This study reveals some unique features associated with the impact of Typhoon Chanchu on the South China Sea (SCS) summer monsoon (SCSSM) onset in 2006. With Typhoon Chanchu entering the SCS in mid-May, southwesterlies were induced over the SCS and its upstream region due to a thermally forced Rossby wave response, leading to a reversal of low-level winds and a positive meridional temperature gradient (MTG) in the upper troposphere over the SCS. Typhoon Chanchu thus acted as an immediate trigger for the SCSSM onset, and the onset date was suggested to occur in the third pentad of May. After the typhoon landed in the coast of southern China, it still contributed to the establishment of the SCSSM due to warming the air column over the northern SCS, resulting in a persistent positive MTG with the ridge surface tilting northward.
 The South China Sea (SCS) summer monsoon (SCSSM) onset indicates the advent of the rainy season over East Asia [Tao and Chen, 1987], thus various indices have been proposed to define the SCSSM onset [e.g., Wang and Wu, 1997; Mao et al., 2004]. For example, Wang et al.  suggested the area-averaged 850-hPa zonal winds over the central SCS (5°–15°N, 110°–120°E) as a concise yet pertinent index to identify the monsoon onset date for an individual year, but in a few years the onset dates can not be readily determined due to vigorous intraseasonal oscillations or flat transitions. It is well known that a stronger typhoon Chanchu, which was the first tropical cyclone to form over the western North Pacific in 2006, entered the SCS on 13 May and landed in the coast of southern China on 18 May. It was also the worst and most intense typhoon on record to enter the SCS in May. After the decay of such a typhoon, the low-level winds over the SCS exhibited somewhat submonthly fluctuations with vague-onset in this year. Therefore, it is necessary for such a special case to conduct an examination to determine its onset date and to identify the roles of the typhoon played in this onset.
 Many triggering mechanisms have been proposed in terms of synoptic disturbances and intraseasonal events. Chang and Chen  found that the equatorward intrusion of a midlatitude front led to the SCSSM onset. A similar situation occurred during the 1998 SCSSM onset [e.g., Chan et al., 2000]. Such a mechanism was supported by the numerical modeling of Liu et al. , who suggested that deep convective heating over the Bay of Bengal produces a band of westerlies and condensation heating along the coast of southern China in association with a Rossby wave train, bringing cold air intrusions to the SCS. Of course, the mechanical and thermal forcing of the Tibetan Plateau should be the basic mechanism of the seasonal transition from the winter to the summer monsoon since its sensible heating eventually leads to a reversal of the atmospheric meridional temperature gradient (MTG) [e.g., Li and Yanai, 1996; Wu and Zhang, 1998]. Lau et al.  and Zhou and Chan  noted that the Madden-Julian Oscillation (MJO) [Madden and Julian, 1972] can trigger the SCSSM onset. Krishnamurti et al.  suggested the establishment of the “onset vortex” signaling the Indian summer monsoon onset. Here we mainly identify the impact of Typhoon Chanchu as a trigger on the 2006 SCSSM onset, which helps us to reveal a triggering mechanism, not recognized so far, responsible for the SCSSM onset.
2. Data and Methods
 The primary circulation data used in this study are extracted from daily European Centre for Medium-Range Weather Forecasts (ECMWF) operational analysis products [ECMWF, 1995]. Also used are daily NCEP/NCAR reanalysis data [Kalnay et al., 1996]. The Climate Prediction Center (CPC) Merged Analysis of Precipitation (CMAP) [Xie and Arkin, 1997] is used to describe the monsoonal rainfall since it merges multisource estimates with the uncertainties in each individual estimate reduced significantly. Due to CMAP rainfall available in pentads, the pentad means for the circulation variables are then constructed from the daily means. Daily outgoing longwave radiation (OLR) as a commonly used proxy of tropical convective activity is employed to reflect propagations of MJO and the thermally induced Rossby wave [Gill, 1980].
 As suggested by Wang et al. , the corresponding principal components (PCs) of the dominant empirical orthogonal function (EOF) modes can be used as an auxiliary index to identify the SCSSM onset date. Thus the EOF analysis for the 850-hPa zonal winds and the multivariable EOF analysis for the CMAP rainfall and 850-hPa winds are performed over the East Asian summer monsoon (EASM) domain (0°–40°N, 100°–140°E), with length of all variable fields from the second pentad of April to the first pentad of July, as was done by Wang et al. .
