1. Introduction
Intraseasonal oscillations are an important part of the monsoon variability within the annual cycle. They significantly affect the local weather and climate as well as the global atmospheric circulation. Generally, the summer monsoon activities undergo several active and break periods. Meteorologically, the Indian and East Asian monsoon systems interact over the South China Sea (SCS) region (Chen and Chen 1995), so that it is of significant interest to study the intraseasonal oscillations of the SCS summer monsoon (SM).
Climatologically, the East Asian summer monsoon is mainly characterized by the mei-yu (in eastern China)–baiu (in Japan) front, which is one of the major convergence zones within the global atmospheric circulation (Chen et al. 2000). In addition, Tao and Chen (1987) have suggested that the onset of the SCS SM signifies the beginning of the summer monsoon system in East Asia (see their Fig. 8, in which the climatological positions of the SCS monsoon trough and mei-yu–baiu front are shown).
With Madden and Julian’s (1971, 1972) discovery of the 40–50-day tropical oscillation, the Monsoon Experiment in 1978–79 provided an opportunity to explore further the intraseasonal oscillations that are involved in the Asian summer monsoon activities. Based on the observations, the intraseasonal oscillations in the broad period ranging from 30 to 60 days (referred to interchangeably as the 30–50-day or 40–50-day oscillations) have been identified in the Asian summer monsoon regime (e.g., Krishnamurti and Subrahmanyam 1982; Lorenc 1984; Murakami and Nakazawa 1984, 1985; Lau and Peng 1987). These researchers pointed out that during the northern summer monsoon season, in addition to the eastward propagation, the 30–60-day oscillation appears to migrate northward. They are linked to regional characteristics, such as monsoon onsets and breaks, and the fluctuation of the low-level jets over Indian/Southeast Asia.
Krishnamurti and Ardanuy (1980) also found the 10–20-day westward-propagating oscillation in the Indian monsoon. Chen and Chen (1993, 1995) investigated successively the intraseasonal variations in the 1979 summer monsoon over India and the SCS. They reported that the 10–20-day monsoon mode exhibits a double-cell (either double high or double low) structure, both of which propagate coherently westward along the Indian monsoon trough and along the equator. They further suggested that the SCS SM in 1979 was regulated by the northward-propagating 30–60-day monsoon trough–ridge, and the SCS SM break occurred simultaneously with a phase lock in the northern SCS between the 30–60-day mode and the 10–20-day mode. They speculated that the 10–20-day mode of the Indian monsoon originated over the SCS–western Pacific region, which suggests the impact of the SCS SM on the Indian summer monsoon.
Lau and Yang (1997) considered the SCS SM as a component of the Southeast Asian monsoon, and surveyed the seasonal and subseasonal progression of the monsoon rain zone, involving the establishment and migration of the SCS monsoon trough and the mei-yu–baiu front. The rainfall along the mei-yu–baiu front is associated with the moisture originating from the SCS–western Pacific region; and the moisture and moisture flux in the SCS–western Pacific fluctuate intraseasonally (Chen et al. 1988). Chen et al. (2000) showed that the temporal evolutions of the filtered monsoon indices that are represented by outgoing longwave radiation (OLR) and 850-hPa zonal winds between the SCS and eastern China exhibit an opposite-phase relation.
Based on the data from the South China Sea Monsoon Experiment, Chan et al. (2002) examined the evolution of the 1998 SCS SM, and found that the maintenance and break of the SCS SM in that year were controlled by the 30–60-day oscillation, and were further modified by the 10–20-day mode. Although both the 30–60-day and the 10–20-day oscillations were found to be distinct in the 1979 (Chen and Chen 1993, 1995 and 1998 (Chan et al. 2002) cases, does such a situation occur in all other years? Are these two oscillations the dominant modes that control the SCS SM activities? These questions are to be addressed in this paper.
Indeed, most of the previous work on the SCS SM was based on case studies. The interannual variability of the intraseasonal oscillations has not been systematically examined. The objective of this paper is, therefore, to explore the intraseasonal oscillations of the SCS SM and their interannual variability by examining data from many years.
The data and method that are used in this study are briefly described in section 2. The technique of wavelet analysis is used to classify the different kinds of dominant oscillations in the SCS SM in section 3. Based on such classification, composite analyses of the large-scale circulation are presented in section 4 to reveal the fundamental features in the configuration and propagation of the 30–60-day and 10–20-day oscillations. As suggested by several studies (e.g., Krishnamurti and Ardanuy 1980; Chen and Chen 1993), the 10–20-day mode of the Indian monsoon originates over the SCS–western Pacific. It will be shown in section 5 through a case study that the 10–20-day oscillation of the SCS SM may be related to a mixed Rossby–gravity (MRG) wave that is observed by Chang et al. (1996, hereafter C96) and Dickinson and Molinari (2002, hereafter DM). The summary and discussion follow in section 6.
