Abstract

Diagnostics from observations and multicentury integrations of a coupled model [Geophysical Fluid Dynamics Laboratory (GFDL) coupled model version 2.1 (CM2.1)] indicate that about 65% of the severe monsoons (rainfall > 1.5 standard deviations of its long-term mean) over South Asia are associated with sea surface temperature (SST) anomalies over the equatorial Pacific during the developing phase of ENSO, and another 30% are associated with SST variations over the tropical Indo-Pacific warm pool. The present research aims to identify the moist processes that initiate the dryness (wetness) and provide a precursor for rainfall anomalies over South Asia in spring during El Niño (La Niña). The hypothesis in this paper, based on CM2.1 composites, is that at low levels El Niño–forced equatorial easterly wind anomalies over the Indian Ocean, resulting from Ekman pumping, promote anticyclonic vorticity over the northern Indian Ocean, whose poleward flank advects dry air from northern latitudes to South Asia. This is tested by performing ensemble simulations with the atmospheric component of CM2.1 (AM2.1) and applying moisture and moist static energy budgets.

During El Niño, AM2.1 solutions capture the anticyclonic vorticity formation over the northern Indian Ocean 20–25 days earlier than organized negative rainfall anomalies over South Asia, and the advection of climatological air of lower moisture content by these anomalous winds initiates the dryness over South Asia from April onward. This long lead time embodied in this precursor signal can be exploited for predicting severe monsoons. During ENSO neutral conditions, the amplitude of regional SST anomalies during spring is insufficient to produce such a precursor signal.

The dominance of the term warrants monitoring the three-dimensional moisture distribution for better understanding, modeling, and predicting of severe monsoons.

1. Introduction

a. Background

The boreal summer seasonal (June–September) rainfall associated with the Asian summer monsoon (ASM) is the main source of water for the agrarian societies of South and Southeast Asia, and therefore the monsoon rainfall dictates the socioeconomic conditions of the world’s densely populated regions. For example, if the monsoon rainfall is below 15% or more of its long-term normal, defined here as severe weak monsoons (Sikka 1999), it directly influences the agricultural yields, resulting in economic stress to millions of small-scale farmers and their families. A successful prediction of severe monsoons a few months ahead would therefore be vital to the economy and food and water securities of the developing nations influenced by the ASM. The focus here is to elucidate the moist processes that initiate and maintain the dryness (wetness) over South Asia during severe weak (strong) monsoons. A particular attention is paid to identify the precursor signal during late spring, and during the developing phase of El Niño (La Niña).

Consistent with the hypothesis embodied in Charney and Shukla (1981), many studies have suggested that the mean and interannual variability of the planetary-scale monsoon are influenced by slowly varying boundary conditions. Of them, the sea surface temperature (SST) anomalies associated with El Niño–Southern Oscillation (ENSO) are found to be the dominant forcing element. However, based on long-term observations and on a statistical averaged picture, certain measures of ENSO during boreal spring do not necessarily foreshadow the ensuing monsoon (e.g., Normand 1953; Webster and Yang 1992), although during summer the simultaneous correlation between monsoon rainfall variations and ENSO is high (e.g., Rasmusson and Carpenter 1983). This led Shukla and Paolino (1983) to indicate that the March–May tendency in ENSO measures provides predictive information.

Motivated by the observational evidence, sensitivity experiments with a suite of atmospheric general circulations models (AGCMs) to identify the mechanisms involved in the ENSO–monsoon association have been carried out. Specifically, the authors forced the AGCMs with aspects of idealized (e.g., Keshavamurty 1982) or observed (e.g., Palmer et al. 1992; Yang and Lau 1998) SST anomaly patterns and examined the changes in planetary- and regional-scale monsoon circulations. Ju and Slingo (1995) found that during spring the broad-scale monsoon circulation index proposed by Webster and Yang (1992) is sensitive to the phase of ENSO. Soman and Slingo (1997) showed that changes during spring are accounted for partly by an anomalous Walker circulation and partly by local SST anomalies. Goswami (1998) emphasized the need to understand the interaction between planetary- and regional-scale circulations. Lau and Nath (2000) stressed the role of ENSO-induced regional SST anomalies in influencing the monsoon circulation during September–October. Annamalai and Liu (2005) examined the sensitivity of the monsoon response to both El Niño intensity and regional SST anomalies and concluded that while El Niño impacts the monsoon during its onset and withdrawal phases, regional SST anomalies counteract its impact during the monsoon’s peak phase (July–August). It is generally agreed that in response to the development of warm SST anomalies over the east-central equatorial Pacific during El Niño, anomalous precipitation and the associated ascending branch of the Walker circulation move eastward with anomalous descent over South Asia, leading to reduction in monsoon rainfall. Sperber and Palmer (1996) assessed the association in many AGCMs and found that the models with realistic mean state tend to have a better ENSO–monsoon representation, consistent with Shukla (1984). For AGCMs to capture the ENSO–monsoon association, realistic simulations of mean monsoon precipitation and diabatic heating anomalies along the equatorial Pacific to imposed SST forcing are necessary ingredients.

The ENSO–monsoon relationship has been examined in coupled general circulation models (CGCMs; Meehl and Arblaster 1998; Wu and Kirtman 2004). Apart from a realistic simulation of mean monsoon precipitation, an additional challenge for CGCMs is to capture the tropical Pacific mean state as pointed out in the modeling study of Turner et al. (2005). Annamalai et al. (2007) analyzed the association in phase 3 of the Coupled Model Intercomparison Project (CMIP3; Meehl et al. 2007) models. They concluded that apart from mean monsoon precipitation, models that correctly simulate the timing and location of SST and diabatic heating anomalies over the equatorial Pacific, and the associated changes to Walker circulation, capture the relationship realistically. While past studies have explored the ENSO–monsoon linkages, to the best of our knowledge no study has focused on elucidating the moist processes responsible for initiating the dryness (or wetness) during severe weak (strong) monsoons over South Asia during the developing phase of ENSO. Meehl (1997) and Meehl and Arblaster (2002) examined the linkage through tropospheric biennial oscillation pathways and we defer that aspect to a future study.

Apart from ENSO, the importance of regional SST anomalies, either induced by ENSO or developed independently due to local air–sea interactions, for the monsoon variability has been studied. Shukla (1975) and Chandrasekar and Kitoh (1998) focused on the role of Arabian Sea and equatorial Indian Ocean SST anomalies, respectively, whereas Soman and Slingo (1997) emphasized the role of boreal spring tropical western Pacific SST anomalies in determining the large-scale monsoon circulation strength. Annamalai and Liu (2005) noted that the magnitudes of monsoon-related rainfall and circulation anomalies are influenced by regional SST anomalies. While ENSO is recognized as the dominant forcing of monsoon interannual variability, are there severe monsoon years influenced by regional SST anomalies during non-ENSO years, and what are the associated moist processes?

b. Present study

Figure 1 shows a scatter diagram between the summer all-India rainfall index (AIR) and Niño-3.4 (5°S–5°N, 120°–170°W) SST anomalies from observation (Fig. 1a) and Geophysical Fluid Dynamics Laboratory (GFDL) coupled model version 2.1 (CM2.1) twentieth-century integrations (20c3m; Fig. 1b). (Details on the data and models used are provided in section 2.) While the ENSO–monsoon relation is complex and nonlinear, both in observations and CM2.1 integrations, about 60%–75% of the severe monsoon years (i.e., AIR anomaly ≥ or ≤ 1.5 standard deviations) cluster with ENSO (highlighted in the top-left and bottom-right boxes in Figs. 1a,b). The composite temporal evolution of Niño-3.4 SST anomalies during ENSO years that occurred with severe monsoons is shown in Figs. 1c and 1d. The evolution covers three years: one year before [Year (−1)], the year of the developing phase [Year (0)], and the year of the decaying phase [Year (+1)] of El Niño or La Niña. While during the early stages in Year (−1) and later part in Year (+1) there are clear differences between observations and CM2.1 particularly for La Niña, beginning from boreal spring of Year (0) the growth of SST during El Niño as well as the decay during La Niña agree quite well. The exception is the “double peak” during El Niño in CM2.1 as already noted (e.g., AchutaRao and Sperber 2006; Joseph and Nigam 2006; Annamalai et al. 2007). The question of interest here is this: can we identify the moist processes leading to severe monsoons during spring of Year (0) (rectangular boxed region in Fig. 1c) when Niño-3.4 SST anomalies have already reached 1.0 standard deviation? The 1.0 standard deviation (~0.8°C) SST anomalies over the Niño-3.4 region are strong enough to perturb the equatorial Pacific precipitation anomalies and subsequent emanation of equatorial waves that could reach the monsoon region.

