The Linkage between Midwinter Suppression of the North Pacific Storm Track and Atmospheric Circulation Features in the Northern Hemisphere

The strongest prevalent high-frequency transient eddies or synoptic-scale disturbances are organized into zonally confined storm tracks in the mid-latitudes (Blackmon, 1976; Wallace et al., 1988; Ren et al., 2007). Storm track activities play vital roles in local weather and climate processes by transporting large amounts of heat, moisture, and momentum poleward (Zhou et al., 2009; Dong et al., 2013; Shaw et al., 2016). In terms of winter climatology, storm track activities are restricted to three areas: the North Pacific, North Atlantic, and Siberia (Blackmon et al., 1977; Hoskins and Hodges, 2002; Fu et al., 2009; Hoskins and Hodges, 2019). Compared with the other two storm tracks, which are located above the ocean, the Siberian storm track has received less attention due to its relatively small intensity.

It has been proven theoretically that the structure and propagation of high-frequency fluctuations are closely related to baroclinic wave activity (Blackmon et al., 1977; Blackmon et al., 1984; Lim and Wallace, 1991). Therefore, storm tracks are not only the areas with the most concentrated synoptic-scale filtered variance but also refer to baroclinic wave activity itself (Wallace et al., 1988). In light of this, baroclinic wave theories, such as the linear theory of baroclinic instability (Charney, 1947; Eady, 1949; Lindzen and Farrell, 1980), have successfully been applied to investigate storm tracks. Regarding annual cycles, both observational and modeling studies have identified that the peak of atmospheric baroclinicity exactly coincides with that of the North Atlantic storm track (NAST) in midwinter; however, the North Pacific storm track (NPST) attains a minimum amplitude in midwinter and peaks in autumn and spring (e.g., Nakamura, 1992; Christoph et al., 1997; Zhang and Held, 1999; Chang, 2001, 2003; Lee et al., 2011; Yang et al., 2021), which is referred to as the well-known “midwinter suppression” (MWS, hereafter). In fact, Klein (1958) noticed this unique feature of the seasonal NPST from the Lagrangian perspective, but he did not link it to atmospheric baroclinicity. Additionally, it is noteworthy that a modest MWS appears to also occur in the NAST when the North Atlantic jet stream is anomalously strong (Deng and Mak, 2006; Penny et al., 2013; Afargan and Kaspi, 2017; Park and Lee, 2020).

For nearly thirty years, this interesting phenomenon of the MWS has attracted the attention of the climate community, and it is still an open issue. In the framework of linear baroclinic instability theory, an intensified storm track is expected to correspond to strengthened atmospheric baroclinicity. The pioneering work of Nakamura (1992) argued that when the velocity of the jet stream exceeds a critical value of 45 meters per second, the relationship between the NPST amplitude and the jet stream intensity changes from positive to negative due to the trapping of eddies by the acceleration of the group velocity (Chang, 2001). This hypothesis was subsequently identified from the results of observations and modeling (Christoph et al., 1997; Lee et al., 2010) and was consistent with the mild MWS of the NAST during enhanced subtropical jet winters (Afargan and Kaspi, 2017). Other studies further suggested that the strengthened atmospheric baroclinicity associated with an intensified subtropical jet stream is strong enough to neutralize the impact of the accelerated group velocity (Nakamura et al., 2002; Harnik and Chang, 2004); therefore, the altered structure of the zonal westerlies should be taken into account (Nakamura and Sampe, 2002; Harnik and Chang, 2004). In addition, the change in eddy structure related to excessively strong westerlies has been suggested to be responsible for the MWS of the NPST. The enhanced jet stream was regarded to be associated with an eddy structure that was inefficient to extract energy from the basic flow, as exhibited by the weaker relation between the eddy temperature and the meridional/vertical velocity fields (Nakamura et al., 2002).

The aforementioned possible mechanisms can be seen as local perspectives; however, upstream (e.g., Penny et al., 2010, 2013; Lee et al., 2013) and global perspectives (Park and Lee, 2020) should also be considered. More specifically, Nakamura (1992) and Robinson et al. (2006) pointed out that some of the baroclinic waves coming from Siberia may play a role in “seeding” the baroclinic disturbances that grow over the western Pacific. By applying feature-tracking techniques, Penny et al. (2010) found that the amplitude and frequency of extratropical cyclones entering the NPST from the Altai-Sayan Mountains were dramatically decreased during midwinter compared with the transitional period. Furthermore, the topographic effect of the Central Asian mountains, particularly the Tibetan Plateau, has been highlighted. Based on a climate model forced with different Tibetan Plateau heights, the MWS of the NPST almost disappeared when the Tibetan Plateau height was largely reduced (Park et al., 2010; Lee et al., 2013). However, Chang and Guo (2012) argued that the interannual variations in upstream seeding and the NPST amplitude are largely uncorrelated; thus, the attenuated upstream seeding may not explain the MWS of the NPST. From a global perspective, Park and Lee (2020) reported that the MWS may partly arise from the consumption of available potential energy by global planetary-scale waves. In addition, much effort has been devoted to understanding the MWS from local energetics diagnostic perspective (e.g., Chang, 2001; Deng and Mak, 2005, 2006; Chen et al., 2013; Zhao and Liang, 2019).

Based on the above review, the critical mechanisms necessary to elucidate the MWS of the NPST have been an active research topic for decades and remain to be determined; these mechanisms are truly a tough issue requiring further study. Given that the changed storm track activities are intimately linked to atmospheric circulation, it is important to have a comprehensive understanding of atmospheric circulation anomalies that occur during the MWS; these anomalies may shed more light on this difficult problem. The research goals of this study include 1) an introduction of a new index for quantifying the MWS and 2) a report on the MWS-related atmospheric circulation characteristics in the Northern Hemisphere. The structure of the present study is as follows: brief descriptions of the data and methods are presented in section 2. Basic features of the midwinter NPST (MNPST) are presented in section 3. Atmospheric circulation anomalies associated with the MWS are explored in section 4. Finally, conclusions and discussions are given in section 5.

