Temporal and Spatial Evolution of the Asian Summer Monsoon in the Seasonal Cycle of Synoptic Fields

Transcription

1 3630 JOURNAL OF CLIMATE Temporal and Spatial Evolution of the Asian Summer Monsoon in the Seasonal Cycle of Synoptic Fields YOUNG-KWON LIM AND KWANG-YUL KIM Department of Meteorology, The Florida State University, Tallahassee, Florida HEE-SANG LEE School of Earth and Environmental Sciences, Seoul National University, Seoul, South Korea (Manuscript received 13 December 2001, in final form 8 July 2002) ABSTRACT The principal mode of the seasonal variation of the Asian summer monsoon (ASM) and the temporal and spatial evolution of the corresponding synoptic fields are investigated via cyclostationary EOF analysis. This study uses the 21-yr ( ) Xie Arkin precipitation pentad data and National Centers for Environmental Prediction daily reanalysis data focusing on the period 21 May to 28 August, which covers the prominent life cycle of the ASM. The first mode, representing the seasonal cycle, explains about 20% 40% of the total variability in the parameters considered in the study. The pronounced feature in the present study is that the seasonal evolution of the sea level pressure anomaly contrast between the Asian continent and the surrounding oceans is the governing mechanism for the ASM evolution. The northward migration of the low pressure anomaly from the Indian Ocean and the moisture transport by a low-level westerly (Somali jet) toward India describes the evolution of the precipitation field over India, the Bay of Bengal, and the Indochina peninsula. Over the Pacific Ocean, a high pressure anomaly to the south of the precipitation band between 25 and 40 N pushes it northward, characterizing the onset and evolution of regional monsoons (mei-yu, baiu, and changma) in east Asia. It is shown that the high pressure anomaly intruding into southern China and farther west to the Bay of Bengal from mid-june to mid-july accelerates the low-level wind along the coastal line of Asia, where the pressure gradient becomes maximum. It provides a favorable condition for moisture transport toward the east Asian monsoon region. Two major moisture sources, the Indian Ocean and the western Pacific Ocean, are found during the monsoon period. Evolutions of anomalous moisture transport patterns toward India, the Indochina peninsula, and east Asia show unique characteristics from one another. While precipitation over the Indian region is persistently affected by the Indian Ocean, the Indochina peninsula is dominated by the western Pacific Ocean in the early monsoon period followed by the effect of the Indian Ocean (mid June to mid July) and the cooperation of the two sources afterward. The east Asian monsoon regions are influenced by both sources from the onset of their monsoons. After late July, the latter source affects the east Asian monsoon more due to the northward movement of a high pressure anomaly (late July) together with the development of a low pressure anomaly over the subtropical western Pacific region. 1. Introduction The monsoon is a global circulation system (Krishnamurti 1985; Lau et al. 1988) caused by the sea land contrast of sea level pressure. Monsoons dominate the three-dimensional large-scale circulation over many regions (Subbaramayya and Ramanadham 1981; Hidore and Oliver 1993). The Asian summer monsoon (ASM) is the most important one and defines the rainy season of the enormous area ranging from India to northeast Asia. It can be classified into regional monsoons based Corresponding author address: Kwang-Yul Kim, Department of Meteorology, 420 Love Bldg., The Florida State University, Tallahassee, FL kkim@met.fsu.edu on their occurrence locations: Indian monsoon, southeast Asian monsoon, South China Sea monsoon, the mei-yu in central China, the baiu in Japan, and the changma in Korea (Lau et al. 1988; Kang et al. 1999). A typical monsoon circulation in the Indian sector is subsequently followed by complex circulation changes in the surrounding Asian countries. Many countries located in south Asia, the Bay of Bengal, the Indochina peninsula, and east Asia are influenced by the ASM (Lau and Yang 1996). The ASM is closely related to well-known large-scale circulation structures (Matsumoto 1992; Nagazawa 1992; Murakami and Matsumoto 1994; Lau and Li 1984; Tanaka 1992; Ueda et al. 1995; Lau and Yang 1996; Li and Yanai 1996). In the lower level, the circulation over the ASM region is controlled by the cross American Meteorological Society

