Interaction between the Summer Monsoons in East Asia and the South China Sea: Intraseasonal Monsoon...

1 MAY 2000 1373C H E N E T A L .

q 2000 American Meteorological Society

Interaction between the Summer Monsoons in East Asia and the South China Sea:Intraseasonal Monsoon Modes

TSING-CHANG CHEN, MING-CHENG YEN,* AND SHU-PING WENG

Atmospheric Science Program, Department of Geological and Atmospheric Sciences, Iowa State University, Ames, Iowa

(Manuscript received 2 September 1998, in final form 3 May 1999)

ABSTRACT

The summer monsoons in East and Southeast Asia are characterized, respectively, by the Mei-yu (in easternChina)–Baiu (in Japan) front (MBF) and by the monsoon trough stretching from northern Indochina to thePhilippine Sea. These two major monsoon elements are separated by the North Pacific anticyclone. As indicatedby the 850-mb zonal wind and cumulus convection over some key areas, a distinct opposite-phase intraseasonalvariation exists between the two monsoon elements. Two approaches are adopted to explore the cause of thisopposite-phase variation (which reflects the coupling between the two monsoon components): 1) the correlationcoefficient patterns between the 850-mb zonal-wind monsoon index and the 850-mb streamfunction field and2) the composite 850-mb streamline charts and the 1208E zonal-wind cross sections. It is shown that the opposite-phase variation between the two monsoon elements is caused by the anomalous circulation associated with thenorthward-migrating 30–60-day monsoon trough/ridge from the equator to 208N and with the westward-prop-agating 12–24-day monsoon low–high along the latitude of ;158–208N. Results obtained in this study are usedto address two often discussed phenomena of the East Asian monsoon: 1) the rapid northward shift of the MBFacross the Yangtze River basin during the Mei-yu onset is related to the north–south meridional oscillation ofthe MBF, and 2) the three longitudinally oriented location zones of extremely heavy rain events in eastern Chinaare formed by the alternation of deep cumulus convection zones associated with the intraseasonal monsoonvortices (centered in the northern part of the South China Sea) between extreme monsoon conditions.

1. Introduction

The East Asian monsoon is characterized by the Mei-yu–Baiu front (MBF), which is climatologically one ofthe major convergence zones within the global atmo-spheric circulation. During the summer monsoon sea-son, the MBF extends from southern China to Japan andthe Aleutian Islands, and exhibits a definite moisturegradient (but no temperature gradient). Water vapororiginating from the South China Sea (SCS)–westerntropical Pacific (WTP) region is advected by the mon-soon southwesterlies along the MBF (Murakami 1959).Moisture and precipitation associated with the front aremaintained by water vapor flux convergence toward thisfront (Akiyama 1973). This hydrological process is driv-en by the large-scale divergent circulation (Chen et al.1988a). Following the seasonal march, the MBF mi-

∗Current affiliation: Department of Atmospheric Science, NationalCentral University, Chung-Li, Taiwan.

Corresponding author address: Dr. Tsing-Chang Chen, Atmo-spheric Science Program, 3010 Agronomy Hall, Department of Geo-logical and Atmospheric Sciences, Iowa State University, Ames, IA50011.E-mail: [email protected]

grates northward from northern Indochina during Mayto the Yangtze River basin and southern Japan by earlyJune, triggering the onset of the East Asian summermonsoon. However, this northward progression is notuniform, but undergoes a north–south intraseasonal os-cillation induced by the eastward propagation of theintraseasonal global divergent circulation (Chen andMurakami 1988). Conceivably, the seasonal evolutionof the East Asian monsoon is modulated by this merid-ional oscillation.

Analyzing temporal evolution of the East Asian sum-mer monsoon rainfall, Lau et al. (1988) identified twoonsets of major monsoon rain: the first (known as theMei-Yu) occurs in central China during the first half ofJune, and the second occurs over northeastern Chinaduring late July. The multiple onsets of the East Asiansummer monsoon rain transpire as major rainbands un-dergo rapid transitions that progress wavelike fromsouth to north between three quasi-stationary locations:southern China (premonsoon rain), central China (theMei-yu front), and northeastern China (the polar front).As suggested by Lau et al.’s empirical orthogonal func-tion analysis of the East Asian summer monsoon rain-fall, the rapid transitions of these major rainbands mayresult from the phase lock between the 30–60-day and20-day intraseasonal monsoon modes.

1374 VOLUME 57J O U R N A L O F T H E A T M O S P H E R I C S C I E N C E S

The water vapor in the SCS–WTP region is suppliedby the convergence of water vapor flux transported bythe monsoon westerlies and by the trade easterlies (e.g.,Chen et al. 1988a). It was observed that the intensityof south Asian monsoon westerlies (an indicator of themonsoon life cycle) is modulated by the northward-propagating 30–60-day monsoon trough/ridge (Krish-namurti and Subrahmanyam 1982; Krishnamurti et al.1985) and that the trade easterlies associated with theNorth Pacific anticyclone are oscillated by the eastward-propagating, global 30–60-day mode (Chen 1987). Forthese reasons, water vapor and water vapor flux in theSCS–WTP region fluctuate intraseasonally (Chen et al.1988b). Since rainfall is maintained by the convergenceof water vapor flux, it is expected that the SCS–WTPmonsoon rain undergoes an intraseasonal fluctuation.Actually, this inference is substantiated by a dipolestructure of tropical convection with an intraseasonaltime scale identified by Lau and Chan’s (1986) EOFanalysis of the outgoing longwave radiation (OLR) be-tween the Indian Ocean and the SCS–WTP regions.

Examining the time variation of the SCS summermonsoon, Chen and Chen (1995) showed that the 30–60-day monsoon trough–ridge in the SCS–WTP regionis coupled with the 30–60-day global divergent circu-lation in a way similar to the type of coupling that occursover the Indian monsoon region. The life cycle of theSCS summer monsoon is regulated by the 30–60-daymonsoon trough–ridge in such a manner that the mon-soon circulation reaches its maximum (minimum) in-tensity when this trough (ridge) moves to approximately158–208N. The SCS summer monsoon break occurs si-multaneously with a phase lock in the north SCS be-tween the northward-propagating 30–60-day monsoonridge and the westward-propagating 12–24-day closedhigh. The most northern parts of the anomalous circu-lation associated with these two intraseasonal monsoonmodes may reach the Yangtze River basin.

As shown in previous studies, intraseasonal oscilla-tions exist in both the summer monsoons in East Asiaand the SCS–WTP region. For East Asia, the MBF un-dergoes a north–south intraseasonal oscillation, and themonsoon rainfall, subsequently, exhibits an intrasea-sonal fluctuation. In the SCS–WTP region, the summermonsoon life cycle, cumulus convection, and hydrolog-ical processes are regulated by the intraseasonal mon-soon mode. In view of the intraseasonal oscillations ofthe monsoons over the two regions, a question is raised:Is there a physical and systematic link between intra-seasonal oscillations of the summer monsoon in EastAsia and the SCS–WTP region? In their examinationof the evolution of the upper-level circulation duringthe Mei-Yu onset in eastern China, Yeh et al. (1959)presented a series of 3-km streamline charts. Synopticdevelopment shown in these charts suggest that the Mei-yu onset is related in some way to the northward mi-gration of the monsoon trough, which extends longi-tudinally from the Indian subcontinent to the SCS.

Based on the y–t diagram of a convective index along1408E, Chen and Murakami (1988) pointed out that thecumulus convection associated with the ITCZ in theWTP region is enhanced (suppressed) when the MBFmigrates southward (northward). Conceivably, thereshould be a physical link between the intraseasonal os-cillations of the summers monsoons in East Asia andthe SCS–WTP region.

So far, the mechanism responsible for the physicallink between the monsoons in these two regions has notbeen systematically examined. In view of the researchfindings by Yeh et al. (1959) and Chen and Murakami(1988), the possible coupling of the two monsoon com-ponents may be explained through the coherent intra-seasonal oscillations of the MBF and the life cycle ofthe SCS summer monsoon. A research task along thisline is undertaken with two major data sources: the Na-tional Centers for Environmental Prediction–NationalCenter for Atmospheric Research NCEP–NCAR re-analysis data (Kalnay et al. 1996) for the period 1979–93 and the equivalent blackbody temperature (TBB) ob-served by the Japanese Geostationary MeteorologicalSatellite (GMS) for the period 1980–93. Results of thisstudy are presented in the following arrangement. Thepossible coupling of the two monsoon components isdemonstrated by the coherent intraseasonal oscillationsof monsoon westerlies and cumulus convection in sec-tion 2. The coupling mechanism of the two monsooncomponents is illustrated by correlation coefficient pat-terns, composite synoptic charts, and zonal-wind crosssections in section 3. Section 4 offers a discussion onthe possible effects of the coupling of the two monsoonson two often discussed phenomena of the East Asianmonsoon: 1) the rapid shifts of upper westerlies and theMBF during the Mei-yu onset and 2) the formation ofthree east–west-oriented zones of extremely heavy rainevents in eastern China. Finally, section 5 is devoted toconcluding remarks.

