2023 年 101 巻 1 号 p. 21-37
The turnabout of air temperature anomalies over East Asia between the first and second halves of winter 2020/21 was examined from a teleconnection perspective with regionally different convective heating anomalies over the Indo-western Pacific sector. In the first half of winter 2020/21, the air temperature over East Asia was lower than normal, accompanied by a pair of anticyclonic and cyclonic anomalies in the upper troposphere southeast of the Tibetan Plateau and north of Japan, respectively. This dipole pattern is newly referred to as the Southeast Asia–Japan (SAJ) pattern in this study, indicating the propagation of Rossby waves caused by enhanced tropical convection over the eastern Indian Ocean toward the South China Sea. In the second half of winter 2020/21, the enhanced convection shifted eastward to the Philippine Sea. The subsequent anticyclonic anomaly changed its position to the south of Japan, which was similar to the western Pacific (WP)-like teleconnection pattern, causing warmer conditions over East Asia. The composite analysis indicated that the anomalous anticyclone over the southeastern Tibetan Plateau corresponding to the SAJ pattern emerged simultaneously with intensification of convection over the South China Sea. Half of the cases of the WP-like pattern have been accompanied by enhanced convection over the Philippine Sea. The different circulation patterns were reproduced by prescribing the heat source over the South China Sea and Philippine Sea to the linear baroclinic model. Moreover, the vorticity budget analysis suggested that the presence of upper-tropospheric convergence of winds to the southeast of the Tibetan Plateau seen in the climatology is conceivable for the in situ localized anomalous circulation constituting the SAJ pattern due to vortex stretching effects.
The western Pacific (WP) pattern is a tropospheric teleconnection pattern in boreal winter, which consists of a north–south dipole of height anomalies over the Kamchatka Peninsula and the western subtropical North Pacific (Wallace and Gutzler 1981). The WP pattern influences climate variability over East Asia (Takaya and Nakamura 2013; Park and Ahn 2016; Shiozaki et al. 2021; Shiozaki and Enomoto 2021) and the frequency of blocking associated with Rossby wave breaking (Pavan et al. 2000; Rivière 2010). The monthly or seasonal geopotential height anomaly corresponding to the WP pattern was investigated in association with the El Niño–Southern Oscillation (ENSO). Observational evidence and numerical studies have shown that a positive (negative) WP pattern tends to appear more frequently when El Niño (La Niña) occurs (Horel and Wallace 1981; Ferranti et al. 1994; Trenberth et al. 1998; Koide and Kodera 1999; Dai and Tan 2019). However, a positive WP pattern has been observed not only in El Niño months but also in La Niña months (Tanaka et al. 2016). In this study, the positive WP pattern is defined as when the southern part of the dipole is high and the northern part is low. It has also been indicated that the emergence of the WP pattern, linked to ENSO, may be affected by anomalous circulation over the Eurasian continent during early winter (Kodera 1998). Tanaka et al. (2016) showed that the WP pattern could be sustained by baroclinic energy conversion associated with heat transport by energetic analysis based on monthly mean data and concluded that the WP pattern could be maintained even without external forcing.
As for the interannual variability of the East Asian winter monsoon, especially in air temperature over the Far East, a more effective north–south dipole pattern could shift slightly westward compared with that of the canonical WP pattern (Takaya and Nakamura 2013). Takaya and Nakamura (2013) referred to the upper-tropospheric pattern as the “WP-like” pattern. These facts indicate that WP-like dipole patterns can appear at slightly different positions from the action center shown by Wallace and Gutzler (1981), and its locality is an important factor influencing climate variability over East Asia.
The variability of the East Asian winter monsoon is closely related to the modulation of tropical convective activity over the Indo-western Pacific sector (Hong and Li 2009; Sakai and Kawamura 2009; Ueda et al. 2015; Abdillah et al. 2017; Kuramochi et al. 2021). They showed a southwest-northeast dipole pattern (or wave train) of circulation anomalies from southern China to Japan in the upper troposphere caused by anomalous diabatic heating around the Maritime Continent, which is a dominant spatial mode over Asia during the winter season (Zheng et al. 2013). For convenience, hereafter, we refer to the southwest–northeast dipole pattern as the “Southeast Asia–Japan (SAJ) pattern.” The SAJ pattern, especially in the southwestern portion of the dipole, is salient in the upper troposphere (∼ 300–200 hPa); however, it becomes obscure in the mid-troposphere (∼ 500 hPa) (Ueda et al. 2015; Abdillah et al. 2017). This vertical structure appears to be consistent with the baroclinic mode of the heat-induced Rossby wave response (Gill 1980; Wu et al. 2015a, b) and differs from the characteristics of the WP pattern.
The winter climate variability over East Asia is affected by both WP(-like) and SAJ patterns. The spatial distributions of the two types of dipole patterns differ, indicating dynamically different excitation mechanisms. Our main objective is to assess the structure and dynamics of the two teleconnection patterns.
