The stratospheric polar vortex (SPV) in the Northern Hemisphere is a cold-core low-pressure system that exists from the stratosphere downward into the mid troposphere. The SPV shows significant intraseasonal variabi-lity from November to next March. The breakdown or abrupt weakening of the SPV is an extreme event known as a Sudden Stratospheric Warming (SSW) (Charlton and Polvani, 2007). An SSW event has taken place almost every year since 2000 (Reichler et al., 2012). Although the SPV is mainly confined to the stratosphere, it shows strong coupling with the troposphere (Thompson et al., 2002; Lu and Ding, 2013; Mitchell et al., 2013). It has been suggested that the SPV can be disturbed by planetary-scale Rossby waves originating from the troposphere. The stratosphere, in turn, can alter the conditions for tropospheric planetary-wave propagation and lead to changes in mean circulation by anomalous poleward meridional eddy momentum flux (Baldwin and Dunkerton, 2001). Therefore, the downward propagation of stratospheric anomalies can likely have a significant influence on surface weather and climate (Baldwin et al., 2003; Gu and Yang, 2006; Wei et al., 2011; Li et al., 2015; Lu and Ding, 2015). For example, Kidston et al. (2015) revealed that, within the 60-day interval following a weakening SPV, the amplitude of the tropospheric jet stream increases (more meandering). The Atlantic blocking is more persistent, leading to extremely low temperature over northern Europe and eastern USA. Kolstad et al. (2010) found a statistical correlation between weak SPVs and the frequency of cold-air outbreaks in the Northern Hemisphere on a weekly timescale. The impact of SPV anomalies in winter on the large-scale circulation and weather patterns in the following spring, however, has barely been explored.

On the other hand, the spatial distribution of surface temperature trends in the Northern Hemisphere is not uniform. Kug et al. (2015) found that the winter surface warming at high latitudes has been getting stronger since 1998; whereas, by contrast, strong cooling trends are evident over parts of the extratropical continents. This feature has been investigated by many studies in recent years (e.g., Cohen et al., 2012; Outten and Esau, 2012). Meanwhile, the spring surface temperature trends over Eurasia also show a non-uniform distribution. For example, Li and Wang (2013) revealed significant cooling trends along the East Asian coast since the 1990s. Considering the persistence of the SPV signal, we speculate that the variability of spring surface temperature over Eurasia may be related to the preceding SPV.

In the present study, we explore the time-lagged stratospheric influence of the winter SPV on the early spring non-uniform variation in surface temperature over Eurasia. We highlight the stratosphere’s role in tropospheric weather and climate forecasts.

Multiple observational datasets are used in the present study: daily datasets of 17-level geopotential height produced by NCEP-2 (Kanamitsu et al., 2002); monthly datasets of tropopause pressure from NCEP-1 (Kalnay et al., 1996); and atmospheric fields from ERA-Interim (Dee et al., 2011). The analysis period is from December 1979 to April 2016. The winter of 1980 refers to December 1979 to February 1980 (December–next February, DJF). The early spring of 1980 refers to March–April (MA) 1980.

We know from previous studies (Hoskins et al., 1985; Ambaum and Hoskins, 2002; Mitchell et al., 2013, Seviour et al., 2016) that a positive (negative) potential vorticity anomaly in the stratosphere will result in an elevated (sinking) tropopause. To represent the characteristics of the SPV, we use two categories of SPV indices defined by directly using the tropopause height. One describes the circumpolar vortex (SPVC; Fig. 1a) calculated as the DJF tropopause height averaged in the polar cap (north of 65°N, multiplied by –1.0). The other describes the SPV displacements (SPVD; Fig. 1b) computed as the DJF tropopause height averaged within 30°–90°N, 90°–179°E minus that averaged within 30°–90°N, 180°–360° (multiplied by –1.0; the regions selected for computation based on Fig. 3c). The correlation coefficient between the SPVC and SPVD is –0.31 (> 90% confidence level). In addition, the correlation coefficient between the SPVC/SPVD and DJF Arctic Oscillation (AO) is 0.39 (> 95% confidence level)/0.21 (statistically insignificant). The surface (1000 hPa) AO index is provided by the Climate Prediction Center (). The three indices are calculated from the monthly data, and normalized. The linear trend is first removed from all data and indices.

Figure 1 Schematic diagrams of the two categories of SPV: (a) SPVC and (b) SPVD.