2. Indication of possible coupling

a. Time series of monsoon indices

To indicate the temporal variation of monsoon inten-sity and life cycle, rainfall and low-level wind are com-monly adopted as monsoon indices (e.g., Murakami1972; Krishnamurti 1985). For this reason, an inferenceof the possible coupling of the summer monsoons inEast and Southeast Asia may be made with time seriesof these indices at some key climatological locationsassociated with the MBF and with the SCS–WTP mon-soon trough.

Since there are no precipitation data available overthe open seas, some precipitation proxy generated fromsatellite data will be used. Arkin and Ardanuy (1989)proposed that tropical rainfall may be estimated withOLR values below the threshold of 235 W m22. Nittaand Sekine (1994) inferred that deep cumulus convec-

1 MAY 2000 1375C H E N E T A L .

tion was associated with cloud tops above 400 mb withTBB values below 250 K. Instead of this TBB threshold,Chen and Chen (1995) used the TBB value of 270 K asa threshold to include low convective clouds. For therainfall and cumulus convection proxies, we define thefollowing two convective indices:

22DOLR 5 235 W m 2 OLR (50 if DOLR # 0)

and

DT 5 270 K 2 T (50 if DT # 0).BB BB BB

In addition to the rainfall and cumulus convection prox-ies, the 1-yr rainfall estimations (P) for 1979 generatedfrom the IR data by Susskind and Pfaendtner (1989) arealso incorporated into our analysis.

Shown in Fig. 1 are histograms of P, DOLR, DTBB

and time series of u(850 mb) for the summers of 1979(left) and for 1989 (right) at locations of maximum rms(root-mean-square) value centers for these variablesover the SCS and eastern China. The selection of thesetwo summers will be explained later. Affected by theinterannual variation of the summer circulation in theEast Asia–western Pacific region (e.g., Nitta 1987; Chenand Weng 1998), locations of the rms centers may notalways be the same in every summer. In Fig. 1, theupper histograms and time series are located at or southof the Yangtze River basin, and the lower ones are ator south of the SCS–WTP summer monsoon trough. Forcomparison, the upper histograms and time series areplotted upside down. Variations of the summer mon-soons in East Asia and the SCS–WTP region are wellindicated by these monsoon indices. Previous studiescited in the introduction report that variations of the twomonsoon components are modulated by the 30–60-dayand 12–24-day monsoon modes. To illustrate the rolesof these two intraseasonal modes in the variations ofthe two monsoon components, a combination of theseasonal mean value, the 30–60-day, and the 12–24-daybandpass filtered monsoon indices [solid lines in Fig.1, denoted by for later discussion] are superimposed( · )on histograms and time series. The second-order But-terworth bandpass filter (Murakami 1979) is used toisolate these two temporal regimes. Salient features ofthese indices include the following.

1) The MBF undergoes a north–south intraseasonal os-cillation, and the East Asian monsoon rain exhibitsan intraseasonal fluctuation that is caused by thenorthward migration of intraseasonal monsoonmodes (Lau et al. 1988). These temporal variationsin the East Asian monsoon are reflected by the his-tograms of P, DTBB, DOLR, and the time series ofu(850 mb).

2) The life cycle of the SCS–WTP summer monsoonis regulated basically by the northward-migrating30–60-day monsoon trough–ridge, and the onset(break) of this monsoon is often the result of a phaselock between the 30–60-day monsoon trough (ridge)

and the 12–24-day monsoon low (high) [Chen andChen (1995); Chen and Weng 1997)—however, tomake the paper more self-contained, the propagationproperties of the two intraseasonal monsoon modesare provided in the appendix]. Since cumulus con-vection in the SCS–WTP region undergoes a pro-nounced intraseasonal fluctuation (e.g., Lau andChan 1986), it is not surprising to see that a clearintraseasonal fluctuation emerges from the monsoonindices in the SCS–WTP region.

3) The spectral analysis of all monsoon indices revealsthat both the 30–60-day and 12–24-day signals aredistinct1 but may not always exist simultaneously inboth monsoon components. One intraseasonal modemay be more significant than the other during onemonsoon season, while the reverse situation may betrue in another monsoon season. The contrast be-tween the bandpass filtered monsoon indices of thetwo monsoon seasons shows clearly that, in the sum-mer of 1979, the 30–60-day mode is more pro-nounced than the 12–24-day mode, while the 12–24-day mode is the dominant one in the summer of1989. Due to this, we selected these two summersto illustrate the possible coupling of the two mon-soons by these two intraseasonal monsoon modes.

4) The most interesting and important feature emergingfrom Fig. 1 is the opposite-phase variation of thesame monsoon indices between the two monsoons.The correlation coefficients of all correspondingbandpass-filtered monsoon indices during the twomonsoons are above 20.7 in the summer of 1989.For the summer of 1979, the correlation coefficientsof u(850 mb), p, and OLR between the two mon-soons are 20.7, 20.65, and slightly below 20.5,respectively. The opposite-phase variations are byno means accidental, but indicate strongly a physicalcoupling of the two monsoons through some mech-anism. In view of contributions of the 30–60-dayand 12–24-day modes to the temporal variations ofall monsoon indices, it is conceivable that they playa vital role in the coupling of the two monsoons.

Presented in Fig. 1 are monsoon indices of only twomonsoon seasons. One may question whether the op-posite-phase variations of monsoon indices between thetwo monsoons exist in other summers. Showing histo-grams and time series of monsoon indices for all seasonsmay not be a practical way to answer this question.Thus, we show correlation coefficients of the bandpassfiltered monsoon indices between the two for all summerseasons (Fig. 2). For most monsoon seasons, correlationcoefficients of u(850 mb), DOLR, and DTBB are over20.7, 20.6, and 20.5, respectively. Evidently, the op-posite-phase relationship between the temporal evolu-

1 The power spectra of monsoon indices averaged over 15 summers(1979–93) are given in the appendix.

1 MAY 2000 1377C H E N E T A L .

FIG. 2. Correlation coefficients between the combined 30–60-dayand 12–24-day monsoon indices in the East Asian and SCS regions.The time series of a given monsoon index is constructed with amonsoon variable at the center of its maximum root-mean-squarevalue for the entire monsoon season (May–Aug) over the East Asianand SCS regions.

←

FIG. 1. Indices of the monsoon westerlies and cumulus convection (or precipitation) for East Asia (upper part of each) and the SCS (lowerpart of each) during the summers of 1979 (left) and 1989 (right) are displayed. (left) A daily histogram of Goddard precipitation estimation(top), DOLR (5235 W m22 2 OLR) (middle), and the time series of daily u(850 mb) (bottom) for the summer of 1979; (right) dailyhistograms of DOLR (top) and of DTBB (5270 K 2 TBB) (middle), and the time series of daily u(850 mb) (bottom) for the summer of 1989.The location of each of the monsoon indices is shown. Superimposed on both the histogram and the time series is the combined 30–60-dayand 12–24-day filtered time series of this variable (thick solid line).

tion of the two monsoons is common to most monsoonseasons.

b. The y–t diagrams of monsoon zonal flow

An inference of the north–south intraseasonal oscil-lation of the MBF may be made with the u(850 mb) orwith the cumulus convection index of the East Asianmonsoon. Since this monsoon index is measured at afixed location, a latitudinal location index of this frontmust be constructed in order to answer the followingquestion: Can the north–south intraseasonal oscillationof the MBF latitudinal location coincide with the tem-poral evolution of the SCS–WTP summer monsoon?This question is answered in terms of the y–t diagramsof u(850 mb) at 1158E for the summers of 1979 and1989 (Fig. 3). To facilitate our discussion, we super-impose these y–t diagrams of u(850 mb) with two mon-soon indices: 1) the u(850 mb) index at 7.58N, 1158E(the lower solid line in Figs. 3a,b) and 2) the filteredmaximum u(850 mb) location index of the MBF at1158E (the upper dashed line in Figs. 3a,b). The laterindex is formed by a combination of the 30–60-day and12–24-day filtered latitudinal locations of the maximumu(850 mb).