2 15 DECEMBER 2002 LIM ET AL equatorial flow, the low-level jet including the Somali jet, the cyclone vortex over India, and the anticyclone over the western Pacific. In the upper troposphere, the Tibetan anticyclone, the easterly jet over south Asia, and the westerly jet north of the Tibetan anticyclone are the primary characteristics (Krishnamurti 1971; Lau and Li 1984; Krishnamurti 1985). According to Krishnamurti (1985), seasonal variation of monsoon rainfall is consistent with these planetary-scale circulations and the migration of convective zones. In addition to understanding the fundamental circulation characteristics of the ASM, it is important to describe the evolution of the monsoon from its onset to depression. The onset of the South China Sea monsoon during mid-may is considered the beginning of the ASM (Tao and Chen 1987; Chang and Chen 1995; Hsu et al. 1999). The Indian monsoon starts with the migration of the heating region from northern Burma in May to the foothills of the Himalayas (eastern Tibet) in June. This transition between the two monsoons can be seen in locations of the heat-induced climatological upper anticyclone, which makes a northward motion from northern Malaysia in May toward the Tibetan plateau (Krishnamurti 1985). Then, the corresponding low-level circulation in south Asia and the diabatic heating over the eastern Tibetan plateau are developing. Yanai et al. (1992) and Hsu et al. (1999) showed that this diabatic heating is responsible for tropospheric warming over the eastern Tibetan plateau. In particular, the seasonal variation of diabatic heating of that region influences the evolution of the ASM (Yanai et al. 1992). Hoskins and Rodwell (1995) investigated the effects of diabatic heating and mountains in the ASM circulation using a time-dependent primitive equation model. After the early stage of ASM, its evolution can be described by the south Asian monsoon and the east Asian monsoon. The former includes the Indian monsoon and southeast Asian monsoon, whereas the latter includes the South China Sea monsoon, mei-yu (China), baiu (Japan), and changma (Korea). Although the linkage of the monsoon regions over India and over east Asia is clearly seen in the zonally elongated outgoing longwave radiation pattern (Lau and Chan 1986; Lau et al. 1988), there is also difference in detailed evolving features between the Indian monsoon and the east Asian monsoon. While the Indian monsoon is heralded by southwesterly low-level moist flow such as the Somali jet from the Arabian Sea, the east ASM is influenced by the subtropical western Pacific high (SWPH). The movement of the major rainband associated with the east Asian monsoon is related to the variation in the SWPH (Lau et al. 1988; Huang and Sun 1992; Murakami and Matsumoto 1994; Kang et al. 1999). The Indian monsoon also appears to be related to the tropical Pacific and the tropospheric biennial oscillation (Meehl 1987, 1997). The current study uses independent computational modes to document the seasonal evolution of the ASM from its onset until the depression. These computational modes will be extracted from key physical variables such that they are physically consistent, thereby facilitating an understanding of the physical interactions among the different variables involved in the ASM evolution. To this end, a cyclostationary EOF (CSEOF) analysis technique is employed (Kim and North 1997). This technique is useful in extracting physically evolving spatial patterns (Kim and Chung 2000). This study primarily focuses on the seasonal cycle of the ASM, the most conspicuous component of the ASM seasonal evolution. It is hoped that results of the analysis clarify key features of the large-scale ASM evolution and the relations among the regional monsoons. The data employed in this study are described in section 2, followed by a brief description of cyclostationary EOF technique in section 3. Section 4 describes the seasonal cycle of the ASM in the precipitation and the sea level pressure fields. The seasonal cycle of the lowlevel circulation structures and the moisture transport is discussed in section 5, followed by the upper-level circulation structures in section 6. Concluding remarks and the summary are presented in section Data Data used in this study are Xie Arkin precipitation pentad data (Xie and Arkin 1996) and daily sea level pressure, low-level (850 hpa) wind, upper-level (200 hpa) wind, and specific humidity archived at the National Centers for Environmental Prediction (NCEP; Kalnay et al. 1996). The Xie Arkin precipitation and NCEP dataset in the present study extend from 1979 to Only the data for the period of late May to late August were employed, thereby covering the prominent ASM evolution cycle. A 5-day mean dataset was constructed by performing nonoverlapping 5-day average of the daily data. The data is on a latitude longitude grid. 3. Method of analysis The CSEOF analysis used here is described in detail by Kim and North (1997). In CSEOF analysis, a space time datum is represented as T(rˆ, t) T (t)b (rˆ, t), (1) n n where B n (rˆ, t) are cyclostationary loading vectors (CSLVs) and T(t) are their corresponding principal component (PC) time series. While the eigenfunctions obtained from regular EOF analysis do not have time dependence, CSLVs in (1) are time dependent. Further, CSLVs are periodic in time: B n(rˆ, t) B n (rˆ, t d), (2) where d is the nested period representing the inherent periodicity in the statistics. The CSLVs are periodic and n

3 3632 JOURNAL OF CLIMATE FIG. 1. The first mode PC time series of each variable during the analysis period. The temporal grid points of PCs are 420 (20 pentads 21 yr). Values of PC are normalized. time dependent because they are eigenfunctions of a periodic and time-dependent covariance statistics. The main motivation for employing this tool in the present study is to investigate the physical processes associated with the evolution of the ASM from observational data. The time dependence in CSLVs describes physical evolution as opposed to statistical (stochastic) evolution contained in PC time series. This technique is somewhat different from other conceptually similar tools such as extended EOFs or complex EOFs. A limited number of comparison tests can be found in Kim and Wu (1999). An important point when performing a CSEOF analysis is to determine the nested period, that is, the inherent period of covariance statistics. We assumed that physical processes associated with the ASM repeat every year although their strength may vary from one year to another. Based on the prominent ASM period of late May to late August, the nested period is determined to be 20 pentads from 21 May to 28 August (100 days). As a result, CSLVs describe physical processes evolving through the given period and repeating each year. On the other hand, PC time series describe stochastic undulation of physical processes at timescales longer than the nested period. The set of PC time series shown in Fig. 1 represents the normalized first PC of each variable analyzed in the present study. The number of temporal grid points of PCs is 420 (20 pentads 21 yr). It shows that all PCs undulate around 1, indicating not only the repetition of seasonal cycle every year but the interannual variation of the strength of the seasonal cycle. To find consistent patterns of other variables with the first PC of precipitation, the first 10 PCs of other variables are regressed onto the first PC of precipitation. For this regression, the precipitation is regarded as the predictand whereas other variables are regarded as the predictors. The regression coefficients and R 2 (correlation squared) for each variable are shown in Table 1. The structures shown in Fig. 3 and Figs. 5 through 8 in sections 4, 5, and 6 are the regression patterns of other variables physically consistent with those of the precipitation (Fig. 2). The time-dependent patterns in Figs. 2 8 are connected with each other by the dynamics of the ASM seasonal cycle. Results of CSEOF analyses indicate that the first mode of each variable represents the seasonal cycle, which is the dominant mode of variability. This mode explains between 18% and 39% of total variability in different parameters. As mentioned earlier, the remaining discussion mainly focuses on the seasonal cycle of the ASM. 4. Evolution of precipitation and sea level pressure anomaly fields Figures 2 and 3 show the first CSEOFs of the precipitation anomaly and the sea level pressure anomaly TABLE 1. Regression coefficients and R 2 values of each variables regressed on the first PC time series of the precipitation. The first 10 PC times series of each variable are used for regression. Mode number Sea level pressure Wind at 850 hpa Moisture transport Wind at 200 hpa Relative vorticity advection R