2. Data and methodology
a. Data
The daily mean variable fields for the period of 1975–2002 from the National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) reanalysis (Kalnay et al. 1996) are the basic data used in this study. Also used are the daily mean OLR data derived from the National Oceanic and Atmospheric Administration for the period of 1975–2002 (except 1978). Because the rainfall over the SCS during August is, to a great extent, affected by typhoons rather than monsoonal precipitation, the OLR records from 1 May to 31 July every year are extracted to investigate the features of intraseasonal variability of the SCSSM.
b. Methodology
The intraseasonal oscillations (ISOs) are characterized in this study by variations in tropical convection, with the daily mean OLR dataset being used as a proxy for large-scale tropical convective activity (e.g., Waliser et al. 1993; Liebmann and Smith 1996; Wu et al. 1999). Many studies have used OLR values to define the onset and development of the summer monsoon (e.g., Xu and Chan 2001). The SCS domain is chosen to be 5°–20°N, 110°–120°E. In their study of the 1998 SCS SM, Chan et al. (2002) found that the temporal variations of the 850-hPa relative vorticity averaged over the SCS had a large negative correlation with those of the OLR. Such a correlation may be intuitively obvious because strong low-level positive relative vorticity in the Tropics is generally associated with strong low-level convergence, which leads to strong convection, and, hence, low OLR values. The area-averaged 850-hPa relative vorticity over the SCS is, therefore, also used as a subsidiary index to identify the intraseasonal oscillations of the SCS SM in order to confirm the results derived from the OLR index.
Wavelet analysis is a common tool for detecting time–frequency variations within a time series. Because the wavelet transform is a bandpass filter with a known response function (the wavelet function), it is also a powerful filtering technique (e.g., Lau and Weng 1995; Torrence and Compo 1998). The wavelet analysis method is, therefore, employed to identify the dominant ISO modes and to isolate the ISO components. The wavelet basis function is the m-order derivatives of a Gaussian (Torrence and Compo 1998), where m is chosen to be 6. Torrence and Compo (1998) discussed several factors in choosing the wavelet function. The current choice is made because the time series of OLR averaged over the SCS exhibits sharp jumps, and such a real wavelet function is suitable to isolate peaks or discontinuities.
3. Characteristics of intraseasonal oscillations in SCSSM
a. Identifying intraseasonal oscillations
Following Chan et al. (2002), the area-averaged OLR over the SCS is chosen to be the basic index for identifying the intraseasonal variability. The anomaly of this index is defined as the departure from the mean averaged from 1 May to 31 July, and this anomaly time series is used for the wavelet analysis. The 27 cases between 1975 and 2002, except for 1978, are examined. The wavelet spectrum suggests that in most years, two main oscillations are present—one with a period of 30–60 days and another with a period of 10–20 days. The time series of the 30–60-day oscillation (hereafter the 3/6 mode) and 10–20-day oscillation (hereafter the 1/2 mode) are reconstructed by the inverse transform over scales between 30 and 60 days as well as between 10 and 20 days. However, the relative amplitudes of these two oscillations vary from year to year.
Figures 1– 3 give three examples that represent the different categories of the dominant modes. A typical year with a strong 3/6 mode is 1976 (Fig. 1). In Fig. 1b, larger values of wavelet spectral coefficients are concentrated within the Madden–Jullian oscillation (MJO) band of 30–60-days. The amplitude of the 3/6 mode is much larger than that of the 1/2 mode (Fig. 1a). Presented in Fig. 2 is the typical case with a strong 1/2 mode. In 1989, although the spectral values at the 30–60-day period are appreciable, they are much less than those at 10–20-day period (Fig. 2b). The amplitudes of the 1/2 mode are particularly pronounced (Fig. 2a). The 1979 the SCS summer monsoon was widely studied (see review in section 1), and could be considered as a case in which both the 3/6 and the 1/2 modes are significant. From Fig. 3b, two frequency bands can easily be discerned: 30–60 day and 10–20 day. The amplitudes of these two modes are also comparable (Fig. 3a). These results are similar to those reported in Chen and Chen (1995), who conducted power spectral analyses of the precipitation and 850-hPa zonal winds.
b. Interannual variability of the ISOs
As discussed above, different oscillations dominate in different years. Which mode is dominant in a particular year can be examined by its explained variance. The square of the correlation coefficient between the raw OLR anomaly and an ISO component denotes the ratio of the variance of this mode to the total variance. It is found that both the 3/6 mode and 1/2 mode are present, but their explained variance varies (Fig. 4). In some years, the 3/6 mode is more significant than the 1/2 mode, while in other years the 1/2 mode is dominant. A third situation also exists in which the explained variances by these two modes do not differ much; in other words, both the 3/6 mode and 1/2 mode are important in these years. For most years, these two modes together explain a large percentage of the total variance, which suggests that these two intraseasonal modes are the essential oscillations that control the behavior of the SCS SM activity.
