Cold surges occur frequently over the South China Sea (SCS) in winter, and most of them last only a few days. However, some cold surge events can persist longer, for instance, more than 5 days. This study focuses on these long-lived cold surge events and investigates the associated extratropical circulation anomalies. The results indicate that long-lived cold surges, characterized as strong northerlies over the SCS, can be triggered by a successive high anomaly center over East Asia. Accompanying this is an anomalously extensive and quasi-stationary anticyclone over Siberia in the midtroposphere, hinting at a more frequent occurrence of Siberian blocking. Further analyses reveal that the blocking frequency is indeed significantly high over 90°–150°E from day −4 to day +2 relative to the onset of long-lived cold surge events. Furthermore, there exist significant correlations between the leading occurrence of Siberian blocking and the sea level pressure (SLP) anomalies over East Asia, which are directly related to long-lived cold surges. The intensification of the high SLP anomaly over East Asia is found to mainly result from cold advection induced by the anomalous northerly winds along the southeastern edge of the Siberian blocking.
A cold surge over the South China Sea (SCS) features an abrupt outbreak of northerly or northeasterly winds and is a typical but severe weather phenomenon during the boreal winter (Ramage 1971; Chang et al. 1983; Lau and Chang 1987; Chang et al. 2006, 2011). As a form of southward propagation of northerlies from the midlatitudes, cold surges can result in a drop of air temperature (e.g., Boyle and Chen 1987; Lu and Chang 2009), an intensification of tropical convection (e.g., Chang et al. 1979; Slingo 1998; Chang et al. 2005; Lim et al. 2017), and extreme rainfall and floods over Southeast Asia (e.g., Johnson and Houze 1987; Tangang et al. 2008; Pullen et al. 2015). In general, cold surges are characterized by a synoptic disturbance, which keeps its strength for 4 days on average (Chang et al. 2006). However, it is reported that some cold surges can last for a much longer time. For example, in February 2008, a record-breaking cold anomaly caused by cold surges lasted for almost one month over Southeast Asia and resulted in numerous casualties and losses in agriculture and fishery (Hong and Li 2009). Such long-persisting cold surges bring pronounced cold advection and heavy precipitation continuously, thus posing a severer threat to human society than a transient cold surge outbreak.
Previous studies have documented the general characteristics of cold surges and their linkages to midlatitude circulation disturbances. Typically, a cold surge is triggered by intensification and southward extension of the Siberian high (Ding 1990; Wu and Chan 1995; Zhang et al. 1997; Chan and Li 2004; Chang et al. 2006), as well as a deepening of the East Asian trough (Boyle 1986; Wu and Chan 1997; Lu and Chang 2009). In addition, our previous work showed the extratropical circulation patterns associated with the outbreak of cold surges and the crucial role of sea level pressure (SLP) anomalies in inducing cold surges over the SCS (Pang and Lu 2019). However, these findings were based on the composite analyses of all cold surge events, and thus mainly revealed the transient or synoptic features of short-lived cold surges, which appear overwhelmingly often in comparison with their long-lived counterparts. Necessary and specific evidence is still lacking for long-lived cold surges that exert more crucial influence and probably with quite different causes than short-lived ones.
Atmospheric blocking, which refers to a quasi-stationary anticyclone over the extratropics, is a typically long-lasting anomalous circulation and thus may be a candidate for persistent enhancement of the Siberian high and long-lived cold surges over the SCS. Actually, various previous studies have indicated that blockings play an important role in causing long-lasting extreme weathers over East Asia in winters (e.g., Lau and Li 1984; Lu and Chang 2009; Zhou et al. 2009; Cheung et al. 2012; Yao et al. 2017). For instance, persistent blockings can induce the amplification of the Siberian high (Takaya and Nakamura 2005) and result in a cold-air outbreak over East Asia to the east of blocking ridge (Joung and Hitchman 1982; Lau and Lau 1984; Ding 1990; Cheung et al. 2013a; Park et al. 2014; Yang et al. 2018). Therefore, it can be hypothesized that long-lived cold surges over the SCS, which are closely related to the Siberian high, may be affected by midlatitude blocking.
