1. Introduction
The seasonal mean rainfall associated with the Asian summer monsoon (ASM) dictates the livelihood of the world’s most populous countries. Given their anticipated population rise, the countries influenced by the ASM will surely face increased stress in the near future, which will seriously impact the demand for drinking water, among other factors. For a variety of reasons, it is desirable to know whether the mean monsoon precipitation will increase or decrease as climate changes and to understand its implication on the intensity of monsoon depressions, the main rain-bearing systems over the plains of central India.
Climate models’ predictions or projections of the magnitude and spatial distribution of mean monsoon precipitation crucially depend on the parameterization of poorly understood small-scale processes that determine the nature of climate feedbacks and changes in general circulation. Furthermore, the sensitivity of a model’s response to any forcing depends on its fidelity in simulating the basic state in the control experiments (Shukla 1984; Fennessy et al. 1994; Sperber and Palmer 1996; Annamalai and Liu 2005; Turner et al. 2005). Therefore, it is not surprising to note diverse results among past modeling studies that examined the ASM response in experiments where the present-day concentrations of greenhouse gases were doubled (Meehl and Washington 1993; Mahfouf et al. 1994; Kitoh et al. 1997; Hu et al. 2000; May 2002; Ashrit et al. 2005). To gain confidence in the future monsoon projections, it is prudent to examine the runs in those models that demonstrate skill in simulating the current monsoon climate.
The availability of multicentury, multiple realizations with numerous contemporary global coupled models conducted recently for the Intergovernmental Panel on Climate Change (IPCC) Fourth Assessment Report (AR4) exercise provides an opportunity for a systematic analysis of the current and future monsoon climates. Annamalai et al. (2007) investigated the ability of the models in capturing the mean monsoon and its relationship with El Niño–Southern Oscillation (ENSO), both in the twentieth century (20c3m) and in the “1pctto2x” experiments, which impose a 1% year−1 CO2 increase to doubling of present-day concentration (2xCO2). As a prerequisite, in the 20c3m integrations, they examined the seasonal average (June–September) precipitation climatology constructed from the last 30 yr (1971–2000) for each of the 18 coupled models, both over India (7°–30°N, 65°–95°E) and for the larger monsoon domain (25°S–40°N, 40°E–180°). The authors estimated metrics such as pattern correlation and root-mean-square differences (RMSDs) relative to observed precipitation for the period 1979–2003. They found that only 6 of the 18 models show larger pattern correlation and smaller RMSD with observations; that is, only one-third of the models depicted a reasonable representation of the current monsoon climate. In the 2xCO2 experiments, despite the usage of different numerics and physical parameterizations in these six models, all of them predicted an increase (by 5%–25%) in the seasonal mean rainfall over India, implying qualitative agreement in the response. Although they all predicted an increase in land–sea thermal contrast over the ASM region, Annamalai et al. (2007) noted a lack of consistency in the predicted spatial distribution of precipitation over the larger monsoon region (their Fig. 7). Determining the reasons for differences among models is a very complicated task given the myriad of parameterizations incorporated into each model. Therefore, no attempt was made in their study to identify the reasons for the rainfall increase over India.
At subseasonal time scales, synoptic systems such as monsoon depressions having a horizontal scale of about 2000–3000 km form over the quasi-stationary seasonal monsoon trough, and they are, by far, the most important components of the monsoon circulation. During the peak monsoon season (July–August), a majority of them form over the warm waters of northern Bay of Bengal, and move in a west-northwesterly direction with an average phase of about 3 m s−1, and on average 4–6 systems form each year. Salient observational features include maximum moisture convergence at 850 hPa over the southwest quadrant and the vertical extent can reach up to 300 hPa. In regard to the thermal field, a cold core is prevalent in the lower troposphere while a warm core is observed at 300 hPa. Their observational characteristics are well documented and the readers are referred to Sikka (2006). In most places over central India, the rainfall associated with the depressions contributes to about 50% of the seasonal mean, and almost all extreme rainfall events there are attested to depressions (Dhar and Nandargi 1995; Sikka 2000). Therefore, to assess the future changes in the expected number of flood days, it is important to examine climate models’ projections in the depressions’ strength. Recently, from observations, Goswami et al. (2006) found an increasing trend in the number of extreme rainfall events since 1976 over central India. Given the fact that the number of observed monsoon depressions shows a sharp declining trend since 1976 (cf. Fig. 1), it is unclear what processes led to the positive trend in extreme rainfall events noted in Goswami et al. (2006).