Fig. 1.

(a),(b) Scatter diagram between standardized anomalies of the all-India rainfall index (AIR; averaged over 7°–30°N, 65°–95°E) and Niño-3.4 SST (5°S–5°N, 120°–170°W) from (a) observations and (b) CM2.1. (c),(d) Temporal evolution of composite monthly Niño-3.4 standardized anomalies for severe weak monsoons and El Niño (solid line), and severe strong monsoons and La Niña (dashed lines), from (c) observations and (d) CM2.1. Year (0) refers to the developing phase; Year (−1) and Year (+1) represent 1 yr before and after the developing phase, respectively. The red boxes in top-left quarter of (a) and (b) represent the severe weak monsoon years occurring with El Niño and those in the bottom-right quarter indicate the severe strong monsoon years occurring with La Niña. The time period enclosed by the rectangles in (c) and (d) indicates the spring season of Year (0). The results are based on the period 1871–2008 for observations and 1861–2000 for CM2.1.

Fig. 1.

(a),(b) Scatter diagram between standardized anomalies of the all-India rainfall index (AIR; averaged over 7°–30°N, 65°–95°E) and Niño-3.4 SST (5°S–5°N, 120°–170°W) from (a) observations and (b) CM2.1. (c),(d) Temporal evolution of composite monthly Niño-3.4 standardized anomalies for severe weak monsoons and El Niño (solid line), and severe strong monsoons and La Niña (dashed lines), from (c) observations and (d) CM2.1. Year (0) refers to the developing phase; Year (−1) and Year (+1) represent 1 yr before and after the developing phase, respectively. The red boxes in top-left quarter of (a) and (b) represent the severe weak monsoon years occurring with El Niño and those in the bottom-right quarter indicate the severe strong monsoon years occurring with La Niña. The time period enclosed by the rectangles in (c) and (d) indicates the spring season of Year (0). The results are based on the period 1871–2008 for observations and 1861–2000 for CM2.1.

Because of a lack of sustained precipitation observations over the tropical oceans, we constructed April–May averaged composites of anomalous rainfall and 850-hPa streamlines from CM2.1 integrations (Fig. 2a). Note that years of severe weak monsoons that occurred with the developing phase of El Niño are only considered here. As expected, in response to warm SST anomalies rainfall is increased along the equatorial central Pacific, with a local maximum around the date line, and reduced over the Maritime Continent, in agreement with the results shown in Annamalai et al. (2007). A marginal reduction in rainfall (~1.0 mm day−1) is also noticeable over parts of South Asia extending into the South China Sea. In association with rainfall or diabatic heating anomalies, CM2.1 captures westerly wind anomalies over the entire equatorial Pacific whereas easterly wind anomalies cover the equatorial Atlantic and Indian Oceans together with a divergence center over the Maritime Continent. The equatorial easterly anomalies are interpreted as partly due to Kelvin wave response to positive rainfall anomalies over the equatorial Pacific and partly due to Rossby wave response to negative rainfall anomalies over the Maritime Continent. As part of this Rossby wave response, an anticyclone covers the entire southern tropical Indian Ocean. Another anticyclone feature is captured over the northern Indian Ocean (10°–30°N). Our working hypothesis is that El Niño–induced easterly wind anomalies over the near-equatorial Indian Ocean, due to Ekman pumping, lead to anticyclonic vorticity over the northern Indian Ocean, promoting dry air intrusion from the north to the South Asian monsoon region. In addition, the anticyclonic circulation over the southern tropical Indian Ocean opposes the cross-equatorial monsoon flow and thereby prevents moisture inflow. Because of a lack of a sufficient length of high temporal (daily or pentad) model outputs in CM2.1, and to isolate ENSO’s exclusive impact, we perform a series of AGCM experiments to demonstrate our hypothesis. Further, Fig. 1b suggests that about 30% of the severe monsoons are not associated with ENSO. In these years, we examine the role of regional SST variations.

Fig. 2.

Anomalous precipitation (shaded, mm day−1) and 850-hPa streamlines composited for severe weak monsoons accompanied with El Niño in CM2.1 and averaged for (a) April–May and (b) June–September. Only anomalies above the 90% confidence level are plotted.

Fig. 2.

Anomalous precipitation (shaded, mm day−1) and 850-hPa streamlines composited for severe weak monsoons accompanied with El Niño in CM2.1 and averaged for (a) April–May and (b) June–September. Only anomalies above the 90% confidence level are plotted.

To identify the possible moist processes, we perform moisture and moist static energy (MSE) budgets. We follow closely the procedures adopted in Su and Neelin (2002), Neelin and Su (2005), and Annamalai (2010). A difference here is that unlike the less complicated models used in Su and Neelin (2002) or linear model employed in Annamalai (2010), budget diagnostics are performed for solutions obtained in a fully nonlinear AGCM and CM2.1 integrations.

The remainder of the paper is organized as follows. Section 2 deals with the datasets along with a brief description of the models used. Details of the sensitivity experiments and diagnostics methodology are also included. In section 3, major results from the sensitivity experiments together with budget diagnostics and validation of the proposed hypothesis are documented. The potential role of regional SST anomalies in causing severe monsoons is addressed in section 4. A summary and implications of the results are provided in section 5.

2. Data, method, model, and experiments

a. Data and method

To identify observed severe monsoon years, the traditional AIR index, constructed from a weighted average of 306 stations spread over the whole of the Indian subcontinent (Parthasarathy et al. 1994), has been used. To construct observed ENSO evolution, the Hadley Centre interpolated SST (HadISST) datasets of Rayner et al. (2003) are used. Both AIR and SST datasets cover the period 1871–2008, and five (eight) severe strong (weak) monsoons occurring with La Niña (El Niño) are identified. Severe monsoons are based on threshold cutoffs of ±1.5 standard deviations, while El Niño and La Niña events are categorized based on Niño-3.4 SST anomalies exceeding ±1.0 standard deviation during summer. To compare the model basic states, precipitation from the Climate Prediction Center (CPC) Merged Analysis Precipitation (CMAP; Xie and Arkin 1996) and 850-hPa wind from the National Centers for Environmental Prediction (NCEP)–National Center for Atmospheric Research (NCAR) reanalysis (Kalnay et al. 1996) for the period 1979–2008 are used.

Because of limited samples in observations, we have examined the five-member ensemble 20c3m integrations performed with CM2.1. The runs spanning the period 1861–2000 were conducted at GFDL as part of CMIP3 (Meehl et al. 2007). In the model runs, monthly anomalies were created with respect to the 1861–2000 climatology, and summer season rainfall averaged over the region 7°–30°N, 65°–95°E is considered to represent the South Asian monsoon variability. Based on the threshold of ±1.5 standard deviations, 26 cases of severe weak and 14 severe strong monsoons associated with El Niño and La Niña were identified, respectively. Then, seasonally stratified composite analysis, separately for severe weak and strong cases, is performed. These results are then subjected to a t test for statistical significance and discussions are confined to those regions where the significance is greater than 90%. The 20c3m integrations have been extensively used in ENSO (e.g., Joseph and Nigam 2006; Wittenberg et al. 2006), monsoon (e.g., Annamalai et al. 2007), and tropical Indian Ocean climate variability (e.g., Annamalai et al. 2010) studies.

b. Model

A brief description of CM2.1 is provided here, and formulation details are documented in Delworth et al. (2006). The atmosphere component of CM2.1 (AM2.1; Anderson et al. 2004) has a finite volume dynamical core (Lin 2004) with 24 vertical levels and horizontal resolution of 2.5° longitude and 2.0° latitude. The land surface model has an isothermal surface, three water sources (snow, root zone, and groundwater), and 18 soil temperature levels to 6-m depth. The ocean component is based on the Modular Ocean Model (MOM) version 4 (Griffies et al. 2003) and has 50 vertical levels (10-m resolution in the upper 220 m). The horizontal resolution is 1° × 1° but its meridional resolution reduces to ⅓° × ⅓° equatorward of 30°. Different components of the CM2.1 are coupled through the Flexible Modeling System, without applying flux corrections. The atmosphere, ocean, land, and sea ice exchange fluxes every 2 h.

c. AGCM experimental designs

The AGCM used for the sensitivity experiments is AM2.1. Anderson et al. (2004) performed ensemble simulation with observed SST during 1951–2000 and highlighted the ability of AM2.1 to capture the ENSO–monsoon relationship. The difference here is that we use CM2.1 SST as forcing to address our hypothesis. For the control (CTL) simulation, monthly varying climatological SST constructed from CM2.1 was prescribed.