The reminder of the paper is organized as follows. Section 2 describes the data, definitions, and methods. Section 3 presents extratropical circulation anomalies associated with long-lived cold surge events. Section 4 investigates the features of blocking and its relationship with long-lived cold surges. Section 5 further diagnoses the mechanism that blocking influence long-lived cold surges. Conclusions and discussion are provided in section 6.
2. Data, definitions, and methods
Daily data from the European Centre for Medium-Range Weather Forecasts (ECMWF) interim reanalysis (ERA-Interim) are used for the period of 1979–2016 with a 0.75° × 0.75° horizontal resolution (Dee et al. 2011). This analysis focuses on the active season of cold surges from November to February (NDJF; Lim et al. 2017) and omits 29 February in leap years, in agreement with Pang and Lu (2019). Daily anomalies in this study are computed by deducting the climatology of that particular day from the raw data.
The definitions of cold surge are the same as in Pang and Lu (2019). The cold surge index is defined as meridional wind averaged over 110°–117.5°E along 15°N at 925 hPa (Chang et al. 2005). A cold surge day is detected when the index below 0.75 standard deviations from the winter mean (Lim et al. 2017). The results are not sensitive to the threshold of cold surge day. This definition yields a total of 1060 cold surge days, nearly one-quarter of the whole study periods. Furthermore, successive cold surge days are merged into one cold surge event. Thus, 344 cold surge events are obtained, and the average duration is about 3 days.
A two-dimensional blocking index is used to identify the blocking by calculating the daily meridional gradients of 500-hPa geopotential height Z from south (GHGS) to north (GHGN), following Scherrer et al. (2006) and Davini et al. (2012); this is an extension of the work of Tibaldi and Molteni (1990) and has been widely used in previous studies (e.g., Luo et al. 2015; Woollings et al. 2018; Li et al. 2019):
where λ0 and ϕ0 represent the longitude and latitude, which range from 0° to 360° and 30° to 75°N, respectively; ϕS = ϕ0 − 15°, ϕN = ϕ0 + 15°. For a given grid (λ0, ϕ0), a large-scale blocking is detected when GHGS > 0 and GHGN < −10 m (° lat)−1 are both satisfied for at least 15° continuous longitude. Also, blocking detected south of 40°N is excluded, as these low-latitude blockings are found to be linked to northward displacement of the subtropical high (Davini et al. 2012). For a specific longitude or region, a blocking day is defined when any grid within is detected as a large-scale blocking and the blocking frequency is counted as the percentage of blocking days.
Composite analyses of circulation anomalies are conducted to investigate the responsible temporal and spatial evolutions, where day 0 refers to the first occurrence day of each cold surge event. The two-tailed Student’s t test is used as the significance test for each variable at each grid point (Wilks 2006). The effective sample size N* at each grid point is evaluated by , where N is the original length of the time series and r1 refers to the lag-1 autocorrelation coefficient (Zwiers and von Storch 1995). In addition, the false discovery rate (FDR) was used in significance testing (Wilks 2016). The threshold for significance level is defined as , where pi is the ith smallest p value, which is evaluated at each grid point for a composite map, n is the total number of grid points, and αFDR is control level, which is set as 0.05.
We follow Pang and Lu (2019) to diagnose the variations in SLP associated with the long-lived cold surges:
where the subscripts 5 and 8 denote vertical pressure levels of 500 and 850 hPa, respectively; V and ∇ refer to the horizontal winds and gradient operator, respectively; p, ζ, and f are SLP, relative vorticity, and planetary vorticity, respectively; and h represents the thickness between 500 and 1000 hPa. The parameters a and b are estimated to be 2.6 × 105 hPa s−1 and 0.08 hPa m−1, respectively, and are calculated as follows:
where f0 (10−4 s−1), K0 (8 m hPa−1), and g (9.8 m s−2) are constants representing the feature scale of the Coriolis parameter, the hypsometric constant, and the acceleration of gravity, respectively; L and LR are the wavelength and Rossby radius of deformation, estimated as 5000 and 1000 km, respectively; and Δp3, Δp5, and p* refer to 300, 500, and 1000 hPa.
