Strengthened connection between meridional location of winter polar front jet and surface air temperature since the mid-1990s

This study reveals a marked enhancement in the relationship between the variations in location of the winter East Asian Polar Front Jet (EAPJ) and the surface air temperature (SAT) in Eurasia since the mid-1990s. Before the mid-1990s, an evident wave train related to the meridional location of the EAPJ exhibited an anticyclonic anomaly over northern Europe and a cyclonic anomaly in northwestern Asia. With an equatorward shift of the EAPJ after the mid-1990s, the wave train experiences a notable adjustment that is conducive to East Asian cooling, displaying an anticyclonic anomaly around the Kara-Laptev Seas and a cyclonic anomaly near northeastern Asia. Arctic warming anomalies and sea ice loss contributed significantly to these decadal changes. Simulation experiments forced by observed Arctic sea-ice variability further confirm this result. Since the mid-1990s, Arctic sea ice loss (or Arctic warming anomaly) has contributed to a reduction in westerly winds in high latitudes by modulating the meridional temperature gradient. The deaccelerated winds intensify the Arctic cold air propagating to the south, enhancing the atmospheric baroclinicity and the westerly flow in the upper level at the south side of the EAPJ, favoring the southward shift of the EAPJ. With the equatorward shift of the EAPJ, the corresponding SAT anomalies in East Asia are more salient.


Introduction
The jet stream has an important effect on weather and climate (Dao 1958;Wu and Su 2015;Yao et al. 2016). There exists a branch of strong and narrow jet streams in the upper troposphere and lower stratosphere over midlatitude East Asia on the north side of the Tibetan Plateau, which is called the eddy-driven jet or the East Asian polar front jet (EAPJ) (Palmin and Newton 1948;Held 1975;Panetta 1993;Lee and Kim 2003). The EAPJ extends eastwards from the northern side of the Tibetan Plateau and converges with the southern subtropical jet over the western Pacific, forming the strongest jet stream (Cressman 1981;Sheng 1986;Liang and Wang 1998;Hudson 2012). As an integral component of the East Asian atmospheric circulation system, the EAPJ has a significant influence on the weather and climate over East Asia (Yang et al. 2002;Gong and Ho 2003;Jhun and Lee 2004;Wang and Zhang 2015a, b). The EAPJ occupies the frontal areas in the upper and lower troposphere, where much atmospheric disturbance and many storms develop and lead to severe weather and climate events such as heavy rainfall, snowstorms and cold surges Zhang 2006, 2013;Luo and Zhang 2015;Xiao et al. 2016). Therefore, it is of great significance to investigate the variations in the EAPJ to gain a better understanding of recent climate regime shifts over the Eurasian continent.
The variations in the EAPJ are characterized by seasonal evolution in the intensity and location, which are often accompanied by transitions in the atmospheric circulation pattern (Zhang et al. 2008;Schiemann et al. 2009;Ye et al. 2014). Many studies have focused on the influence of the EAPJ on summer weather and climate (Huang et al. 2014;Li et al. 2014;Hall et al. 2017;Wang 2019, 2021;Wu and Francis 2019). However, the intensity and range 1 3 of the EAPJ in winter are larger than those in summer, and its transient characteristics are also more obvious, which is closely related to the activity of cold air in the high latitudes (Ren et al. 2010). Therefore, studies should pay more attention to the winter polar front jet. As the polar front and subtropical jets are both dominated by the zonal westerly, the clear geographical boundary between the two jets cannot be obtained via the analysis of monthly mean data (Yang et al. 2002;Lee and Kim 2003;Zhang 2008;Ren et al. 2011), making it difficult to analyze the variation characteristics of the EAPJ. Previous studies before the 2000s have no consensus on the characteristics of the EAPJ center location and active region. Some studies after the late 2000s showed that a distinct boundary separating the EAPJ from the subtropical jet exhibits in Eurasia by using daily wind data to calculate the wintertime occurrence frequencies of the jet stream at 300 hPa (Zhang et al. 2008;Ren et al. 2010). Based on the calculation of jet stream occurrence frequencies, a series of studies on the synergetic changes of two jet streams and their relationship with atmospheric circulation evolution and climate anomaly have been carried out (Liao and Zhang 2013;Li and Zhang 2014;Huang et al. 2014;Wang and Zhang 2015a, b;Xue et al. 2017;Yin and Zhang 2021). Xue and Zhang (2017) discovered that the interannual variation in the 500 hPa geopotential height anomaly related to the location of the two jet streams is similar to the Eurasian teleconnection (EU) pattern, while the pattern of sea level pressure (SLP) anomalies showed an Arctic Oscillation (AO)-like pattern that affected