In this study, we first focus on a change in air temperature from cold to warm over East Asia during the winter season in 2020/21, linked with the modulation of atmospheric circulation fields (Japan Meteorological Agency 2021a, b). Figure 1a shows the time series of the air temperature anomaly at 850 hPa averaged over East Asia [25–45°N, 100–140°E] from December 1, 2020, to February 28, 2021. The air temperature anomaly changed from negative to positive around January 14, 2021. In the current study, we defined two periods: the colder period (Period 1: from December 13, 2020, to January 11, 2021) and the warmer period (Period 2: from January 23, 2021, to February 21, 2021). The anomalous coldness in Period 1 was associated with the SAJ pattern, whereas the warmth in Period 2 was the WP-like pattern, as discussed below in Section 3. The present study examines the difference between these two periods and the factor of air temperature transition from the teleconnection perspective; moreover, the inherent physical mechanism involved in the SAJ and WP(-like) patterns anchored with anomalous convection over the tropical western Pacific is investigated by comparing their differences.
The data used in this study were monthly mean and 6-hourly data from the Japanese 55-year Reanalysis dataset (JRA-55) with a horizontal resolution of 1.25° and 37 pressure levels (Kobayashi et al. 2015), daily data from the National Oceanic and Atmospheric Administration (NOAA) Interpolated Outgoing Longwave Radiation (OLR) dataset with a horizontal resolution of 2.5° (Liebmann and Smith 1996), and monthly mean sea surface temperature (SST) data from the Centennial in Observation-Based Estimate-SST (COBESST; Ishii et al. 2005) for December–February (DJF) during the 47 years from 1974/75 to 2020/21. To examine daily oceanic conditions, the NOAA Daily Optimum Interpolation SST (OISST; Reynolds et al. 2007) from 1981 to 2021 was also utilized. To abstract high-frequency components, an appropriate running mean was applied to the 6-hourly and daily data. In this study, climatological mean states were defined as the 47-year average during the study period (41-year average only for the OISST), and anomalies are defined as deviations from the climatological mean states.
The propagation of stationary Rossby wave packets was analyzed using the quasi-geostrophic wave-activity flux defined by Takaya and Nakamura (2001). The horizontal component of the flux W is defined as follows:
 |
To assess the role of tropical convection in the excitation and maintenance of circulation patterns, vorticity budget analysis was conducted. The linearized barotropic vorticity equation is written as follows (Sardeshmukh and Hoskins 1988):
 |
 |
To verify the linear response to tropical diabatic heating, we used the linear baroclinic model (LBM; Watanabe and Kimoto 2000). The LBM is a spectral model based on linearized primitive equations and has 20 sigma levels with a horizontal resolution of T42. The model employed del-forth horizontal diffusion, Rayleigh friction, and Newtonian thermal damping. The e-folding decay time of the diffusion coefficient was set to 4 h. The damping coefficient was set at two days for the lowest three levels, one day for the uppermost two levels, and 30 days elsewhere. The LBM was forced with externally imposed heating and time-integrated to a steady state. The details of the experimental settings, including the prescribed heating and background states, are mentioned in Section 4.3.
The first half of the winter of 2020/21 was colder than normal over East Asia. In contrast, the latter half became warmer (Fig. 1a). Figures 1b and 1c show the circulation and temperature anomalies in the lower troposphere during the two defined periods. In Period 1, East Asia and Siberia were colder than the climatological state, and the anomalous northwesterly flow was indicative of intensification of the East Asian winter monsoon, together with the deepened northern part of the Aleutian low. However, in Period 2, the cold air remained alive over northern Eurasia; nevertheless, the air temperature over East Asia became warmer than normal, which is associated with an anticyclonic circulation anomaly extending to Japan from the North Pacific. The remote influence of tropical modulations is suggested explaining the variation in air temperature anomalies.
(a) Time series of 5-day running mean air temperature anomaly at 850 hPa over East Asia [25–45°N, 100–140°E] (red and blue shading) and SST indices (blue and orange lines) (unit: °C). The indices are defined as SST anomalies averaged as follows: Niño-3.4 as [5°S–5°N, 170°E–120°W] and Niño-WP as [0–10°N, 130–150°E]. (b) Anomalies of streamfunction (contours; unit: 106 m2 s−1), air temperature (shading; unit: °C), and horizontal wind (vectors; unit: m s−1) at 850 hPa averaged in Period 1 (December 13, 2020–January 11, 2021). (c) Same as (b), but for Period 2 (January 23, 2021–February 21, 2021). Purple rectangle indicates the East Asia region in this study.