The choice of a 10% difference is to ensure enough cases for each category. Based on this classification, the years can be grouped as follows:
3/6-mode category: 1975, 1976, 1985, 1988, 1996, 1997, 1998;
1/2-mode category: 1982, 1983, 1986, 1989, 1993, 1994, 1995, 2000;
dual-mode category: 1979, 1980, 1981, 1984, 1987, 1990, 1992, 1999, 2001, 2002.
The years 1977 and 1991 are not included in the classification, because in each of these 2 yr neither the 3/6 mode nor the 1/2 mode was active.
It is obvious from this classification and the result in Fig. 4 that there is strong year-to-year variability in ISO activity of the SCS SM. Chen and Chen (1995) suggested that the 10–20-day mode becomes a major agent in regulating the cycle of the Indian monsoon when the 30–60-day mode is weak. Our classification of the ISOs of the SCS SM indirectly verifies their speculation, at least, for the SCS SM. The essential characteristics of the ISOs for the different categories and the physical processes associated with their formation and propagation are discussed in the next section.
4. Structure and propagation of ISO
a. Compositing technique
The lower-tropospheric flow patterns associated with the 3/6 mode and 1/2 mode have been shown in case studies (e.g., Chen and Chen 1995; Chan et al. 2002). To capture the essential features in the configuration of these two modes, composite analyses are made.
The temporal variations of the filtered OLR shown in Figs. 1– 3 are much like a sinusoidal function. In general, an ISO cycle can be defined as one with a positive and negative anomaly (or an inactive and an active period), both of which must have a peak amplitude that is greater than one standard deviation from zero. Based on this selection criterion, 10 cycles are isolated for the 3/6-mode category, 23 cycles for the 1/2-mode category, and 12 cycles for the dual-mode category during the period of 1975–2002 (except 1978). Once the cycles are isolated, each cycle is broken into several different phases for compositing, as in Chan et al. (2002). That is, each active–inactive cycle is divided into nine phases (see Fig. 1). Phase 1 represents the transition from an active to an inactive period, while phase 3 is the peak of the inactive period. Phase 5 indicates the transition from the inactive to the active period, and phase 7 is the peak of the active period. Phases 2, 4, 6, and 8 occur at the time when the oscillation reaches half of its maximum or minimum value. Phase 9 is same as phase 1. Composites are then made of various meteorological quantities for those cases in which a particular mode is dominant.
Significance of the composite results at each grid point is estimated based on the Student’s t test (null hypothesis is that the composite value is not significantly different from zero). An area within which the t values are greater than the threshold values for significance (at the 95% level) suggests that the parameter associated with the 3/6 mode or 1/2 mode in this area is not likely due to random chance, and composite results are likely to be representative of the individual cases.
As mentioned in section 3b, each year falls into one of three categories. In the following, only the 3/6- and 1/2-mode categories are discussed. The evolution of the situation in the dual-mode category follows much like that for the 3/6-mode category, but is modified by the 1/2 mode. The processes are found to be similar to those in the 3/6-mode category and the patterns associated with the dual-mode category are not discussed.
b. 30–60-day mode
Composites of the 3/6-mode OLR and 850-hPa wind fields show that in phase 1, an east–west elongated convection zone (negative OLR anomalies) clearly exists over the northern SCS, the northern Bay of Bengal, and northwestern Pacific, accompanied by strong westerly or southwesterly anomalies (Fig. 5a). But to the south of this convection belt, an anomalous anticyclone covers the Tropics from the Indian Ocean to the western Pacific. In phase 2, convection, together with anomalous westerlies, migrates northward into southern China and its adjacent oceans to the east (Fig. 5b). At the same time, an anticyclone dominates much of the SCS and the Philippine Sea so that the convection is suppressed. By phase 3, the anomalous anticyclone over the SCS appears to have become more intense (Fig. 5c). Notice also that deep convection begins to develop along the equatorial area. In addition, the mei-yu–baiu front is well developed, with enhanced convection in a band extending from central China to southern Japan. Thus, an obvious cyclonic–anticyclonic–cyclonic circulation pattern is set up from the equator to the north, together with the alternating strong–weak–strong convection zones. This pattern is similar to the situation for the SCS monsoon break phase illustrated by Chen et al. (2000), who gave only the charts of the two extreme conditions based on the 850-hPa streamlines, superimposed with changes in the brightness temperature.