Figure options

We analyze the dominant features of MA-mean 2-m temperature (T_2m) anomalies over Eurasia (15°–75°N, 0°–150°E) by using empirical orthogonal function (EOF) analysis (Fig. 2). The first three EOF modes (EOF1, EOF2, and EOF3) explain 37.0%, 13.8%, and 11.1% of the total temperature variance, respectively. Analysis of the eigenvalues indicates that the three leading modes in T_2m are significantly different from each other and from the rest of the eigenvectors in terms of sampling error above the 95% confidence level. The EOF1 represents a consistent warming (cooling) pattern, with large variations over high-latitude Eurasia (Fig. 2a). The EOF2 displays a north–south variation with the zero line along 60°N (Fig. 2b). The EOF3 is characterized by a west–east variation, with the zero line along 80°E (Fig. 2c). In the following analysis, we focus mainly on EOF1 and EOF3, which are related to the SPVC and SPVD, respectively.

Figure 2 (a) EOF1, (b) EOF2, and (c) EOF3 (multiplied by 100) (detrended) of 2-m temperature over Eurasia (15°–75°N, 0°–150°E) during March–April of 1980–2016.

Figure options

As shown in Fig. 3a, EOF1 corresponds to an enhanced SPVC: the tropopause height anomalies are largely zonally symmetric with negative/positive values in the polar cap/midlatitudes. The spatial distribution resembles that regressed upon the SPVC index (Fig. 3c). The corresponding principal component of T_2m EOF1 (T_2m PC1; Fig. 4a, black line) is statistically correlated with the SPVC index (blue line) over the 36-yr record (R = 0.55, > 99% confidence level). By contrast, EOF3 (Fig. 3b) is associated with SPVD: the tropopause height anomalies are southwest–northeast oriented (negative/positive values were mainly in the Asian–North Pacific region/North Atlantic), which is similar to that regressed upon the SPVD index (Fig. 3d). The correlation coefficient between T_2m PC3 (Fig. 4b, black line) and the SPVD index (red line) is 0.44 (> 99% confidence level). We also note that T_2m PC2 is unrelated to both the SPVC and SPVD indices (R = 0.07 and –0.13, respectively; statistically insignificant). The results suggest that the west–east surface temperature seesaw over Eurasia in MA can be linked to SPVD in DJF.

Figure 3 Regressions of DJF tropopause height (hPa) anomalies upon the (a) first and (b) third principal components of 2-m temperature. Shading indicates significance at the 90%, 95%, and 99% confidence levels according to a two-tailed Student’s t test. (c, d) As in (a, b), but for the SPVC and SPVD indices.

Figure options

Figure 4 Temporal evolutions of (a)T2m PC1 (black line) and SPVC (blue line) and (b)T2m PC3 (black line) and SPVD (red line).

Figure options

How does the preceding winter SPV impact the early spring surface climate?Garfinkel et al. (2013) argued that a change in the SPV is usually accompanied by changes in both the polar and midlatitude jet streams in the troposphere. We begin in this part of the study by presenting the regressions of DJF- and MA-mean 200-hPa zonal wind (U200) anomalies upon the SPVC and SPVD indices (Fig. 5). In DJF, the SPVC is associated with positive/negative U200 anomalies around the polar cap/in the midlatitudes (Fig. 5a). This implies an accelerated polar jet stream in the upper troposphere. In MA, the positive values around the polar cap (i.e., the polar jet stream) are much weaker than those in DJF and shift northward (Fig. 5b). In comparison, in DJF, SPVD is associated with negative/positive U200 anomalies around the polar cap/in the East Asian–North Pacific region (Fig. 5c). This indicates a decelerated polar jet stream, together with an enhanced East Asian jet stream in the upper troposphere. It is worth noting that the positive values in the East Asian–North Pacific region (i.e., the enhanced East Asian jet stream) continue to MA (Fig. 5d).

Figure 5 Regressions of (a) DJF and (b) MA 200-hPa zonal wind (m s–1) anomalies upon the SPVC. Shading indicates significance at the 90%, 95%, and 99% confidence levels according to a two-tailed Student’s t test. (c, d) As in (a, b), but for the SPVD. Black frame indicates the East Asian–North Pacific region.

Figure options

Figure 6 shows similar results as Fig. 5, but for sea level pressure (SLP). Figure 7 illustrates the regressions of MA-mean 1000-hPa meridional wind (V1000) and T_2m anomalies upon the SPVC and SPVD indices. In DJF, the SPVC is related to negative/positive SLP anomalies in the polar cap/North Atlantic and midlatitude Eurasia (Fig. 6a), resembling the typical AO pattern (Thompson and Wallace, 2001). In MA, the north–south oscillation weakens, and is replaced by largely negative values in North Pacific (Fig. 6b). The weakened north–south oscillation in SLP favors anomalous southerly winds over most parts of Eurasia (Fig. 7a), which may lead to a unified warming pattern (Fig. 7c). The results are consistent with previous studies (Baldwin and Dun-kerton, 1999; Reichler et al., 2012; He and Wang, 2013; Kidston et al., 2015). The correlation coefficient between T_2m PC1 and DJF AO is 0.44 (> 99% confidence level).

