This study investigated the extratropical circulation anomalies responsible for cold surges over the South China Sea in winter. The surge events were identified by the intensity of northerly winds over 110°–117.5°E along 15°N at 925 hPa. Two distinct patterns of sea level pressure (SLP) anomalies in East Asia were found to have a crucial role in inducing cold surges over the South China Sea. Accordingly, the cold surge events were classified into two types. The first type of cold surge is characterized by a pair of SLP anomalies with positive and negative ones centered over China and Japan, respectively, whereas the second type of cold surge is characterized by widespread and persistent positive SLP anomalies over East Asia. Furthermore, the first type of cold surge is accompanied by a deepened East Asian trough and precursory Rossby wave trains across the Eurasian continent in the mid- and upper troposphere, but the latter is not. Prior to both types of the cold surges, the Siberian high is significantly intensified. However, diagnosis of the SLP tendency indicates that the intensification is related to different physical processes. In the first type of cold surge, the Rossby wave trains favor negative vorticity advection and cold advection, inducing intensification of the Siberian high. By contrast, in the second type of cold surge, vorticity advection can be ignored due to the lack of Rossby wave trains, and only the lower-tropospheric cold advection induced by anomalous northerly winds, resulting from the anomalous Siberian high, contributes to the further intensification of the Siberian high.
A cold surge over the South China Sea (SCS) is characterized by a period of strong winds and is one of the most hazardous weather phenomena occurring during the Asian winter monsoon (Ramage 1971; Lau and Chang 1987; Chang et al. 2006; Chang et al. 2011). The arrival of a cold surge is often accompanied by the outbreak of northerly winds, an increase in surface pressure, and a decrease in the surface air temperature (Chang et al. 1983; Boyle and Chen 1992; Chan and Li 2004; Samah et al. 2016). A cold surge originating from the extratropical region can lead to a drop of air temperature over the northern SCS, such as at Hong Kong (e.g., Chang et al. 2006; Lu and Chang 2009). With the southward extensions of northerly winds, cold surges can also result in the intensification of atmospheric convection over the Maritime Continent (e.g., Chang et al. 1979; Slingo 1998; Chang et al. 2005; Lim et al. 2017; Pang et al. 2018) and extreme rainfall events and floods over Vietnam (Yokoi and Matsumoto 2008), the Philippines (Pullen et al. 2015), Malaysia (Johnson and Houze 1987; Johnson and Chang 2007; Tangang et al. 2008), and Indonesia (Wu et al. 2007).
The occurrence of a cold surge over the SCS is associated with disturbances in midlatitude circulation patterns. The onset of a cold surge is recognized to be a result of an amplification of the Siberian high and its southward expansion (Ding 1990; Wu and Chan 1995; Zhang et al. 1997). With prevailing northeasterly winds over the southeastern flank of the region of high pressure, cold surges carry intensely cold air into the SCS (Ding 1990; Wu and Chan 1995; Zhang et al. 1997; Chan and Li 2004; Chang et al. 2006). An intensified East Asian trough can also affect cold surges by inducing northerly flows, which carry cold air into lower latitudes to the west of the trough (Boyle 1986; Wu and Chan 1997; Lu and Chang 2009).
The Siberian high and the East Asian trough, which both have a strong influence on the East Asian winter climate (e.g., Gong et al. 2001; Wu and Wang 2002; Yang et al. 2002; Jhun and Lee 2004; Chen et al. 2005; Wu et al. 2006; Wang et al. 2009; Ha et al. 2012; He et al. 2017), have been widely investigated. Climatologically, the development of the Siberian high is attributed to strong cooling (Ding and Krishnamurti 1987; Ding 1990). However, the evolution of the Siberian high is related to the eastward propagation of synoptic wave packets in the middle and upper troposphere, which amplifies ridges over northern Siberia (Wu and Chan 1995; Takaya and Nakamura 2005a; Park et al. 2008; Zhou et al. 2009; Lin et al. 2018; Shi et al. 2019). Variations in the East Asian trough are also affected by extratropical disturbances or wave trains over the Eurasian continent (Joung and Hitchman 1982; Takaya and Nakamura 2005b; Hong et al. 2009; Park et al. 2014; Song and Wu 2017). The Siberian high and the East Asian trough show concurrent variations (Ding and Krishnamurti 1987; Zhang et al. 1997; Yang et al. 2002; Jhun and Lee 2004).