With an appreciable number of in situ observations over the Northern Hemisphere, the amplitudes of Northern Hemisphere storm track activities are intimately consistent among different reanalysis datasets compared with those of the Southern Hemisphere (Guo and Chang, 2008; Guo et al., 2009; Chang et al., 2016). The daily-mean and monthly atmospheric data used in the present study were provided by the National Centers for Environmental Prediction/National Center for Atmospheric Research (NCEP/NCAR) global reanalysis dataset (Kalnay et al., 1996). The daily variables include three-dimensional winds ($ u $, $ v, $ and $ \omega $ denote zonal, meridional, and vertical wind, respectively), air temperature ($ T $), and geopotential heights ($ h $). The monthly variables include sea-level pressure (SLP), meridional wind and air temperature at 850 hPa, geopotential heights at 500 hPa, and zonal wind at 250 hPa. The horizontal resolution is 2.5° × 2.5°, and the time span is from 1948 to 2018.

Compared with the Lagrangian objective algorithm that tracks individual cyclones and anticyclones accompanied by long and complicated programs (Kuwano-Yoshida, 2014), the alternative Eulerian definition of time-filtered eddy variance is easy to calculate and can reflect the interaction between a storm track and time-mean flow well (Chang, 2009). The Lanczos bandpass filter (Duchon, 1979) is used to isolate synoptic-scale (2.5–6 day) disturbances from daily-mean data. Although storm track activities have a three-dimensional structure from the sea surface to the tropopause, the standard deviation of the filtered geopotential height at 250 hPa ($\sqrt{\overline{z{'}z{'}}}$), one of the most commonly used Eulerian diagnostics, is selected in this study to characterize storm track activities owing to the more obvious phenomenon of the MWS in the upper levels of the troposphere. The prime denotes the synoptic-scale (2.5–6 days) disturbances and the overbar denotes a monthly average.

Given that eddy kinetic energy (EKE), like $\sqrt{\overline{z{'}z{'}}}$, is also widely used to represent storm track activities, the vertically integrated (from 925 to 250 hPa) eddy available potential energy (EAPE) is selected to represent the eddy energy. According to Lorenz (1955) and Takahashi and Shirooka (2014), the EAPE can be characterized by:

where ${C}_{p}$ represents the specific heat of dry air at constant pressure, $ \gamma $ manifests as $-{g} {{\rm{d}}{\theta }^{\mathrm{*}}}/{{C}_{p}}{{\rm{d}}z}$, $ g $ is the acceleration of gravity, $ T $ and $ \theta $ signify the temperature and potential temperature, respectively, the prime denotes the synoptic-scale (2.5–6 days) disturbances, and the asterisk denotes the global average.

The monthly storm track strength index (STSI) used in the present study is the same as that of Yang et al. (2020a) and sets a threshold that is the median of the NPST amplitudes of all of the grids within a domain of (25°–65°N, 130°E–120°W) where is the main body of the NPST. The mean of the values greater than the threshold value among all the grids is defined as the STSI, which can be expressed as:

where $ N $ denotes the number of grid points on which the amplitude exceeds the median of the NPST strength of all grids within the domain, and $ {\mathrm{N}\mathrm{P}\mathrm{S}\mathrm{T}}_{\mathrm{a}\mathrm{m}\mathrm{p}} $ denotes the amplitude of $\sqrt{\overline{z{'}z{'}}}$ at each grid point. The threshold is set each month and is applied to all grid points. Given the large interannual variation in the NPST center position (Li et al., 2010; Zhu et al. 2013), compared with the average of a fixed area, the use of the median strength threshold can better represent storm track activities by dynamically capturing the main body of the NPST in each calculation.

Figure 1a shows the year–month cross section of interannual changes in the monthly mean march of the NPST. We can see that the NPST has conspicuous sub-seasonal and interannual variabilities. There is an obvious interdecadal enhancement of the NPST amplitude (Nakamura et al., 2002; Lee et al., 2012). Although the NPST occasionally attains its maximum amplitude around wintertime, it generally peaks in late autumn (October or November) and early spring (March or April). More specifically, the NPST reaches its maximum amplitude fifteen times in October, twenty-two times in November, nine times in December, two times in January, three times in February, nine times in March, and ten times in April. In only five of the seventy years did the maximum amplitude occur in midwinter. From the latitude–month cross section of the monthly mean march in the NPST amplitude (Fig. 1b), it can be clearly seen that the two climatological NPST amplitude peaks are located at approximately 45°N in October-November and April. To further investigate the longitudinal location and evolution of the climatological NPST, Fig. 1c shows the month-longitude cross section of the seasonal march in the NPST amplitude. The two climatological NPST amplitude peaks occur around the International Date Line. The climatological NPST amplitude in November is noticeably larger than that in October; thus, the two climatological NPST amplitude peaks can be identified in November and April. The lower-tropospheric atmospheric baroclinicity is measured by the vertically integrated (from 925 to 700 hPa) maximum Eady growth rate (EGR; Lindzen and Farrell, 1980), which can be expressed as:

Figure 1.  (a) Year–month cross section of interannual changes in the seasonal march in the NPST amplitude averaged over 25°–65°N, 140°E–140°W, (b) latitude–month cross section of the seasonal marches in the NPST amplitude (shading; units: m) and vertically integrated (from 925 to 700 hPa) EGR (contours; units: 10^–6 s^–1) values averaged over longitude between 140°E–140°W and 130°E–160°W, respectively, and (c) month-longitude cross section of the seasonal march in the NPST amplitude averaged over latitude between 25°–65°N. Note that July to December refers to months in the current year, and January to June refers to months in the next year; the results are calculated for a time span from July 1948 to June 2018.

