• Bhalme, H. N., , and D. A. Mooley, 1980: Large-scale drought/floods and monsoon circulation. Mon. Wea. Rev., 108 , 11971211.

  • Chen, T-C., , and M-C. Yen, 1994: Interannual variation of the Indian monsoon simulated by the NCAR community climate model: Effect of the tropical Pacific SST. J. Climate, 7 , 14031415.

    • Search Google Scholar
    • Export Citation
  • Goswami, B. N., 1998: Interannual variations of Indian summer monsoon in a GCM: External conditions versus internal feedbacks. J. Climate, 11 , 501522.

    • Search Google Scholar
    • Export Citation
  • Hsu, H. H., , C. T. Terng, , and C. T. Chen, 1999: Evolution of large scale circulation and heating during the first transition of Asian summer monsoon. J. Climate, 12 , 793810.

    • Search Google Scholar
    • Export Citation
  • Joseph, P. V., , J. K. Eischeid, , and R. J. Pyle, 1994: Interannual variability of the onset of the Indian summer monsoon and its association with atmospheric features, El Niño, and sea surface temperature anomalies. J. Climate, 7 , 81105.

    • Search Google Scholar
    • Export Citation
  • Ju, J., , and J. Slingo, 1995: The Asian summer monsoon and ENSO. Quart. J. Roy. Meteor. Soc., 121 , 11331168.

  • Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77 , 437471.

  • Kang, I. S., , C. H. Ho, , Y. K. Lim, , and K-M. Lau, 1999: Principal modes of climatological seasonal and intraseasonal variation of the Asian summer monsoon. Mon. Wea. Rev., 127 , 322340.

    • Search Google Scholar
    • Export Citation
  • Kawamura, R., , T. Matsuura, , and S. Iizuka, 2001: Interannual atmosphere-ocean variations in the tropical western north Pacific relevant to the Asian summer monsoon-ENSO coupling. J. Meteor. Soc. Japan, 79 , 883898.

    • Search Google Scholar
    • Export Citation
  • Kim, K. Y., , and G. North, 1997: EOFs of harmonizable cyclostationary processes. J. Atmos. Sci., 54 , 24162427.

  • Kim, K. Y., , and Q. Wu, 1999: A comparison study of EOF techniques: Analysis of nonstationary data with periodic statistics. J. Climate, 12 , 185199.

    • Search Google Scholar
    • Export Citation
  • Krishnamurthy, V., , and J. Shukla, 2000: Intraseasonal and interannual variability of rainfall over India. J. Climate, 13 , 43664377.

  • Krishnamurti, T. N., , P. Ardanuy, , Y. Ramanathan, , and R. Pasch, 1981: On the onset vortex of the summer monsoon. Mon. Wea. Rev., 109 , 344363.

    • Search Google Scholar
    • Export Citation
  • Lau, N-C., , and M. J. Nath, 2000: Impact of ENSO on the variability of the Asian–Australian monsoons as simulated in GCM experiments. J. Climate, 13 , 42874309.

    • Search Google Scholar
    • Export Citation
  • Lau, N-C., , and H. T. Wu, 2001: Principal modes of rainfall-SST variability of the Asian summer monsoon: A reassessment of the monsoon–ENSO relationship. J. Climate, 14 , 28802895.

    • Search Google Scholar
    • Export Citation
  • Lau, N-C., , and M. J. Nath, 2003: Atmosphere–ocean variations in the Indo-Pacific sector during ENSO episodes. J. Climate, 16 , 320.

  • Lau, N-C., , G. T. Yang, , and S. H. Shen, 1988: Seasonal and intraseasonal climatology of summer monsoon rainfall over East Asia. Mon. Wea. Rev., 116 , 1837.

    • Search Google Scholar
    • Export Citation
  • Li, C., , and M. Yanai, 1996: The onset and interannual variability of the Asian summer monsoon in relation to land–sea thermal contrast. J. Climate, 9 , 358375.

    • Search Google Scholar
    • Export Citation
  • Lim, Y-K., , K-Y. Kim, , and H-S. Lee, 2002: Temporal and spatial evolution of the Asian summer monsoon in the seasonal cycle of synoptic fields. J. Climate, 15 , 36303644.

    • Search Google Scholar
    • Export Citation
  • Madden, R. A., , and P. R. Julian, 1972: Description of global-scale circulation cells in the Tropics with a 40–50 day period. J. Atmos. Sci., 29 , 11091123.

    • Search Google Scholar
    • Export Citation
  • Meehl, G. A., , and J. M. Arblaster, 1998: The Asian–Australian monsoon and El Niño–Southern Oscillation in the NCAR climate system model. J. Climate, 11 , 13561385.

    • Search Google Scholar
    • Export Citation
  • Nigam, S., 1994: On the dynamical basis for the Asian summer monsoon rainfall–El Niño relationship. J. Climate, 7 , 17501771.

  • Philander, S. G. H., 1990: El Niño, La Niña, and the Southern Oscillation. 1st ed. Academic Press, 293 pp.

  • Rasmusson, E. M., , and T. H. Carpenter, 1983: The relationship between eastern equatorial Pacific sea surface temperatures and rainfall over India and Sri Lanka. Mon. Wea. Rev., 111 , 517528.

    • Search Google Scholar
    • Export Citation
  • Seo, K-H., , and K-Y. Kim, 2003: Propagation and initiation mechanisms of the Madden-Julian oscillation. J. Geophys. Res., 108 .4384, doi:10.1029/2002JD002876.

    • Search Google Scholar
    • Export Citation
  • Shukla, J., , and D. A. Paolino, 1983: The Southern Oscillation and long-range forecasting of the summer monsoon rainfall over India. Mon. Wea. Rev., 111 , 18301837.

    • Search Google Scholar
    • Export Citation
  • Slingo, J. M., , and H. Annamalai, 2000: 1997: The El Niño of the century and the response of the Indian summer monsoon. Mon. Wea. Rev., 128 , 17781797.

    • Search Google Scholar
    • Export Citation
  • Sperber, K. R., 2003: Propagation and the vertical structure of the Madden–Julian Oscillation. Mon. Wea. Rev., 131 , 30183037.

  • Torrence, C., , and P. J. Webster, 1999: Interdecadal changes in the ENSO–monsoon system. J. Climate, 12 , 26792690.

  • Wang, B., and LinHo, 2002: Rainy season of the Asian–Pacific summer monsoon. J. Climate, 15 , 386398.

  • Wang, B., , R. Wu, , and X. Fu, 2000: Pacific–East Asian teleconnection: How does ENSO affect East Asian climate? J. Climate, 13 , 15171536.

    • Search Google Scholar
    • Export Citation
  • Wang, B., , R. Wu, , and K-M. Lau, 2001: Interannual variability of the Asian summer monsoon: Contrasts between the Indian and the western north Pacific–East Asian monsoons. J. Climate, 14 , 40734090.

    • Search Google Scholar
    • Export Citation
  • Weare, B. C., , and J. N. Nasstrom, 1982: Examples of extended empirical orthogonal function analysis. Mon. Wea. Rev., 110 , 481485.

  • Webster, P. J., , and S. Yang, 1992: Monsoon and ENSO; Selectively interactive systems. Quart. J. Roy. Meteor. Soc., 118 , 877926.

  • Wu, R., , Z-Z. Hu, , and B. P. Kirtman, 2003: Evolution of ENSO-related rainfall anomalies in east Asia. J. Climate, 16 , 37423758.

  • Xie, P., , and P. A. Arkin, 1996: Analyses of global monthly precipitation using gauge observations, satellite estimates, and numerical model predictions. J. Climate, 9 , 840858.

    • Search Google Scholar
    • Export Citation
  • Yanai, M., , C. Li, , and Z. Song, 1992: Seasonal heating of the Tibetan Plateau and its effects on the evolution of the Asian summer monsoon. J. Meteor. Soc. Japan, 70 , 1B. 319351.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    (a)–(d) The seasonal cycle modes of the boreal ASM precipitation extracted by different analysis techniques. (bottom) The PC time series of the CSEOF (dashed line) and the two extended EOFs (first mode—solid line; second mode—dotted line).

  • View in gallery

    Normalized PC time series (dotted line) of the ENSO mode (fourth CSEOF) of precipitation for 23 yr of the ASM period. The solid line denotes the monthly Niño-3 index time series. The triangles represent El Niño events (along the top) and La Niña events (along the bottom).

  • View in gallery

    The fourth CSEOF (ENSO mode) of the Xie–Arkin precipitation. The figure shows how the ENSO signal evolves in time and space throughout the prominent ASM period from 21 May to 12 September. The contour interval (CI) is 0.5 mm day−1, with values greater than 0.5 heavily shaded and values less than −0.5 lightly shaded.