Based on previous studies, it can be hypothesized that cold surges over the SCS, which are favored by an intensified Siberian high and deepened East Asian trough, may be affected by extratropical disturbances or wave trains over the Eurasian continent. We considered the large-scale circulations responsible for northerly flows over East Asia, which are a crucial precursor of cold surge events over the SCS, and found that there are two patterns of large-scale circulation over East Asia that favor cold surges over the SCS. We then investigated the extratropical disturbances over the upstream Eurasian continent corresponding to these two circulation patterns and the impact of these upstream upper tropospheric disturbances on the intensification of the Siberian high.
2. Data and methods
The data used in this study are obtained from the European Centre for Medium-Range Weather Forecasts (ECMWF) interim reanalysis (ERA-Interim) dataset (Dee et al. 2011) with a horizontal resolution of 0.75° × 0.75°. The daily data are averaged from the 6-h data. Our analysis covers the time period 1979–2016, with the omission of 29 February in leap years. Daily anomalies are obtained by removing the 38-yr mean of that particular day from the raw data. The boreal winter is defined as the period from November to February (NDJF) when cold surges are active (Lim et al. 2017).
The cold surge index is computed as the 925-hPa meridional wind averaged over longitudes 110°–117.5°E along 15°N (Chang et al. 2005). For a cold surge day to be counted, a surge index <0.75 standard deviations from the climatological mean (−5.35 m s−1), or −8.55 m s−1, is required (Lim et al. 2017). Accordingly, we identified 1060 cold surge days during the time periods of analysis in this study, accounting for 23% of the total days. This is roughly consistent with the value (20%) obtained by Chang et al. (2005), who used the National Centers for Environmental Prediction (NCEP) Reanalysis data for the time period 1979–2001 for DJF. The continuous surge days are viewed as a cold surge event and the first occurrence day is used as the reference (day 0). A total of 344 events are detected during the 38 boreal winters, which corresponds to nine events per year with a mean duration of about 3 days.
A composite analysis is performed to investigate the large-scale anomalies associated with cold surge events, where days −N and N refer to the composite at N days before and after the reference day, respectively. The effective degree of freedom is calculated for each variable at each grid. The two-tailed Student’s t test is used as the significance test by comparing the composite days with the remaining days during the boreal winter. A composite wind vector is considered significant if either its zonal or meridional component passes the test.
To diagnose the variations in sea level pressure (SLP) associated with cold surges, we adopt a simplified version of the pressure tendency equation following Carlson (1991):
where the subscripts 5 and 0 denote heights of 500 and 1000 hPa, respectively; p is the SLP; V = (u, υ) represents the horizontal winds; ∇ represents the horizontal gradient operator; and ζ, f, and Z are relative vorticity, planetary vorticity, and geopotential height, respectively. Also, a and b are parameters defined as
The parameters a and b, representing the amplification factors for advection, are estimated to be 2.6 × 105 hPa s−1 and 0.08 hPa m−1, respectively, assuming wavelength L = 5000 km, Rossby radius of deformation LR = 1000 km, feature scale of Coriolis parameter f0 = 10−4 s−1, acceleration of gravity g = 9.8 m s−2, hypsometric constant K0 = 8 m hPa−1, Δp3 = 300 hPa, Δp5 = 500 hPa, and p* = 1000 hPa.
The two terms on the right-hand side of Eq. (1) represent the contributions of the midtropospheric absolute vorticity advection and lower tropospheric temperature advection, respectively. Equation (1) shows that the SLP increases (decreases) in response to negative (positive) vorticity advection and cold (warm) advection.
We also examine the individual terms contributing to the advection by dividing the variables into their climatological winter means and anomalies. For instance, the temperature advection can be written as
where h is the thickness between 500 and 1000 hPa (h = Z5 − Z0). The overbar symbol denotes the mean of whole winter days and the prime symbol denotes the perturbation obtained by removing the winter mean from the raw data. By removing the winter mean (), the remaining three terms on the right-hand side of the equation refer to the temperature advection induced by the winter-mean winds and anomalous thickness, by the anomalous winds and winter-mean thickness, and by the wind and thickness anomalies, respectively.