One noteworthy feature of this winter was the continued La Niña condition, which can be seen from the negative values of the Niño-3.4 SST anomaly (Fig. 1a). Niño-3.4, a widely used index of ENSO, is defined as the averaged SST anomaly over the equatorial eastern Pacific [5°S–5°N, 170–120°W] (Trenberth 1997). The positive values of the SST anomaly in the Niño-WP [0–10°N, 130–150°E] (Hoell and Funk 2013) also exhibit La Niña conditions (Fig. 1a), which set up an easy-to-enhance environment for convection over the tropical western Pacific owing to the in situ anomalous warm SST. In a La Niña winter, the East Asian winter monsoon tends to be stronger than normal (Wang et al. 2000; Zhou and Wu 2010), although the air temperature over East Asia was warmer in the latter half of winter 2020/21. This inconsistency suggests that intraseasonal variability contributed to the dramatic turnabout of temperature anomalies, rather than prolonged SST anomalies.
The anomalies of the upper-tropospheric circulation and OLR, together with the wave-activity flux in the two periods, are shown in Figs. 2a and 2b. The negative values of the OLR anomaly in the tropics correspond to the enhancement of tropical convection. In Period 1, anticyclonic and cyclonic circulation anomalies emerged over the region from the southeast of the Tibetan Plateau to southern China and over the northern part of Japan, respectively, exhibiting a southwest–northeast dipole pattern. The northeastward wave-activity flux emanating from the dipole's southwestern anomaly indicates the propagation of stationary Rossby waves. In addition, southeastward propagation from the Caspian Sea to the northern part of India was confirmed. Concurrently, in the tropics, enhanced convection was discernible over the South China Sea toward the eastern Indian Ocean. These modulated atmospheric fields are responsible for the colder-than-normal conditions and heavy snowfall in Japan during winter 2020/21 (Japan Meteorological Agency 2021a) and are almost consistent with the features of the SAJ pattern as well as the typical mode of the cold East Asian winter (Hong and Li 2009; Sakai and Kawamura 2009; Ueda et al. 2015; Abdillah et al. 2017; Shiozaki and Enomoto 2021). Corresponding to the enhanced equatorial convection, anomalous divergent flows occurred in the upper troposphere over the Philippines and southern India (Fig. 2c).
(a) Averaged anomalies of streamfunction (contours with an interval of 3.0 × 106 m2 s−1) at 250 hPa and OLR (shading; unit: W m−2) in Period 1. Gray vectors indicate the wave-activity flux (unit: m2 s−2) at 250 hPa. (b) Same as (a), but for Period 2. (c) Averaged anomalies of velocity potential (contours; unit: 106 m2 s−1) and divergent wind (vectors; unit: m s−1) at 200 hPa in Period 1. (d) Same as (c), but for Period 2. Scale for arrows is given on the lower-right side.
In Period 2, the location of the anticyclonic anomaly around 30°N shifted eastward, causing anomalously warm conditions to associate with attenuated cold air outbreaks (Fig. 1). The meridional dipole pattern across the Japanese islands resembles the WP pattern (Fig. 2b); however, this pattern is referred to as the “WP-like” pattern in the present study because the southern anomaly of the dipole is located westward compared to the action center of the canonical WP pattern (160°E). Notably, the northern cyclonic anomaly was not as clear as the southern anticyclonic anomaly. We could not find any noticeable wave packet propagation from upstream, which seems consistent with the WP-like blocking analyzed by Takaya and Nakamura (2005). The center of the enhanced convection was located from the Philippine Sea to Indonesia, and convective activity over the South China Sea toward the eastern Indian Ocean had already decayed. This zonal shift of the anomalous convection center on the intraseasonal timescale, such as the Madden–Julian oscillation (MJO: Madden and Julian 1972), can also be confirmed from the time- longitude cross-section (Fig. 3). The negative OLR anomaly persisted around 80–120°E from mid-December 2020 until January 2021 (Period 1), and around 120–150°E in February 2021 (Period 2). The MJO phase transition induces different uppertropospheric circulations through modulated divergent fields (Cassou 2008; Seo and Son 2012). Thus, the anomalous divergence and its winds (Figs. 2c, d) likely caused the different positions of the subtropical anticyclonic anomaly of the dipole patterns between the two periods, which will be discussed in Section 3.2.
Time–longitude cross-sections of 7-day running mean OLR anomaly averaged over the tropics 10°S–10°N in the Indo–Pacific region during the winter of 2020/21. The unit is W m−2.
Figure 4 shows the climatological mean state of the upper troposphere over the western North Pacific region. The strong westerly jet and northward positive meridional gradient of absolute vorticity are the most important features in East Asia during winter (Fig. 4a). Moreover, the southern Tibetan Plateau in the upper troposphere is a very intense convergence area of winds associated with deep convection over the Indo-western Pacific sector (Fig. 4b). These unique background states play essential roles in anomalous vorticity generation in vorticity budget analysis.