  • View in gallery

    Time–longitude sections of the (a), (c) seasonal cycle and (b), (d) the El Niño mode of precipitation for (a), (b) 70°–90°E band (India) and (c), (d) 115°–130°E band (East Asia). The thick dashed lines indicate the onset of the typical precipitation spell (top) over India and (bottom) over central China.

  • View in gallery

    Streamlines of anomalous Walker circulation composites along the equator for ENSO developing years: July of (a) El Niño years (1982, 1986, 1991, 1994, and 1997) and (b) La Niña years (1984, 1988, 1995, 1998, and 1999). The ordinate denotes the vertical pressure level in hPa. (c), (d) The same as (a), (b), respectively, but for the anomalous Hadley circulation composites at 125°E (western Pacific). (a)–(d) Significant at 95% level in the sense that the likelihood of observing these patterns during non-ENSO years is less than 5%.

  • View in gallery

    Composites of sea level pressure anomaly in (a) early ASM stage (May and June) and (b) late July and early August in El Niño(0) years. (c), (d) The composites for La Niña(0) years. These patterns are significant at 95% level. The CI is 0.2 hPa, with values greater than 0.4 heavily shaded and values less than −0.4 lightly shaded.

  • View in gallery

    Precipitation anomaly (mm day−1) explained by the ENSO mode at four different regions for the ASM period. The four regions are (a), (b) India, the Bay of Bengal and Bangladesh (7.5°–25°N, 72.5°–90°E); (c), (d) Indonesia (10°S–10°N, 100°–130°E); (e), (f) central China (22.5°–32.5°N, 100°–120°E); and (g), (h) northeast Asia (30°–40°N, 125°–140°E). Precipitation in (a), (c), (e), (g) for El Niño years and in (b), (d), (f), (h) for La Niña years. The period is 21 May–17 Sep of ENSO(0) year and 21 May–9 Jul of ENSO(1) year. There is a small gap between ENSO(0) and ENSO(1) years.

  • View in gallery

    Temporal and spatial evolution of 850-hPa wind (m s−1) and relative vorticity (× 10−6 s−1) of the ENSO mode in Fig. 3. Positive (negative) vorticity anomalies greater (less) than 1 × 10−6 s−1 (−1 × 10−6 s−1) are heavily (lightly) shaded. Wind vectors less than 0.3 m s−1 are omitted with the arrow scale at the bottom.

  • View in gallery

    Same as in Fig. 4, but for 850-hPa zonal wind (m s−1).

  • View in gallery

    Same as in Fig. 8, but for 850-hPa moisture transport (kg kg−1 m s−1) and its convergence (kg kg−1 s−1). The arrow scale is at the bottom, and vectors less than 3 × 10−3 are omitted. Convergence values greater than 1 × 10−9 are heavily shaded, and divergence values less than −1 × 10−9 are lightly shaded.

  • View in gallery

    Same as in Fig. 8, but for 200-hPa wind (m s−1) and its divergence (× 10−7 s−1). Divergence (convergence) anomalies greater (less) than 2 × 10−7 s−1 (−2 × 10−7 s−1) are heavily (lightly) shaded. Wind vectors less than 1 m s−1 are omitted with the arrow scale at the bottom.

  • View in gallery

    The ENSO component (solid line) of the ASM precipitation (mm day−1) plotted against the ±1 pentad running-averaged Indian precipitation anomalies without climatology (bar) at the four regions defined in Fig. 7. (a), (b) The dashed curve is the regional monsoon index defined in Goswami (1998). Precipitation (a), (c), (e), (g) for El Niño years and (b), (d), (f), (h) for La Niña years. The period is 21 May–17 Sep of ENSO(0) year and 21 May–9 Jul of ENSO(1) year. There is a small gap between ENSO(0) and ENSO(1) years.

  • View in gallery

    Correlation of precipitation between (a) observation and the seasonal cycle, (b) observation without the seasonal cycle and the ISO mode, and (c) observational without the seasonal cycle and the ENSO mode. The 23-yr (1979–2001) observed ASM precipitation was ±1 pentad running averaged and then smoothed spatially over 3 × 3 grid boxes.

  • View in gallery

    Schematic diagram showing the three-dimensional circulation structure of the El Niño component of the ASM: (a) early stage (late May–June), and (b) middle stage (July): (top) the 200-hPa level and (bottom) the 850-hPa level. In late July, the lower-level negative pressure anomaly over the subtropical western Pacific extends toward the Bay of Bengal and northern India [see dotted line at the bottom of (b)].

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 63 63 15
PDF Downloads 44 44 13

ENSO Impact on the Space–Time Evolution of the Regional Asian Summer Monsoons

View More View Less
  • 1 Center for Ocean–Atmospheric Prediction Studies, The Florida State University, Tallahassee, Florida
  • 2 Department of Meteorology, The Florida State University, Tallahassee, Florida
© Get Permissions
Full access

Abstract

This study investigates how ENSO affects the space–time evolution of the Asian summer monsoon (ASM) precipitation and synoptic variables on a 5-day resolution over the entire ASM area. Cyclostationary EOF and regression methods were used to investigate the detailed evolution features associated with ENSO during the prominent life cycle of the ASM (21 May–17 September). This ENSO mode is identified as the third largest component (next to the seasonal cycle and the intraseasonal oscillations with a 40–50-day period) of the ASM rainfall variation.

The ENSO mode reveals that the individual regional monsoons over the ASM domain respond to ENSO in a complex manner. 1) Under the El Niño condition, the early monsoon stage over India, the Bay of Bengal, and the Indochina peninsula is characterized by rainfall deficit, along with a delayed monsoon onset by one or two pentads. This is the result of weakened diabatic heating over the Asian continent and meridional pressure gradient over the Indian Ocean, causing a weak low-tropospheric westerly monsoonal flow and the ensuing moisture transport decrease toward the regional monsoon areas. Onsets of the subsequent regional monsoons are delayed successively by this poorly developed ASM system in the early stage. 2) The Walker circulation anomaly persistently induces an enhanced subsidence over the Maritime Continent, resulting in a drought condition over this region for the entire ASM period. 3) The Hadley circulation anomaly linked to the Walker circulation anomaly over the Tropics drives a rising motion over the subtropical western Pacific, resulting in a wetter south China monsoon. The negative sea level pressure anomaly over the subtropical western Pacific associated with this anomalous Hadley circulation provides an unfavorable condition for the moisture transport toward East Asia, causing drier monsoons over north China, Japan, and Korea regions. 4) This negative sea level pressure anomaly intrudes into India, the Bay of Bengal, and the Indochina peninsula during late July and early August, developing a brief wet period over these regions. In contrast, the physical changes including the onset variation and the monsoon strength addressed above are reversed during La Niña events.

In reality, the observed ASM rainfall anomaly does not necessarily follow the ENSO-related patterns addressed above because of other impacts contributing to the rainfall variations. Although the impact of ENSO is moderately important, a comparison with other impacts demonstrates that the rainfall variations are controlled more by regional-scale intraseasonal oscillations.

* Current affiliation: Environmental Forecasts and Value-Oriented Research, Inc., Tallahassee, Florida

Corresponding author address: Kwang-Yul Kim, Department of Meteorology, The Florida State University, 404 Love Bldg., Tallahassee, FL 32306-4250. Email: kwang56@gmail.com

Abstract

This study investigates how ENSO affects the space–time evolution of the Asian summer monsoon (ASM) precipitation and synoptic variables on a 5-day resolution over the entire ASM area. Cyclostationary EOF and regression methods were used to investigate the detailed evolution features associated with ENSO during the prominent life cycle of the ASM (21 May–17 September). This ENSO mode is identified as the third largest component (next to the seasonal cycle and the intraseasonal oscillations with a 40–50-day period) of the ASM rainfall variation.

The ENSO mode reveals that the individual regional monsoons over the ASM domain respond to ENSO in a complex manner. 1) Under the El Niño condition, the early monsoon stage over India, the Bay of Bengal, and the Indochina peninsula is characterized by rainfall deficit, along with a delayed monsoon onset by one or two pentads. This is the result of weakened diabatic heating over the Asian continent and meridional pressure gradient over the Indian Ocean, causing a weak low-tropospheric westerly monsoonal flow and the ensuing moisture transport decrease toward the regional monsoon areas. Onsets of the subsequent regional monsoons are delayed successively by this poorly developed ASM system in the early stage. 2) The Walker circulation anomaly persistently induces an enhanced subsidence over the Maritime Continent, resulting in a drought condition over this region for the entire ASM period. 3) The Hadley circulation anomaly linked to the Walker circulation anomaly over the Tropics drives a rising motion over the subtropical western Pacific, resulting in a wetter south China monsoon. The negative sea level pressure anomaly over the subtropical western Pacific associated with this anomalous Hadley circulation provides an unfavorable condition for the moisture transport toward East Asia, causing drier monsoons over north China, Japan, and Korea regions. 4) This negative sea level pressure anomaly intrudes into India, the Bay of Bengal, and the Indochina peninsula during late July and early August, developing a brief wet period over these regions. In contrast, the physical changes including the onset variation and the monsoon strength addressed above are reversed during La Niña events.

