Abstract

In this paper, the authors present a description of the internal dynamics and boundary forcing characteristics of two major subcomponents of the Asian summer monsoon (ASM), that is, the South Asian monsoon (SAM) and the East–Southeast Asian monsoon (EAM). The description is based on a new monsoon-climate paradigm in which the variability of ASM is considered as the outcome of the interplay of a “fast” and an “intermediate” monsoon subsystem, under the influence of “slow” external forcings. Two sets of regional monsoon indices derived from dynamically consistent rainfall and wind data are used in this study. Results show that the internal dynamics of SAM are representative of a “classical” monsoon system in which the anomalous circulation is governed by Rossby wave dynamics, where anomalous vorticity induced by an off-equatorial heat source is balanced by the advection of planetary vorticity. On the other hand, the internal dynamics of EAM are characterized by a “hybrid” monsoon system featuring multicellular meridional circulation over the East Asian sector, extending from the deep Tropics to the midlatitudes. These meridional cells link tropical heating to extratropical circulation systems via the East Asian jet stream and are responsible for the observed zonally oriented anomalous rainfall patterns over East and Southeast Asia and the subtropical western Pacific. In the extratropical regions, the major upper-level vorticity balance is between the advection and generation by anomalous divergent circulation and basic-state circulation. A consequence of the different dynamical underpinnings is that EAM is associated with stronger extratropical teleconnection patterns to regions outside ASM compared to SAM.

The interannual variability of SAM is linked to basin-scale SST fluctuation with pronounced signals in the equatorial eastern Pacific. During the boreal spring, warming of the Arabian Sea and the subtropical western Pacific may lead to a strong SAM. For EAM, interannual variability is tied to SST anomalies over the East China Sea, the Sea of Japan (East Sea), and the South China Sea regions, while the linkage to equatorial basin-scale SST anomaly is weak at best. A strong EAM is foreshadowed by a large-scale SST anomaly dipole with warming (cooling) in the subtropical central (eastern) Pacific.

Comparison with the P. J. Webster and S. Yang (WY) monsoon index shows that WY is not significantly correlated with either the SAM or EAM regional-scale rainfall separately. It is demonstrated that WY can be considered as a measure of the large-scale atmospheric circulation state over the Indian/Pacific Ocean basin, including the integrated heat source over the ASM region. As such, the regional monsoon indices developed in this paper and WY provide a complementary description of the broadscale and regional aspects of the ASM.

1. Introduction

It is well known that the Asian summer monsoon (ASM) is an extremely complex phenomenon that encompasses variabilities over a wide range of spatial and temporal scales. Prediction of the monsoon is one of the major challenges of climate research. Recently, numerous studies have suggested that the El Niño–Southern Oscillation (ENSO) may have an influence on the year-to-year variability of the ASM (e.g., Rasmusson and Carpenter 1983; Webster and Yang 1992; Ju and Slingo 1995; Lau and Bua 1998; Yang and Lau 1998). These studies have also shown that many extreme ASM droughts and floods were not linked to ENSO. By some estimate, more than half of the major anomalies in the ASM are not related to ENSO, suggesting that there are other factors that may contribute to the predictability (or lack thereof) of the ASM (Webster et al. 1998).

Heuristically, variabilities of the ASM can be classified in terms of those due to internal dynamics and those forced by boundary conditions such as sea surface temperature (SST) and land surface conditions (Shukla 1984; Palmer 1994). For seasonal-to-interannual predictions, the former constitutes the chaotic or unpredictable component, while the latter, the potentially predictable component of the ASM. The degree to which the ASM is controlled by each component determines the overall predictability of the system. One difficulty encountered in the predictability of the ASM is that monsoon variability itself cannot be easily quantified because this variability is a strong function of space and time and depends highly on the parameters used. For long-term monsoon variability, many authors used the all-India monsoon rainfall (AIMR) as an index for the ASM (e.g., Mooley et al. 1986; Parthasarathy et al. 1992; Lau and Yang 1996). However, because the Indian subcontinent covers only a small fraction of the entire ASM region, the Indian monsoon may not be representative of the ASM as a whole. Likewise, results from studies using East Asian rainfall (e.g., Shen and Lau 1995; Nitta and Hu 1996; Weng et al. 1999) may not be applicable to other regions of the ASM. During the warm event of 1997, rainfall averaged over the entire Indian subcontinent was near normal. Yet, the monsoon over the East Asian and the Southeast Asian regions was extremely abnormal with record-breaking droughts and floods occurring in the different parts of East Asia (C.-Y. Li 1998, personal communication). This is so, despite that the Indian monsoon is known statistically to have a stronger ENSO influence compared to the East Asian monsoon (Ropelewski and Halpert 1988). The need for better defining the components of Asian summer monsoon in order to resolve the relationship of ASM with ENSO and with other global climate anomalies is therefore paramount. In this paper, we will present preliminary results and provide a framework for a classification of regional components of the ASM in terms of their dynamical and boundary-forcing characteristics.

2. Preliminary considerations

In an attempt to quantify monsoon climate variability, many authors have used different monsoon indices to measure different components of the ASM (Webster and Yang 1992; Goswami et al. 1999; ,Wang and Fan 1999; ,Ailikun and Yasunari 2000, manuscript submitted to J. Meteor. Soc. Japan; Chen et al. 2000). The present paper is an attempt to synthesize and to expand on some of the concepts developed in previous studies.

a. A monsoon climate paradigm

To begin, we propose a new paradigm for classification of the ASM based on the characteristics of its internal dynamics and relationship to local and remote boundary forcing functions. Figure 1 encapsulates a monsoon-centric view of the climate system including its subcomponents. This schematic presumes a basic knowledge of the ASM. For a detailed description of the individual phenomena, the readers are referred to various review papers on the ASM in the literature (e.g., Lau and Li 1984; Krishnamurti 1985; Webster et al. 1998). At the center of the climate system shown in Fig. 1 is the ASM that, for completeness, includes both the summer and the winter components, labeled here as the annual cycle. The ASM is affected by a “fast” component, depicted in the lower left-hand corner of the schematic. This is the internal dynamic subsystem that includes moisture recycling and hydrologic feedback in the atmosphere. This fast system is dominated by synoptic (3–5 days) and subsynoptic cloud scales (<1 day) and can be identified with monsoon depressions, easterly waves, convective cloud clusters, and mesoscale complexes within the ASM regions. Additionally, the monsoon climate is strongly affected by surface conditions in the adjacent oceans and landmasses, represented as the “intermediate” subsystem in the lower right-land corner of Fig. 1. This subsystem possesses both fast and slow (weeks to months) timescales. The key elements of this subsystem consist of SST, wind, and surface humidity that determine heat and moisture fluxes over the adjacent oceans. Also included as elements in the “intermediate” subsystem are soil moisture, vegetation, and snow cover that control the heat and moisture fluxes over the land regions of ASM.

Fig. 1.

Schematic showing the various components and interlinkages of a monsoon-centric climate system.

Fig. 1.

Schematic showing the various components and interlinkages of a monsoon-centric climate system.

Connecting the fast and intermediate subsystems is the “intraseasonal variability” that comprises the Madden–Julian oscillation and other atmospheric low-frequency variability such as the biweekly oscillation, which has been linked to onset, breaks, and revival of the ASM. Both the fast and intermediate subsystems are under the influence of remote forcings from planetary-scale phenomena such as ENSO and other long-term secular variations of the global climate system from decadal variability to global warming. These remote forcings constitute the “slow” subsystem that is considered as external to the ASM, but yet can play an important role in altering the basic states of the monsoon atmosphere, the adjacent oceans, and land surfaces. These altered states may then lead to modulations of the probability distribution of monsoon states within the fast component. It is also plausible that the coupling of the fast and intermediate subsystems may influence the slow system. These two-way interactions are represented by the linkages labeled “teleconnections” in Fig. 1. As an example, SST anomalies in the Indian Ocean and the western Pacific are coupled to those in the eastern equatorial Pacific through east–west overturnings in the equatorial atmosphere and oceans. Another example is the possible linkage between the equatorial east–west SST gradient across the Pacific basin and the meridional thermocline gradient through the gyre circulation and/or subduction processes originating in the extratropical oceans (Gu and Philander 1995; Liu and Huang 1997). These gradients define the basic oceanic state that governs the long-term behavior of ENSO–monsoon interactions.