4 15 DECEMBER 2002 LIM ET AL FIG. 2. The seasonal cycle (first CSEOF) of the precipitation anomalies over the Asian region. The nested period is 20 pentads (100 days) from 21 May to 28 Aug. The unit of contour lines is mm day 1 ; positive values greater than 0.5 are heavily shaded and negative values less than 0.5 are lightly shaded.

5 3634 JOURNAL OF CLIMATE FIG. 3. The seasonal cycle of the sea level pressure anomalies over the Asian region. The contour interval is 0.4 hpa with dark shading for positive values greater than 0.4 and light shading for negative values less than 0.4.

6 15 DECEMBER 2002 LIM ET AL regressed onto the first mode of precipitation anomaly, respectively. The anomaly is defined with respect to the mean of the total record. The first mode, representing the seasonal cycle of the ASM, explains 18% and 27% of the total variance of precipitation and sea level pressure, respectively. The use of the 100-day nested period with the 5-day interval produces the 20 spatial patterns for each mode, which depict the temporal evolution of the precipitation and the sea level pressure anomaly fields for the same period (21 May 28 August). In the following figures, however, spatial patterns are shown at every other time step. The temporal evolution of the ASM as shown in the seasonal cycle is classified into three stages. a. Early stage of monsoon The ASM typically begins in mid-may (Tao and Chen 1987; Chang and Chen 1995; Hsu et al. 1999). By late May, positive rainfall anomalies are limited mainly to the western part of the Indochina peninsula, the Indian Ocean, and the western Pacific at around 30 N (Figs. 2a,b). During this period, a positive precipitation anomaly is not formed over most of the Asian continent and the tropical western Pacific (0 15 N), consistent with a large-scale high pressure anomaly system dominating over those regions (Figs. 3a,b). By early June, positive rainfall anomalies begin to develop west of the Indochina peninsula, the Bay of Bengal, and the Indian peninsula, marking the onset of the Indian monsoon (Figs. 2b,c). Seasonal cycle of precipitation anomaly averaged over India at this time shows a sharp increase, and the first peak appears during the fifth pentad (10 14 June) (Fig. 4a). This transition to the Indian monsoon is characterized by the development of the low-level cyclonic circulation anomaly (Figs. 5a,b), the upper-level anticyclone, and convective activity in south and southeast Asia (Hsu et al. 1999). The location of these positive precipitation anomalies is intimately related to the development and the northward progression of the low pressure anomaly over India and the Indochina peninsula (Figs. 3b d). This explains the wet period over the Indochina peninsula during the sixth through eighth pentads (15 29 June) as shown in Fig. 4b. The patterns associated with the rainfall anomalies in the early monsoon phase shown here have good agreement with the previous studies (Krishnamurti et al. 1981; Ju and Slingo 1995; Li and Yanai 1996; Hsu et al. 1999; Wu et al. 1999; Krishnamurthy and Shukla 2000). Two positive rainfall anomalies over the tropical western Pacific near the equator and over the midlatitude Pacific in Figs. 2a d are related respectively to the intertropical convergence zone (ITCZ) and to the high pressure anomaly system in the western Pacific. The low pressure anomaly system at N intrudes upon the two high pressure anomaly regions, the Okhotsk high and the SWPH (Figs. 3b d). Since May, the Okhotsk high, originating from the North Pacific FIG. 4. The time series of the area-averaged precipitation anomaly derived from the seasonal cycle mode over (a) India ( N, E), (b) the Indochina peninsula ( N, E), (c) China ( N, E), and (d) southern part of Korea ( N, E; solid) and Japan (30 35 N, E; dashed), respectively. The abscissa indicates the anomaly and the ordinate the 20 pentads starting from 21 May. ( 50 N), begins to diminish due to the expansion of the low pressure system (Figs. 3b d), allowing a favorable condition for precipitation over the east Asian countries. In fact, the evolution of the SWPH ( N) and the Okhotsk high ( 50 N) is an important factor in characterizing the early evolution of the monsoon system over east Asia as shown below. b. Middle stage of monsoon Monsoon in east Asia During the developing phase of the monsoon, the progression of the precipitation pattern appears strongly related to the evolution of sea level pressure. Two main features of the sea level pressure change during this stage are the temporal evolution of the high pressure anomaly over the western Pacific (SWPH), and the development of low pressure anomaly over the Asian continent and the Sea of Okhotsk. The first feature represents the northwestward movement of the high pressure anomaly over the western Pacific and is clearly depicted in Figs. 3d f. This high pressure anomaly represents the variation of the climatological mean large-scale anticyclonic flow over the western Pacific (Lau et al. 2000). The second feature, the low pressure anomaly over the Asian continent, is due to continental warming (Figs. 3d h) (Yanai et al. 1992; Li and Yanai 1996). These

7 3636 JOURNAL OF CLIMATE FIG. 5. The seasonal cycle of the lower-level (850 hpa) wind anomalies. The arrow scale is denoted below figures and values lower than 0.4 are excluded. The dark shading represents the region of positive vorticity and the light shading the region of negative vorticity.