Such a flow pattern continues to migrate northward so that the anticyclone over the SCS weakens and convection over the southern part of the SCS is significantly enhanced (phase 4, Fig. 5d). By phase 5, the southern SCS is covered by deep convection (Fig. 5e). Further northward migration of this pattern results in an extensive cyclonic circulation, as well as enhanced convection over the SCS and the northern Philippines in phase 6 (Fig. 5f). Note also the westerly anomalies extending from the Bay of Bengal through Indochina to the Philippine Sea. During the most active phase (phase 7, Fig. 5g), an anomalous cyclone accompanied by strong convection is present over the SCS. The meridional juxtaposition of circulation and convection patterns is opposite to that in phase 3. Afterward, the SCS cyclonic circulation continues to shift northward, along with weakening convection (phase 8, Fig. 5h).
The most prominent feature in Fig. 5 is that the anomalous cyclones (anticyclones), along with enhanced (suppressed) convection, migrate northward from the equator to the midlatitudes, which is the so-called trough–ridge seesaw in which the monsoon trough and subtropical ridge exist alternatively over the SCS, with the former providing an enhanced convection environment and the latter a suppressed one. Such a seesaw phenomenon manifests not only between two latitudinal locations, but also among three latitudinal locations, especially during the peak phases, suggesting the linkage between the Tropics and extratropics on a 30–60-day time scale.
The northward propagation of the 3/6 mode can also be seen from the phase–latitude cross section (Fig. 6a). Alternate regions of positive and negative relative vorticity that evidently originate from the equatorial region migrate northward with time. Propagations of the relative vorticity and OLR anomalies, however, are not apparent in the longitudinal direction (Fig. 6b). Even if westward propagation exists, its magnitude and speed are much smaller than those in Fig. 6a.
Chen and Chen (1995) suggested that, based on the 1979 summer, the low-level 3/6-mode monsoon trough and ridge were associated with the 3/6-mode global divergent circulation. Chan et al. (2002) also found that in 1998, the variations in the 850-hPa relative vorticity were in phase with those in the 200-hPa divergence on the 30–60-day time scale. It is, therefore, of interest to examine the evolutions of the 3/6-mode upper-level divergent circulations. A convergent center of the 3/6 mode is found to the southeast of Sri Lanka near the equator in phase 1 (Fig. 7a). Subsequently, this center propagates eastward to reach the SCS by phase 2 (Fig. 7b). The convergent circulation then starts to weaken in phase 3 (Fig. 7c), and the convergent center does not move much northeastward. By phase 4 (Fig. 7d), the convergent center almost dissipates. However, in phase 5, anomalous divergent easterlies occur over the equatorial Indian Ocean between 10°S and 10°N (Fig. 7e). By phase 6, except east of the SCS, divergent winds emanating from the SCS become very prominent (Fig. 7f). During the most active monsoon phase (phase 7), the divergent center is located over the eastern SCS and the Philippine Sea (Fig. 7g), which suggests that both the east–west and local Hadley circulations strengthen over the SCS region. Later, the divergent circulation weakens (phase 8, Fig. 7h).
It can be seen from Fig. 7 that the upper-level convergent center propagates eastward from the equatorial Indian Ocean through the SCS and the Philippines during the inactive monsoon period, while divergent winds extend eastward during the active period. As compared with the 850-hPa circulation shown in Fig. 5, the upper-level convergent (divergent) center well corresponds to the low-level anticyclonic (cyclonic) center over the SCS and the Philippines. These results demonstrate that the northward-migrating 3/6-mode monsoon trough–ridge in the lower troposphere is coupled with the eastward-propagating 3/6-mode divergence–convergence in the upper troposphere.
c. 10–20-day mode
Composite evolutions of the 10–20-day oscillation for the 1/2-mode category suggest that the migrations of the circulation and convection patterns are very different from those of the 3/6 mode. During phase 1, a strong cyclone that is accompanied by significant convection occurs over the Bay of Bengal and western SCS, with an anticyclone immediately to its east (Fig. 8a). Another weak cyclone is situated northeast of the anticyclone to the southeast of Japan. These three systems exhibit roughly a southwest–northeast orientation. The anticyclone then strengthens and its center moves to the Bashi Channel (∼20°N, 122°E), while the cyclonic circulation over the Bay of Bengal weakens almost completely, and convection over the western SCS is suppressed (phase 2, Fig. 8b). By phase 3 (the peak of the inactive period), the anomalous anticyclone over the SCS extends westward into the western part of the Indochina Peninsula (Fig. 8c). The strong southwesterlies with deep convection to the north of the SCS appear to be associated with the mei-yu front system. Also noteworthy is that the westerlies near the equator between 130° and 150°E begin to strengthen, and convection starts to extend northward. The anomalous anticyclone over the SCS continues to migrate further westward, and becomes the major zonal structure extending from the SCS to the Indian Peninsula (phase 4, Fig. 8d). More importantly, the equatorial westerlies east of 130°E strengthen further and are accompanied by enhanced convection. The relationship between the enhanced convection and the equatorial westerlies will be discussed further in section 5.