In the aforementioned paradigm, the interannual predictability of the monsoon climate is determined by the strength of the linkages provided by the intermediate and slow subsystems. The predictability of the monsoon climate evolves as the result of an interplay between the relative influence of the fast, intermediate, and slow subsystems. The stronger the control by the fast system, the less predictable the ASM. The more influence by the slower subsystems, the more the ASM is predictable. In this view, the predictability of the monsoon evolves as a function of the slowly varying basic states and therefore will wax and wane over long timescales (Mehta and Lau 1997).

b. Monsoon climate indices

A first step toward achieving a better understanding of monsoon predictability is the development of quantitative measures to characterize the ASM climate and its subsystems. This requires the construction of appropriate measures of the different and closely related aspects of the ASM climate. Up to now, there have been several monsoon indices that are commonly used for the ASM based either on rainfall or on circulation (Mooley and Parthasarathy 1984; Webster and Yang 1992; Shen and Lau 1995). Most of these indices are developed based on different considerations including data availability, dynamical consistency, and regional relevance. In the following, we summarize the main attributes that need to be considered for a monsoon climate index in the context of the aforementioned paradigm. They are

  • representativeness of the annual cycle as well as the variability (interannual or longer) of the ASM or its subcomponents;

  • relevance in regional (geographical, because of society impacts) as well as global context;

  • simplicity for construction, preferably from direct observables such as rainfall, wind, temperature, and moisture;

  • internal dynamical consistency, if different parameters are used;

  • robustness and reproducibility by different datasets (for the same variables); and

  • expandability to include long-term, multidecadal timescales.

Obviously, because of practical constraints, not all the conditions can be met. In this paper, we have considered these conditions, to the extent possible, in our choice of monsoon indices and the descriptions based on the indices. We need to stress that the monsoon indices that we used in this paper are not meant to supplant, but rather to complement, previous monsoon indices. The main focus of this work is not on the development of new monsoon indices but rather on the better definition of regional aspects of the ASM.

c. Data description

To establish the monsoon rainfall climatology and variability, we used 18 yr (1979–96) of global rainfall estimate from the Climate Prediction Center Merged Analysis of Precipitation (CMAP) (Xie and Arkin 1997). This dataset is derived from multiple sources including gauge observations, satellite infrared (IR) and microwave estimates, and the National Center for Environmental Predictions–National Atmospheric Research Center (NCEP–NCAR) reanalysis. For comparison, we have also included the Global Precipitation Climatology Project (GPCP) version 1 combined precipitation dataset (Huffman et al. 1997) and other commonly used satellite rainfall products including the Microwave Sounding Unit (MSU) and the Geosynchronous Operational Environmental Satellites Precipitation Index (GPI). The GPCP combined dataset covering the period 1987–96 has a similar data source to the CMAP but differs in the treatment of the blending of the satellite and the in situ observations. The MSU satellite rainfall has been merged with rain gauge data, and the GPI is an IR-based satellite-only rainfall estimate.

For the development of the monsoon circulation indices and associated dynamical characteristics, we used 40 yr (1958–97) of the wind reanalysis from NCEP–NCAR. This is the only multidecadal reanalysis dataset available at the time of this study. A description of the dataset can be found in Kalnay et al. (1996). Other datasets used include SST from NCEP, National Aeronautics and Space Administration (NASA) Goddard Earth Observing System reanalysls, European Centre for Medium-Range Weather Forecasts (ECMWF) reanalysis, and the all-India monsoon rainfall.

3. Regional monsoon characteristics

a. Rainfall and circulation climatologies

In this section, we describe the development of regional monsoon climate indices that will be used in subsequent analysis to characterize the climate variability of the ASM and its subcomponents. We begin with a description of the climatological rainfall and circulation fields. Even though the basic climatological features of the ASM are well known, they are shown here to provide references for later discussions.

The CMAP (Fig. 2a) rainfall climatology shows clearly the broad features of monsoon precipitation over the warm pool of the Indian Ocean and the Pacific. Four monsoon rain centers, each anchored to the west side of the subcontinental or maritime landmasses, can be identified near the South China Sea/Philippines region, southern Indochina, Bay of Bengal, and the west coast of the Indian subcontinent. Climatological CMAP rainfall amounts in these regions exceed 16 mm day−1, with a maximum over the Bay of Bengal in excess of 20 mm day−1. These features are also found in GPCP (Fig. 2b), whose peak amplitude is about 10%–20% smaller than that of CMAP. The MSU (Fig. 2c) and the GPI (Fig. 2d) products yield heavier precipitation over the Bay of Bengal but less well-defined signals over the other centers. The first two centers over the South China Sea and Indochina appear to be linked to an extensive region of precipitation over the western Pacific intertropical convergence zone and to a branch of moderate rain over the East China Sea across southern Japan. This can be identified as the model manifestation of the well-known Mei-yu or Baiu rainband. In the CMAP climatology, this monsoon rainband is very weak. The oceanic component of the Mei-yu is better defined in GPCP and in MSU, but weaker and almost absent in GPI. This may be due to the larger proportion of warm rain in the Mei-yu rain, which produces a signature in the microwave and the IR fields that is different from that due to deep tropical convection. Interestingly, all four rainfall products indicate that there is a region of enhanced precipitation just south of the equator over the central Indian Ocean during the peak summer months. The presence of this midoceanic precipitation is consistent with the upper-level divergence field (see Fig. 3d).

Fig. 2.

JJA monsoon rainfall climatology based on (a) CMAP, (b) GPCP, (c) MSU, and (d) GPI. Units are in mm day−1. Contour interval is 2 mm day−1. Areas with rain rate exceeding 6 mm day−1 are shaded.

Fig. 2.

JJA monsoon rainfall climatology based on (a) CMAP, (b) GPCP, (c) MSU, and (d) GPI. Units are in mm day−1. Contour interval is 2 mm day−1. Areas with rain rate exceeding 6 mm day−1 are shaded.

Fig. 3.

JJA climatology of the Asian monsoon circulation: (a) 850-mb streamline, (b) 200-mb streamline, (c) 200-mb vorticity, and (d) 200-mb divergence. Units and contour intervals are as indicated.

Fig. 3.

JJA climatology of the Asian monsoon circulation: (a) 850-mb streamline, (b) 200-mb streamline, (c) 200-mb vorticity, and (d) 200-mb divergence. Units and contour intervals are as indicated.

Figures 3a and 3b show the June–August (JJA) 850- and 200-mb large-scale circulation patterns associated with the ASM. Evident in the large-scale flow is the low-level interhemispheric gyre circulation (IGC) that is characterized by easterlies over the southern Indian Ocean (about 10°S), connected via the Somali jet to the westerlies across the Indian subcontinent. This IGC evolves into the southwesterly flow over the South China Sea and East Asia. Worth noting is a closed cyclonic circulation over the Indian Ocean just south of the equator, between 70° and 90°E. This feature is dynamically consistent with the mid-Indian Ocean precipitation in the Southern Hemisphere noted in the rainfall climatologies shown in Fig. 2 and in previous reports (Krishnamurti 1985). Another dominant circulation pattern is the large-scale low-level anticyclonic flow associated with the western Pacific subtropical high (WPSH). This circulation regime features prevailing low-level easterlies in the tropical western Pacific, northerlies from the South China Sea to central East Asia, and southwesterlies across the Korean peninsula and southern Japan. The IGC and the WPSH are the two dominant low-level circulation regimes that control major fluctuations of the entire ASM.

At 200 mb, the monsoon circulation is dominated by the Tibetan high, which has two centers, one over the northern Bay of Bengal and the other over the northern Arabian Sea (Fig. 3b). In the region from 60° to 120°E, and from 10°N to the equator, a strong vertical shear is noted with strong upper-level easterlies overlying low-level westerlies. The 200-mb vorticity field (Fig. 3c) shows that the basic state has strong negative (anticyclonic) vorticity in a zonally oriented broad band across the monsoon region, with a center to the east of the Tibetan Plateau. There are strong vorticity gradients with opposite signs to the north and the south of the subtropical jet stream near 40°N and 20°–30°N, stretching zonally across the whole domain. The vorticity distribution, which is mainly due to the variation of the basic-state zonal wind with latitude, will have an important implication in determining the dynamical characteristics of anomalous monsoon circulation (see discussions in sections 4 and 5). The 200-mb climatological divergence (Fig. 3d) has maximum centers over the Bay of Bengal, the eastern Arabia Sea, the South China Sea, and over the equatorial Indian Ocean. These are in agreement with the precipitation centers noted in Fig. 2. The results provide further reassurance that the NCEP wind fields and the satellite rainfall fields are dynamically consistent.

b. Regional monsoon indices

As discussed above, the ASM precipitation centers are strongly controlled by two large-scale low-level circulation systems, that is, the IGC and the WPSH. In the following, we use the rainfall averaged over two areas, (10°–30°N, 70°–100°E) and (5°–25°N, 100°–130°E), as a starting point for studying monsoon climate variability governed by these two circulation systems (see outlined boxes in Fig. 2a). The areas are chosen based on considerations of geography and coherent rainfall systems. We have experimented with different box sizes used in defining the monsoon rainfall indices and found that provided the major climatological convective centers are included, the results are not very sensitive to the domain size. In the following, the JJA rainfall time series over the above regions are denoted by PSA and PSEA, respectively, where SA stands for South Asia and SEA for Southeast Asia. It is important to note that in our definition of the South Asia rainfall, the convection over the Bay of Bengal is included, in addition to that over the land region of the Indian subcontinent. For the Southeast Asian rainfall, the averaging domain covers the South China Sea/western Pacific and Indochina. As we shall show later, the two areas represent distinct source regions of monsoon variability in South Asia and East–Southeast Asia, respectively. The rainfall patterns associated with the two rainfall indices are also similar to the dominant modes obtained by singular value decomposition of the regional ASM rainfall and SST (Lau and Wu 1999).