3. Classification of cold surge events
Figure 1 shows the climatological winter-mean SLP, 925-hPa horizontal wind, and 850-hPa air temperature. The pressure is characterized by the Siberian high and Aleutian low. Correspondingly, northwesterly winds prevail over East Asia, conditions known in this region as the winter monsoon. Along the southeastern flank of the Siberian high, prevailing northeasterly winds are found over southern China and the adjacent seas, including the SCS. These northeasterly winds merge into easterly winds in the tropics. These East Asian winter monsoonal flows mean that East Asia is much cooler than other regions at the same latitude and result in the strongest meridional gradient appearing over coastal East Asia. In addition to the winter monsoonal flows, strong prevailing winds appear over the middle to high latitudes on the Eurasian continent, along the northern flank of the Siberian high.
Figure 2 shows the composite circulations at the onset of the cold surge (day 0). The northeasterly or northerly anomalies are dominant along 120°E and penetrate into the SCS (Fig. 2a). These northerly anomalies are attributed to the strong zonal gradients in SLP over East Asia. There is a strong and positive SLP anomaly over eastern China, which is associated with the southeastward extension of the Siberian high, while a negative SLP anomaly lies over Japan and the western North Pacific.
Significant circulation anomalies also appear in the mid- and upper troposphere (Figs. 2b,c). The negative geopotential height anomalies lie over East Asia (more exactly, over the region of the climatological East Asian trough), indicating an intensification of the trough (Fig. 2b). In addition, there are positive geopotential height anomalies over the central Eurasian continent, accompanied by a northeast–southwest tilted ridge, similar to the results of Bueh et al. (2011). These negative and positive anomalies align with the climatological trough and ridge, suggesting that the cold surges over the SCS are favored by meandering flows in the mid- and upper troposphere over the Eurasian continent. There are significant upper tropospheric zonal wind anomalies (Fig. 2c) corresponding to the anomalies in geopotential height. A positive zonal wind anomaly over subtropical East Asia indicates an enhanced East Asian upper tropospheric jet stream, which is consistent with the intensification of the East Asian trough. There are also positive zonal wind anomalies extending northeastward from the Caspian Sea and negative anomalies centered over northeastern Asia. The surface air temperature anomalies are significantly negative in East Asia and positive over the region extending from the Caspian Sea northeastward to the central Siberian plateau (Fig. 2d). The negative temperature anomalies are consistent with the higher SLP and northerly wind anomalies (Fig. 2a). Positive temperature anomalies appear to the northwest of the anticyclonic anomaly (Fig. 2b), suggesting that meridional flows may also be responsible for these temperature anomalies.
The circulation and temperature anomalies associated with the onset of the cold surge in the SCS are generally consistent with previous results obtained for stronger East Asian winter monsoons (e.g., Chang and Lau 1980; Jhun and Lee 2004; Ryoo et al. 2005; Chang et al. 2006; Hong et al. 2009; Lu and Chang 2009). However, the circulation and temperature anomalies reported in these previous studies exhibited varied features depending on the monsoon regions studied and the time scales of monsoonal variability considered.
Figure 3 shows a scatterplot of two indices of SLP anomalies at day 0 for all 344 cold surge events. Here, the western index (W-index) is defined as the average SLP anomalies over 15°–45°N, 100°–120°E and the eastern index (E-index) as the average anomalies over 15°–45°N, 130°–150°E. These two regions are shown as blue and red boxes in Fig. 2a. The results shown in Fig. 3 confirm that the cold surge events are associated with a pair of SLP anomalies (as shown in Fig. 2a). In total, 162 (47%) cold surge events show a positive W-index and a negative E-index (illustrated by the dots in the second quadrant).
A large number (110) of cold surge events show both a positive W-index and a positive E-index (illustrated by the dots in the first quadrant). This indicates that many events are distinct from the composite results shown in Fig. 2. The W-index averaged over all the cold surge events is 2.7 hPa, much stronger than the average E-index (−1.2 hPa). The W-index is positive in most of the cold surge events (80%), but the E-index does not show such a strong tendency, with 60% of the events being negative and 40% positive. These results suggest that the positive SLP anomalies over the western region have a crucial role in inducing cold surge events, whereas the SLP anomalies over the eastern region show a greater diversity. The correlation coefficient between the two indices is only 0.07, suggesting that the variations in SLP over these two regions are roughly independent.