cold winters in Eurasia. Interdecadal changes in the synergetic variation of the two jet streams have also been studied. He and Wang (2012) found that the weakening of the East Asia winter monsoon around the mid-1980s was probably connected to the change in the meridional westerly shear in upper levels. Luo and Zhang (2015) indicated that the concurrent variability patterns of the two jets experienced noticeable decadal changes in both location and intensity in addition to the interannual variations. Huang et al. (2017) showed a possible linkage to the meridional displacements of the two jets for the decadal change in precipitation over East China around the late 1990s. Yin and Zhang (2021) further explained that the concurrent variation in the location of two jets was accompanied by the enhanced Siberian high and tilted East Asia trough after the late 1990s, benefitting the cold anomalies over northern Eurasia with the northern mode of the East Asia winter monsoon intensified. These studies revealed the interannual and interdecadal changes in the concurrent variations of the two jet streams, which are often accompanied by variations in the atmospheric circulation system. Most of these studies highlighted the importance of the concurrent variation in the EAPJ and subtropical jet and its effect on climate. However, the variation features of a single jet and its influence on weather and climate are completely different from the synergistic variation of the two jets. Lu et al. (2010) found that under the single-jet circulation regime, the polar front jet shifts monotonically poleward with enhancement of both the global mean and the equator-to-pole gradient of the sea surface temperature (SST). In a double-jet circulation regime, the position of the polar front jet and SST show an opposite relationship. Hence, it is important to explore the variation in a single EAPJ and its impact on winter climate to improve the predictions of weather events and climate variability over Eurasia. What is the feature of the EAPJ single-jet location? Does the surface air temperature (SAT) anomaly pattern, which is related to the EAPJ location, change? Thus, here, we mainly focus on the relationship between the EAPJ location and SAT over Eurasia. We attempt to answer the following questions in this study: (1) Has the relationship between the meridional location of the EAPJ and East Asian SAT changed in recent decades? and (2) What are the possible mechanisms for this varying relationship?
The structure of the paper is as follows. Section 2 describes the data and method employed in this study. Section 3 presents observational evidence for the interdecadal variation in the linkage between the winter EAPJ and SAT in the Eurasian continent. Section 4 discusses the possible reasons for the changes in the winter EAPJ-SAT relationship. Section 5 provides a summary and discussion.

Data and analysis methods
Monthly mean horizontal winds, geopotential height, sea level pressure (SLP), and SAT are extracted from the National Centers for Environmental Prediction-National Center for Atmospheric Research (NCEP-NCAR) reanalysis from 1960 to 2019 (Kalnay et al. 1996). The spatial resolution of atmospheric variables is 2.5 × 2.5° in latitude and longitude, with 17 levels in the vertical direction, whereas SAT is on 192 × 94 Gaussian grids. The AO index is obtained from the National Oceanic and Atmospheric Administration (NOAA). Here, winter is defined as December-January-February (DJF), and the winter of 1960 refers to December 1960 and January and February 1961.
Empirical orthogonal function (EOF; Lorenz 1956;North et al. 1982;Thompson and Wallace 2000;Wilks 1995) analysis is employed to extract the dominant patterns of EAPJ variability. Linear regression and correlation analyses are used to examine the relationship between the principal component time series and the spatial patterns. The statistical significance of the correlation was evaluated using the twotailed Student's t test.
The selection of the EAPJ key region is based on the results of Zhang et al. (2008) and Ren et al. (2008). They calculated the occurrence numbers of the jet stream at 300 hPa at each grid point during winter by using the daily wind data to identify the maximum wind speed centers over East Asia. If the wind speed at a specific grid point satisfied the following two conditions: (1) was equal to or greater than 30 m s −1 ; (2) was the local maximum in the surrounding 8 grid points, then this point was regarded as one point of jet core occurrence, and the corresponding latitude and longitude were recorded. Via a similar method, Xue et al. (2017) further used the 6 hourly data of longer periods to calculate the number of jet core occurrences in winter. The results showed that there existed a region over 50°E-100°E, 42.5°N-62.5°N with large jet occurrence numbers over the poleward side of the Tibetan Plateau. Thus, this study takes the same area as the key region for the EAPJ.
