Extended Impact of Cold Air Invasions in East Asia in Response to a Warm South China Sea and Philippine Sea

Cold surges are a prominent phenomenon in East Asia during boreal winter. Cold air originating from Siberia periodically migrates southeastward, leading to profound drops in temperature over East Asia on synoptic timescales. The invasion of cold surges can be associated with significant socioeconomic impacts and mortality (Zhou et al., 2011; Sun et al., 2022). Hence, previous studies have investigated the influence of cold events in East Asia and their possible causes (Zhou et al., 2009; Cheung et al., 2016; Zhang et al., 2021b; Bueh et al., 2022; Zhang et al., 2022).

Temperature variations on synoptic timescales in East Asia are closely related to horizontal temperature advection in the lower troposphere in conjunction with the passage of extra-tropical cyclones, which is accompanied by altered wind direction and temperature advection on synoptic tempo-spatial scales (Leung et al., 2015; Leung and Zhou 2016a; Song et al., 2016; Ma and Zhu 2021). Extra-tropical cyclone genesis often occurs on the leeward side of mountains in Mongolia and the Tibetan Plateau, whose evolution is aided by the strong surface temperature gradient over the Kuroshio Current (Cho et al., 2018; Lee et al., 2020). In the developing stage, the intensification of extra-tropical cyclones is modulated by baroclinic instability, diabatic processes, and the efficiency of eddy growth, which contribute to seasonal variations in the intensity of extra-tropical cyclones (Chang and Song, 2006; Leung and Zhou, 2016b; Schemm and Rivière, 2019; Liu et al., 2020; Yang et al., 2022). Therefore, extra-tropical cyclones are modulated by low-frequency variations in atmospheric and oceanic circulations.

The El Niño/Southern Oscillation (ENSO) can induce anomalous atmospheric circulation in East Asia through air-sea interactions during boreal winter (Wang et al., 2000). Recent studies have demonstrated ENSO forcing on the intensity of synoptic temperature variation in eastern China (Leung and Zhou, 2016b; Jian et al., 2021a). However, it has also been pointed out that ENSO forcing on synoptic-scale temperatures in China is substantially weaker after the 1980s, in association with a change in the teleconnection pattern of ENSO after the 1980s (Jian et al., 2021b). Consequently, ENSO-related temperature patterns and baroclinicity in East Asia are also altered. Additionally, interannual variability in sea surface temperature over the Kuroshio Current could influence the seasonal prediction of the life cycle of baroclinic wave activity in the Northwest Pacific (Zhang et al., 2021a). Therefore, a multi-scale change in the upper ocean could be crucial to synoptic temperature variations in East Asia. In addition, the effects of local oceanic forcing on the variation in East Asia remain uncertain.

Anthropogenic forcing is likely to drive ocean warming by changing the radiative forcing. Observational data indicate a considerable spatial difference in the rate of ocean warming (Wang et al., 2016). For instance, the tropical western Pacific warms substantially faster than the eastern Pacific because of strengthening trade winds and the cooling effect of upwelling in the eastern Pacific (Amaya et al., 2015; Seager et al., 2019). It is also noted that the warming rate in the offshore China seas accelerated after 2011, along with the Interdecadal Oscillation (Tang et al., 2020). Warming of the offshore China seas and the western Pacific may enhance baroclinicity in the lower troposphere, which may, in turn, affect extra-tropical cyclones in East Asia.

The frequency of cold events is expected to decrease with global warming, but extreme cold events may still occur as a result of natural variability in the climate system (Qian et al., 2018; Hu et al., 2020). Apart from extreme temperatures, recent studies have shown that the magnitude of temperature fluctuations could influence public health and agriculture (Ikram et al., 2015; Xu et al., 2020). Thus, there is a pressing need to investigate the mechanism controlling the intensity of strong synoptic cold days. This study aims to identify the trend in the intensity of strong synoptic cold days in East Asia during boreal winter and the degree to which ocean warming contributes to this trend.

The remainder of this paper is organized as follows. Section 2 presents the data and methods, section 3 presents the results. Section 4 provides a discussion before a conclusion is presented in section 5.

