Asian Summer Monsoon—ENSO Feedback on the Cane–Zebiak Model ENSO

Chul Chung Department of Meteorology, University of Maryland at College Park, College Park, Maryland

Search for other papers by Chul Chung in
Current site
Google Scholar
PubMed
Close
and
Sumant Nigam Department of Meteorology, University of Maryland at College Park, College Park, Maryland

Search for other papers by Sumant Nigam in
Current site
Google Scholar
PubMed
Close
Full access

Abstract

The Asian summer monsoon heating anomalies are parameterized in terms of the concurrent ENSO SST anomalies and used as additional forcing in the Cane–Zebiak (CZ) Pacific ocean–atmosphere anomaly model. The Asian heating parameterization is developed from the rotated principal component analysis of combined interannual variability of the tropical Pacific SSTs, residually diagnosed tropical diabatic heating at 400 mb (from ECMWF’s analyses), and the 1000-mb tropical winds during the 1979–97 summer months of June, July, and August.

Analysis of the 95 000-yr-long model integrations conducted with and without the interactive Asian sector heating anomalies reveals that their influence on the Pacific surface winds leads to increased ENSO occurrence—an extra ENSO event every 20 yr or so. An examination of the ENSO distribution w.r.t. the peak SST anomaly in the eastern equatorial Pacific shows increased El Niño occurrence in the 2.2–3.6 K range (and −1.0 to −1.6 K range in case of cold events) along with a modest reduction in the 0.6–1.2 K range, that is, a population shift due to the strengthening of weak El Niños in the monsoon run. The interaction of ENSO-related Asian summer monsoon heating with the CZ model’s ocean–atmosphere also results in a wider period distribution of ENSO variability, but with the El Niño peak phase remaining seasonally locked with the northern winter months.

The above modeling results confirm the positive feedback between Asian summer monsoon and ENSO suggested by previous empirical and diagnostic modeling studies; the feedback is generated primarily by the diabatic heating changes in the Asian Tropics.

Corresponding author address: Dr. Sumant Nigam, Room 3403, Computer and Space Sciences Bldg., Dept. of Meteorology, University of Maryland at College Park, College Park, MD 20742.

Email: nigam@atmos.umd.edu

Abstract

The Asian summer monsoon heating anomalies are parameterized in terms of the concurrent ENSO SST anomalies and used as additional forcing in the Cane–Zebiak (CZ) Pacific ocean–atmosphere anomaly model. The Asian heating parameterization is developed from the rotated principal component analysis of combined interannual variability of the tropical Pacific SSTs, residually diagnosed tropical diabatic heating at 400 mb (from ECMWF’s analyses), and the 1000-mb tropical winds during the 1979–97 summer months of June, July, and August.

Analysis of the 95 000-yr-long model integrations conducted with and without the interactive Asian sector heating anomalies reveals that their influence on the Pacific surface winds leads to increased ENSO occurrence—an extra ENSO event every 20 yr or so. An examination of the ENSO distribution w.r.t. the peak SST anomaly in the eastern equatorial Pacific shows increased El Niño occurrence in the 2.2–3.6 K range (and −1.0 to −1.6 K range in case of cold events) along with a modest reduction in the 0.6–1.2 K range, that is, a population shift due to the strengthening of weak El Niños in the monsoon run. The interaction of ENSO-related Asian summer monsoon heating with the CZ model’s ocean–atmosphere also results in a wider period distribution of ENSO variability, but with the El Niño peak phase remaining seasonally locked with the northern winter months.

The above modeling results confirm the positive feedback between Asian summer monsoon and ENSO suggested by previous empirical and diagnostic modeling studies; the feedback is generated primarily by the diabatic heating changes in the Asian Tropics.

Corresponding author address: Dr. Sumant Nigam, Room 3403, Computer and Space Sciences Bldg., Dept. of Meteorology, University of Maryland at College Park, College Park, MD 20742.

Email: nigam@atmos.umd.edu

1. Introduction

The El Niño/La Niña phenomena—the opposite phases of the Southern Oscillation—have been the focus of intense analysis, coupled modeling, and prediction research in view of their widespread impact on atmospheric circulation and global precipitation (Bjerknes 1969; Horel and Wallace 1981; Rasmusson and Wallace 1983; Philander 1985; Ropelewski and Halpert 1987) during both the winter and summer seasons. The extratropical circulation anomalies in the Pacific–North American sector during northern winter (Horel and Wallace 1981) and the Asian summer monsoon rainfall anomalies (Rasmusson and Carpenter 1983) are leading examples of the El Niño–Southern Oscillation’s (ENSO) impact on midlatitude and tropical climates.

The phrase “Asian summer monsoon” inclusively represents the seasonal circulation and rainfall over the Asian continent and the tropical Indian ocean basin; the inclusion of both landmasses and adjoining seas in the monsoon domain in this and most other studies (e.g., Ramage 1971; Webster and Yang 1992)1 follows from the canonical definition of the monsoon that involves seasonal land–sea contrasts in sensible and, ultimately, deep latent heating.

The linkage between Asian summer monsoon and ENSO—the two global-scale climate systems—has been examined using both observations and dynamical models. Rasmusson and Carpenter (1983) showed that extreme summertime droughts in India tend to occur during the moderate/strong El Niño years, and later Bhalme et al. (1990) showed the eastern equatorial Pacific SST to be inversely related to the Indian summer monsoon rainfall. The impact of Asian monsoon circulation and precipitation anomalies on subsequent ENSO evolution has also been investigated; empirical analysis (Meehl 1987; Yasunari 1991; Webster and Yang 1992) and diagnostic modeling experiments (Nigam 1994) indicate that Asian summer monsoon rainfall anomalies impact the Pacific trade winds and therefore, potentially, the ENSO evolution. Lag-correlation analysis does show the Indian summer monsoon rainfall variations to be most strongly (and negatively) correlated with the central–eastern tropical Pacific SST anomalies in the post–summer monsoon season than during any other period (Verma 1992; Lau and Yang 1996), indicating a positive feedback role of the Asian summer monsoon anomalies in ENSO evolution. The active role of the Indian monsoon in the Southern Oscillation evolution was first recognized by Sir Gilbert Walker (1923, 1924), who noted that the Southern Oscillation index during June–August was more strongly correlated with the following winter’s index (+0.8) than with the preceding winter’s one (−0.2). Sir Charles Normand (1953) succinctly summarized the role of the Indian summer monsoon in the Southern Oscillation evolution in his presidential address to the Royal Meteorological Society in 1953 titled “Monsoon Seasonal Forecasting.”

Unfortunately for India, the southern oscillation in June–August—at the height of the monsoon—has many significant correlations with later events and relatively few with earlier events. . . . The Indian monsoon therefore stands out as an active, not a passive feature in world weather, more efficient as a broadcasting tool than as an event to be forecast.

Thus Asian summer monsoon variability, particularly when ENSO related, warrants inclusion in a theory explaining ENSO evolution and in ENSO prediction models. This effort has, however, been hindered by the incomplete understanding of the mechanism(s) through which ENSO-related tropical Pacific basin SST (heating) anomalies influence the Asian monsoon precipitation and circulation. An important unresolved issue is whether the summer monsoon is influenced only by the concurrent El Niño SST (heating) anomalies, or are the preceding period SST anomalies also important, potentially, through their impact on the Asian land surface and Indian ocean circulation, both of which have at least a season-long memory? The efficacy of various predictors is discussed in a recent review of the Indian summer monsoon rainfall prediction by statistical methods (Hastenrath 1990).

Even in case of the concurrent influence, the dynamical basis for ENSO’s impact on the monsoon is yet to be fully delineated. Diagnostic modeling of monsoon circulation anomalies during the contrasting Indian summer monsoon rainfall years 1987 (weak) and 1988 (strong) by Nigam (1994) indicated that the ENSO-related spring-to-summer Pacific basin heating anomalies regulate the low-level monsoon westerlies, and thus the large-scale moisture flux into Indochina, while the convergence of this moisture flux was controlled by the anomalous lower-tropospheric stationary waves generated by the interaction of the zonal mean circulation anomalies and the Himalayan–Tibetan orography. Ju and Slingo’s (1995) analysis of the GCM simulation of Asian monsoon variability in ENSO years corroborates the importance of the spring season Pacific heating anomalies; their modeling results suggest that latitudinal displacements of the Intertropical Convergence Zone (ITCZ, or deep convective heating anomalies) in the western Pacific, in particular, are linked to the modulation of the low-level southwesterly flow over the equatorial Indian ocean, and that the ENSO impact on the monsoon circulation is strongest during the monsoon onset months of May and June.

The interannual variability of the Asian summer monsoon is, however, not all related to ENSO; a significant portion of it apparently results from internal dynamics (Palmer et al. 1992) and from interactions with the underlying land surface properties, such as soil moisture (Meehl 1994) and the Eurasian snow cover (Dickson 1984; Vernekar et al. 1995). The ENSO-unrelated variability of the Asian summer monsoon can also impact subsequent ENSO evolution, contributing, perhaps, to the aperiodicity of the oscillation (Wainer and Webster 1996), but such monsoon variability is not modeled in this study because of our incomplete understanding of the involved processes and because of difficulties in representing it in intermediate coupled models.

The impact of anomalous Asian summer monsoon on ENSO evolution has been examined by Yamagata and Masumoto (1989), Wainer and Webster (1996), and An (1996). Except for the first one, these studies do not address the feedback issue as the monsoon heating anomalies are prescribed and are therefore unresponsive to the state of the Pacific SSTs; furthermore, the structure of monsoon heating anomalies is highly idealized in all three investigations; for example, in the second study, the anomalies are generated by the amplitude variation of the climatological Asian–African heating pattern, which is assumed to have a meridionally dipolar structure. On the other hand, the positive feedback of the Asian summer monsoon on ENSO evolution must be implicitly included in Barnett et al.’s (1993) hybrid coupled model and coupled GCMs, but the impact of the monsoon feedback itself is difficult to isolate, for example, given the empirical parameterization of the Pacific surface wind stress anomalies in terms of the Pacific SST anomalies in the first case.

In this pilot-phase modeling study, the difficulties in dynamically modeling ENSO’s impact on the Asian summer monsoon are circumvented by using an empirical parameterization of the Asian sector heating anomalies, based on the concurrent Pacific basin SST anomalies; this study was motivated by the desire to determine the feedback of realistic ENSO-related Asian monsoon heating anomalies on subsequent ENSO evolution.