8 15 DECEMBER 2002 LIM ET AL features strongly determine the shape of the further evolution of the ASM, particularly the east Asian monsoon, as will be described below. The particular manner in which sea level pressure anomaly evolves as briefly addressed above seems to be responsible for moisture availability in the far eastern Asian countries, resulting in the development of the east ASM. That is, the high pressure anomaly over the subtropical western Pacific induces the zonally elongated precipitation patterns to the north and south of the pressure anomaly pattern (Figs. 2d,e). That the formation of the zonally elongated precipitation field over east Asia and its evolution are associated with the SWPH anomaly was previously suggested by Kang et al. (1999). The migration of the SWPH anomaly toward Southeast Asia and China along with the development of the low pressure anomaly over the Asian continent appears to determine the pathway of moisture during the mature stage of the monsoon (Figs. 3d f, 6d f). Specifically, a strong pressure gradient along the coast of southeast China produces a strong southwesterly wind, which serves as a conduit for moisture transport from the tropical Pacific (Figs. 5d f, 6d f). The typical pressure anomaly configuration of the summer monsoon (high pressure over ocean and low pressure over land) also allows moisture transport to the eastern part of the Asian continent from the Indian Ocean (Figs. 3d f, 6d f). Therefore, it is considered that the change in the timing, position, and strength of these pressure anomaly patterns is an important factor of precipitation variability over east and south Asia. Note that the pathway of moisture is pushed to the north as the SWPH anomaly migrates northwestward, resulting in the northward migration of the midlatitude precipitation band (Figs. 2d f). This precipitation band determines the wettest period in east Asia as shown in Figs. 4c and 4d. The evolution of the monsoon system described above is accompanied by the gradual northward movement of the precipitation band in east Asia (mei-yu, baiu, and changma). The onset of mei-yu (in China), baiu (in Japan), and changma (in Korea) depends on the arrival of this precipitation band in each region (Figs. 2c e). The spatial patterns in Fig. 2 show that the precipitation band reaches the Yangtze River, China, in June; this indicates the onset of mei-yu, the Chinese summer monsoon (Tanaka 1992; Lau and Yang 1996; Lau et al. 1988; Liang and Wang 1998) (Fig. 2c). Much precipitation occurs during the onset period and the rainy period lasts until mid-july (12th pentad) as seen in Fig. 4c. The precipitation band arrives at Japan in mid-june (Fig. 2c) and the southern part of the Korean peninsula in late June (Fig. 2d), starting baiu (Ninomiya 1984; Nagazawa 1992; Murakami and Matsumoto 1994; Ueda et al. 1995) and changma (Kang et al. 1999), respectively (see the sixth through eighth pentads in Fig. 4d). This rainband over Korea and Japan is almost stagnant, causing a long period (about a month) of rainfall (Fig. 4d). These spatial patterns of the precipitation band and the sea level pressure anomaly suggest that the SWPH anomaly plays a crucial role of pushing the rainband northward (Figs. 2d f, 3d f). Thus, the maintenance and the northwestward intrusion of the SWPH anomaly are important mechanisms for the summertime rainfall in east Asia. A similar result can be also found in Figs. 5 and 6 in Kang et al. (1999). c. Later stage of monsoon The monsoon over east Asia undergoes its decaying phase after mid-july (Figs. 3g j). It is seen that a low pressure anomaly develops over the subtropical western Pacific whereas a high pressure anomaly intensifies northeast of the low pressure anomaly (Figs. 3g i). These two anomalies define a southwestward transport of moisture from the Pacific Ocean, which, on the one hand, results in increased precipitation near the South China Sea and the Indochina peninsula (Fig. 4b) and, on the other hand, terminates moisture transport from the equatorial Pacific into east Asia (Figs. 6g j). Nonetheless, precipitation persists over northeast Asia with a limited amount of moisture transport coming primarily from the midlatitude Pacific Ocean (Figs. 2g j, 6g j). With the low pressure anomaly over the Asian continent retreating northward as the Indian Ocean cools, moisture is transported farther inland over the Indian continent from the Indian Ocean compared to the transport before July (Figs. 6g,h). As a result, the precipitation region migrates farther inland (Figs. 2g,h). At the same time, precipitation decreases to the west of India (Figs. 2g,h, 3g,h) and over the Indian Ocean due to the northward progression of the high pressure anomaly. As the continent continues to cool, a weaker pressure gradient begins to form between the Indian continent and the Indian Ocean (Figs. 3i,j). This tends to limit the availability of moisture from the Indian Ocean to the Indian continent. As a result, rainfall over India weakens significantly as the high pressure anomaly begins to dominate the Asian continent (Figs. 2j and 3j). The seasonal variation of Indian rainfall shown here agrees well with the area-averaged rainfall time series over India in Fig. 9 of Vernekar and Ji (1999). The low pressure anomaly over the subtropical western Pacific is associated with a positive subtropical rainfall anomaly over the western Pacific throughout the rest of the summer (Figs. 2g j). This low pressure anomaly pattern becomes stronger until it moves away from the Asian continent in late August (Figs. 3g j). The invigoration of the subtropical Pacific low pressure anomaly and the onset of the high pressure anomaly over the Asian continent develop a well-defined pressure gradient along the southeast coast of China (Figs. 3i,j). This mechanism, which induces strong northeasterly moisture transport, continues to be an effective means for precipitation over the East China Sea and the South China Sea (Figs. 2i,j; 5i,j; 6i,j). As the subtropical Pacific low pressure anomaly reaches east Asia, the sharp