DJF climatological mean (a) absolute vorticity (shading; unit: 10−5 s−1), zonal wind (white contours; unit: m s−1), (b) divergence of wind (shading; unit: 10−6 s−1), and divergent wind (vectors; unit: m s−1) at 250 hPa. Scale for arrows is given on the lower-right side.
The upper-tropospheric vorticity budget was determined based on the steady state of Eq. (2), that is,
 |
(a) RWS and (b) rotational terms in Eq. (2) based on time-averaged flow at 250 hPa in Period 1 (shading). (c)–(f) Decomposition of RWS into (c) anomalous divergence term (S1; −̄ζD′), (d) anomalous advection term (S2; −v′χ · ∇̄ζ), (e) climatological divergence term (S3; −ζ′̄D), and (f) climatological advection term (S4; −̄vχ · ∇ζ′) in Eq. (3). Intervals of shading are 0.4 × 10−10 s−2. Black contours indicate the vorticity anomaly averaged in the period (unit: 10−5 s−1).
Same as Fig. 5, but for Period 2.
The climatological divergence term (S3; Fig. 5e) is found over the center of the anticyclonic vorticity anomaly, which is related to vortex stretching due to the intense background convergence over the southern Tibetan Plateau (Fig. 4b) acting on the anomalous vorticity. S3 seems to counterbalance S1 over the center of the anomalous vorticity rather than their nodes, canceling out the mutual stretching effect (Fig. 5c) because S3 cannot balance with the rotational terms. The contribution of S3 may explain the localized anomalous circulation over the southeastern Tibetan Plateau toward southern China when considering a steady state. Indeed, the climatological convergence region can produce a vorticity source on the perturbation vorticity (Sardeshmukh and Hoskins 1988; Trenberth et al. 1998), and its importance has been confirmed by numerical studies (Grimm and Silva 1995a, b). The climatological advection term (S4; Fig. 5f) dominated the southeastern part of the RWS, particularly the positive values around the anomalous anticyclone (Fig. 5a). This term acts to advect the circulation anomaly poleward.
During Period 2 (Fig. 6), the negative vorticity anomaly corresponding to the southern part of the WP-like pattern was located south of Japan. Again, we focused on subtropical anticyclonic anomalies. The RWS (Fig. 6a) counterbalanced the rotational terms (Fig. 6b) over the northwest of the anticyclonic vorticity anomaly. The anomalous divergence (S1; Fig. 6c) and advection terms (S2; Fig. 6d) were the main components of the RWS (Fig. 6a). While, the contributions of climatological divergence and its flow (S3 and S4) were small around the anomalous anticyclonic vorticity (Figs. 6e, f). The rotational terms were predominant in the southeastern portion of the vorticity anomaly owing to the strong westerly winds, which may have been compensated by the residuals.
Composite analysis was performed to obtain and compare the statistical characteristics of the SAJ and WP-like patterns. Here, we define two indices that represent the dipole patterns to extract and composite the teleconnection patterns. The index of the SAJ pattern is defined as the deviation of the normalized anomalous geopotential height at 250 hPa (Z250) averaged from the southeast of the Tibetan Plateau to southern China [20–30°N, 80–120°E] and the corresponding anomaly over northern Japan [40–50°N, 130–160°E], with reference to Ueda et al. (2015). The index of the WP-like pattern is defined as the deviation of the normalized Z250 anomaly averaged over the western North Pacific [20–40°N, 130–160°E] and over the Sea of Okhotsk [45–65°N, 130–160°E]. Thus, the northern-low southern-high height anomaly patterns correspond to positive phases. These two indices were calculated from the monthly mean data, and normalization was based on the standard deviation of each month. The wintertime months for which each index exceeded +1 (−1) standard deviation for the 141 (DJF × 47 years) calendar months were used as those for the positive (negative) phase of the dipole patterns. For the composite of the WP-like pattern, additional filtering was applied using the OLR anomaly averaged over the domain of [5°S–10°N, 120–160°E], to determine whether convection was enhanced over the tropical western Pacific. In this analysis, we focused on the positive (negative) WP-like pattern accompanied by enhanced (suppressed) convection over the tropical western Pacific, such as in the case of winter 2020/21. Based on these criteria, 20 (21) months were selected as the positive (negative) SAJ pattern months and 11 (9) months were selected as the positive (negative) phase of the WP-like pattern accompanied by a negative (positive) OLR anomaly (Table 1). Interestingly, the months of the WP-like pattern were divided by almost half by OLR filtering.
Figure 7 shows the composite anomalies of geopotential height, OLR, and air temperature between the positive and negative phases based on the monthly mean data. The positive phase of the SAJ pattern was significantly correlated with enhanced convection over the Maritime Continent, and its maximum was discernible over the South China Sea (Fig. 7a). The anomalous northwesterly flow in the lower troposphere indicates an intensified East Asian winter monsoon, causing colder-than-normal conditions over the Far East (Fig. 7c). These features bear a considerable resemblance to those described in previous studies; however, the positions of the anomalous convection and northern circulation of the dipole in our results differ slightly from their results based on the DJF mean. When the SAJ index is applied to the DJF mean data (Fig. S2), the anomalous circulation was almost the same as that indicated previously (Zheng et al. 2013; Ueda et al. 2015; Abdillah et al. 2017).