To isolate the exclusive role of ENSO on the monsoon evolution, the following sensitivity runs are carried out: (i) TPO_WARM and (ii) TPO_COLD. Here, the forcing is the monthly varying SST anomalies constructed from CM2.1 composites of severe weak monsoons during the developing phase of El Niño for TPO_WARM, and severe strong monsoons during the developing phase of La Niña for TPO_COLD (Figs. 1c,d). For both experiments, prescribed anomalies together with CM2.1 climatology cover the tropical Pacific Ocean (30°S–30°N, 120°E–80°W). In the rest of the oceans, monthly varying climatological SSTs are prescribed. To understand severe monsoons caused by regional SST anomalies, a third experiment named EQWPAC_WARM, with SST anomalies prescribed over the equatorial western Pacific, is performed. In CM2.1, 14 severe weak monsoon years are associated with warming over the equatorial western Pacific. All the experiments are conducted for the period 1 March–30 November. In each experiment, to account for atmospheric internal variability, a 25-member ensemble is carried out. The members differ only in their initial conditions that span the period from 15 February to 14 March obtained from CM2.1 archives for the CTL run and from the CTL run for all the sensitivity experiments. A summary of these experiments is provided in Table 1.

Table 1.

Details of various AM2.1 experiments conducted, including the SST forcing region and forcing function.

Details of various AM2.1 experiments conducted, including the SST forcing region and forcing function.
Details of various AM2.1 experiments conducted, including the SST forcing region and forcing function.

To illustrate the spatial extent and strength of anomalous SST prescribed in the experiments, the April–May averaged composites for the three cases are shown in Fig. 3. During El Niño (Fig. 3a), SST warming greater than 0.5°C is noted along the entire equatorial Pacific with a peak over the eastern basin. While the warm anomalies during El Niño extend beyond the equatorial belt, cold SST anomalies during La Niña are merely confined at the equator (Fig. 3b), possibly due to skewness in ENSO properties. In contrast, in the western Pacific case, any noticeable warm anomalies (~1.0°C) are situated only around the date line. A comparison between the TPO_WARM and EQWPAC_WARM solutions is expected to show the differences in the simulated equatorial diabatic heating and the associated tropical circulation anomalies to geographical location of prescribed maximum SST anomalies, and subsequently in the source or pathways impacting the dryness over South Asia.

Fig. 3.

April–May average composite SST anomalies (°C) used as boundary forcing for AM2.1 sensitivity experiments: (a) TPO_WARM, (b) TPO_COLD, and (c) EQWPAC_WARM. Warm SST anomalies are shaded progressively and cold anomalies are shown in contours with an interval of 0.5°C. Only anomalies above the 90% confidence level are plotted. The composites are based on (a) severe weak monsoon and El Niño years, (b) severe strong monsoon and La Niña years, and (c) severe weak monsoon and equatorial western Pacific warming years.

Fig. 3.

April–May average composite SST anomalies (°C) used as boundary forcing for AM2.1 sensitivity experiments: (a) TPO_WARM, (b) TPO_COLD, and (c) EQWPAC_WARM. Warm SST anomalies are shaded progressively and cold anomalies are shown in contours with an interval of 0.5°C. Only anomalies above the 90% confidence level are plotted. The composites are based on (a) severe weak monsoon and El Niño years, (b) severe strong monsoon and La Niña years, and (c) severe weak monsoon and equatorial western Pacific warming years.

d. Moisture and moist static energy budget formulations

Here, we briefly provide the formulations and readers are referred to Su and Neelin (2002) for details. The vertical integrated temperature T and moisture q equations for the perturbations have the following form:

 
formula
 
formula

where both T′ and q′ are in energy units (W m−2) after absorbing the heat capacity at constant pressure Cp and latent heat of condensation L, respectively. Also, s′ = T′ + Φ′ is dry static energy, with Φ′ being the geopotential; Qc and Qq are the anomalous convective heating and moisture sink, respectively; g is acceleration due to gravity; ω is vertical pressure velocity; PT is the reference pressure depth of the troposphere; and is the net radiative flux convergence into the atmospheric column. The surface sensible and latent heat fluxes are H′ and E′, respectively. Angle brackets indicate vertical integration. The operators DT and Dq include both horizontal advection and diffusion terms:

 
formula

where KH is the diffusion coefficient, which is the amount of substance (here, T for DT and q for Dq) that diffuses across a unit area in 1 s under the gradient of one unit.

Combining Eqs. (1) and (2), the vertically integrated anomalous MSE equation under steady-state conditions is

 
formula

where is the anomalous MSE convergence. Also h = s + q is the MSE. Here, the large terms in individual temperature and moisture equations, and , cancel each other and the resultant balance is between net flux into the column and stability (Annamalai 2010).

Since , anomalous precipitation can be calculated from Eq. (2).

3. Moist teleconnection between the tropical Pacific and monsoon

a. Climatology and composite patterns

Figure 4 shows (left) April–May and (right) June–September average rainfall and 850-hPa wind climatologies from observations (Figs. 4a,d), CM2.1 (Figs. 4b,e), and the AM2.1 CTL run (Figs. 4c,f). Note that differences in the basic states between CM2.1 and AM2.1 suggest the role of air–sea interaction. During spring, in agreement with observations, the models capture the large-scale patterns in rainfall including the Pacific and Indian Ocean intertropical convergence zone (ITCZ) and the South Pacific convergence zone (SPCZ). Yet, there are differences in the magnitude along with the unrealistic double ITCZ feature in CM2.1. As regards to low-level flow, easterly trades over the tropical Pacific and Atlantic, as well as the weak westerlies along the equatorial Indian Ocean, are ably captured.

Fig. 4.

Climatological precipitation (shaded, mm day−1) and 850-hPa winds (vector, m s−1) averaged for (a)–(c) April–May and (d)–(f) June–September from (a),(d) observations, (b),(e) CM2.1, and (c),(f) AM2.1. The reference wind vector (15 m s−1) is also shown.

Fig. 4.

Climatological precipitation (shaded, mm day−1) and 850-hPa winds (vector, m s−1) averaged for (a)–(c) April–May and (d)–(f) June–September from (a),(d) observations, (b),(e) CM2.1, and (c),(f) AM2.1. The reference wind vector (15 m s−1) is also shown.

An important feature of the monsoon annual cycle is the north-northwestward migration of rainfall and the establishment of a strong cross-equatorial flow along the western Indian Ocean with subsequent development of westerlies over the northern Indian Ocean during summer. While the salient features are simulated by both CM2.1 (Fig. 4e) and AM2.1 (Fig. 4f), compared to observations the coupled model has a relatively stronger (weaker) precipitation over the eastern equatorial Indian Ocean (monsoon trough around 20°N), but the opposite is true for AM2.1. In particular, in AM2.1 rainfall strength over the Arabian Sea is comparable to that over the Bay of Bengal and the simulated rainfall does not extend beyond 20°N (Fig. 4f). In CM2.1, the equatorial easterlies over the Atlantic have further weakened. All these model systematic errors are taken into account while interpreting our results.

The anomalous composite patterns during late spring (Fig. 2a) for severe monsoons are already discussed in section 1. The striking feature during summer (Fig. 2b) is the persistence of the anomalies over the entire tropical Indo-Pacific basins. One difference is the increase in rainfall over the tropical northwestern Pacific (10°–20°N, 120°–160°E). A possible interpretation is that at low levels the El Niño–induced Rossby wave response results in cyclonic vorticity and that, together with westerlies stemming from the Bay of Bengal, it leads to rainfall enhancement (Annamalai and Liu 2005). Next, AM2.1 solutions are presented to isolate the exclusive role of ENSO.

b. ENSO-induced teleconnection anomalies

In the TPO_WARM experiment (Fig. 5a), in response to imposed warm SST anomalies, increase in rainfall and low-level westerly wind anomalies along the equatorial east-central Pacific are accompanied by reduced rainfall over the equatorial western Pacific and easterly wind anomalies along the equatorial Indian Ocean. These perturbations in rainfall or diabatic heating and low-level circulation fields over the tropical Indo-Pacific are robust features during the developing phase of El Niño (e.g., Soman and Slingo 1997; Lau and Nath 2000; Annamalai and Liu 2005; Turner et al. 2005). As noted in CM2.1 composites (Fig. 2a), over the tropical Indian Ocean two anticyclonic circulation cells (one in each hemisphere), together with reduced rainfall over the northern Indian Ocean extending to tropical western Pacific, are also simulated. The agreement in the spatial structure of rainfall and circulation features between CM2.1 composites (Fig. 2a) and AM2.1 forced with warm SST anomalies only over the tropical Pacific (Fig. 5a) confirms that the reduction in rainfall and anticyclonic vorticity over the northern Indian Ocean appear to be “remotely” forced.

Fig. 5.