In reality, the observed ASM rainfall anomaly does not necessarily follow the ENSO-related patterns addressed above because of other impacts contributing to the rainfall variations. Although the impact of ENSO is moderately important, a comparison with other impacts demonstrates that the rainfall variations are controlled more by regional-scale intraseasonal oscillations.

* Current affiliation: Environmental Forecasts and Value-Oriented Research, Inc., Tallahassee, Florida

Corresponding author address: Kwang-Yul Kim, Department of Meteorology, The Florida State University, 404 Love Bldg., Tallahassee, FL 32306-4250. Email: kwang56@gmail.com

Keywords: ENSO; Asia; Monsoons

1. Introduction

The El Niño–Southern Oscillation (ENSO) is one of the most important factors governing the interannual variation of the Asian summer monsoon (ASM). The relationship between ENSO and the ASM has been studied more actively than other aspects of the ASM variations (Joseph et al. 1994; Chen and Yen 1994; Ju and Slingo 1995; Li and Yanai 1996; Goswami 1998; Meehl and Arblaster 1998; Torrence and Webster 1999; Lau and Nath 2000; Wang et al. 2001; Lau and Wu 2001; Lau and Nath 2003). The basic feature of this relationship is that the enhanced upward motion over the central and eastern Pacific Ocean during a warm ENSO event weakens the typical Walker circulation pattern by producing a downward motion anomaly over the tropical western Pacific and the Indian Oceans; this results in a weak ASM over India and the Indonesia.

This mechanism from previous studies, however, is a gross feature on a seasonal mean basis specifically for the tropical region. More detailed features of how ENSO affects the monsoon have not been clearly understood for the whole ASM area and the period. While many previous studies referred to above have focused on the strength of the ASM response to ENSO, variations of each regional monsoon onset and termination times in association with ENSO have been studied little. Joseph et al. (1994) and Ju and Slingo (1995) reported that El Niño is associated with the delayed onset of the Indian monsoon. These papers described only the Indian monsoon onset, defined as the arrival of low-level kinetic energy, always being delayed during a warm ENSO event. Variations of monsoon onset in other ASM regions have not been sufficiently discussed so far. Also, it is not clear whether the impact of ENSO on the ASM rainfall is dominant over that of intraseasonal oscillations (ISO; e.g., the Madden–Julian oscillation; Madden and Julian 1972). That the observed rainfall anomaly would be always negative (positive) in every El Niño (La Niña) year, in fact, may be a serious misstatement when non-ENSO factors affecting the monsoon precipitation are strong. In this respect, the statement that the Indian monsoon onset is always delayed in an El Niño year (Joseph et al. 1994; Ju and Slingo 1995) needs to be reassessed. Finally, in many previous studies on the ENSO–ASM relationship focus is restricted to tropical regional monsoon regions such as India, the Bay of Bengal, and the Indochina peninsula (Chen and Yen 1994; Goswami 1998; Krishnamurthy and Shukla 2000; Slingo and Annamalai 2000; Lau and Nath 2003), where the ASM rainfall variability is relatively clear-cut compared to such midlatitude Asian monsoon regions as China, Japan, and Korea. The impact of ENSO on the East Asian monsoons (China, Japan, and Korea) has not been adequately understood because of its complexity and indirectness (Wang et al. 2000; Wu et al. 2003). Thus, comprehensive analyses and descriptions on a high temporal (e.g., 5 day) resolution over the entire ASM domain are necessary to clarify these aspects of the ENSO–ASM relationship in an accurate manner.

To address these questions appropriately, observational data in this study will be analyzed from a different viewpoint than in the past. Namely, variability in the data is viewed as a superposition of individual physical modes (e.g., seasonal cycle, ENSO, ISO, etc.) with temporally varying amplitudes. This new perspective facilitates the dynamical and thermodynamical interpretations of variability and understanding of how individual modes interact with one another. To this end, the cyclostationary EOF (CSEOF) technique is employed (Kim and North 1997) to extract physically evolving spatial patterns of each mode; the main motivation for employing CSEOF analysis in the present study is to better investigate the physical evolution in space and time (Lim et al. 2002; Seo and Kim 2003). The ENSO-loading vector, in particular, describes the physical evolution of precipitation associated with ENSO over the ASM domain, whereas the corresponding principal component (PC) time series represents the long-term variation of its amplitude. This distinction of “physical evolution” and “amplitude fluctuation” facilitates the physical interpretation of the evolution each mode represents and its contribution to the overall variation of the ASM.

Several key questions this study attempts to answer are as follows. 1) How do the regional monsoons respond differently to ENSO and evolve during the entire ASM period? 2) How does the regional monsoon onset time vary with the ENSO phase? 3) How significant is the influence of ENSO on the ASM rainfall variations compared to other impacts? 4) What is the impact of ENSO on the extratropical East Asian monsoons?

Section 2 introduces the dataset used in this study. Methods of analysis are described in section 3. Results and interpretations on the ASM features associated with ENSO are addressed in sections 4, 5, and 6, followed by a summary and concluding remarks in section 7.

2. Data

Data used in this study are the 23-yr (1979–2001) Xie–Arkin precipitation pentad data (Xie and Arkin 1996), and several synoptic variables including daily sea level pressure, wind at all standard levels up to 100 hPa, and specific humidity archived at the National Centers for Environmental Prediction (NCEP; Kalnay et al. 1996). As derived variables, moisture transport, and its convergence at the 850-hPa level, vorticity and its advection are calculated via the finite-difference method.

The spatiotemporal evolution of the ASM is focused on the period of late May (21 May) to mid-September (17 September), thereby covering the prominent ASM evolution cycle, over the domain 10°S–40°N × 50°–140°E. A 5-day mean dataset with a 2. 5° × 2.5° spatial resolution was constructed for each variable by performing nonoverlapping 5-day average of the daily data. Then, CSEOF analysis was conducted on the pentad precipitation data to identify the ENSO mode of variability, and the other pentad datasets were regressed onto the ENSO mode of precipitation as explained below.

3. Methods of analysis

a. CSEOF analysis

Cyclostationary EOF (Kim and North 1997) is an analysis technique for extracting the spatiotemporal evolution of physical modes (e.g., seasonal cycle, ENSO, ISO, etc.) and their long-term amplitude variations. Space–time data in CSEOF analysis are represented as
i1520-0442-20-11-2397-e1
where Bn(r, t) are cyclostationary loading vectors (CSLVs) and Sn(t) are their corresponding PC time series. While the eigenfunctions obtained from conventional EOF analysis do not have time dependence, CSLVs in (1) are time dependent. Furthermore, CSLVs are periodic in time:
i1520-0442-20-11-2397-e2
where d is the nested period representing the inherent periodicity of the statistics. Thus, Bn(r, t) describes the evolution of spatial patterns over the period of d as in extended EOFs (Weare and Nasstrom 1982). The Sn(t) represents the amplitude of Bn(r, t); the amplitude time series fluctuates slowly showing how the strength of evolution depicted in Bn(r, t) varies on a time scale longer than that of Bn(r, t). Since the amplitude, Sn(t), does not vary significantly over the period of d, Bn(r, t) describes the evolution of correlated spatial patterns [i.e., Bn(r, t = 1), Bn(r, t = 2), . . . , Bn(r, t = d)]. This is interpreted as “physical” evolution since the correlation of successive spatial patterns comes only from a physical process repeating in time in a consistent manner. For example, the evolution of ENSO is reasonably similar from one event to another; this can be described in Bn(r, t) with the nested period of d = 1 yr. The strength of ENSO, however, varies from one year to another; this can be described in Sn(t).
While CSEOFs are similar in structure to extended EOFs including the orthogonality of loading vectors and the uncorrelatedness of the PC time series, the rendition of covariance statistics is very different between the two techniques (Seo and Kim 2003). In CSEOF analysis, covariance function is computed by
i1520-0442-20-11-2397-e3
Kim and North (1997) discuss in detail how this function can be estimated under the cyclostationarity assumption. In extended EOF analysis, data is artificially augmented in a lagged manner:
i1520-0442-20-11-2397-e4
where l is an arbitrary lag. Then, spatial covariance function (matrix) is computed via
i1520-0442-20-11-2397-e5
where the ensemble averaging, 〈 〉, can be interpreted as time averaging under the stationarity assumption. Temporal covariance is introduced in the covariance function since data is augmented with a lag as explained above.

