Relative Roles of Intraseasonal and Above-seasonal Components in the South China Sea Summer Monsoon Onset

The South China Sea (SCS) summer monsoon (SCSSM) is an important component of the East Asian summer monsoon (EASM), and its onset represents the commencement of the EASM. The year-to-year variations of the timing of the SCSSM establishment and its northward propagation, as well as its intensity might directly affect the spatial distribution and evolution of the rain belt in eastern China. The onset time of the SCSSM exhibits obvious interannual and interdecadal variations, and the onset abnormality not only has influence on water resources and occurrence of flood disasters in southern China, but also affects the global climate through teleconnections (Huang and Sun^[1]). Therefore, it is imperative to investigate the characteristics and mechanisms of the interannual and interdecadal variations of the SCSSM onset.

The establishment of SCSSM is featured by the rapid seasonal transition of low-level zonal winds from easterlies to westerlies and upper-level zonal winds from westerlies to easterlies over the SCS. In fact, the seasonal evolution of the monsoon circulation is attributed to the collaborative effects from the multiscale components. For this reason, to reveal the characteristics and mechanisms of the SCSSM onset variability requires a clear understanding of the relative roles of different timescale components in the seasonal transition of the circulation over the SCS monsoon region. The intraseasonal oscillations (ISO) are considered as the dominant intraseasonal variability in the tropics, and their activities have a significant influence on weather and climate in many regions. In the early 1970s, Madden and Julian^[2-3] discovered the obvious 30-60-day ISO in the tropics, which is characterized by eastward propagation in the tropical Indian Ocean (TIO) and northward propagation in the tropical western Pacific (Li^[4]). In recent years, a large number of studies have focused on the relationship between ISO and the SCSSM onset. Some studies have proposed that the eastward propagation of ISO over the TIO, the formation and development of low-frequency cyclones over the ocean east of the Philippines, as well as the movement of cyclones to the SCS have substantial impacts on the SCSSM onset. Moreover, some studies have emphasized the collaborative influence of 30-60-day and 10-20-day ISOs on the SCSSM onset (Chen and Chen^[5]; Mu and Li^[6]; Chan et al.^[7]; Wen et al.^[8]; Zhou and Chan^[9]; Mao and Chan^[10]; Lin et al.^[11]; Huangfu et al.^[12]). For example, the collaboration of northeastward propagation of the 30-60-day oscillations and northwestward propagation of the 10-20-day oscillations can cause the SCSSM to outbreak earlier by weakening the subtropical high ridge (Zhou and Chan^[9]). Shao et al.^[13] revealed that the SCSSM onset is closely related to the phase transition of ISO from inactive to active convection, and the onset always occurs in the ISO developing phase. The earlier enhancing of ISO, with earlier appearance of low-level westerly wind anomalies and active convection anomalies over the SCS, is favorable for an earlier SCSSM onset. Furthermore, Wang et al.^[14] suggested that the phase transition of ISO from dry to wet has significant impact on the SCSSM onset, and more importantly, the ISO evolution and mechanisms of monsoon onset are diverse in different types of SCSSM onset.

Additionally, the atmospheric ISO also exhibits considerable interannual variation. Some studies suggested that the central and western tropical Pacific regions may play an important role in the interannual variation of tropical ISO (Li and Zhou^[15]), and the Walker circulation strength in spring as well as the strength of convection over the SCS might be treated as early signals of the interannual variation of ISO in summer (Ju et al.^[16]). Besides, the ISO signal originating from the tropical western Indian Ocean can provide useful information for the prediction of the SCSSM onset, while sea surface temperature (SST) in the Northwest Pacific, associated with El Niño-Southern Oscillation (ENSO) and the Pacific Decadal Oscillation, can provide a basis for predicting the types of ISO evolution before the monsoon outbreaks (Wang et al.^[14]).

In addition to studies on the ISO, previous studies have shown that above 90-day component can also exert significant impact on the SCSSM onset. Ju et al.^[17] examined the relationship between Asian summer monsoon and ENSO by using observational data and results of model simulation, and proposed that there were obvious interannual changes in both monsoon intensity and monsoon onset date, showing as weak monsoon circulation and late SCSSM onset in El Niño years but strong monsoon circulation and earlier onset in La Niña years. Zhou and Chan^[18] obtained similar results in a study of the SCSSM. Some studies further showed that the decadal variation of SST in Indo-Pacific oceans may affect the decadal variation of the SCSSM onset date evidently (Kajikawa and Wang^[19]; Yuan and Chen^[20]; Lin and Zhang^[21]; You et al.^[22]; Zeng et al.^[23]).