This study employs the 2-meter air temperature (T2m) and 10-meter meridional wind (V10m) of the ERA-5 reanalysis (Hersbach et al., 2020). Daily values are obtained by averaging the hourly temperatures. For the sea surface temperature (SST), ERSSTv5 data are utilized (Huang et al., 2017). To identify a strong cold air invasion, a cold surge is traditionally defined as a temperature drop exceeding a specific threshold. In this study, we quantify the intensity of strong cold synoptic days using the average of the lowest 10% synoptic T2m (L10ST) for every winter. A Lanczos filter is first applied to T2m to extract the synoptic signal (≤ 14 days) (Duchon, 1979). The filter is also applied to the synoptic V10m. The study period includes winters from 1979/80 to 2020/21 (December to February). Winters are labeled according to the year of their December; for example, the 1979 winter denotes the 1979/80 winter.

To investigate the variation in L10ST along with oceanic forcing, we use a simplified atmospheric general circulation model, SPEEDY, from the International Centre for Theoretical Physics (Molteni, 2003; Kucharski et al., 2006, 2013). This model is widely used to study the atmospheric response to anomalous sea surface temperature (Jian et al., 2020; Leung et al., 2020, 2022a, b; Cheung et al., 2021; Feng et al., 2022). It is a hydrostatic model with a semi-implicit treatment of gravity waves, eight vertical levels in σ-coordinates, and T30 spectral truncation resolution. The model is driven first by the climatological mean sea surface temperature. Subsequently, anomalously warm and cold sea surface temperature is prescribed to drive the sensitivity simulations. To mitigate the influence of specific initial conditions on the simulations, a composite of 135 different initial conditions is computed and investigated. Additionally, the output of the simulations is examined by a paired z-test.

The climatological mean of winter L10ST over East Asia is shown in Fig. 1a. It is noted that L10ST is relatively stronger in the midlatitudes of East Asia, except for a center of local maximum in Southeast China (SEC; 25°–35°N and 110°–120°E). For the linear trend in L10ST (Fig. 1b), a significant decreasing trend is found in the East China Sea (ECS; 25°–35°N and 125°–135°E) and the Indochina Peninsula (ICP; 15°–25°N and 95°–110°E). The negative trend in the East China Sea and the Indochina Peninsula indicate stronger synoptic cold days in the two regions. This also implies that the center of the local L10ST minimum in Southeast China extends northeastward and southwestward, possibly subjecting this area to the influence of cold invasion.

Figure 1.  (a) Climatological mean L10ST (units: K). (b) Linear trend in L10ST, in units of K (10 yr^–1); the yellow contour lines indicate p-values = 0.05. (c) Dashed lines show the time series of the area-averaged L10ST over the SEC, ECS, and ICP; solid lines show their corresponding linear trend.

The time series of areal averaged L10ST in the three regions are presented in Fig. 1c. The decreasing L10ST trend in Southeast China is approximately equivalent to that in the East China Sea and Indochina Peninsula. However, the trend in Southeast China cannot pass the significance test (Fig. 1c) because of the stronger interannual variation of L10ST in Southeast China relative to the East China Sea and the Indochina Peninsula. This results in a relatively lower signal-to-noise ratio for the decreasing trend of L10ST in Southeast China.

To investigate the possible cause of decreasing L10ST, a negated time series of L10ST in the East China Sea and Indochina Peninsula is utilized in the regression reanalysis. As shown in Figs. 2a and b, L10ST in the East China Sea and Indochina Peninsula is negatively correlated to the lowest 10% synoptic meridional wind intensity. This indicates that stronger cold synoptic days in these two regions occur along with stronger northerly winds on synoptic timescales. In Fig. 2c, a positive relationship is found between L10ST in the East China Sea, and the winter mean T2m over eastern China and the South China Sea and Philippine Sea (SCSPS; 5°S–30°N and 110°–145°E). For the regression of L10ST in the Indochina Peninsula (Fig. 2d), a positive center is found in the SCSPS. Hence, the warmer near-surface air temperature over the SCSPS is concurrent with the stronger synoptic cold days in the two regions, which results in a significantly steeper meridional gradient upstream of the East China Sea (Fig. 2e) and a southward shift of the stronger baroclinic zone in the Indochina Peninsula (Fig. 2f). Thus, the warmer near-surface temperature in the SCSPS may enhance the synoptic meridional wind and synoptic cold days in the two regions by altering regional baroclinicity.