The diversity of the E-index suggests that different physical mechanisms may be responsible for the cold surge events. There are fewer events with a negative W-index, illustrated by the dots in the third and fourth quadrants. We found that these events tend to be related to tropical circulation anomalies (not shown). Because we focus on the extratropical circulations associated with cold surge events in this study, we exclude the cold surge events with a negative W-index from further analysis. We classify the cold surge events with a positive W-index into two types (positive and negative) based on the sign of the E-index—that is, 110 and 162 events in the first and second quadrants, respectively.
Figure 4 compares the day 0 composite circulations between the two types of cold surge events. The negative type is featured by a pair of SLP anomalies (Fig. 4a), similar to Fig. 2a, but with a much greater intensity in both the western and eastern regions. The W-index and E-index averaged for this type of cold surge are 4.0 and −3.6 hPa, respectively (versus 2.7 and −1.2 hPa averaged over all the cold surge events). In contrast with the pair of SLP anomalies in the negative type of cold surge event, the positive type is characterized by widespread high pressure anomalies covering most of East Asia, with two centers over eastern China and Japan, respectively (Fig. 4b). Anomalous northerly winds appear over East Asia for the negative cold surge events as a result of the greater pressure gradient associated with a pair of the SLP anomalies (Fig. 4a). These northerly anomalies extend southward from northeast Asia into the SCS. By contrast, for the positive cold surge events, the northerly anomalies are limited to the south of 30°N along coastal East Asia and anticyclonic wind anomalies are centered over Japan in association with the eastward extension of the high pressure anomalies (Fig. 4b).
The most remarkable differences in the mid- and upper troposphere between the two types are seen in the East Asian trough and westerly jet. For the negative type of cold surge event, the East Asian trough is intensified, indicated by the significant negative geopotential height anomalies over East Asia (Fig. 4c) and in agreement with the stronger East Asian trough. The East Asian jet escalates, indicated by the significant positive zonal wind anomalies to the south of Japan (Fig. 4e). However, there is no negative geopotential height anomaly over East Asia in the positive cold surge events (Fig. 4d) and the positive zonal wind anomalies are weak to the south of Japan (Fig. 4f), suggesting that both the East Asian trough and East Asian jet are not significantly changed. The significant and positive zonal wind anomalies appear to the east of Japan, corresponding to the eastward shift of the core of the East Asian jet relative to the negative type of cold surge event (Fig. 4e). Both types are characterized by cooler temperatures in East Asia and warmer temperatures in the central Eurasian continent (Figs. 4g,h), similar to the composite result for all cold surge events (Fig. 2d).
4. Time evolution of extratropical circulation anomalies associated with cold surge events
The differences in circulation described in the preceding section suggest that the negative and positive types of cold surge event may be induced by different physical processes. In this section, we investigate the evolution of extratropical circulation associated with these two types of event. Figure 5 shows the evolution of SLP and 925-hPa wind anomalies for the negative type of cold surge event. A pair of pressure anomalies can be traced back to the North Atlantic on day −6, with an anticyclonic anomaly over the North Atlantic and a cyclonic anomaly over western Europe (Fig. 5c). An anomalous region of high pressure is triggered downstream over western Siberia on day −4, accompanied by a slight eastward propagation of the pair of SLP anomalies upstream (Fig. 5d). On day −2, the anomalous high pressure region expands eastward over Siberia and reaches its maximum strength, indicating amplification of the Siberian high (Fig. 5e). At the same time, northerly anomalies appear over Mongolia and eastern China. The Siberian high moves southward and an anomalous area of low pressure and a cyclone are formed over the Sea of Japan on day −1 (not shown). On the onset days of the cold surge events (day 0), the cyclonic flows are further enhanced and the anomalous northerly winds are dominant along 120°E (Fig. 5f). These circulation anomalies are similar to those associated with outbreaks of cold air in midlatitude East Asia (e.g., Shoji et al. 2014, their Fig. 6; Abdillah et al. 2017, their Fig. A1), but show a farther southward distribution focused on the cold surge in the SCS. The anticyclone then moves southeastward and declines over the East China Sea, while the cyclone propagates northeastward to the Aleutian Islands (Fig. 5h).