Previous studies mainly aim at the effects of intraseasonal or longer time-scale oscillations on the interannual variability of SCSSM onset, but the relative contributions of various time-scale components to the SCSSM onset remain unclear at present. Thus, in this study, we aim to analyze the relative contributions of different time-scale components to the interannual variation of the SCSSM onset, and further discuss the influence processes of the key components on the SCSSM onset and their causal factors and mechanisms.

The remainder of this paper is organized as follows. Section 2 is the introduction of the dataset and methods used in this study. In Section 3, we explore the relative contributions of different components to the anomalies of SCSSM onset. The evolution and mechanism of ISO and above-90-day component are discussed in Section 4 and Section 5, respectively. Finally, conclusions and discussion are presented in Section 6.

The datasets used in this study include daily mean winds and geopotential height from the European Centre for Medium Range Weather Forecasts (ECMWF) Interim reanalysis dataset (ERA-interim; Dee et al.^[24]) during the period 1980-2016 and outgoing longwave radiation (OLR) data from the National Oceanic and Atmospheric Administration (NOAA) satellites (Liebmann and Smith^[25]), with a horizontal resolution of 2.5°×2.5°. In addition, the daily SST data is from the Met Office Hadley Centre, with a horizontal resolution of 1° × 1° (Rayner et al.^[26]).

Because the SCSSM onset is marked by the stable establishment of southwesterly winds at lower level and northeasterly winds at upper level over the SCS, this paper selects the area-averaged 850-hPa daily zonal winds (U850) in (5°-17.5°N, 110°-120°E) of the SCS as the index for the convenience of subsequent analyses (Wen et al.^[8]). The SCSSM onset date is defined as the index satisfying the following criteria: the filtered U850, with below 7-day component removed, shifts from negative to positive after April 21, and the persistently positive value duration is longer than any following persistently negative value duration occurring in the period from April 21 to June 30, then the first day of the aforementioned persistently positive value duration is defined as the SCSSM onset date.

The Lanczos filtering procedure (Duchon^[27]) is adopted in this paper for the ISO filtering.

According to the aforementioned definition of monsoon onset date, we have obtained the SCSSM onset dates in 1980-2016 (table omitted). The climatological average SCSSM onset date is May 16, with a standard deviation of 11.3 days. Then, we identified the early years (late years) of the SCSSM onset with the onset date 5 days earlier (later) than May 16, with standard deviations being nearly 0.5. Following this criterion, we classified 16 normal years with the onset date between May 11 and May 21, 12 early years with the onset date earlier than May 11, and 9 late years with the onset date later than May 21 (Tab. 1). Fig. 1 presents the composite time series of U850 averaged over the SCS for the early, normal, and late onset groups. The evolution of U850 shows a sharp shift from negative to positive and enhancement around the onset stages in the early and normal onset cases, but a slow change in the late case.

Figure 1.  Composite time series of U850 averaged over the SCS (5° - 17.5° N, 110° - 120° E) for early onset years (red dashed line), normal onset years (black dot line), and late onset years (blue dashed line). The black solid line denotes the climatological mean state.

To identify the dominant components affecting the SCSSM onset, the time series of area-average daily zonal wind anomalies over the SCS (5°-17.5°N, 110°- 120°E) from 1980 to 2016 was first processed by a threepoint running average procedure with the weighting coefficients 0.25, 0.5 and 0.25, respectively, to filter out the high-frequency signals. Then, the time series of daily zonal wind anomalies from April to June each year were selected for the power spectrum analysis. Based on Fig. 2, we can divide the temporal domain into four subdomains, namely 2-7-day, 8-30-day and 30-90-day, and above 90-day, which are associated with high-frequency synoptic process, two intraseasonal oscillations, and lowfrequency component on above-seasonal scale, respectively, while two intraseasonal oscillations are most dominant among them. Actually, the 2-7-day component of high-frequency synoptic process makes very little contribution to the abnormal onset cases; therefore, we pay attention to the roles of other three main components in the SCSSM onset.