(a) Composited deviations of geopotential height (contours; unit: m) at 250 hPa and OLR (shading; unit: W m−2) between the 20 strongest positive months and the 21 negative months of the SAJ index. (b) Same as (a), but for the deviations between the 11 strongest positive months and the 9 negative months of the WP-like index that satisfied the criteria of tropical convection (see text for definitions of the indices and details of the criteria). Rectangles indicate the regions used for the indices. (c, d) Same as (a, b), respectively, but for geopotential height (contours; unit: m), air temperature (shading; unit: °C), and horizontal wind (vectors; unit: m s−1) at 850 hPa. Black dots indicate the statistical significance at the 95 % level of the shading quantities. Vectors are plotted only where the zonal or meridional component is statistically significant at the 95 % confidence level.
The composited Z250 anomaly based on the WP-like index filtered with the OLR anomaly (Fig. 7b) is more zonally elongated than that of winter 2020/21 (Fig. 2b) and resembles the canonical WP pattern. The defined WP-like index is significantly correlated with the WP index defined at 500 hPa by Wallace and Gutzler (1981) (r = −0.89); note that the sign is the opposite. The weak baroclinic structure of the southern pole of the dipole as a southward tilt with height (Figs. 7b, d) was consistent with that proposed by Tanaka et al. (2016). Anomalous convective activity was distributed only around the Philippine Sea, and no modulated convection was confirmed over the eastern Indian Ocean and the Malay Peninsula (Fig. 7b), which may be important when considering the effective wave source. Cold anomalies were evident around the Sea of Okhotsk to the far eastern Siberia, and oppositely air temperatures over western Japan tended to be warmer, associated with anomalous southeasterly winds (Fig. 7d).
The correlation coefficients between the Z250 and ENSO indices are listed in Table 2. The SAJ pattern is correlated with ENSO, especially tropical western Pacific SST. These correlations became more potent on the seasonal (DJF) timescale. Indeed, the upper-tropospheric height anomaly over southern China, an action center of the SAJ pattern, is recognizable in the response to the wintertime ENSO (Wang et al. 2000; Shiozaki and Enomoto 2021). A recent study showed that SAJ-type anomalies tend to appear in the first winter of the multi-year La Niña, which is more influential on the East Asian air temperature than the second winter (Nishihira and Sugimoto 2022). In contrast, the WP-like index (not filtered) showed no significant correlation with ENSO.
In Section 3.2, it was implied that the SAJ and WPlike patterns were attributed to the diff erent vorticity balances. We also performed the vorticity budget analysis for the composited anomalies. Figure 8a shows a comparison of the vorticity tendency terms in Eq. (4), averaged over the center and upstream nodes of the subtropical anticyclonic anomaly in the positive phase of the SAJ pattern. At the center of the anomaly (heavy blue bars), the S3 term counterbalances the S1 term, which can be written as:
 |
(a) (Upper) Composited anomaly of vorticity at 250 hPa in the positive months of the SAJ index. Solid and dashed contours with an interval of 1.0 × 10−5 s−1 indicate the positive and negative values, respectively, and zero contours are omitted. Heavy solid and light dashed color rectangles denote the region of the center [20–30°N, 90–110°E] and upstream node [20–30°N, 75–95°E] of the anticyclonic eddy, respectively. (Lower) Area-averaged vorticity tendency terms at 250 hPa based on Eq. (4) (unit: 10−10 s−2). Heavy and light color bars indicate the center and upstream node of the eddy. (b) Schematic of the vorticity budget of the anticyclonic eddy (AC′) over the southeast of the Tibetan Plateau in the upper troposphere associated with the positive phase of the SAJ pattern. Underlines denote the negative value of the term in the vorticity balance. See Sections 4.2 and 6 for details.
Figure 9 shows the vorticity budget of the positive WP-like pattern accompanied by enhanced convection over the tropical western Pacific. Here, emphasis is placed on the center of the subtropical eddy; however, the composited vorticity anomaly is zonally elongated and its center is unclear (Fig. 9, upper panel). The vorticity budget shows the predominant contributions of the rotational, S1, and residual terms. The intense jet stream is responsible for the large magnitude of the rotational terms, and S1 attempts to balance them. The residual term may be associated with relatively short-period eddies. Indeed, Tanaka et al. (2016) and Sekizawa et al. (2021) indicate the importance of feedback forcing by submonthly and transient eddies in the maintenance of the WP(-like) pattern. Similar to the 2020/21 case (Fig. 6), the relatively small negative S2 term represents advection by anomalous divergent winds induced by the enhanced convection over the Philippine Sea (Fig. 7b). The S3 term is almost zero because of the in situ non-divergence of the background winds (Fig. 4b), which differs from the balance of the SAJ pattern (Fig. 8).