The intensity variation of the monsoon circulation is

reflected by the low-level monsoon westerlies. Thus, thetemporal evolutions of the 1979 and 1989 SCS summermonsoons are revealed clearly from both the u(850 mb)y–t diagram south of 208N and the u(850 mb) indicesof these two monsoon seasons. For the East Asian sum-mer monsoon, let us use the latitudinal location of themaximum u(850 mb) to indicate approximately the coreof the monsoon southwesterlies associated with theMBF. With this approximation, the north–south oscil-lation of the MBF latitudinal location is reflected by thenorth–south oscillation of the maximum u(850 mb) lo-cation [revealed from the u(850 mb) y–t diagram northof 208N].

The question posed above may be answered by asimple contrast between the SCS u(850 mb) index andthe filtered MBF location index.

1) The amplitude of the MBF’s location index is about208 latitude, which is comparable to the north–southextent covered by the northward migration of the30–60-day monsoon trough/ridge in the SCS region.For the 12–24-day monsoon mode, its latitudinal ex-tent has not yet been presented (but will be illustratedlater). Actually, the circulation pattern and size ofthis intraseasonal mode in the SCS region are similarto these associated with the 30–60-day monsoontrough–ridge.

2) For both the 1979 and 1989 summer monsoon sea-sons, the MBF location index oscillated in the north–south direction coherently with the intraseasonal var-iation of the SCS u(850 mb) index. The correlationcoefficients between these two time series are 0.91and 0.95, respectively. The coherent relationship be-tween the two indices, shown in Fig. 3, is presentin every summer (not shown) analyzed in this study.

The coherent intraseasonal variations in the monsoonindices of the two monsoon components presented inthis section and in section 2a lead us to a basic question:What is the physical mechanism responsible for the co-herent northward migration (southward retreat) of theMBF in East Asia and the intraseasonal variation of thesummer monsoon intensity in the SCS–WTP region? Inother words, what is responsible for the coupling of themonsoons over these two regions?

3. Coupling mechanism

Analyzing the lower-tropospheric streamline chartsduring the second half of May and early June 1956, Yehet al. (1959) pointed out that the Mei-Yu onset (triggered


FIG. 3. The latitude–time (y–t) diagrams of u(850 mb) (1158E) for (a) the summer of 1979 and (b) the summer of 1989. Thethick solid line is a combination of the 30–60-day and 12–24-day bandpass filtered u(850 mb) time series at the center of theroot-mean-square value of Du(850 mb) [5u(850 mb) 2 u (850 mb); ( ) 5 summer mean( )] in the SCS region. The thick dashedline is the combined bandpass filtered 30–60-day and 12–24-day maximum u(850 mb) location associated with the MBF. Thecontour interval of u(850 mb) is 2.5 m s21. Values of 2.5 m s21 # u(850 mb) , 5 m s21 and 5 m s21 # u(850 mb) are lightlyand heavily stippled, respectively.

by the northward movement of a shear line) is linkedto the northward migration and eastward extension ofthe monsoon trough from the Indian subcontinent to theSCS region. With a careful examination of Yeh et al.’sstreamline charts (their Fig. 16), one can also see thatthe lower-tropospheric high retreats from the SCS regioninto the western North Pacific when the monsoon troughmoves northward. As observed by previous studies (e.g.,Chen and Chen 1995), the northward-migrating 30–60-day monsoon trough/ridge over the SCS–WTP regionis associated with an anomalous cyclonic (anticyclonic)circulation, and the westward-propagating 12–24-daymonsoon mode (which plays an important role in theSCS monsoon onset and break) possesses a double-celled structure with its northern cell moving along158;208N and its southern cell along the equator. The

opposite-phase intraseasonal variations of monsoon in-dices between the two monsoons may be establishedthrough their coupling by the anomalous circulationcells associated with these two intraseasonal monsoonmodes. This argument will be substantiated by usingtwo approaches in the following two sections: 1) cor-relation coefficient patterns between the u(850 mb)monsoon index and the lower-tropospheric streamfunc-tion and (2) composite synoptic charts of lower-tropo-spheric streamline and zonal-wind cross sections.

a. Correlation coefficient pattern

The spatial structure of summer monsoon distur-bances associated with the intraseasonal fluctuation ofa monsoon index may be inferred from the correlation

1 MAY 2000 1379C H E N E T A L .

coefficient pattern between this index and streamfunc-tion. Therefore, we construct the correlation coefficientpatterns between the monsoon index of the 850-mbmonsoon westerlies and the 850-mb streamfunctionfields for three temporal regimes: 1) the departure fromthe seasonal cycle, D( ); 2) the 30–60-day mode, (;);and 3) the 12–24-day mode, (^). The seasonal cycle issimply removed by a least-squares-fit approach, whilethe two intraseasonal modes are isolated by the first-order Butterworth filter (Murakami 1979). For conve-nience of later discussion, we denote the correlationcoefficients of the three temporal regimes as sDuDc,

, and .s suc uc

The procedure for computing the three aforemen-tioned correlation coefficients is outlined as follows.

1) The monsoon westerly indices [Du(850 mb), u(850mb), and u(850 mb)] are formed at rms[Du(850 mb)]centers over the SCS and eastern China for eachsummer monsoon season.

2) We select in each monsoon season the periods when|Du(850 mb)| $ 0.8SD1 and |u(850 mb)| $ 0.8SD2

[SD1 and SD2 are standard deviations of u(850 mb)and u(850 mb), respectively, over the entire monsoonseason] for constructing the correlation coefficientpatterns.

3) The filtered monsoon westerly index and stream-function of selected periods for all 15 monsoon sea-sons in each temporal regime are assembled se-quentially (following the order of years) to form anew set of data. Correlation coefficient patterns,sDuDc, , and , of three temporal regimes ares suc uDc

constructed with the three newly formed time seriesof the filtered monsoon westerly index and the cor-responding streamfunction anomalies.

Correlation coefficient patterns of the three temporalregimes with the SCS (East Asia) monsoon westerlyindices are displayed in Figs. 4a–c (Figs. 4d–f). Salientfeatures of these patterns are as follows.

1) A double-celled structure dominates all three typesof correlation coefficient patterns. As revealed fromFig. 1, the most distinctive intraseasonal oscillationsin the Du(850 mb) index are the 30–60-day and 12–24-day monsoon modes. The resemblance of the spa-tial pattern between sDuDc and the other two (i.e.,

and ) indicates that the intraseasonal variations suc uc

of Dc(850 mb) fields is dominated primarily by thesetwo intraseasonal monsoon modes.

2) The Du(850 mb) index in the SCS reaches its max-imum (minimum) value when the 30–60-day mon-soon trough (ridge) arrives at 158N;208N and a new30–60-day monsoon ridge (trough) appears near theequator. The double-celled structure of s u manifestsc

when the anomalous cyclonic (anticyclonic) circu-lation centered in the north SCS attains its maximumintensity as the 30–60-day monsoon trough (ridge)arrives there. At this stage, the anomalous circulation

at the equator should be opposite to its northerncounterpart.

3) Because of the spatial structure of the 12–24 daymonsoon mode, the double-celled pattern of iss uc

not a surprise. In view of the differences of propa-gation properties and life cycle between both thisintraseasonal mode and the 30–60-day monsoonmode, the similarity in the spatial structures of s uc

and is surprising and may reflect two possibili-s uc

ties: (a) the relationship between the u(850 mb) mon-soon index and the anomalous circulation associatedwith the 12–24-day monsoon mode behaves in a waysimilar to the 30–60-day monsoon mode, and (b)The spatial structure of the anomalous circulation ofthe 12–24-day monsoon mode during its maturestage over the SCS resembles that of the 30–60-daymonsoon mode.

4) If opposite-phase intraseasonal variations of themonsoon indices in the two monsoons are inducedby the same monsoon disturbance, the following maybe expected: the spatial patterns of and , con-s suc uc

structed with the 850-mb zonal-wind indices of boththe SCS and East Asian monsoons should resembleeach other but have opposite spatial structures. Thisexpectation is confirmed by the contrast of corre-sponding correlation coefficient patterns in Figs.4a–c and Figs. 4d–f.

In summary, feature 4 of the correlation coefficientpatterns shown in Fig. 4 implies that the intraseasonaloscillations of the two monsoons are coupled by theanomalous circulations associated with the northward-migrating 30–60-day monsoon trough–ridge and withthe westward-propagating 12–24-day monsoon low–high. However, a more quantitative assessment of con-tributions of the two intraseasonal modes to the intra-seasonal variations of monsoon indices cannot be at-tained directly from the correlation coefficient patterns.Instead, this assessment may be made through the useof composite streamline charts and zonal-wind crosssections.

b. Composite charts

To illustrate the possible coupling of the two monsooncomponents, composite charts are prepared for threetemporal regimes: total, 30–60-day, and 12–24-day. Thecomposite procedure starts with the case selection usingthe following criteria.