Despite the extensive knowledge gained from observations, only few studies have evaluated the representation of depressions in climate models, partly because of resolution constraints and/or because of model systematic errors. In both the National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) reanalysis and the 15-yr European Centre for Medium-Range Weather Forecasts (ECMWF) Re-Analysis (ERA-15), Annamalai et al. (1999) noted that relevant statistics such as growth/decay rate and genesis/lysis locations are in good agreement between the two reanalyses although the intensity is higher in ERA-15 where the reanalyses is performed at a higher (T106) spatial resolution. Sabre et al. (2000) examined the Laboratoire de Météorologie Dynamique (LMD) atmospheric general circulation model and found that the simulated number of disturbances is too low, and their movement is more often toward the foothills of the Himalayas. Bengtsson et al. (2006) found that the simulated depressions’ intensity in ECHAM5 model is stronger than those in the 40-yr ECMWF Re-Analysis (ERA-40). Extending their analysis with ECHAM5 coupled model simulations, the authors note an increase in the depressions’ intensity in the twenty-first century compared to twentieth-century integrations, a result that needs to be verified by other models.
Observational and theoretical studies show that the mean conditions play necessary, although by no means sufficient, conditions for the formation and growth of monsoon depressions. As mentioned earlier, usually depressions develop over the central-northern Bay of Bengal, a region where the meridional gradient of potential vorticity vanishes (Shukla 1978), one of the necessary conditions for the zonal jet instability mechanism of Charney and Stern (1962) to operate. Employing quasigeostrophic models, the individual and combined effects of conditional instability of the second kind (CISK) and barotropic and baroclinic instabilities on the genesis and intensification of monsoon depressions have been studied (Shukla 1978; Moorthi and Arakawa 1985). Despite an increase in SST over the Bay of Bengal (Sikka 2006), observational records indicate for a decline in the number of depressions over the Bay of Bengal since 1976 (Xavier and Joseph 2000), and various factors are attributed to this trend that includes weakening of the low-level westerly flow over the Arabian Sea (Joseph and Simon 2005) and decrease in the horizontal and vertical wind shears as well as in moisture and convection over the Bay of Bengal (Mandke and Bhide 2003; Dash et al. 2004). Given these facts, the lack of monsoon depression simulations in the study of Sabre et al. (2000) can be due to systematic errors in the simulation of mean monsoon rainfall in their model (their Fig. 2a). In summary, as underscored by theoretical and observational studies, the “mean monsoon” conditions play vital role in the formation and amplification of monsoon depressions.
Of the six models studied in detail by Annamalai et al. (2007), the latest version of the Geophysical Fluid Dynamics Laboratory coupled model, version 2.1 (GFDL CM2.1, hereafter referred to as CM2.1), turned out to be the “best” (relative to other models), in terms of capturing both the mean monsoon and the details in the ENSO–monsoon relationship. In a recent study, Sperber and Annamalai (2008) find that the boreal summer intraseasonal variability is also realistically represented in this model. Motivated by the fact that the mean and the dominant modes of interannual and intraseasonal variations are well represented, in the present study, we concentrate on CM2.1 integrations and assess the ASM response to a quadrupling of CO2 forcing (1pctto4x or 4xCO2). Then, together with other models that showed reasonable skill in capturing mean monsoon precipitation (Annamalai et al. 2007), we point out the overall skill in CM2.1 in representing the monsoon depressions through employing a diagnostic tool exclusively developed to track the tropical synoptic systems (Stowasser et al. 2007). We propose a hypothesis that, in the 4xCO2 integrations, an increase in time-mean midtroposphere temperature and associated increase in moisture content provide additional potential for latent heat release during the buildup of low pressure systems, thereby possibly intensifying monsoon depressions. Since coarse-resolution global models are unable to capture the high-intensity rainfall amounts associated with mesoscale convective systems embedded in depressions, the International Pacific Research Center Regional Climate Model (IPRC RegCM) is integrated for 20 yr with lateral and lower boundary conditions taken from CM2.1.