Note in (3) that covariance between two different times (e.g., January and February) are specifically rendered in the covariance statistics. The covariance between January and February is different from that between, say, June and July. In (5), however, covariance statistics are averaged in time. Therefore, there is no distinction between January–February covariance and June–July covariance in conventional extended EOF analysis. This is a significant difference because there may be essential difference in the covariance between January–February and June–July. This difference often results in significantly disparate sets of eigenvectors and eigenvalues.

Figure 1 shows an example that compares the two analysis techniques in rendering the seasonal cycle of the Asian summer monsoon precipitation. The longitude– time sections along 20°N show that the spatial patterns extracted by the CSEOF analysis technique are similar to those of the composite analysis as applied to the precipitation data in the present study. On the other hand, the extended EOF technique extracts the seasonal cycle as two sets of evolution patterns. The first two extended EOFs are 90° out of phase to each other. The corresponding PC time series show that the two patterns have a significant lagged correlation (Fig. 1b). Figure 1 reflects the fact that the extended EOF technique cannot render a temporally asymmetric covariance function. As can be seen in the figure, seasonal cycle is not quite symmetric over the given interval. Note also that the extended EOF patterns on the eastern side of the domain have the opposite sign to those of the composite patterns whereas the patterns on the western side have the same sign.

In fact, one cannot find any extended EOF mode having a PC time series similar to that in Fig. 2 from the boreal summer precipitation data. (It is not possible to extract Figs. 8 –11 from the data in terms of extended EOF analysis, neither is it possible to derive Figs. 2 –4 and 6 –12 using EOF or complex EOF analysis.) Thus, it is essential to conduct CSEOF analysis in the present study to extract a reasonable picture of evolution associated with the ENSO mode. Detailed comparisons of the two techniques are considered beyond the scope of this paper, but can be found in Kim and Wu (1999) and Seo and Kim (2003).

An important step when performing CSEOF analysis is to determine the nested period (i.e., the inherent period of covariance statistics). The CSLVs are periodic and time dependent because they are eigenfunctions of a time-dependent and periodic covariance statistics. The data employed in this study is 24 pentads (120 days) long from 21 May to 17 September for each year. According to the data interval, the nested period is set to be 24 pentads. This is based on the assumption that the statistical properties of the dominant modes of the ASM variability such as the seasonal cycle, ENSO, and ISOs do not vary from one year to another. This does not mean that the evolution of each mode (e.g., ENSO) is identical from one year to another; each year one sees a different realization of the same statistics in the observational data. With the nested period defined here, CSLVs describe physical modes evolving through the given period (24 pentads) and repeating each year with varying amplitude.

A physical mode exhibiting a discrete periodicity such as the seasonal cycle is captured as a single mode in the CSEOF analysis. It should be pointed out, however, that a physical mode with a broad spectral density and/or variable phases cannot be represented as a single mode (e.g., intraseasonal oscillations); in such a case, prominent characteristics of a physical mode are captured typically in terms of two modes.

b. Regression analysis

Two sets of CSEOFs computed independently for two variables may not necessarily be consistent with each other. To find physically and dynamically consistent modes among different variables, one reference variable needs to be declared as a target variable followed by regression of other variables (predictors) onto the target variable. For this regression, the PC time series of the first few significant modes of a predictor variable are regressed onto a certain PC time series of the target variable (multiple regression). That is,
i1520-0442-20-11-2397-e6
where PCTn(t) are the target PC time series, αi are the regression coefficients, and PCPi(t) are the predictor PC time series. Regression coefficients are determined such that the variance of regression error, Var[ε(t)], is minimized. The R2 value, which measures the degree of fitting, is given by 1 − 〈ε2(t)〉/〈PCT2n(t)〉, where the angle brackets denote ensemble averaging. If the value of R2 approaches unity, regression fitting is perfect; there is no fitting when R2 = 0.
The spatial pattern of the predictor variable, CSRn(r, t), consistent with the pattern of the target variable, CSTn(r, t), is obtained by
i1520-0442-20-11-2397-e7
where CSPi(r, t) are the cyclostationary eigenfunctions of the predictor variable. Note that the regression method here only imposes that the amplitude fluctuations of the regressed pattern be identical with that of the target pattern. Spatiotemporal evolutions in CSTn(r, t) and CSRn(r, t) should satisfy the dynamical relationship between the two. This approach should be contrasted with the projection method, regression in EOF space, in which two spatial patterns evolve in an identical manner according to the identical PC time series.
For the regression analysis in this study, the ENSO mode of precipitation is regarded as the target variable (predictand) whereas other variables (e.g., low-level wind and vorticity, moisture transport and its convergence, and upper-level circulation and divergence) as the predictors. The first 10 PC time series of the predictor variables, Pn(t), which explain the majority of variability, are used for regression (i.e., i = 1, . . . , 10):
i1520-0442-20-11-2397-e8
where S(t) is the PC time series of B(r, t), the ENSO mode of precipitation. Typically, the regression error, Var[ε(t)], is less than 5% of Var[S(t)]. Then the evolution of a predictor variable is obtained by using the first 10 CSEOFs of the predictor variable as delineated in (7):
i1520-0442-20-11-2397-e9
where Cn(r, t) are CSEOFs of the predictor variable. Then, B(r, t) (precipitation) and P(r, t) (predictor variable) are consistent in the sense that they have nearly the same amplitude time series, S(t). The resulting regressed patterns of predictor variables (presented in Figs. 8, 10 and 11) are considered physically and dynamically consistent with the target patterns in Fig. 3.

4. Principal component time series associated with the ENSO mode

The CSEOF analysis was conducted for the entire ASM domain and period. The ENSO-related evolution of precipitation over the ASM domain was reasonably separated from the observed precipitation data. The ENSO mode of the ASM rainfall variation appeared as the fourth mode, which explains ∼7% of the total variance. Figure 2 shows the PC time series of the ENSO mode. The time series is fairly similar to the El Niño and La Niña occurrence record (not shown), with positive amplitude in El Niño years and vice versa. The solid line in Fig. 2 represents the monthly Niño-3 sea surface temperature (SST) index time series. Note that the PC time series are only for the monsoon period whereas the Niño-3 time series are for the entire year. As can be seen, the two time series are reasonably close to each other; correlation between the two time series is 0.71 with the PC time series leading the Niño-3 time series by 5 months. Two strong El Niño events (1982 and 1997) and two strong La Niña events (1988 and 1998) are clearly shown in both time series.

The PC time series recognizes the moderate ENSO events (e.g., 1984, 1986, 1987, 1991, 1994, 1995, and 1999) better than the Niño-3 time series. For example, the year 1984 was reported to be a La Niña year although it almost looks like a non-ENSO year according to the Niño-3 time series. In another case, the Niño-3 time series suggests four successive years of El Niño (1991–94), however, noticeable El Niño events only occurred in 1991 and 1994. The PC time series correctly identifies these warm events (1991 and 1994) and the cold event (1984). A faithful reproduction of the ENSO occurrences by the PC time series demonstrates that the impact of ENSO events is clearly felt in the ASM domain.

5. Evolution of the ASM associated with ENSO

a. Precipitation anomaly

Temporal and spatial evolution of the precipitation field associated with ENSO is presented in Fig. 3. Spatial patterns were obtained with a 5-day interval but are shown here only at every other time step. The sign of the anomalies in this figure represents the El Niño condition. A notable impact of the ENSO mode is the delayed progression of the ASM system; the timing of the positive anomaly in the ENSO-related precipitation change at each regional monsoon area is about two pentads later than the typical onset time of the regional monsoon. For example, the typical onset time of the Indian monsoon as reflected in the seasonal cycle is early June (Lim et al. 2002; Wang and LinHo 2002). In Fig. 3, however, the positive precipitation anomaly is not seen over India until late June. This means that an El Niño event tends to delay the onset of the Indian monsoon.

Joseph et al. (1994) and Ju and Slingo (1995) also noted that the Indian summer monsoon appears to start later than usual in El Niño years. They suggested that a southward displacement of the ITCZ is associated with the delayed onset of the Indian monsoon. The increased SST, the decreased outgoing longwave radiation, and the decreased sea level pressure over the Indian Ocean (Fig. 6a) are responsible for the southward displacement of the ITCZ from its climatological location (Ju and Slingo 1995). Consequently, the northward migration of convective maximum over the Indian Ocean is delayed resulting in delayed monsoon onsets over the ASM region. Diabatic heating is also important for the development of continental heat-low and upper-level anticyclone (Yanai et al. 1992; Hsu et al. 1999), but the sea level pressure anomaly in the early ASM stage (Figs. 6a,c) and the upper-level circulation and divergence (Figs. 11a–d) suggests a poor development of diabatic heating over the Asian continent; this decelerates the moisture-laden low-level westerlies toward the continent thereby delaying the onset of the regional monsoons.