Dominant terms in moisture budget from TPO_WARM experiment and averaged for April–May. The anomalies are calculated as difference between the ensemble means of TPO_WARM and CTL runs. (a) Precipitation (shaded) and 850-hPa winds (vector, m s−1), (b) vertically integrated moisture convergence, (c) vertically integrated horizontal moisture advection, and (d) evaporation. The red (blue) box represents the South Asian monsoon (equatorial eastern Pacific) region, where the budget analysis is focused. Units are W m−2, but the scales vary by panel. Reference wind vector (8 m s−1) is provided in (a). The area covered by red (blue) box is 7°–20°N, 60°–100°E (5°S–5°N, 180°–90°W).

Fig. 5.

Dominant terms in moisture budget from TPO_WARM experiment and averaged for April–May. The anomalies are calculated as difference between the ensemble means of TPO_WARM and CTL runs. (a) Precipitation (shaded) and 850-hPa winds (vector, m s−1), (b) vertically integrated moisture convergence, (c) vertically integrated horizontal moisture advection, and (d) evaporation. The red (blue) box represents the South Asian monsoon (equatorial eastern Pacific) region, where the budget analysis is focused. Units are W m−2, but the scales vary by panel. Reference wind vector (8 m s−1) is provided in (a). The area covered by red (blue) box is 7°–20°N, 60°–100°E (5°S–5°N, 180°–90°W).

During summer (Fig. 6a), the rainfall and circulation anomalies intensify as a result of further El Niño–related warming (Fig. 1d) and the positive feedback between diabatic heating and circulation anomalies. An additional feature is the simulation of positive rainfall anomalies over the tropical northwestern Pacific replacing the reduced rainfall noted during late spring. Again, CM2.1 composites (Fig. 2b) are in good agreement with results obtained from this experiment. It should be noted here that due to the climatological rainfall bias in AM2.1 (Fig. 4f), the negative rainfall anomalies are concentrated over the Arabian Sea region than over the monsoon land areas. The results from TPO_COLD experiment are mirror images to those of TPO_WARM and are briefly discussed later (section 3e). Next, we perform budget diagnostics to identify the moist processes that initiate and maintain the dryness over South Asia.

Fig. 6.

As in Fig. 5, but for June–September. Negative anomalies are progressively shaded and positive anomalies are represented by contours. Scales and contour intervals vary by panel. In (a) and (b) the contour interval is 50 W m−2, but the first contour is 20 W m−2, while in (c) it is 20 W m−2 and in (d) the contour interval is 10 W m−2. In (a) the wind (vector) is in m s−1 and a reference vector (8 m s−1) is also given. All other parameters are expressed in W m−2.

Fig. 6.

As in Fig. 5, but for June–September. Negative anomalies are progressively shaded and positive anomalies are represented by contours. Scales and contour intervals vary by panel. In (a) and (b) the contour interval is 50 W m−2, but the first contour is 20 W m−2, while in (c) it is 20 W m−2 and in (d) the contour interval is 10 W m−2. In (a) the wind (vector) is in m s−1 and a reference vector (8 m s−1) is also given. All other parameters are expressed in W m−2.

c. Moisture and moist static energy budgets

The budget analyses are focused over two regions (shown in Fig. 5): South Asia (area outlined with red solid lines covering 7°–20°N, 60°–100°E) and the equatorial Pacific (area outlined with blue lines covering 5°S–5°N, 180°–90°W). Budget diagnostics are confined to the TPO_WARM experiment because the simulated precipitation anomalies over both target regions (Fig. 5a) agree qualitatively with CM2.1 composites (Fig. 2). Results shown are anomalies for the ensemble run relative to the CTL ensemble. Brief budget discussions of the TPO_COLD experiment and CM2.1 composites are provided in sections 3d and 3e, respectively.

1) Moisture budget

Figures 5a–d show the four terms of the moisture budget [Eq. (2)] averaged for April–May months. They are precipitation , moisture convergence , moisture advection , and evaporation . It should be noted here that in Fig. 5b a positive sign means net moisture convergence into the column and vice versa. Similarly, moist advection (Fig. 5c) into the column is positive and dry advection is negative.

For the wet conditions over the equatorial Pacific, anomalous moisture convergence contributes about 75% while the remaining 25% is due to increased evaporation (Fig. 5d). Moisture convergence and increase in evaporation are expected over prescribed warm SST anomaly regions, and our budget results for spring agree closely with those presented in Su and Neelin (2002) for the boreal winter season.

For the dry conditions over South Asia, about 75% is explained by moisture divergence, nearly 20% is accounted for by dry advection, and the remaining 5% is contributed by reduced evaporation. In this experiment, it needs to be recollected that only monthly varying climatological SST was inserted over the tropical Indian Ocean. Therefore, the decrease in evaporation is primarily due to weakened cross-equatorial monsoon flow. Further, the anomalous northeasterlies associated with the northern Indian Ocean anticyclone advect moist air from the Arabian Sea to the horn of Africa that is nearly balanced by reduced evaporation. Over South Asia, dry advection plays a dominant role, and this aspect is discussed in detail later.

The moisture budget terms for summer are shown in Fig. 6. Over the entire tropical Indo-Pacific basin, the anomalies have intensified with some particular changes from spring to summer. For instance, largely due to moisture convergence (Fig. 6b), positive rainfall anomalies have replaced the spring season reduced rainfall (Fig. 5b) over the tropical western Pacific (10°–20°N, 120°–160°E). The off-equatorial reduced rainfall anomalies over the tropical Indian Ocean, in both hemispheres, have maximum amplitude around 10°N/S. The low-level divergence center has shifted westward from the Maritime Continent into the eastern equatorial Indian Ocean (Fig. 6b). While precipitation and moisture divergence are the dominant balancing terms over South Asia, dry advection from the north by the anomalous northerlies (27%) and reduction in evaporation (12%) due to weakened monsoon flow make substantial contributions in maintaining the dryness.

In the deep tropics, while moisture convergence dominates the moisture budget, it is not the “cause” but a “feedback” in the chain of moist convective response to SST forcing (cf. Su and Neelin 2002; Annamalai 2010). As mentioned earlier, MSE budget [Eq. (4)] analysis provides a better approach to understanding thermodynamic balance because it combines physical processes influencing temperature and moisture (Neelin and Su 2005). In particular, in the MSE budget the effect of other processes in determining convergence can be brought out because moisture convergence and adiabatic cooling associated with rising motions cancel out substantially.

2) Moist static energy budget

Figure 7 shows the prominent terms of the MSE budget averaged for April–May, namely MSE convergence or export , sensible heat flux H′, temperature advection , and net radiation . Note that and E′ are already shown in Figs. 5c and 5d, respectively. Over convective regions, cancellation between convergence of dry static energy and moisture results in MSE convergence that has less magnitude compared to moisture convergence but retains the same pattern. In Fig. 7a, positive values over the equatorial east-central Pacific indicate net convergence of high-MSE air in the lower troposphere and hence export out of the region at the upper troposphere, again consistent with that shown for boreal winter in Su and Neelin (2002). In contrast, negative values over South Asian and western Pacific monsoon regions indicate import of low-MSE air (Fig. 7a). Let us now quantify the contributions from individual terms over the target regions.

Fig. 7.

Dominant terms in MSE budget from TPO_WARM averaged for April–May. The anomalies are calculated as the difference between the ensemble means of TPO_WARM and CTL. (a) Vertically integrated MSE divergence, (b) sensible heat flux, (c) vertically integrated horizontal temperature advection, and (d) net radiative flux. Units are W m−2, but scales vary by panel.

Fig. 7.

Dominant terms in MSE budget from TPO_WARM averaged for April–May. The anomalies are calculated as the difference between the ensemble means of TPO_WARM and CTL. (a) Vertically integrated MSE divergence, (b) sensible heat flux, (c) vertically integrated horizontal temperature advection, and (d) net radiative flux. Units are W m−2, but scales vary by panel.

Over the equatorial Pacific, the good agreement in spatial patterns between rainfall (Fig. 5a) and net radiation (Fig. 7d) suggests the dominance of cloud-radiative warming, primarily due to longwave warming. Temperature advection (Fig. 7c) has a wavy nature in the extratropics (Fig. 7c) and sensible heat flux (Fig. 7b) over the equatorial Pacific contributes about 10% to the budget. Therefore, over the El Niño region, the overall MSE export by the divergent flow is balanced by evaporation (Fig. 5d). The results obtained from AGCM solutions here are consistent with the intermediate model solutions of Su and Neelin (2002).