9 3638 JOURNAL OF CLIMATE FIG. 6. The seasonal cycle of the moisture transport by the lower-level (850 hpa) wind anomalies and the corresponding moisture convergence field. The arrow scale ( 10 9 ) is denoted below figures and values lower than are excluded. The dark shading indicates the convergence of moisture greater than kg kg 1 s 1 and the light shading the divergence of moisture less than kg kg 1 s 1.

10 15 DECEMBER 2002 LIM ET AL pressure gradient moves slightly north, resulting in the secondary period of summertime rainfall for the southern part of Japan and Korea (Figs. 2j and 3j). The time series of the area-averaged precipitation anomaly over the Korean peninsula and Japan derived from the seasonal cycle mode shows that the second wet period appears from late August especially over the southern part of the Korean peninsula and Japan (Fig. 4d). 5. Characteristic features at lower level a. Horizontal circulation at 850-hPa level The first CSEOF (Fig. 5) of the low-level (850 hpa) wind, which represents the seasonal cycle, explains 24% of the total variance. The circulation field at 850-hPa level appears intimately related to the sea level pressure anomaly in Fig. 3. The geostrophic relationship is rather obvious in the intercomparison of Figs. 3 and 5. A resulting strong wind, the so-called low-level jet, develops along the coastal line of the Asian continent during the middle stage of the ASM. This low-level jet is the result of the strong pressure gradient between the continent and the ocean and the western Pacific pressure anomaly patterns. The jet is instrumental in producing moisture transport from the oceans to the interior of the continent, becoming conspicuous during the onset stages of the regional monsoons (Figs. 5b d). One should note also that the precipitation anomaly patterns in Fig. 2 are remarkably similar to the vorticity patterns of the lowlevel jet anomaly. Thus, the low-level jet is an important agent of moisture transport and precipitation (Ju and Slingo 1995; Li and Yanai 1996; Lau et al. 2000) The early stage of the Indian monsoon (Figs. 5a c) is characterized by the cyclonic circulation over the western Indian Ocean, which has been noted by Krishnamurti (1985). A weak cyclonic vortex is also found west of the Indochina peninsula. Precipitation is associated with these cyclonic vortices. They migrate northward as the low-level westerlies strengthen due to the meridional pressure difference between the Indian Ocean and the continent (Figs. 5d f; Vernekar and Ji 1999). The cyclonic vortices help the low-level jet transport moisture eastward along the coast and northward into the Asian continent (Figs. 5d f, 6d f). This characteristic pattern continues until mid-july when the westerly wind anomaly becomes confined to India and the Bay of Bengal (Figs. 5g i). This confinement of the low-level jet is related to the development of the subtropical western Pacific low pressure anomaly (Figs. 3g i), implying that the low-level inflow of moisture from the Indian Ocean to east Asia weakens significantly at this time. After early August, the southwesterly wind anomaly along the southern coast of China is replaced by the strong northeasterly wind anomaly as the low (high) pressure anomaly develops over the Pacific Ocean (continent). Moisture for the secondary precipitation over the southern parts of Korea and Japan during late August is primarily transported from the midlatitude western Pacific due to this configuration of the pressure anomalies. Another important feature in Fig. 5 is the evolution of circulation in the subtropical Pacific; this evolution affects the east ASM (Huang and Sun 1992). Due to the SWPH, anticyclonic flow develops near 15 N (Fig. 5a). Precipitation bands form along the northern flanks of this anticyclone, where the low-level wind converges (Figs. 2a,b, 5a,b). This anticyclone moves northward to about N and also westward toward east Asia until mid-july (Figs. 5a f). Strong northeastward wind along the western flank of the anticyclone is an important source of moisture over east Asia. As discussed above, moisture transport from the Indian Ocean remains an important source for the east Asian countries until mid-july. Moisture transport from the Indian Ocean and the western Pacific sources forms a confluent zone north of the Indochina peninsula, from which confluent flow develops northeastward from late-june to mid-july (Figs. 5d f, 6d f). The resulting moisture convergence over east Asia seems responsible for the formation and maintenance of a frontal system. After late July, cyclonic flow associated with the low pressure anomaly develops over the subtropical Pacific (Figs. 3g j, 5g j). Outside of the northern and the southern boundaries of this cyclonic circulation, where the anticyclonic low-level wind anomaly is observed, corresponds to the bands of decreased precipitation. At the center of the cyclonic circulation, precipitation increases due to low-level convergence. As the pressure difference increases between this low pressure anomaly and the high pressure anomaly over the northeast Asian continent (Fig. 3j), a strong easterly reaches the southern tips of Korea and Japan, causing a brief secondary rainy period. b. Transport of moisture by low-level wind at 850 hpa Moisture transport is essential for the formation and maintenance of precipitation associated with the ASM. The Indian Ocean and the subtropical western Pacific are the major moisture sources. The first mode of the moisture transport anomaly explains 27% of the total variability. The moisture transport toward India and the Bay of Bengal comes from the Indian Ocean throughout the monsoon period. For the Indochina peninsula and the east Asian countries, the source of moisture varies with time. For example, for the Indochina peninsula, moisture transport anomaly comes from the tropical western Pacific until mid-june. As the low pressure anomaly pattern develops over the Asian continent afterward (Fig. 3d), moisture comes from the Indian Ocean until mid-july (Figs. 6d f), when moisture transport to the peninsula begins to weaken. After early August, moisture is transported from both the Indian Ocean and the western Pacific (Figs. 6h j). For the east Asian