where $ \sigma $ signifies the EGR, $ f $ denotes the Coriolis parameter, $ N $ denotes the Brunt-Väisälä frequency which is defined as $\sqrt{g{\partial ln\theta }/{\partial z}}$, and $ \partial u/\partial z $ is the vertical shear of zonal wind. The climatological NPST reaches its minimum amplitude during midwinter despite the existence of the most vigorous atmospheric baroclinicity during this time (Fig. 1b), corresponding to the well-known MWS of the NPST, which is in agreement with previous results (e.g., Zhao and Liang, 2019; Yang et al., 2021). Notably, Nakamura et al. (2002) found that the MWS of the low-level NPST almost disappeared after the mid-1980s. However, the MWS of the upper-level NPST still clearly exists (Fig.1a).

By using an empirical orthogonal function analysis for the midwinters of the entire period (1948–2017), Fig. 2 shows the spatial distribution of the first two principal components of the MNPST marked by $\sqrt{\overline{z{'}z {'}}}$. The first principal component exhibits a monopole characterizing the holistic variability of the MNPST amplitude (Fig. 2a). The second principal component presents a meridional dipole that reflects the meridional shift of the MNPST (Fig. 2b). The first and second principal components explain 27% and 20% of the MNPST variance, respectively. The linear trends of the standardized time coefficients of the first (Fig. 2c) and second (Fig. 2d) principal components are 0.24 and 0.02 per decade, respectively. However, only the linear trend for the first principal component of the MNPST is significant, indicating that the MNPST exhibits an increasing trend. To better examine the interdecadal anomalies of the MNPST, Fig. 2e presents the difference in the MNPST from 1990–2015 and 1955–80. Significantly positive MNPST anomalies are predominantly located in the peak area and along the southern flank of the climatological midwinter mean, which characterizes the interdecadal enhancement and southward migration of the MNPST. Note that the linear trend is removed in the following sections.

Figure 2.  Spatial patterns (shading; units: m) of the (a) first principal component and (b) second principal component of the MNPST, and the corresponding standardized time series (black lines) and linear trends (red lines) for the (c) first and (d) second principal component of the MNPST, and (e) the difference in the MNPST between 1990–2015 and 1955–80. The contours represent the climatology of the MNPST with an interval of 10 m. The variance contribution is shown in the upper right corner of each panel. Stippling denotes anomalies that are significant at the 95% confidence level as determined by a Student’s t-test.

Before studying the MWS-related atmospheric circulation anomalies, it is necessary to select a Midwinter Suppression Index (MSI) to quantitatively measure the MWS of the NPST. To our knowledge, there are two candidate indices, one is the Intraseasonal Variability Index (IVI; Deng and Mak, 2006), and the other is the time coefficient of the first principal component of the MNPST (PC1; Nakamura et al., 2002). According to Deng and Mak (2006), the IVI can be written as follows:

where $ {\mathrm{S}\mathrm{T}}_{\mathrm{N}\mathrm{o}\mathrm{v}} $, $ {\mathrm{S}\mathrm{T}}_{\mathrm{J}\mathrm{a}\mathrm{n}}, $ and $ {\mathrm{S}\mathrm{T}}_{\mathrm{M}\mathrm{a}\mathrm{r}} $ signify the strength of the NPST, quantified by the box average over 30°–60°N and 140°E–130°W, in November, January, and March, respectively. When the IVI is negative, the midwinter suppression phenomenon is thought to occur. The larger the absolute value of the SVI is, the stronger the midwinter suppression. In fact, the IVI is defined based on the NPST which is measured by the root-mean-square of the filtered 300 hPa geopotential height field ($ h{'} $300). Note that after examining the latitude–time cross section of the monthly mean march in the$h {'}$300, we found that the $ h{'} $ 300 peaks in November and April (figure not shown).

Motivated by the considerations that $\sqrt{\overline{z{'}z{'}}}$ attains its maximum amplitudes in November and April (Fig. 1b) and that midwinter includes January and February, we define MSI by modifying the IVI; the modified MSI can be expressed as follows:

where $ {\mathrm{S}\mathrm{T}\mathrm{S}\mathrm{I}}_{\mathrm{N}\mathrm{o}\mathrm{v}} $ and $ {\mathrm{S}\mathrm{T}\mathrm{S}\mathrm{I}}_{\mathrm{A}\mathrm{p}\mathrm{r}} $ denote the STSI of the NPST in November and April, respectively, and $ {\mathrm{S}\mathrm{T}\mathrm{S}\mathrm{I}}_{\mathrm{m}\mathrm{w}} $ and $ {\mathrm{S}\mathrm{T}\mathrm{S}\mathrm{I}}_{\mathrm{c}\mathrm{o}\mathrm{l}\mathrm{d}} $ denote the averaged STSI during the midwinter and cold season (November to April), respectively. A positive (negative) MSI indicates a minimum (maximum) intensity of the NPST in midwinter. The MSI is expected to be more suitable to characterize the MWS of $\sqrt{\overline{z {'}z {'}}}$ than the IVI is and to provide an alternative option with which to measure the MWS. Based on the MSI, occurrence frequencies of the MWS during 1948–2018 and 1990–2015 are 0.84 and 0.81, respectively.