Equation (2) is a simplified version of the pressure tendency equation obtained from Carlson (1991), but replacing 1000-hPa horizontal winds to 850-hPa ones in the second terms on the right-hand side considering the topographic effect. It indicates that the increase (decrease) of SLP results from both negative (positive) vorticity advection at the middle troposphere and cold (warm) advection at the lower troposphere. The equation is used to diagnose the variations of SLP in the quasigeostrophic system. Temperature advection is evaluated over 500–1000 hPa as the vertical mean tends to be coincident with advection throughout the lower troposphere but to decrease in magnitude with height.
As mentioned in Pang and Lu (2019), by dividing each variable A into its winter mean and anomaly A′, the anomalous vorticity and temperature advections are further written as
where ζa refers to absolute vorticity.
3. Extratropical circulation anomalies associated with long-lived cold surge events
a. Comparisons between short- and long-lived cold surge events over the SCS
Figure 1 shows the occurrence of cold surge events with different durations. In general, the number of events is declining with the growth in duration. A large proportion of cold surge events last for a period shorter than 5 days, indicating the transient nature of many cold surges. However, there are numbers of events that persist for a long period, with the longest one lasting for 18 days (from 17 November to 5 December 1983). Cold surge events persisting longer than 5 days are classified as long-lived events and the results are not sensitive to the threshold values. We have repeated the following analyses by using other criteria, such as 4 or 6 days, and obtained similar results. Therefore, the conclusions shown in this paper do not depend on the threshold of long-lived events. In total, there are 42 long-lived and 302 short-lived cold surge events. As the short-lived events account for the majority (88%) of all events, the analysis results of the short-lived events is similar to the results of all cold surge events, presented in Pang and Lu (2019).
Figure 2 displays the evolutions of cold surge index relative to the onset of long- and short-lived events. On average, long-lived events can last for approximately 10 days, whereas the average short-lived ones last for less than 3 days. The cold surge index for long-lived events is able to maintain its strength below −10 m s−1 from day 0 to day +6, indicating the strong and successive northerlies over the SCS. Besides, the long- and short-lived events do not show an apparent difference in intensity before the outbreak of a cold surge. More specifically, the intensity of long-lived events (−10.4 m s−1) is comparable to that of short-lived ones (−10.2 m s−1) on day 0. Therefore, the initial strength of cold surge cannot explain the persistent duration of these long-lived events.
b. Evolutions of extratropical circulation anomalies associated with long-lived cold surges
Figure 3 demonstrates the spatial evolutions of SLP and 925-hPa horizontal wind anomalies for long-lived cold surge events. The northerly or northeasterly anomalies extend southward from Siberia to the Maritime Continent, accompanied by synchronous extension of high pressure anomalies. The SLP anomalies can be traced back to central Siberia on day −4 (Fig. 3a); they develop to the strongest intensity on day −2 (Fig. 3b) and then move southward to East Asia when cold surge outbreaks occur over the SCS (Fig. 3c). The above processes, recognized as an amplification and southward expansion of the Siberian high, are similar to the results for all cold surge events (e.g., Ding 1990; Wu and Chan 1995; Zhang et al. 1997). However, distinct evolutions of circulation anomalies occur after the onset of cold surge, in agreement with the variation of cold surge index (Fig. 2). Contrary to the rapid decaying of circulation anomalies for the short-lived events (figure not shown), the SLP anomalies for long-lived events maintain their strength over East Asia until day +6 (Fig. 3f). Consequently, the northerly or northeasterly anomalies are able to sustain continuously over the SCS. Therefore, the duration of cold surge is mainly attributed to the persistence of anomalous high pressure over East Asia. Besides, the successive SLP anomalies associated with long-lived cold surges are located farther southward compared to the ones related to cold-air outbreaks in the midlatitudes (e.g., Shoji et al. 2014; Abdillah et al. 2017).