To further investigate the influence of Arctic sea-ice loss on the decadal change of the EAPJ location and its relation to Eurasian SAT, simulation experiments are conducted using the Community Atmosphere Model version 5.4 (CAM5.4) model, which is the atmospheric component of the Community Earth System Model (CESM1.2.2). The spatial resolution is 1.9° in latitude and 2.5° in longitude, with 30 levels in the vertical direction (Neale et al. 2012). A 39 year control run is performed with the climatological monthly SST and sea ice concentrations (SICs). The climatology is calculated as the mean of November 1981 to February 2006. The simulation experiment employs the same SST boundary condition as the control run but is forced by observed Arctic SICs for the period of January 1979 to December 2017. The simulation experiments were repeated 39 times with different initial conditions, which were derived from 1 January of each year of the 39-year control run. We obtain 39 winters in the simulation experiments. The greenhouse gas emissions and the aerosol forcing in both experiments are fastened to 2000, according to the data of the model itself (Wu et al. 2022).

Changes in the linkage between the EAPJ and winter SAT
Although the same EAPJ key area is selected in this paper, the objective of the present study is to analyze the changes in the winter EAPJ-EOF2 and Eurasian SAT connection (as described in the introduction), which is distinct from previous studies (Luo and Zhang 2015;Xue et al. 2017;Yin and Zhang 2021). To investigate the features of winter EAPJ variability, EOF analysis is performed for winter mean zonal wind variability at 300 hPa over the key area of the EAPJ (solid purple box in Fig. 1a, c). Figure 1 shows the first two EOF spatial patterns of 300-hPa zonal wind variability and the corresponding normalized time series (PCs). The first principal mode (accounting for 48.1% of the total variance) exhibits a center located along the EAPJ. This mode is associated with the strengthening or weaking of the upper-level jet stream. The corresponding time series (PC1) shows dominant positive phases from the early 1960s to the early 1980s and is then followed by frequent negative phases from the mid-1980s to the early 2000s. However, frequent positive phases since the early 2000s indicate that the EAPJ has been weakened significantly. EOF2 accounts for 31.5% of the total variance, and it is statistically separable from EOF1 (North et al. 1982). This mode depicts a structure with two anomalous centers located on the north and south sides of the jet. In the positive (negative) phase of PC2, the enhanced westerly winds on the south (north) side of the jet are conducive to the equatorward (poleward) shift of the EAPJ. Thus, EOF2 represents a north-south shift in the position of the EAPJ. This structure of the upper tropospheric zonal wind anomalies provides favorable dynamical conditions for the cyclonic/anticyclonic anomaly in the mid-latitude over East Asia. PC2 shows dominant negative phases from the early 1970s to the late 1980s and is then followed by frequent positive phases with some interannual variations. This indicates that the westerly winds on the south side of the EAPJ have strengthened in recent years, and the EAPJ shifts significantly equatorward. Previous studies (Lu et al. 2001(Lu et al. , 2002Lin and Lu 2005) have shown that the meridional displacement of the jet stream has a more direct influence on climate variability over East Asia than its intensity variations. Here, we mainly focus on the second mode (EOF2), which depicts the meridional displacement of the EAPJ. The change in the relationship between EOF2 of the where SAT is mainly related to EAPJ-EOF2 in this paper EAPJ and winter air temperature is the major objective of this paper, which is a topic that has rarely been addressed in previous studies.