Table 1 shows that the PSA is significantly correlated to AIMR with a correlation coefficient r equal to 0.43. The main difference between PSA and AIMR is that PSA includes the Bay of Bengal convection that should be considered as part of the South Asia monsoon subsystem. Note that PSEA is uncorrelated with either AIMR (r = 0.06) or PSA (r = 0.03), indicating a distinct rainfall regime. The rainfall that is averaged over both the PSA and the PSEA domains has a higher correlation with PSEA than with PSA.

Table 1.

Cross correlation of PSA, PSEA (both derived from CMAP), and AIMR. Correlation exceeding 99% (95%) significance is bold-faced (in parentheses).

Cross correlation of PSA, PSEA (both derived from CMAP), and AIMR. Correlation exceeding 99% (95%) significance is bold-faced (in parentheses).
Cross correlation of PSA, PSEA (both derived from CMAP), and AIMR. Correlation exceeding 99% (95%) significance is bold-faced (in parentheses).

Because a reliable rainfall estimate over the oceanic regions does not exist before the era of satellites, it is necessary to find dynamically consistent circulation indices that share the same regional information contained in PSA and PSEA so as to extend the monsoon indices back in time to increase samples. To identify circulation features that are most strongly correlated with PSA and PSEA, we have obtained one-point correlation maps of the anomalous (with seasonal cycle removed) zonal wind, meridional wind, and the associated wind shears with PSA and PSEA, respectively. For convenience, only the fields that are used in the subsequent definition of monsoon circulation indices will be shown. These fields display the most coherent and well-defined features associated with each of the rainfall time series. Figure 4a shows the correlation pattern of the meridional wind shear (υ850mbυ200mb) with PSA. The most prominent feature is a large area of positive shear signaling a direct meridional circulation with low-level poleward and upper-level equatorward flows over the Indian subcontinent. Coupled to this meridional circulation is a reversed meridional circulation over northern China and the western Pacific east of the Philippines. In contrast to PSA, the most well-defined feature correlated with PSEA is an alternation of upper-level easterly and westerly in the longitudinal sector from 110° to 150°E (Fig. 4b). In the extratropics, the zonal regimes span the entire Eurasian sector, with a secondary maximum to the northwest of the Himalayas. The association of the East Asian upper-level jet with tropical convection over the East Asian sectors has been noted in a number of previous studies (e.g., Lau and Li 1984; Yang and Webster 1990; Liang and Wang 1998). Based on the correlation maps shown in Figs. 4a and 4b, we define two regional monsoon (RM) circulation indices as

 
formula

where the zonal and meridional winds are averaged over the geographic regions as indicated (outlined boxes in Fig. 4). Based on the NCEP reanalysis, the above indices have a record length of 40 yr, thus permitting a reasonable statistical estimate of the ASM variability. We should point out that RM1 is identical to the monsoon index proposed by Goswami et al. (1999). However, both RM1 and RM2 are significantly different from the monsoon index used by Webster and Yang (1992, hereafter referred to as WY). Table 2 shows a preliminary comparison among RM1, RM2, and WY with respect to PSA and PSEA. A more detailed comparison will be provided in section 5. By design, the correlation of RM1 to PSA (r = 0.64) and RM2 to PSEA (r = 0.74) is highly significant at the 99% confidence level. It is important to note that the cross correlation between RM1 and PSEA, and between RM2 and PSA is very small, confirming that the indices PSA/RM1 and PSEA/RM2 indeed represent distinct regional and dynamical phenomena. The regional context of the monsoon indices is supported by the significant correlation (r = 0.53) between RM1 and the AIMR, while there is negligible correlation (r = 0.13) between RM2 and AIMR. Note also that the WY index has only poor correlation with AIMR (r = 0.24) and PSA (r = 0.18), respectively. It has higher correlation with PSEA (r = 0.42), which just falls short of the 95% confidence level. This result suggests that WY is not an appropriate regional index for the Indian monsoon and Bay of Bengal rainfall variability. On the other hand, the WY index appears to be significantly correlated with rainfall anomalies summed over South and Southeast Asia, PSEA+SA. This is consistent with the recent findings of Ailikun and Yasunari (2000, manuscript submitted to J. Meteor. Soc. Japan). Further comparison of RM1, RM2, and WY will be presented in section 6.

Fig. 4.

Correlation patterns of (a) PSA and meridional wind shear (850–200 mb), and (b) PSEA and 200-mb zonal wind. Boxes indicate regions where the averaged values of the winds are taken to construct the monsoon circulation indices, RM1 and RM2. Contour interval is 0.2. Areas with correlation exceeding the 95% confidence level are shaded.

Fig. 4.

Correlation patterns of (a) PSA and meridional wind shear (850–200 mb), and (b) PSEA and 200-mb zonal wind. Boxes indicate regions where the averaged values of the winds are taken to construct the monsoon circulation indices, RM1 and RM2. Contour interval is 0.2. Areas with correlation exceeding the 95% confidence level are shaded.

Table 2.

Same as in Table 1 except for the correlation among AIMR, RM1, RM2, and WY.

Same as in Table 1 except for the correlation among AIMR, RM1, RM2, and WY.
Same as in Table 1 except for the correlation among AIMR, RM1, RM2, and WY.

4. Dynamical characteristics

This section is devoted to a more detailed description of the dynamical characteristics of the two monsoon subsystems, that is, the South Asian monsoon and the East–Southeast Asian monsoon, as depicted by RM1 and RM2.

a. The South Asian monsoon (SAM)

To illustrate the regional context of RM1, the correlation map of RM1 with precipitation anomalies at every grid point in the greater ASM region is shown in Fig. 5a. As expected, the areas of largest positive rainfall anomalies are found over South Asia, including the Indian subcontinent, the Bay of Bengal, northern Indochina, and southern East Asia. Negative rainfall anomalies are found in the surrounding oceanic areas. The strongest negative anomalies appear over the southeastern Indian Ocean, the South China Sea, and the subtropical and tropical western Pacific. Associated with the positive rainfall anomalies are strong low-level westerly and southwesterly flows over the Arabian Sea and the Bay of Bengal (Fig. 5b). This circulation pattern suggests an enhancement of the climatological IGC (Fig. 3a). Two branches of cross-equatorial flow can be identified: one along the east coast of Africa and the other from southeastern Indian Ocean. The latter appears to be the main supply of the low-level inflow into the Bay of Bengal. At the upper level, a prominent anticyclone center is found to the northwest of the positive heating (Fig. 5c). This can be identified as the regional manifestation of the well-known South Asia high. The circulation pattern is also consistent with the induced Rossby-type circulation by latent heating located off the equator (Gill 1980; Lau and Lim 1982).

Fig. 5.

Spatial distribution of the regression of RM1 with (a) CMAP rainfall (mm day−1), in 0.2 contour interval, (b) 850-mb wind (m s−1), and (c) 200-mb wind (m s−1). Areas with correlation exceeding the 95% confidence level are shaded.

Fig. 5.

Spatial distribution of the regression of RM1 with (a) CMAP rainfall (mm day−1), in 0.2 contour interval, (b) 850-mb wind (m s−1), and (c) 200-mb wind (m s−1). Areas with correlation exceeding the 95% confidence level are shaded.

The above-mentioned circulation features are linked to well-defined low-level anticyclone circulation over the coastal region of East Asia (Fig. 5b). This anticyclone represents an enhancement and westward extension of the WPSH, which is associated with surface easterlies and subsidence over the South China Sea region (see discussion in section 4b). Figure 5c also shows an upper-level anticyclone located to the northeast of the low-level anticyclone, directly over northern Korea (near 40°N, 120°E). As we shall demonstrate later, the dynamic of the anomalous circulation over this region is underpinned by barotropic interaction with the climatological westerly jet.