By contrast, the evolution of circulation anomalies for the positive type of cold surge event, shown in Fig. 6, indicates a different development process for the Siberian high. Without the wavelike pattern seen upstream in the negative type of cold surge event (Fig. 5), the Siberian high for the positive type is amplified from positive pressure anomalies over western Siberia on day −8 (Fig. 6b). The anomalous area of high pressure develops to its strongest intensity over Siberia around day −2 (Fig. 6e), but is located farther to the south and east than in the negative cold surge events (Fig. 5e). Accordingly, northerly or northeasterly anomalies appear over coastal East Asia and this region is shifted farther south and east than in the negative type of cold surge event, consistent with the difference in location of the positive anomalies in SLP. The anomalous high pressure then divides into two centers: one moves south to eastern China and the other moves east to northern Japan (Fig. 6f). Corresponding to these anomalies in SLP, northwesterly anomalies appear over the midlatitude western North Pacific, becoming easterly anomalies in the subtropics and northeasterly anomalies over South China and the SCS. Both the pressure and wind anomalies are weakened after day 0 (Figs. 6g,h).
The temporal and spatial evolutions of anomalies in SLP are related to variations in the midtropospheric geopotential height anomalies (Figs. 7 and 8). The evolution of height anomalies is seen as an eastward propagation of a Rossby wave train for the negative type of cold surge event (Fig. 7). The wave packets emanate from an anticyclonic anomaly over the North Atlantic and a cyclonic anomaly over western Europe on day −6 (Fig. 7c). These anticyclonic and cyclonic anomalies are enhanced and shift eastward, triggering a downstream anticyclonic anomaly over the Ural Mountains on day −4 (Fig. 7d). Although the upstream anomalies become weaker, the anticyclonic anomaly is intensified over the central and northern Eurasian continent on day −2 (Fig. 7e) and is weakened on day 0 when the cyclonic anomaly over East Asia reaches its maximum (Fig. 7f). The disturbances then propagate eastward into the North Pacific and become weakened (Figs. 7g,h). The wavelike anomalies resemble those related to the wave train type cold surge over East Asia in Park et al. (2014; their Fig. 3), but the positive height anomaly over East Asia in their results is absent in this study.
The evolution of midtropospheric circulation anomalies for the positive type of cold surge event (Fig. 8) is very different from that for the negative type of cold surge event. The most striking differences are that the eastward propagation and energy dispersion of wavelike disturbances are not as clear as those for the negative type of cold surge event and there is less intensification of the East Asian trough. The anticyclonic anomaly over the central and northern Eurasian continent is relatively weaker than that for the negative type of cold surge event, but it tends to persist for a longer period of time. This anomalous anticyclone appears significantly from day −6 and reaches its maximum on day −2, with a slight eastward shift. These differences in midtropospheric circulation anomalies between the two types of cold surge event are generally consistent with the differences in SLP.
We calculated the indices of Park et al. (2014), Shoji et al. (2014), and Abdillah et al. (2017), which have been used to depict the cold events over midlatitude East Asia. The results of these indices indicate that the midlatitude circulations tend to be stronger prior to the SCS cold surge events, including both the negative and positive types, with a peak approximately on day −1. However, they all display a great diversity among the SCS cold surge events, due to the different focuses between these previous studies and this study.
5. Possible mechanisms: Analysis of SLP tendency
The preceding section reported that there are large differences in upstream circulation between the two types of cold surge event. The contrasting SLP anomalies over Japan between the two types of event are associated with different features of the East Asian trough in the midtroposphere. However, both types of event are related to the intensification of the Siberian high about two days before the onset of a cold surge (Figs. 5e and 6e). It is therefore reasonable to hypothesize that the distinct upstream disturbances may influence the enhancement of the Siberian high via different mechanisms. We now diagnose the tendency of the SLP, which is widely used to describe the Siberian high, to investigate the mechanisms responsible for the intensification of the Siberian high.
Figure 9 shows the area-averaged evolution of the SLP tendency and two advection terms to measure their contributions. Based on Eq. (1), the variations in SLP are determined by the midtropospheric vorticity advection and lower tropospheric temperature advection. The key areas of the Siberian high are selected as 40°–70°N, 70°–120°E for the negative type of cold surge event (Fig. 5e; green rectangle) and as 25°–55°N, 90°–130°E for the positive type of cold surge event (Fig. 6e; green rectangle), considering the discrepancy in the regions of strong SLP anomalies between the two types. We have shifted these key areas in the south or north directions, and found that the conclusions are not sensitive to the average areas, both for the negative and positive types. The tendency is calculated by a centered difference—for instance, the values on day 0 are obtained from the difference between day 1 and day −1. For both types of cold surge event, the SLP tendency is generally in agreement with the changes in vorticity and temperature advection around day −2. This confirms the validity of Eq. (1) and suggests that vorticity and temperature advection both have a crucial role in the evolution of the Siberian high. Some discrepancies, such as clearly weaker advection compared with the SLP tendency, are expected considering that Eq. (1) is a simplification and thus only suitable for qualitatively estimating the contribution of vorticity and temperature advection.