Over South Asia, the increase in sensible heat flux over continental India and extending to plains of Indo-China (Fig. 7b) is perhaps due to lack of rainfall, and is roughly balanced by cold air advection (Fig. 7c). Climatologically, sensible heat flux and the associated land surface warming peaks during April–May (e.g., Annamalai et al. 1999), and a reduction in premonsoon rainfall, further amplify the signal. The contributions from evaporation (21%; Fig. 5d) and net radiation (9%; Fig. 7d) are important but are relatively less so compared to dry advection (~70%; Fig. 5c). Over the horn of Africa, moist advection is reduced compared to the sum of evaporation and net radiation (radiative cooling), resulting in descent and import of low-MSE air. An examination of the summer MSE budget (Fig. 8) indicates persistence of the spring patterns, with dry air advection still dominating (70%; Fig. 6c) over South Asia.

Fig. 8.

As in Fig. 7, but for June–September. Negative anomalies are progressively shaded and positive anomalies are represented by contours. Units are W m−2, but scales and contour intervals vary by panel. In (a) the contour interval is 30 W m−2, but the first contour is 10 W m−2. In (b) the contour interval is 5 W m−2 and in (c) and (d) it is 10 W m−2.

Fig. 8.

As in Fig. 7, but for June–September. Negative anomalies are progressively shaded and positive anomalies are represented by contours. Units are W m−2, but scales and contour intervals vary by panel. In (a) the contour interval is 30 W m−2, but the first contour is 10 W m−2. In (b) the contour interval is 5 W m−2 and in (c) and (d) it is 10 W m−2.

3) Moist mechanism

During spring, MSE analysis underlines the role of dry advection in initiating the dryness over South Asia during the developing phase of El Niño. This motivates us to examine the role of moisture advection in more detail. To understand the relative contributions from horizontal wind and moisture gradient, the moisture advection can be partitioned into

 
formula

The first term on the right-hand side denotes advection associated with climatological wind acting on anomalous moisture gradient, and the second term represents advection due to anomalous wind acting on climatological moisture gradient. The third term is a contribution from advection associated with anomalous wind acting on anomalous moisture gradient, and the last term is residual due to transient variability.

Figures 9a–c show moisture advection terms averaged for April–May. To aid in their interpretation, Fig. 9d shows April–May averaged climatology of the vertically integrated specific humidity from the CTL run together with 850-hPa wind anomalies from the TPO_WARM experiment. Over South Asia, is the principal contributor to the moisture advection equation. This is interpreted as follows: the poleward flank of the anomalous anticyclonic circulation over the Arabian Sea sector (Fig. 9d) advects climatological air of low moisture content from the north, resulting in dry advection. In contrast, the southern flank of this anticyclone advects climatological air of high moisture content from the Arabian Sea to the horn of Africa. The other two terms’ contribution is marginal over South Asia but they tend to cancel each other over the region 20°S–10°N, 30°–80°E. A further examination suggests that the anomalous meridional wind acting on climatological moisture gradient plays a dominant role in the dry advection over South Asia (not shown).

Fig. 9.

(a)–(c) Dominant terms of the moisture advection from TPO_WARM experiment averaged for April–May: (a) anomalous winds acting on climatological moisture gradient; (b) climatological wind acting on anomalous moisture gradient; (c) anomalous wind acting on anomalous moisture gradient. Units are W m−2 with different scales in each panel. (d) Vertically integrated specific humidity (kg m−2, shaded) from CTL and 850-hPa anomalous wind (vector, m s−1) as the difference between the ensemble means of TPO_WARM and CTL. A reference wind vector (3 m s−1) is also given.

Fig. 9.

(a)–(c) Dominant terms of the moisture advection from TPO_WARM experiment averaged for April–May: (a) anomalous winds acting on climatological moisture gradient; (b) climatological wind acting on anomalous moisture gradient; (c) anomalous wind acting on anomalous moisture gradient. Units are W m−2 with different scales in each panel. (d) Vertically integrated specific humidity (kg m−2, shaded) from CTL and 850-hPa anomalous wind (vector, m s−1) as the difference between the ensemble means of TPO_WARM and CTL. A reference wind vector (3 m s−1) is also given.

A question of interest here is this: is the low-level anticyclone over the Arabian Sea forced by in situ negative rainfall anomalies or triggered by the equatorial Indian Ocean easterlies forced by El Niño? In other words, does the anticyclone induced dry advection cause the initiation of negative rainfall anomalies over South Asia? Figure 10 shows the daily evolution of ensemble mean rainfall averaged over South Asia (5°–25°N, 60°–100°E) and relative vorticity at 850 hPa over the northern Indian Ocean (0°–20°N, 40°–80°E) from 16 March. Within a few days of the start of the sensitivity experiment, the vorticity signature becomes anticyclonic and is maintained throughout with occasional fluctuations. In contrast, negative rainfall anomalies begin only after 16 April (i.e., nearly a month after the anticyclonic signature). Thus, budget diagnostics performed with TPO_WARM solutions support our hypothesis that dry advection initiates the dryness, and the long lead time provides a useful predictive signal for severe weak monsoons during El Niño. The persistence of the anticyclonic circulation together with continuous dry air intrusion during summer then maintains the severe weak monsoons but the lead–lag relationship between vorticity and rainfall reverses around early May (Fig. 10), suggesting anomalous circulation response to rainfall anomalies. Annamalai (2010) noted the prevalence of moisture advection in the linkage between the equatorial Indian Ocean SST anomalies and the South Asian monsoon trough.

Fig. 10.

Five-day running mean of temporal evolution of anomalous rainfall (mm day−1, dotted line) over the South Asian monsoon region (5°–25°N, 60°–100°E) and relative vorticity at 850 hPa (10−5 s−1, solid line) over the northern Indian Ocean (0°–20°N, 40°–80°E) from day 16 of model integrations. The anomalies are calculated as the difference between the ensemble means of TPO_WARM and CTL.

Fig. 10.

Five-day running mean of temporal evolution of anomalous rainfall (mm day−1, dotted line) over the South Asian monsoon region (5°–25°N, 60°–100°E) and relative vorticity at 850 hPa (10−5 s−1, solid line) over the northern Indian Ocean (0°–20°N, 40°–80°E) from day 16 of model integrations. The anomalies are calculated as the difference between the ensemble means of TPO_WARM and CTL.

d. TPO_COLD solutions

Figure 11 shows the major terms of moisture and MSE budgets of April–May from the TPO_COLD experiment. Over the equatorial Pacific, as expected, negative rainfall anomalies (Fig. 11a) are balanced by moisture divergence (75%; Fig. 11b) and reduced evaporation (18%; Fig. 11d). Over South Asia, while moisture convergence (65%) dominates the rainfall increase, contributions from evaporation (20%) as well as from moist advection (15%) are noteworthy. The low-level circulation anomalies (Fig. 11a) over the tropical Indo-Pacific basins depict a mirror image to those from TPO_WARM (Fig. 5a). In particular, cyclonic circulations on either side of the equator, anomalous cross-equatorial flow along the East African highlands, and westerly anomalies over the equatorial western Indian Ocean are prominent features. As noted earlier (Fig. 5), here too rainfall anomalies are stronger over the Arabian Sea, perhaps associated with the bias in the basic state (Fig. 4f).

Fig. 11.

Dominant terms in moisture and MSE budget averaged for April–May from TPO_COLD. The anomalies are calculated as difference between the ensemble means of TPO_COLD and CTL: (a) precipitation (shaded) and 850-hPa winds (vector, m s−1), (b) vertically integrated moisture convergence, (c) vertically integrated horizontal moisture advection, (d) evaporation, (e) vertically integrated MSE divergence, and (f) net radiative flux. Shading in (e) and (f) is as in (c) and (d). Units are W m−2. Positive anomalies are progressively shaded and negative anomalies are represented by contours. Contour interval is 50 W m−2 in (a). It is 30 W m−2 in (b), but first contour is 10 W m−2. Contour interval is 10 W m−2 in (c)–(e) and 5 W m−2 in (f). Units are W m−2. Reference wind vector (2 m s−1) is shown in (a).

Fig. 11.

Dominant terms in moisture and MSE budget averaged for April–May from TPO_COLD. The anomalies are calculated as difference between the ensemble means of TPO_COLD and CTL: (a) precipitation (shaded) and 850-hPa winds (vector, m s−1), (b) vertically integrated moisture convergence, (c) vertically integrated horizontal moisture advection, (d) evaporation, (e) vertically integrated MSE divergence, and (f) net radiative flux. Shading in (e) and (f) is as in (c) and (d). Units are W m−2. Positive anomalies are progressively shaded and negative anomalies are represented by contours. Contour interval is 50 W m−2 in (a). It is 30 W m−2 in (b), but first contour is 10 W m−2. Contour interval is 10 W m−2 in (c)–(e) and 5 W m−2 in (f). Units are W m−2. Reference wind vector (2 m s−1) is shown in (a).