Figure 2.  Regression of the intensity of synoptic meridional wind onto standardized and negated (a) ECS L10ST and (b) ICP L10ST. Panels (c) and (d), as in (a) and (b), but for near-surface temperature. Panels (e) and (f), as in (a) and (b), but for the meridional temperature gradient (shading). In (e) and (f), dashed and solid contours indicate the seasonal mean of the meridional temperature gradient with values of –1 × 10^–5 and –2 × 10^–5 K m^–1, respectively. For shading, only values exceeding the 0.05 significance level are depicted. Purple and green boxes indicate the regions of ECS and ICP, respectively.

In this study, as portrayed in Figs. 3a and b, we also examine the regression of sea surface temperature onto negated L10ST in the East China Sea and Indochina Peninsula. There is a significant positive surface temperature anomaly over the SCSPS, which may warm the air above. The area-averaged winter sea surface temperature in the SCSPS is depicted in Fig. 3c. A notable warming trend occurred after the mid-1990s. As mentioned above, a positive relationship exists between the SCSPS sea surface temperature and the intensity of strong synoptic cold days in the East China Sea and the Indochina Peninsula. Hence, the positive trend in the SCSPS sea surface temperature supports the trends in L10ST in the two regions (Figs. 3c and d). To deduce the oceanic forcing of the SCSPS on synoptic cold days, we define warm and cold SCSPS winters as the standardized sea surface temperature in the SCSPS that are > 1 and < –1, respectively. Accordingly, ten warm and cold SCSPS winters are identified, as shown in Fig. 3c.

Figure 3.  Regression of SST onto the standardized and negated (a) ECS L10ST and (b) ICP L10ST. (c) Average SST in SCSPS (black; units: K); purple dashed lines indicate the range from ±1 standard deviation; red rectangles and blue triangles represent winters with SST larger than 1 standard deviation and less than −1 standard deviation, respectively. (d) Linear trend in SST. In (a), (b), and (d), purple contours indicate p-values = 0.05.

The L10ST days in Southeast China for positive and negative SCSPS winters are identified. Accordingly, the composite of synoptic T2m and V10m on days –1, 1, and 2 is presented in Fig. 4. For positive SCSPS winters, a cold center is located in northeast Asia on day –1 (Fig. 4a). The negative temperature center migrates southward and extends southwestward and northeastward in the following days in conjunction with the migration of the synoptic meridional wind (Figs. 4b and c). This illustrates the passage of cold air originating from Siberia, which results in near-surface temperature fluctuation on a daily timescale in Southeast China, the East China Sea, and the Indochina Peninsula. For negative SCSPS winters, a similar migration of a negative temperature center from Siberia is observed (Figs. 4d and f). However, notably weaker magnitudes are noted in the East China Sea and Indochina Peninsula when the cold air migrates across East Asia. Therefore, the impact of the cold air invasion widens with warmer sea surface temperatures in the SCSPS.

Figure 4.  (a–c) Composite of T2m (shading; units: K) and V10m at the levels of –1.5 and 1.5 (contours; units: m s^–1) on days –1, 1, and 2, corresponding to L10ST days in South China, for positive SCSPS winters. Panels (d–f), as in (a–c), but for negative SCSPS winters. The gray box indicates the region of SEC.

The probability distribution function of synoptic temperature in the East China Sea and Indochina Peninsula is presented in Figs. 5a and b. In assessing the difference between warm and cold SCSPS conditions in the two regions, we find that the number of strong synoptic cold days is significantly larger with warm SCSPS, which is associated with a reduction in weak cold days. In addition, strong synoptic warm days in the East China Sea and Indochina Peninsula also increase along with warm SCSPS. Hence, the number of strong synoptic cold days increases along with greater synoptic temperature variation during warm SCSPS events, possibly due to the stronger baroclinicity and synoptic meridional wind in South China.

Figure 5.  Probability distribution function of synoptic temperature in a warm SCSPS (dashed red line) and a cold SCSPS (dashed blue line) in the (a) ECS and (b) ICP, with the difference between warm and cold SCSPS (green bars); black stars indicate differences exceeding the 0.05 significance level. Solid red and blue vertical lines indicate the average of the first decile of synoptic temperature for warm and cold SCSPS, respectively.