The Cane–Zebiak Pacific ocean–atmosphere anomaly model (the CZ model; Zebiak and Cane 1987), a widely used intermediate-level model for ENSO simulation and prediction research, was chosen as the investigative tool in the present study as none of the proposed monsoon–ENSO linkage mechanisms posit an active role for the Indian Ocean SST anomalies. The suitability of using a “Pacific basin only” ocean model in this investigation might be called into question, given the phenomenal SST and outgoing longwave radiation anomalies (of both signs) in the Indian ocean during the 1997 fall and 1997/98 winter months (Webster et al. 1999).

The objective of this study is to determine the modulation of the CZ model’s ENSO characteristics (amplitude, frequency, episode duration, etc.) from inclusion of the Asian summer monsoon feedback in this coupled model. The interactive Asian sector heating anomalies, included additionally in the Gill-type (Gill 1980) CZ atmospheric model, are determined at each model time step from an empirical parameterization based on the rotated principal component analysis of combined interannual variability of the Pacific basin SST, 1000-mb winds, and the residually diagnosed 400-mb diabatic heating in the near-global Tropics. Although similar in concept to the empirical parameterization of surface wind stress (Barnett et al. 1993), our scheme, based on combined rather than individual empirical orthogonal functions, provides a parameterization of the forcing (monsoon heating anomalies) rather than the response. The CZ model’s SSTs at each time step are projected on the SST components of the first few leading combined modes, and these projections are then used to construct the associated 400-mb heating anomaly over the Asian sector during the northern summer season.

The datasets and the residual diagnosis of the three-dimensional diabatic heating field are described in section 2, while aspects of the CZ model are discussed in section 3. The empirical parameterization of the Asian sector heating anomalies in terms of concurrent Pacific basin ENSO SST anomalies is described in section 4. The differences between the “control” (original CZ model) and “monsoon” (CZ model with interactive Asian summer monsoon heating anomalies) integrations in the peak-amplitude distribution of ENSO events and their evolution, and the SST spectrum are examined in section 5; this section also discusses the sensitivity of the peak-amplitude distribution to the Asian sector heating amplitude. Summary and concluding remarks follow in section 6.

2. Datasets and the residual diagnosis of diabatic heating

a. Sea surface temperature

The sea surface temperature data was obtained from the Climate Data Library of the Lamont-Doherty Earth Observatory (LDEO), where it is archived as the Climate Analysis Center sea surface temperature (CAC SST). The monthly record extends from January 1970 to February 1998, and the data are available on a 2° × 2° latitude–longitude grid from 29°S to 29°N, and from 124°E to 70°W. The LDEO’s CAC SST record is obtained by appending the optimum interpolation SST (Reynolds and Smith 1994) to the January 1970–October 1981 period in situ and blended analysis of SST (Reynolds 1988). The CAC SSTs are linearly interpolated in this study on to the 2° × 5.625° latitude–longitude CZ model’s SST grid after an application of a 1–2–1 filter in the longitudinal direction.

b. Xie–Arkin precipitation

The Climate Prediction Center’s merged analysis of global monthly precipitation produced by Xie and Arkin (1997) on a 2.5° × 2.5° global grid for the 1979 onward period was obtained from the National Centers for Environmental Prediction (NCEP). The merged precipitation is generated by combining gauge observations with the satellite estimates derived from the infrared, outgoing longwave, and microwave scattering and emission-based precipitation indices; for additional algorithm details and comparison of the merged product with gauge observations, see Xie and Arkin (1997).

c. ECMWF reanalyses and operational analyses

The 6-h initialized2 European Centre for Medium-Range Weather Forecasts (ECMWF) reanalyses archived at the National Center for Atmospheric Research (NCAR) on a 2.5° × 2.5° global grid and at 17 pressure levels for the January 1979–December 1993 period were used to compute the monthly mean analysis fields and the submonthly thermal transients from which the 3D diabatic heating was diagnosed using the method described in the next section.

In the interest of analyzing summertime interannual variability in a more expanded period, particularly one that included the phenomenal 1997/98 El Niño, the 15-yr-long ECMWF reanalysis–based circulation and diagnosed heating record was extended beyond December 1993 using ECMWF’s uninitialized operational analyses, which are available on a 2.5° × 2.5° global grid at 15 pressure levels from NCAR for the January 1985 onward period.

d. Residual diagnosis of diabatic heating

The 3D diabatic heating is diagnosed as a residual in the thermodynamic equation (e.g., Hoskins et al. 1989;Nigam 1994):
i1520-0442-12-9-2787-e1
This diagnosis, however, does not provide any information about the constituent sensible, latent, and radiative heating components whose knowledge should be helpful in understanding the monsoon–ENSO linkage.

The residual calculation (1) requires a consistent vertical velocity (ωdp/dt), which is available in the initialized reanalyses; however, ω diagnosed from the uninitialized analyses is often suspect. In view of the dominant balance in the Tropics between and ω[T(∂ lnϒ)/∂p] in (1), that is, between and (−N2ω) (e.g., Holton 1992), the monthly averaged ω was diagnostically recalculated using the time-averaged continuity equation and the mass-balance constraint when diagnosing heating from the ECMWF’s 1985–97 period uninitialized analyses; the mass-balance constraint was imposed using the technique first suggested by O’Brien (1970).

The 9-yr overlap (1985–93) between ECMWF’s initialized reanalyses (1979–93) and the uninitialized operational analyses (1985–present) provides the opportunity to compare the 3D heating diagnosed from these datasets; to be sure, there are significant differences in the heating strength, with the tropical heating (and divergent circulations) stronger in the ECMWF reanalysis. In the absence of an observational counterpart, the credibility of a heating estimate can be gauged only by the extent of its dynamical consistency with the observed/analyzed large-scale circulation, that is, through modeling, and such an evaluation shows the uninitialized analyses–based heating (Nigam 1994, 1997) and, even more, the ECMWF reanalyses–based counterpart (Nigam and Chung 1998, manuscript submitted to J. Climate) to be reasonably consistent estimates.3 The diagnosed heating anomalies in an El Niño and La Niña summer month are shown in Figs. 2 and 3, and discussed in the next section.

3. Cane–Zebiak ocean–atmosphere anomaly model

The atmospheric component of the CZ coupled model is essentially a Gill-type steady-state linear shallow water model on an equatorial β plane (i.e., the Coriolis parameter = βy) that solves for the horizontal flow associated with the first internal baroclinic mode, that is, for flows having opposite polarities in the upper and lower troposphere. The deep convective heating profile in the tropical Pacific precipitation regions has largest amplitude at approximately the 400-mb level (e.g., Reed and Recker 1971; Hartmann et al. 1984; Nigam 1994), leading to convergence (divergence) below (above) this level, that is, to a flow structure similar to that represented by the first internal baroclinic mode. The momentum dissipation is modeled as Rayleigh friction while the radiative cooling is represented by the damping of geopotential perturbations in the model.

The CZ coupled model has many parts and therefore different computational domains: for example, the SST and atmospheric heating anomalies are computed over the tropical Pacific (19°S–19°N, 129.375°E–84.375°W), while the atmospheric model calculates the global response to the tropical Pacific heating anomalies. We expand the model’s heating domain in both the zonal and meridional directions (31°S–39°N, 56.25°E–11.25°W) to include the Asian, and potentially the American, monsoon regions. The original and expanded heating domains are shown in Fig. 1, which also shows the Asian monsoon region (shaded) where the empirically parameterized heating anomalies are specified during June, July, and August. Over the tropical Pacific, however, the “expanded domain CZ model” generates heating anomalies all through the year, exactly as in the original model; thus if the parameterized Asian sector heating were set to zero, the modified model would reduce to the original CZ model.

Zebiak and Cane (1987) discuss two versions of their coupled model that differ in the memory of the heating calculation algorithm. We obtained the 1996 version of the CZ model code directly from the LDEO, and in this version the memory is reset whenever the NINO3 SST anomaly index [defined as the SST anomaly average in the 5°S–5°N, 150°–90°W sector; see Kousky (1998)] is between −0.1 and 0.1 K; if this index exceeds ±0.1 K, however, the heating anomalies are simply incremented.

The deep convective heating is parameterized in the original CZ model using both evaporation from local SST anomalies and the low-level moisture convergence feedback; the latter generates deep heating only when the sum of the climatological and anomalous low-level winds is convergent. Figure 2 compares the model parameterized and the residually diagnosed heating anomalies with the Xie–Arkin precipitation anomalies in July 1997, an El Niño summer month as evident from the SST anomalies (Fig. 2d); the 400-mb diabatic heating (Q400) is diagnosed from ECMWF’s July 1997 operational analyses, and the anomalies are computed w.r.t. the 1985–97 climatology.

The striking feature in this comparison is the strong spatial coherence of the CZ model heating and SST anomalies (Figs. 2c,d), indicating dominance of the thermodynamic mechanism in the generation of the CZ model heating anomalies. The diagnosed Q400 (Fig. 2b), on the other hand, has very little resemblance with the underlying SSTs, particularly over the western tropical Pacific where Q400 is large but SST anomalies insignificant, over the southeastern tropical Pacific where Q400 is notably weak in the presence of large SST anomalies, and over the northern Tropics (10°–20°N) where the Q400 and SST anomalies even have the opposite sign. The oppositely signed Q400 anomalies represent equatorward shift of the ITCZ, but such an ITCZ shift is conspicously absent in the CZ model heating anomalies. This lack of resemblance is indicative of a much more prominent pole of the tropical circulation (e.g., lower-tropospheric moisture flux convergence, upper-tropospheric divergence and subsidence) in the generation of ENSO-related deep heating anomalies, in nature. Note that the CZ model parameterized deep heating anomalies are also 2–3 times stronger than diagnosed Q400 anomalies. The Xie–Arkin precipitation anomaly in July 1997,4 shown in the top panel of Fig. 2, corroborates most of the diagnosed Q400 heating features; in fact, the correspondence between the two anomaly patterns is quite striking.

During July 1988 (a La Niña summer month), the two heating anomalies (Figs. 3b,c) again exhibit substantial differences in structure and amplitude over the western and eastern tropical Pacific sectors, for example, the CZ model erroneously generates deep diabatic cooling over the eastern and, particularly, the southeastern tropical Pacific where the CZ model’s heating is overly sensitive to the underlying SST anomalies.

The above comparisons indicate that the CZ model’s parameterization of midtropospheric heating anomalies from the Pacific basin SST anomalies can be significantly improved. The development of an improved deep heating parameterization is, however, not the objective of this investigation (and is being pursued separately), and all coupled modeling experiments reported in this study were undertaken using the CZ model’s own parameterization for the Pacific basin heating anomalies, along with an empirical parameterization of the ENSO-related Asian sector heating anomalies. The empirical heating parameterization was not used over the CZ model’s Pacific basin in order to draw upon the large body of extant research on the stability and ENSO characteristics of the CZ model (e.g., Mantua and Battisti 1995; An 1996), the “control” case in this study.