1) The SCS zonal-wind monsoon index [u(850 mb) oru(850 mb)] exceeds 0.8 of its standard deviation(SD1 or SD2) over an entire monsoon season.

2) A distinct closed cyclonic (anticyclonic) cell is cen-tered at the northern SCS when the zonal-wind index[u(850 mb) or u(850 mb)] meets criterion 1.

3) For the 30–60-day monsoon mode, a clear trough(ridge) appears in the northern SCS when the u(850mb) index satisfies criterion 1.

All cases selected (see Tables 1 and 2) are averaged


FIG. 4. Listed in Table 1(2) are the dates when the maximum and minimum u(850 mb) [u(850mb)] in the SCS are larger than 0.8s and smaller than 0.8s, respectively. Here, s is the standarddeviation of the u(850 mb) or u(850 mb) time series over a summer (May–Aug). The selectedcycles of u(850 mb) [or u(850 mb)] time series are combined chronologically to form a newtime series. The correlation coefficients between the newly formed time series and the corre-sponding [ ] fields are shown in (b) [(c)]. The SCS Du(850 mb) time seriesc(850 mb) c(850 mb)covered by the selected u(850 mb) and u(850 mb) cycles are lined up to form a new Du(850mb) time series that is correlated with the corresponding Dc(850 mb) fields. The correlationcoefficients generated from the selected Du(850 mb) time series and Dc(850 mb) fields aredisplayed in (a). Shown in (d)–(f ) are the correlation coefficient patterns corresponding to thosein (a)–(c) with the zonal-wind index over East Asia and the 850-mb filtered streamfunction.

over a 3-day window with its center date coincidingwith the maximum/minimum u(850 mb) or u(850 mb)index. After this composite procedure is complete, twotypes of composite charts are constructed:1) the 850-mb streamline charts superimposed with the

TBB index and2) the zonal-wind cross sections at 1208E superimposed

with the latitudinal distribution of the TBB indexalong this longitude.

1) THE 30–60-DAY MODE

(i) The 850-mb streamline chartsIn order to save space, shown in Fig. 5 are only the

streamline and TBB anomaly charts of the two extreme

monsoon conditions when the SCS–u(850 mb) indexreaches its maximum (minimum) value and the differ-ences between them.

1) Total 850-mb winds. Because of the northwestwardextension of the North Pacific anticyclone’s ridgeline into northern China during the active monsoonphase (Fig. 5a), the MBF convection zone retreatsnortheastward to cover only the region stretchingfrom the Yangtze Delta to the northwest Pacific. Themonsoon trough (south of the North Pacific anti-cyclone) is deepened and stretched from northernIndochina to the Philippine Sea. The deepening ofthis trough enhances cumulus convection and alsostrengthens the monsoon westerlies associated with

1 MAY 2000 1381C H E N E T A L .

TABLE 1. Dates of the selected active and break phases of the South China Sea summer monsoon. Note that this monsoon exhibits threelife cycles. In order to distinguish every monsoon life cycle in the same monsoon season, we attach them chronologically to every activeand break monsoon phase.

Year Active1 Break1 Active2 Break2 Active3 Break3

197919801981198219831984198519861987198819891990199119921993

22 May20 May—20 May———19 May——22 May22 May13 May—20 May

10 Jun—27 May29 May—1 Jun

——3 Jun

16 May—5 Jun

22 May—3 Jun

3 Jul—8 Jun

10 Jun—22 Jun18 Jun8 Jul

13 Jun1 Jun

—20 Jun13 Jun23 Jun21 Jun

20 Jul15 Jun25 Jun16 Jun15 Jun—12 Jul24 Jul5 Jul

—2 Jul

14 Jul3 Jul

—29 Jun

11 Aug—19 Jul26 Jun16 Jul7 Aug7 Aug

13 Aug13 Jul31 Jul—26 Jul22 Jul——

28 Aug—25 Jul20 Jul30 Jul———1 Aug

20 Aug—10 Aug2 Aug

——

TABLE 2. Dates of the selected maximum and minimum u (850mb) indices in the South China Sea region.

Year Min u Max u Year Min u Max u

19791979197919791979198019801981198119811982198219821982198219821983198319831983198319831983198419841984198419841985198519851985

10 Jun28 Jun13 Jul10 Aug28 Aug12 May28 May26 Jun11 Jul26 Jul12 May27 May18 Jun08 Jul20 Jul03 Aug17 May31 May14 Jun02 Jul21 Jul02 Aug20 Aug16 May03 Jun03 Jul06 Aug24 Aug19 Jun07 Jul03 Aug18 Aug

21 Jun05 Jul

16 Aug

21 May

04 Jul19 Jul

20 May

28 Jun

28 Jul

23 May06 Jun24 Jun15 Jul

14 Aug

26 May18 Jun08 Jul18 Aug31 Aug29 Jun

11 Aug

198619861987198719871987198719871988198819881989198919891989199019901990199019911991199119911992199219931993199319931993

07 Jul25 Jul04 Jun22 Jun09 Jul24 Jul10 Aug26 Aug04 Jul22 Jul06 Aug14 May31 May15 Jun04 Jul07 Jun25 Jun21 Jul10 Aug28 Jun17 Jul02 Aug28 Aug14 Jun02 Jul14 Jun30 Jun16 Jul02 Aug17 Aug

15 Jul

13 Jun01 Jul16 Jul30 Jul20 Aug

12 Jul30 Jul

22 May08 Jun24 Jun

16 Jun

30 Jul

11 Jul25 Jul16 Aug

25 Jun

23 Jun09 Jul26 Jul09 Aug

them. During the break monsoon phase, the ridgeline of the North Pacific anticyclone (Fig. 5b) in-trudes southward into the northern part of the SCS.The MBF convection zone is extended southwest-ward to cover southern China in such a way that theSCS monsoon trough is filled and cumulus convec-tion is suppressed over the northern part of the South

China Sea and the Philippine Sea. Thus, the monsoontrough in the SCS becomes oriented southeastward(across the sea to reach Borneo) in such a way thatthe SCS convection zone is moved equatorward andthe monsoon westerlies are significantly weakened.

The opposite-phase intraseasonal variation of themonsoon indices in the SCS and East Asian regioncan be inferred directly from the comparison of themonsoon circulation between the two extreme mon-soon conditions. This comparison is accomplishedeasily by the difference between them (Fig. 5c). Likethe sDuDc pattern (Fig. 4a), the difference streamlinechart of DV(850 mb) between the active and breakmonsoon phases exhibits a double-celled structure.A meridional juxtaposition of the east–west-elon-gated positive and negative DTBB zones can be seenclearly, although major convective zones are asso-ciated with the SCS monsoon trough and the MBF.A reversal of the DV(850 mb) monsoon flow patternand the meridional juxtaposition of convection zonesare expected when the contrast of the two extrememonsoon conditions is reversed. It becomes evidentthat the northward (southward) shifts of the MBFand the convection zone associated with the MBFin eastern China follow the strengthening (weaken-ing) of the SCS monsoon westerlies and the en-hancement (suppression) of cumulus convectionalong the monsoon trough.

2) The 30–60-day filtered winds. The role played bythe 30–60-day monsoon mode is illustrated by thecomposite charts of the 30–60-day filtered stream-line, and DTBB anomalies corresponding to Figs. 4a–care displayed in Figs. 5d–f. The double-celled struc-ture of the DV(850 mb) streamline and the meridionaljuxtaposition of DTBB in Fig. 5c appear again in the[V(850 mb), DTBB] chart of the active monsoon phase(Fig. 5d). In contrast, the directions of the DV(850mb) flow pattern and of the meridional DTBB jux-taposition are reversed during the break monsoon


FIG. 5. Composite streamline charts superimposed with DTBB [5270 K 2 TBB $ 0 or 50 if 270 K 2 TBB , 0] for (a) activephase, (b) break phase, and (c) the difference between (a) and (b). The corresponding 30–60-day filtered streamline and DTBB

charts are displayed in (d)–(f ). Any 30–60-day cycle of the SCS u(850 mb) index $ 0.8s [s the standard deviation of the SCSu(850 mb) index] during any given summer is selected for composite. The composite procedure includes a 3-day window withthe centered date corresponding to date of the maximum or minimum u(850 mb) index. Values of 0 # DTBB # 15 K and 0 #DTBB # 58 are lightly stippled, while those of 15 K # DTBB and 5 K # DTBB are heavily stippled.

phase (Fig. 5e). Recall that the SCS summer mon-soon attains its maximum (minimum) intensity whena 30–60-day monsoon trough (ridge) arrives at 208Nwith a cyclonic (anticyclonic) cell and a maximum

(minimum) convection zone (Chen and Chen 1995).The reversal of the DV(850 mb) circulation patternand of the meridional DTBB juxtaposition betweenthe two extreme monsoon conditions (Figs. 5a,b) re-

1 MAY 2000 1383C H E N E T A L .

FIG. 6. Composite latitude–height cross sections of u(1208E) and u(1208E) correspond to Fig. 5. DTBB(1208E)and DTBB(1208E) are also displayed on each cross section with the scale on the right side of each. Contourintervals of u(1208E) and Du(1208E) are 2.5 and 1.5 m s21, respectively, while those of u(1208E) and Du(1208E)are 0.5 and 1 m s21, respectively. Positive values of all zonal-wind components are stippled.

sults from the alternative northward migration of the30–60-day monsoon trough and ridge from the equa-tor to near 208N. To strengthen this argument further,displayed in Fig. 5f is the difference in [V(850 mb),DTBB] between the two extreme monsoon conditions.The close resemblance of circulation and DTBB

anomalies between Figs. 5c and 5f indicates that the30–60-day monsoon mode is a major agent respon-sible for the change in synoptic conditions betweenthe active and break monsoon phases and, in turn,for the opposite-phase variation in the monsoon in-dices of the SCS and East Asian monsoons.