Hovmöller diagrams in Fig. 4 clearly show the delayed onset of the ASM over India. Long-dashed lines represent the typical onset time of the Indian monsoon (Fig. 4a). At this time, contribution from El Niño is significantly negative (Fig. 4b) causing the delayed onset of the Indian monsoon.

Similarly, other regional monsoons including the Indochina peninsula, China, and northeast Asia (Korea and Japan) areas are also delayed (Figs. 3a–f). The sign of the precipitation anomaly due to El Niño is negative for about two pentads from the typical onset times of regional monsoons in these areas, which are respectively early June, mid-June, and late June in the seasonal cycle (Fig. 2 in Lim et al. 2002; Wang and LinHo 2002). As shown in Figs. 4c,d, however, the amount of precipitation change due to El Niño is much smaller than the amplitude of the seasonal cycle in East Asia. Thus, the impact of the ENSO over these areas may be much less than that in the Indian monsoon region.

The Asian summer monsoons are expected to start earlier than normal during La Niña years. Figure 3 with an opposite sign indicates that positive precipitation anomalies appear earlier than suggested by the seasonal cycle in the areas of regional monsoons, leading to an earlier onset of the wet period. Sea level pressure anomalies (shown in Fig. 6c) and other synoptic fields (shown in Figs. 8, 10 and 11), with their signs reversed, imply a favorable condition for continental heating and the development of the low-level westerlies and the ensuing moisture transport during the early ASM stage.

The effect of El Niño on the strength of the monsoon varies from one region to another. While the Maritime Continent (Indonesia) experiences a drought during an El Niño event due to a subsidence of the anomalous Walker circulation (Fig. 5a; see also the negative vortex in Figs. 8e,f), the south China area (20°–30°N) and the Philippines have a more active monsoon phase (Figs. 3e–j), which is in good agreement with Kawamura et al. (2001). As shown in Figs. 5a,c, this north–south contrast is a direct result of the well-established anomalous Walker circulation (sinking branch at ∼120°E) and the anomalous Hadley circulation (rising branch at ∼30°N) associated with El Niño. Many studies documented that the Indian monsoon becomes weak during an El Niño event (Bhalme and Mooley 1980; Rasmusson and Carpenter 1983; Shukla and Paolino 1983; Philander 1990; Webster and Yang 1992; Chen and Yen 1994). In the present study, however, the early rainfall deficit over such subtropical regions as India and the Bay of Bengal is followed by periods of rainfall surplus (Figs. 3d,e and g–i). The rainfall surplus in late July and early August of El Niño years is associated with the westward expansion of the negative sea level pressure anomaly from the western Pacific (Figs. 6a,b; also see Fig. 14b), where the anomaly is connected with the upward branch of the anomalous Hadley circulation.

b. Precipitation anomaly over four regions

Figure 7 describes the precipitation anomalies over four different regions for five El Niño and four La Niña events. This figure is obtained by computing precipitation at specified locations:
i1520-0442-20-11-2397-e10
where B(r, t) is the ENSO mode identified in the present study and S(t) is the corresponding amplitude (PC) time series. Precipitation anomalies were computed from 21 May to 17 September in ENSO starting years [hereafter ENSO(0)] and from 21 May to 9 July of the following years [hereafter ENSO(+1)].

The negative precipitation anomaly over India, the Bay of Bengal, and the Indochina peninsula characterizes the early and the late stages of the ASM in the ENSO(0) year during an El Niño event (Figs. 7a,b) and the opposite characteristics during a La Niña event. The same pattern persists until the early ASM stage of the ENSO(+1) year; this appears to be due to the persistence of ENSO effect until the spring of ENSO(+1) year. Notable exceptions to these characteristic features are 1995 and 1998—two La Niña years preceded by El Niño years. Figure 7 shows that the early stage of the ASM is more affected by the preceded ENSO condition than the developing ENSO condition. Thus, the early monsoon precipitation in 1995 and 1998 reflects the El Niño condition (negative precipitation) rather than the La Niña condition (positive precipitation). During the middle stage of the ASM (late June–early July, late July–early August), the sign of precipitation anomaly is reversed (wet during El Niño and dry during La Niña) from that of the early stage with a brief intermission (or reversal) in mid-July yielding a bimodal structure of precipitation anomaly.

The Indonesian precipitation anomaly associated with ENSO is negative during El Niño events and is positive during La Niña events (Figs. 7c,d). Severe drought or flood is likely to occur over this region particularly in ENSO(0) years.

The East Asian monsoon regions show contrasting precipitation anomaly patterns. While the south China and Yangtze River region shows a positive anomaly (Fig. 7e), northeast Asia tends to experience a drier monsoon (Figs. 7g) in El Niño years. This is due to the negative sea level pressure anomaly over the subtropical western Pacific (Fig. 6b) in association with the anomalous Hadley circulation (Fig. 5c). The upward branch of the anomalous Hadley circulation serves as a barrier for the northeast moisture transport along the eastern coast of the Asian continent (Kang et al. 1999; Lim et al. 2002), causing drier monsoons over the regions of north China, Japan, and Korea. The situation is reversed in La Niña years (Figs. 7f,h and 5d).

c. Lower-level (850 hPa) horizontal circulation

The space–time evolution of the low-level wind field consistent with El Niño was extracted via regression analysis (Fig. 8). The onset of the Indian monsoon is regarded as the arrival of the low-level westerly jet and positive vortex (Krishnamurti et al. 1981). During warm ENSO events, the arrival of the low-level westerly jet is delayed as shown in Figs. 8a–c. Also, the onset vortex migrates northward from the western Indian Ocean reaching India in late June for the first time during the ASM life cycle; this represents a delayed onset. The delayed development of low-level westerly jet is due to a poor continental heating and a decreased south–north gradient of sea level pressure anomalies during El Niño (Fig. 6a), which weakens and delays the growth of the north–south sea level pressure gradient in the seasonal cycle, which is instrumental for the development of westerly jet. The negative pressure anomaly over the Indian Ocean is a result of the increased SST and the resulting southward displacement of ITCZ (Ju and Slingo 1995), whereas the positive pressure anomaly over the continent is driven by a poor diabatic heating (Yanai et al. 1992; Li and Yanai 1996; Hsu et al. 1999). The upper-level anomalous cyclonic circulation over the continent (shown in Fig. 11) also indicates poor diabatic heating and the resulting weak Tibetan high during El Niño events. The El Niño–related diabatic heating anomalies oppose the climatological low-level westerly monsoon flow over the northern Indian Ocean (Nigam 1994). Low-level easterly anomalies arising from the anomalous Walker circulation (Fig. 5) over the Maritime Continent also counteract the westerly anomalies from the western Indian Ocean, hampering the enhancement of the low-level westerly jet (Sperber 2003).

The delayed progression of the monsoon system successively affects the onset times of the subsequent regional monsoons over East Asia (Figs. 8d,e). According to the seasonal cycle, the westerly jet and positive vorticity should be well established along the coast of south China so that the East Asian regional monsoons can begin (Lau et al. 1988; Kang et al. 1999; Lim et al. 2002). While the typical initiation of this westerly wind anomaly is no later than mid-June based on the seasonal cycle (Fig. 5 in Lim et al. 2002), the ENSO mode does not exhibit a westerly wind anomaly and positive vortex until late June or early July (Figs. 8d,e).

The time–latitude sections of the low-level wind under the El Niño condition are presented in Fig. 9 in comparison with the seasonal cycle. Over the Indian region, easterly wind anomaly is enhanced during El Niño near the onset of the Indian monsoon so that the appearance of westerly wind anomaly (shaded) is delayed (Figs. 9a,b). Similarly, over the East Asian region, the westerly wind anomaly in the seasonal cycle is weakened by the easterly wind anomaly under the El Niño condition at least in the early stage of the East Asian monsoon (see ∼30°N in Figs. 9c,d).

Under the El Niño condition, a stagnant anticyclonic circulation dominates over the eastern Indian Ocean including Indonesia from late June (Figs. 8d–i). This stagnant anticyclone, a consequence of a strong sinking motion due to the anomalous Walker circulation, results in a drier condition over the region. The cyclonic circulation over the subtropical western Pacific in July and August, on the other hand, is associated with a wetter south China monsoon (Kawamura et al. 2001). This cyclonic circulation is linked with the Hadley circulation anomaly (see the upward motion over the subtropics in Fig. 5c) and the negative sea level pressure anomaly over the subtropical western Pacific (Fig. 6b).