In phase 5, the westerly jet east of 130°E extends northward into the southern part of the Philippines, with convection enhanced over the eastern SCS and the Philippine Sea (Fig. 8e). A large cyclonic circulation can also be identified to east of the SCS based on the significant southerlies over the south of Japan, northeasterlies over eastern China, and northwesterlies over the SCS. This extended cyclonic circulation, though still quite weak, seems to result from a merger of the frontal system from central China and the system to the south that is associated with the westerly jet, which suggests that disturbances from the Tropics and extratropics could have impacts on the SCS 1/2 mode. This circulation soon develops into a well-organized cyclone, accompanied by enhanced convection over the eastern SCS (phase 6, Fig. 8f). Associated with the OLR anomalies in the vicinity of the Philippines is the strengthening of a westerly jet on the southern side of the cyclone. By phase 7 (the peak of the active period), the cyclone that is accompanied by strong convection evidently strengthens and starts to propagate westward so that the SCS monsoon now reaches its maximum state (Fig. 8g). Note also that the westerlies are now found farther west over the Bay of Bengal. The cyclone over the SCS then further extends westward, and convection around the eastern SCS begins to be suppressed, which indicates that the 1/2 mode weakens (phases 8, Fig. 8h). Note also the pronounced southerlies over the Bay of Bengal from the Southern Hemisphere. The double-low structure reported by Chen and Chen (1993) appears to be identified in this phase.
The sequence of events described above points to the different propagation characteristics of the 1/2 mode compared with that of the 3/6 mode. The anticyclonic flow over the SCS during the suppressed phase (phase 3) appears to propagate westward. The cyclonic flow that subsequently migrates into the SCS in phase 7 seems to come from, and is accompanied by strong convection over, the tropical western North Pacific (WNP). In other words, the 1/2 mode has a largely (north) westward propagation.
The westward propagation of the 1/2 mode is illustrated more clearly from the phase–longitude cross section (Fig. 9b). The positive relative vorticity region accompanied by negative OLR extends westward from phase 3, which suggests that the “genesis” of the 1/2 mode takes place at least east of 140°E. An analysis of the meridional migrations of the 1/2 mode (Fig. 9a) shows that, in addition to the northward propagation from the equator to the northern SCS, a perturbation from the north moves southward, and it merges with the convection coming from the southeast.
Composite evolutions of the 1/2-mode upper-level divergent circulation indicate clearly that a convergent center continuously migrates northwestward from east of the Philippines (Figs. 10a and 10b) so that by phase 3, it arrives at the SCS (Fig. 10c). It seems that in this phase a divergent center initiates around the equator east of 150°E. This divergent center becomes pronounced in phase 4 (Fig. 10d) and migrates northwestward (Figs. 10e and f) so that by phase 7, it is located in the SCS (Fig. 10g). In comparison with Fig. 8, it is obvious that for the 1/2 mode, a strong coupling exists between the upper-level convergence (divergence) and low-level anticyclone (cyclone), and these systems in the upper and lower troposphere all migrate northwestward. In particular, the upper-level divergent center well corresponds to the region where convection is more active, which suggests that dynamically, the upper-level divergence and its coupling with the low-level cyclonic circulation are favorable for the development of deep convection.
5. Mechanism responsible for the 10–20-day mode
Several studies (e.g., Krishnamurti and Ardanuy 1980; Chen and Chen 1993) have suggested that the 1/2- mode of the Indian monsoon originates over the western Pacific and the SCS. If this is true, where does the 10–20-day oscillation of the SCS SM originate? Many attempts have also been made to develop a theoretical framework to explain the propagation characteristics of the 30–60-day oscillation such as wave-conditional stability of the second king (CISK) mechanism (e.g., Lau and Peng 1987) and evaporation–wind feedbacks (Emanuel 1987; Neelin et al. 1987). What, then, is the mechanism responsible for the propagation of the 1/2 mode?
The composite evolutions presented in section 4c indicate that the 1/2 mode of the SCS SM appears to originate from the equatorial region and then propagates northwestward. The composite patterns shown in Fig. 8 resemble the observational case studies described in C96 and DM. That such a northwestward propagation is associated with a MRG wave was suggested by DM. Because the period of oscillation varies from year to year, it might be more meaningful to compare the results of these previous studies with the evolution in an individual year. A cycle in 1982 is chosen for comparison because the origin and propagation of the 1/2 mode for this case are quite distinct.
On 13 June, strong convection accompanied by strong westerlies from the west occurred near the equator east of 160°E, with weak easterlies near 5°N (Fig. 11a). On 15 June, the westerlies between 150° and 160°E became stronger than 2 days before (Fig. 11b), and a cyclonic circulation has developed to the northwest of the area of strongest convection, which is similar to what C96 and DM observed. Three days later (18 June, Fig. 11c), the oscillation of the SCS SM reached the most inactive period when the SCS and the western North Pacific were dominated by a huge anticyclone. However, an almost closed cyclone formed to southeast of the Philippines. The westerlies strengthened further and extended westward to reach 130°E, accompanied by deep convection in a band.