Figure 6 As in Fig. 5, but for SLP (hPa).

Figure options

Figure 7 Regressions of March–April (a) 1000-hPa meridional wind (m s–1) and (c) 2-m temperature (°C) anomalies upon the SPVC. Shading indicates significance at the 90%, 95%, and 99% confidence levels according to a two-tailed Student’s t test. (b, d) As in (a, c), but for the SPVD.

Figure options

However, in DJF, the SPVD is related to negative/positive SLP anomalies in central–eastern Eurasia/ North Pacific (Fig. 6c). In MA, negative/positive values in Europe and East Asia/ North Pacific continue (Fig. 6d). Anomalous southerly/northerly winds prevail over western/eastern Eurasia (the zero line along 80°N) (Fig. 7b), which will result in a west–east surface temperature seesaw pattern (Fig. 7d). We note that the warm and cold centers are located north of the Caspian Sea and Lake Baikal, respectively—similar to those in Fig. 2c. The correlation coefficient between T_2m PC3 and DJF AO is nearly zero (R = 0.07, statistically insignificant). Thus, we suggest that SPVD may account for an accelerated East Asian jet stream in DJF. This will persist into MA and act as a bridge connecting DJF SPVD and MA surface climate over Eurasia.

We perform composite analysis on the daily polar vortex variability [geopotential height anomalies averaged in the polar cap (north of 65°N), PCH] between high and low SPVC and SPVD years (Fig. 8). Negative/positive PCH, which are normalized by the standard deviation for each of the 17 pressure levels (Baldwin and Thompson, 2009; Kim et al., 2014), indicate enhanced/weakened polar vortex variability. The high and low SPVC/SPVD years are determined when the indices are above 1.0 or below –1.0 standard deviation. As shown in Fig. 8a, the SPVC corresponds to negative PCH anomalies in the stratosphere in DJF, which indicates an intensified SPV. The SPVC signal can persist into MA and propagate downward from the upper to the lower troposphere. By contrast, the SPVD-related positive PCH anomalies in the stratosphere in DJF are relatively weak, representing a weakened SPV. It is also detected in the entire troposphere in MA (Fig. 8b). The results are consistent with those in the polar cap in Fig. 6, and confirm the persistence of stratospheric influences from DJF to MA.

Figure 8 (a) Composite differences in the sub-seasonal evolution of geopotential height anomalies averaged in the polar cap as a function of pressure between high (1983, 1993, 1996, 2000, 2005, 2014, and 2016) and low (1985, 1987, 1999, 2004, 2006, 2009, and 2013) SPVC years. Shading indicates significance at the 90%, 95%, and 99% confidence levels according to a two-tailed Student’s t test. (b) As in (a), but for high (1995, 1999, 2001, 2002, and 2004) and low (1986, 1997, 2003, 2005, and 2014) SPVD.

Figure options

Based on ERA-Interim, NCEP-1, and NCEP-2 reanalysis data, this study identifies a significant relationship between preceding winter SPVD and early spring non-uniform variation in surface temperature over Eurasia. The mechanism accounting for this relationship can be explained by the persistence of an anomalous East Asian jet stream. Specifically, a positive change in SPVD can be linked to a decelerated polar jet stream, accompanied by an accelerated East Asian jet stream in the troposphere in DJF. The enhanced East Asian jet stream will persist into MA. The analysis of V1000 anomalies shows that anomalous southerly/northerly winds prevail over western/eastern Eurasia, and may result in the west–east surface temperature seesaw over Eurasia.

In addition, SPVD can be influenced by various factors including cryospheric conditions (Arctic sea ice, Eurasian snow), tropical forcing (the Quasi-Biennial Oscillation, El Niño–Southern Oscillation), and solar activity (Gray et al., 2004; Garfinkel et al., 2010; Jaiser et al., 2013; Kim et al., 2014; Peings and Magnusdottir, 2014). Previous studies have also suggested the modulation of SSW events by the Pacific Decadal Oscillation (Jadin et al., 2010; Hurwitz et al., 2012; Woo et al., 2015). However, the contributions of these factors are not examined in the present study. This issue needs to be addressed in future work.

Acknowledgments. Thanks to Prof. Ke Wei for helpful discussions.