Figure 2.  Power spectrum (solid line) of regional mean U850 over the SCS (5°-17.5°N, 110°-120°E). The red dashed line denotes the power spectrum for red noise at a confidence level of 95%.

In fact, all the above-mentioned components of zonal winds over the SCS always exert simultaneously their impacts on the SCSSM onset, meaning that their respective roles in the SCSSM onset process are always accompanied with effects of other components. Given this situation, the"deduction procedure"is adopted to detect the relative contribution of each component to the early and late onset cases. The"deduction procedure"is performed as follows: for one year, we can deduct a concerned component from the filtered U850 with below 7-day component removed, and then identify a new onset date (named as"deducted date") based on the deducted U850 series. The difference between the original onset date and the"deducted date"represents relative contribution of the concerned component to the discussed SCSSM onset case. Specifically, a negative (positive) difference represents advanced (delayed) contribution by the concerned component to the SCSSM onset.

As shown in Table 2, in early onset years, both above-90-day and 30-90-day components make important contributions to the earlier onset, and the latter one is more important, while the 8-30-day component has very small and nearly negligible effect on the SCSSM onset abnormality. Similarly, in late onset years, both the above-90-day and 30-90-day components also play key roles in the later onset, while the contribution of 8-30-day component is very little. In summary, the quasi-biweekly component has little influence on the SCSSM onset abnormality, consistent with the conclusion by Chen^[28], while both above-90-day and 30-90-day components play major roles in the SCSSM onset abnormality, and the latter contributes about twice as much as the former. In addition, since the average onset date is May 16, and the total contribution of all components in early and late onset years is 10.8 days earlier and 14.6 days later than the average onset date, respectively, so anomalous onset date caused by them are very consistent with the actual average onset dates of the early and late group, which are May 4 and May 30, respectively.

Figure 3 shows the evolution of the composite anomalies of three components in the early and late onset cases. In early case (Fig. 3a), from day -2 to day 2, the positive anomalies of above-90-day and 30-90-day components are pronounced and their amplitudes are quite comparative, which coordinatively makes the SCSSM outbreak in early May in advance, while the anomalies of the 8-30-day component are negative within 7 days before day 0 and very small around day 0, so its effect on the early SCSSM onset is negligible. In late onset years (Fig. 3b), during day -15 to day 0, both above-90-day and 30-90-day components show continuously remarkable negative anomalies, and the amplitude of the latter component is more pronounced, so its role in delaying onset is more obvious than the former one. In contrast, the anomalies of the 8-30-day component maintain negative only within 9 days before the onset date, and its amplitude is also much smaller than the 30-90-day component and becomes comparative to the latter one only within 3 days before day 0; therefore, it has less effect on the onset abnormality. These facts further confirm the results in Table 2.

Figure 3.  Composite time series of different components of U850 averaged over the SCS (5°-17.5°N, 110°-120°E) in (a) early onset years and (b) late onset years. The vertical dashed line indicates the onset date (Day 0) of early or late onset years, and"Day-5"or"Day+5"on the abscissa denotes 5 days before or after the onset date, respectively.

The spatio-temporal evolution of the SCSMM onset process may be different in early/late onset years. Next, the influence of the circulation evolution of 30-90-day and above-90-day components on SCSSM onset in early/late years is discussed. The 8-30-day component will not be discussed, because it has a negligible impact on monsoon onset suggested by the previous discussion.

Figure 4 shows the composite evolution patterns of 850-hPa wind and OLR anomalies on 30-90-day time scale around the SCSSM onset stage in early onset years. From day -20 to day -5 (Figs. 4a-d) before the onset, the low-frequency anticyclone over South China moves southeastward and develops, and its ridge line controls the northern SCS, accompanied with positive OLR anomalies and weak convective activities as well as low-frequency easterly wind prevailing in the SCS. Meanwhile, negative OLR anomaly center (active convection center) propagates eastward from the tropical western Indian Ocean to the tropical eastern Indian Ocean and strengthens remarkably, with a pair of lowfrequency cyclones existing on both sides of the equatorial eastern Indian Ocean (Fig. 4d), and the easterly wind shifts to westerly over the TIO, while the region from Maritime Continent to western tropical Pacific is still controlled by the easterly wind. On the onset day (Fig. 4e), the cyclones over the eastern equatorial Indian Ocean move poleward and the cyclone originating from mid-latitude area of East Asia moves southward to South China. Moreover, negative OLR anomalies near the SCS-Philippines strengthen, meaning enhanced convection and giving rise to the eastward withdraw of the anti-cyclone from the SCS, and the westerly wind begins to occupy the SCS monsoon region, causing the SCSSM to outbreak. From day 0 to day + 15 (Figs. 4e-h), the cyclone over South China moves eastward, strengthens, and expands over time. It is noticed that the evolution of 30-90-day circulation over the TIO during day 0 to day + 15 is almost opposite to that in day -20 to day -5. In short, from late April, the coordination of these low-frequency circulations is conducive to the SCSMM onset in early May in advance, like the 30-90-day cyclone over East Asia moving southward to the northern SCS, a pair of cyclones forming over the eastern TIO and moving poleward, and inhibited convection as well as lowfrequency westerly winds prevailing over the TIO.