On 20 June, the cyclonic circulation intensified further and became a vortex with the trough line exhibiting a southwest–northeast orientation (Fig. 11d). Another prominent feature is the convergence between the southwesterlies of this vortex and the southeasterlies from the equator east of 150°E, leading to enhanced convection there. The vortex became the dominant system over the western Pacific by 22 June (Fig. 11e). It largely displaced the previous anticyclone in such a way that deep convection occurred over the southeastern SCS. It was found that the westerly or southwesterly flow on the southern side of the vortex was more intense than the northeasterly flow on the northern side of the vortex. Two days later (25 June, Fig. 11g), the entire circulation pattern propagated westward and one of the vortex centers entered into the SCS, so that convection over the SCS reached its maximum. Afterward, only one vortex center existed by 27 June (Fig. 11h).
This cycle of the 10–20-day oscillation suggests that the vortex, which caused the SCS SM to be active, originated from the central equatorial Pacific in the transition phase of the inactive period, and then migrated northwestward into the SCS. This sequence of events, and the associated wind and convection patterns, is very similar to that described in both C96 and DM. It is, therefore, possible that the 1/2 mode is in a form of an MRG wave, as discussed in DM.
Webster (1983) suggested that quasi-biweekly oscillations, which might be related to the 40–50-day oscillations, could result from feedback between the hydrological cycle and the dynamics over the monsoon regions. A significant 30–60-day convection mode actually formed along the equator over 130°–160°E during the evolution of this 1/2 mode. Although the 3/6 mode over the SCS was weak, it was not weak over the equator (not shown). It is, therefore, possible that the convection associated with the 3/6 mode may contribute toward the development of the 1/2 mode through the excitation of an MRG wave (as discussed in DM), which might then be responsible for the 10–20-day oscillation of the SCS SM. All of the 23 cycles in the 1/2-mode category have been examined, and the case of 1982 shows the most distinct signal. For most of the other cycles, some general qualitative agreement to this pattern is also found.
6. Summary and discussion
The National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) reanalysis data are used to study the intraseasonal variability of the South China Sea (SCS) summer monsoon (SM). The OLR anomaly is selected as the reference parameter to identify the intraseasonal oscillations of the SCS SM. The objective of this paper is to explore the interannual variability and propagating mechanisms in the intraseasonal oscillations of the SCS SM by examining data from many years.
Wavelet analyses show that the 30–60-day and 10–20-day intraseasonal modes are the essential oscillations that control the behavior of the SCS SM activities. Both the 3/6 and 1/2 modes are distinct, but may not always be comparable in magnitude during one individual year. Interannual variability in ISO activity of the SCS SM exhibits three categories. In some years, the 3/6 mode is more significant than the 1/2 mode, while in other years the 1/2 mode is dominant. The third situation is that in some years both the 3/6 and 1/2 modes are pronounced.
Composite analyses based on the 3/6-mode category cases indicate that the 30–60-day oscillation of the SCS SM exhibits a trough–ridge seesaw in which the monsoon trough and subtropical ridge exist alternatively over the SCS. During a cycle, anomalous cyclones (anticyclones) along with enhanced (suppressed) convection migrate northward from the equator to the midlatitudes, with the cyclonic anomaly providing an enhanced convection environment and the anticyclonic anomaly a suppressed one. Such a seesaw phenomenon manifests not only between two latitudinal locations, but also among three latitudinal locations, especially during the peak phases, suggesting the linkage between the SCS SM and other systems over the Tropics and midlatitudes on a 30–60-day time scale. Composite results indicate that the northward-migrating 3/6-mode monsoon trough–ridge in the lower troposphere is linked with the eastward-propagating 3/6-mode divergence–convergence in the upper troposphere, probably through deep convection.
Composites of the 1/2-mode category show the 10–20-day oscillation as a westward-propagating anticyclone–cyclone system that is largely zonally oriented. A close coupling exists between the upper-level convergence (divergence) and low-level anticyclone (cyclone). Moreover, the upper-level divergent center well corresponds to the convectively active region, which suggests that dynamically, the upper-level divergence and the low-level cyclonic circulation are associated with deep convection. Noteworthy is that the 1/2 mode of the SCS SM originates over the equatorial Pacific east of 140°E. A disturbance from the northeast of the SCS also appears to contribute to the 1/2 mode. The case of 1982 is examined in detail to examine the possible propagation mechanism. It is found that the convection and flow patterns in this case resemble those associated with the propagation of a mixed Rossby–gravity wave found by C96 and DM. Of course, this may be only one of the possibilities in explaining the origin of the 10–20-day oscillation of the SCS SM, and is based on a qualitative comparison with previous observational studies. Kang and Kimura (2003) have proposed that meridionally propagating waves over the western North Pacific can also result from SST anomalies. Whether such anomalies can couple with the MJO (e.g., Watterson 2002) to trigger waves that propagate into the SCS need to be further ascertained. Numerical experiments, as well as further theoretical analyses, are necessary to identify the mechanisms further.