3. Identifying the SCSSM Onset
 The onset of the SCSSM is signaled by a large-scale wind shift in the lower troposphere from easterly to westerly and outburst of deep convection over the SCS [e.g., Ding and Liu, 2001]. In 2006, southwesterlies between 5°–15°N were firstly observed from the third pentad of May (Figures 1a and 1b), with heavy precipitation above 6 mm day−1 (Figure 1c). Note that the maximum precipitation rate even exceeded 30 mm day−1. Such an intense rainfall was caused by Typhoon Chanchu. Subsequently, two breaks of southwesterlies occurred respectively around the fifth pentad of May and the fifth pentad of June. Such a fluctuation can be seen from the area-averaged zonal winds over the SCS, with different positive durations during June in Figures 1a and 1b, indicating that the NCEP/NCAR data may underestimate 850-hPa winds, thus we mainly use the ECMWF data in this study. Based on the definition of Wang et al. , Figure 1b shows that the 2006 SCSSM onset date might be the sixth pentad of June because both previously positive durations did not keep three consecutive pentads, while Figure 1a indicates that the date should be the first pentad of June. Since it could not be readily identified using a single zonal wind index, other essential indices associated with large-scale circulation have to be examined to understand the onset process.
 The first mode EOF 1 of the single zonal wind field for the transitional period in 2006 explained 30.8% of total variance, and its spatial pattern exhibited a tripole structure with the negative wind anomalies over southern China being out of phase with those over the SCS and Yangtze Basin (not shown). The PC1 changed its sign for the first time from negative to positive on the second pentad of May, keeping positive anomalies for three pentads. The second mode EOF2 showed the westerly zone extending northward to the north of 20°N, with positive PC2 values from the six pentad of May to the third pentad of June, representing an intraseasonal northward shift of the EASM. Similar results were obtained from the multivariate EOF analysis. The sign reversals of the PC1 and the area-averaged zonal wind over the SCS (Figure 1a) were almost simultaneous, indicating that the 2006 SCSSM was established concurrently with the EASM commencement.
 Since the vertical tilt of the ridge surface of the subtropical anticyclone depends on the horizontal MTG under the constraint of the thermal wind relation, Mao et al.  proposed the area-averaged upper tropospheric (200–500 hPa) MTG near the ridge surface as an index to measure the SCSSM onset. The MTG evolutions (Figures 1a and 1b) show that it basically kept positive since the second pentad of May, though the 850-hPa zonal wind was sometimes negative, signaling that a complete switch over the circulation from the winter to the summer pattern had occurred since that pentad. This feature will be further discussed based on the ridge surface (see Figure 3).
 To ascertain the 2006 SCSSM onset date, Figure 2 displays the evolutions of 850-hPa wind fields from 1–5 May to 26–30 May. On the first pentad of May, the SCS was dominated by the western Pacific subtropical anticyclone with the ridgeline along 20°N, while Typhoon Chanchu was observed east of 140°E and slightly north of the equator. At this time, no significant cross-equatorial flows were found over the upstream areas except for the equatorial Indian Ocean westerlies between 70°–90°E. By the second pentad, Chanchu migrated northwestward into the Philippine Sea, with its center around 7.5°N and 133°E. But the SCS was still under the influence of the subtropical anticyclone. With Chanchu entering the SCS on the third pentad, the subtropical anticyclone retreated to the east of 120°E, with its ridgeline shifting to the north of 25°N. Meanwhile, the Somali cross-equatorial flows and the equatorial Indian Ocean westerlies started to accelerate and tended to converge toward the SCS, with the entire flow pattern exhibiting a typical SCSSM signature. When Chanchu reached northern SCS (forth pentad), the southwesterlies over the Indian Ocean strengthened significantly and penetrated the Indochina Peninsula into the SCS, with heavy rainfall occurring over the eastern Bay of Bengal. Such strong southwesterlies could be attributed to a Rossby wave response to the off-equatorial convective heating over the SCS [Gill, 1980]. Also observed were enhanced cross-equatorial flows west of 110°E, which was also a manifestation of the atmospheric response to this off-equatorial heat source [Gill, 1980].
 The typhoon continued to migrate northward, and by the fifth pentad of May it landed in the coast of southern China so that the subtropical anticyclone extended southwestward. The westerlies over the SCS were thus replaced transitorily by the easterlies and rainfall was suppressed. Note that significant rainfall resulted from Chanchu over the coast of southern China may be important for the establishment and maintenance of the SCSSM. Such a condensational heating over the northern SCS, based on the model simulation [Liu et al., 2002], could cause the overturning of the MTG over the SCS (as discussed below). Again, the southwesterlies dominated the SCS in subsequent pentads.