Figure 4.  Composite 30-90-day oscillation of 850-hPa wind (streamline) and OLR anomalies (shading, units: W m^-2) in early onset years."Day+5"means 5 days after the SCSSM onset date (Day 0). The stippled areas denote the OLR anomalies significant at the 95% confidence level.

Next, we discuss the influence of above-90-day component in early SCSSM onset cases. In April, a cyclonic circulation of above-90-day component occupies the lower troposphere over the SCS to the subtropical northwestern Pacific (Fig. 5a). Furthermore, the anomalous cross-equatorial flow over the western TIO, which originates from the northern flank of the anti-cyclone over southwestern Australia, moves northward and turns to the westerly wind with eastward expansion, converges with a northerly airflow from East Asia over the south of Indo-China Peninsula and flows eastward to the northwestern Pacific. Under the control of above circulation, intensified convection indicated by pronounced negative OLR anomalies prevails over the southern Bay of Bengal, the central to southern SCS and the western tropical Pacific. All the above-mentioned anomalous features in the above-90-day circulation and convection around the SCS are potentially conducive to the early SCSSM onset. Therefore, some striking features of anomalous circulation in April, such as the enhanced cross-equatorial flow over the western TIO and anomalous westerlies over the northern TIO, intensified convection over the SCS and surrounding regions, and suppressed convection over the southern Australia, can be regarded as precursory signals of the early SCSSM onset.

Figure 5.  Composite above-90-day component of 850-hPa wind (streamline) and OLR anomalies (shading, units: W m^-2) in (a) April and (b) May in early onset years. The stippled areas denote the OLR anomalies significant at the 95% confidence level.

The distribution of composite above-90-day winds and OLR anomalies in May is similar to that in April, but the northern boundary of the westerly wind over the Asian region including the SCS extends a bit northward (Fig. 5b), which is a feature of anomalous winds caused by the early monsoon onset.

Figure 6 shows the composite evolution pattern of 30-90-day-filtered 850-hPa wind and OLR anomalies in late onset years. In day - 20 to day - 5 (Figs. 6a-d), before the onset, the SCS-subtropical northwestern Pacific is continuously controlled by the low-level anticyclonic circulation anomalies, and the central to southern SCS is mainly occupied by the easterlies. Meanwhile, accompanied with anomalous convection (negative OLR anomalies) propagating eastward and strengthening from the western TIO to the eastern TIO and Maritime Continent, the low-level 30-90-day easterlies over the TIO also shifts to westerlies. Moreover, the suppressed convection zone with positive OLR anomalies over the Bay of Bengal also moves eastward to the SCS-Philippine Sea, matching with the persistent anticyclonic circulation in this region.

Figure 6.  Same as Fig. 4, but for late onset years.

However, during day -5 to day 0 (Fig. 6d-e), the anomalous convection over the TIO weakens and moves northward. The convection over the South China strengthens while the convection activities over the SCS are less suppressed, with the west boundary of easterlies over the tropical Asia retreating eastward to 100° E nearby, which indicates that active convection over the TIO-Maritime Continent is not conducive to the SCSMM onset and maintenance. Nevertheless, considering the SCS monsoon region is dominated by significant westerly winds since mid-May under climatological mean condition (inferred from Fig. 1), as the suppression effect by the circulation and convection of 30-90-day component weakens to a certain extent around the end of May and cannot resist the climatologically seasonal intensification of westerly winds in situ, the SCSSM outbreaks finally. With the onset of the SCSSM and the emergence of suppressed convection phase in the western TIO and the Indian Peninsula, the convection activities and westerly winds over the SCS and the Philippine Sea are enhanced (Fig. 6f), which is conducive to the maintenance of the SCSSM after onset.