To examine the SAT anomalies in boreal winter associated with location changes in the EAPJ, the SAT anomalies derived from a linear regression of the normalized PC2 are shown in (Fig. 2a). A positive phase of PC2 corresponds to negative SAT anomalies in the mid-high latitudes of Eurasia and positive SAT anomalies in marginal seas centered on the Barents-Kara Seas, even though those anomalies are weak. In addition, a significant cold anomaly occurs in western North America, implying that the SAT anomalies related to EAPJ-EOF2 may not be limited to Eurasia. Figure 2b-d exhibits atmospheric circulation anomalies associated with EAPJ-EOF2, including several meteorological variables, such as SLP, 1000-hPa winds, and 500-hPa and 300-hPa geopotential heights, regressed on the normalized PC2 time series. A positive SLP anomaly occurs from the Greenland Sea to the Laptev Sea and the Siberian marginal seas, northern Eurasia, and the northern North Pacific, while a negative SLP anomaly is observed in the middle and low latitudes of Asia and the North Atlantic (Fig. 2b). The Siberian High (SH) is weakened significantly. Northerly surface wind anomalies appear over the northern Eurasian continent and bring about plenty of cold air (Fig. 2b). The 500-hPa geopotential height anomalies show a wave train structure over the North Atlantic, Arctic, Eurasia and Japan. A significantly negative geopotential height anomaly occupies most of the mid-high latitudes of Asia and the North Atlantic. Positive geopotential height anomalies cover northern Europe and the Siberian marginal seas, resulting in a weakened polar vortex at 500 hPa. Positive height anomalies over Japan and the northwestern Pacific indicate a weakened East Asian trough (EAT). The result here is different from previous studies (Luo and Zhang 2015;Yin and Zhang 2021), which emphasized that the equatorward shift of the EAPJ and the poleward migration of EASJ jointly enhanced the EAT. The weakened SH and EAT indicate that Arctic warming and Eurasian cooling do not necessarily correspond to the intensified East Asian winter monsoon. This result contradicts the common opinion found in previous studies (Wu et al. 1999(Wu et al. , 2011Petoukhov and Semenov 2010;Inoue et al. 2012), which generally consider that the warm Arctic-cold Eurasian pattern corresponds to the intensified East Asian winter monsoon. Furthermore, although this wave train structure at 500 hPa is similar to the EU teleconnection pattern, the relationship between the EU index and EAPJ-PC2 has gradually weakened in recent years. For example, the correlation coefficient between the EU index and EAPJ-PC2 is 0.74 during 1960-1990 but weakened to 0.37 during 1990-2019. The calculation method of EU index is based on the previous studies (Wang and Zhang 2015a, b). Meanwhile, there is no EU-like pattern in a later study (Figs. 6,8 below). Thus, in our following study, we do not consider the role of EU. At the higher level, the wave train structure is similar to that at 500 hPa.
To further analyze the relationship between the EAPJ and Eurasian SAT more precisely, we selected the solid green box (60°E-120°E, 45°N -60°N) in Fig. 2a as the key SAT regions related to EAPJ-EOF2 and then calculated the areamean SAT anomalies. Figure 3a displays the normalized EAPJ-PC2 (bar) and the detrended normalized winter SAT time series in the key region (black line). Obviously, the out-phase interannual variation in SAT and PC2 of the EAPJ is more frequent after than before the mid-1990s, implying a plausible interdecadal change in the PC2-SAT connection. In particular, large positive (negative) PC2 values in 2004PC2 values in , 2005PC2 values in , 2009PC2 values in and 2012PC2 values in (2001PC2 values in and 2003 correspond to evident winter cooling (warming). The varying PC2-SAT relationship is further confirmed in Fig. 3b, which shows the 21 year sliding correlations between EAPJ-PC2 and winter SAT. The years shown in Fig. 3b represent the central year of the 21-year running window. The winter PC2-SAT linkage has experienced an enhancement since the 1990s. Other lengths of running windows (such as 23 and 25 years) will obtain similar results (not shown). Therefore, in the following, we select the highest correlation epoch and the lowest correlation epoch before and after the 1990s, respectively, to examine the reasons for the interdecadal change in the PC2-SAT connection. Based on the sliding correlations  dow (1972-1992) when the negative correlation is most insignificant, and 2004 (the solid red dot after the mid-1990s in Fig. 3b) is the central year of the 21 year window  when the negative correlation is the most significant. As such, the largest contrast of the winter PC2-SAT correlation is observed between 1972-1992 and 1994-2014. Spatially, winter SAT anomalies associated with EAPJ-EOF2 display notable differences over Eurasia during 1972Eurasia during -1992Eurasia during and 1994Eurasia during -2014. Figure 4 shows regression maps of winter SAT anomalies on the normalized PC2 time series during 1972-1992 and 1994-2014. During the first phase   (Fig. 4a), negative SAT anomalies are less significant in East Asia. Only a small area northwest of Lake Balkhash has a significant negative SAT, but anomalous amplitudes are very small. Warm anomalies occur in the Tibetan Plateau and parts of East Asia. The negative SAT anomalies in the