As noted earlier, a significant second wet period is found over India and the Bay of Bengal during late July through early August in association with an El Niño event (Fig. 3h). As shown in Figs. 8g–i, the development of the low-level westerly wind anomaly and positive vortex cells in these areas signals the secondary wet period. A comparison with the seasonal cycle indicates that the El Niño condition delays this secondary wet period (see also Figs. 4a,b)

d. Evolution of the low-level moisture transport

Convergence of the moisture transport from the main source regions (e.g., the Indian Ocean and the subtropical western Pacific) is essential for the formation and maintenance of the precipitation field. The regressed patterns of the moisture transport convergence depicted in Fig. 10 confirm the delay of the northward progression of the ASM system during El Niño events. For example, during late May and early June, the typical onset period, the divergence of moisture transport characterizes the regions of India, the Bay of Bengal, and Indochina (Figs. 10a,b). The divergence of moisture and the moisture transport vectors indicate that moisture supply from the Indian Ocean and the convergence of moisture are both reduced over these regions. This feature is dynamically consistent with the sea level pressure anomalies and the low-level wind fields as discussed earlier (Figs. 6 and 8). The low-level flow driven by El Niño heating over the tropical Pacific and Indian Ocean (Nigam 1994) counters the climatological westerly monsoon flow over the tropical Indian Ocean thereby reducing moisture transport toward India and Indochina. Two well-developed convergence patterns are seen over India in late June (Fig. 10d) and around early August (Figs. 10g–i), during which significant positive precipitation anomalies are also observed in the same area (Figs. 3d,h).

According to the seasonal cycle (Fig. 6 in Lim et al. 2002), the increased moisture transport to the Indonesian region comes from the equatorial western Pacific in the early monsoon stage, from the Indian Ocean in the middle stage, and from both sources in the later stage. During El Niño events, moisture transport anomaly vectors are westward along the equator (Indonesia) whereas they are eastward along the southern tip of the Asian continent (∼10°N; Figs. 10d–j). This anticyclonic moisture transport anomaly and the resulting moisture divergence over Indonesia are linked with the sinking motion driven by the anomalous Walker circulation. Thus, the Indonesian region experiences a severe drought.

Convergence of moisture transport over East Asia from mid-June through late July is weaker than normal in the El Niño condition (Figs. 10c–g). The convergence zone is generally shifted southward and covers southeast China, the South China Sea, Taiwan, Hong Kong, the Philippines, and even the Indochina peninsula (Figs. 10e,f). This feature of weak moisture transport over East Asia in El Niño years is caused by a poor development of the subtropical western Pacific high pressure (SWPH) anomaly (Kang et al. 1999; Lim et al. 2002), which contributes to the moisture supply toward the East Asian monsoon regions. The reduced moisture transport tends to make the East Asian monsoon weaker than normal.

e. Horizontal circulation at the upper level (200 hPa)

The evolution features of the regressed upper-level wind field associated with El Niño are shown in Fig. 11. A comparison with Fig. 8 reveals that the upper-level circulation is strongly tied with the low-level circulation. Low-level convergence is generally translated into upper-level divergence (dark shade) and vice versa. A large-scale upper-level divergence zone is seen in the 10°–20°N band in late June (Fig. 11d). A divergence zone over midlatitude East Asia (∼30°N) develops from early July (Fig. 11e), approximately two pentads later than the seasonal cycle. The divergence pattern over subtropical East Asia is remarkably different from that in the seasonal cycle in such a way that it occupies the entire subtropical western Pacific centered around the Philippines since late June (Figs. 11d–h), whereas the divergence pattern in the seasonal cycle is concentrated along the midlatitude frontal zone (Figs. 7c–e in Lim et al. 2002). The former is related to the suppression of the SWPH, which weakens the East Asian summer monsoons (North China, Japan, and Korea). As depicted in Figs. 5c and 6b, the weakening of the SWPH is linked to the anomalous Hadley circulation during an El Niño condition.

Another remarkable feature is that the westerly wind anomaly is sustained over the Indian sector throughout the monsoon period. While a well-developed anticyclone with the associated easterlies south of 30°N is seen in the seasonal cycle (Figs. 7d, e in Lim et al. 2002), this anticyclone is not clearly seen in the El Niño–related circulation in the early ASM stage (Figs. 11a–d). This anticyclone has specific meanings: 1) it is a reflection of the Tibetan high developed by diabatic heating (Yanai et al. 1992; Li and Yanai 1996) and 2) this large-scale monsoon circulation affects the strength of the ASM (Webster and Yang 1992). The diabatic heating over the Tibetan Plateau and the adjacent elevated areas, to a large extent, represents sensible heating (Yanai et al. 1992; Hsu et al. 1999) and contributes to the formation of the heat low and the resulting circulation (Nigam 1994) near the surface. This, in turn, makes the sea–land pressure contrast stronger, resulting in a stronger monsoon circulation. After all, a strong monsoon is characterized by strong low-level inflow from the ocean and upper-level outflow toward the ocean in the form of an anticyclone (Webster and Yang 1992; Lim et al. 2002). Near the onset time of the Indian monsoon, the El Niño mode produces an upper-level cyclone (convergence) and the associated westerly wind anomalies over the Indian sector as shown in Fig. 11c; this weakens the anticyclone in the seasonal cycle and henceforth results in a weak monsoon. In La Niña years, on the other hand, the intensification of an anticyclone is clearly observed over the Tibetan Plateau and Indian interior (figure not shown).

6. Impact of ENSO on the variation of the observed ASM rainfall

The discussion so far has focused on how the onset and strength of regional monsoons are affected by the ENSO mode. Here, we examine the significance of the ENSO contribution to the observed ASM rainfall variation. Figure 12 shows the ENSO component of precipitation (solid line) together with the observed precipitation anomalies (shaded bars) over four selected regions during ENSO events. As can be seen, the ENSO-related rainfall variation, in general, does not follow the observed precipitation although the significance of the ENSO effect is moderately seen over the Maritime Continent region (Indonesia). The ENSO effect is in no way robust in other regions.

Except for the Maritime Continent region, the impact of ENSO seems much less pronounced than that of intraseasonal oscillations. For example, the observed precipitation over India, Bangladesh, and the Bay of Bengal region is relatively close to the regional monsoon index (dashed curve in Fig. 12; Goswami 1998), which primarily measures the impact of intraseasonal oscillations (Figs. 12a,b). During the early monsoon period of 1982, 1991, and 1994 specifically, El Niño precipitation is of the opposite sign to the observation, whereas the regional monsoon index is consistent with the observation. For the rest of the monsoon period, the regional monsoon index also reasonably describes the observed precipitation except for 1997 and 1998—the strongest ENSO years; the strong ENSO impact on precipitation dominates the ISO effect in those two years. As can be seen in Figs. 12e–h, a significant discrepancy also exists between the ENSO precipitation and the observation throughout the monsoon period in the south China area and over the northeast Asian region.

Figure 13 depicts the geographic distribution of the correlation coefficients between the first three dominant modes and the observed rainfall. Undoubtedly, the seasonal cycle alone explains a significant amount of precipitation variability (Fig. 13a). Correlations with the climatology-removed precipitation anomaly show that the ISO component (Fig. 13b) plays a far more significant role in determining the ASM variation than the ENSO mode (Fig. 13c). The ISO component exhibits high correlation (>0.5) over the subtropical and tropical monsoon regions. The East Asian monsoon region shows correlations greater than 0.3 (Fig. 13b). The correlation with the ENSO mode is generally much lower than the correlation with the ISO mode. It is rather obvious that ENSO alone does not determine the sign of the observed precipitation anomaly. The ENSO-related precipitation variability nonetheless is moderately important over the tropical monsoon region including southern India and the Indochina peninsula, and the Maritime Continent region.

7. Summary and concluding remarks

Detailed evolution features of the ENSO component of the ASM variability have been investigated in order to understand how ENSO affects the onset and strength of the ASM during the entire life cycle. The present study shows that the ASM–ENSO relationship is not uniform over the entire ASM domain and the period. The results of analyses for El Niño years are summarized: 1) El Niño acts to delay the onset of all regional monsoons by one or two pentads; 2) the monsoon strength over Indonesia including the Maritime Continent, and northeast Asia is weaker than normal; 3) South China including the Philippines experiences a wetter monsoon during the entire ASM period; and 4) over India, the Bay of Bengal, and the Indochina peninsula, the monsoon is weaker than normal during the early ASM stage but exhibits a brief wet period in late July and early August. The impact of La Niña is essentially opposite to that of El Niño; therefore, only El Niño–related features will be addressed below.