4. Empirical parameterization of ENSO-related monsoon heating anomalies

The difficulties in dynamically modeling the impact of ENSO on Asian summer monsoon rainfall and the associated deep heating are circumvented in this pilot-phase study by developing an empirical parameterization for the June, July, and August midtropospheric (400 mb) heating anomalies over the Asian sector in terms of the concurrent tropical Pacific basin SST anomalies. The assumption here is that the dynamic response of the atmosphere is established quite rapidly in comparison with the slowly evolving ENSO SST anomalies.

The intent here is to include only the feedback effect of the ENSO-related Asian summer monsoon variability on ENSO evolution. Thus, only those monsoon heating anomalies that are generated from ENSO variability need to be parameterized. As discussed in the introduction, the Asian summer monsoon variability can result also from the anomalous winter/spring Eurasian snow cover, that is, from anomalies in the land surface properties in the premonsoon seasons. Although the linkage of these anomalies with ENSO variability remains uncertain, the feedback of these anomalies on ENSO evolution will be implicitly included through the modulation of the Asian summer monsoon to the extent these land surface anomalies are themselves related to ENSO evolution.

a. Extraction of ENSO covariant Asian sector heating anomalies

The ENSO covariant Asian monsoon heating anomalies are extracted from a rotated principal component analysis (RPCA) of combined interannual variability of the tropical Pacific SST, and the 1000-mb zonal and meridional winds and 400-mb diabatic heating in the near-global Tropics during the 1979–97 summer months of June, July, and August. The 1000-mb winds are included in the combined analysis to ascertain the structure of ENSO covariant winds during the northern summer that, unlike the corresponding December–February structure, is not widely known, particularly, over the western tropical Pacific sector. The 1000-mb winds and diagnosed heating are obtained principally from the ECMWF reanalyses, that is, data for 15 of the 19 summers comes from the reanalysis fields.

The RPCA technique extracts recurrent modes of combined variability by simultaneously analyzing the structure of autocovariance and cross-covariance matrices. The efficacy of this method in extracting the truly coupled variability modes increases with the number of variables in the combination (Nigam and Shen 1993; Nigam and Chao 1996); in case of a four-variable combination, as in the undertaken analysis, the cross-covariance submatrices outnumber the autocovariance ones by 3 to 2. When analyzing combined variability, the individual variables are put on par with each other by dividing each of them by the square root of the sum of temporal variance over that variable’s spatial grid. Note that this kind of normalization is preferable to the one based on local standard deviation, as discussed in section 2 of Nigam and Shen (1993), and even to the one based on spatially averaged variance when the variables differ significantly in the number of spatial grid points. The results presented in this study are obtained from orthogonal rotation of 10 loading vectors using the VARIMAX criterion (e.g., Horel 1981), which destroys the spatial orthogonality of the vectors but leaves their temporal orthogonality intact.

In view of the earlier noted differences between the ECMWF reanalyses and operational analyses–based heating estimates, the heating (and 1000-mb wind) anomalies are calculated in the two datasets from their own period monthly climatologies (1979–93 and 1985–97, respectively), before being appended together to form the 1979–97 anomaly record. In order to ascertain the impact of any remaining temporal discontinuity in the extended anomaly record, the variability was analyzed first using 15 yr of ECMWF reanalyses (1979–93) augmented by 4 yr of operational analyses (1994–97) and again using 6 yr of reanalyses (1979–84) supplemented by 13 yr of operational analyses (1985–97). The ENSO-related combined variability modes obtained from these somewhat different 1979–97 anomaly records were almost identical, and the leading mode extracted from the former record is displayed in Fig. 4.

The 400-mb diabatic heating, Pacific SST, and the 1000-mb wind anomalies associated with the leading summertime interannual variability mode are displayed in Fig. 4; the SST anomalies and the principal component (i.e., time series) indicate the obvious linkage of this pattern with ENSO variability. The 1982/83, 1987, and the phenomenal 1997 El Niño are all captured, including the anomalous evolution of the 1982/83 event, in which the eastern equatorial Pacific warmed up later during the 1983 spring–summer (e.g., Gill and Rasmusson 1983); the weak SST cooling during 1981 and 1994–96 summers, a somewhat stronger cooling during 1984–85, and the 1988 La Niña are also evident in the time series.

The 400-mb heating anomalies depict the southward retreat of the ITCZ over the central/eastern Pacific sector, and an impressive reduction of deep heating (convective precipitation) over the eastern Indian ocean and the northern Tropics of the western Pacific during summer of the El Niño year. There is also some evidence of heating (precipitation) redistribution over Indochina, with reduced deep heating over northern India, and strengthened heating over the central and northeastern China; note that the heating reduction over northern India is in qualitative accord with Rasmusson and Carpenter’s (1983) finding of a linkage between the moderate to severely deficient Indian monsoon rainfall years and El Niño events. A substantial attenuation of deep heating is also evident over Central America, along with hints of a weakened Mexican monsoon. Finally, the representativeness of ENSO-related Pacific heating anomalies (Fig. 4a) can be gauged from comparisons with heating anomalies diagnosed for individual ENSO events (Figs. 2–3).

b. Impact of Asian sector heating anomalies in the CZ atmospheric model

The ENSO-related surface wind anomalies contain strong equatorial westerlies in the western/central Pacific sector (Fig. 4c), and it is clearly of interest to determine the contribution of deep heating anomalies in the Asian sector in the generation of these summertime westerly anomalies. Although evidence for the impact of anomalous Asian summer monsoon on Pacific trade winds has been presented in earlier analysis and modeling studies (cited in the introduction), we examine here the contribution of realistic ENSO-related deep heating anomalies in the Asian sector (the shaded region in Fig. 1) in generating the Pacific basin surface winds, using the CZ atmospheric model. Figure 5a displays the surface zonal wind forced by the Asian sector heating anomalies, while Fig. 5b shows the response of ENSO heating anomalies confined to the CZ model’s original heating domain over the Pacific;5 to facilitate compar- ison, the ENSO covariant 1000-mb zonal wind (Fig. 4c) is contoured in Fig. 5c. Note that in all cases, the wind response is shown only over the tropical Pacific, because it is only here that the SSTs are computed in the CZ model.

It is quite apparent from Fig. 5 that the ENSO-related deep heating anomalies over the Asian sector, not included in the original CZ coupled model, can significantly impact the zonal winds over the western Pacific (till the date line). In fact, in the 150°–160°E sector, where the Pacific basin forced westerlies are the strongest (∼0.6 m s−1 in Fig. 5b), the Asian sector forced response (∼0.2 m s−1) is about one-third as large.6 Further modeling decomposition shows that the zonal wind response in Fig. 5a is forced largely by the tropical part of the Asian sector heating anomalies (e.g., those within ±20°); the contribution of the subtropical part (20°–39°N) was evaluated using both the equatorial β-plane and variable-f versions of the model, and in both cases, the Pacific surface zonal winds were found to be at least an order of magnitude smaller than those forced by the tropical heating anomalies. The tropical heating anomalies in the Asian sector, for example, over the Indian ocean, are, of course, not independent of the anomalies over the Asian continent, for they are both generated as part of the larger ENSO-related Asian monsoon anomaly pattern, as evident from the leading variability mode structure (Fig. 4a).

Can the surface winds forced by ENSO-related Asian sector heating anomalies influence coupled evolution of the CZ model, in particular, the ENSO characteristics? In order to investigate such questions, an empirical parameterization of the ENSO-related Asian sector heating anomalies in terms of Pacific SST anomalies is developed in the next section.

c. Empirical parameterization of Asian sector heating

The task of obtaining ENSO-related deep heating anomalies by projecting a given SST anomaly on the combined modes’ SST components is not entirely straightforward, as the rotated modes (or loading vectors) are not spatially orthogonal. If rotation was not performed, the combined modes would be orthogonal, but even so, the SST components need not be spatially orthogonal. In the parameterization developed here, the projection is calculated using the least squares method, applied over the SST subspace of the chosen rotated modes of combined variability; our preliminary analysis indicated that the combined RPCA technique with subsequent least squares projection was as effective in anomaly reconstruction as the individual EOF based method of Barnett et al. (1993) and the SVD approach of Syu et al. (1995).

The least squares based reconstruction of SST anomalies works as follows. If the (N × M) matrix E consists of the SST components of the combined modes, [T1, T2, . . . , TM], where each column vector T is the “N grid points” long SST component of a mode, and M is the number of modes being considered (2 in our case), then any arbitrary SST anomaly vector (S) can be expressed as S = ΣMαiTi, with the least squares projections αi = (ETE)−1(ETS)i. Note that this method does not require any additional scaling.

The undertaken RPCA analysis of summertime interannual variability reveals that the ENSO-related variability is not all captured by the leading mode (Fig. 4), which mainly represents the variability associated with SST variations in the eastern equatorial Pacific, or the NINO3 sector. The second leading mode (Fig. 6), on the other hand, appears linked to interannual SST variations in the central equatorial Pacific, or the NINON3.4 sector (5°S–5°N, 170°–120°W), and explains about 10% of the combined variance.

The second mode’s principal component (Fig. 6d) is strongly positive during JJA 1982 and negative in JJA 1983, in contrast with the leading mode’s evolution in these summers, indicating that this mode represents variability resulting from the departure of individual El Niño/La Niña evolutions from their canonical pattern (Fig. 4). On the other hand, the projections after 1988 represent even lower-frequency interannual variability—perhaps aliased decadal timescale variability, given the 19-yr-long data record and the confinement of SST analysis to the tropical Pacific basin alone.7

The heating anomalies associated with the two modes are broadly similar, but the one in the second mode (Fig. 6a) is more strongly focused over the Maritime Continent and western Pacific longitudes. Over Indochina, the heating anomalies in the second mode are weaker but the reduction in Indian summer monsoon rainfall is more widespread, in contrast with the redistribution seen in the leading mode heating anomalies (Fig. 4a). There are interesting differences in the surface wind anomalies as well, with the one linked to the second mode (Fig. 6c) exhibiting greater zonality, particularly in the central and eastern equatorial Pacific, consistent with the weaker meridional SST gradients in Fig. 6b in that region (Lindzen and Nigam 1987).

The robust zonal wind variability in the western/central Pacific basin in conjunction with large variations of deep heating over the eastern Tropics and the representation of aspects of ENSO variability mandate the inclusion of the second leading mode in the parameterizaton of ENSO-related Asian sector heating anomalies.