Kara-Laptev Seas indicate that the relationship of EAPJ-EOF2 and Arctic warming is weak in the first period. During the second phase (1994-2014), significant negative SAT anomalies almost occupy northern Eurasia and extend to Japan, with anomalous amplitudes below − 1.5 °C (Fig. 4b). At the same time, the Arctic warming anomaly is more prominent, implying that Arctic warming seems to be in stronger connection with the equatorward shift of the EAPJ since the mid-1990s. Consequently, the relationship between the EAPJ location and Eurasian winter SAT experienced noticeable decadal changes around the mid-1990s. It is worth noting that the relationship between the EAPJ meridional location and Eurasian SAT is strengthened accompanied by the equatorward shift of the EAPJ. Especially in the latter epoch  in our study, it is confirmed that the EAPJ shifts southward as the zonal winds increased on the south side and weakened on the north side of the jet (Fig. 5). Differences of winter zonal wind anomalies at 300 hPa (1994-2014 composite minus 1972-1992 composite) are more significant on the north and south sides of the EAPJ (not shown). This indicates a dependence of the relationship between the EAPJ and SAT on the decadal change in the meridional displacement of the jet.

Possible mechanism for the relationship change between EAPJ and SAT
The above analyses suggest that the relationship between winter Eurasian SAT and EAPJ location experienced an enhancement with an equatorward shift of the EAPJ. An accompanying issue to be addressed is: what is the possible factor contributing to this interdecadal change? What are the possible mechanisms in this process? First, we compare the atmospheric circulation anomalies related to the anomalous EAPJ pattern between the high-and low-correlation epochs. Figure 6 displays the various atmospheric circulation variables in winter regressed on the normalized PC2 index during 1972-1992 and 1994-2014. Before the mid-1990s, a belt of upper tropospheric 300-hPa zonal wind anomalies is observed north of 40°N, with negative anomalies over eastern Europe to northern Russia and positive anomalies over the mid-latitude and the Barents Sea (Fig. 6a). The significantly weakening westerly anomalies in northwestern Eurasia intensify the cold air propagating to the south, corresponding to the negative SAT anomaly northwest of Lake Balkhash (Fig. 4a). This zonal wind anomaly pattern was accompanied by an anomalous anticyclone over Europe and an anomalous cyclone over northern Asia (Fig. 6c). Furthermore, a westerly wind anomaly occurs over the Okhotsk Sea, and an easterly wind anomaly occupies the northwestern Pacific. An anticyclone anomaly occurs in the northwestern Pacific, implying the weakening of the EAT. In the lower troposphere (at 850 hPa), the anticyclone anomaly mainly occurs over northern Europe, Japan and the northwestern Pacific, while the cyclone anomaly occupies Asia and the Arctic (Fig. 6e). The SLP anomaly pattern (not shown) is similar to that at 850 hPa. The negative SLP anomaly also occurs in the mid-high latitudes of Asia, indicating the weakening of the SH. The weakening of the SH and EAT implies the weak East Asian winter monsoon. Furthermore, compared with the means averaged over 1960-2019, the positive SLP anomaly in Europe is weaker, and the location is further south before the mid-1990s, bringing about relatively less cold air to Asia. Thus, negative SAT anomalies are weaker and confined to the northwest of Lake Balkhash (Fig. 4a). These atmospheric circulation anomalies are completely different from those after the mid-1990s. During 1994-2014, the pattern of zonal wind anomalies at 300 hPa clearly shows a belt structure with a negative anomaly over 60°N-80°N and a positive anomaly over 40°N-60°N and 80°N-90°N (Fig. 6b). With decreased westerly wind in high latitudes, the Arctic cold air propagating to the south is strengthened. In addition, with the cold air propagating to the south, the baroclinicity of the atmosphere at middle and high latitudes is strengthened, which is conducive to the strengthening of the westerly wind at high levels. The anticyclone anomaly occurs over the whole Arctic, and the cyclone anomaly occurs over the middle and high latitudes (Fig. 6d). In the lower troposphere, the cyclone anomaly occupies all extra Arctic regions (Fig. 6f). This spatial distribution of atmospheric circulation anomalies is largely similar to the negative phase of AO, indicating a strong connection of the winter EAPJ-EOF2 with the simultaneous AO during 1994-2014. In order to eliminate the effect of AO on SAT, and to better understand the relationship between the EAPJ location itself and SAT, the results are further analyzed after removing the AO variability as follows. Figure 7 shows SAT anomalies linked to PC2 after removing the AO signal in the latter epoch. During 1994-2014, the cold anomalies in the mid-high latitudes of Eurasia still exist, and the significant negative SAT anomalies can also extend to Japan after removing AO variability, confirming that the relationship between Eurasia SAT (especially East Asia SAT) and the meridional location of EAPJ enhanced since the mid-1990s. Meanwhile, Arctic warming is more significant, implying that the relationship between the equatorward shift of the EAPJ and Arctic warming is strengthened after removing AO variability since the mid-1990s.