The present study shows that the thermal contrast between the Indian Ocean and the Asian continent varies during ENSO events. The resulting changes in the meridional pressure gradient and the low-level zonal flow over the Indian Ocean are the primary reasons for the onset variation of the ASM (Fig. 14a). During El Niño years, the negative pressure anomaly over the Indian Ocean (Fig. 14a) and the poor continental heat low causes a weaker north–south pressure gradient in the seasonal cycle. This weakened pressure gradient delays the development of the low-level westerly anomalies over the Indian Ocean by one or two pentads. As a result, the westerly monsoonal flow and the moisture transport toward India are subsequently delayed (Fig. 14a). These low-tropospheric structures are linked vertically with a weaker upward motion over the Asian continent and a weaker upper-level anticyclone (Fig. 14a).

The late establishment of the early monsoon circulation and the moisture transport over the Indian Ocean due to these factors affects the subsequent evolution of the ASM and acts to delay the onset of regional monsoons over the whole ASM domain. For example, the East Asian monsoons over China, Japan, and Korea tend to be delayed during El Niño years. The weaker and later development of SWPH anomaly is primarily responsible for the delayed onset and drier monsoons over these regions although the impact of ENSO on the monsoon strength looks smaller than that over the Indian monsoon region.

The anomalous Walker circulation is closely linked with changes over the subtropical ASM domain. During El Niño events, the anomalous Walker circulation is established with subsidence over the Indian Ocean and the western Pacific, and with ascent over the eastern Pacific. This planetary-scale subsidence gives rise to a weaker ASM over the Maritime Continent (Indonesia) in El Niño years (Fig. 14b).

The East Asian region (China north of 30°N, Japan, and Korea) also experiences a weaker monsoon due to a weaker SWPH and the corresponding deficit of moisture influx in El Niño years. This particular sea level pressure anomaly pattern is a manifestation of an anomalous Hadley circulation (Fig. 14b). The negative SWPH anomaly, on the other hand, causes a wetter monsoon directly over south China and the Philippines (Fig. 14b). This anomaly appears to expand inland during late July and early August, which results in a brief wet period over India, the Bay of Bengal, and the Indochina peninsula (Fig. 14b).

This study reveals that the impact of ENSO is not uniform but complex with much regional peculiarity throughout the ASM domain and the period. Also, the observed rainfall and the onset of regional monsoons do not always follow the ENSO–ASM relationships described in this study as also noted by Slingo and Annamalai (2000). One crucial reason is that the observed rainfall is affected more by regional-scale intraseasonal oscillations. It should be borne in mind that ENSO is just one of the prominent modes constituting the ASM variability.

Acknowledgments

The authors are thankful for the useful and constructive comments from the reviewers. COAPS receives its base support from the Applied Research Center, which is funded by the NOAA Climate Program Office. This work was also partially supported by NSF Grant ATM-0353494.

REFERENCES

  • Bhalme, H. N., , and D. A. Mooley, 1980: Large-scale drought/floods and monsoon circulation. Mon. Wea. Rev., 108 , 11971211.

  • Chen, T-C., , and M-C. Yen, 1994: Interannual variation of the Indian monsoon simulated by the NCAR community climate model: Effect of the tropical Pacific SST. J. Climate, 7 , 14031415.

    • Search Google Scholar
    • Export Citation
  • Goswami, B. N., 1998: Interannual variations of Indian summer monsoon in a GCM: External conditions versus internal feedbacks. J. Climate, 11 , 501522.

    • Search Google Scholar
    • Export Citation
  • Hsu, H. H., , C. T. Terng, , and C. T. Chen, 1999: Evolution of large scale circulation and heating during the first transition of Asian summer monsoon. J. Climate, 12 , 793810.

    • Search Google Scholar
    • Export Citation
  • Joseph, P. V., , J. K. Eischeid, , and R. J. Pyle, 1994: Interannual variability of the onset of the Indian summer monsoon and its association with atmospheric features, El Niño, and sea surface temperature anomalies. J. Climate, 7 , 81105.

    • Search Google Scholar
    • Export Citation
  • Ju, J., , and J. Slingo, 1995: The Asian summer monsoon and ENSO. Quart. J. Roy. Meteor. Soc., 121 , 11331168.

  • Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77 , 437471.

  • Kang, I. S., , C. H. Ho, , Y. K. Lim, , and K-M. Lau, 1999: Principal modes of climatological seasonal and intraseasonal variation of the Asian summer monsoon. Mon. Wea. Rev., 127 , 322340.

    • Search Google Scholar
    • Export Citation
  • Kawamura, R., , T. Matsuura, , and S. Iizuka, 2001: Interannual atmosphere-ocean variations in the tropical western north Pacific relevant to the Asian summer monsoon-ENSO coupling. J. Meteor. Soc. Japan, 79 , 883898.

    • Search Google Scholar
    • Export Citation
  • Kim, K. Y., , and G. North, 1997: EOFs of harmonizable cyclostationary processes. J. Atmos. Sci., 54 , 24162427.

  • Kim, K. Y., , and Q. Wu, 1999: A comparison study of EOF techniques: Analysis of nonstationary data with periodic statistics. J. Climate, 12 , 185199.

    • Search Google Scholar
    • Export Citation
  • Krishnamurthy, V., , and J. Shukla, 2000: Intraseasonal and interannual variability of rainfall over India. J. Climate, 13 , 43664377.

  • Krishnamurti, T. N., , P. Ardanuy, , Y. Ramanathan, , and R. Pasch, 1981: On the onset vortex of the summer monsoon. Mon. Wea. Rev., 109 , 344363.

    • Search Google Scholar
    • Export Citation
  • Lau, N-C., , and M. J. Nath, 2000: Impact of ENSO on the variability of the Asian–Australian monsoons as simulated in GCM experiments. J. Climate, 13 , 42874309.

    • Search Google Scholar
    • Export Citation
  • Lau, N-C., , and H. T. Wu, 2001: Principal modes of rainfall-SST variability of the Asian summer monsoon: A reassessment of the monsoon–ENSO relationship. J. Climate, 14 , 28802895.

    • Search Google Scholar
    • Export Citation
  • Lau, N-C., , and M. J. Nath, 2003: Atmosphere–ocean variations in the Indo-Pacific sector during ENSO episodes. J. Climate, 16 , 320.

  • Lau, N-C., , G. T. Yang, , and S. H. Shen, 1988: Seasonal and intraseasonal climatology of summer monsoon rainfall over East Asia. Mon. Wea. Rev., 116 , 1837.

    • Search Google Scholar
    • Export Citation
  • Li, C., , and M. Yanai, 1996: The onset and interannual variability of the Asian summer monsoon in relation to land–sea thermal contrast. J. Climate, 9 , 358375.

    • Search Google Scholar
    • Export Citation
  • Lim, Y-K., , K-Y. Kim, , and H-S. Lee, 2002: Temporal and spatial evolution of the Asian summer monsoon in the seasonal cycle of synoptic fields. J. Climate, 15 , 36303644.

    • Search Google Scholar
    • Export Citation
  • Madden, R. A., , and P. R. Julian, 1972: Description of global-scale circulation cells in the Tropics with a 40–50 day period. J. Atmos. Sci., 29 , 11091123.

    • Search Google Scholar
    • Export Citation
  • Meehl, G. A., , and J. M. Arblaster, 1998: The Asian–Australian monsoon and El Niño–Southern Oscillation in the NCAR climate system model. J. Climate, 11 , 13561385.

    • Search Google Scholar
    • Export Citation
  • Nigam, S., 1994: On the dynamical basis for the Asian summer monsoon rainfall–El Niño relationship. J. Climate, 7 , 17501771.

  • Philander, S. G. H., 1990: El Niño, La Niña, and the Southern Oscillation. 1st ed. Academic Press, 293 pp.

  • Rasmusson, E. M., , and T. H. Carpenter, 1983: The relationship between eastern equatorial Pacific sea surface temperatures and rainfall over India and Sri Lanka. Mon. Wea. Rev., 111 , 517528.

    • Search Google Scholar
    • Export Citation
  • Seo, K-H., , and K-Y. Kim, 2003: Propagation and initiation mechanisms of the Madden-Julian oscillation. J. Geophys. Res., 108 .4384, doi:10.1029/2002JD002876.

    • Search Google Scholar
    • Export Citation
  • Shukla, J., , and D. A. Paolino, 1983: The Southern Oscillation and long-range forecasting of the summer monsoon rainfall over India. Mon. Wea. Rev., 111 , 18301837.

    • Search Google Scholar
    • Export Citation
  • Slingo, J. M., , and H. Annamalai, 2000: 1997: The El Niño of the century and the response of the Indian summer monsoon. Mon. Wea. Rev., 128 , 17781797.