Acknowledgments
The NCEP–NCAR reanalysis data are provided by the NOAA–CIRES Climate Diagnostics Center, Boulder, Colorado, from their Web site (http://www.cdc.noaa.gov/). Comments from the three anonymous reviewers are much appreciated. This research is supported by City University of Hong Kong Grants 7001336 and 7010010.
REFERENCES
Chan, J. C. L., W. Ai, and J. Xu, 2002: Mechanisms responsible for the maintenance of the 1998 South China Sea summer monsoon. J. Meteor. Soc. Japan, 80 , 1103–1113.
Chang, C-P., J. M. Chen, P. A. Harr, and L. E. Carr, 1996: Northwestward-propagating wave patterns over the tropical western North Pacific during summer. Mon. Wea. Rev., 124 , 2245–2266.
Chen, T-C., and J-M. Chen, 1993: The 10–20-day mode of the 1979 Indian monsoon: Its relation with the time variation of monsoon rainfall. Mon. Wea. Rev., 121 , 2465–2482.
Chen, T-C., and J-M. Chen, 1995: An observational study of the South China Sea monsoon during the 1979 summer: Onset and life cycle. Mon. Wea. Rev., 123 , 2295–2318.
Chen, T-C., M-C. Yen, and M. Murakami, 1988: The water vapor transport associated with the 30–50 day oscillation over the Asian monsoon regions during 1979 summer. Mon. Wea. Rev., 116 , 1983–2002.
Chen, T-C., M-C. Yen, and S. P. Weng, 2000: Interaction between the summer monsoon in East Asia and the South China Sea: Intraseasonal monsoon modes. J. Atmos. Sci., 57 , 1373–1392.
Dickinson, M., and J. Molinari, 2002: Mixed Rossby–gravity waves and western Pacific tropical cyclogenesis. Part I: Synoptic evolution. J. Atmos. Sci., 59 , 2183–2196.
Emanuel, K. A., 1987: An air–sea interaction model of intraseasonal oscillations in the Tropics. J. Atmos. Sci., 44 , 2324–2340.
Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77 , 437–471.
Kang, S-D., and F. Kimura, 2003: Effect of tropical SST on the northwest Pacific subtropical anticyclone. Part I: Linear Rossby wave propagation. J. Meteor. Soc. Japan, 81 , 1225–1242.
Krishnamurti, T. N., and P. Ardanuy, 1980: The 10–20-day westward propagating mode and breaks in the monsoons. Tellus, 32 , 15–26.
Krishnamurti, T. N., and D. Subrahmanyam, 1982: The 30–50 day mode at 850 mb during MONEX. J. Atmos. Sci., 39 , 2088–2095.
Lau, K. M., and L. Peng, 1987: Origin of low-frequency (intraseasonal) oscillations in the tropical atmosphere. Part I: Basic theory. J. Atmos. Sci., 44 , 950–972.
Lau, K. M., and H. Weng, 1995: Climate signal detection using wavelet transform: How to make a time series sing. Bull. Amer. Meteor. Soc., 76 , 2391–2402.
Lau, K. M., and S. Yang, 1997: Climatology and interannual variability of the Southeast Asian summer monsoon. Adv. Atmos. Sci., 14 , 141–162.
Liebmann, B., and C. A. Smith, 1996: Description of a complete (interpolated) outgoing longwave radiation dataset. Bull. Amer. Meteor. Soc., 77 , 1275–1277.
Lorenc, A. C., 1984: The evolution of planetary-scale 200-mb divergent flow during the FGGE year. Quart. J. Roy. Meteor. Soc., 110 , 427–441.
Madden, R. A., and P. R. Julian, 1971: Detection of a 40–50 day oscillation in the zonal wind in the tropical Pacific. J. Atmos. Sci., 28 , 702–708.
Madden, R. A., and P. R. Julian, 1972: Description of global-scale circulation cells in the tropics with a 40–50 day period. J. Atmos. Sci., 29 , 1109–1123.
Murakami, T., and T. Nakazawa, 1984: On the 40–50 day oscillation during the 1979 Northern Hemisphere summer, Part I: Phase propagation. J. Meteor. Soc. Japan, 62 , 440–468.
Murakami, T., and T. Nakazawa, 1985: Tropical 45-day oscillations during the 1979 Northern Hemisphere summer. J. Atmos. Sci., 42 , 1107–1122.