The regressions of atmospheric circulation anomalies with respect to EAPJ after removing the AO variability are calculated during the two periods (Fig. 8). The results show that the corresponding atmospheric circulation anomalies related to the EAPJ location are completely different during the two periods. Before the mid-1990s, the zonal wind anomaly pattern (Fig. 8a) and the wave train structure (Fig. 8c) in the upper level were similar to those in Fig. 6a-c, indicating that the AO had little relation to the EAPJ location before the mid-1990s. However, there are still some subtle differences in the atmospheric circulation after removing AO variability. The easterly wind anomaly at high latitudes can extend westward into the North Atlantic (Fig. 8c). The range of anticyclonic anomalies over Europe expanded north to the  Arctic, and the anticyclonic anomaly over the northwestern Pacific was weaker than that in Fig. 6c. In the lower level (Fig. 8e), the extent of the anticyclonic anomaly over Europe expands to the Arctic and is larger than that in (Fig. 6e). After the mid-1990s, zonal wind anomalies present a belt structure in the upper level ( Fig. 8b) with an easterly wind anomaly over 55°N-75°N and westerly wind anomalies south of 40°N-55°N and north of 75°N (Fig. 8d). The westerly anomaly weakened significantly in northern Eurasia, corresponding to the cold anomaly in Eurasia. A wave train structure with a cyclonic anomaly near northeastern Asia and the anticyclonic anomaly around the Kara-Laptev Seas and mid-low latitudes (Fig. 8d) occupies Eurasia at 300 hPa. Geopotential height anomalies in the lower level (Fig. 8e) exhibit a dipole structure with an anticyclonic anomaly over the Kara-Laptev Seas and a cyclonic anomaly over northeast Asia. This anomalous circulation is different from those in previous studies (Luo and Zhang 2015;Xue and Zhang 2017). The above results indicate that with the decadal shift in the EAPJ meridional location and the change in the corresponding wave train anomalies, the SAT anomalies related to EAPJ location in East Asia are more significant after the mid-1990s.