The parameterized deep heating anomalies in an El Niño (July 1997) and La Niña (July 1988) summer month are shown in Fig. 7. Over the Asian and Pacific sectors, the heating anomalies are rather similar except for the sign difference, as the corresponding Pacific SST anomalies are likewise (Figs. 2d and 3d). Comparison of the parameterized (Figs. 7a,b) and the residually diagnosed deep heating anomalies (which contain both ENSO-related and unrelated components; Figs. 2b and 3b, respectively) indicates that the heating anomalies are better modeled in the El Niño case, particularly, over the Maritime Continent and the eastern Indian ocean sector, perhaps, because of greater occurrence of El Niños in the 1979–97 record (on which the parameterization is based).

Finally, given the large Pacific heating amplitudes generated by the CZ model during both El Niño and La Niña events—about 3 times larger than the diagnosed 400-mb heating anomalies over the central equatorial Pacific in July 1997 and July 1988 (cf. Figs. 2 and 3)—the parameterized ENSO-related Asian sector heating anomalies are scaled by a factor of 3 before inclusion in the CZ atmospheric model. The modeling analysis of section 4b indicates that such scaling of observed heating anomaly produces reasonable surface wind in the presence of large Rayleigh damping in the CZ model. The heating amplitude discrepancy will be larger in case of stronger ENSO events, as the CZ model heating is dominated by the evaporative flux contribution, and so the use of constant scaling may diminish the impact of interactive Asian heating anomalies on the evolution of stronger ENSO events in the CZ model.

The ENSO-related Asian sector heating anomalies were inserted at each model time step [Δt = (1/36) yr] during the summer months of June–August.

5. Impact of interactive Asian monsoon heating on CZ model’s ENSO

The impact of ENSO related Asian summer monsoon heating anomalies on the CZ model’s ENSO characteristics is evaluated through extended coupled model integrations. The standard CZ model is first integrated for 95 000 yr (the control run), archiving the monthly SST, zonal wind, upper-layer thickness (or thermocline depth), and deep heating in the equatorial Pacific sector. The modified model, with ENSO-related Asian sector heating parameterized in terms of Pacific basin SST anomalies, is also run for 95 000 yr (the monsoon run), archiving the above fields and the deep heating anomalies in the Asian longitudes.

This study employs the Japan Meteorological Agency’s (JMA) SST index (e.g., O’Brien et al. 1996), which is the SST anomaly averaged in the 4°S–4°N, 150°–90°W equatorial box.8 The standard deviations of the JMA SST index are 1.359 and 1.465 K in the control and monsoon runs, respectively; the increased standard deviation is indicative of the positive feedback between Asian summer monsoon and ENSO. The modulation of ENSO characteristics are analyzed for warm and cold events separately, and these are identified using the five-month running average of the JMA SST index (hereafter simply the JMA index); a warm event is defined to occur in the CZ model integrations when the JMA index exceeds 0.5 K for at least six consecutive months, while the cold event occurrence is based on this index being less than −0.4 K for at least six successive months. The slight asymmetry in index threshold in the warm and cold event definitions is due to the CZ model’s bias toward stronger warm anomalies.

a. Impact on CZ model’s ENSO frequency

We first examine the distribution of warm and cold events w.r.t. the peak JMA index in Fig. 8, which shows the distributions for both the control and monsoon runs using a 0.2 K index interval. In the control run, there are 38 205 ENSO events, with the warm events distributed over a much wider amplitude range than the cold events; the model’s bias toward stronger warm events is also evident from the sizable population of warm events in the 3.0–5.0 K index range. Interestingly, both the warm and cold event distributions exhibit bimodality in the control case, but this feature is considerably attenuated in the monsoon case distributions, which are displayed using solid dots (connected by a line).

The 95 000-yr monsoon run has 42 655 ENSO events, which represent, approximately, a 12% increase in ENSO occurrence over the control case; the increase in ENSO occurrence is statistically significant because the standard error of the number of ENSO events in our 95 000-yr long control case sample is estimated to be less than 400.9 A comparison of the two distributions shows that ENSO occurrence is enhanced in the monsoon run primarily in the 2.2–3.6 K index range in case of warm events, and the −1.0 to −1.6 K range in case of cold events. In both ranges, the enhancements are statistically significant as the error bars representing ±2.101 std dev (the std dev was inferred from the variance of ENSO occurrence in 19 nonoverlapping 5000-yr-long subsamples, and the factor 2.101 denotes the 95% confidence level in t statistics at 18 degrees of freedom) are well separated. The concurrent reduction in the number of warm events in the 0.6–1.2 K range indicates that the above-noted population increase results, in part, from a population shift, due to strengthening of El Niños in the monsoon run. Interestingly, the inclusion of ENSO-related Asian sector heating in the CZ model results in a diminution of the bimodality seen in the control case distributions.

In summary, there are more ENSO events in the monsoon run, with the frequency increasing from 0.40 yr−1 in the control case to 0.45 yr−1 in the monsoon run, that is, an extra ENSO event every 20 yr or so. The ENSO events will also likely be strong, for example, with peak JMA indices between 2.2 and 3.6 K in case of warm events, and between −1.0 and −1.6 K for cold events.

b. Impact on CZ model’s ENSO evolution

An assessment of the impact of the interactively generated ENSO-related Asian sector heating anomalies on CZ model’s evolution poses somewhat of a dilemma, given the significant changes in ENSO strength distribution between the control and monsoon runs. If composites based on the peak JMA index are compared, the potential impact on SST amplitudes may not be clearly revealed; on the other hand, if compositing is based on the percentile range of the peak JMA index, the impact on evolution may be obscured as ENSOs with different peak amplitudes have, intrinsically, somewhat different evolutions.

We begin by displaying the composited evolution of the warm events whose peak JMA indices are within the 23d and 27th percentiles; note that this percentile range corresponds to a peak JMA index range of 1.41–1.57 K in the control distribution and 1.71–1.89 K in the monsoon case, and has 781 and 861 members, respectively. Figure 9 displays the composited equatorial (4°S–4°N) SST evolutions over a 3-yr period beginning in January of the year preceding the one in which the JMA index peaks. The evolution difference, shown in the right panel, shows warmer eastern Pacific SSTs in the monsoon run, certainly in the post-month 18 period (i.e., beyond June of the El Niño year); during El Niño’s mature phase in December (month 24), the peak SSTs are warmer in the monsoon composite by ∼0.4 K (or ∼25%), consistent with the above noted correspondence between the percentile range and peak JMA indices in the two runs. The composited El Niño in the monsoon run is not only stronger but also somewhat longer lived due to the presence of warm SST anomalies well into the “post mature phase” summer season in the control case (left panel), which generates another round of positive feedback from the ENSO-related Asian sector heating anomalies.

The warmer El Niño SSTs in the 23d–27th percentile range of the monsoon run’s distribution suggest only that the inclusion of ENSO-related Asian summer monsoon heating in the CZ model enhances the relative population at higher peak JMA index values. Does this imply strengthening of El Niño events in the monsoon run? Not necessarily, unless there is evidence of both a relative population increase at higher index values and a decrease in relative population at lower index values, that is, of a population shift. Although the distributions shown in Fig. 8 are unnormalized, they are strongly indicative of a population shift in the monsoon run and hence of strengthening of “weak” El Niños due to the feedback of ENSO-related Asian summer monsoon heating.

Examination of the composited (23d–27th percentile) evolution of midtropospheric heating, surface zonal wind, and upper-ocean-layer depth differences in Fig. 10 provide insight into the dynamics of the Asian sector heating induced feedback on CZ model’s ENSO evolution. The composited difference in the equatorial heating evolution (left panel) graphically illustrates the duration and magnitude of the parameterized Asian sector heating anomalies; the El Niños in this percentile range generate strongest negative heating anomalies during late summer of the El Niño year, with peak amplitudes of about 1.5 K day−1 over the eastern equatorial Indian Ocean and 3.0 K day−1 over the Maritime Continent. Negative heating anomalies are evident over the Asian longitudes also in the following summer as the El Niño SSTs in the control and, particularly, the monsoon run have not dissipated altogether by then, as noted earlier. Interestingly, parameterized heating anomalies of the opposite sign (i.e., positive) are present in the summer preceding the El Niño year, consistent with the cold Pacific SST anomalies during that summer (cf. Fig. 9). The composited deep heating anomalies eastward of the date line, on the other hand, are strongly linked to the local SST evolution (cf. Fig. 9c), through the CZ model’s heating parameterization, which emphasizes the evaporative contribution.

The difference in zonal wind evolution between the control and monsoon run El Niños in the 23d–27th percentile range is shown in the second panel of Fig. 10. The surface westerlies over the western equatorial Pacific (west of the date line) are temporally in sync with the Asian sector heating anomalies as they are generated in the coupled model from a Gill-type steady state atmospheric model, which produces surface westerlies to the east of the deep cooling region, much as in Fig. 10;the westerly anomalies in the central/eastern Pacific, on the other hand, are forced by the local deep heating anomalies and are thus positioned westward of the heating field. The Pacific basin SST anomalies thus control the surface winds everywhere, given that the parameterization of ENSO-related Asian sector heating anomalies is also dependent on their structure (cf. section 4c). Although the spatio–temporal structure of zonal wind anomalies in the central Pacific sector may give the impression of westward propagation and intensification, no such propagation, in fact, occurs, for reasons discussed next.

The composited difference in the evolution of upper-layer depth10 between the control and monsoon runs is shown in the right panel of Fig. 10. Although this is the field of choice to analyze the propagation of oceanic Kelvin and Rossby waves, the compositing of a large number of cases is not very helpful in this regard. A correspondence between the upper-layer depth and SST (Fig. 9c) variations is, of course, evident in the eastern equatorial Pacific, where SSTs are largely controlled by the thermocline depth variations, with deepening leading to warmer SSTs. In the western Pacific, on the other hand, the upper-layer depth changes are larger but do not produce SST variations, as expected. The western Pacific depth variations cannot be uniquely associated with the local wind variations (Fig. 10b) either, as they are produced by both contemporaneous surface zonal wind anomalies over the western basin and oceanic wave propagation initiated in the past. In the CZ model, the Kelvin wave takes about two months to traverse the Pacific, while the Rossby wave takes at least three times longer to cross back, which together with the boundary reflection yields a circuit travel time of ∼1 yr; the related oceanic variability and the summertime forcing of the coupled system by ENSO-related Asian sector heating anomalies together create the impression of westward propagation of the surface zonal wind anomalies in Fig. 10b.

c. Impact on CZ model’s SST spectrum and phase locking

In contrast with the previous sections, which focused exclusively on the modulation of CZ model’s ENSO amplitude, evolution, and average frequency, this section examines the impact of ENSO-related Asian summer monsoon heating on preferred periodicities of the CZ model generated interannual variability and its phase locking with the seasonal cycle. The power spectral density (or spectrum) of the raw JMA monthly SST index (i.e., without the 5-month running average) was estimated for both the control and monsoon runs as a function of period (τ); the estimation method is described in the appendix. The conventional frequency spectrum I(ν) is related to the period spectrum I′(τ) as follows:
i1520-0442-12-9-2787-eq1
with the unnormalized spectrum having units of K2 month−1.

