.,2008: The seasonal cycle of interhemispheric oscillations in mass field of the global atmosphere.Chin.Sci. Bull.,53,3226–3234.

Lu Chuhan,Guan Zhaoyong,Li Yonghua,et al.,2013: Interdecadal linkages between the Pacific decadal oscillation and interhemispheric air mass oscillation and their possible connections with East Asian monsoon.Chinese J.Geophys.,56,117–128.

Man Wenmin and Zhou Tianjun,2011:Forced response of atmospheric oscillations during the last millennium simulated by a climate system model.Chin. Sci.Bull.,56,3042–3052.

Mantua,N.J.,S.R.Hare,Y.Zhang,et al.,1997:A Pacific interdecadal climate oscillation with impacts on salmon production.Bull.Amer.Meteor.Soc., 78,1069–1079.

Meehl,G.A.,C.Covey,K.E.Taylor,et al.,2007:The WCRP CMIP3 multimodel dataset:A new era in climate change research.Bull.Amer.Meteor.Soc., 88,1383–1394.

Sardeshmukh,P.D.,and B.J.Hoskins,1988:The generation of global rotational flow by steady idealized tropical divergence.J.Atmos.Sci.,45,1228–1251.

Shi Neng and Zhu Qiangen,1993:Studies on the Northern early summer teleconnection patterns,their interannual variations and relation to drought/flood in China.Adv.Atmos.Sci.,10,155–168.

Shao,P.,X.B.Zeng,K.Sakaguchi,et al.2013.Terrestrial carbon cycle:Climate relations in eight CMIP5 earth system models.J.Climate,26,8744–8764.

Tan Benkui and Chen Wen,2014:Progress in the study of the dynamics of extratropical atmospheric teleconnection patterns and their impacts on East Asian climate.J.Meteor.Res.,28.780–802.

Taylor,K.E.,R.J.Stouffer,and G.A.Meehl,2012: An overview of CMIP5 and the experiment design. Bull.Amer.Meteor.Soc.,93,485–498.

Thompson,S.L.,and D.Pollard,1995:A global climate model(GENESIS)with a land-surface(LSX).Part I:Present climate simulation.J.Climate,8,732–761.

Thompson,D.W.J.,and J.M.Wallace,1998:The Arctic oscillation signature in the wintertime geopotential height and temperature fields.Geophys.Res. Lett.,25,1297–1300.

Thompson,D.W.J.,and J.M.Wallace,2000:Annular modes in the extratropical circulation.Part I: Month-to-month variability.J.Climate,13,1000–1016.

Wallace,J.M.,2000:North Atlantic oscillation/annual mode:Two paradigms–one phenomenon.Quart.J. Roy.Meteor.Soc.,126,791–805.

Wallace,J.M.,and D.S.Gutzler,1981:Teleconnections in the geopotential height field during the Northern Hemisphere winter.Mon.Wea.Rev.,109,784–812.

Wang Panxing,Guan Zhaoyong,and Xu Jianjun,1994: Odd and even symmetry of atmospheric circulation. Part II:Measurement and anomaly analysis.Scientia Meteor.Sinica,15,1–8.

Wang Bin,Zhou Tianjun,Yu Yongqiang,et al.,2008: A perspective on earth system model development. Acta Meteor.Sinica,66,857–869.(in Chinese)

Wang Shaowu,Wen Xinyu,and Huang Jianbin,2013: Are the changes of winter and summer monsoons over East Asia out of phase?Adv.Climate Change Res.,9,231–233.(in Chinese)

Wang,Z.M.,J.Turner,B.Sun,et al.,2014:Cycloneinduced rapid creation of extreme Antarctic sea ice conditions.Scientific Reports,4,5317.

Watanabe,M.,2004:Asian jet waveguide and a downstream extension of the North Atlantic oscillation. J.Climate,17,4674–4691.

Weng,H.Y.,G.X.Wu,Y.M.Liu,et al.,2011:Anomalous summer climate in China influenced by the tropical Indo-Pacific Oceans. Climate Dyn.,36, 769–782.

Wu Guoxiong and Meng Wen,1998:Gearing between the Indo-monsoon circulation and the Pacific-Walker circulation and the ENSO.Part I:Data analyses. Scientia Atmos.Sinica,22,470–480.(in Chinese)

Xiao Dong,Li Jianping,Zhou Xiuji,et al.,2010: An evaluation of the spatiotemporal structures of the decadal abrupt changes in the global oceanatmosphere system simulated by the coupled climate models.Acta Meteor.Sinica,68,763–778.(in Chinese)

Xie,P.P.,and P.A.Arkin,1997:Global precipitation: A 17-year monthly analysis based on gauge observation,satellite estimates,and numerical model outputs.Bull.Amer.Meteor.Soc.,78,2539–2558.

Yuan Yuan,Yang Hui,and Li Chongyin,2012:Study of El Ni˜no events of different types and their potential impact on the following summer precipitation in China.Acta Meteor.Sinica,70,467–478.(in Chinese)

Yuan Yuan,Li Chongyin,and Yang Song,2014:Decadal anomalies of winter precipitation over southern China in association with El Ni˜no and La Ni˜na. J.Meteor.Res.,28,91–110.

Zeng Qingcun and Li Jianping,2002:Interactions between the Northern and Southern Hemispheric atmospheres and the essence of monsoon.Chinese J. Atmos.Sci.,26,433–448.(in Chinese)

Zhang Gang,2012:Study of East Asian winter climate predictability by using European multi-model ensemble prediction.Acta Meteor.Sinica,70,690–703.(in Chinese)

Zhang Jie,Li Lijuan,Zhou Tianjun,et al.,2013:Variation of surface temperature during the last millennium in a simulation with the FGOALS-g1 climate system model.Adv.Atmos.Sci.,30,699–712.

Zhou Tianjun,Zou Liwei,Wu Bo,et al.,2014:Development of Earth/Climate System Models in China: A review from the coupled intercomparison project perspective.J.Meteor.Res.,28,762–779.

:Tang Weiya and Guan Zhaoyong,2015:ENSO-independent contemporaneous variations of anomalous circulations in the Northern and Southern Hemispheres:The polar-tropical seesaw mode. J.Meteor.Res.,29(6),917–934,

10.1007/s13351-015-5095-6.

Supported by the National Natural Science Foundation of China(41175062 and 41330425)and Science Innovation Program of the State Key Laboratory of Severe Weather,Chinese Academy of Meteorological Sciences(2015LASW-A03).

∗Corresponding author:guanzy@nuist.edu.cn.

©The Chinese Meteorological Society and Springer-Verlag Berlin Heidelberg 2015

(Received May 19,2015;in final form October 15,2015)