The results reveal the different contributors to the intensification of the Siberian high between the two types of cold surge event. Both vorticity and temperature advection make comparable contributions to the increase in SLP for the negative type of cold surge event, although the former tends to be slightly stronger than the latter (Fig. 9a). By contrast, for the positive type of cold surge event, only temperature advection contributes to the evolution of the SLP tendency, including the intensification of the Siberian high, whereas vorticity advection can be ignored (Fig. 9b). The distinct roles of vorticity advection between the two types of cold surge event can also be seen in Figs. 10a and 10b, which show the composites of the 500-hPa anomalous vorticity and winter-mean horizontal winds averaged over the peak period of SLP tendency (from day −4 to day −2). The prevailing westerly or northwesterly winds generally correspond to a strong gradient of anomalous vorticity over the key area for the negative type of cold surge event (Fig. 10a). By contrast, for the positive type of cold surge event, the vorticity anomalies are very weak over the key area (Fig. 10b), so that the vorticity advection can be ignored, although the prevailing winds are stronger over the key area, which is different from the key area for the negative type of cold surge event. The distinct strength of anomalous vorticity is consistent with that of the anticyclonic anomaly over the central and northern Eurasian continent. The cyclonic anomaly to the southeast of this anticyclonic anomaly is much stronger for the negative than for the positive type of cold surge event (Figs. 7d,e and 8d,e). Note that the ordinate scale for the positive type of cold surge event is half that for the negative type, indicating that the SLP tendency for the positive type of cold surge event is relatively weaker. On the other hand, the SLP for the positive type of cold surge event tends to increase earlier than for the negative type of cold surge event, consistent with the persistent enhancement of anomalous SLP over Siberia (Fig. 6).
Figure 11 shows the individual terms of temperature advection, which make important contributions to the intensification of the Siberian high for both types of cold surge event. The results are area-averaged over the same key areas with that in Fig. 9. Subscripts x and y denote partial derivatives in the zonal and meridional directions, respectively. In general, the meridional components are much larger than the zonal components, indicating that northerly winds are important in cold advection. For the negative type of cold surge event, both and contribute to cold advection (Fig. 11a), but for the positive type of cold surge event, is the main source of cold advection (Fig. 11b). This result is consistent with Figs. 10c–f. In these figures, only meridional winds or their anomalies are shown as vectors based on the overwhelming contribution of the meridional components.
For the negative type of cold surge event, the winter-mean southerly winds over western Siberia, which are associated with the prevailing southwesterly winds in this region (Fig. 1), lead to strong cold advection over the key area when there is a strong northward gradient of anomalous thickness (Fig. 10c). This strong northward gradient is associated with a pair of positive and negative thickness anomalies—which result from a 500-hPa positive height anomaly over the central and northern Eurasian continent and a negative anomaly to the southeast of this positive anomaly (Fig. 7)—and positive SLP anomalies in the key area (Fig. 5). The northerly anomalies over the southern part of the key area result in cold advection, but the southerly anomalies in the northwestern part lead to warm advection (Fig. 10e). The area-weighted mean of advection in the key area is cold advection (Fig. 11a). Although the southerly anomalies in the northwestern part of the key area are stronger than the northerly anomalies over the southern part, the latter correspond to a stronger gradient of winter-mean thickness than the former, resulting in overall cold advection.
By contrast, for the positive type of cold surge event, the anomalous northerly winds and the large southward gradient in winter-mean thickness over East Asia are responsible for cold advection and the resultant increase in SLP (Fig. 10f). In addition, the negative thickness anomalies correspond to the winter-mean northerly winds over the key area (Fig. 10d), leading to a much smaller value of (Fig. 11b).