(ii) Latitude–height cross sections

The latitude–height cross sections of u(1208E) andu(1208E) superimposed with the corresponding TBB

anomalies are displayed in Fig. 6 for the following rea-

sons: 1) to understand the three-dimensional structuresof both the monsoon circulation and the 30–60-daymonsoon mode and 2) to attain a more quantitative mea-surement of the 30–60-day monsoon mode’s contribu-tion to the intraseasonal variation of monsoon indices.

1) Total zonal wind. The comparison between Figs. 6aand 6b shows that the intraseasonal changes of theu(1208E) cross section in the extratropics and Tropicsdiffer from each other. For the former region, thelower-tropospheric monsoon flow undergoes a dis-tinct and quasiperiodic oscillation between the twoextreme monsoon conditions, which are the MBFmonsoon westerlies centered at 408N (258N) in theactive (break) monsoon. This intraseasonal oscilla-tion went undetected by previous studies (Yeh et al.1959; Lau and Yang 1996), which stressed the rapidnorthward shifts of both the upper-level westerlies


and the location of the MBF during the Mei-Yu onsetin eastern China. Concerning the Tropics, a sharpcontrast between the lower SCS monsoon westerliesand the upper tropical easterly jet during the activemonsoon is present. Although during the break mon-soon, easterly flow dominates the entire troposphere.This flow regime change is consistent with that ofthe Indian monsoon (Chen and Yen 1991). As forcumulus convection, the active convection zones [in-dicated by the large DTBB(1208E) index (solid lines)]along the MBF and the SCS monsoon trough oscil-late coherently following these two monsoon ele-ments. In Fig. 5c, a well-organized cyclonic vortexwith a clear meridional juxtaposition of the convec-tion zones emerges from the change in monsoon flowbetween the two extreme conditions. The verticalstructure of this monsoon vortex, depicted by theDu(1208E) cross section in Fig. 6c, can reach as highas 300 mb. Apparently, the opposite-phase variationin the monsoon indices between the SCS and EastAsia (Fig. 1) and the in-phase variations between theSCS Du(850 mb) and the MBF location indices areattributed to the direction alternation of this monsoonvortex.

2) The 30–60-day filtered zonal wind. During the active(break) monsoon phase, a well-developed cyclonic(anticyclonic) vortex centered at about 208N emergesfrom the u(1208E) cross section [Fig. 6d] [(Fig. 6e)]and is coupled with positive (negative) DTBB anom-alies at its center near 408N. The intraseasonal al-ternations of the u(1208E) and TBB(1208E) cross sec-tions described in Figs. 6a,b are coincident withthose of the u(1208E) and DTBB(1208E) in Figs. 6d,e.This claim is further substantiated by the close re-semblance in the structures of D(u, DTBB)(1208E)(Fig. 6c) and D(u, DTBB)(1208E) (Fig. 6f), althoughamplitudes of the latter are only about half of theformer values. Evidently, the anomalous circulationand cumulus convection associated with the north-ward migration of the 30–60-day monsoon trough–ridge are major factors in causing the intraseasonalopposite-phase variation of the monsoon indices be-tween the two monsoons. The coherent intraseasonaloscillation observed by Chen and Murakami (1988)between the north–south oscillation of the MBF con-vection zone and the western tropical Pacific con-vection zone (ITCZ) is consistent with our analysisshown in Fig. 6.

2) THE 12–24-DAY MODE

The characteristics of the 12–24-day monsoon modediffer in several ways from those of the 30–60-day mon-soon mode. The 30–60-day monsoon trough–ridge mi-grates northward from near the equator to about 208N,while the 12–24-day monsoon low–high propagateswestward along two tracks (the equator and 158;208N)(Chen and Chen 1995; Chen and Weng 1997). Because

of its link with the monsoon life cycle, the 30–60-daymonsoon trough–ridge exhibits a quasiperiodical north-ward migration. Although the occurrence of the 12–24-day monsoon mode may not be quasiperiodic, this in-traseasonal mode plays a substantial role in the onsetand break of the SCS summer monsoon. Conceivably,the intraseasonal variations of monsoon indices (Figs.1 and 2) may be partially caused by the 12–24-daymonsoon mode. Despite the similar structure of the

and patterns (Fig. 4), it may not be legitimates sDuDc DuDc

to claim that the 12–24-day monsoon mode plays littleor no part compared to the 30–60-day monsoon mode.This concern is clarified by composite charts of V(850mb) streamlines and DTBB anomalies following the sameprocedure as in the 30–60-day monsoon mode shownin section 3b(1).

(i) The 850-mb streamline charts

The monsoon trough in the northern part of the SCSdeepens during the maximum SCS u(850 mb) index. Atthis time, the western part of the ridge line of the NorthPacific anticyclone extends into northern China, and aridge line appears near the equator (Fig. 7a). In contrast,during the minimum u(850 mb) index phase, the ridgeline of the North Pacific anticyclone intrudes south-westward into the northern SCS region, and the mon-soon trough becomes NW–SE oriented across the SCS(Fig. 7b). The synoptic change of the monsoon flowbetween the two extreme u(850 mb) indices is similarto that between the two extreme monsoon conditionsshown in Figs. 5a and 5b. The convection zones asso-ciated with the monsoon trough and the MBF undergoa similar meridional oscillation as these two monsoonelements do between the active and break SCS summermonsoon. The possible structure of disturbances causingthe variation of u(850 mb) and DTBB indices may beinferred from the patterns (Figs. 4c,f), but thesesDuDc

disturbances can be depicted well by the D[ ,y(850 mb)DTBB] chart between the two extreme u(850 mb) indicesin Fig. 7c. A well-defined double-celled structure showsup in this difference streamline chart, and a meridionaljuxtaposition of positive and negative DTBB zones is alsodiscernible. The direction alternation in the northern cellof the difference streamline chart and in the associatedconvection zones results in the opposite-phase varia-tions of the zonal wind and TBB indices between the twomonsoons with the timescale of 12–24 days.

For the 12–24-day monsoon mode (Figs. 7d–f), a cleardouble-celled structure, like the pattern, appears insDuDc

the DV(850 mb) streamline during both the maximumand minimum u(850 mb) indices, with the exception ofa reversal of anomalous circulation direction. Just as inDTBB anomalies shown in Figs. 5d and 5f, the DTBB anom-alies exhibit a north–south juxtaposition. The appearanceof the cyclonic (anticyclonic) DV(850 mb) northern cellfacilitates the northeastward (southwestward) retreat (in-trusion) of the ridge line out of (into) the northern SCS

1 MAY 2000 1385C H E N E T A L .

FIG. 7. Same as Fig. 5, except for the maximum and minimum phases of the SCS u(850 mb) index. The u(850 mb) cyclesselected for the composite are listed in Table 2.

and the deepening (filling) of the monsoon trough. Theeffect of the 12–24-day monsoon mode on the changeof monsoon flow is further substantiated by the structureresemblance between D[V(850 mb), DTBB] (Fig. 7c) andD[V(850 mb), TBB] (Fig. 7f). Conceivably, the 12–24-dayintraseasonal variations in the monsoon westerlies andcumulus convection of the two monsoon components can

be induced by the westward-propagating 12–24-day mon-soon mode.