The paper is organized as follows. In section 2, a brief description of the models, experiments performed, and the efficacy of the tracking method are presented. In section 3, a possible mechanism for the increase (decrease) in mean monsoon precipitation (circulation) in a warmer climate is provided. The statistics related to monsoon depressions are presented in section 4. In section 5, summary and implications of the present results are discussed.
2. Data, method, models, and experiments
a. Data
First, we examine the 20c3m output of the CM2.1 model [briefly described in section 2c(1)]. The 20c3m simulations attempt to replicate the overall climate variations during the period ∼1850–present by imposing the best estimates of natural (solar irradiance, volcanic aerosols) and anthropogenic (greenhouse gases, sulfate aerosols, ozone) climate forcing during this period. Then, we analyze the output from the “4xCO2” experiments which impose a 1%/year CO2 increase to quadrupling of initial concentration (∼140 yr) and then hold the CO2 constant for an additional ∼150 yr of simulation. We examine only the period in the run after the CO2 has been stabilized at 4 times the preindustrial day value. The time-mean climatology constructed is based on the last 30 yr of the simulations. Because of the limited length of daily output available from the 4xCO2 runs (∼20 yr), the statistics of monsoon depressions are estimated for the last 20 yr, 1981–2000 (271–290) in the 20c3m (4xCO2) integrations.
For model validation, we use the ECMWF global reanalysis data for the period 1958–2001 (ERA-40; Simmons and Gibson 2004), together with Climate Prediction Center Merged Analysis of Precipitation (CMAP) dataset of Xie and Arkin (1996), and the National Oceanic and Atmospheric Administration (NOAA) interpolated SST products of Reynolds and Smith (1994).
b. Tracking method
The method adopted for the detection and tracking of the model synoptic systems follows the one developed and described in Stowasser et al. (2007). The criteria used in their study to track tropical cyclones in the western North Pacific are relaxed to account for the weaker synoptic systems we are interested in the present study. The following criteria must be satisfied for a system that we identify and track:
A local vorticity maximum at 850 hPa exceeds 5 × 10−5 s−1 (Annamalai et al. 1999).
A local pressure minimum exists within a radius of 250 km of the vorticity maximum; this minimum pressure is taken as defining the center of the storm system. To be considered as a model storm trajectory, a storm must last at least 2 days.
As mentioned in the introduction, recent observational studies present evidence for a decreasing trend in the frequency of occurrence of monsoon depressions over the Bay of Bengal with a sharp drop in recent years bringing the average seasonal frequency down to 2–3 in the most recent decade of 1993–2002, from its long period average value 1976 onwards of 7–8 (Xavier and Jospeh 2000). To test our tracking method’s ability to reproduce this observed downward trend, the tracking algorithm was first applied to the ERA-40 data for the period 1958–2001. Figure 1 shows the yearwise number of identified systems and their 11-yr running mean (smoothed line) along with the observed number of depressions and deep depressions over the northern Indian Ocean.
The overall number of depressions is overestimated in the data deduced from the ERA-40 by about two storms per year. It should be mentioned here that ERA-40 is originally performed at a high resolution of T159 (∼1.0° resolution), and then linearly interpolated to 2.5° × 2.5°. Earlier studies suggest that reanalysis carried out at higher resolution (e.g., ERA-15) captures many storms compared to those performed at coarse resolutions (e.g., NCEP; Annamalai et al. 1999). Nevertheless, the diagnostic tool correctly reproduces the decreasing trend after 1980 that is seen in the observations, although there are differences in the total number of storms in a particular year. Since we are especially interested in the relative changes of the number of storms in a global warming case, we are confident that in a warmer climate any changes to the statistical properties of monsoon depressions detected by our technique deserve attention since a small difference in the number of storms identified by the tracking method compared to observations will not influence our findings. To elucidate the dependence of the results on the chosen threshold criteria we have performed sensitivity tests and results are provided in section 4d.