(Upper) Composited anomaly of vorticity at 250 hPa in the positive months of the WP-like index satisfied with the OLR− criteria. Solid and dashed contours with an interval of 1.0 × 10−5 s−1 indicate the positive and negative values, respectively, and zero contours are omitted. Rectangle denotes the area-averaged region [20–30°N, 130–160°E]. (Lower) Area-averaged vorticity tendency terms at 250 hPa based on Eq. (4) (unit: 10−10 s−2).
The previous sections indicated that regionally different heating anomalies in the tropics could be responsible for the two types of teleconnection patterns. To confirm this hypothesis, sensitivity experiments for tropical convective heating were conducted using the LBM described by Watanabe and Kimoto (2000). We performed two experiments that differed only in the positions of the prescribed heating: one was over the South China Sea (the center was at 8°N, 110°E), and the other was over the Philippine Sea (the center was at 10°N, 135°E), which corresponded to the composited OLR peaks (Figs. 7a, b). The imposed heating exhibited an oval shape, with its heating maxima at its center (as shown in Fig. 10). The heating also had a vertical structure that peaked at σ = 0.45, where the maximum heating rate was approximately +1.0 K day−1, which imitated the vertical profile of the composite anomalous diabatic heating calculated as the residual of the thermodynamic equations (Yanai et al. 1973). The background atmospheric state was DJF climatology derived from JRA-55, and the response on day 15 is shown when the model reached a quasi-steady state.
Atmospheric response on day 15 in the LBM to the prescribed heating centered at (a) the South China Sea (8°N, 110°E) and (b) the Philippine Sea (10°N, 135°E). The geopotential height at 250 hPa is plotted (contour intervals: 2.0 m). Purple solid contours indicate the prescribed heating (+0.6 and +0.3 K day−1) at σ = 0.45. Black dots are the anticyclone center specified as the maximum of the streamfunction at 250 hPa on days 1, 2, 3, 5, 10, and 15 in the LBM. The numbers near the dots denote the days.
Figure 10 shows the steady response of the upper troposphere to imposed tropical heating. The response to South China Sea heating roughly reproduced the wave train of height anomalies corresponding to the SAJ pattern (Fig. 10a). The anomalous anticyclone north of the Indian subcontinent could be interpreted as the western margins of the Matsuno–Gill type heat-induced atmospheric response (Matsuno 1966; Gill 1980) when the vertical gradient of the heating is considered (Wu et al. 2015a). It should be noted here that the anticyclone seen in the geopotential height field locates slightly westward in comparison with the reanalysis (Fig. 7a). This may be partly due to the lack of extratropical forcing as well as a model bias (Tseng et al. 2020). The vorticity balance indicated in the present study (Figs. 5, 8) is inherently caused by the combined effect of climatological mean flows and external forcing. Hence, the SAJ-like steady perturbation vorticity tends to emerge over the region of convergence area of winds and under the subtropical jet in the climatological mean (Fig. 4). As for the atmospheric response to heating over the Philippine Sea (Fig. 10b), the positive height anomaly is recognizable to the south of Japan. Its position is the same as that of the southern part of the WP-like pattern. Note that the northern cyclonic anomaly of the WP(-like) pattern was not clearly reproduced.
The noteworthy difference between the results of the two experiments is the subsequent steady positions of the generated anticyclonic anomalies, although the imposed heat sources were only 25° longitude apart from each other. If we pay attention to its time evolution, after the anticyclone was generated over the northwest of the heat source as a first response in the South China Sea experiment, it shifted northward and consequently stayed north of the Indian subcontinent (Figs. 10a, S3). Similarly, the anticyclonic anomaly seen in the results of the Philippine Sea experiment was also first generated northwest of the heat source, after which, however, it moved northeastward and stayed north of the heat source (Figs. 10b, S3). These meridional shifts of the anomalous circulations may be explained by the effects of the background divergent wind. As mentioned in the RWS analysis, the climatological divergent flow advects the anomalous vorticity poleward in the subtropics (S4, Figs. 5, 6). Using numerical experiments, Sekizawa et al. (2021) demonstrated that poleward vorticity advection is an essential role of the background divergent wind in the establishment of the subtropical anomalous anticyclone. Based on this idea, the additional eastward shift of the anomalous anticyclone of the WP(-like) pattern can be attributed to advection by the background jet core represented by the rotational terms in Eq. (2). Thus, it is suggested that the unique background winds, as well as the regionally different anomalous heating, are responsible for the significantly different steady locations of the anomalous anticyclones and the related teleconnection patterns.