(ii) Latitude–height cross section

During the maximum u(850 mb) index, the u(1208E)structure (Fig. 8a) is characterized by the following sa-


FIG. 8. Same as Fig. 6, except for the maximum and minimum phases of the SCS u(850 mb) index. Theu(850 mb) cycles selected for the composite are listed in Table 2.

lient features: the tropical monsoon westerlies, the trop-ical easterly jet, and the midlatitude westerlies coupledwith the MBF centered at about 408N. This structure issimilar to that of the active monsoon phase (Fig. 6a). Incontrast, during the minimum u(850 mb) index phase(Fig. 8b) the monsoon westerlies south of 208N disappear,and the midlatitude westerlies extend down to couplewith the MBF, just as the u(1208E) cross section behavesduring the break monsoon phase. The cumulus convec-tion is active in the maximum u(850 mb) index phase atabout 208 and 358N but is somewhat opposite and lessorganized during the minimum u(850 mb) index phase.

Shown by our previous study (Chen and Chen 1995),the vertical extent of the 12–24-day monsoon mode canreach to a level near 300;400 mb. A well-organizedu(1208E) vortex with this vertical extent (Fig. 8c) iscentered at about 208N. The amplitudes of u(1208E) andDTBB(1208E) are about two-thirds those of Du(1208E)and DTBB(1208E). The resemblances between cross sec-

tions of both Du(1208E) and Du(1208E) andDTBB(1208E) and DTBB(1208E) reveal that the intrasea-sonal changes of the monsoon zonal flow and cumulusconvection are caused by the westward propagation ofthe 12–24-day monsoon mode.

Recall that D[u, DTBB](1208E) in Fig. 6f may reachamplitudes of about one-half that of D[u, DTBB](1208E)(Fig. 6c). In spite of the differences in the propagationproperty and the occurrence frequency between the 30–60-day and 12–24-day monsoon modes, there is a phaselock between these two intraseasonal monsoon modesduring the SCS monsoon onset and break. Our analysisof 1979–93 in this study (not shown) reveals that thisphase lock occurs in more than 75% of the SCS mon-soon onsets and breaks. Compared to the amplitudes ofu and TBB anomalies, the 30–60-day and 12–24-daymonsoon modes contribute the most to the intraseasonalvariations in the Du and DTBB indices of the two mon-soons during the active and break monsoon phases.

1 MAY 2000 1387C H E N E T A L .

FIG. 9. Locations of extremely heavy rain events (amount exceed-ing 200 mm day21) are marked by different symbols according totheir magnitude. Cases with a rainfall amount larger than 200, 600,800, and 1000 mm day21 are marked by crosses, open circles, soliddots, and solid squares, respectively. This chart is extracted from Taoand Ding (1981, their Fig. 2) with only a slight modification.

4. Discussion

The East Asian summer monsoon is a complex cir-culation system that contains a number of interestingphenomena during its evolution (e.g., Lau and Li 1984;Tao and Chen 1987; Murakami 1987). The followingtwo phenomena are often discussed in literature.

1) During the Mei-Yu onset, the large-scale summercirculation over East Asia undergoes some abruptchanges (e.g., Yeh et al. 1959; Tao and Ding 1981).These changes consist of several stepwise northwardshifts of the upper-level westerlies over East Asiaand the northward shift of the lower-level MBF froma position south of the Yangtze River to north of theriver.

2) As shown in Fig. 9, three east–west-oriented locationzones of extremely heavy rain events emerge overthe eastern half of China (Tao and Ding 1981). Re-gardless of the mechanism generating the heavy rain,Tao and Ding suggested that these three longitudinalzones are related, during the Mei-yu onset, to thestepwise northward shifts of the upper-level west-erlies over East Asia and to the northward shift ofMBF across the Yangtze River.

Concerning the first phenomenon, the northward shiftof the upper westerlies (covering only a limited timeperiod of the entire monsoon season) is a part of theseasonal march of the East Asian circulation, which doesnot reverse its course until late summer. On the other

hand, the MBF oscillates meridionally in a coherentmanner with the SCS summer monsoon. Since the MBFshould couple in a certain way with the upper westerlies,it would be of interest to explore how the northwardshift of the MBF across the Yangtze River (in concertwith the tropical northward shifts of upper westerlies)during the Mei-Yu onset occurs within the context ofthe MBF’s north–south intraseasonal oscillation. Forthis purpose, let us use the coherent intraseasonal os-cillations of the filtered MBF location (thick dashedline) and the SCS u(850 mb) (thick solid line) indicesin Fig. 3 to illustrate the role played by the intraseasonalmonsoon modes in the northward shifts of the MBFlocation across the Yangtze River basin. The climato-logical onset date of the Mei-yu season in the YangtzeRiver basin is approximately 10–30 June (Tao and Chen1987). As indicated by the SCS u(850 mb) index (Fig.3a), the SCS monsoon westerlies reintensify in the sec-ond life cycle of this monsoon after 10 June. Accordingto observations at this stage by Chen and Chen (1995)a 30–60-day monsoon ridge reaches the northern SCSat about 208N, and a newly originated 30–60-day mon-soon trough emerges near the equator. The location in-dex shows that the MBF is located south of the YangtzeRiver. About a week to 10 days later, the MBF shiftedacross the Yangtze River. Accompanying the MBF’snorthward shift, the strong zonal wind associated withthis front (particularly below 500 mb) also moves north-ward [indicated by the u(1208E) cross section in Fig.6b] over the Yangtze River. Let us next compare theu(1208E) and u(1208E) cross sections in Fig. 6. Duringthe active (break) monsoon phase, it is inferred fromthe coincidence of negative (positive) u anomaliesaround 208;358N with the weak (strong) MBF west-erlies that the anomalous circulation associated with the30–60-day monsoon trough (ridge) plays a significantrole in the northward shift of the MBF across the Yang-tze River basin. This argument is further supported bythe resemblance of the Du(1208E) (Fig. 6c) andDu(1208E) (Fig. 6f) cross sections. For the 1989 summer(Fig. 3b), the zonal-wind monsoon index indicated thatthe intensity of the 12–24-day monsoon mode (revealedfrom the zonal-wind monsoon index) was stronger thanthat of the 30–60-day monsoon mode. As shown by thelocation index, the MBF moved rapidly across 308N inearly June primarily due to the 12–24-day monsoonmode [also inferred from the comparison between theu(1208E) and u(1208E) cross sections in Fig. 8].

Our discussion so far focuses only on the lower halfof the troposphere. Nevertheless, the northward shift ofupper westerlies observed by Yeh et al. (1959) for the1956 Mei-yu onset is not easily perceived from Figs. 6and 8. This northward shift of the upper westerlies as-sociated with the Mei-yu onset may not occur at theexact same latitudinal location and cover the samenorth–south extent in every summer monsoon season.It is likely that the northward shifts of upper westerlieswith the Mei-yu onset of 1979–93 are obscured by the


composite procedure. Yeh et al. (1959) stressed in theirstudy only the rapid northward shifts of the upper west-erlies during the Mei-yu onset. In contrast, we haveshown in this paper that the two intraseasonal monsoonmodes are vital to the coupling, that is, the opposite-phase oscillation between the lower-tropospheric cir-culations of the SCS and East Asian summer monsoons,which was neglected previously.

Recently, Lau and Yang (1996) suggested that duringthe East Asian monsoon onset, the rapid northward shiftof the upper-level westerlies over East Asia is linked tothe abrupt northward migration of the ascending branchof the local Hadley circulation over the SCS region.This migration may be induced by the symmetric in-stability of the basic flow in May. Actually, the north-ward migration of the ascending branch of the localHadley circulation can be coupled with that of the twointraseasonal monsoon modes (Chen and Chen 1995).Whether the northward migration of the 30–60-daymonsoon trough–ridge and the westward propagation ofthe 12–24-day monsoon mode are related to the sym-metric instability of the monsoon flow is beyond thescope of this study. Nevertheless, the discussion abovesupports the argument that the two intraseasonal mon-soon modes are indispensable components of the EastAsian monsoon to the northward shifts of the MBF lo-cation.

For the second phenomenon, the rainstorms in easternChina during the summer monsoon season are steeredby the upper westerlies to move along the MBF. Perhapsbecause of this, Tao and Ding (1981) suggested a pos-sible link between the three east–west-oriented locationzones (around 208, 308, and 408N) and the stepwisenorthward shift of the upper westerlies over eastern Chi-na; however, it was not made clear in their study howthese three zones of heavy rain events are linked to thenorthward shifts of the MBF and the upper westerliesduring the Mei-yu onset. Actually, these three locationzones presented by Tao and Ding’s heavy rain eventsmatch well with Lau et al.’s (1988) quasi-stationary lo-cations of three rainbands formed by multiple onsets ofthe East Asian summer monsoon. Lau et al.’s analysisstrongly indicates that the three location zones of heavyrain events in eastern China may not be linked only tothe Mei-yu onset. Therefore, some questions are raisedfrom the difference between the observations of thesetwo studies. Why should there be three location zonesfor heavy rain events? What is the dynamical mecha-nism responsible for the formation of these three zones?Answers to these two questions may be derived fromresults presented in sections 3 and 4.