Figure 4 shows the corresponding evolutions of 500-hPa geopotential height and wind anomalies, which are highly linked to the variations of SLP anomalies. The midtropospheric circulation features a persistent and widespread anticyclonic anomaly over northern Asia. The anomalous anticyclone emanates from western Siberia on day −4, accompanied with a weak cyclone to the south (Fig. 4a). Then the anticyclonic anomaly intensifies its strength and extends eastward to the Okhotsk Sea on day 0, while the cyclonic anomaly is found over East Asia (Fig. 4c). Then, the anticyclonic anomaly maintains its strength and extension over Siberia until day +6 (Fig. 4f), in agreement with the persistent high pressure anomalies over East Asia (Fig. 3). Combined with the evolution of SLP anomalies (Fig. 3), it is found that the anticyclone exhibits an equivalent barotropic structure over Siberia but a baroclinic structure downstream over East Asia, consistent with Takaya and Nakamura (2005), who suggested that the baroclinic structure over East Asia favors the growth of the surface Siberian high. The persistent and quasi-stationary anticyclonic anomaly in the mid- to high latitudes is reminiscent of the occurrence of blocking. Therefore, it is reasonable to hypothesize that the anomalous anticyclone may be attributed to midlatitude blockings, which may further influence SLP anomalies over East Asia.
4. Blocking occurrence associated with long-lived cold surge events
Extratropical circulation anomalies described in the preceding section suggest the possibility of midlatitude blocking associated with long-lived cold surge events. In this section, we investigate the occurrence of blocking and the relevant influence on circulation anomalies over East Asia. Figure 5 shows the longitudinal distribution of blocking frequency from 40° to 75°N. Climatologically, there are two peaks of blocking frequency over the Euro-Atlantic (~0°) and northern Pacific (~180°) regions, respectively, in agreement with previous studies (e.g., Barriopedro et al. 2006; Diao et al. 2006; Davini et al. 2012; Cheung et al. 2013b; Li et al. 2019). However, during the evolution of long-lived events, the east side peak moves westward from the northern Pacific (~180°) to Siberia (~120°E). More specifically, the blocking frequency is significantly high from day −4 to day +2 over 90°–150°E, where blockings seldom occur under normal circumstances (Figs. 5b–e). Here, the blocking frequency reaches its maximum around 120°E with the amplitude of 31.6% at day −2, which is more than 3 times than the climatological mean (Fig. 5c). In contrast, the blocking frequency for the short-lived events shows no obvious difference compared to the winter mean, suggesting that the short-lived cold surges may not be related to the Siberian blocking. Besides, the frequency over the Atlantic also increases at day −4 (Fig. 5b), consistent with a positive height anomaly at 500 hPa over the northern Atlantic from day −5 to day −1 (not shown in Fig. 4). Previous studies have found that the Atlantic blockings affect the development of Siberian high by triggering Rossby wave activity downstream (Lu and Chang 2009). Considering the relatively shorter duration of the Atlantic blocking, we focus on the major peak of Siberia blocking in the following.
Figure 6 further demonstrates the distribution of blocking frequency anomalies averaged from day −4 to day +2, by removing the climatological mean at each longitude or grid. The period from day −4 to day +2 corresponds to the strong blocking frequency anomalies over Siberia (Fig. 5). It is clear that a wide range of positive anomaly occurs over 90°–150°E around 60°N for the long-lived events. The geographic distribution of anomalous blocking frequency is identical to the anomalous anticyclone shown in Fig. 4. On the other hand, the anomalies for the short-lived ones are much weaker, with maximum amplitude of 4.0% around 150°E. Based on the results in Figs. 5 and 6, it can be inferred that the occurrence of Siberian blocking may be linked to the following circulations associated with long-lived cold surge. Hereafter, Siberian blockings in the following analysis refer to those occurring over 90°–150°E and from day −4 to day +2.