We examined all the MSE budget terms and found that the contributions from sensible heat flux and temperature advection are indeed too small and hence not shown. Briefly, over the equatorial Pacific reduced evaporation and radiative cooling account for dry air import and anomalous descent, and hence negative rainfall anomalies. Over South Asia, cloud-radiative warming (20%), enhanced evaporation (45%; again primarily due to increase in wind speed), and moist advection (32%) contribute to net convergence of MSE air at low levels. Specifically, the low-level cyclonic flow over the northern Indian Ocean advects air of high moisture content from the equatorial regions to fuel convection over South Asia. Our interpretation is that the enhanced rainfall during late spring or the premonsoon and the associated diabatic heating anomalies would anchor the ensuing monsoon circulation.

e. Budget analysis in CM2.1 composites

Over the two target regions, we examined the budget with the available outputs in CM2.1 archives. The terms that make substantial contributions during April–May are shown in Fig. 12 (Fig. 13) for severe weak (strong) monsoons and El Niño (La Niña) categories. During severe weak monsoon years, regarding the moisture budget over the equatorial Pacific, moisture convergence (85%) and moist advection (12%) account for the rainfall variations. Surprisingly, the contribution from evaporation is confined to the far eastern Pacific. Over South Asia, moisture divergence (75%), dry advection (10%), and reduced evaporation (5%) make up the budget. Again, the reasoning for dry air advection over South Asia is similar to that provided in section 3c but accumulated dry air concentration is higher over eastern India and the Bay of Bengal. Turning to the MSE budget, over the equatorial Pacific moist advection (50%) and net radiation (35%) dominate while over South Asia evaporation (40%) and dry advection (50%) contribute. A comparison of diagnostics between forced (Fig. 5) and coupled (Fig. 12) integrations reveal a striking difference in the role of evaporation over the equatorial Pacific. Budget diagnostics performed for summer (not shown) suggest that dry advection contributes about 50% to the MSE divergence over South Asia.

Fig. 12.

As in Fig. 11, but for the composites of severe weak monsoons occurring with El Niño in CM2.1.

Fig. 12.

As in Fig. 11, but for the composites of severe weak monsoons occurring with El Niño in CM2.1.

Fig. 13.

As in Fig. 12, but for the composite of severe strong monsoon years occurring with La Niña. Positive anomalies are progressively shaded and negative anomalies are represented by contours. Contour interval is 20 W m−2 in (a) and (b).

Fig. 13.

As in Fig. 12, but for the composite of severe strong monsoon years occurring with La Niña. Positive anomalies are progressively shaded and negative anomalies are represented by contours. Contour interval is 20 W m−2 in (a) and (b).

Figure 13 shows budget terms for severe strong monsoons occurring with La Niña. Over the equatorial Pacific, moisture divergence (77%) dominates the moisture budget while modest contribution by dry advection (20%) to the west of the date line is noteworthy. Here, too, evaporation makes little impact. Over South Asia, moisture convergence (85%) and moist advection (13%) make up the moisture budget. However, unlike in the forced experiment (Fig. 11a), the source of moist advection stems from plains of Indo-China and the tropical western Pacific. It is unclear whether this difference arises from differences in the basic state (Fig. 4) or the fact that the La Niña–monsoon association in coupled models is different. These aspects will be examined in a future study.

4. Moist teleconnection between regional SST and severe monsoons

As mentioned in section 1b, severe monsoons are also noted during non-ENSO years in 20c3m integrations (Fig. 1b). Now we focus on the SST variations over the Indo-Pacific warm pool regions in causing them. For non-ENSO years, if summer area-averaged (5°–30°N, 60°–100°E) rainfall anomalies over South Asia exceed 1.5 standard deviations, and if summer SST anomalies over either the equatorial Indian Ocean (10°S–0°, 90°–110°E) or western Pacific (5°S–5°N, 150°E–180°) exceed 1.0 standard deviation, then those years are considered for composite analysis and also to perform numerical experiments. In 20c3m, we note 6 and 16 severe weak monsoon years associated with regional SST anomalies over equatorial Indian Ocean and western Pacific, respectively. One difference between ENSO and regional SST is that the anomalies in the latter do not develop with sufficient amplitude until summer. Therefore, discussions are confined to summer and the western Pacific case where the sample is relatively higher (16 events).

The anomalous rainfall together with 850-hPa wind (Fig. 14a) and SST (Fig. 14b) suggest that modest warming is concentrated to the west of the date line while anomalous westerly winds and positive rainfall extend over the tropical western Pacific. Over South Asia, anomalous negative rainfall and anticyclonic circulation anomalies together with below-normal rainfall over Maritime Continent are prominent. To understand if this modest warming, compared to basinwide warming during El Niño (Fig. 3a), is sufficient enough to cause severe weak monsoons, the EQWPAC_WARM (Table 1) experiment is performed.

Fig. 14.

(a),(b) June–September CM2.1 composite anomaly of severe weak monsoon occurring with equatorial western Pacific warming for (a) rainfall (shaded, W m−2) and 850-hPa winds (vector, m s−1) and (b) SST (°C). Reference vector (2 m s−1) is provided with (a). (c) Anomalous rainfall (shaded, W m−2) and 850-hPa winds (vector, m s−1) from EQWPAC_WARM. Reference vector (5 m s−1) is also provided. The anomalies are calculated as difference between the ensemble means of EQWPAC_WARM and CTL for (c). Positive anomalies are represented by contours and negative anomalies are progressively shaded. Contour interval is 30 W m−2 in (a) and is 50 W m−2 in (c).

Fig. 14.

(a),(b) June–September CM2.1 composite anomaly of severe weak monsoon occurring with equatorial western Pacific warming for (a) rainfall (shaded, W m−2) and 850-hPa winds (vector, m s−1) and (b) SST (°C). Reference vector (2 m s−1) is provided with (a). (c) Anomalous rainfall (shaded, W m−2) and 850-hPa winds (vector, m s−1) from EQWPAC_WARM. Reference vector (5 m s−1) is also provided. The anomalies are calculated as difference between the ensemble means of EQWPAC_WARM and CTL for (c). Positive anomalies are represented by contours and negative anomalies are progressively shaded. Contour interval is 30 W m−2 in (a) and is 50 W m−2 in (c).

AM2.1 solutions capture enhanced rainfall over the equatorial western Pacific and decreased rainfall over the Maritime Continent (Fig. 14c), and the near-equatorial circulation anomalies are consistent with these changes in diabatic heating. Also captured are rainfall increase and anomalous low-level cyclonic vorticity over the tropical western Pacific, and weakened rainfall and anticyclonic circulation anomalies over South Asia. Barring magnitude, the overall spatial patterns in budget terms from EQWPAC_WARM closely resemble those from TPO_WARM (Figs. 5 and 7). This indicates that in both experiments, the mechanism operating in summer seems to be similar. The difference is that the daily evolution of ensemble mean rainfall over South Asia and relative vorticity at 850 hPa over the northern Indian Ocean shows that in EQWPAC_WARM, vorticity leads rainfall only by 4–5 days (not shown) instead of 25 days in TPO_WARM (Fig. 10). However, due to differences in basic states between CM2.1 and AM2.1 (Fig. 4), simulated regions of maximum rainfall differ (cf. Figs. 4a and 4c).

5. Summary and discussion

a. Summary

In the present research moisture and MSE budget diagnostics are applied to identify moist processes that initiate and maintain the dryness (wetness) over South Asia during severe weak (strong) monsoon years that are associated with the developing phase of El Niño (La Niña; Figs. 1 and 3). A particular focus is to identify a precursor signal during boreal spring. Because of the paucity of observations related to moist processes, we diagnosed CM2.1 coupled model integrations and performed sensitivity experiments with AM2.1 (Table 1).

Over the tropical Indo-Pacific, AM2.1 solutions to El Niño forcing (TPO_WARM) capture the essential features (Fig. 5a) noted in CM2.1 composites (Fig. 2a). This consistency permits budget analyses in AM2.1 solutions over the two target regions, namely the equatorial Pacific and South Asia. Both during spring and summer, anomalous precipitation and moisture divergence (70%–80%) balance the moisture budget (Figs. 5 and 6), shadowing the contributions from moisture advection and evaporation. The cancellation of large terms in the MSE budget permits analysis of balances among terms that might seem small in the individual moisture and temperature equations but are vital in moist teleconnection mechanisms.