Figure 3 shows the time series of the standardized MSI and IVI. There is a significant out–of–phase relationship between the MSI and the IVI and their correlation coefficient is –0.60 which is significant at the 99% confidence level. Note that the correlation coefficient between the MSI and PC1 is –0.38, which implies that the smaller the MNPST amplitude is, the more obvious the MWS is. Considering that the amplitude of the NPST during midwinter has a distinct interannual variation (Fig. 1a) and the phenomena of the MWS are reflected by comparisons of the NPST amplitudes during midwinter and the other two peaks, a smaller amplitude of the MNPST does not necessarily correspond to the occurrence of the MWS. Therefore, it is not appropriate to regard PC1 as an index with which to characterize the midwinter suppression of the NPST. In addition, if the linear trend is not removed, PC1 is manifested as an obvious interdecadal intensification (Fig. 2c), which is indicative of a no longer apparent or even a vanished midwinter suppression of the NPST. However, from Fig. 1a, the relative sizes between the amplitude of the MNPST and those of the autumn and spring NPSTs still exist, indicating that the MWS is alive. In contrast, removing the linear trend has little impact on both the MSI and the IVI (figure not shown).

Figure 3.  Standardized time series of the midwinter suppression index (MSI; blue) and the intraseasonal variability index (IVI; red) from 1948 to 2017. The capital R denotes the correlation coefficient between the MSI and the IVI.

Next, we would like to demonstrate the usefulness of the MSI and make a comparison by applying the IVI to $\sqrt{\overline{z{'}z{'}}}$. Figure 4 shows the regressed NPST and vertically integrated (from 925 to 250 hPa) EAPE values with respect to the standardized MSI. With an obvious MWS phenomenon, significantly positive NPST anomalies emerge on the northern flank of the climatological NPST in November (Fig. 4a). In contrast, prominently negative NPST anomalies occur in the western and northern flanks of the climatological midwinter-mean NPST (Fig. 4b). Additionally, in April, prominently positive NPST anomalies are located in the peak area and north side of its climatology (Fig. 4c). The spatial distributions of the anomalous EAPE values associated with the MWS phenomenon (Figs. 4d–f) agree well with those of the anomalous NPST values. These anomalous patterns display the contrast between the intensities of the NPST and eddy energy values during midwinter and in November/April.

Figure 4.  NPST (a–c; shading; units: m) and vertically integrated (from 925 to 250 hPa) EAPE (d–f; shading; units: m² s^–2) anomalies in (a, d) November, (b, e) midwinter, and (c, f) April regressed upon the standardized MSI. The contours represent the climatology of the NPST (a–c) and EAPE (d–f) in November, midwinter, and April with intervals of (a–c) 15 m and (d–f) 1.5 m² s^–2. Stippling denotes anomalies that are significant at the 95% confidence level as determined by a Student’s t-test.

To better prove that the MSI is more suitable as a measure to represent the midwinter suppression of the NPST than the IVI, Fig. 5 shows the composite latitude–time cross section of the monthly mean march in the NPST amplitude by taking one standard deviation as the criterion. In strong MSI years, the climatologically sub-seasonal NPST attains its minimum and maximum amplitudes in midwinter and November/April, respectively, and the difference between the maximum and minimum amplitudes is larger than 27 m (Fig. 5a). In contrast, a sufficiently large increase and decrease in NPST amplitudes are detected during midwinter and November/April, respectively, during weak MSI years (Fig. 5b), which is associated with a less conspicuous amplitude contrast despite the relatively weak MNPST amplitudes compared with those of the November/April NPSTs. It is worth noting that the MWS of the NPST is more apparent in the upper troposphere than in the lower troposphere (Nakamura, 1992; Lee et al., 2011). During weak IVI years, the minimum amplitude of the NPST occurs in January (Fig. 5c); although the climatological midwinter-mean NPST amplitudes are suppressed, the NPST amplitudes in February are evidently larger than those in January. However, the amplitude contrast between midwinter and November/April is more remarkable during strong IVI years (Fig. 5d) than during weak MSI years (Fig. 5b), indicating that the IVI has less skill in describing the MWS than the MSI. In the following section, regression analysis is used to examine the atmospheric circulation anomalies associated with the midwinter suppression of the NPST based on this well-defined MSI.

Figure 5.  Composite latitude–time cross section of the seasonal march in the NPST amplitude (units: m) averaged over longitude between 140°E–140°W based on (a) strong MSIs, (b) weak MSIs, (c) weak IVIs, and (d) strong IVIs.

Figure 6 shows the regressed SLP values and 500 hPa geopotential heights with respect to the standardized MSI. The midwinter suppression of the NPST is related to significantly positive SLP anomalies on Eurasia, north of 45°N, and prominently negative anomalies over the North Pacific (Fig. 6a). Although the peak areas in which SLP anomalies are located are not consistent with the climatological centers of the Siberian High and Aleutian Low, the strengthened east–west pressure gradient tends to be accompanied by an intensified East Asian Winter Monsoon (EAWM). Note that such anomalous SLP values regressed on the MSI are basically in agreement with the anomalous 850 hPa geopotential heights regressed upon the PC1 of the low-level meridional eddy heat flux (Nakamura et al., 2002), but the signs of these two anomalies are opposite, suggesting that the variation in MNPST amplitude may be a major contributing factor to the MWS compared to the amplitude variations in November and April.

Figure 6.  SLP (a; shading; units: hPa) and 500 hPa geopotential height (b; shading; units: m) anomalies in midwinter regressed upon the standardized MSI. The contours represent the climatology of the midwinter SLP and 500 hPa geopotential height with intervals of (a) 4 hPa and (b) 100 m, respectively. Stippling denotes anomalies that are significant at the 95% confidence level as determined by a Student’s t-test.