In late onset years, characteristics of the composite above-90-day component of 850-hPa circulation and OLR anomalies in April and May (Fig. 7) are generally opposite to early onset years, showing that tropical Asia is dominated by the easterly anomalies but the western Maritime Continent and South China are affected by negative OLR anomalies. The SCS monsoon region is affected by abnormal easterly winds, and the Bay of Bengal, the SCS as well as the tropical northwestern Pacific are dominated by positive OLR anomalies (weakened convection activities), restraining the monsoon onset.

Figure 7.  Same as Fig. 5, but for late onset years.

The above results show that for different types of anomalous SCSSM onsets, the evolution process of 30-90-day oscillation is mainly manifested by the difference in the origination and the occurrence time of anomalous westerly wind over the SCS monsoon region, while the evolution process of above-90-day component is reflected in the difference in zonal wind and convection anomalies in April and May over the Asian-Australian monsoon region.

Section 3 has shown that the evolutions of both 30-90-day oscillation and above-90-day component have significant influence on the anomalous SCSSM onset by modulating the transition time from low-level easterlies to westerlies over the SCS in the monsoon onset stage from late April to May. Therefore, to understand the mechanisms for the interannual variation of the SCSSM onset, it is essential to explore the relationships of SST in the Indo-Pacific Oceans in preceding winter and following spring with the interannual variation of the negative-positive phase transition time (NPPTT) of 30- 90-day zonal wind during April to June and above-90- day zonal wind in April and May over the SCS monsoon region, respectively.

Figure 8 shows the correlation coefficients of NPPTT of the 30-90-day U850 around the SCSSM onset stage with SST, 850-hPa wind, and OLR. The results show that there is a significant negative correlation between NPPTT and SST in the tropical western Pacific in preceding boreal winter and following spring (Figs. 8a, b), which means that the earlier / later transition is related to abnormally warmer/colder SST in the tropical western Pacific in preceding winter and following spring, accompanied with an anomalous cyclonic / anticyclonic circulation in the lower troposphere over the SCS and surrounding areas. In addition, there exists pronounced positive correlation between NPPTT and OLR over the SCS and surrounding regions, indicating that the earlier / later transition is associated with strengthened / weakened convection activities in the SCS and its vicinity (Figs. 8c and d). Above statistical facts suggest that the abnormally warmer SSTs in tropical western Pacific in winter and spring induce an abnormal low-level cyclone over the SCS and enhanced convection through thermal forcing and Rossby-wave response (Gill^[29]), leading to the earlier NPPTT which is conducive to the SCSSM outbreaks in advance. The situation is opposite when the SST in tropical western Pacific is abnormally colder in winter and spring.

Figure 8.  Correlation coefficients of the negative-positive phase transition time (NPPTT) of 30-90-day U850 during the SCSSM onset stage with (a, b) SST and 850-hPa wind, and (c, d) OLR in preceding (a, c) winter (December-February) and (b, d) spring (March-May). Stippled areas are significant at the 90% confidence level.

Possible physical linkage between the background anomalous circulations (e. g., low-level cyclonic anomaly and enhanced convection) over the SCS and surrounding regions and NPPTT (e.g., advanced) on the 30-90-day time scale might be illustrated as follows. The low-level cyclonic wind anomalies and enhanced convection over the SCS and surrounding areas, which are induced by the positive SST anomalies in the tropical western Pacific in spring via Rossby-wave response, give rise to low-level anomalous westerlies over the eastern TIO and northerlies over East Asia (Figs. 5a, 8b, d). These background anomalous winds favor the advection of the anomalous convection and cyclonic circulation on the 30-90-day time scale to the SCS from the eastern TIO and East Asia, respectively, inducing the development of the 30-90-day cyclonic circulation over the SCS. Then, this process will benefit an early occurrence of the negative-positive phase transition of 30-90-day zonal wind over the SCS in spring (Figs. 4c-e). In contrast, the background anomalous circulation is not conducive to the early phase transition of 30-90-day zonal wind over the SCS under the condition of negative SST anomalies in the tropical western Pacific in spring.