Recall that during the active SCS monsoon phase thecenter and northern periphery of the intraseasonal mon-soon cyclonic vortex are located at about 208 and 358N(Fig. 6d), respectively. As inferred from DTBB(1208E)(Fig. 6a) or DTBB(1208E) (Fig. 6d), deep cumulus con-vection associated with the 30–60-day monsoon troughand along the MBF exists around the two aforemen-

tioned latitudinal locations. In contrast, the intraseasonalmonsoon vortex becomes anticyclonic during the SCSbreak monsoon phase (Fig. 6e). Although the anticy-clonic vortex center is still located at about 208N, themaximum 850-mb zonal wind of the MBF is shiftedsouthward to about 258N (Fig. 6b), and the southernperiphery of this vortex reaches the equator (Fig. 6e).At this stage, the deep cumulus convection positionedalong the MBF and associated with the 30–60-day mon-soon trough appears near the maximum 850-mb zonalwind and the southern rim of the vortex, respectively.As far as the 12–24-day monsoon mode is concerned,the convection zones associated with the vortex of thismonsoon mode (shown in Fig. 8) for the two extremeu(850 mb) indices alternate in the same manner as the30–60-day monsoon mode.

The locations of maximum DTBB (1208E) andDTBB(1208E) [or DTBB(1208E)] during the SCS activemonsoon phase [or the maximum SCS u(850 mb) phase]coincide with the latitudinal locations of Tao and Ding’s(1981) heavy rain zones around 208N and near 408N.Contrarily, the activation of the MBF during the breakSCS monsoon phase [or the minimum SCS u(850 mb)phase] enables heavy rain events to occur around theYangtze River (;308N). We cannot deny that heavy rainevents in eastern China during the Mei-yu onset maybe related to the northward shifts of the MBF and upperwesterlies. Since these shifts occur over only a limitedtime period during the East Asian monsoon season, thethree zones of heavy rain events over eastern China areunlikely to be the results of only these stepwise north-ward shifts. By engaging the following comparisons:

1) the zonal wind and DTBB cross sections at 1208E inFigs. 6 and 8,

2) the coherent intraseasonal oscillation of the SCSu(850 mb) and the MBF location indices in Fig. 3,and

3) the coherent intraseasonal oscillations of TBB(1408E)between the ITCZ and the MBF observed by Chenand Murakami (1988, their Fig. 1),

we are able to connect the zoning of the eastern Chinaheavy rain locations to the maximum convection zoneassociated with the MBF and the monsoon trough duringextreme SCS monsoon conditions indicated by theu(850 mb) index. In other words, the three heavy rainzones shown by Tao and Ding are a result of the in-traseasonal north–south oscillations of the MBF cou-pled with the northward migration of the intraseasonalmonsoon trough–ridge in the SCS region. This argumentsheds some light on the question posed above.

5. Concluding remarks

Although the establishment of the summer monsoonlife cycle by the 30–60-day and 12–24-day monsoonmodes in the SCS–WTP region has been documentedin previous studies, the possible links of this monsoon

1 MAY 2000 1389C H E N E T A L .

component to the monsoon over East Asia and to anylikely effect of the former monsoon component on thelatter monsoon component were not systematically ex-plored in the past. However, findings from several pre-vious studies lead us to hypothesize a link between themonsoons in these two regions.

1) The isoline of monsoon onset dates in East Asia,compiled by Tao and Chen (1987, their Fig. 3.9),shows a clear northward advancement of the EastAsian monsoon from southern to eastern China fol-lowing the northward seasonal march of the sun.

2) The TBB(1408E) y–t diagram of Chen and Murakami(1988) revealed that a coherent intraseasonal north–south oscillation exists between the MBF and theITCZ in the western North Pacific.

3) The Indian monsoon onset is triggered by the arrivalof the northward-migrating 30–60-day monsoontrough at 158;208N (Krishnamurti and Subrahman-yam 1982; Chen et al. 1988b). This migrating intra-seasonal monsoon trough extends eastward to thewestern tropical Pacific, causing the second life cycleof the SCS summer monsoon to synchronize withthe first life cycle of the Indian monsoon during theactive monsoon phase (Chen and Chen 1995). Sincethe Mei-yu onset follows the rapid northward shiftof the upper westerlies across the Tibetan Plateau,there should be some connection between the mon-soons in Southeast and East Asia.

4) A number of previous studies cited in the introduc-tion have observed the existence of 12–24-day and30–60-day intraseasonal oscillations in the EastAsian and SCS summer monsoons. The former in-traseasonal oscillation was found to be more re-gional, while the latter intraseasonal oscillation maybe coupled with the eastward-propagating Madden–Julian oscillation (MJO; Madden and Julian 1971,1972). In light of the coupling between the regional30–60-day monsoon mode and the global-scaleMJO, a coherent intraseasonal oscillation may existbetween the summer monsoons in East Asia and theSCS.

The NCEP–NCAR reanalysis data, OLR, and GMSTBB of Japan for the period 1979–93 were analyzed inthis study to seek the possible link between the summermonsoons in East Asia and the SCS. The major findingsof this study are as follows:

1) an opposite-phase intraseasonal oscillation of a) the850-mb monsoon zonal-wind index and b) the cu-mulus convection proxy (i.e., DTBB and DOLR) ofthe two monsoons components, and

2) in-phase intraseasonal oscillations of a) the MBF andthe ITCZ and b) the MBF latitudinal location andthe intensity of the SCS monsoon westerlies.

Regardless of the separation of the summer monsooncomponents in East Asia and the SCS by the NorthPacific anticyclone, these coherent intraseasonal oscil-

lations between the two monsoon components arecaused essentially by the intraseasonal flip-flop oscil-lation between the anomalous cyclonic and anticyclonicvortices associated with (a) the northward-migrating30–60-day monsoon trough and ridge, respectively, and(b) the westward propagating 12–24-day monsoon lowand high, respectively. Evidently, intraseasonal oscil-lation of summer monsoons in East Asia and the SCSis coupled through these monsoon vortices. Observa-tions made in the South China Sea Monsoon Experimentduring the summer of 1998 (Lau et al. 1998) providedhigh quality data over the Southeast–East Asian regionand offer a unique opportunity for us to test further thesuggested link between the intraseasonal oscillations ofthe summer monsoons in East and Southeast Asia.

Acknowledgments. This study was supported by theNational Science Foundation through the NSF GrantsATM-9416954 and ATM-9906454. A part of M.-C.Yen’s effort in Taiwan was supported by NSC Grant 89-2111-M-008-033 of Taiwan. Comments offered by Dr.K. M. Lau of Goddard Space Flight Center and tworeviewers were helpful in improving the presentation ofthe paper. The typing support by Mrs. Reatha Diedrichsand the editorial assistance by Mrs. Susan Carr are high-ly appreciated.

APPENDIX

Propagation Properties of the Two IntraseasonalMonsoon Modes and Averaged Power Spectra of

Monsoon Indices

a. Propagation properties of the two intraseasonalmonsoon modes

The northward propagation of the 30–60-day mon-soon mode and the westward propagation of the 12–24-day monsoon mode were depicted in terms of laggedcorrelation coefficient patterns (between the filtered850-mb zonal-wind monsoon index and streamfunction)by Chen and Chen (1995, their Figs. 9 and 15) for thesummer of 1979 and by Chen and Weng (1997) for 15summers during the period 1979–93. The lagged cor-relation coefficient pattern certainly provides a statis-tical summary for propagation properties and for thestructure of intraseasonal monsoon modes. However, thelagged correlation coefficient patterns are only on ex-pansion of the correlation coefficient patterns (shownin Fig. 4) in time. It may be informative to illustratethe propagation properties of the intraseasonal monsoonmodes by some direct approach.

Using the 30–60-day filtered 850-mb wind fields,Chen and Chen (1995) portrayed the 30–60-day mon-soon trough/ridge over the SCS–WTP region. The sameapproach was adopted to determine latitudinal locationsof these monsoon troughs/ridges in this study. A y–tdiagram of latitudinal locations of the 30–60-day mon-soon troughs (solid stars) and ridges (open circles) and


FIG. A1. The y–t diagram of latitudinal locations of the 30–60-day monsoon trough (star) and ridge (opencircle) and DTBB anomalies averaged between 1158 and 1208E over 15 summers (1979–93). Positive valuesof DTBB are stippled with a contour interval of 0.5 W m22.