A comparison of the unnormalized spectra of the control and monsoon runs (Fig. 11) indicates reduced power in the 45–50-month period range in the monsoon run; the power reduction at the center of the 2–7-yr ENSO period range is, however, compensated by modest increases in power in the wings, particularly, in the 30–45-month period range. This broadening of the spectrum is, however, not due to the breakdown of the seasonal phase locking of ENSO evolution in the monsoon integration, as evident from Fig. 12, which compares ENSO peak-phase timings in the CZ control and monsoon runs. The onset, peak, and termination of observed ENSO events are known to occur during specific phases of the seasonal cycle (Rasmusson and Carpenter 1982;An 1996), and the seasonal phase locking is evident in the CZ control case as well (Fig. 12) but with the La Niña peak phase occuring during August–September instead of boreal winter. Figure 12 shows that the inclusion of interactive Asian summer monsoon has little impact on the seasonal phase locking aspect of ENSO evolution in the CZ model.

In summary, the dynamic interaction of the ENSO-related Asian summer monsoon heating with the CZ model’s ocean–atmosphere system results in a wider period distribution of ENSO variability, but with the El Niño peak phase remaining seasonally locked with boreal winter months. Our finding of a broadened spectrum in the monsoon run appears to be in general agreement with Wainer and Webster’s (1996) finding that inclusion of Asian summer monsoon interannual variability (from whatever causes) in their coupled model led to considerably more aperiodicity of ENSO-like oscillations.

It is worth noting that the monsoon run’s spectrum shows diminished power also at the biennial (24 month) period, for reasons that are unclear at the present time.

d. Sensitivity to Asian sector heating amplitude

The sensitivity of the ENSO peak-amplitude distribution to the scaling of ENSO-related Asian summer monsoon deep heating anomalies is assessed in this section by changing the scaling factor from 3 to 2 and 4; the new distributions of peak JMA index for warm events is shown in Fig. 13. As discussed earlier in section 4, the empirically parameterized Asian sector deep heating needs to be scaled as the CZ model’s parameterization produces ENSO-related Pacific deep heating anomalies that are much stronger than those diagnosed from observations (Figs. 2 and 3); the CZ model needs the stronger heating anomalies in order to generate reasonable surface wind amplitudes in the presence of strong Rayleigh dissipation.

The 95 000-yr monsoon runs were repeated using a scaling factor of 2 and 4, and the obtained distributions of the peak JMA index are shown in Fig. 13 using the‘+’ and ‘○’ symbols, respectively. It is apparent from this figure that the strengthening of the ENSO-related Asian sector heating anomalies leads to greater impact on ENSO amplitude and average frequency; note that scaling factors of 2 and 4 yield 42 026 and 44 218 ENSO events, respectively, and that larger scaling leads to a bigger shift in ENSO population toward higher-peak JMA SST, for example, the scaling factor of 4 results in a pronounced decrease in population in the 0.8–1.6 K index range and a substantial increase in population in the 2.2–3.8 K range.

6. Summary and concluding remarks

The interaction of the two largest climate systems on earth—the Asian summer monsoon and the El Niño–Southern Oscillation—is modeled using the intermediate-level Cane–Zebiak ocean–atmosphere anomaly model, to determine the impact of the Asian summer monsoon’s feedback on the model’s ENSO characteristics (amplitude, frequency, evolution, etc.). In this study, the ENSO’s impact on the Asian monsoon is not dynamically modeled but is obtained from a parameterization based on the rotated principal component analysis of combined variability of the Pacific SST, diagnosed diabatic heating (at 400 mb), and the 1000-mb wind anomalies during the 1979–97 summer months. Although inclusion of orography and a simple land surface model (minimally needed to dynamically simulate Asian monsoon variability) in the CZ coupled model would have been preferable, the empirical parameterization strategy was used in order to realistically represent the ENSO-related monsoon variations in an intermediate model context.

Analysis of the 95 000-yr long CZ model integrations conducted both with and without the interactive Asian summer monsoon heating anomalies reveals the following.

  • The influence of the June–August ENSO-related Asian heating anomalies on the Pacific surface winds increases ENSO occurrence in the CZ model, producing an extra ENSO event every 20 yr or so.

  • The ENSO occurrence is enhanced in the monsoon run primarily in the 2.2–3.6 K JMA index range in case of warm events, and the −1.0 to −1.6 K range in case of cold events, with the enhancements being statistically significant at the 95% confidence level. The modest reduction in the number of warm events in the 0.6–1.2 K range in the monsoon run indicates that the population increase in the 2.2–3.6 K index range results, in part, from a population shift, due to the strengthening of weak El Niños in the monsoon run. Interestingly, the inclusion of ENSO-related Asian sector heating in the CZ coupled model results in a diminution of the bimodality evident in the control case distributions.

  • The interactive ENSO-related Asian heating impacts SST evolution of weak El Niños (e.g., in the 23d–27th percentile range) primarily in the eastern Pacific longitudes (east of 140°W) and in the period beginning with the summer of the year in which the JMA SST index peaks.

  • The interaction of ENSO-related Asian summer monsoon heating with the CZ model’s ocean–atmosphere results in a wider period distribution of ENSO variability, for the monsoon run exhibits reduced spectral power near the center of the 2–7-yr ENSO period range (∼48 months) but more power in the wings, particularly, in the 30–45-month range; the broadening of the spectrum is, however, not due to the breakdown in the seasonal phase-locking behavior of ENSO events, as the monsoon run exhibited no significant changes in this property.

The above findings attest to the positive feedback between Asian summer monsoon and ENSO, which has been suggested by several previous empirical and diagnostic modeling studies. We have recently become aware of Kirtman and Shukla’s study (1999) that also examines the impact of the Indian summer monsoon–ENSO feedback on the Cane–Zebiak model ENSO. These authors determine the all-India summer monsoon rainfall anomaly from the CZ model’s NINO3 SST using empirical linkage and include the monsoon feedback on Pacific surface winds using regressions obtained from atmospheric GCM integrations forced by climatological SST. Their study corroborates the existence of positive feedback between the Asian monsoon and ENSO in the CZ model.

Given the significance of our findings, the feedback of the Australian monsoon and the year-round American monsoon on ENSO also need to be examined. If our findings are substantiated using other coupled models, it may be of interest to explore if the predictive skill of these models can be improved by the inclusion of the monsoon–ENSO feedback.

Acknowledgments

This work is part of the doctoral research being conducted by Chul Chung, whose graduate studies and research have been supported by NASA Graduate Student Fellowship in Global Change Research (Award NGT 30276). This research effort was also supported by NOAA Grant NA46GP0194 and NSF Grants ATM 9316278 and 9422507 to S. Nigam.

The authors would like to thank the editor, Neville Nicholls, and David Stephenson and another anonymous reviewer for a careful reading of the manuscript and constructive suggestions, Steve Zebiak of LDEO for advice on aspects of the CZ coupled model, and Zhengxi Zhou of NOAA/NODC for help in the local installation of this model. They also thank Ilana Stern of the Data Support Group at NCAR for help in accessing the ECMWF operational analysis and reanalysis datasets archived at NCAR, and the Scientific Computing Division at NCAR for providing computational resources.

Preliminary findings of study were presented at the 21st Annual Climate Diagnostic and Prediction Workshop held in Huntsville, Alabama [28 October–1 November 1996; see pp. 288–291 of the conference proceedings (PB97-159164)], at the American Meteorological Society’s Seventh Conference on Climate Variations held in Long Beach, California (2–7 February 1997), and at the 22d European Geophysical Society’s Symposium on Climate Variability held in Vienna, Austria (21–25 April 1997).

REFERENCES

  • An, S., 1996: A critical examination of the current theories on the ENSO dynamics using an intermediate coupled ocean–atmosphere model. Ph.D. dissertation, Seoul National University, 171 pp. [Available from Department of Atmospheric Sciences, Seoul National University, 151-742, Korea.].

  • Barnett, T. P., M. Latif, N. E. Graham, M. Flügel, S. Pazan, and W. White, 1993: ENSO and ENSO-related predictability. Part I: Prediction of equatorial Pacific sea surface temperature with a hybrid coupled ocean–atmosphere model. J. Climate,6, 1545–1566.

  • Bhalme, H. N., A. B. Sikder, and S. K. Jadhav, 1990: Coupling between the El Niño and planetary-scale waves and their linkage with the Indian monsoon rainfall. Meteor. Atmos. Phys.,44, 293–305.

  • Bjerknes, J., 1969: Atmospheric teleconnections from the equatorial pacific. Mon. Wea. Rev.,97, 163–172.

  • Dickson, R. R., 1984: Eurasian snow cover versus Indian monsoon rainfall—An extension of the Hahn-Shukla results. J. Climate Appl. Meteor.,23, 171–173.

  • Gill, A. E., 1980: Some simple solutions for heat-induced tropical circulations. Quart. J. Roy. Meteor. Soc.,106, 447–462.

  • ——, and E. M. Rasmusson, 1983: The 1982–83 climate anomaly in the equatorial Pacific. Nature,305, 229–234.

  • Hartmann, D. L., H. H. Hendon, and R. A. Houze Jr., 1984: Some implications of the mesoscale circulations in tropical could clusters for large-scale dynamics and climate. J. Atmos. Sci.,41, 113–121.

  • Hastenrath, S., 1990: Tropical climate prediction: A progress report, 1985–1990. Bull. Amer. Meteor. Soc.,71, 819–825.

  • Holton, J. R., 1992: An Introduction to Dynamic Meteorology. Academic Press, 510 pp.

  • Horel, J. D., 1981: A rotated principal component analysis of the interannual variability of the Northern Hemisphere 500 mb height field. Mon. Wea. Rev.,109, 2080–2092.

  • ——, and J. M. Wallace, 1981: Planetary-scale atmospheric phenomena associated with the Southern Oscillation. Mon. Wea. Rev.,109, 813–829.

  • Hoskins, B. J., H. H. Hsu, I. N. James, M. Masutani, P. D. Sardeshmukh, and G. H. White, 1989: Diagnostics of the Global Atmospheric Circulation based on ECMWF analyses 1979–1989. WCRP-27, WMO/TD 326, 217 pp.

  • Ju, J., and J. Slingo, 1995: The Asian summer monsoon and ENSO. Quart. J. Roy. Meteor. Soc.,121, 1133–1168.

  • Kirtman, B. P., and J. Shukla, 1999: Influence of the Indian Summer Monsoon on ENSO. Quart. J. Roy. Meteor. Soc., in press.