What factor contributes to this decadal change? Francis and Skific (2015) revealed that the Arctic warming event occurred in the mid-1990s. Yin and Zhang (2021) found that the first dominant pattern of Arctic sea-ice concentration (SIC) in December experienced a phase transition in the 1990s. Wu et al. (2012) inferred that the summer Arctic surface wind pattern exhibited interdecadal variability during the late 1990s, which is consistent with the rapid declining trend of sea ice extent. And many previous studies revealed the effect of Arctic sea ice variation on jet and SAT (Cohen et al. 2014;Wu et al. 2015Wu et al. , 2019Francis and Vavrus 2015;Coumou et al. 2018;Alizadeh and Lin 2021). Do Arctic warming anomalies and Arctic sea-ice loss have contribution to the recent decadal changes of EAPJ location and its link to Eurasian SAT? Simulation experiments are conducted to examine these decadal changes in response to recent Arctic sea-ice loss. The simulated EOF2 for the EAPJ (Fig. 9a), which accounts for 28.9% of the total variance, is similar to the observation (Fig. 1c) with a positive zonal wind anomaly centered at the south side of the EAPJ and a negative zonal wind anomaly centered at the north side of the EAPJ. Hence, the simulation result of EOF2 also represents the meridional location of the EAPJ. The corresponding time series (PC2) shows a decadal change in the mid-1990s (Fig. 9b). The accumulative anomalies of PC2 show that positive phases have frequently occurred since 1995. Although the decadal change in simulation is later   1995 (1995-2017 composite minus 1979-1994 composite). The solid purple box indicates the key region of EAPJ. Stippled regions represent anomalies at the 95% confidence level than that in observation (approximately 1986 and Figure is not shown), it cannot deny the contribution of Arctic seaice loss to the southward shift of the EAPJ in recent years. Simulated winter decadal changes of the meridional temperature gradient (MTG) at 1000 hPa and zonal wind at 300 hPa approximately 1995 also affirm that the EAPJ shifts southward around mid-1990s (Fig. 10). The MTG increased at the south side of the EAPJ and decreased in the Arctic north of the EAPJ since the mid-1990s. The zonal wind structure is similar to that of the MTG, with intensifying westerly winds centered on the south side of the EAPJ. With the decadal change in the meridional displacement of the EAPJ, has the relationship between the EAPJ location and the Eurasia SAT also been changed in simulation results? Fig. 11 shows simulated 1000 hPa winter air temperature anomalies over Eurasia associated with the simulation result of PC2 for the EAPJ during 1979EAPJ during -1994EAPJ during and 1995EAPJ during -2017. Similar to the observations (as shown in Fig. 4), the simulated winter cooling related to the EAPJ location also experiences notable enhancement over Eurasia (particularly in East Asia).
Based on the above simulation experiment analyses, the result has been confirmed that Arctic sea-ice loss is an important contributor to the decadal change of EAPJ shift and its relation to the Eurasia SAT anomalies. The linking mechanisms can be interpreted as follows. Arctic sea ice loss (or Arctic warming anomaly) decreases the meridional temperature gradient at high latitudes and has probably contributed to a reduction in westerly winds at high latitudes. The deaccelerated winds intensify the Arctic cold air propagating to the south, enhancing the atmospheric baroclinicity at the south side of the EAPJ and benefitting the strengthening of westerly winds in the upper level at the south side of the EAPJ. Finally, a zonal wind anomaly pattern with the decreasing westerly wind to the north side of the EAPJ and increasing westerly wind to the south side of the jet has formed, favoring the southward shift of the EAPJ.
With the equatorward shift of the EAPJ, the corresponding wave train structure over Eurasia and SAT anomaly pattern also changed. The wave train structure that occupied western Eurasia before the mid-1990s changed into a wave train structure that occupied eastern Eurasia after the mid-1990s, and these circulation changes were accompanied by SAT anomalies extending into East Asia.

Summary and discussion
Previous studies have demonstrated that the winter EAPJ has a significant impact on the SAT in the mid-high latitudes of Eurasia (Yang et al. 2002;Jhun and Lee 2004;Luo and Zhang 2015;Xue and Zhang 2017). This study focuses on the second mode of EOF analysis (EOF2) for the EAPJ, which represents the variation in the meridional location of the EAPJ. The spatiotemporal features of winter EAPJ-EOF2 are elaborated in detail with NCEP/NCAR reanalysis data. The changes in the relationship between the winter EAPJ-EOF2 and Eurasia SAT are also investigated. Then, we try to explain the possible mechanisms for the change in the winter EOF2-SAT connection. The main results are summarized as follows.
EOF2 for the zonal wind at 300 hPa over East Asia features the meridional displacement variations in the EAPJ. The enhancement of the westerlies south of the jet makes the EAPJ shift southward in the positive phase of PC2. The amplitude of the positive phase of PC2 is very strong in the last two decades, implying that the EAPJ has shifted equatorward significantly in recent years. The meridional displacement variations in the EAPJ have a salient impact on the SAT. The equatorward shift of the EAPJ corresponds to the warm Arctic-cold Eurasia pattern. The SH and EAT are weakened. The weakening of the East Asian winter monsoon indicates that the warm Arctic-cold Eurasia pattern is Fig. 11 Simulated winter air temperature anomalies at 1000 hPa (unit: K), derived from a linear regression onto the simulation result of normalized PC2 time series for EAPJ during (a) 1979-1994 and (b) 1995-2017. Stippled regions represent anomalies at the 95% confidence level not necessarily consistent with the intensified East Asian winter monsoon. This result is different from those of previous studies (Wu et al. 1999(Wu et al. , 2011Petoukhov and Semenov 2010;Inoue et al. 2012), which generally consider that the warm Arctic-cold Eurasia pattern is accompanied by the intensified East Asian winter monsoon.