Considering that the thermal contrast between the Asian continent and the adjacent oceans marked by the Siberian High and the Aleutian Low is the primary aspect of the EAWM (Wang and Chen, 2014) and that MNPST amplitudes can be modulated by the EAWM (Nakamura et al., 2002), we hypothesize that there may be a connection between the MWS and the EAWM. To better identify the relationship between the MWS of the NPST and the EAWM, we calculate the SLP-based EAWM index proposed by Wang and Chen (2014). Figure 7a shows the scatterplot of the MSI and EAWM index values. There is a significant in-phase correlation between the MSI and the EAWM index and their correlation coefficient is 0.32 which is significant at the 99% confidence level, indicating that stronger EAWM values are related to more obvious MWSs. To see the spatial pattern of anomalous MNPSTs associated with the EAWM, Fig. 8a shows the regressed MNPSTs with respect to the standardized EAWM index. Significant negative MNPST anomalies appear along the northwestern and southeastern flanks of the climatological MNPST; these anomalies are conducive to the occurrence of the MWS and imply that a weakened NPST is accompanied by a strengthened EAWM during midwinter (Nakamura et al., 2002).

Figure 7.  Scatterplot diagrams and correlation coefficients between (a–d; abscissa) the MSI and the (a; ordinate) EAWM index, (b; ordinate) East Asian trough index, (c; ordinate) North Pacific atmospheric blocking index, and (d; ordinate) lower-tropospheric atmospheric baroclinicity represented by the vertically integrated (from 925 to 700 hPa) box-averaged EGR over 25°–50°N, 120°E–170°W in midwinter. Double asterisks indicate significance at the 99% confidence level as determined by a Student’s t-test.

Figure 8.  MNPST anomalies (shading; units: m) regressed upon the standardized (a) EAWM index, (b) East Asian trough index, (c) North Pacific atmospheric blocking index, and (d) EGR index values. The contours represent the climatology of the MNPST. Stippling denotes anomalies that are significant at the 95% confidence level as determined by a Student’s t-test.

As shown in Fig. 6b, there are significant negative geopotential height anomalies related to the MWS occurring over eastern China that correspond to the enhanced East Asian trough. Following the methods used in previous studies (Wang and He, 2012; Yang et al., 2020b), the East Asian trough index is defined as the opposite of the standardized box-averaged, midwinter-mean 500 hPa geopotential height in the area of 25°–45°N, 110°–145°E. The East Asian trough index is positively correlated with the MSI and their correlation coefficient is 0.42 which is significant at the 99% confidence level (Fig. 7b). The intensified East Asian trough during midwinter is associated with significantly negative MNPST anomalies over the east side of Japan (Fig. 8b). Given that the intensified East Asian trough can weaken the winter NPST by attenuating baroclinic energy conversion (Yang et al., 2020b), the strengthened East Asian trough during midwinter may be responsible for the obviously suppressed MNPST.

In addition, the anomalous geopotential heights have a meridional dipole structure, with positive anomalies emerging over the East Siberian Sea and negative anomalies arising over the region south of the Aleutian Islands. These anomalous geopotential heights coincide with the typical spatial pattern of enhanced North Pacific atmospheric blocking (Cheung et al., 2013). By calculating the two-dimensional instantaneous blockings (Tibaldi and Molteni, 1990; Scherrer et al., 2006), the North Pacific atmospheric blocking index is defined by midwinter-mean instantaneous blocking averaged over 55°–75°N, 110°E–140°W. The correlation coefficient between the North Pacific atmospheric blocking index and the MSI is 0.34 with a 99% confidence level (Fig. 7c). There are significantly negative MNPST anomalies related to enhanced North Pacific atmospheric blocking in the northern flank of the climatological MNPST (Fig. 8c); these anomalies favor the occurrence of the MWS. Recently, Hwang et al. (2020) found that vorticity fluxes induced by high-frequency transient eddies contribute to the intensification of North Pacific atmospheric blocking through the interaction between the time-mean flow and transient eddies. In this way, strengthened North Pacific atmospheric blocking may potentially contribute to the MNPST anomalies associated with midwinter suppression.

Apart from the anomalous geopotential heights observed over the North Pacific, it is noteworthy that significantly positive geopotential height anomalies are also located over the Ural Mountains, which implies an intensified Ural blocking. Some previous studies emphasized the role of an alteration upstream of the midwinter suppression of the NPST (e.g., Nakamura, 1992; Robinson et al., 2006; Penny et al., 2010, 2013; Lee et al., 2013). For example, Penny et al. (2010) argued that the midwinter suppression was related to the considerably decreased amplitudes of synoptic-scale disturbances entering the NPST from midlatitude Asia near the Altai-Sayan Mountains. As a quasi-stationary and persistent anticyclonic circulation system in the mid-high latitudes, Ural blocking is identified as a key factor that is linked with the Siberian High and downstream East Asian climatic response (Li and Gu, 2010; Wang et al., 2010). The enhanced meridional circulation induced by strengthened Ural blocking allows for more cold air in higher latitudes to be transported southward and tends to intensify the Siberian High. Thus, anomalous Ural blocking may serve as a potential candidate accounting for the midwinter suppression of the NPST by modulating its downstream climate.

The importance of low-level atmospheric baroclinicity to the development of high-frequency transient eddies has been widely emphasized by a large number of previous studies (e.g., Hoskins and Valdes, 1990; Nakamura et al., 2008; Machado et al., 2021). To examine the association between low-level atmospheric baroclinicity and the MWS of the NPST, Fig. 9a shows the regressed vertically integrated (from 925 to 700 hPa) EGR with respect to the standardized MSI. The decreased EGR related to the MWS over midlatitude Central Asia may correspond to the weakened upstream seeding effect (Orlanski, 2005), which tends to result in the occurrence of the MWS of the NPST (Penny et al., 2010, 2013). The significantly positive EGR anomalies over the south side of Japan can be explained by the strengthened East Asian trough (Fig. 6b) which guides cold air farther south. In this way, warm air encounters equatorward shifting cold air at relatively low latitudes, resulting in an enhanced meridional air temperature gradient. MWS-related EGR anomalies present a dipole structure over the North Pacific, with a positive anomaly emerging over the northwest side of the Hawaiian Islands and a negative anomaly located over the area south of the Bering Strait. The positive anomalies can be attributed to the conspicuously negative SLP or to geopotential height anomalies over the North Pacific (Fig. 6), which are accompanied by a strengthened Aleutian low or an anomalous cyclone. The westerly wind is intensified (weakened) on the south (north) side of the anomalous cyclone, corresponding to the positive (negative) EGR anomalies over the North Pacific.