Neelin, J. D., I. M. Held, and K. H. Cook, 1987: Evaporation–wind feedback and low-frequency variability in the tropical atmosphere. J. Atmos. Sci., 44 , 2341–2348.
Tao, S. Y., and L. X. Chen, 1987: A review of recent research on the East Asian summer monsoon in China. Monsoon Meteorology, C. P. Chang and T. N. Krishnamurti, Eds., Oxford University Press, 60–92.
Torrence, C., and G. P. Compo, 1998: A practical guide to wavelet analysis. Bull. Amer. Meteor. Soc., 79 , 61–78.
Waliser, D. E., N. E. Graham, and C. Gautier, 1993: Comparison of the highly reflective cloud and outgoing longwave radiation datasets for use in estimating tropical deep convection. J. Climate, 6 , 331–353.
Watterson, I. G., 2002: The sensitivity of subannual and intraseasonal tropical variability to model ocean mixed layer depth. J. Geophys. Res., 107 .4020, doi:10.1029/2001JD000671.
Webster, P. J., 1983: Mechanics of monsoon low-frequency variability: Surface hydrological effects. J. Atmos. Sci., 40 , 2110–2124.
Wu, M. L. C., S. Schubert, and N. E. Huang, 1999: The development of the south Asian summer monsoon and the intraseasonal oscillation. J. Climate, 12 , 2054–2075.
Xu, J., and J. C. L. Chan, 2001: First transition of the Asian summer monsoon in 1998 and the effect of the Tibet-tropical ocean thermal contrast. J. Meteor. Soc. Japan, 79 , 241–253.
(a) Time series of 1976 OLR anomalies (W m−2) over the SCS (5°–20°N, 110°–120°E) for the period 1 May to 31 Jul. The solid line indicates raw OLR anomalies, the thick dashed line is the 30–60-day oscillation, and the thin dashed line is the 10–20-day oscillation. Numbers 1, 3, 5, 7, and 9 represent the phases of the 30–60-day mode (see section 4a for details). (b) Wavelet spectrum of raw OLR time series. The contours denote the wavelet spectral coefficient. Thick long-dashed line indicates the cone of influence outside of which edge effects become important.
Citation: Journal of Climate 18, 13; 10.1175/JCLI3395.1
Explained variance (percent of the total variance) by the 30–60-day mode (solid line) and the 10–20-day mode (dashed line) during 1975–2002.
Citation: Journal of Climate 18, 13; 10.1175/JCLI3395.1
Composite evolutions of the 30–60-day-filtered 850-hPa winds (vectors, m s−1) and OLR anomalies (shading, W m−2) during an ISO cycle for the 3/6-mode category. (a)–(h) Phases 1–8 are displayed, respectively. Negative OLR anomalies are shaded, light shading represents the OLR values between 0 and −10, and heavy shading represents OLR values less than −10. Open circles indicate grid points where the wind anomalies are significantly different from zero at the 95% level (based on the Student’s t test) in at least one of the wind components (zonal or meridional).
Citation: Journal of Climate 18, 13; 10.1175/JCLI3395.1
(a) Phase–latitude (averaged between 110° and 120°E) and (b) phase–longitude (averaged between 5° and 20°N) cross sections of composite 30–60-day-filtered 850-hPa relative vorticity (contours, 10−6 s−1) and OLR (shading, W m−2). OLR shading is the same as in Fig. 5.
Citation: Journal of Climate 18, 13; 10.1175/JCLI3395.1
Composite evolutions of the 30–60-day-filtered upper-level (indicated by σ surface with σ = 0.2101) velocity potential (contours, 105 m2 s−1) and divergent winds (vectors, m s−1) during an ISO cycle for the 3/6-mode category. (a)–(h) Phases 1–8 are displayed, respectively. Open circles indicate grid points where the wind anomalies are significantly different from zero at the 95% level (based on the Student’s t test).
Citation: Journal of Climate 18, 13; 10.1175/JCLI3395.1
As in Fig. 5, except for the 1/2-mode category. (a)–(h) The eight phases correspond to those of the 10–20-day mode.
Citation: Journal of Climate 18, 13; 10.1175/JCLI3395.1
As in Fig. 6, except for the 1/2-mode category. (a)–(h) The eight phases correspond to those of the 10–20-day mode.
Citation: Journal of Climate 18, 13; 10.1175/JCLI3395.1
As in Fig. 7, except for the 1/2-mode category. (a)–(h) The eight phases correspond to those of the 10–20-day mode.
Citation: Journal of Climate 18, 13; 10.1175/JCLI3395.1
As in Fig. 5, except for a 10–20-day oscillation cycle of 1982 from 13 to 27 Jun. Note that all of the fields have been passed through a 10–20-day filter, and the domain is extended eastward to the date line to identify the origin of this oscillation.
Citation: Journal of Climate 18, 13; 10.1175/JCLI3395.1