11 3640 JOURNAL OF CLIMATE countries, moisture transport is not well established until mid-june. As the pathway along the southeast coast of China becomes established (Figs. 3d and 5d), moisture is transported from the Indian Ocean (Figs. 6d f) as well as from the subtropical western Pacific. The moisture transport from the Indian Ocean weakens significantly by mid-july (Fig. 6g), when the low pressure anomaly develops over the western Pacific (Fig. 3g). The subsequent merger of the Pacific low pressure anomaly with that over the Asian continent (Figs. 3g h) blocks moisture transport from the Indian Ocean. As a result of this blocking, moisture originating from the Indian Ocean accumulates over the Bay of Bengal and the Indochina peninsula and precipitation significantly increases over those areas (Figs. 6g j). Moisture transport from the western Pacific still remains important for the east Asian countries (Figs. 6h j); this moisture transport anomaly is now associated with the cyclonic circulation anomaly over the subtropical western Pacific. This moisture transport anomaly also results in the increase of precipitation over the Indochina peninsula (Figs. 6h j). As discussed earlier, the location of the subtropical western Pacific pressure anomaly pattern is crucial for regional precipitation fluctuation. Specifically, the result shows that the northeastward migration of the convergence zone of the moisture transport essentially determines the location of the precipitation front; this, in turn, determines the onset of the east Asian local monsoons such as mei-yu, baiu, and changma. 6. Circulation features at upper level Figure 7 shows the seasonally evolving streamline of the upper-level wind anomaly at 200 hpa. This mode explains about 39% of the total variability. During the early summer, strong westerly jet and midlatitude cyclonic circulation develop north of the Tibetan plateau, Korea, and Japan by early July (Figs. 7a e). Over the Asian continent, anticyclonic circulation anomaly forms from mid-june (15 June) to early July characterizing the westerlies in midlatitudes and the easterlies south of the Tibetan plateau (Figs. 7d,e). The circulation anomaly propagates northwestward from southeast China since mid-june (15 June; figure not shown). This anticyclonic circulation, also seen in the upper-level circulation climatology in Lau et al. (2000), is due to heating over land and ocean. Further, Li and Yanai (1996) showed that the oceanic heat source migrating northward merges with the heat source over the land and the resulting heating maximum reaches 20 N. During the early stage of the ASM, the upper-level westerly jet anomaly appears over a broad latitude band (Figs. 7a,b). As warming progresses over the continent, the pressure gradient between the continent and the Indian Ocean becomes weaker, resulting in the reversal of meridional temperature gradient, causing the appearance of anomalous easterly jet south of 30 N (Figs. 7e,f; Li and Yanai 1996). The band of westerly jet anomaly in Figs. 7a and 7b moves farther to the north. By mid-june, the trough of the jet crosses the southern tip of Japan and Korea (Fig. 7c). The downstream side of the jet stream trough is associated with the upper-level divergence that may cause the rising motion, while the upstream side of the trough shows convergence (sinking motion). It appears that positive vorticity is advected downstream of the trough, inducing upper-level divergence (Holton 1992). Figure 8 shows relative vorticity advection by the seasonal cycle of the upper-level wind in Fig. 7. Although a few regions show the rather weak dynamical relationship between positive (negative) relative vorticity advection and divergence (convergence), positive advection generally occurs downstream of the westerly jet stream trough from May to early July (Figs. 8a e). On the other hand, the upstream area of the trough is characterized by negative vorticity advection. After mid-june, the cyclonic pattern over east Asia (Korea and Japan) elongates in the zonal direction with the center moving eastward (Figs. 7d,e). At the same time, the westerly jet weakens significantly and is replaced eventually by an easterly jet anomaly near 20 N. This anomaly is related to the tropospheric warming over the continent (Krishnamurti 1985) and the seasonal cycle of the intensity of the westerly jet, which weakens as summer matures. Then, the vertical structure associated with the first internal mode in the Tropics, westerly at lower levels and easterly at upper levels, is established over south Asia and the Indian Ocean (Figs. 5e i, 7e i). This strong wind shear may enhance the instability of the atmospheric column, helping to maintain the precipitation over the Indian Ocean, particularly over the Bay of Bengal and south Asia. It appears that the development of baroclinic waves along the westerly jet stream facilitates convection (Figs. 7a e, 8a e). An assessment of how significantly the upper atmosphere contributes to the evolution of precipitation during the ASM is yet to be done. The present study shows that the patterns of the upper-level circulation anomaly and its divergence are consistent with the precipitation pattern. In particular, positive vorticity advection appears to be consistent with precipitation pattern over the midlatitude east Asian monsoon region. 7. Summary and concluding remarks The seasonal cycle of the ASM and the associated temporal and spatial evolution of synoptic fields have been investigated using cyclostationary EOF analysis over a 21-yr ( ) Xie Arkin precipitation data and NCEP NCAR reanalysis data. This study focused only on the prominent ASM period (21 May 28 August). The first mode describes the typical seasonal evolution of spatial patterns associated with the Indian mon-