Focusing back on the sector of the Indian subcontinent, Fig. 6 shows the meridional vertical cross sections of the regression of zonal, meridional wind, and vertical motion fields associated with RM1, averaged between the longitudes 70°–110°E. The zonal wind cross section (Fig. 6a) indicates a coupling of the upper easterlies to low-level westerlies over the main ASM region (10°–20°N). Over the equator weak easterly wind coupled with tropospheric easterly shear can be found. Poleward of the ASM region, an alternation of barotropic easterly and westerly patterns is developed, consistent with the flow around the South Asia high. The strongest signal of RM1 is found in the meridional circulation field (Fig. 6b). Here, a strong negative (equatorward) meridional flow coupled to middle to low-level poleward flow is evident. Over the Southern Hemisphere, the meridional winds reverse direction and are much reduced in scope. The meridional winds are a part of the main branch of the local Hadley circulation that spawns large-scale ascending motion over the South Asian region (10°–25°N) and descending motion at 30°–40°N and 10°S–0° (Fig. 6c).

Fig. 6.

Latitude–height cross section showing covariance of RM1 with (a) zonal wind, (b) meridional wind, and (c) vertical velocity. Contour intervals are 0.2 m s−1 for zonal wind, 10−5 mb s−1 for vertical velocity, and 0.1 m s−1 for meridional wind.

Fig. 6.

Latitude–height cross section showing covariance of RM1 with (a) zonal wind, (b) meridional wind, and (c) vertical velocity. Contour intervals are 0.2 m s−1 for zonal wind, 10−5 mb s−1 for vertical velocity, and 0.1 m s−1 for meridional wind.

In summary, RM1 portrays a “classical” monsoon circulation system associated with latent heating over the Indian subcontinent and adjacent oceans including the Bay of Bengal. RM1 also depicts significant precipitation and circulation anomalies in the East and Southeast Asian regions, suggesting that the two monsoon subsystems are dynamically linked.

b. The East–Southeast Asia monsoon (EAM)

The regional context of RM2 is shown in the correlation map of precipitation with RM2 in Fig. 7a. In contrast to RM1, the precipitation pattern shows an east–west banded structure, indicating alternate dry and wet conditions over the subtropical western Pacific and East Asia, with a very weak signal over the Indian subcontinent. The meridional extent of the monsoon-related rainfall anomalies is very large, spanning from the equator to 50°N. In Fig. 7b, the low-level circulation field indicates a strong influence from the WPSH in the subtropical region. When convection is strongly developed over the Southeast Asia/South China Sea region, low-level westerly flow prevails from the southern Bay of Bengal to the western Pacific, across Indochina and the central South China Sea. The circulation pattern signals a northward advance of the WPSH over East Asia. The easterly inflow near 25°–30°N is associated with local sinking motion and the returning westerly flow is linked to the rising motion near 35°–45°N (see Fig. 8c). At 200 mb (Fig. 7c), the most prominent feature is the presence of an intensive anticyclone, whose main effect is to cause the axis of the climatological subtropical jet to migrate northward by about 10°–15° in latitude. This represents a remarkable response of the subtropical upper-level flow to tropical heating in the Southeast Asian region. The East Asian anticyclone is connected upstream to an anomalous anticyclone to the northwest of northern India, indicating again that the East Asian monsoon is dynamically linked to its South Asian counterpart.

Fig. 7.

Spatial distribution of the regression of RM2 with (a) CMAP rainfall (mm day−1), in 0.2 contour interval, (b) 850-mb wind (m s−1), and (c) 200-mb wind (m s−1). Areas with correlation exceeding the 95% confidence level are shaded.

Fig. 7.

Spatial distribution of the regression of RM2 with (a) CMAP rainfall (mm day−1), in 0.2 contour interval, (b) 850-mb wind (m s−1), and (c) 200-mb wind (m s−1). Areas with correlation exceeding the 95% confidence level are shaded.

Fig. 8.

Latitude–height cross section showing covariance of RM2 with (a) zonal wind, (b) meridional wind, and (c) vertical velocity. Contour intervals are the same as Fig. 6 except the contour interval is 0.5 m s−1 for the zonal wind component.

Fig. 8.

Latitude–height cross section showing covariance of RM2 with (a) zonal wind, (b) meridional wind, and (c) vertical velocity. Contour intervals are the same as Fig. 6 except the contour interval is 0.5 m s−1 for the zonal wind component.

Figure 8 shows vertical cross sections of the covariance of the wind fields, which further illustrate the fundamentally different dynamical structure of EAM versus SAM. The dominant feature in the zonal wind cross section (Fig. 8a) is a barotropic structure signaling a northward shift of the upper-tropospheric jet stream, which creates a strong local anomalous vorticity gradient. The shift in the zonal wind is associated with strong poleward flow between the surface and 300 mb and equatorward flow above 300 mb, just south of the zonal wind maximum (Fig. 8b). This indicates a strong meridional circulation transverse to the climatological jet (near 40°N, see Fig. 3b). This transverse meridional flow advects basic-state vorticity from the south to the north of the climatological jet while transporting momentum poleward. A more detailed comparison between the vorticity balance for SAM and EAM will be presented in section 4c.

In the tropical region, characteristics of the “classical” monsoon circulation featuring upper-level easterlies–equatorward flows and low-level westerlies–poleward flows can still be discerned up to 20°N (Figs. 8a and 8b). However, the local latent heat-induced Hadley circulation appears to be squeezed into a narrow latitudinal span with rising motion centered near 20°N and subsiding motions at 10° and 30°N, as shown in Fig. 8c. Here, the multicellular structure of the meridional vertical motion field in EAM is in sharp contrast to the broad ascent in SAM. Because of the relatively smaller magnitude of the rainfall in the extratropics compared to that in the Tropics, the anomalous ascent centered at 40°N is unlikely to be caused by latent heating, but rather induced by meridional wind advection of vorticity associated with the northward migration of the jet stream (see section 4c).

c. Vorticity balance and teleconnection

The aforementioned dynamical characteristics of SAM and EAM are further illustrated by examining the vorticity balance over these regions. The major contribution to the vorticity anomaly (defined with respect to the climatological mean) at 200 mb is given by the vorticity equation.

 
formula

where an overbar represents a climatological mean and a prime denotes an anomaly from the mean. The symbols have their usual meanings. The residual represents contribution by twisting terms and by vertical motions, which are not considered here. Figure 9 shows the time series of the two most dominant terms of the vorticity equation averaged over the RM1 and RM2 domains. For RM2, both the lower and higher latitude domains are included (see Fig. 4b). The root-mean-square (rms) values of all terms shown in Eq. (3) are shown in Table 3. The sign in front of the rms value is determined by the sign of the correlation coefficient with the βν′ term in each domain.

Fig. 9.

Time series of dominant terms of the vorticity equation for 200-mb flow in (a) the RM1 region, (b) the subtropical region of RM2, and (c) the extratropical region of RM2.

Fig. 9.

Time series of dominant terms of the vorticity equation for 200-mb flow in (a) the RM1 region, (b) the subtropical region of RM2, and (c) the extratropical region of RM2.

Table 3.

Rms values of major terms contributing to the 200-mb vorticity balance in Eq. (3) in different regions. The sign of the correlation of each term with respect to the planetary vorticity balance (βν′) is shown inside a bracket to indicate the sense of the balance. The bold-faced values indicate the most dominant terms for each domain.

Rms values of major terms contributing to the 200-mb vorticity balance in Eq. (3) in different regions. The sign of the correlation of each term with respect to the planetary vorticity balance (βν′) is shown inside a bracket to indicate the sense of the balance. The bold-faced values indicate the most dominant terms for each domain.
Rms values of major terms contributing to the 200-mb vorticity balance in Eq. (3) in different regions. The sign of the correlation of each term with respect to the planetary vorticity balance (βν′) is shown inside a bracket to indicate the sense of the balance. The bold-faced values indicate the most dominant terms for each domain.

For the RM1 region, Fig. 9a shows an approximately inverse variation between the meridional advection of planetary-scale vorticity and the generation by the divergence of anomalous flow. The rms values of the two terms, as indicated in the first row of Table 3, are about the same magnitude but with opposite sign. These suggest the vorticity balance is of the Sverdrup type, that is,

 
βν′ + (
ζ
+ f) · V′ ≈ 0.
(4)

Table 3 also shows a close balance between the two advection terms (columns 3 and 4 in the table) so that the total advection effect in this region is effectively small. The vorticity balance in (4) also supports the use of the meridional wind as an index for the local heat source for SAM.