During spring, over the equatorial Pacific evaporation (60%) and radiative warming due to cloud-radiative forcing (36%) balance the MSE export. Over South Asia, dry advection (70%) controls the MSE divergence, and the contributions from evaporation (20%) and sensible heat flux (5%) are secondary. Specifically, dry air intrusion from northern latitudes appears as the single dominant element in initiating the dryness over South Asia. A further examination of the individual terms of the moisture advection equation (Fig. 9) shows that the advection of climatological air of lower moisture content by the anomalous wind is the principal reason for initiating the dryness during spring. A closer examination of daily evolution confirms that the anticyclonic vorticity over the northern Indian Ocean develops about 20–25 days earlier than the organization of negative rainfall anomalies over South Asia (Fig. 10). Similar dry advection dominates the MSE budget in cases where severe weak monsoon years are associated with equatorial western Pacific warm anomalies (not shown). The solutions from TPO_COLD (i.e., severe strong monsoons associated with developing La Niña) are mirror images to those obtained from TPO_WARM. In this case, moist advection from the equatorial region preconditions the monsoon atmosphere toward a severe strong monsoon year. In CM2.1 budgets, the pathways through which moist advection initiates wetness (Fig. 13) are different from the forced runs.

In the severe weak monsoon years that occurred with regional SST anomalies in the equatorial western Pacific, SST anomalies attain 1.0 standard deviation magnitude only by July, which may not be sufficient enough to produce a precursor during the spring season (not shown). Yet, the total SST in both experiments (TPO_WARM and EQWPAC_WARM) has a maximum over the central Pacific. Therefore, the rainfall and circulation response are similar in both cases over the equatorial Pacific (Figs. 6a and 14c). We suspect that this may be the reason for identical response over the monsoon region.

The identified mechanism is summarized in Fig. 15. Briefly, during late spring El Niño–related Pacific rainfall anomalies force equatorial waves. The easterly wind anomalies over the equatorial Indian Ocean, due to Ekman pumping, promote an anticyclonic vorticity over the northern Indian Ocean at low levels. The northwesterly winds at the poleward flank of this anticyclone advect air of lower moisture content from northern latitudes over South Asia. Advection of dry air impacts the convective heating and precipitation anomalies several times larger due to feedbacks.

Fig. 15.

A schematic illustrating the moist teleconnection from the equatorial Pacific to the South Asian monsoon region, particularly during the developing phase of El Niño associated with severe weak monsoon years. Diagnostics from CM2.1 and AM2.1 solutions show that during the developing phase of El Niño, pronounced SST warming over the equatorial east-central Pacific promotes enhanced rainfall that in turn forces equatorial Kelvin and Rossby waves. The concentrated easterly anomalies along the equatorial Indian Ocean are interpreted as due partly to Kelvin waves and partly to Rossby waves. These easterly anomalies, due to Ekman pumping, force an anticyclonic vorticity at low levels over the northern Indian Ocean. The northwesterly component of this anticyclone at its poleward flank advect air of lower moisture content from the north, leading to dryness over South Asia and subsequently a reduction in monsoon rainfall. Dry advection is initiated during late spring and continues into the following summer. Shaded region represents positive rainfall anomalies; hatched regions represent negative rainfall anomalies.

Fig. 15.

A schematic illustrating the moist teleconnection from the equatorial Pacific to the South Asian monsoon region, particularly during the developing phase of El Niño associated with severe weak monsoon years. Diagnostics from CM2.1 and AM2.1 solutions show that during the developing phase of El Niño, pronounced SST warming over the equatorial east-central Pacific promotes enhanced rainfall that in turn forces equatorial Kelvin and Rossby waves. The concentrated easterly anomalies along the equatorial Indian Ocean are interpreted as due partly to Kelvin waves and partly to Rossby waves. These easterly anomalies, due to Ekman pumping, force an anticyclonic vorticity at low levels over the northern Indian Ocean. The northwesterly component of this anticyclone at its poleward flank advect air of lower moisture content from the north, leading to dryness over South Asia and subsequently a reduction in monsoon rainfall. Dry advection is initiated during late spring and continues into the following summer. Shaded region represents positive rainfall anomalies; hatched regions represent negative rainfall anomalies.

b. Discussion

A close monitoring of the development of ENSO (McPhaden et al. 2008) led to better understanding, modeling, and successful prediction of many aspects of ENSO by CGCMs (Saha et al. 2006; Kirtman 2003). While most models capture the perturbations to large-scale circulations during the developing phase of ENSO (e.g., Ju and Slingo 1995), regional variations in moisture fields that depend on the moist convective parameterization schemes are indeed difficult to quantify, particularly over the ASM region (e.g., Annamalai et al. 2007). This problem is exacerbated by the presence of multiple regional heat sources and their interaction over the ASM (Annamalai and Sperber 2005; Annamalai 2010).

The budget diagnostics performed here identifies the role of dry (moist) advection as a precursor signal during El Niño (La Niña). In a recent modeling study, Annamalai et al. (2011, manuscript submitted to Nat. Geosci.) noted the role of dry air intrusion in causing the long-term drying trend over South Asia. The dominance of the term warrants that models need to capture the details in the climatological moisture distribution before and during the monsoon, apart from the details in the location and intensity of diabatic heating anomalies along the equatorial Pacific (Turner et al. 2005; Annamalai et al. 2007). The AGCM used here also has some limitations in the regional distribution of monsoon precipitation climatology (Fig. 4). Specifically, in contrast to observations the simulated rainfall has a local maximum over the Arabian Sea (Fig. 4f), perhaps leading to more anomalous precipitation there in all the sensitivity experiments (e.g., Figs. 5 and 6). Therefore, adjoining oceanic regions are included in the budget diagnostics. Hence, experiments with models that have a more realistic rainfall basic state may confirm our hypothesis. The unavailability of sufficiently long-term moisture and other flux datasets over the monsoon region limits the verification of the mechanism in real-time observation. A direct implication of our research is that observational efforts are necessary to monitor the three-dimensional moisture distribution over the ASM region that would aid in better understanding, modeling, and predicting severe monsoons. One caveat is that AM2.1 uses sigma coordinates and the output is interpolated to standard pressure levels, and MSE can be sensitive to interpolations. In addition, the numerical schemes employed for advection may be different in the original model codes than the ones employed by us. In the future we plan to compute the budget quantities in the native model coordinates. Our future research will focus on the response of the ASM to the decaying phase of ENSO and will also identify mechanisms of severe monsoons unrelated to boundary forcing.

Acknowledgments

The authors express sincere gratitude to Jan Hafner for his help in setting up the AM2.1 experiments and providing the codes to estimate the budget diagnostics. Comments from anonymous reviewers are acknowledged. The authors also acknowledge National Energy Research Scientific Computing Centre (NERSC), which is supported by Department of Energy, for the computational resources. This work is supported by the Office of Science (BER) U.S. Department of Energy, Grant DEFG02-07ER6445, and also by the three institutional grants (JAMSTEC, NOAA, and NASA) of the IPRC.