Figure 11 indicates that the temperature advection caused by the wind and thickness anomalies is very weak for both types of cold surge event. We also examined vorticity advection at 500 hPa using a similar approach and found that the vorticity advection caused by wind and vorticity anomalies is also very weak. Therefore we conclude that the intensification of the Siberian high mainly depends on the configuration of the disturbances and winter-mean states, rather than the structure of the disturbances.
6. Summary and discussion
We investigated the extratropical circulation anomalies associated with cold surge events over the SCS during the boreal winters of 1979–2016 and found two distinct patterns of SLP anomalies in East Asia responsible for these events. The first pattern is characterized by positive and negative SLP anomalies centered over China and Japan, respectively, and by northerly anomalies over coastal East Asia. The second pattern is characterized by widespread positive SLP anomalies over East Asia and northeasterly anomalies along coastal East Asia limited to the south of 30°N. We refer to the cold surge events over the SCS associated with these two patterns as “negative type” and “positive type” events, respectively, according to the signs of the SLP anomalies centered over Japan. In addition to these lower tropospheric circulation anomalies, circulation anomalies in the mid- and upper troposphere also show distinct features in association with these two types of cold surge event. The East Asian trough and upper tropospheric westerly jet are significantly intensified in the negative type of cold surge event, but not in the positive type.
We made further comparisons in an attempt to find the mechanisms of development of the extratropical circulation anomalies for these two types of cold surge event. There is an eastward propagation of midtropospheric Rossby waves in the negative type of cold surge event, which can be traced back to disturbances over the North Atlantic. The upstream disturbances trigger an anticyclonic anomaly over central and northern Eurasia and a cyclonic anomaly over East Asia, leading to intensification of the Siberian high and East Asian trough. Further analysis showed that the amplification of the Siberian high results from negative vorticity advection in the midtroposphere and cold advection in the lower troposphere, both of which are related to wavelike anomalies in the extratropical circulation. The vorticity advection is mainly induced by a strong gradient of anomalous vorticity and prevailing northwesterly winds over Siberia. The temperature advection is attributed to the meridional components of both the anomalies and the winter-mean states of the temperature gradient and winds.
By contrast, for the positive type of cold surge event, the wavelike anomalies are weakened or absent over the Eurasian continent and the East Asian trough is not deepened. Under these conditions, the Siberian high is significantly intensified about two days before the onset of the cold surges, similar to the negative type of cold surge event. The mechanism for the intensification of the Siberian high is different, however. The vorticity advection can be ignored due to the weak circulation anomalies in the midtroposphere, whereas the anomalous northerly winds and strong southward gradient of the winter-mean temperature in the lower troposphere over East Asia cause strong cold advection and an increase in the SLP.
This study focuses on the extratropical precursors of the two types of cold surge event over the SCS. However, there are strong tropical–extratropical interactions over East Asia in winter. For instance, the Madden–Julian oscillation over the Maritime Continent and the western Pacific Ocean can modulate cold surges over East Asia (Jeong et al. 2005; He et al. 2011; Abdillah et al. 2018) and these cold surges, in turn, may trigger or affect the Madden–Julian oscillation (Chen et al. 2017; Hong et al. 2017; Pang et al. 2018). Therefore it would be interesting to investigate whether tropical–extratropical interactions show distinct features over East Asia and the western North Pacific in these two types of cold surge event, particularly under the different phases of the Madden–Julian oscillation. Furthermore, previous works have also documented the influence of El Niño–Southern Oscillation (ENSO) on the cold surges over East Asia (e.g., Zhang et al. 1997; Chen et al. 2004; Abdillah et al. 2017). So it is necessary to further discuss the response of cold surges over the SCS to ENSO. Finally, the basic states over the Eurasia can be affected by long-term changes, such as the Atlantic multidecadal oscillation or the decline of Arctic sea ice (e.g., Osborne et al. 2017; Li et al. 2018). The change in basic states might play a role in modulating the propagation of wave trains associated with the cold surges in the SCS (i.e., the present results may exhibit a long-term change). This hypothesis also requires further investigations in the future. Such studies may help us to better understand the physical mechanisms for the cold surge events in terms of both tropical–extratropical interactions and interactions on multiple time scales.
We sincerely thank Prof. Bueh Cholaw for his constructive suggestions and two anonymous reviewers for their comments and suggestions, which greatly helped us to improve the presentation of this paper. This work was supported by the National Natural Science Foundation of China (Grant 41721004).