The wave train pattern referred to as the SAJ pattern in this study had been indicated in association with East Asian climate variability during winter in previous research. To confirm its significance for the interannual variability of atmospheric circulation in Asia, we performed an empirical orthogonal function (EOF) analysis to the seasonal (DJF) mean Z250 over the domain of [10°S–70°N, 10–180°E], including the tropical and extratropical upstream regions. The spatial distributions of the anomalies of streamfunction and OLR corresponding to the first (EOF1) and second EOF modes (EOF2) are illustrated in Fig. 11. EOF1 explained approximately 23.4 % of the total variance for the period 1974/75–2020/21, whereas EOF2 explained approximately 20.0 %. The anomalies of upper-tropospheric circulation and convective activity associated with EOF1 resemble those based on the SAJ index (Figs. 7a, c, S2). The correlation coefficient between the seasonal mean SAJ index and EOF1 is 0.74. Furthermore, EOF1 is similar to the height anomalies associated with the dominant variation of the East Asian winter monsoon (Sakai and Kawamura 2009) and rainfall over the Indo-western Pacific sector (Zheng et al. 2013). In addition, we noticed a similar anomaly around southern China in EOF2 (Fig. 11b), although its amplitude is smaller than that of EOF1. These results imply that the region from the southeast of the Tibetan Plateau to southern China is an important fluctuation center of the upper-tropospheric circulation from the teleconnection perspective, which may be attributed to the intense background convergence of winds as well as the climatological subtropical jet (Figs. 4, 5, 8).
Upper-tropospheric perturbation associated with the EOF modes of the interannual variation of geopotential height at 250 hPa over Asia [10°S–70°N, 10–180°E] from 1974/75 to 2020/21. Black contours denote the regression coefficients between the streamfunction at 250 hPa and the (a) first and (b) second principal component indices (unit: 106 m2 s−1). Red and blue shading denote the positive and negative correlations exceeding the 90 % confidence level, respectively. (c, d) Same as (a, b), respectively, but for OLR (shading; unit: W m−2). Black dots indicate the statistical significance at the 90 % level. The signs of second mode were adjusted to the opposite.
It should be pointed out that the circulation anomalies around Japan of EOF1 and EOF2 differ remarkably from each other (Figs. 11a, b), despite the similar anomaly of convective activity around the Maritime Continent and the resultant anomalous circulation over southern China (Figs. 11c, d). The wave train northeastward from the Maritime Continent via Japan to the Kamchatka Peninsula seen in EOF2 (Fig. 11b) can be identified as a WP(-like) pattern accompanied by anomalous circulation over southern China rather than the SAJ pattern. The circulation anomaly over southern China can form the WP(-like) pattern by acting as a vorticity source for the emanation of northeastward-propagating stationary waves. Indeed, the modulated circulation field of EOF2 bears a striking resemblance to the geopotential height anomalies associated with variation in the Australian summer monsoon (Sekizawa et al. 2021) and El Niño (Shiozaki et al. 2021), which were reported to induce a circulation anomaly over southern China and the subsequent WP pattern. The OLR anomaly of EOF2 consistently extended to the Southern Hemisphere (Fig. 11d) compared to that of EOF1.
An essential difference between EOF1 and EOF2 appears to be the wave train at higher latitudes. In EOF1 (Fig. 11a), the wave train along the polar jet from northern Europe via Siberia to Japan can be identified as the Eurasian (EU) pattern (Wallace and Gutzler 1981). The EU pattern shares the anomalous circulation over Japan with the SAJ pattern. It is suggested that this coherent amplification along the polar and subtropical jets is an important factor in generating the different circulation anomalies over Japan compared with EOF2. To assess the relationship between the wave propagation along the two main jet streams is awaited, which are linked with the EU and SAJ patterns respectively.
5.2 Relationship between the WP(-like) pattern and the tropical modulationsAs for the controversial issue regarding the WP pattern, its linkage with tropical variations such as ENSO remains unsolved. As shown in Table 1, not all of the positive (negative) WP-like patterns were caused by enhanced (suppressed) convection in the south. Thus, the correlation with tropical modulations is unclear (Table 2). Nevertheless, it is still influential in determining the phase of the WP pattern that is likely to be triggered (Tanaka et al. 2016; Dai and Tan 2019). Our study provides a possible mechanism by which tropical modulation directly induces the WP(-like) pattern by providing a wave source for the southern portion of the dipole.
The anomalous circulation associated with EOF2 indicates another type of WP(-like) pattern concurrent with the opposite convection anomaly, that is, the cyclonic anomaly of the southern part of the dipole against the enhanced convection over the tropical western Pacific (Figs. 11b, d). As mentioned in the preceding subsection, this WP(-like) pattern is accompanied by an anticyclonic circulation anomaly over southern China and intensified heating around the Maritime Continent. The anomalous circulation over southern China could be an important factor in generating the WP(-like) anomaly (Sekizawa et al. 2021). Thus, in the case of EOF2, the enhanced convection around the Maritime Continent might indirectly induce the negative WP(-like) pattern through wave propagation. The relationship between the phase of the WP(-like) pattern and convective activity over the tropical western Pacific should be surveyed in greater detail in future studies.