    • Search Google Scholar
    • Export Citation
  • Sperber, K. R., 2003: Propagation and the vertical structure of the Madden–Julian Oscillation. Mon. Wea. Rev., 131 , 30183037.

  • Torrence, C., , and P. J. Webster, 1999: Interdecadal changes in the ENSO–monsoon system. J. Climate, 12 , 26792690.

  • Wang, B., and LinHo, 2002: Rainy season of the Asian–Pacific summer monsoon. J. Climate, 15 , 386398.

  • Wang, B., , R. Wu, , and X. Fu, 2000: Pacific–East Asian teleconnection: How does ENSO affect East Asian climate? J. Climate, 13 , 15171536.

    • Search Google Scholar
    • Export Citation
  • Wang, B., , R. Wu, , and K-M. Lau, 2001: Interannual variability of the Asian summer monsoon: Contrasts between the Indian and the western north Pacific–East Asian monsoons. J. Climate, 14 , 40734090.

    • Search Google Scholar
    • Export Citation
  • Weare, B. C., , and J. N. Nasstrom, 1982: Examples of extended empirical orthogonal function analysis. Mon. Wea. Rev., 110 , 481485.

  • Webster, P. J., , and S. Yang, 1992: Monsoon and ENSO; Selectively interactive systems. Quart. J. Roy. Meteor. Soc., 118 , 877926.

  • Wu, R., , Z-Z. Hu, , and B. P. Kirtman, 2003: Evolution of ENSO-related rainfall anomalies in east Asia. J. Climate, 16 , 37423758.

  • Xie, P., , and P. A. Arkin, 1996: Analyses of global monthly precipitation using gauge observations, satellite estimates, and numerical model predictions. J. Climate, 9 , 840858.

    • Search Google Scholar
    • Export Citation
  • Yanai, M., , C. Li, , and Z. Song, 1992: Seasonal heating of the Tibetan Plateau and its effects on the evolution of the Asian summer monsoon. J. Meteor. Soc. Japan, 70 , 1B. 319351.

    • Search Google Scholar
    • Export Citation
Fig. 1.
Fig. 1.

(a)–(d) The seasonal cycle modes of the boreal ASM precipitation extracted by different analysis techniques. (bottom) The PC time series of the CSEOF (dashed line) and the two extended EOFs (first mode—solid line; second mode—dotted line).

Citation: Journal of Climate 20, 11; 10.1175/JCLI4120.1

Fig. 2.
Fig. 2.

Normalized PC time series (dotted line) of the ENSO mode (fourth CSEOF) of precipitation for 23 yr of the ASM period. The solid line denotes the monthly Niño-3 index time series. The triangles represent El Niño events (along the top) and La Niña events (along the bottom).

Citation: Journal of Climate 20, 11; 10.1175/JCLI4120.1

Fig. 3.
Fig. 3.

The fourth CSEOF (ENSO mode) of the Xie–Arkin precipitation. The figure shows how the ENSO signal evolves in time and space throughout the prominent ASM period from 21 May to 12 September. The contour interval (CI) is 0.5 mm day−1, with values greater than 0.5 heavily shaded and values less than −0.5 lightly shaded.

Citation: Journal of Climate 20, 11; 10.1175/JCLI4120.1

Fig. 4.
Fig. 4.

Time–longitude sections of the (a), (c) seasonal cycle and (b), (d) the El Niño mode of precipitation for (a), (b) 70°–90°E band (India) and (c), (d) 115°–130°E band (East Asia). The thick dashed lines indicate the onset of the typical precipitation spell (top) over India and (bottom) over central China.

Citation: Journal of Climate 20, 11; 10.1175/JCLI4120.1

Fig. 5.
Fig. 5.

Streamlines of anomalous Walker circulation composites along the equator for ENSO developing years: July of (a) El Niño years (1982, 1986, 1991, 1994, and 1997) and (b) La Niña years (1984, 1988, 1995, 1998, and 1999). The ordinate denotes the vertical pressure level in hPa. (c), (d) The same as (a), (b), respectively, but for the anomalous Hadley circulation composites at 125°E (western Pacific). (a)–(d) Significant at 95% level in the sense that the likelihood of observing these patterns during non-ENSO years is less than 5%.

Citation: Journal of Climate 20, 11; 10.1175/JCLI4120.1

Fig. 6.
Fig. 6.

Composites of sea level pressure anomaly in (a) early ASM stage (May and June) and (b) late July and early August in El Niño(0) years. (c), (d) The composites for La Niña(0) years. These patterns are significant at 95% level. The CI is 0.2 hPa, with values greater than 0.4 heavily shaded and values less than −0.4 lightly shaded.

Citation: Journal of Climate 20, 11; 10.1175/JCLI4120.1

Fig. 7.
Fig. 7.

Precipitation anomaly (mm day−1) explained by the ENSO mode at four different regions for the ASM period. The four regions are (a), (b) India, the Bay of Bengal and Bangladesh (7.5°–25°N, 72.5°–90°E); (c), (d) Indonesia (10°S–10°N, 100°–130°E); (e), (f) central China (22.5°–32.5°N, 100°–120°E); and (g), (h) northeast Asia (30°–40°N, 125°–140°E). Precipitation in (a), (c), (e), (g) for El Niño years and in (b), (d), (f), (h) for La Niña years. The period is 21 May–17 Sep of ENSO(0) year and 21 May–9 Jul of ENSO(1) year. There is a small gap between ENSO(0) and ENSO(1) years.

Citation: Journal of Climate 20, 11; 10.1175/JCLI4120.1

Fig. 8.
Fig. 8.

Temporal and spatial evolution of 850-hPa wind (m s−1) and relative vorticity (× 10−6 s−1) of the ENSO mode in Fig. 3. Positive (negative) vorticity anomalies greater (less) than 1 × 10−6 s−1 (−1 × 10−6 s−1) are heavily (lightly) shaded. Wind vectors less than 0.3 m s−1 are omitted with the arrow scale at the bottom.

Citation: Journal of Climate 20, 11; 10.1175/JCLI4120.1

Fig. 9.
Fig. 9.

Same as in Fig. 4, but for 850-hPa zonal wind (m s−1).

Citation: Journal of Climate 20, 11; 10.1175/JCLI4120.1

Fig. 10.
Fig. 10.

Same as in Fig. 8, but for 850-hPa moisture transport (kg kg−1 m s−1) and its convergence (kg kg−1 s−1). The arrow scale is at the bottom, and vectors less than 3 × 10−3 are omitted. Convergence values greater than 1 × 10−9 are heavily shaded, and divergence values less than −1 × 10−9 are lightly shaded.

Citation: Journal of Climate 20, 11; 10.1175/JCLI4120.1

Fig. 11.
Fig. 11.

Same as in Fig. 8, but for 200-hPa wind (m s−1) and its divergence (× 10−7 s−1). Divergence (convergence) anomalies greater (less) than 2 × 10−7 s−1 (−2 × 10−7 s−1) are heavily (lightly) shaded. Wind vectors less than 1 m s−1 are omitted with the arrow scale at the bottom.

Citation: Journal of Climate 20, 11; 10.1175/JCLI4120.1

Fig. 12.
Fig. 12.

The ENSO component (solid line) of the ASM precipitation (mm day−1) plotted against the ±1 pentad running-averaged Indian precipitation anomalies without climatology (bar) at the four regions defined in Fig. 7. (a), (b) The dashed curve is the regional monsoon index defined in Goswami (1998). Precipitation (a), (c), (e), (g) for El Niño years and (b), (d), (f), (h) for La Niña years. The period is 21 May–17 Sep of ENSO(0) year and 21 May–9 Jul of ENSO(1) year. There is a small gap between ENSO(0) and ENSO(1) years.

Citation: Journal of Climate 20, 11; 10.1175/JCLI4120.1

Fig. 13.
Fig. 13.

Correlation of precipitation between (a) observation and the seasonal cycle, (b) observation without the seasonal cycle and the ISO mode, and (c) observational without the seasonal cycle and the ENSO mode. The 23-yr (1979–2001) observed ASM precipitation was ±1 pentad running averaged and then smoothed spatially over 3 × 3 grid boxes.

Citation: Journal of Climate 20, 11; 10.1175/JCLI4120.1

Fig. 14.
Fig. 14.

Schematic diagram showing the three-dimensional circulation structure of the El Niño component of the ASM: (a) early stage (late May–June), and (b) middle stage (July): (top) the 200-hPa level and (bottom) the 850-hPa level. In late July, the lower-level negative pressure anomaly over the subtropical western Pacific extends toward the Bay of Bengal and northern India [see dotted line at the bottom of (b)].

Citation: Journal of Climate 20, 11; 10.1175/JCLI4120.1

Save