In the RM2 region, near the climatological position of the East Asian jet (Fig. 3b), the advection by the mean circulation becomes important. Figure 9b shows that there is clearly an inverse relationship between the vorticity generation and advection by the basic-state divergent circulation. Table 3 (second row) shows that the major vorticity balance is shared among three terms, that is,

 
βν′ + (
ζ
+ f) · V′ +
V
 · ζ′ ≈ 0.
(5)

From Fig. 9c and Table 3 (third row), it can be seen that poleward of the climatological jet stream position (in the northern domain of RM2 shown in Fig. 4b), the dominant balance is between the vorticity generation induced by divergence and the advection of the anomalous vorticity by the mean circulation, that is,

 
(
ζ
+ f) · V′ +
V
 · ζ′ = 0.
(6)

Given the different vorticity balance between SAM and EAM, one important question to ask is whether the two monsoon subsystems sustain similar or different teleconnection patterns. Figure 10 shows the correlation maps of RM1 and RM2 with the 300-mb geopotential field. Areas with correlation coefficients equal to or exceeding the 95% significance level are shaded. In RM1 (Fig. 10a), the anticyclone over northwestern India due to heating over South Asia appears as a part of an elongated ridge system extending from northern Africa. The South Asian anticyclone is linked to a high over northeastern Asia. As discussed previously, this high is associated with the fluctuation of the WPSH. The teleconnection pattern of RM2 is broadly similar to that of RM1 but much stronger and more well defined (Fig. 10b). This feature indicates that the fluctuation in RM2 is associated with a strong meridionally oriented trough–ridge system spanning the coastal region of East Asia, Japan, and the central North Pacific. This system is apparently connected to the north–south migration of the subtropical jet stream associated with heating over the South China Sea and western Pacific region. The similarities between the RM1 and RM2 teleconnection patterns in the Asian and western Pacific sectors indicate that the northward shift of the jet stream and the fluctuation of the WPSH may be tied to the development of the South Asian high. Similar wave train signals are also noted over North America, downstream of the North Pacific anomalies. The result that similar downstream teleconnection signals are excited independent of the sources suggests that the teleconnection pattern may be manifestation of intrinsic barotropic modes in the northern summer basic-state flow (Lau and Peng 1992). The RM2 teleconnection pattern is also similar to the Pacific–Japan pattern associated with tropical heating near the Philippines and zonally banded convective pattern over the subtropical western Pacific reported by Nitta (1987).

Fig. 10.

Correlation map showing the teleconnection pattern of the 300-mb geopotential height associated with (a) RM1 and (b) RM2. Units are nondimensional with a contour interval of 0.1.

Fig. 10.

Correlation map showing the teleconnection pattern of the 300-mb geopotential height associated with (a) RM1 and (b) RM2. Units are nondimensional with a contour interval of 0.1.

5. SST relationships

As discussed in section 2, the dynamical structure of the monsoon climate is strongly influenced by intermediate and slow processes, that is, lower boundary conditions in both the adjacent regions and those far from the monsoon region. Whereas the largest interannual basinscale SST anomalies (SSTAs) occur during ENSO, these SSTAs are not necessarily the maximal (in the sense of providing the most sensitivity) in influencing ASM variability. For regional monsoon prediction, it is necessary to identify first the geographical locations of SSTAs that are most effective in causing monsoon fluctuations and then ask the question how these SSTAs relate to ENSO. In this section, we discuss the sensitivity of SSTA forcing for SAM and EAM.

a. SSTA–monsoon covariability

Figure 11 shows the contemporaneous and lagged covariances of RM1 with SSTA over the Indo-Pacific region. The label MAM (DJF) indicates SST patterns one season (two seasons) preceding SAM. The areas exceeding 95% significance level are shaded. A strong SAM is preceded in the northern winter by a dipole SSTA pattern in the North Pacific with a positive anomaly over the western North Pacific and a negative anomaly over the eastern North Pacific (Fig. 11a). During the preceding MAM, the SSTA is similar to that of a La Niña with a cold tongue in the eastern equatorial Pacific, a positive SSTA in subtropical mid-Pacific, and a negative SSTA in the eastern North Pacific (Fig. 11b). Notice that the Arabian Sea and the Bay of Bengal have an above-normal SST at this time. The JJA correlation indicates that the subtropical SSTA further strengthens and becomes anchored to the East China Sea and the Sea of Japan. A strong SAM is associated with simultaneous cooling of the western Indian Ocean along the coast of Somalia. This may reflect the enhanced oceanic upwelling due to increased monsoon wind. The eastern Pacific SSTA becomes increasingly well developed in JJA (Fig. 11c). Except for the western Pacific and the Indian Ocean, the JJA SSTA projects strongly on the canonical SSTA pattern observed during La Niña. We have also computed the same SSTA correlations with AIMR. The basic SSTA patterns (not shown) remain unchanged. The main difference is that the subtropical western Pacific SSTA pattern is much weaker than that shown in Fig. 11. Since the major difference between AIMR and RM1/PSA is in the inclusion of the convection over the Bay of Bengal, the result suggests that the Bay of Bengal convection may play a key role in linking the ASM and the EAM.

Fig. 11.

Covariance map of RM1 (JJA) with SST of (a) preceding DJF, (b) preceding MAM, and (c) contemporaneous JJA. Contour interval is 0.05°C. Shaded areas indicate confidence level exceeding 95%.

Fig. 11.

Covariance map of RM1 (JJA) with SST of (a) preceding DJF, (b) preceding MAM, and (c) contemporaneous JJA. Contour interval is 0.05°C. Shaded areas indicate confidence level exceeding 95%.

The SST correlation with RM2 shows that EAM is associated with distinctly different SST anomaly patterns (Fig. 12). Enhanced convection over Southeast Asia is preceded by a well-defined dipole of an SSTA pattern, in the previous cold season, in the central subtropical Pacific, oriented in a southwest–northeast direction (Fig. 12a). In MAM, the SSTA pattern expands but remains essentially stationary (Fig. 12b). Before a strong EAM, warming is noticed along the coastal region of China, while overall cooling occurs in the Indian Ocean. In contrast to the SAM, the EAM appears to be most sensitive to the mid-Pacific SSTA dipole pattern, and there is only a weak resemblance to the El Niño SST signal in equatorial western Pacific. During JJA, the maximum positive SST anomaly is locked on to the extratropics of the western Pacific, encompassing the Yellow Sea, the Sea of Japan, and the central North Pacific (Fig. 12c). The cooling of the Indian Ocean has spread to cover the South China Sea and the western Pacific up to 25°N. The sharp north–south SST gradient in the subtropical western Pacific appears to be coupled with the rainfall and circulation anomalies (see Fig. 7), with warm water overlaid by dry surface easterlies and with cooler water to the south overlaid by wet surface westerlies. It is noted that while the overall SSTA pattern has some resemblance to La Niña in the eastern Pacific, areas of strongest SSTA are found over the oceans adjacent to the monsoon landmasses. The importance of the SSTA in monsoon regions, in addition to that in the Niño-3 region, in affecting ASM variability has been noted by and Soman and Slingo (1997).

Fig. 12.

Same as Fig. 11 except for RM2.

Fig. 12.

Same as Fig. 11 except for RM2.

The JJA correlation patterns of SST (Fig. 12c), rainfall (Fig. 7a), and circulation (Fig. 7b) over the subtropical western Pacific are suggestive of strong ocean–atmosphere feedback associated with the fluctuation of the WPSH and convection over the South China Sea, that is, the PSEA region. One possible scenario is that the increased convection over the South China Sea–Philippines region enhances subsidence in the subtropical dry zone near 30°–35°N, resulting in cloud-free sky SST (see Fig. 7a). Increased insolation in the dry zone may lead to the rapid shoaling of the oceanic mixed layer and warming of the surface waters of the subtropical western Pacific. The increased anticyclonic flow from the WPSH will bring more moisture to southern China and return a drier air mass over the dry/warm water zone, thus accelerating the SST warming in that region. On the other hand, the extensive cooling of the Indian Ocean and the South China Sea in JJA is likely due to the increased upwelling and surface evaporation. The differential heating and cooling of the subtropical western Pacific and the Indian Ocean/South China Sea may increase the north–south thermal contrast, inhibiting (favoring) convection over the South China Sea (western subtropical Pacific) and thus providing a negative feedback on EAM. Note that while SSTA patterns associated with RM1 and RM2 are quite different in MAM, they both show a strong SSTA signal over the subtropical western Pacific in JJA, suggesting the SSTA in that region may be a common link to both SAM and EAM.