Figure 7 shows the lag composite of circulation anomalies for blocking days relevant to long-lived cold surge events. Here, 119 blocking days are detected as lag 0 when Siberian blocking is found from day −4 to day +2 relative to the onset of long-lived cold surges. The results reveal that the occurrence of Siberian blocking has a prominent impact on the following circulation anomalies over East Asia. In the midtroposphere, it is characterized by an omega-like height pattern with a northeastward extension of the ridge and deepening of the East Asian trough (Fig. 7a). Meanwhile, in the lower troposphere, a strong high pressure anomaly covers the majority of northern Asia, indicating the intensification of the Siberian high (Fig. 7b). These circulation anomalies can also be seen by compositing for all the blocking days in winter (figure not shown), suggesting their reliable relationship with SLP anomalies. Significant circulation anomalies are persistent after the occurrence of Siberia blockings in both the middle and lower troposphere (Figs. 7c–f). In particular, the high pressure anomaly moves southward from Siberia to East Asia, accompanied with a southward extension of northerly or northeasterly anomaly. Therefore, the occurrence of Siberia blocking has a lag impact on SLP anomaly over East Asia. Actually, previous studies have discussed the role of blockings on cold events over East Asia or their relevant circulations, such as the Siberian high (Takaya and Nakamura 2005; Lu and Chang 2009; Zhou et al. 2009; Yao et al. 2017). However, blockings here show a more eastward distribution than those responsible for cold events (i.e., ~60°E in their Fig. 4; Zhou et al. 2009). Therefore, the impact of blocking on long-lived cold surges over the SCS exerts unique features.
Figure 8 shows the scatterplot of the SLP index and blocking frequency for long-lived cold surge events with blocking occurrence. Here, the SLP index is defined as the average SLP anomalies over 15°–50°N, 100°–125°E, shown as a blue box in Fig. 7f. Among the total 42 long-lived cold surge events, there are 30 events accompanied with Siberia blocking (40°–75°N, 90°–150°E) from day −4 to day +2 relative to the onset of long-lived cold surges, shown as dots in the figure. First, a lag correlation between the blocking frequency and the SLP index is calculated because of the lag influence shown in Fig. 7. The result shows that coefficients are significant from day −2 to day +4 (figure not shown). Thus, the SLP index is averaged from day −2 to day +4. The result reveals a prominent positive relationship between the SLP index and blocking frequency (r = 0.52; p < 0.01) and indicates that the occurrence of Siberia blocking is responsible for the enhancement of SLP over East Asia, which further influences the duration of cold surge over the SCS. Besides, most events obtain higher blocking frequency than the climatological mean (28.6%), with an averaged frequency of 56.7% for these 30 long-lived events.
5. Possible mechanisms: Analysis of SLP tendency
The preceding section reported the occurrence of Siberian blocking and its close relationship with high pressure anomalies over East Asia associated with long-lived cold surge events. This section aims to investigate the mechanism by which the Siberian blocking influences the intensification of anomalous pressure by diagnosing the SLP tendency. To highlight the effect of Siberian blocking, 30 long-lived cold surge events accompanied with blocking occurrence are investigated in the following.
Figure 9 shows the evolutions of blocking frequency, the SLP index, and two advection terms of the SLP tendency in Eq. (2). The key area averaged by the advection terms is consistent with the SLP index (15°–50°N, 100°–125°E). On the one hand, the variation of the SLP index lags behind the blocking frequency (Fig. 9a), in agreement with the results in Fig. 7. More importantly, the SLP index starts to increase from day −4 and reaches its peak around day +2, when the blocking frequency is significantly high. Thus, to further investigate the influence of Siberian blocking on the intensification of SLP, the following diagnosis concentrates on the key period of SLP rising from day −4 to day +2.
On the other hand, Fig. 9b reveals that temperature advection makes an overwhelming contribution to the development of SLP over East Asia, whereas the vorticity advection is quite small and almost negative during the key period from day −4 to day +2. The distinct roles between two advection terms can be inferred from the spatial evolutions of extratropical circulation anomalies shown in Fig. 7, as the circulation anomalies in the lower troposphere are prominent over East Asia, whereas the ones in the midtroposphere are mainly located to the north of the key area. Thus, our results suggest that the Siberian blockings favor the intensification of SLP over East Asia through cold advection. Besides, the residual term, calculated by subtracting the above two advection terms from SLP tendency, is found to be much smaller than the temperature advection, suggesting the validity of diagnosis results.