In the present study, the modulated atmospheric circulation associated with the turnabout of air temperature anomalies over East Asia between the first and latter halves of the winter of 2020/21 was examined. In the colder-than-normal first half, a pair of anticyclonic and cyclonic circulation anomalies appeared in the upper troposphere over the southeastern Tibetan Plateau and northern Japan, respectively. This southwest-northeast dipole pattern (referred to as the SAJ pattern in this study) was accompanied by enhanced convection over the South China Sea toward the eastern Indian Ocean and the stationary Rossby wave propagation along the subtropical jet. However, in the latter half of winter, the anticyclonic anomaly in the subtropics shifted to the south of Japan, and the resultant meridional dipole was identified as the WPlike pattern. It altered the cold airflow around East Asia, causing warmer-than-normal conditions. Simultaneously, the center of the reinforced convection shifted eastward to the Philippine Sea. The subsequent divergence and flow between the two periods related to the intraseasonal eastward shift of convective activity induced different anomalous vorticity fields in the upper troposphere, satisfying each vorticity balance.
Further analysis revealed the inherent spatial structures and dynamics of the two teleconnection patterns. The composite analysis showed that the SAJ pattern significantly influenced the East Asian winter monsoon, and the magnitude of the anomalous air temperature around the Sea of Japan was larger than that of the WP-like pattern. The center of the modulated convection emerged over the South China Sea when the SAJ pattern appeared on the monthly timescale. The dynamical structures of the SAJ pattern, focusing on the vorticity balance of the anticyclonic anomaly corresponding to the southwestern pole of the dipole, are schematically shown in Fig. 8b. Anomalous diabatic heating around the Maritime Continent induces an anticyclonic anomaly in the upper troposphere southeast of the Tibetan Plateau with the aid of climatological divergent winds advecting the anomalous vorticity poleward (S4). At the center of the anticyclonic eddy, the negative vorticity perturbation on the climatological convergence of winds (S3) counterbalances the anomalous convergence associated with the enhanced convection (S1). The negative vorticity advection by anomalous divergent flows crossing the subtropical jet northward (S2) acts to compensate for the positive advection by the mean rotational winds (rot.), together with anomalous divergence to maintain the wave train structure in the upstream nodes of the eddy. The anticyclonic anomaly is responsible for the cyclonic anomaly around Japan through the propagation of stationary Rossby waves. The vorticity budget suggests the critical role of the background convergence of the winds over the southeast of the Tibetan Plateau in the emergence and maintenance of the spatially phase-locked anomaly of the SAJ pattern.
In contrast, half of the positive (negative) WP-like patterns were accompanied by enhanced (suppressed) convection over the Philippine Sea, which could directly induce the southern pole of the WP-like pattern. The dominant vorticity advection by the rotational winds and residual terms played essential roles in the vorticity balance of the WP-like pattern. The results of the LBM experiments support the regionally different heat sources and background winds responsible for the two types of teleconnection patterns.
Although we attempted to understand the winter-time climate variability over East Asia in relation to some teleconnection patterns, especially from the perspective of tropical forcing, it is also necessary to investigate it from mid- and higher latitudes. Further studies focusing on tropical–extratropical interactions are required to better understand the dynamics of teleconnections and to improve seasonal forecasting.
The JRA-55 and COBE-SST datasets are provided by Japan Meteorological Agency (https://jra.kishou.go.jp/JRA-55/index_en.html; https://ds.data.jma.go.jp/tcc/tcc/index.html), and the Interpolated OLR and OISST datasets by NOAA PSL (https://psl.noaa.gov/). Documents and codes for the LBM are available at https://ccsr.aori.u-tokyo.ac.jp/∼lbm/sub/lbm.html.
Supplement 1 shows spatial distributions of decomposed rotational terms in Eq. (2) and the β term during Period 1. Supplement 2 shows Z250 anomaly associated with the DJF mean SAJ index. Supplement 3 shows time evolution of upper-tropospheric circulation anomalies in the LBM experiments.
The authors are grateful to Dr. Meiji Honda and Dr. Koutarou Takaya for their helpful comments and discussion. We also thank Dr. Masaru Inatsu and the anonymous reviewers for their constructive comments on the earlier version of this paper. This work was supported by the Japan Society for the Promotion of Science (JSPS) Grants-in-Aid for Scientific Research (KAKENHI), Grant Number 21H00626K, the industry-academia collaboration project for the anomalous warm winter 2019/2020 of University of Tsukuba, and the Global Environment Research Fund (4-2102) of the Ministry of the Environment, Japan.