Further correlation analyses affirm that the interannual relationship between the EAPJ-PC2 and northern Eurasian SAT in winter experienced a notable interdecadal change around the mid-1990s. After the mid-1990s, there existed a significant connection between the meridional location of the EAPJ and SAT over Eurasia. In contrast, before the mid-1990s, this connection is weak and statistically insignificant. In addition, the relationship between the equatorward shift in the EAPJ and Arctic warming has also strengthened since the mid-1990s.
The possible mechanisms for this decadal change are discussed. Simulation experiments are carried out using the CAM5.4 model to investigate the impact of Arctic sea-ice loss on the decadal change of EAPJ location and its link to Eurasia SAT. The simulation experiments are forced by observed monthly Arctic SICs. Based on the simulation experiments and observed analyses, the recent Arctic warming anomalies and Arctic sea-ice loss contributed to these decadal changes around the mid-1990s. The possible physical processes can be summarized as follows. Arctic warming anomalies and sea ice loss decrease westerly winds at high latitudes by modulating the meridional temperature gradient at high latitudes. With the Arctic cold air propagating to the south, the anomalous baroclinicity and the westerly flow in the upper level at the south side of the EAPJ were enhanced. An anomalous zonal wind pattern with a decreasing westerly wind to the north side of the EAPJ and an increasing westerly wind to the south side of the EAPJ was generated, benefitting the southward shift of the EAPJ around the mid-1990s. With the equatorward shift of the EAPJ, the corresponding wave train structure over Eurasia also changed around the mid-1990s. Before the mid-1990s, a wave train with an anticyclonic anomaly over northern Europe and a cyclonic anomaly over northern Asia was established. However, after the mid-1990s, the geopotential height anomaly exhibits a wave train pattern with an anticyclonic anomaly over the Kara-Laptev Sea and a cyclonic anomaly over northeast Asia, which is distinct from that in previous studies (Luo and Zhang 2015;Xiao et al. 2016;Xue and Zhang 2017). Such changes have been accompanied by Eurasian (especially East Asian) cooling since the mid-1990s. In brief, a different relation between the meridional displacement of the EAPJ and Eurasian SAT anomalies in different periods was found in this study, which has not been addressed well in previous literature. This confirmed that recent Arctic sea ice loss contributed to these differences. These results complement previous works on the relationship between the EAPJ location and Eurasian cooling.
In our study, the winter Arctic sea ice loss plays an important role in influencing the EAPJ meridional displacement. However, Arctic sea-ice may be one of the contributors to the effect of the EAPJ location. Xiao et al. (2021) considered that a negative AO phase may push the EAPJ equatorward and cause Eurasian cooling. We also find a strong connection of the winter EAPJ location with AO after the mid-1990s (as shown in Fig. 6f). However, as members of the atmospheric circulation, the changes in the AO and EAPJ may have been simultaneous. Both the jet and AO may be influenced by other factors (such as Arctic sea-ice in this study). The emergence of meridionally oriented EAPJs since the mid-1990s cannot be simply attributed to the decadal change in AO. It is a challenge to distinguish a cause-andeffect relationship between the EAPJ location variations and AO. Further efforts are needed to answer this question.
Funding The authors thank the NCEP-NCAR, and the NOAA/Climate Prediction Center for providing atmospheric reanalysis data, and AO index. This study is supported by the National Key Basic Research Project of China (2019YFA0607002), the Major Program of the National Natural Science Foundation (41790472), the National Natural Science Foundation of China (41730959), the program of CAMS (2015CB453202) and the National Natural Science Foundation of China (Grant 41905058).

Conflict of interest All authors declared no competing interests.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.