REFERENCES

REFERENCES
AchutaRao
,
K.
, and
K. R.
Sperber
,
2006
:
ENSO simulation in coupled ocean–atmosphere models: Are the current models better?
Climate Dyn.
,
27
,
1
15
.
Anderson
,
J. L.
, and
Coauthors
,
2004
:
The new GFDL global atmosphere and land model AM2-LM2: Evaluation with prescribed SST simulations
.
J. Climate
,
17
,
4641
4673
.
Annamalai
,
H.
,
2010
:
Moist dynamical linkage between the equatorial Indian Ocean and the South Asian monsoon trough
.
J. Atmos. Sci.
,
67
,
589
610
.
Annamalai
,
H.
, and
P.
Liu
,
2005
:
Response of the Asian summer monsoon to changes in El Niño properties
.
Quart. J. Roy. Meteor. Soc.
,
131
,
805
831
.
Annamalai
,
H.
, and
K. R.
Sperber
,
2005
:
Regional heat sources and the active and break phases of boreal summer intraseasonal (30–50 day) variability
.
J. Atmos. Sci.
,
62
,
2726
2748
.
Annamalai
,
H.
,
J. M.
Slingo
,
K. R.
Sperber
, and
K.
Hodges
,
1999
:
The mean evolution and variability of the Asian summer monsoon: Comparison of ECMWF and NCEP–NCAR reanalyses
.
Mon. Wea. Rev.
,
127
,
1157
1186
.
Annamalai
,
H.
,
K.
Hamilton
, and
K. R.
Sperber
,
2007
:
The South Asian summer monsoon and its relationship with ENSO in the IPCC AR4 simulations
.
J. Climate
,
20
,
1071
1092
.
Annamalai
,
H.
,
S.
Kida
, and
J.
Hafner
,
2010
:
Potential impact of tropical Indian Ocean–Indonesian Seas on El Niño characteristics
.
J. Climate
,
23
,
3933
3952
.
Chandrasekar
,
A.
, and
A.
Kitoh
,
1998
:
Impact of localized sea surface temperature anomalies over the equatorial Indian Ocean on the Indian summer monsoon
.
J. Meteor. Soc. Japan
,
76
,
841
853
.
Charney
,
J. G.
, and
J.
Shukla
,
1981
:
Predictability of monsoons
.
Monsoon Dynamics, J. Lighthill and R. P. Pearce, Eds., Cambridge University Press, 99–109
.
Delworth
,
T. L.
, and
Coauthors
,
2006
:
GFDL’s CM2 global coupled climate models. Part I: Formulation and simulation characteristics
.
J. Climate
,
19
,
643
674
.
Goswami
,
B. N.
,
1998
:
Interannual variations of Indian summer monsoon in a GCM: External conditions versus internal feedbacks
.
J. Climate
,
11
,
501
522
.
Griffies
,
S. M.
,
M. J.
Harrison
,
R. C.
Pacanowski
, and
A. R.
Rosati
,
2003
:
A technical guide to MOM4
.
GFDL Ocean Group Tech. Rep. 5, 295 pp
.
Joseph
,
R.
, and
S.
Nigam
,
2006
:
ENSO evolution and teleconnection in IPCC’s twentieth-century climate simulations: Realistic representation?
J. Climate
,
19
,
4360
4377
.
Ju
,
J.
, and
J. M.
Slingo
,
1995
:
The Asian summer monsoon and ENSO
.
Quart. J. Roy. Meteor. Soc.
,
121
,
1133
1168
.
Kalnay
,
E.
, and
Coauthors
,
1996
:
The NCEP/NCAR 40-Year Reanalysis Project
.
Bull. Amer. Meteor. Soc.
,
77
,
437
471
.
Keshavamurty
,
R. N.
,
1982
:
Response of the atmosphere to sea surface temperature anomalies over the equatorial Pacific and the teleconnections of the Southern Oscillation
.
J. Atmos. Sci.
,
39
,
1241
1259
.
Kirtman
,
B. P.
,
2003
:
The COLA anomaly coupled model: Ensemble ENSO prediction
.
Mon. Wea. Rev.
,
131
,
2324
2341
.
Lau
,
N. C.
, and
M. J.
Nath
,
2000
:
Impact of ENSO on the variability of the Asian–Australian monsoon as simulated in GCM experiments
.
J. Climate
,
13
,
4287
4309
.
Lin
,
S. J.
,
2004
:
A “vertically Lagrangian” finite-volume dynamical core for global models
.
Mon. Wea. Rev.
,
132
,
2293
2307
.
McPhaden
,
M. J.
,
2008
:
Evolution of the 2006–2007 El Niño: The role of intraseasonal to interannual time scale dynamics
.
Adv. Geosci.
,
14
,
219
230
.
Meehl
,
G. A.
,
1997
:
The South Asian monsoon and the tropospheric biennial oscillation
.
J. Climate
,
10
,
1921
1943
.
Meehl
,
G. A.
, and
J. M.
Arblaster
,
1998
:
The Asian–Australian monsoon and El Niño–Southern Oscillation in the NCAR climate system model
.
J. Climate
,
11
,
1356
1385
.
Meehl
,
G. A.
, and
J. M.
Arblaster
,
2002
:
Indian monsoon GCM sensitivity experiments testing tropospheric biennial oscillation transition conditions
.
J. Climate
,
15
,
923
944
.
Meehl
,
G. A.
,
C.
Covey
,
K. E.
Taylor
,
T.
Delworth
,
R. J.
Stouffer
,
M.
Latif
,
B.
McAvaney
, and
J. F. B.
Mitchell
,
2007
:
The WCRP CMIP3 multimodel dataset: A new era in climate change research
.
Bull. Amer. Meteor. Soc.
,
88
,
1383
1394
.
Neelin
,
J. D.
, and
H.
Su
,
2005
:
Moist teleconnection mechanisms for the tropical South American and Atlantic sector
.
J. Climate
,
18
,
3928
3950
.
Normand
,
C.
,
1953
:
Monsoon seasonal forecasting
.
Quart. J. Roy. Meteor. Soc.
,
79
,
463
473
.
Palmer
,
T. N.
,
C.
Brankovic
,
P.
Viterbo
, and
M. J.
Miller
,
1992
:
Modeling interannual variations of summer monsoon
.
J. Climate
,
5
,
399
417
.
Parthasarathy
,
B.
,
A. A.
Munot
, and
D. R.
Kothawale
,
1994
:
All-India monthly and seasonal rainfall series: 1871–1993
.
Theor. Appl. Climatol.
,
49
,
217
224
.
Rasmusson
,
E. M.
, and
T. H.
Carpenter
,
1983
:
The relationship between eastern equatorial Pacific sea surface temperature and surface wind fields associated with the Southern Oscillation/El Niño
.
Mon. Wea. Rev.
,
111
,
517
528
.
Rayner
,
N. A.
,
D. E.
Parker
,
E. B.
Horton
,
C. K.
Folland
,
L. V.
Alexander
,
D. P.
Rowell
,
E. C.
Kent
, and
A.
Kaplan
,
2003
:
Globally complete analyses of sea surface temperature, sea ice and night marine air temperature, 1871–2000
.
J. Geophys. Res.
,
108
,
4407
,
doi:10.1029/2002JD002670
.
Saha
,
S.
, and
Coauthors
,
2006
:
The NCEP climate forecasting system
.
J. Climate
,
19
,
3483
3517
.
Shukla
,
J.
,
1975
:
Effect of Arabian Sea surface temperature anomaly on Indian summer monsoon: A numerical experiment with the GFDL model
.
J. Atmos. Sci.
,
32
,
503
511
.
Shukla
,
J.
,
1984
:
Predictability of time averages
.
Part II: The influence of boundary forcing. Problems and Prospects in Long and Medium-Range Forecasting, D. M. Burridge and E. Kallen, Eds., Springer-Verlag, 155–206
.
Shukla
,
J.
, and
D. A.
Paolino
,
1983
:
The Southern Oscillation and long range forecasting of the summer monsoon rainfall over India
.
Mon. Wea. Rev.
,
111
,
1830
1837
.
Sikka
,
D. R.
,
1999
:
Monsoon drought over India
.
Joint COLA/COARE Tech. Rep. 2, 270 pp. [Available from Center for Ocean–Land–Atmosphere Studies, 4041 Powder Mill Road, Suite 302, Calverton, MD 20705-3106.]
Soman
,
M. K.
, and
J. M.
Slingo
,
1997
:
Sensitivity of the Asian summer monsoon to aspects of sea surface temperature anomalies in the tropical Pacific Ocean
.
Quart. J. Roy. Meteor. Soc.
,
123
,
309
336
.
Sperber
,
K. R.
, and
T. N.
Palmer
,
1996
:
Interannual tropical rainfall variability in general circulation model simulations associated with the atmospheric model simulations associated with the Atmospheric Model Intercomparison Project
.
J. Climate
,
9
,
2727
2750
.
Su
,
H.
, and
J. D.
Neelin
,
2002
:
Teleconnection mechanism for tropical Pacific descent anomalies during El Niño
.
J. Atmos. Sci.
,
59
,
2694
2712
.
Turner
,
A. G.
,
P. M.
Inness
, and
J. M.
Slingo
,
2005
:
The role of basic state in the ENSO–monsoon relationship and implications for predictability
.
Quart. J. Roy. Meteor. Soc.
,
131
,
781
804
.
Webster
,
P. J.
, and
S.
Yang
,
1992
:
Monsoon and ENSO: Selectively interactive systems
.
Quart. J. Roy. Meteor. Soc.
,
118
,
877
926
.
Wittenberg
,
A. T.
,
A.
Rosati
,
N. C.
Lau
, and
J. J.
Ploshay
,
2006
:
GFDL’s CM2 global coupled climate models. Part III: Tropical Pacific climate and ENSO
.
J. Climate
,
19
,
698
722
.
Wu
,
R.
, and
B. P.
Kirtman
,
2004
:
Impact of the Indian Ocean on the Indian summer monsoon–ENSO relationship
.
J. Climate
,
17
,
3037
3054
.
Xie
,
P.
, and
P.
Arkin
,
1996
:
Analysis of global monthly precipitation using gauge observations, satellite estimates, and numerical model predictions
.
J. Climate
,
9
,
840
858
.
Yang
,
S.
, and
K.-M.
Lau
,
1998
:
Influences of sea surface temperature and ground wetness on Asian summer monsoon
.
J. Climate
,
11
,
3230
3246
.

Footnotes

*

International Pacific Research Center Publication Number 807.