b. Annual cycle–ENSO phase relationship

Since the monsoon is a key contributor to the annual cycle variation of the tropical ocean–atmosphere and the ENSO signal is tied to the annual cycle, it is instructive to view the relationship between the monsoon and basin-scale SST in the context of the entire annual cycle. Figure 13 shows the phase portraits of the regional monsoon indices with respect to the Niño-3 SST. In this figure, the solid curve represents the 40-yr climatological annual cycle trajectory that describes the annual variations of the monsoon index and Niño-3 SST. The composite annual cycles for six warm events (dotted trajectory) and six cold events (dot–dashed trajectory) are also shown. The interannual variability of the monsoon index is marked by the shaded area around the annual cycle trajectory, based on one standard deviation from the climatology. The annual cycle trajectory indicates that South Asian monsoon activity, as reflected in RM1, commences in April when the Niño-3 SST is warmest (Fig. 13a). The monsoon strength rapidly increases from May through July and reaches its peak in July, while the Niño-3 SST falls as the cold tongue develops. Note that RM1 changes sign from negative to positive between May and June, signaling the establishment of the direct monsoon meridional cell with low-level poleward and upper-level equatorward flows over South Asia. The SAM withdraws very rapidly from August to December when the cold tongue is well developed and the Niño-3 SST remains below 25°C. The reversal of the local meridional cell, which is one of the defining characteristics of SAM, can be seen in the negative extremes of RM1 in the cold season DJFM. Note that the interannual variability of RM1 is much larger in DJF than in JJA. During warm events, the annual cycle is considerably reduced with RM1 reaching a lower peak value in July and the annual swing in Niño-3 SST less than 1.5°C (compared to about 2.5°C in the climatological mean). In contrast, during cold events, the annual cycle is much accentuated with a peak-to-peak swing of about 4.5°C. Here, RM1 is noticeably enhanced. The annual trajectories of the cold and warm events are clearly distinct from the climatological trajectory, evidencing the impact of remote SSTA forcing on the course of the evolution of the regional monsoon with respect to Niño-3 SST.

Fig. 13.

Phase portrait of the annual cycles of (a) RM1 vs Niño-3 SST and (b) RM2 vs Niño-3 SST. Shaded region indicates 1σ bound of RM1 and RM2, respectively. Units are in m s−1 for the regional monsoon indices and °C for SST. Solid, dotted, and dot–dashed lines indicate climatology, warm events, and cold events, respectively.

Fig. 13.

Phase portrait of the annual cycles of (a) RM1 vs Niño-3 SST and (b) RM2 vs Niño-3 SST. Shaded region indicates 1σ bound of RM1 and RM2, respectively. Units are in m s−1 for the regional monsoon indices and °C for SST. Solid, dotted, and dot–dashed lines indicate climatology, warm events, and cold events, respectively.

For RM2 (Fig. 13b), the annual cycle variation with respect to the Niño-3 SST is similar to that of RM1. Here, the abrupt transition from negative to positive RM2 (northward jump of the subtropical jet) occurs from June to July, reaching a peak in August. This is consistent with the different stages in the development of the regional monsoon precipitation system, that is, the Mei-yu in China, Changma in Korea, and Biau in Japan. During warm events (dotted trajectory), the annual cycle amplitude of SST is substantially reduced compared to cold events (dot–dashed trajectory). There is a small but noticeable increase of RM2 in the months of June–September from warm to cold events. A noteworthy feature is that the advance of the EAM from April through June appears to be delayed, while the Niño-3 SST cools substantially. During cold events, the monsoon transition from June to July in RM2 is very marked.

6. Further discussions

In this section, we present results of a comparison of the basic properties of the regional monsoon indices developed in the previous sections to those associated with WY. As noted before, WY is used by many recent investigators as an index for the interannual fluctuations of the Asian summer monsoon.

a. Rainfall and circulation associated with WY

The WY monsoon index is constructed from the zonal wind shear between the 850- and 200-mb wind in the region of the climatological JJA upper-tropospheric easterly jet overlying low-level westerlies, spanning the Indian Ocean and the western Pacific (0°–20°N, 40°–110°E). Figure 14a shows the correlation of WY with the CMAP rainfall over the ASM region. It is obvious that WY captures aspects of both the RM1 and RM2 rainfall patterns (Figs. 5a and 7a). It depicts the east–west banded rainfall pattern in the western Pacific similar to RM2. It also depicts the inverse relationship between rainfall over the equatorial Indian Ocean and the Indian subcontinent. However, the Bay of Bengal and the Indian subcontinent no longer appear as a major center of action. Notice that the region of WY encompasses all the major climatological heat sources discussed in section 2. In a broad sense, the WY appears to capture the wind response associated with the integrated heat source spanning the Indian Ocean to the western Pacific, with perhaps stronger contribution from the western Pacific. The results here are in agreement with those of Goswami et al. (1999) and Ailikun and Yasunari (2000, manuscript submitted to J. Meteor. Soc. Japan), who found that WY is not significantly related to Indian rainfall but more to convection over the western Pacific.

Fig. 14.

Spatial distribution of the regression of WY with (a) CMAP rainfall (mm day−1), in 0.2 contour interval, (b) 850-mb wind (m s−1), and (c) 200-mb wind (m s−1). Areas with correlation exceeding the 95% confidence level are shaded.

Fig. 14.

Spatial distribution of the regression of WY with (a) CMAP rainfall (mm day−1), in 0.2 contour interval, (b) 850-mb wind (m s−1), and (c) 200-mb wind (m s−1). Areas with correlation exceeding the 95% confidence level are shaded.

The circulation pattern associated with WY is also shown in Fig. 14. At 850 mb, WY depicts an elongated band of westerly wind anomaly centered at 10°N, stretching from the east coast of Africa to the South China Sea (Fig. 14b). Also noted is a band of easterly anomaly from the date line to 140°E, forming a convergent center near (0°N, 120°–140°E). At 200 mb, WY is linked to the South Asia high with easterly flow over the Indian Ocean and the Arabian Sea (Fig. 14c). A downstream wave train pattern over East Asia and Japan is also noted. The circulation pattern associated with WY is somewhat similar to that of RM1, but the spatial scale appears much larger, encompassing the entire Indian Ocean and much of the western Pacific. The SST correlation patterns (Fig. 15) show evolution features that resemble those of the ENSO cycle with the positive phase of WY coinciding with a cold event. Strong SST anomalies are concentrated over the two extreme ends of the Indo-Pacific basin, that is, the equatorial eastern Pacific, the western Indian Ocean, and the Arabian Sea (Fig. 15c). Comparing the associated rainfall, circulation, and SST fields, WY appears to possess aspects of both RM1 and RM2, but lacks in regional details.

Fig. 15.

Covariance of WY with SST of (a) preceding DJF, (b) preceding MAM, and (c) simultaneous JJA. Areas that exceed the 95% significance level are shaded. Contour interval is 0.05°C.

Fig. 15.

Covariance of WY with SST of (a) preceding DJF, (b) preceding MAM, and (c) simultaneous JJA. Areas that exceed the 95% significance level are shaded. Contour interval is 0.05°C.

b. Intercomparison of monsoon indices

As suggested in preceding discussions, RM1 and RM2 differ from WY in that the former two are measures of the local features of subcomponents, that is, SAM and EAM, while WY is representative of the broadscale features. These differences are reflected in the temporal characteristics of the indices. The autocorrelation functions of RM1, RM2, WY, and the Southern Oscillation index (SOI) for different seasons are shown in Fig. 16. It is clear that RM1 and RM2 have very little seasonal memory on their own, as indicated by the rapid decline in the autocorrelation after one season (Figs. 16a and 16b). This is consistent with the notion that these are regional indices in which internal dynamics play an important role in determining their intrinsic temporal properties. On the other hand, the WY index (Fig. 16c) shows a much slower decline, with significant autocorrelation up to two or three seasons. Notice that both the SOI and WY suggest a hint of a “spring predictability barrier” (Webster and Yang 1992;Lau and Yang 1996). Anomalies for MAM appear to have the longest memory, remaining quite high after two seasons (Figs. 16c and 16d). For all other seasons, the memory decreases rather abruptly in passing through MAM. Thus, it appears that WY has similar temporal characteristics to the SOI and is more likely to be dependent upon the basin-scale SST variations than RM1 and RM2.

Fig. 16.

Autocorrelation functions for different seasons for (a) RM1, (b) RM2, (c) WY, and (d) SOI. Units are nondimensional. The initial seasons are DJF (solid), MAM (dotted), JJA (dashed), and SON (dashed–dot).

Fig. 16.

Autocorrelation functions for different seasons for (a) RM1, (b) RM2, (c) WY, and (d) SOI. Units are nondimensional. The initial seasons are DJF (solid), MAM (dotted), JJA (dashed), and SON (dashed–dot).