To verify the forcing of sea surface temperature in the SCSPS on the size of the cold air invasion in East Asia, two numerical simulations are carried out using the ICTP AGCM. By considering a similar regression pattern in Figs. 3a and b, the two simulations are driven by the regression pattern of sea surface temperature in SCSPS (Pos_Run; black box in Fig. 3b) and its negated pattern (Neg_Run), respectively.

The difference in near-surface L10ST between the two simulations (Pos_Run — Neg_Run) is presented in Fig. 6a. Notably, significant negative differences are found in the East China Sea and the Indochina Peninsula. This demonstrates that warmer sea surface temperatures can induce significantly stronger synoptic cold days in the SCSPS. In addition, warmer sea surface temperature also results in colder temperatures in South China and a negative difference in the meridional temperature gradient (a steeper gradient) extending from the East China Sea to the Indochina Peninsula (Figs. 6b and c). Therefore, the results of the numerical simulation prove that warm sea surface temperatures in the SCSPS heat the near-surface air, leading to a stronger land-sea thermal contrast and baroclinicity. This ultimately induces a change in the synoptic meridional wind intensity and an extension of the cold air invasion into East Asia (Fig. 6a).

Figure 6.  (a) Difference in L10ST between Pos_Run and Neg_Run simulations; Panels (b) and (c), as in (a), but for near-surface temperature and its meridional gradient; yellow contours indicate a z-score = 1.96.

The variation in the intensity of strong synoptic cold days in East Asia is documented in this study. Significant decreasing temperature trends in the East China Sea and Indochina Peninsula imply larger areas of cold air invasion in East Asia. These trends are attributed to sea surface warming over the South China Sea and Philippine Sea. This warming heats the air near the surface, which enhances the land-sea thermal contrast between East Asia and the western Pacific, which leads to a steeper meridional temperature gradient near the surface, extending from the East China Sea to the Indochina Peninsula. This induces changes in the synoptic meridional wind intensity and L10ST. The forcing by a warm SCSPS on L10ST in the East China Sea and Indochina Peninsula is generally reproduced by a simplified atmospheric general circulation model. Apart from the change in synoptic meridional wind intensity, the stronger near-surface temperature gradient in Southeast Asia, on a seasonal timescale concurrent with a warm SCSPS, could enhance horizontal temperature advection and the subsequent invasion of cold air.

Previous studies show that the influence of cold surges in East Asia is determined by their intensity, pathway, and frequency. As indicated by Leung et al. (2019), the intensity of synoptic temperature variations in the Indochina Peninsula is modulated by the strength of extra-tropical eddies, along with variation in baroclinic energy conversion from eddy potential energy to eddy kinetic energy. Thus, this study points out the importance of oceanic forcing in lower tropospheric baroclinicity and the size of cold air invasion in East Asia. In addition, this study suggests that the synergy between the intensity, frequency, and size of cold air invasions should be considered when evaluating their socioeconomic impacts in association with climate change. On the other hand, extra-tropical cyclones could reduce the warm sea surface temperature in SCSPS via sensible and latent heat fluxes, which play a negative feedback role in anomalous temperature in SCSPS (Abdillah et al., 2017; Dacre et al., 2020). Therefore, the intensity of strong synoptic cold days is determined, in part, by the ocean heat content in the upper SCSPS.

Based on numerical simulations, the forcing of the SCSPS sea surface temperature on the intensity of strong synoptic cold days in East Asia is examined. The results suggest that a warm SCSPS can induce stronger-than-normal synoptic cold days in the East China Sea and Indochina Peninsula. In addition, the SCSPS sea surface temperature demonstrates a robust, increasing trend. Previous studies have attributed the interdecadal variation and trend in the SCSPS sea surface temperature to changes in ocean advection, which are driven by a weakening East Asian winter monsoon and a strengthening western North Pacific subtropical high (Wang et al., 2002; Qu et al., 2005; Cai et al., 2017; Tan et al., 2021; Liang et al., 2022). Hence, these variations in atmospheric and oceanic circulations could influence the intensity of strong synoptic cold days in East Asia by altering the sea surface temperature in the SCSPS.

Acknowledgements. This study was jointly supported by the National Natural Science Foundation of China (Grant Nos. 42120104001, 41805042), the Science and Technology Program of Guangzhou, China (Grant No. 202102020939), and the Fundamental Research Funds for the Central University, Sun Yat-Sen University (Grant No. 22qntd2202), and a project of the Center for Ocean Research in Hong Kong and Macau (CORE).