Figure 9.  Vertically integrated (from 925 to 700 hPa) EGR (a; shading; units: 10^–6 s) and 250 hPa zonal wind (b; shading; units: m s^–1) anomalies in midwinter regressed upon the standardized MSI. The contours represent the climatology of the midwinter lower-tropospheric EGR and 250 hPa zonal wind with intervals of (a) 2 × 10^–6 s and (b) 10 m s^–1, respectively. Stippling denotes anomalies that are significant at the 95% confidence level as determined by a Student’s t-test.

According to the linear theory of baroclinic instability (Charney, 1947; Eady, 1949), a positive EGR anomaly coinciding with enhanced atmospheric baroclinicity is expected to feed high-frequency transient eddies and shape a well-organized storm track (Hoskins and Valdes, 1990). However, the significantly positive EGR anomalies over the North Pacific are not accompanied by at strengthened MNPST (Fig. 4b). In fact, except for the MWS detected by Nakamura (1992), some other previous studies have also noticed that such linear theory does not necessarily reflect the actual behavior of highly nonlinear high-frequency transient eddy activities. For example, when investigating the NPST response to mesoscale sea surface temperature anomalies, Kuwano-Yoshida and Minobe (2017) and Ma et al. (2017) found substantial NPST anomalies despite little change in the EGR. Additionally, mismatches between the anomalous NPST and EGR values also took place when studying the roles of topography (Park et al., 2010; Lee et al., 2013) and two types of El-Niño events (Yang et al., 2020c) on the NPST. To better examine the relationship between the MNPST and low-level atmospheric baroclinicity, we define the EGR index as the average of the peak domain of the EGR over 25°–50°N, 120°E–170°W. The correlation coefficient between the MSI and the EGR index is –0.07 (Fig. 7d) and is insignificant. In addition, there are some significantly positive MNPST anomalies observed in the regressed pattern with respect to the standardized EGR index (Fig. 8d), which implies that the linear theory of baroclinic instability may not be appliable to explain the variation in the MNPST.

Many previous studies have paid particular attention to the role of westerlies on the MWS of the NPST (e.g., Nakamura, 1992; Christoph et al., 1997; Chang, 2001; Nakamura and Sampe, 2002; Harnik and Chang, 2004). The spatial patterns of anomalous upper-tropospheric zonal winds that are related to the MWS (Fig. 9b) are basically consistent with those of the EGR due to the definition of the EGR. Although there are few significant anomalies in the vicinity of the upper-level jet core, distinctly positive upper-level westerly wind anomalies are found along the southern flank, particularly downstream of the climatological midwinter upper-level jet. The increased upper-level jet associated with the MWS observed in this study agrees with the results of Christoph et al. (1997) and Nakamura (1992), who argued that when the zonal wind speed exceeds 45 meters per second, the intensified upper-level jet tends to favor the occurrence of the MWS due to the advection of synoptic-scale eddies by an accelerated group velocity. In addition, some significantly negative westerly wind anomalies occurring on the northern flank of the upper-level jet contribute to decreasing the zonal wind speed. Note that this regressed pattern of midwinter upper-level westerlies over the North Pacific (Fig. 9b) is broadly in alignment with the distribution of regression coefficients obtained by regressing the 300 hPa zonal wind values in January upon the IVI (Deng and Mak, 2006).

Figure 10 presents the regressed patterns of the 850 hPa air temperature and meridional wind anomalies with respect to the standardized MSI. Significantly negative lower-tropospheric air temperature anomalies (Fig. 10a) accompanied by anomalous northerlies (Fig. 10b) lie over eastern China and in the area upstream of the MNPST, which may be attributed to the intensified East Asian winter monsoon and East Asian trough (Fig. 6). According to previous studies on the upstream seeding effect (Orlanski, 2005; Lee et al., 2010; Penny et al., 2013; Yang et al., 2020c), anomalous low-level cooling upstream of the MNPST may result in an attenuation of the seeding effect with an increase in lower-tropospheric static stability. The weakened upstream seeding effect generates fewer baroclinic disturbances that enter the storm track (Orlanski, 2005; Penny et al., 2013), thus suppressing the development of the MNPST. The meridional dipole of air temperature anomalies over the North Pacific coincides with enhanced North Pacific atmospheric blocking. Note that the spatial pattern of anomalous air temperature is generally consistent with that of the geopotential height pattern (Fig. 6b) but with an opposite sign. The anomalous northerlies roughly alternate with anomalous southerlies along the entire latitude circle, which may imply that the MWS is associated with some kind of global wave train. This is in agreement with the findings of Park and Lee (2020), who believed the MWS should be investigated from a global perspective.

Figure 10.  850 hPa air temperature (a; shading; units: °C) and 850 hPa meridional wind (b; shading; units: m s^–1) anomalies in midwinter regressed upon the standardized MSI. The contours represent the climatology of the midwinter 850 hPa air temperature and meridional wind with intervals of (a) 5°C and (b) 2 m s^–1, respectively. Stippling denotes anomalies that are significant at the 95% confidence level as determined by a Student’s t-test.