Figure 10 shows the individual terms of both temperature and vorticity advections based on Eqs. (3) and (4) averaged over the key period from day −4 to day +2 within the same region in Fig. 9b. The sums of two advections confirm the leading role of temperature advection on SLP development. Among all the budget terms of temperature advection, is the main source, which accounts for 97.1% of the sum of temperature advection (Fig. 10a). The dominant contribution of can also be seen in Fig. 11, which shows the spatial distribution of the anomalous horizontal winds at 850 hPa and winter-mean thickness. The widespread northerly or northeasterly anomalies correspond to the large southward gradient in winter-mean thickness over the key area, and thus lead to extensive cold-air advection and resultant increase in SLP over East Asia. These anomalous northerlies or northeasterlies are associated with high pressure anomalies over Siberia, which result from the occurrence of blockings there (Fig. 7b). Besides, we also diagnosed the equation for the 12 long-lived cold surge events without Siberian blocking, and found a much weaker amplitude of SLP tendency averaged from day −4 to day +2 (0.3 × 10−5 hPa s−1) compared to the 30 events with blocking occurrence (1.0 × 10−5 hPa s−1). The anomalous northerlies, which are crucial for temperature advection and resultant SLP development, are limited to the south of 30°N for the 12 long-lived events without blocking (figure not shown), but extend into the mid- to high latitudes for the 30 events with blocking occurrence (Fig. 11).
For the vorticity advection, all the terms are much smaller than , and the negative value in the sum suggests its suppressed effect of SLP development (Fig. 10b). We also examined the spatial distribution of vorticity advection terms and found that the anomalies are relatively weak and concentrated on the northern part of the key area (figure not shown). Therefore, it can be concluded that vorticity advection in midtroposphere makes less contribution to the amplification of SLP over East Asia.
6. Summary and discussion
The large-scale extratropical circulation anomalies contributing to the persistence of long-lived cold surge events over the SCS are investigated in this study. The northerly winds over the SCS manage to maintain their strength for approximately 10 days on average for long-lived events, which is 3 times longer than for short-lived ones. The long-lasting northerly anomalies are found to result from successive circulation anomalies in the extratropics. In the lower troposphere, there is a persistent high pressure anomaly over East Asia, which is extended from an intensified Siberian high in midlatitude. Meanwhile, an extensive and quasi-stationary height anomaly is found over Siberia, accompanied with anticyclonic flows in the midtroposphere. Compared to the rapid movement of wavelike anomalies for the short-lived events (figure not shown), the anomalous circulations for long-lived events are more expansive and stable.
The persistent anticyclone over Siberia implies more occurrence of Siberian blocking. Therefore, further analysis is conducted on the midlatitude blockings associated with long-lived cold surges. The results show that blockings are more frequent over 90°–150°E from day −4 to day +2 relative to the onset of long-lived cold surge events. Among the total 42 long-lived events, there are 30 events accompanying the occurrence of blocking over there, with a much higher blocking frequency on average than a climatological mean. Under normal circumstances, the blocking frequency over Siberia (i.e., 90°–150°E) is much lower than over the Urals or the western Pacific (Cheung et al. 2013b), and thus Siberian blocking has received little concern in previous works. However, this work implies that Siberian blockings have important effects on long-lived cold surges over the SCS and more attention should be paid to them in future works.
Further diagnosis of SLP tendency is examined to analyze the physical mechanism whereby the Siberia blockings affect long-lived cold surge events. First, circulation anomalies composited by blocking days show a lag influence on SLP anomaly over East Asia. Accompanied with a successive anticyclone anomaly over Siberia in the midtroposphere, anomalous high pressure extends from Siberia to East Asia. In addition, SLP anomalies over East Asia increase persistently from day −4 to day +2 when the blocking frequency is prominently high, indicating a crucial role of blockings in the enhancement of SLP. Further analysis shows that the amplification of SLP results from cold advection in the lower troposphere, which is related to persistent northerly anomalies triggered by a blocking high and a strong southward gradient of the winter-mean temperature over East Asia. Besides, the effect of vorticity advection is negligible as the midtropospheric circulation anomalies are quite weak over East Asia.
This work was supported by the National Natural Science Foundation of China (Grant 41721004).