  • Kousky, V. E., Ed., 1998: Climate Diagnostics Bulletin. NOAA/NWS/NCEP/CPC, U.S. Dept. of Commerce, 81 pp.

  • Lau, K.-M., and S. Yang, 1996: The Asian monsoon and predictability of the tropical ocean–atmosphere system. Quart. J. Roy. Meteor. Soc.,122, 945–957.

  • Lindzen, R. S., and S. Nigam, 1987: On the role of sea surface temperature gradients in forcing low-level winds and convergence in the Tropics. J. Atmos. Sci.,44, 2418–2436.

  • Mantua, N. J., and D. S. Battisti, 1995: Aperiodic variability in the Zebiak–Cane coupled ocean–atmosphere model: Air–sea interactions in the western equatorial Pacific. J. Climate,8, 2897–2927.

  • Meehl, G. A., 1987: The annual cycle and interannual variability in the tropical Pacific and Indian Ocean regions. Mon. Wea. Rev.,115, 27–50.

  • ——, 1994: Coupled land–ocean–atmosphere processes and South Asian monsoon variability. Science,265, 263–267.

  • Nigam, S., 1994: On the dynamical basis for the Asian summer monsoon rainfall–El Niño relationship. J. Climate,7, 1750–1771.

  • ——, 1997: The annual warm to cold phase transition in the eastern equatorial Pacific: Diagnosis of the role of stratus cloud-top cooling. J. Climate,10, 2447–2467.

  • ——, and H.-S. Shen, 1993: Structure of oceanic and atmospheric low-frequency variability over the tropical Pacific and Indian oceans. Part I: COADS observations. J. Climate,6, 657–676.

  • ——, and Y. Chao, 1996: Evolution-dynamics of tropical ocean–atmosphere annual-cycle variability. J. Climate,9, 3187–3205.

  • Normand, C., 1953: Monsoon seasonal forecasting. Quart. J. Roy. Meteor. Soc.,79, 463–473.

  • O’Brien, J. J., 1970: Alternative solutions to the classical vertical velocity problem. J. Appl. Meteor.,9, 197–203.

  • ——, T. S. Richards, and A. C. Davis, 1996: Climatology and multinational sporting events, Atlanta 1996. Bull. Amer. Meteor. Soc.,77, 771–774.

  • Palmer, T. N., C. Brankovic, P. Viterbo, and M. J. Miller, 1992: Modeling interannual variations of summer monsoons. J. Climate,5, 399–417.

  • Philander, S. G. H., 1985: El Niño and La Niña. J. Atmos. Sci.,42, 2652–2662.

  • Ramage, C. S., 1971: Monsoon Meteorology. International Geophysics Series, Vol. 15, Academic Press, 296 pp.

  • Rasmusson, E. M., and T. H. Carpenter, 1982: Variations in tropical sea surface temperature and surface wind fields associated with the Southern Oscillation/El Niño. Mon. Wea. Rev.,110, 354–384.

  • ——, and ——, 1983: The relationship between eastern equatorial Pacific sea surface temperatures and summer monsoon rainfall over India and Sri Lanka. Mon. Wea. Rev.,111, 517–528.

  • ——, and J. M. Wallace, 1983: Meteorological aspects of the El Niño/Southern Oscillation. Science,222, 1195–1202.

  • Reed, R. J., and E. E. Recker, 1971: Structure and properties of synoptic-scale wave disturbances in the equatorial western Pacific. J. Atmos. Sci.,28, 1117–1133.

  • Reynolds, R. W., 1988: A real-time global sea surface temperature analysis. J. Climate,1, 75–86.

  • ——, and T. M. Smith, 1994: Improved global sea surface temperature analyses using optimum interpolation. J. Climate,7, 929–948.

  • Ropelewski, C. F., and M. S. Halpert, 1987: Global and regional-scale precipitation associated with El Niño/Southern Oscillation. Mon. Wea. Rev.,115, 1606–1626.

  • Syu, H.-H., J. D. Neelin, and D. Gutzler, 1995: Seasonal and interannual variability in a hybrid couled GCM. J. Climate,8, 2121–2143.

  • Troup, A. J., 1965: The Southern Oscillation. Quart. J. Roy. Meteor. Soc.,91, 490–506.

  • Verma, R. K., 1992: ENSO—Monsoon linkages as evidenced from Pacific SST correlations with monsoon precipitation. TOGA Notes,6, 1–3.

  • Vernekar, A. D., J. Zhou, and J. Shukla, 1995. The effect of Eurasian snow cover on the Indian monsoon. J. Climate,8, 248–266.

  • Wainer, I., and P. J. Webster, 1996: Monsoon/El Niño–Southern Oscillation relationships in a simple coupled ocean–atmosphere model. J. Geophys. Res.,101 (C11), 25 599–25 614.

  • Walker, G. T., 1923: Correlation in seasonal variations of weather. III: A preliminary study of world weather. Memo. Indian Meteor. Dept.,24, 75–131.

  • ——, 1924: Correlation in seasonal variations of weather. IV: A further study of world weather. Memo. Indian Meteor. Dept.,24, 275–332.

  • Webster, P. J., and S. Yang, 1992: Monsoon and ENSO: Selectively interactive systems. Quart. J. Roy. Meteor. Soc.,118, 877–900.

  • ——, J. P. Loschnigg, A. M. Moore, and R. R. Leben, 1999: The Great Indian Ocean Warming of 1997–1998: Evidence of coupled oceanic-atmospheric instabilities. Nature, in press.

  • 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.

  • Yamagata, T., and Y. Masumoto, 1989: A simple ocean–atmosphere coupled model for the origin of a warm El Niño Southern Oscillation event. Philos. Trans. Roy. Soc. London,A329, 225–236.

  • Yasunari, T., 1991: The Monsoon Year—A new concept of the Climate Year in the Tropics. Bull. Amer. Meteor. Soc.,72, 1331–1338.

  • Zebiak, S. E., and M. A. Cane, 1987: A model El Niño–Southern Oscillation. Mon. Wea. Rev.,115, 2262–2278.

  • Zhang, Y., J. M. Wallace, and D. S. Battisti, 1997: ENSO-like interdecadal variability: 1900–93. J. Climate,10, 1004–1020.

APPENDIX

Estimation of Power Spectral Density

The finite sample-size and discrete sampling introduce errors in estimation of continuous power spectral density; the aliasing error due to monthly sampling of the CZ model integration should be negligible because of the long-wave approximation in ocean dynamics. The leakage error was reduced by smoothing the FFT-based spectrum with the Modified Bartlett spectral window whose bandwidth was close to that of the side lobes. The time series was padded with 10 000 yr of data and 30 000 yr of zeroes. (The IMSL subroutine SSWD was used to calculate the spectrum.) The obtained spectrum will deviate from the true spectrum also due to the remaining statistical variability of the PSD (likely associated with unbiased errors), and the aperiodicity of the model output. In view of this, the spectrum computation was repeated 100 times over partially overlapping 10 000-yr subperiods of the 95 000-yr run; the obtained spectra were then averaged to lessen the unbiased part of the error.

Fig. 1.
Fig. 1.

The original and expanded heating domains in the CZ coupled model. The unshaded rectangle is the original model domain in which SST and atmospheric heating anomalies are computed over the Pacific basin, while the shaded area to the left represents the region in which Asian summer monsoon heating anomalies are additionally specified in the monsoon runs.

Citation: Journal of Climate 12, 9; 10.1175/1520-0442(1999)012<2787:ASMEFO>2.0.CO;2

Fig. 2.
Fig. 2.

The precipitation, heating, and SST anomalies during Jul 1997, an El Niño summer month, as evident from the SST panel: (a) Xie–Arkin precipitation anomalies based upon the 1979–97 climatology; (b) 400-mb diabatic heating anomaly w.r.t. the 1985–97 climatology, diagnosed residually from the ECMWF operational analyses; (c) Pacific deep heating, generated from the original CZ model parameterization driven by the July 1997 SST anomalies; and (d) CAC SST anomalies w.r.t. the 1979–97 climatology. The contour interval and shading threshold is 1.0 mm day−1 for precipitation, 0.5 K day−1 for heating, and 0.3 K for SST, and the zero contour is thickened in all panels.

Citation: Journal of Climate 12, 9; 10.1175/1520-0442(1999)012<2787:ASMEFO>2.0.CO;2

Fig. 3.
Fig. 3.

As in Fig. 2 except for Jul 1988, a La Niña summer month. The heating anomaly in this month is, however, diagnosed from the ECMWF reanalyses and w.r.t. the 1979–93 climatology.

Citation: Journal of Climate 12, 9; 10.1175/1520-0442(1999)012<2787:ASMEFO>2.0.CO;2

Fig. 4.
Fig. 4.

The leading mode of combined interannual variability of tropical Pacific SST, and the 1000-mb zonal and meridional winds and 400-mb diabatic heating (Q400) in the near-global Tropics during the 1979–97 summer months of Jun, Jul, and Aug:(a) Q400, (b) SST, (c) V1000, and (d) the common principal component. The contour interval and shading threshold is 0.1 K day−1 for heating and 0.1 K for SST. This leading rotated mode is associated with ENSO variability and explains about 20% of the combined domain averaged variance.

Citation: Journal of Climate 12, 9; 10.1175/1520-0442(1999)012<2787:ASMEFO>2.0.CO;2

Fig. 5.
Fig. 5.

The surface zonal wind response of the CZ atmospheric model when forced by the ENSO related deep heating anomalies (Fig. 4a) in (a) the Asian sector (shaded region in Fig. 1) and (b) the original CZ model domain (rectangular box in Fig. 1); note that heating is smoothed and scaled prior to use as model forcing, as described in footnote 5. The bottom panel (c) displays the zonal wind component of Fig. 4c using contours to facilitate comparison with the above panels. The contour interval and shading threshold is 0.1 m s−1 in all panels, and the zero contour is thickened as before.

Citation: Journal of Climate 12, 9; 10.1175/1520-0442(1999)012<2787:ASMEFO>2.0.CO;2

Fig. 6.
Fig. 6.

The second leading mode of combined interannual variability of tropical Pacific SST, V1000, and Q400 during the 1979–97 summer months of Jun, Jul, and Aug. This mode is linked to interannual SST variability in the central equatorial Pacific (NINO3.4 region) and explains about 10% of the combined variance. The rest as in Fig. 4.

Citation: Journal of Climate 12, 9; 10.1175/1520-0442(1999)012<2787:ASMEFO>2.0.CO;2

Fig. 7.
Fig. 7.

The parameterized Q400 anomalies in (a) Jul 1997 and (b) Jul 1988, based on the two-leading rotated modes of combined variability (Figs. 4 and 6); the projection method is described in section 4c. The contour interval and shading threshold is 0.5 K day−1, as in Figs. 2b and 3b, with which this figure is compared.