12 15 DECEMBER 2002 LIM ET AL FIG. 7. The seasonal cycle of the streamline of the upper-level (200 hpa) wind anomalies. The dark shading denotes the region of divergence and the light shading the region of convergence. soon from late May, mei-yu (China) in June, baiu (Japan) from mid-june to mid-july, and changma (Korea) from late June to late July. The complex nature of the seasonal evolution is obvious in the extracted modes of precipitation, sea level pressure, low-level (850 hpa) and upper-level (200 hpa) winds and moisture transport (see Fig. 9). The seasonal cycle typically explains about 20% 40% of the total variability.

13 3642 JOURNAL OF CLIMATE FIG. 8. The relative vorticity advection by the seasonal cycle of the upper-level wind anomalies (Fig. 7). The dark shading represents the positive vorticity advection and the light shading negative vorticity advection.

14 15 DECEMBER 2002 LIM ET AL FIG. 9. A schematic diagram showing the configuration of the surface pressure anomaly, precipitation anomaly, moisture transport anomaly, and the resulting moisture transport convergence band during the four stage of the Asian summer monsoon: (a) early (initiation) stage, (b) middle stage, (c) transition stage, and (d) later stage. The solid lines with H or L represent boundaries of surface high and low pressure anomalies, respectively. The short hollow arrows describe the movement of surface pressure anomaly patterns. The shaded areas represent dominant precipitation regions. The long arrows depict the direction of moisture transport anomalies carried by low-level wind anomalies. The filled bars represent dominant moisture transport convergence regions, the bars with dotted boundary implying weakening convergence bands. The short filled arrows describe the movement direction of convergence bands. One of the key features in the evolution of the ASM seasonal cycle is the sea level pressure change over the Asian continent and the resulting sea land pressure contrast. The development of the pressure anomaly pattern in the subtropical western Pacific also plays a crucial role in the evolution of the ASM system. The complex nature of low-level wind and the resulting moisture transport is mainly due to the differential timing and strength, and the location of these two pressure system (Fig. 9). Specifically, the development of a low pressure anomaly over India and Southeast Asia and the northwestward movement of the SWPH anomaly from early June open a passageway for moisture transport from the Indian Ocean to the southeast Asian countries. This results in increased precipitation over the Bay of Bengal and Southeast Asia. The northwestward migration of the SWPH anomaly also helps transport the moisture of the Pacific to the east and northeast Asian countries a key process for the onset of local monsoons. The increased convergence of moisture transport at the lower level makes the atmospheric column unstable, resulting in the development of convective instability; this appears to create a favorable condition for low-level air moving upward during the mature stage of the east Asian monsoon. The continuous supply of moisture from the Indian Ocean and the subtropical western Pacific is, of course, instrumental for maintaining a long spell of precipitation over east Asia. Ascending air following the low-level convergence along the frontal region is typically associated with divergence at the upper level. The reversal of the subtropical western Pacific pressure anomaly in mid-july marks the cessation of moisture transport from the Indian Ocean to northeast Asia (Fig. 9c). Precipitation in the east Asian countries lingers for a few more weeks with the transport of moisture from the midlatitude western Pacific. Without moisture transport from the Indian Ocean, the location of precipitation is primarily due to the location of the subtropical western Pacific low pressure anomaly pattern. With the shutdown of the northeastward moisture passage, moisture from the Indian Ocean is transported mainly to the Bay of Bengal and the Indochina peninsula; this results in increased precipitation in these areas from mid-july to late August (Fig. 9c).