 As suggested by Mao et al. , one of the significant changes in circulation during the monsoon onset is in the configuration of the subtropical anticyclone. Its ridge surface tilts southward with increasing height before the onset, but northward after the onset. From Figure 3, it is found that before the SCSSM onset (first pentad), the ridgelines at different isobaric surfaces distributed southward from the lower troposphere to the upper troposphere over the SCS and its east, indicating a southward tilt of the ridge surface, which represents a typical winter monsoon pattern. However, the opposite situation occurred over the Bay of Bengal and Indochina Peninsula, implying that the summer monsoon onset might have taken place there. Although the ridgeline at 850 hPa still dominated the Indochina Peninsula and the SCS in the second pentad of May, the northward tilt of the ridge surface occurred over the SCS, suggesting that the vertical shear of zonal winds had started overturning over the SCS, which was consistent with the seasonal transition of large-scale circulation pattern in the lower troposphere derived from EOF analyses. By the third pentad, the ridge surface was moved to the coast of southern China due to Chanchu entering the SCS. It even broke up totally over the SCS in the forth pentad. After the typhoon made landfall (fifth pentad), it released lots of latent heating and warmed air column over the northern SCS, the MTG thus remained positive over the SCS. Although the ridgeline at 850 hPa intruded into the SCS again, the summer monsoon pattern with the ridge surface tilting northward had been truly established over the SCS. Subsequently, such a typical summer pattern became more significant, which can also be seen from Figures 1a and 1b. Considering the rainfall condition and the MTG persistence, the 2006 SCSSM onset was suggested to occur in the third pentad of May, though short breaks of low-level winds took place in the fifth pentad of both May and June. After all, persistent summer circulation pattern in the MTG and vertical shear of zonal wind in the troposphere reflect the essential features of the summer monsoon establishment.
4. Impacts of Typhoon Chanchu on the SCSSM Onset
 The above analyses show that Typhoon Chanchu migrated northwestward from the equatorial western Pacific into the SCS in the third pentad, leading to the SCSSM onset. Thus the typhoon was an immediate trigger of this onset. Such a triggering mechanism appears to be associated with the westward propagation of thermally forced Rossby wave [Gill, 1980], and further the genesis of the typhoon may be initiated by the eastward passage of the equatorial MJO. These are evident in the time-longitude sections of OLR. From the beginning of May, a deep convection episode (with OLR values less than 220 W m−2) was observed to propagate westward (Figure 4a), indicating a Rossby wave. Actually, the westward migration of the typhoon may partly be a manifestation of this Rossby wave before the typhoon made landfall. After the monsoon onset, the upstream southwesterlies and deep convection became more significant. In the entire episode, the thermally forced Rossby wave response [Gill, 1980] may play an important role. As shown in Figure 2, in early May the typhoon located east of 140°E and around the equator. It was at this time that an equatorial MJO arrived there (Figure 4b), indicating a connection between the development of the typhoon and the triggering of MJO. Maloney and Hartmann  suggested that low-level westerly anomalies associated with MJO favor the tropical cyclone formation due to barotropic eddy kinetic energy conversion from the mean flow.
 After the typhoon landed in the coast of southern China, on the other hand, the release of deep condensation heating in the troposphere warmed the air column over the northern SCS, leading to a positive MTG, and hence a northward tilt of the ridge surface over the SCS. This effect is similar to that of deep convection over the northern SCS induced by the Bay of Bengal convection [Liu et al., 2002]. Therefore, the typhoon still contributed to the maintenance of the SCSSM.
5. Summary and Discussion
 This paper examines the influences of the typhoon on the SCSSM onset. The best case of such a phenomenon was the development of the summer monsoon in 2006. Some unique features appeared during the SCSSM onset, with Typhoon Chanchu entering the SCS in mid-May and with submonthly fluctuations in zonal winds, indicating a vogue-onset. Thus other essential indices associated with large-scale circulation were investigated to identify the 2006 SCSSM onset date because it could not be determined readily based on a single 850-hPa zonal wind index. The EOF analyses for the low-level winds and rainfall show that the seasonal transition of large-scale circulation pattern occurred in the second pentad of May, with the MTG in the upper troposphere changing simultaneously its sign from negative to positive over the SCS. Such a slightly early MTG reversal might favor the development of deep convection, thus the typhoon went into the SCS in the third pentad, with the southwesterlies being induced over the SCS and its upstream region due to a thermally forced Rossby wave response, leading to a reversal of low-level winds and a positive MTG over the SCS, thereby forming a typical SCSSM circulation pattern with the ridge surface tilting northward. Therefore, Typhoon Chanchu acted as an immediate trigger for the SCSSM onset, and the onset date was suggested to occur in the third pentad of May. Such a triggering process was somehow different from the “onset vortex” accompanied with the Indian monsoon onset in that the onset vortex results largely from the barotropic instability of the basic flow [Mak and Kao, 1982]. After Chanchu landed in the coast of south China, on the other hand, latent heating released from the typhoon was conducive to a persistent positive MTG over the SCS, suggesting another impact of the typhoon on the establishment and maintenance of the SCSSM. In the normal onset case, significant southwesterlies usually exist over the northern Indian Ocean before the onset, thus the onset is completed by stronger heating over the land surface north of the SCS. But in the case of 2006, the full development of the significant southwesterlies concurred with the typhoon intrusion. The generation of the typhoon was found to be linked with the triggering of MJO. Such a connection needs to be further studied.
 ECMWF data used in this study have been obtained from the ECMWF data server. NCEP reanalysis data have been obtained from NOAA-CIRES Climate Diagnostics Center, Boulder, Colorado. This research was supported by National Basic Research program of China (grants 2006CB403603 and 2005CB42204) and Natural Science Foundation of China grants 40523001 and 40221503.