Citation: Journal of Climate 12, 9; 10.1175/1520-0442(1999)012<2787:ASMEFO>2.0.CO;2

Fig. 8.
Fig. 8.

Distribution of the number of ENSO events w.r.t. their peak JMA index in the 95 000-yr control (bar graph) and monsoon (solid dots connected by line) runs; there are 38 205 ENSO events in the control run and 42 655 events in the monsoon run. Each error bar represents the standard error of the corresponding interval mean at the 95% confidence level; see section 5a for additional details on error estimation.

Citation: Journal of Climate 12, 9; 10.1175/1520-0442(1999)012<2787:ASMEFO>2.0.CO;2

Fig. 9.
Fig. 9.

El Niño SST evolution in the 4°S–4°N equatorial band, based on compositing in the 23d–27th percentile range of the peak JMA index: (a) control run, (b) monsoon run, and (c) monsoon–control. The composite is based on 781 events (out of 19 571 warm events) with peak JMA SST index in the 1.41–1.57 K range in the control case, and 861 events (out of 21 501 warm events) with JMA index in the 1.71–1.89 K range in the monsoon case. Note that peak El Niño SST amplitudes are attained close to month 24, that is, at the end of year 2; the month markings are at the middle of the months. The contour interval and shading threshold is 0.1 K in all three panels, and the zero contour is thickened as before.

Citation: Journal of Climate 12, 9; 10.1175/1520-0442(1999)012<2787:ASMEFO>2.0.CO;2

Fig. 10.
Fig. 10.

The difference between the CZ model’s monsoon and control case El Niño evolutions in the equatorial band, based on compositing in the 23d–27th percentile range of the peak JMA index: (a) deep heating in the Asian and Pacific longitudes, (b) surface zonal wind in the Pacific sector, and (c) ocean upper-layer depth. The contour interval and shading threshold is 0.2 K day−1 for heating, 0.1 m s−1 for surface zonal wind, and 1.0 m for the upper-layer depth differences. Zero contour is omitted in the left panel.

Citation: Journal of Climate 12, 9; 10.1175/1520-0442(1999)012<2787:ASMEFO>2.0.CO;2

Fig. 11.
Fig. 11.

Power spectral density (K2 month−1) as a function of period (in months) for both the control (thin line) and monsoon (thick line) runs; see the appendix for details on the estimation method.

Citation: Journal of Climate 12, 9; 10.1175/1520-0442(1999)012<2787:ASMEFO>2.0.CO;2

Fig. 12.
Fig. 12.

The seasonal timing of ENSO peak phase in the CZ model’s control and monsoon runs; the y axis represents occurrence as percentage of the total number of warm (or cold) events in each run.

Citation: Journal of Climate 12, 9; 10.1175/1520-0442(1999)012<2787:ASMEFO>2.0.CO;2

Fig. 13.
Fig. 13.

Sensitivity of the El Niño strength distribution to the scaling of the Asian sector deep heating anomalies. The distribution obtained using a scaling factor of 4 in a 95 000-yr monsoon run is shown by the thin solid line connecting open circles (“○”), while that obtained with a scaling of 2 is shown by the dotted line connecting plus signs (“+”). For reference, the standard monsoon distribution, obtained with a scaling factor of 3, is shown by the thick solid line connecting solid dots, and the control run (with no Asian sector heating) is shown using vertical bars. Note that the standard error of each interval mean at the 95% confidence level is shown only for the two new runs.

Citation: Journal of Climate 12, 9; 10.1175/1520-0442(1999)012<2787:ASMEFO>2.0.CO;2

1

Ramage (1971), for example, defines the monsoonal region to include the oceanic and land surface regions between 20°S–40°N and 20°W–170°E (see Fig. 1a in Webster and Yang 1992).

2

The initialization procedure, in essence, modifies the objective analysis of the mass and wind fields to minimize the generation of gravity wave noise in weather forecasts. The “uninitialized” qualifier indicates that the circulation data has been objectively analyzed, that is, checked for accuracy and interpolated to a regular grid, but not otherwise modified.

3

The evaluation was performed using a steady global primitive equation model having high horizontal and vertical resolution: Δθ = 2.5°, zonal Fourier truncation at wavenumber 15 or 30, and 18 vertical sigma levels.

4

The near-normal rainfall over India during July 1997 has sometimes been interpreted as being indicative of the breakdown of the monsoon–ENSO linkage. It is worth noting, however, that while the Indian monsoon rainfall was near normal, the Asian monsoon as a whole was not, for Southeast Asia and southern–eastern China had excessive rainfall while the eastern equatorial Indian ocean had deficient rainfall in 1997 (Fig. 2a). Thus, the Asian monsoon—ENSO linkage is seemingly intact on the large scale, even if it is fractured on the regional scale (e.g., over India); the Indian monsoon–ENSO linkage is expected to be less robust, given Troup’s (1965) analysis, which showed the Indian subcontinent to be near the node (or zero line) of the Southern Oscillation surface pressure distribution during the northern summer season.

5

Prior to use as model forcing, the 400-mb heating anomaly (Fig. 4a) is smoothed using a 1–2–1 filter in the zonal and meridional directions, and scaled by a factor of 3. Such scaling is needed in the CZ atmospheric model in order to elicit a reasonable surface wind amplitude in the presence of large Rayleigh dissipation.

6

A comparison of the sum of the two model responses (Figs. 5a, b) with the ENSO covariant zonal winds (Figs. 4c or 5c), on the other hand, allows an evaluation of the Gill-type atmospheric model as the model’s forcing in this case (the deep-heating anomalies in Fig. 4a) can be considered to be “perfect,” that is, dynamically consistent with the model’s target (the 1000-mb wind anomalies in Figs. 4c or 5c), for they are both obtained from the same combined variability analysis. The comparison shows that the Gill-model response is quite reasonable in the western equatorial Pacific and the off-equatorial northern Tropics, but rather deficient in the eastern equatorial Pacific and the southern Tropics, indicating that a high-quality parameterization of the Pacific deep heating anomalies alone will not suffice in improving the surface wind simulation of the CZ model—an objective not pursued in this study.

7

The interdecadal variability in the Pacific is characterized by comparable SST amplitudes in both the tropical and northern extratropical latitudes (e.g., Zhang et al. 1997).

8

The JMA and NINO3 SST indices differ only in the extent of latitudinal averaging, 5°S–5°N in the latter case; however, the former index is used in this study as the CZ model SSTs were stored at only four meridional grid points (3°S, 1°S, 1°N, and 3°N) in the 95 000-yr runs.

9

This estimate is obtained by analyzing the ENSO occurrence variability in nonoverlapping subsamples of various lengths, ranging from 1000 to 5000 yr.

10

The upper-layer depth is defined as the sum of the surface and subsurface layer depths; in the CZ ocean model, the suface layer has a fixed depth of 50 m whereas the subsurface layer has an average depth of 100 m.

Save
  • An, S., 1996: A critical examination of the current theories on the ENSO dynamics using an intermediate coupled ocean–atmosphere model. Ph.D. dissertation, Seoul National University, 171 pp. [Available from Department of Atmospheric Sciences, Seoul National University, 151-742, Korea.].

  • Barnett, T. P., M. Latif, N. E. Graham, M. Flügel, S. Pazan, and W. White, 1993: ENSO and ENSO-related predictability. Part I: Prediction of equatorial Pacific sea surface temperature with a hybrid coupled ocean–atmosphere model. J. Climate,6, 1545–1566.

  • Bhalme, H. N., A. B. Sikder, and S. K. Jadhav, 1990: Coupling between the El Niño and planetary-scale waves and their linkage with the Indian monsoon rainfall. Meteor. Atmos. Phys.,44, 293–305.

  • Bjerknes, J., 1969: Atmospheric teleconnections from the equatorial pacific. Mon. Wea. Rev.,97, 163–172.

  • Dickson, R. R., 1984: Eurasian snow cover versus Indian monsoon rainfall—An extension of the Hahn-Shukla results. J. Climate Appl. Meteor.,23, 171–173.

  • Gill, A. E., 1980: Some simple solutions for heat-induced tropical circulations. Quart. J. Roy. Meteor. Soc.,106, 447–462.

  • ——, and E. M. Rasmusson, 1983: The 1982–83 climate anomaly in the equatorial Pacific. Nature,305, 229–234.

  • Hartmann, D. L., H. H. Hendon, and R. A. Houze Jr., 1984: Some implications of the mesoscale circulations in tropical could clusters for large-scale dynamics and climate. J. Atmos. Sci.,41, 113–121.

  • Hastenrath, S., 1990: Tropical climate prediction: A progress report, 1985–1990. Bull. Amer. Meteor. Soc.,71, 819–825.

  • Holton, J. R., 1992: An Introduction to Dynamic Meteorology. Academic Press, 510 pp.

  • Horel, J. D., 1981: A rotated principal component analysis of the interannual variability of the Northern Hemisphere 500 mb height field. Mon. Wea. Rev.,109, 2080–2092.

  • ——, and J. M. Wallace, 1981: Planetary-scale atmospheric phenomena associated with the Southern Oscillation. Mon. Wea. Rev.,109, 813–829.

  • Hoskins, B. J., H. H. Hsu, I. N. James, M. Masutani, P. D. Sardeshmukh, and G. H. White, 1989: Diagnostics of the Global Atmospheric Circulation based on ECMWF analyses 1979–1989. WCRP-27, WMO/TD 326, 217 pp.

  • Ju, J., and J. Slingo, 1995: The Asian summer monsoon and ENSO. Quart. J. Roy. Meteor. Soc.,121, 1133–1168.

  • Kirtman, B. P., and J. Shukla, 1999: Influence of the Indian Summer Monsoon on ENSO. Quart. J. Roy. Meteor. Soc., in press.

  • Kousky, V. E., Ed., 1998: Climate Diagnostics Bulletin. NOAA/NWS/NCEP/CPC, U.S. Dept. of Commerce, 81 pp.

  • Lau, K.-M., and S. Yang, 1996: The Asian monsoon and predictability of the tropical ocean–atmosphere system. Quart. J. Roy. Meteor. Soc.,122, 945–957.

  • Lindzen, R. S., and S. Nigam, 1987: On the role of sea surface temperature gradients in forcing low-level winds and convergence in the Tropics. J. Atmos. Sci.,44, 2418–2436.

  • Mantua, N. J., and D. S. Battisti, 1995: Aperiodic variability in the Zebiak–Cane coupled ocean–atmosphere model: Air–sea interactions in the western equatorial Pacific. J. Climate,8, 2897–2927.

  • Meehl, G. A., 1987: The annual cycle and interannual variability in the tropical Pacific and Indian Ocean regions. Mon. Wea. Rev.,115, 27–50.