FIG. A2. The summer (May–Aug) mean 850-mb streamlines and TBB superimposed byoccurrences of 12–24-day monsoon modes (identified by a closed center of the 12–24-dayfiltered streamlines) over 15 summers (1979–93).

DTBB anomalies averaged between 1158 and 1208E areshown for all SCS monsoon life cycles selected over 15summers (1979–93) by Chen and Weng (1997) in Fig.A1. Locations of the 30–60-day monsoon troughs andridges are not the same every summer season. However,as revealed from Fig. A1, the quasiperiodic alterationbetween the 30–60-day monsoon trough and ridge isobvious. The copresence of the 30–60-day monsoontrough and ridge is also true when one of them arrivesat 158;208N.

The occurrence of the 12–24-day monsoon mode inthe SCS–WTP region is identified by the streamlinecharts of the 12–24-day filtered 850-mb wind when aclosed low coincides with a DTBB center. The 12–24-day monsoon mode possesses a double-cell structurewith the same polarity. A clear view of this monsoonmode’s propagation track can be gained from the oc-currence frequency chart in Fig. A2. Low centers of the12–24-day monsoon mode identified every day aremarked with an open circle (solid triangle) for the north-ern (southern) track. The two-track propagation of the

12–24-day monsoon mode is apparent, but the SCSmonsoon is primarily affected by the northern track.This intraseasonal mode propagates westward and rarelyappears east of 1508E.

b. Averaged power spectra of monsoon indices

The u(850 mb), TBB, and OLR monsoon indices as-sociated with the SCS monsoon and the MBF were sub-jected to spectral analysis using a scheme proposed byMadden and Julian (1971, 1972). The power spectra (ofthese monsoon indices) averaged over 15 summers(1979–93) are displayed in Fig. A3. A predominant sig-nal of the 30–60-day mode emerges with a confidencelevel close to or higher than 99% (smooth solid curve).In addition to this intraseasonal mode, there is still anoticeable signal within the period between 12 and 24days. To make this signal more discernible in the powerspectra of monsoon indices, we applied simple harmonicanalysis [as Chen and Chen (1995) suggested] to ex-clude signals with periods longer than 30 days from the

1 MAY 2000 1391C H E N E T A L .

FIG. A3. Power spectra of u(850 mb), TBB, and OLR monsoon indices in the SCS and the MBFaveraged over summers of the period indicated in each. The total power spectra are displayedwith solid lines. To make the 12–24-day mode discernible, the power spectra of residual timeseries by removing signal with time period longer than 30 days are shown in dashed lines. Thesmooth, solid, and dashed lines represent the confidence level of 99% for their correspondingspectra.

monsoon indices. The 12–24-day signal with a confi-dence level of 99% (smooth dashed curve) stands outin the power spectra (dashed lines) of the residual timeseries.

REFERENCES

Akiyama, T., 1973: The large-scale aspects of the characteristic fea-tures of the Baiu front. Pap. Meteor. Geophys, 24, 157–188.

Arkin, P. A, and P. E. Ardanuy, 1989: Estimating climatic-scale pre-cipitation from space: A review. J. Climate, 2, 1229–1238.

Chen, T.-C., 1987: 30–50 day oscillation of 200-mb temperature and850-mb height during the 1979 northern summer. Mon. Wea.Rev., 115, 1589–1605.

, and M. Murakami, 1988: The 30–50 day variation of convectiveactivity over the western Pacific Ocean with emphasis on thenorthwestern region. Mon. Wea. Rev., 116, 892–906., and M.-C. Yen, 1991: Intraseasonal variations of the tropicaleasterly jet during the 1979 northern summer. Tellus, 43A, 213–225., and J.-M. Chen, 1995: An observational study of the SouthChina Sea monsoon during the 1979 summer: Onset and lifecycle. Mon. Wea. Rev., 123, 2295–2318., and S.-P. Weng, 1997: A further observational study of theSouth China Sea summer monsoon: Climatology and interannualvariation. Tech. Report, Dept. of Geological and AtmosphericSciences, Iowa State University, 41 pp. [Available from T.-C.Chen, Atmospheric Science Program, 3010 Agronomy Hall,Iowa State University, Ames, IA 50011.], and , 1998: Interannual variation of the summer synoptic-


scale disturbance activity in the western tropical Pacific. Mon.Wea. Rev., 126, 1725–1733., M.-C. Yen, and M. Murakami, 1988a: The water vapor transportassociated with the 30–50 day oscillation over the Asian mon-soon regions during 1979 summer. Mon. Wea. Rev., 116, 1983–2002., R.-Y. Tzeng, and M.-C. Yen, 1988b: Development and life cycleof the Indian monsoon: Effect of the 30–50 day oscillation. Mon.Wea. Rev., 116, 2183–2199.

Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Re-analysis Project. Bull. Amer. Meteor. Soc., 77, 437–471.

Krishnamurti, T.-N., 1985: Summer monsoon experiment—A review.Mon. Wea. Rev., 113, 1590–1626., and D. Subrahmanyam, 1982: The 30 to 50 day mode at 850mb during MONEX. J. Atmos. Sci., 39, 2088–2095., P. K. Jayakumar, J. Sheng, N. Surgi, and A. Kuma, 1985:Divergent circulations on the 30 to 50 day timescale. J. Atmos.Sci., 42, 364–375.

Lau, K. M., and M. T. Li, 1984: The monsoon of East Asia and itsglobal associations—A survey. Bull. Amer. Meteor. Soc., 65,114–125., and P. Chan, 1986: Aspects of the 40–50 day oscillation duringthe northern summer as inferred from outgoing longwave ra-diation. Mon. Wea. Rev., 114, 1354–1367., and S. Yang, 1996: Seasonal variation, abrupt transition, andintraseasonal variability associated with the Asian summer mon-soon in the GLA GCM. J. Climate, 9, 965–985., G.-J. Yang, and S. H. Shen, 1988: Seasonal and intraseasonalclimatology of summer monsoon rain over East Asia. Mon. Wea.Rev., 116, 18–37., H. T. Wu, and S. Yang, 1998: Hydrological processes associatedwith the first transition of the Asian summer monsoon: A pilotsatellite study. Bull. Amer. Meteor. Soc., 79, 1871–1882.

Madden, R. A., and P. R. Julian, 1971: Detection of a 40–50 dayoscillation in the zonal wind in the tropical Pacific. J. Atmos.Sci., 28, 702–708.

, and , 1972: Description of global-scale circulation cellsin the Tropics with a 40–50 day period. J. Atmos. Sci., 29, 1109–1123.

Murakami, M., 1979: Large-scale aspects of deep convective activityover GATE area. Mon. Wea. Rev., 107, 997–1013.

Murakami, T., 1959: The general circulation and water vapor balanceover the Far East during the rainy season. Geophys. Mag., 29,131–171., 1972: Equatorial tropospheric waves induced by diabatic heatsources. J. Atmos. Sci., 29, 827–836., 1987: Effects of the Tibetan Plateau. Monsoon Meteorology,C. P. Chang and T. N. Krishnamurti, Eds., Oxford UniversityPress, 235–270.

Nitta, T., 1987: Convective activities in the tropical western Pacificand their impact on the Northern Hemispheric circulation. J.Meteor. Soc. Japan, 65, 373–390., and S. Sekine, 1994: Diurnal variation of convective activityover the tropical Pacific. J. Meteor. Soc. Japan, 72, 627–641.

Susskind, J., and J. Pfaendtner, 1989: Impact of interactive physicalretrievals on NWP. Report on the Joint ECMWF/EUMETSATWorkshop on the Use of Satellite Data in Operational WeatherPrediction: 1989–1993, T. Hollingsworth, Ed., Vol. 1, EuropeanCentre for Medium-Range Weather Forecasts, 245–270.

Tao, S.-Y., and Y.-H. Ding, 1981: Observational evidence of the in-fluence of the Oinhiu-Xizang (Tibet) plateau on the occurrenceof heavy rain and severe storms in China. Bull. Amer. Meteor.Soc., 62, 23–30., and L. S. Chen, 1987: A review of recent research on the EastAsian summer monsoon in China. Monsoon Meteorology, C. P.Chang and T. N. Krishnamurti, Eds., Oxford University Press,60–92.

Yeh, T.-C., S.-Y. Dao, and M.-C. Li, 1959: The abrupt changes ofcirculation over the Northern Hemisphere during June and Oc-tober. The Atmosphere and the Sea in Motion, Scientific Con-tributions to the Rossby Memorial Volume, Rockefeller Institute,249–267.

Interaction between the Summer Monsoons in East Asia and the South China Sea: Intraseasonal Monsoon...

Documents

Transcript of Interaction between the Summer Monsoons in East Asia and the South China Sea: Intraseasonal Monsoon...