The correlations of regional mean above-90-day U850 over the SCS in April with SST, 850-hPa wind, and OLR in preceding boreal winter and spring are shown in Fig. 9. The correlation with SST is featured by a typical La Niña-like pattern, specifically, with significant negative correlation in the TIO and the equatorial central-eastern Pacific, and horseshoe-shaped positive correlation in the tropical western Pacific and the subtropical North and South Pacific (Figs. 9a-b). The low-level cyclonic correlation pattern (Figs. 9a-b) corresponds to pronounced negative correlation with OLR over the SCS and surrounding regions (Figs. 9c, d). In particular, when the distribution of SST anomaly (SSTA) in the tropical Indo-Pacific oceans in preceding boreal winter presents a zonally positive-negativepositive pattern, i. e., a mature El Niño pattern, an anomalous anticyclone appears in the subtropical northwestern Pacific via Rossby-wave response to the thermal forcing of the tropical Pacific SSTA (Wang et al.^[30]). Meanwhile, the warm SSTA in the TIO in spring, caused by a lagging response to the warm SSTA in the equatorial central-eastern Pacific, can also trigger an eastward-propagating warm Kelvin wave, which favors the persistence of the anomalous anticyclone in subtropical northwestern Pacific (Xie et al.^[31]), suppresses the convection activities over the SCS and causes easterly anomalies of above-90-day low-level wind, resulting in late SCSSM onset eventually. When the distribution of SSTA in the tropical Indo-Pacific oceans in preceding boreal winter and spring resembles a La Niña pattern, the opposite consequence is conducive to early SCSSM onset. The above results are consistent with previous studies on the significant influence of ENSO in the interannual variation of the SCSSM onset (e.g. Chen et al.^[32]; Zhou and Chan^[18]).

Figure 9.  Correlation coefficients of the regional mean above-90-day U850 over the SCS (5°-17.5°N, 110°-120°E) in April with (a, b) SST and 850-hPa wind and (c, d) OLR in (a, c) preceding winter (December-February) and (b, d) spring (March-May). Stippled areas are significant at the 90% confidence level.

This study has examined the relative contribution of different time-scale components of atmospheric circulation to the interannual variation of the SCSSM onset, and further explored the influence of 30-90-day oscillation and above-90-day component on the SCSSM onset, as well as the linkage of interannual variation of two aforementioned components with SST in the tropical Indo-Pacific oceans.

In terms of low-level zonal wind, the phase transition time of 30-90-day component of atmospheric circulation in spring and the intensity of above-90-day component over the SCS have pivotal influences on the interannual variation of the SCSSM onset, and the former one is more important, while the effect of 8-30- day component is negligible.

The evolution of 30-90-day circulation is dramatically different in the early and late SCSSM onset years. In early onset years, the low-frequency westerly winds of 30-90-day component move and extend eastward from the TIO to the SCS monsoon region in late April to early May, and replace the low-frequency easterly winds with the southward movement of lowfrequency cyclone over East Asia, thus facilitating the early SCSSM onset. While in late onset years, not only the 30-90-day westerly signal over the TIO moves and extends eastward later than normal, but also the time of emerging, weakening, as well as moving away of the low-frequency anticyclone over the SCS is also later than normal, leading to the SCSSM onset when the lowfrequency anticyclone is significantly weakened over the SCS.

In early onset years, the above-90-day low-level wind over the SCS monsoon region from April to May presents obvious westerly anomalies and enhanced convection activities, which is conducive to the early onset and maintenance of the SCSSM. While in late onset years, the low-level easterly anomalies and the weakened convection activities of above-90-day component prevail over the SCS and surrounding region in April to May, which significantly inhibit the monsoon onset.

The NPPTT of 30-90-day U850 over the SCS in spring is significantly negatively correlated with the SST in the tropical western Pacific in preceding winter and spring. Moreover, the intensity of above-90-day U850 over the SCS in spring is significantly affected by the decaying stage of ENSO events in a negative correlation.

However, some issues remain to be explored. The impacts of SST anomalies in the tropical western Pacific on the negative-positive phase transition of 30-90-day zonal wind over the SCS in spring still deserve a meticulous study through dynamic diagnosis and numerical simulation. Moreover, the relative contributions of the 30-90-day circulation components in TIO and East Asia to the early or late SCSSM onset cases need to be further clarified.