Previous studies have suggested the presence of two distinct climatic states in the monsoon–ocean–atmosphere system, the transition between them is characterized by an intrinsic quasi-biennial timescale (Meehl 1993; Yasunari and Seki 1992; Shen and Lau 1995; Lau and Yang 1996; Clarke et al. 1998). It is therefore instructive to compare and contrast the three monsoon circulation indices (RM1, RM2, and WY) in such a context. Figure 17 shows the lagged correlation–longitude sections depicting the time evolution of basin-scale SST and the Walker circulation, defined as the difference (850 mb minus 200 mb) between the linearly regressed wind vector with respect to the three indices along the equator of the Indo-Pacific basin. Regions where the correlation between the monsoon index and SSTA is over the 95% confidence level are shaded. Wind vectors are shown only if at least one of the wind components exceeds the 95% confidence level. Figure 17a shows that a strong RM1 coincides with the developing stage of the cold phase of the equatorial eastern Pacific, accompanied by a weak divergent Walker circulation with lower-level easterlies (upper-level westerlies) over the eastern Pacific. The anomalous circulation is reversed over the Indian Ocean. The cool phase becomes fully established over the central Pacific in DJF following a strong RM1. Conversely, a weak RM1 is associated with a developing warm event. Notice that the SST signal is stronger after than before the monsoon. This is consistent with the well-known characteristics that abnormal monsoon activities over India may presage by about two seasons basin-scale SST anomalies in the Pacific region (Shakla and Paolino 1983; Lau and Yang 1996). Also noteworthy is that the cold phase lasts from about APR (0) through APR (+1). This is in agreement with the intrinsic quasi-biennial timescales in the South Asian monsoon found in previous studies (e.g., Meehl 1997). Consistent with the discussion in previous sections, Fig. 17b shows that the relationships of RM2 with basin-scale SSTA and the Walker circulation are insignificant and marginal at best. As shown in section 5, atmospheric activities associated with RM2 are linked to a subtropical Pacific SST dipole, which appears to be distinct from the warm and cold states in the equatorial Indo-Pacific Ocean. In contrast to RM1 and RM2, WY depicts a very strong relationship with the equatorial basin-scale SST variations, with a strong Walker circulation coupling the Indian Ocean and the eastern Pacific (Fig. 17c). Notice that a strong WY is linked contemporaneously to the cold water of the equatorial Pacific Ocean and the relationship has a maximum in JJA. Yet JJA is not the time of strongest development of SST in the eastern Pacific due to El Niño. Contrast this with RM1 that displays a 6-month lag, which is typical of monsoon–ENSO relationship modulated by the annual cycle. Hence, Fig. 17c represents the component of WY due to fluctuation of the equatorial Walker circulation that is tightly coupled to the variation of east–west SST gradient across the Indo-Pacifc basin.

Fig. 17.

Time–longitude section of lagged covariance of monsoon circulation index with basin-scale SST anomalies and wind components. Vectors shown indicate the difference (850 mb minus 200 mb) of the regressed wind at the two levels. Results are shown for (a) RM1, (b) RM2, and (c) WY. Shaded areas indicate statistical significance exceeding 95% level. Contour interval is 0.1°C.

Fig. 17.

Time–longitude section of lagged covariance of monsoon circulation index with basin-scale SST anomalies and wind components. Vectors shown indicate the difference (850 mb minus 200 mb) of the regressed wind at the two levels. Results are shown for (a) RM1, (b) RM2, and (c) WY. Shaded areas indicate statistical significance exceeding 95% level. Contour interval is 0.1°C.

c. Reproducibility and reliability

This subsection provides an assessment of the reproducibility of the monsoon indices when different datasets are used. Figure 18 shows the time variation of rainfall indices PSA and PSEA using the four different rainfall datasets mentioned in section 2. During the period in which at least three of the datasets overlap (1987–96), it can be seen that the variations of the four datasets seem to agree. All indices indicate a weak SAM during 1987, 1989, and 1993, and a strong SAM in 1988 and 1994. The GPI rainfall estimate is consistently the highest, while GPCP data has the lowest rainfall estimate compared to others. Except for a systematic bias of about 2 mm day−1, the CMAP and GPCP variations are quite similar in both regions. Before 1987, the MSU and the CMAP differ significantly in magnitude and variations for both PSA and PSEA. The MSU, being a single satellite, may have suffered from problems arising from orbit changes and instrument drift during the above period (Hurrell and Trenberth 1992). This problem is significantly reduced for CMAP because of the blending of multiple satellites and sensors (Xie and Arkin 1997). Therefore, it is reasonable to assume that CMAP will have higher reliability than MSU in the period prior to 1987.

Fig. 18.

Time series based on CMAP, MSU, GPCP, and GPI for the period 1979–96 for (a) the South Asian region and (b) the Southeast Asian region. Units are in mm day−1.

Fig. 18.

Time series based on CMAP, MSU, GPCP, and GPI for the period 1979–96 for (a) the South Asian region and (b) the Southeast Asian region. Units are in mm day−1.

The regional monsoon circulation indices RM1 and RM2 derived from NCEP, ECMWF, and GEOS for the overlapping period from 1980 to 1994 are shown in Figs. 19a and 19b, respectively. For RM1, the NCEP and ECMWF agree reasonably well overall. The differences between the two datasets appear to be larger before 1984. The GEOS index agrees more with ECMWF for the entire period, but with a negative bias of about 2 m s−1. Since a large portion of the meridional wind shear in RM1 may be due to the divergence component, it may be the reason for the noted discrepancies among the reanalyses. Hence, the results for RM1 may be dependent on the dataset used. On the other hand, Fig. 19b shows that RM2 has a robust and consistent variation among the three datasets, indicating that it is a reliable index independent of data sources. The WY index is also quite reproducible by all three datasets, even though ECMWF appears to consistently underestimate the magnitude by about 2 m s−1 (Fig. 19c).

Fig. 19.

Time series based on reanalyses of NCEP, GEOS, and ECMWF for (a) RM1, (b) RM2, and (c) WY. Units are in m s−1.

Fig. 19.

Time series based on reanalyses of NCEP, GEOS, and ECMWF for (a) RM1, (b) RM2, and (c) WY. Units are in m s−1.

7. Conclusions

We have provided a description of the internal dynamics and boundary forcing characteristics of two major subcomponents, that is, the South Asian and the East–Southeast Asian components of the ASM based on two sets of regional monsoon indices. These regional indices provide a distinctive description of the annual cycle and the interannual variability associated with the South Asian monsoon (SAM) (PSA/RM1) and the East–Southeast Asian monsoon (EAM) (PSEA/RM2), respectively. The dynamical characteristics of the regional monsoon indices and their sensitivities to SST anomalies are compared and contrasted to those represented by the broadscale circulation index of WY.

The dynamical characteristics embodied by the regional monsoon indices (RM1, RM2) and the broadscale index (WY) are shown in Fig. 20. The term PSA/RM1 describes a “classical” monsoon system where the three-dimensional anomalous circulation pattern is primarily determined by a balance between the advection of planetary vorticity and the generation of anomalous vorticity by the divergent flow because of an off-equatorial convective heat source (Fig. 20a). On the other hand, PSEA/RM2 describes a “hybrid” monsoon system characterized by multiple meridional cells over East and Southeast Asia, which span the deep Tropics and midlatitudes (Fig. 20b). The extratropical extension of the meridional circulation is due to the strong interaction of the East Asian jet stream with tropical convection and extratropical disturbances. This interaction gives rise to the characteristic zonal rainfall structure found over the East Asian monsoon region. In the extratropical region, the vorticity advection by the basic-state circulation and generation by the divergent flow contributes to the balance of upper-level anomalous vorticity. A corollary to the strong influence of extratropical dynamics on EAM is that the excitation of teleconnection to regions outside of the Asian summer monsoon by EAM is stronger than that by SAM. In contrast, the WY index appears to be a mixed measure of the integrated heat source over the ASM as well as the large-scale Walker circulation over the Indo-Pacific basin (Fig. 20c). Here WY seems to have some correlation with the EAM rainfall but not with the SAM rainfall. These findings are consistent with those recently reported by Goswami et al. (1999) and Ailikun and Yasunari (2000, manuscript submitted to J. Meteor. Soc. Japan).

Fig. 20.

Schematic showing the basic dynamical underpinnings of the various monsoon indices (a) the South Asia monsoon (RM1/PSA), (b) the East–Southeast Asian monsoon (RM2/PSEA), and (c) the broadscale Indo-Pacific circulation state (WY). Symbols A stands for anticyclone, C for cyclone, H for high, and L for low.

Fig. 20.

Schematic showing the basic dynamical underpinnings of the various monsoon indices (a) the South Asia monsoon (RM1/PSA), (b) the East–Southeast Asian monsoon (RM2/PSEA), and (c) the broadscale Indo-Pacific circulation state (WY). Symbols A stands for anticyclone, C for cyclone, H for high, and L for low.