15 3644 JOURNAL OF CLIMATE During the terminal stage of the ASM, the rainfall over India decreases significantly as a high pressure anomaly develops over the Asian continent. In the tropical Pacific, the low pressure anomaly pattern becomes stronger until it moves away from the Asian continent during late August. The strengthening of the low pressure anomaly in the western Pacific causes moisture transport from the western Pacific to the South China Sea and the Indochina peninsula until the end of August (Fig. 9d). In closing, the physics of the seasonal cycle of the ASM, as the most dominant mode of variability, was faithfully extracted from key synoptic variables in the present investigation; the detailed space time physical evolutions of different physical variables are consistent with each other. We feel that this modal decomposition based on physics will help us understand the complex nature of the ASM and ultimately improve its forecasts. Future study should focus on the higher modes of variability in order to have a fuller understanding of the ASM. Acknowledgments. The authors thank Dr. Song-You Hong for providing them with the dataset. We also thank Prof. Henry Fuelberg and Dr. Kyong-Hwan Seo for carefully proofreading our manuscript. This work was supported by the National Science Foundation (ATM ) and the Department of Energy (DE-FG ). REFERENCES Chang, C. P., and G. T. Chen, 1995: Tropical circulations associated with southwest monsoon onset and westerly surges over the South China Sea. Mon. Wea. Rev., 123, Hidore, J. J., and J. E. Oliver, 1993: Climatology: An Atmospheric Science. 1st ed. Macmillan, 423 pp. Holton, J. R., 1992: An Introduction to Dynamic Meteorology. 3d ed. Academic Press, 511 pp. Hoskins, B. J., and M. J. Rodwell, 1995: A model of the Asian summer monsoon. Part I: The global scale. J. Atmos. Sci., 52, Hsu, H. H., C. T. Terng, and C. T. Chen, 1999: Evolution of largescale circulation and heating during the first transition of Asian summer monsoon. J. Climate, 12, Huang, R., and F. Sun, 1992: Impacts of the tropical western Pacific on the East Asian summer monsoon. J. Meteor. Soc. Japan, 70, Ju, J., and J. Slingo, 1995: The Asian summer monsoon and ENSO. Quart. J. Roy. Meteor. Soc., 121, Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-year Reanalysis Project. Bull. Amer. Meteor. Soc., 77, Kang, I.-S., C.-H. Ho, Y.-K. Lim, and K.-M. Lau, 1999: Principal modes of climatological seasonal and intraseasonal variation of the Asian summer monsoon. Mon. Wea. Rev., 127, Kim, K. Y., and G. North, 1997: EOFs of harmonizable cyclostationary processes. J. Atmos. Sci., 54, , and Q. Wu, 1999: A comparison study of EOF techniques: Analysis of nonstationary data with periodic statistics. J. Climate, 12, , and C. Chung, 2001: On the evolution of the annual cycle in the tropical Pacific. J. Climate, 14, Krishnamurthy, V., and J. Shukla, 2000: Intraseasonal and interannual variability of rainfall over India. J. Climate, 13, Krishnamurti, T. N., 1971: Tropical east-west circulations during northern summer. J. Atmos. Sci., 28, , 1985: Summer Monsoon Experiment A review. Mon. Wea. Rev., 113, , P. Ardanuy, Y. Ramanathan, and R. Pasch, 1981: On the onset vortex of the summer monsoon. Mon. Wea. Rev., 109, Lau, K.-M., and M.-T. Li, 1984: The monsoon of east Asia and its global associations A survey. Bull. Amer. Meteor. Soc., 65, , and P. H. Chan, 1986: Aspects of the day oscillation during the northern summer as inferred from outgoing longwave radiation. Mon. Wea. Rev., 114, , and S. Yang, 1996: Seasonal variation, abrupt transition, and intraseasonal variability associated with the Asian summer monsoon in the GLA GCM. J. Climate, 9, , G. T. Yang, and S. H. Shen, 1988: Seasonal and intraseasonal climatology of summer monsoon rainfall over east Asia. Mon. Wea. Rev., 116, , K.-M. Kim, and S. Yang, 2000: Dynamical and boundary forcing characteristics of regional components of the Asian summer monsoon. J. Climate, 13, Li, C., and M. Yanai, 1996: The onset and interannual variability of the Asian summer monsoon in relation to land sea thermal contrast. J. Climate, 9, Liang, X.-Z., and W. C. Wang, 1998: Associations between China monsoon rainfall and tropospheric jets. Quart. J. Roy. Meteor. Soc., 124, Matsumoto, J., 1992: The seasonal changes in Asian and Australian monsoon regions. J. Meteor. Soc. Japan, 70, Meehl, G. A., 1987: The annual cycle and interannual variability in the tropical Pacific and Indian Ocean region. Mon. Wea. Rev., 115, , 1997: The south Asian monsoon and the tropospheric biennial oscillation (TBO). J. Climate, 10, Murakami, T., and J. Matsumoto, 1994: Summer monsoon over the Asian continent and western North Pacific. J. Meteor. Soc. Japan, 72, Nagazawa, T., 1992: Seasonal phase lock of intraseasonal variation during the Asian summer monsoon. J. Meteor. Soc. Japan, 70, Ninomiya, K., 1984: Characteristics of Baiu front as a predominant subtropical front in the summer northern hemisphere. J. Meteor. Soc. Japan, 62, Subbaramayya, I., and R. Ramanadham, 1981: On the onset of the Indian southwest monsoon and the monsoon general circulation. Monsoon Dynamics, J. Lighthill and R. P. Pearce, Eds., Cambridge University Press, Tanaka, M., 1992: Intraseasonal oscillation and the onset and retreat dates of the summer monsoon over east, southeast Asia and the western Pacific region using GMS high cloud amount data. J. Meteor. Soc. Japan, 70, Tao, S. Y., and L. X. Chen, 1987: A review of recent research of the East Asian summer monsoon in China. Monsoon Meteorology, C.-P. Chang and T. N. Krishnamurti, Eds., Oxford University Press, Ueda, H., T. Yasunari, and R. Kawamura, 1995: Abrupt seasonal change of large-scale convection activity over the western Pacific in northern summer. J. Meteor. Soc. Japan, 73, Vernekar, A. D., and Y. Ji, 1999: Simulation of the onset and intraseasonal variability of two contrasting summer monsoon. J. Climate, 12, 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, Xie, P., and P. A. Arkin, 1996: Analyses of global monthly precipitation using gauge observations, satellite estimates, and numerical model predictions. J. Climate, 9, Yanai, M., C. Li, and Z. Song, 1992: Seasonal heating of the Tibetan Plateau and its effects on the evolution of the Asian summer monsoon. J. Meteor. Soc. Japan, 70 (1B),