  • ——, 1994: Coupled land–ocean–atmosphere processes and South Asian monsoon variability. Science,265, 263–267.

  • Nigam, S., 1994: On the dynamical basis for the Asian summer monsoon rainfall–El Niño relationship. J. Climate,7, 1750–1771.

  • ——, 1997: The annual warm to cold phase transition in the eastern equatorial Pacific: Diagnosis of the role of stratus cloud-top cooling. J. Climate,10, 2447–2467.

  • ——, and H.-S. Shen, 1993: Structure of oceanic and atmospheric low-frequency variability over the tropical Pacific and Indian oceans. Part I: COADS observations. J. Climate,6, 657–676.

  • ——, and Y. Chao, 1996: Evolution-dynamics of tropical ocean–atmosphere annual-cycle variability. J. Climate,9, 3187–3205.

  • Normand, C., 1953: Monsoon seasonal forecasting. Quart. J. Roy. Meteor. Soc.,79, 463–473.

  • O’Brien, J. J., 1970: Alternative solutions to the classical vertical velocity problem. J. Appl. Meteor.,9, 197–203.

  • ——, T. S. Richards, and A. C. Davis, 1996: Climatology and multinational sporting events, Atlanta 1996. Bull. Amer. Meteor. Soc.,77, 771–774.

  • Palmer, T. N., C. Brankovic, P. Viterbo, and M. J. Miller, 1992: Modeling interannual variations of summer monsoons. J. Climate,5, 399–417.

  • Philander, S. G. H., 1985: El Niño and La Niña. J. Atmos. Sci.,42, 2652–2662.

  • Ramage, C. S., 1971: Monsoon Meteorology. International Geophysics Series, Vol. 15, Academic Press, 296 pp.

  • Rasmusson, E. M., and T. H. Carpenter, 1982: Variations in tropical sea surface temperature and surface wind fields associated with the Southern Oscillation/El Niño. Mon. Wea. Rev.,110, 354–384.

  • ——, and ——, 1983: The relationship between eastern equatorial Pacific sea surface temperatures and summer monsoon rainfall over India and Sri Lanka. Mon. Wea. Rev.,111, 517–528.

  • ——, and J. M. Wallace, 1983: Meteorological aspects of the El Niño/Southern Oscillation. Science,222, 1195–1202.

  • Reed, R. J., and E. E. Recker, 1971: Structure and properties of synoptic-scale wave disturbances in the equatorial western Pacific. J. Atmos. Sci.,28, 1117–1133.

  • Reynolds, R. W., 1988: A real-time global sea surface temperature analysis. J. Climate,1, 75–86.

  • ——, and T. M. Smith, 1994: Improved global sea surface temperature analyses using optimum interpolation. J. Climate,7, 929–948.

  • Ropelewski, C. F., and M. S. Halpert, 1987: Global and regional-scale precipitation associated with El Niño/Southern Oscillation. Mon. Wea. Rev.,115, 1606–1626.

  • Syu, H.-H., J. D. Neelin, and D. Gutzler, 1995: Seasonal and interannual variability in a hybrid couled GCM. J. Climate,8, 2121–2143.

  • Troup, A. J., 1965: The Southern Oscillation. Quart. J. Roy. Meteor. Soc.,91, 490–506.

  • Verma, R. K., 1992: ENSO—Monsoon linkages as evidenced from Pacific SST correlations with monsoon precipitation. TOGA Notes,6, 1–3.

  • Vernekar, A. D., J. Zhou, and J. Shukla, 1995. The effect of Eurasian snow cover on the Indian monsoon. J. Climate,8, 248–266.

  • Wainer, I., and P. J. Webster, 1996: Monsoon/El Niño–Southern Oscillation relationships in a simple coupled ocean–atmosphere model. J. Geophys. Res.,101 (C11), 25 599–25 614.

  • Walker, G. T., 1923: Correlation in seasonal variations of weather. III: A preliminary study of world weather. Memo. Indian Meteor. Dept.,24, 75–131.

  • ——, 1924: Correlation in seasonal variations of weather. IV: A further study of world weather. Memo. Indian Meteor. Dept.,24, 275–332.

  • Webster, P. J., and S. Yang, 1992: Monsoon and ENSO: Selectively interactive systems. Quart. J. Roy. Meteor. Soc.,118, 877–900.

  • ——, J. P. Loschnigg, A. M. Moore, and R. R. Leben, 1999: The Great Indian Ocean Warming of 1997–1998: Evidence of coupled oceanic-atmospheric instabilities. Nature, in press.

  • 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.

  • Yamagata, T., and Y. Masumoto, 1989: A simple ocean–atmosphere coupled model for the origin of a warm El Niño Southern Oscillation event. Philos. Trans. Roy. Soc. London,A329, 225–236.

  • Yasunari, T., 1991: The Monsoon Year—A new concept of the Climate Year in the Tropics. Bull. Amer. Meteor. Soc.,72, 1331–1338.

  • Zebiak, S. E., and M. A. Cane, 1987: A model El Niño–Southern Oscillation. Mon. Wea. Rev.,115, 2262–2278.

  • Zhang, Y., J. M. Wallace, and D. S. Battisti, 1997: ENSO-like interdecadal variability: 1900–93. J. Climate,10, 1004–1020.

  • Fig. 1.

    The original and expanded heating domains in the CZ coupled model. The unshaded rectangle is the original model domain in which SST and atmospheric heating anomalies are computed over the Pacific basin, while the shaded area to the left represents the region in which Asian summer monsoon heating anomalies are additionally specified in the monsoon runs.

  • Fig. 2.

    The precipitation, heating, and SST anomalies during Jul 1997, an El Niño summer month, as evident from the SST panel: (a) Xie–Arkin precipitation anomalies based upon the 1979–97 climatology; (b) 400-mb diabatic heating anomaly w.r.t. the 1985–97 climatology, diagnosed residually from the ECMWF operational analyses; (c) Pacific deep heating, generated from the original CZ model parameterization driven by the July 1997 SST anomalies; and (d) CAC SST anomalies w.r.t. the 1979–97 climatology. The contour interval and shading threshold is 1.0 mm day−1 for precipitation, 0.5 K day−1 for heating, and 0.3 K for SST, and the zero contour is thickened in all panels.

  • Fig. 3.

    As in Fig. 2 except for Jul 1988, a La Niña summer month. The heating anomaly in this month is, however, diagnosed from the ECMWF reanalyses and w.r.t. the 1979–93 climatology.

  • Fig. 4.

    The leading mode of combined interannual variability of tropical Pacific SST, and the 1000-mb zonal and meridional winds and 400-mb diabatic heating (Q400) in the near-global Tropics during the 1979–97 summer months of Jun, Jul, and Aug:(a) Q400, (b) SST, (c) V1000, and (d) the common principal component. The contour interval and shading threshold is 0.1 K day−1 for heating and 0.1 K for SST. This leading rotated mode is associated with ENSO variability and explains about 20% of the combined domain averaged variance.

  • Fig. 5.

    The surface zonal wind response of the CZ atmospheric model when forced by the ENSO related deep heating anomalies (Fig. 4a) in (a) the Asian sector (shaded region in Fig. 1) and (b) the original CZ model domain (rectangular box in Fig. 1); note that heating is smoothed and scaled prior to use as model forcing, as described in footnote 5. The bottom panel (c) displays the zonal wind component of Fig. 4c using contours to facilitate comparison with the above panels. The contour interval and shading threshold is 0.1 m s−1 in all panels, and the zero contour is thickened as before.

  • Fig. 6.

    The second leading mode of combined interannual variability of tropical Pacific SST, V1000, and Q400 during the 1979–97 summer months of Jun, Jul, and Aug. This mode is linked to interannual SST variability in the central equatorial Pacific (NINO3.4 region) and explains about 10% of the combined variance. The rest as in Fig. 4.

  • Fig. 7.

    The parameterized Q400 anomalies in (a) Jul 1997 and (b) Jul 1988, based on the two-leading rotated modes of combined variability (Figs. 4 and 6); the projection method is described in section 4c. The contour interval and shading threshold is 0.5 K day−1, as in Figs. 2b and 3b, with which this figure is compared.

  • Fig. 8.

    Distribution of the number of ENSO events w.r.t. their peak JMA index in the 95 000-yr control (bar graph) and monsoon (solid dots connected by line) runs; there are 38 205 ENSO events in the control run and 42 655 events in the monsoon run. Each error bar represents the standard error of the corresponding interval mean at the 95% confidence level; see section 5a for additional details on error estimation.

  • Fig. 9.

    El Niño SST evolution in the 4°S–4°N equatorial band, based on compositing in the 23d–27th percentile range of the peak JMA index: (a) control run, (b) monsoon run, and (c) monsoon–control. The composite is based on 781 events (out of 19 571 warm events) with peak JMA SST index in the 1.41–1.57 K range in the control case, and 861 events (out of 21 501 warm events) with JMA index in the 1.71–1.89 K range in the monsoon case. Note that peak El Niño SST amplitudes are attained close to month 24, that is, at the end of year 2; the month markings are at the middle of the months. The contour interval and shading threshold is 0.1 K in all three panels, and the zero contour is thickened as before.

  • Fig. 10.

    The difference between the CZ model’s monsoon and control case El Niño evolutions in the equatorial band, based on compositing in the 23d–27th percentile range of the peak JMA index: (a) deep heating in the Asian and Pacific longitudes, (b) surface zonal wind in the Pacific sector, and (c) ocean upper-layer depth. The contour interval and shading threshold is 0.2 K day−1 for heating, 0.1 m s−1 for surface zonal wind, and 1.0 m for the upper-layer depth differences. Zero contour is omitted in the left panel.

  • Fig. 11.

    Power spectral density (K2 month−1) as a function of period (in months) for both the control (thin line) and monsoon (thick line) runs; see the appendix for details on the estimation method.

  • Fig. 12.

    The seasonal timing of ENSO peak phase in the CZ model’s control and monsoon runs; the y axis represents occurrence as percentage of the total number of warm (or cold) events in each run.

  • Fig. 13.

    Sensitivity of the El Niño strength distribution to the scaling of the Asian sector deep heating anomalies. The distribution obtained using a scaling factor of 4 in a 95 000-yr monsoon run is shown by the thin solid line connecting open circles (“○”), while that obtained with a scaling of 2 is shown by the dotted line connecting plus signs (“+”). For reference, the standard monsoon distribution, obtained with a scaling factor of 3, is shown by the thick solid line connecting solid dots, and the control run (with no Asian sector heating) is shown using vertical bars. Note that the standard error of each interval mean at the 95% confidence level is shown only for the two new runs.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 509 158 2
PDF Downloads 138 26 2