Our results indicate that the regional monsoon variability is governed not only by basin-scale SST anomalies associated with the “slow” subsystem, but also by SSTA in the adjacent oceans within the “intermediate” subsystem. For SAM, while the influence of ENSO-related SSTA in the equatorial eastern Pacific appears to be statistically significant, the boreal spring warming in the northern Arabian Sea and subtropical western Pacific may also play a role in foreshadowing a strong SAM. For EAM, interannual variability is strongly tied to the SSTA in the Sea of Japan, the East China Sea, and the South China Sea. A strong SSTA dipole pattern with warming (cooling) in the subtropical central (eastern) Pacific during the previous spring and winter (MAM and DJF) presages a strong EAM. However, the summertime East Asian jet variability as represented by RM2 is found to have only a weak relationship with ENSO-related SSTA. Nonetheless, this does not preclude a strong relationship of the East Asian jet with ENSO-related SSTA during northern winter (Yang and Webster 1990; Liang and Wang 1998). We also note that since monsoon–ENSO relationship tends to fluctuate on decadal timescales, the above relationships are valid only for the period of the present analysis (1958–97). Regardless, our results show that the coupling of SSTA in the adjacent oceans to monsoon processes may be the key to enhance predictability of the Asian summer monsoon and should be more fully exploited.

Acknowledgments

This work is partially supported by the Global Modeling and Analysis Program of the Earth Science Enterprise, NASA Headquarters. The authors have benefited from discussions with Profs. B. Wang, I.-S. Kang, B. N. Goswami, and T. Yasunari during different stages of this investigation. Comments from Dr. Neville Nicholls and the two anonymous reviewers were helpful for improving the quality of this paper. One of the authors, K.-M. Kim, was a research associate with the National Research Council at NASA GSFC while this research was conducted.

REFERENCES

REFERENCES
Chen, T. C., M. C. Yen, and S. P. Weng, 2000: Interaction between the summer monsoons of East Asia and South China Sea: Intraseasonal monsoon modes. J. Atmos. Sci., 57, 1373–1392.
Clarke, A. J., X. Liu, and S. V. Gorder, 1998: Dynamics of the biennial oscillation in the equatorial Indian and far western Pacific Oceans. J. Climate, 11, 987–1001.
Gill, A. E., 1980: Some simple solutions of heat-induced tropical circulation. Quart. J. Roy. Meteor. Soc., 106, 447–462.
Goswami, B. N., V. Krishnamurthy, and H. Annamalai, 1999: A broad scale circulation index for the interannual variability of the Indian summer monsoon. Quart. J. Roy. Meteor. Soc., 125, 611–634.
Gu, D., and S. G. H. Philander, 1995: Secular changes of annual and interannual variability in the Tropics during the past century. J. Climate, 8, 864–876.
Huffman, G., and Coauthors, 1997: The Global Precipitation Climatology Project (GPCP) combined precipitation dataset. Bull. Amer. Meteor. Soc., 78, 5–20.
Hurrell, J. W., and K. E. Trenberth, 1992: An evaluation of monthly mean MSU and ECMWF global atmospheric temperatures for monitoring climate. J. Climate, 5, 1424–1440.
Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77, 437–471.
Krishnamurti, T. N., 1985: Summer monsoon experiment—A review. Mon. Wea. Rev., 113, 1590–1626.
Lau, K.-M., and H. Lim, 1982: Thermally driven motions in an equatorial β-plane: Hadley and Walker circulations during the winter monsoon. Mon. Wea. Rev., 110, 1372–1388.
——, and M. T. Li, 1984: The monsoon of East Asia and its global associations—A survey. Bull. Amer. Meteor. Soc., 65, 114–125.
——, and L. Peng, 1992: Dynamics of atmospheric teleconnections during the northern summer. J. Climate, 5, 140–158.
——, and S. Yang, 1996: The Asian monsoon and the predictability of the tropical ocean–atmosphere system. Quart. J. Roy. Meteor. Soc., 122, 945–957.
——, and W. Bua, 1998: Mechanism of monsoon–Southern Oscillation coupling: Insights from GCM experiments. Climate Dyn., 14, 759–779.
——, and H. T. Wu, 1999: An assessment of the impacts of the 1997–98 El Niño on the Asian–Australian monsoon. Geophys. Res. Lett., 26, 1747–1750.
Liang, X.-Z., and W. C. Wang, 1998: Associations between China monsoon rainfall and tropospheric jets. Quart. J. Roy. Meteor. Soc., 124, 2597–2623.
Liu, Z., and B. Huang, 1997: A coupled theory of tropical climatology: Warm pool, cold tongue, and Walker circulation. J. Climate, 10, 1662–1679.
Meehl, G. A., 1993: A coupled air–sea biennial mechanism in the tropical Indian and Pacific region: Role of the ocean. J. Climate, 6, 42–63.
——, 1997: The South Asian monsoon and the tropospheric biennial oscillation. J. Climate, 10, 1921–1943.
Mehta, V., and K.-M. Lau, 1997: Influence of solar irradiance on the Indian monsoon–ENSO relationship at decadal–multidecadal time scales. Geophys. Res. Lett., 24, 159–162.
Mooley, D. A., and B. Parthasarathy, 1984: Fluctuation in all India summer monsoon rainfall during 1871–1978. Climate Change, 6, 287–301.
——, ——, and G. B. Pant, 1986: Relationship between Indian-summer monsoon rainfall and location of the ridge at the 500-mb level along 75°E. J. Climate Appl. Meteor., 25, 633–640.
Nitta, T., 1987: Convective activities in the tropical western Pacific and their impact on the Northern Hemisphere summer circulation. J. Meteor. Soc. Japan, 65, 373–390.
——, and Z. Hu, 1996: Summer climate variability in China and its association with 500 hPa height and tropical convection. J. Meteor. Soc. Japan, 74, 425–445.
Palmer, T. N., 1994: Chaos and predictability in forecasting the monsoons. Proc. Indian Nat. Sci. Acad., 60, 57–66.
Parthasarathy, B., R. R. Kumar, and D. R. Kothawale, 1992: Indian summer monsoon rainfall indices, 1871–1990. Meteor. Mag., 121, 174–186.
Rasmusson, E. M., and T. H. Carpenter, 1983: The relationship between eastern equatorial Pacific seas surface temperature and rainfall over India and Sri Lanka. Mon. Wea. Rev., 111, 517–528.
Ropelewski, C., and J. Halpert, 1988: Global and regional precipitation pattern associated with the El Niño–Southern Oscillation. Mon. Wea. Rev., 115, 1606–1626.
Shen, S.-H., and K. M. Lau, 1995: Biennial oscillation associated with the East Asian summer monsoon and tropical sea surface temperature J. Meteor. Soc. Japan, 73, 105–124.
Shukla, J., 1984: Predictability of time averages. Part II: The influence of boundary forcing. Problems and Prospects in Long and Medium Range Weather Forecasting, D. M. Burridge and E. Kallen, Eds., Springer-Verlag, 115–206.
——, and D. Paolino, 1983: The Southern Oscillation and the long-range forecasting of monsoon rainfall over India. Mon. Wea. Rev., 111, 1830–1837.
Soman, M. K., and J. Slingo, 1997: Sensitivity of Asian summer monsoon to aspects of sea surface temperature anomalies in the tropical Pacific Ocean. Quart. J. Roy. Meteor. Soc., 123, 309–336.
Wang, B., and Z. Fan, 1999: Choice of South Asian summer monsoon indices. Bull. Amer. Meteor. Soc., 80, 629–638.
Webster, P. J., and S. Yang, 1992: Monsoon and ENSO: Selectively interactive systems. Quart. J. Roy. Meteor. Soc., 118, 877–926.
——, T. Palmer, M. Yanai, V. Magana, J. Shukla, and T. Yasunari, 1998: Monsoons: Processes and predictability and prospect for prediction. J. Geophys. Res., 103 (C7), 14 451–14 510.
Weng, H., K.-M. Lau, and Y.-K. Xue, 1999: Multi-scale summer rainfall variability over China and its long-term link to global sea surface temperature variability. J. Meteor. Soc. Japan, 77, 845–857.
Xie, P., and P. A. Arkin, 1997: Global precipitation: A 17-year monthly analysis based on gauge observations, satellite estimates and numerical model outputs. Bull. Amer. Meteor. Soc., 78, 2539–2558.
Yang, S., and P. J. Webster, 1990: The effect of summer tropical heating on the location and intensity of the extratropical westerly jet streams. J. Geophys. Res., 95, 18 705–18 721.
——, and K.-M. Lau, 1998: Influences of sea surface temperature and ground wetness on Asian summer monsoon. J. Climate, 11, 3230–3246.
Yasunari, T., and Y. Seki, 1992: Role of the Asian monsoon on the interannual variability of the global climate system. J. Meteor. Soc. Japan, 70, 177–189.

Footnotes

Corresponding address: Dr. K.-M. Lau, Climate and Radiation Branch, NASA GSFC, Code 913, Greenbelt, MD 20771.