In light of the barotropic governor mechanism (James, 1987), strengthened barotropic wind shear can restrict baroclinic instability and give rise to an inefficiency in the barotropic energy conversion (BTEC) by changing the eddy structure. During midwinter, the barotropic wind shear is enhanced owing to the dipole pattern of anomalous upper-level westerlies related to the MWS (Fig. 9b). Given that climatological winter-mean BTEC calculated from the mean kinetic energy of the time-mean flow to the EKE is negative (Cai et al., 2007), the resultant inefficient BTEC resulting from the intensified barotropic wind shear weakens the EKE loss and suppresses the barotropic damping effect on the MNPST, indicating that the inefficient BTEC is not beneficial to the occurrence of the MWS. Considering the important role of eddy energy in storm track activities (Deng and Mak, 2006; Zhao and Liang, 2019), we next discuss the eddy energy associated with the MWS of the NPST. According to the EAPE and EKE budget equations (Takahashi and Shirooka, 2014), the BTEC from the mean kinetic energy to the EKE, the baroclinic energy conversion (BCEC) from the mean available potential energy to the EAPE, the buoyancy conversion (BC) from the EAPE to the EKE and the generation of EAPE by diabatic heating (GEDH) can be written as follows:

where ${\boldsymbol{V}}$ and ${{\boldsymbol{V}}}_{3}$ are the two- and three-dimensional winds, respectively, $ {\nabla }_{3} $ denotes the gradient operator, $ R $ is the gas constant for dry air, $ P $ is the pressure, $ \omega $ is the vertical wind in the pressure coordinate, and $ Q $ denotes the diabatic heating estimated from the thermodynamic energy equation (Yanai et al., 1973). The prime, hat, tilde, and overbar signify the high-frequency (2.5–6 days), intermediate-frequency (7–29 days), intraseasonal-frequency (30–90 days), and low-frequency fields (more than 90 days), respectively. It should be emphasized that although these local energetics do not conserve energy (Xu and Liang, 2017; Zhao and Liang, 2019), it is still quite useful to make a quantitative discussion.

To identify the eddy energy anomalies related to the MWS, Fig. 11 shows the regressed vertically integrated (from 925 to 250 hPa) aforementioned eddy energy budget terms with respect to the standardized MSI. Significantly positive BTEC anomalies emerge over the International Date Line with increases in the MSI (Fig. 11a), indicating that an increase in the MWS is accompanied by less kinetic energy transfer from high-frequency transient eddies to the time-mean flow. This is consistent with the above analysis based on the barotropic governor mechanism and the results of Lee et al. (2013), who reported that the existence of the Tibetan Plateau accelerates the upper-tropospheric jet and impairs barotropic damping. Although Deng and Mak (2006) concluded that evident intensification of barotropic damping is a primary contributing factor to the MWS, their study was focused on a comparison between early/late winter and midwinter, which is not contradictory to our results. The MWS-related meridional dipole pattern of anomalous upper-level zonal winds over the North Pacific tends to be accompanied by a narrowing of and a slightly southward migrating jet (Fig. 9b), and thus the area where the jet overlaps with the MNPST may decrease, which may be responsible for the inefficient BTEC and reduced EKE loss.

Figure 11.  Vertically integrated (from 925 to 250 hPa) eddy energy budget terms in midwinter (contours; units: m² s^–2 d^–1) and their anomalies (shading) regressed upon the standardized MSI. The (a) BTEC, (b) BCEC, (c) BC, and (d) GEDH values are shown. Stippling denotes anomalies that are significant at the 95% confidence level as determined by a Student’s t-test.

Additionally, there are significantly negative BCEC anomalies that occur in the peak climatological area (Fig. 11b), indicating that less mean available potential energy is transferred to the EAPE when the MWS is evident. The MWS-related dipole structure of EGR anomalies (Fig. 9a) tends to keep the baroclinic zone away from the MNPST, which may reduce the energy obtained by the MNPST from the baroclinic zone and result in a weakened BCEC. Note that the region of the large BCEC anomaly is not consistent with that of the EGR anomaly, which indicates that the connections between atmospheric baroclinicity, baroclinic energy conversion, and storm track amplitudes are extremely complex over the North Pacific (Chang et al., 2002; Nakamura et al., 2002; Park et al., 2010; Lee et al., 2013). Given that the BCEC is determined by contributions from different time scales, to further investigate the influence of which time scale is most important, Fig. 12 shows the regressed, vertically integrated (from 925 to 250 hPa) BCEC contributed by different time scales with respect to the standardized MSI. The anomalous BCEC contributed by the intermediate-frequency time scale, with tiny amplitudes, is relatively noisy and sporadically located over the North Pacific (Fig. 12a). Most anomalies fail to pass the statistical significance test, indicating that the intermediate-frequency time scale seems to be trivial to the BCEC. In addition, negative BCEC anomalies occurring over the Bering Sea from the intraseasonal-frequency time scale (Fig. 12b) slightly contribute to the MWS-related BCEC anomalies. In contrast, the spatial pattern and amplitude of the BCEC anomalies contributed by the low-frequency time scale (Fig. 12c) are almost consistent with those of the MWS-related BCEC anomalies (Fig. 11b), which suggests that the low-frequency time scale is most important to the BCEC.

Figure 12.  Vertically integrated (from 925 to 250 hPa) BCEC contributed by the (a; shading) intermediate-frequency, (b; shading) intraseasonal-frequency and (c; shading) low-frequency time scales in midwinter (contours; units: m² s^–2 d^–1) and their anomalies (shading) regressed upon the standardized MSI. Contour intervals are (a) 0.5, (b) 0.3, and (c) 15 m² s^–2 d^–1, respectively. Stippling denotes anomalies that are significant at the 95% confidence level as determined by a Student’s t-test.

Furthermore, the anomalous spatial distribution of the BC related to the MWS (Fig. 11c) is similar to that of the BCEC, which suggests that high-frequency transient eddies tend to weaken due to the attenuated energy conversion from the EAPE to the EKE. The MWS-related, negative air temperature anomalies over the East China Sea (Fig. 10a) tend to be associated with the intensified static stability, which may suppress BC by limiting vertical motion. The role of diabatic heating in the MWS has been discussed by Chang (2001), who argued that diabatic heating acts to dissipate eddy energy during midwinter. From Fig. 11d, we can see that significantly negative GEDH anomalies associated with the MWS are located in the climatological midwinter-mean center, implying that the EAPE is lost due to diabatic heating during midwinter, which is consistent with Chang (2001).

In the present study, the atmospheric circulation anomalies associated with the midwinter suppression (MWS) of the North Pacific storm track (NPST) in the Northern Hemisphere are investigated based on the NCEP/NCAR global reanalysis dataset spanning 1948 to 2018. Given that the MWS is more remarkable in upper levels, the NPST is quantified by $\sqrt{\overline{z{'}z{'}}}$, which is one of the most commonly used Eulerian diagnostics. The main conclusions are as follows.

The midwinter NPST (MNPST), characterized by $\sqrt{\overline{z{'}z{'}}}$, attains two maximum amplitudes at approximately 45°N in November and April. The MWS phenomenon is apparent throughout the period. The long-term trend in MNPST indicates an interdecadal intensification. Motivated by the intraseasonal variability index defined by Deng and Mak (2006) and the consideration of monthly mean march in the $\sqrt{\overline{z{'}z{'}}}$ amplitude, we define a midwinter suppression index (MSI) that can describe the MWS of the NPST measured by $\sqrt{\overline{z{'}z{'}}}$ as well.

Based on this MSI, regression and correlation analyses are used to examine the atmospheric circulation anomalies and eddy energy budget related to the MWS. In terms of the SLP field, the MWS is accompanied by significantly positive anomalies in Eurasia and negative anomalies over the North Pacific, which correspond to the strengthened East Asian winter monsoon. There are significantly negative 500 hPa geopotential height anomalies related to the MWS occurring over eastern China, corresponding to the enhanced East Asian trough. Furthermore, there is a meridional dipole structure of anomalous geopotential heights with positive anomalies emerging over the East Siberian Sea and negative anomalies arising over the area south of the Aleutian Islands, which coincides with intensified North Pacific atmospheric blocking. Additionally, significantly negative low-level air temperature anomalies, accompanied by anomalous northerlies, lie in the upstream area of the MNPST. The MSI has a strong positive correlation with the EAWM index, East Asian trough index, and North Pacific atmospheric blocking index, indicating that the occurrence of the MWS is indeed associated with an intensified EAWM, East Asian trough, and atmospheric blocking in the North Pacific. However, it should be emphasized that the relationship between the MWS and the low-level atmospheric baroclinicity is somewhat puzzling.

From the perspective of the eddy energy budget, the MWS of the NPST is accompanied by significantly negative BCEC anomalies in its climatological midwinter-mean peak area, suggesting that the occurrence of the MWS is related to lower mean available potential energy transferred to the EAPE. The anomalous spatial pattern of the BC is consistent with that of the BCEC. In addition, there are significantly negative GEDH anomalies associated with the MWS, indicating that the occurrence of the MWS is related to EAPE loss due to diabatic heating during midwinter. However, significantly positive BTEC anomalies are associated with the barotropic governor mechanism, indicating that the BTEC does not relate to the occurrence of the MWS.

In general, our results support that local perspectives (e.g., Nakamura, 1992; Chang, 2001; Nakamura et al., 2002; Deng and Mak, 2006) are important to the MWS of the NPST. To be specific, the MWS of the NPST may be influenced by the EAWM, East Asian trough, North Pacific atmospheric blocking, upper-level jet, baroclinic energy conversion, and diabatic heating. Additionally, the atmospheric circulation anomalies in the Northern Hemisphere suggest that a non-local perspective (Penny et al., 2010; Park and Lee, 2020) is also needed to investigate the MWS of the NPST. More specifically, Ural blocking in the upstream region of the MNPST may be another candidate mechanism associated with the MWS of the NPST. It must be emphasized that since all the regression analyses in this study are simultaneous instead of lead-lag regression analyses and there are interactions between the time-mean flow and the transient eddies (Lau and Holopainen, 1984; Lau, 1988), it is difficult to distinguish their causal relationship. A rigorous causality formalism based on the Liang-Kleeman information flow theory (Liang and Kleeman, 2005; Liang, 2014) which has been widely used for cause-effect analyses may help to reveal the reason behind the MWS in future studies.

In addition, the possible roles of ocean processes are not considered in this paper. Many previous studies highlighted the impacts of oceanic surface heating, mesoscale eddies, or fronts on storm track activities (e.g., Ren et al., 2008; Zhu and Li, 2010; Small et al., 2014; Yuan and Xu, 2016; Kuwano-Yoshida and Minobe, 2017; Ma et al., 2017). In fact, the one-month lead time regression of sea surface temperature (SST) anomalies with respect to the standardized MSI show that significant negative SST anomalies are located on the north side of the Hawaiian Islands, and positive SST anomalies are located in the Kuroshio region (figure not shown). In future studies, the consideration of ocean processes may shed more light on the MWS of the NPST.

Acknowledgements. The authors thank two anonymous reviewers and Associate Editor-in-Chief for their helpful and crucial comments which improved the manuscript substantially. Many thanks are given to the editor for finding excellent reviewers. The NCEP/NCAR reanalysis dataset was obtained online (https://www.esrl.noaa.gov/psd/data/gridded/reanalysis/). This research was supported by the National Key Research and Development Program of China (Grant No. 2018YFC1505901) and the National Natural Science Foundation of China (Grant Nos. 41490642, 4160501, and 41520104008).