c. Models
1) CM2.1
A brief description of the CM2.1 coupled model is provided here, and details can be found in Delworth et al. (2006). The atmosphere and land models have a horizontal resolution of 2° latitude by 2.5° longitude, and the atmospheric model has 24 levels in the vertical. The physical packages employed are described in The GFDL Global Atmospheric Model Development Team (2004). The dynamical core uses finite-volume numerics (Lin 2004) that improved the surface wind stress pattern, and subsequently the climate drift after coupling. The land model incorporated is that of Milly and Shmakin (2002), which includes a river routing scheme. The advantage of this scheme is that the runoff accumulated over the model’s drainage basins is moved to the river mouths, where the freshwater is injected into the model ocean (Delworth et al. 2006). The ocean model is based on the Modular Ocean Model code (MOM4; Griffies et al. 2003), and has a resolution of 1° in latitude and longitude, with meridional resolution equatorward of 30° becoming progressively finer, such that the meridional resolution is 1/3° at the equator. In the vertical, there are 50 levels, with 22 levels of 10-m thickness each at the top 220 m (Gnanadesikan et al. 2006). The CM2.1 uses the Flexible Modeling System coupler for calculating and passing fluxes between its atmosphere, land, sea ice, and ocean components every 2 h.
Figure 2 shows the seasonal mean (June–September) precipitation climatology from observations (Fig. 2a) and 20c3m integrations (Fig. 2b). Also superimposed (in contours) is their respective SST climatology. Observations indicate that, over the ASM region, high precipitation [> 6 mm day−1 or outgoing longwave radiation (OLR) values < 220 W m−2, hereafter referred to as intense rainfall or deep convection] occurs over three regions that represent (i) the Indian summer monsoon (ISM; 10°–25°N, 70°–100°E), (ii) the western North Pacific monsoon (WNPM; 10°–20°N, 110°–150°E), and (iii) the equatorial Indian Ocean (EIO; 10°S–0°, 70°–100°E). As noted in Annamalai et al. (2007), the 20c3m run captures the three centers realistically (Fig. 2b), with some deficiencies outlined below. It is our premise that, in climate models, a reasonable representation of the three centers is important to adequately investigate the ASM variability at intraseasonal (Annamalai and Sperber 2005) and interannual time scales (Annamalai and Liu 2005; Turner et al. 2005), and also in determining the time-mean changes in climate change experiments as demonstrated in section 3a.
In the model (Fig. 3), the lower-troposphere monsoon circulation is characterized by a concentrated cross-equatorial flow with a prominent zonal jet whose core is situated around 10°N in the Arabian Sea (Fig. 3a), and the upwelling favorable near-surface meridional winds (Fig. 3b) cool the SST along the coast off Somalia (Fig. 2a). The upper-level flow (not shown) is dominated by a massive anticyclone with strong easterlies over the southern Arabian Sea. Not surprisingly, the easterly shear in the vertical (Fig. 3d), a measure of the first internal baroclinic mode forced by deep convection, also has its local maximum over the southern Arabian Sea. In general, the simulated circulation features are in good agreement with reanalysis (Annamalai et al. 1999). The moisture that maintains the monsoon convection stems from evaporation caused by stronger surface winds in the southern Indian Ocean (Fig. 3b), while the southern Arabian Sea and Bay of Bengal act as secondary sources (Fig. 3c).
Despite the model’s overall success in capturing the salient features of tropical climatology, systematic errors, both in spatial extent and intensity, still exist. In terms of precipitation, in particular over the three regional heat sources over the ASM region, one notices higher (lower) precipitation over the eastern equatorial Indian Ocean (Bay of Bengal and around the Philippines). In the eastern Pacific, the simulated cold tongue is less good, resulting in the split or double ITCZ, and one implication of this error results in a warmer SST over the Indonesian seas leading to excessive rainfall there (J. Slingo 2007, personal communication). Similarly, compared to reanalysis products, errors in the intensity of the circulation fields over the tropical Indian Ocean (not shown) are apparent. These model systematic errors need to be taken into account when interpreting the monsoon response to global warming.
2) Other models
Annamalai et al. (2007) also noted that both the Meteorological Research Institute Coupled General Circulation Model, version 2.3.2 (MRI CGCM2.3.2) and Max Planck Institute (MPI) ECHAM5 coupled models demonstrated skill in capturing the mean monsoon precipitation. Therefore, we have also examined the ability of these models in simulating synoptic systems, and compare the results with those obtained from CM2.1. The horizontal resolution of the atmospheric model used in MRI CGCM2.3.2 (MPI ECHAM5) is T42 (T63).
To support our diagnostics presented in section 3a, we perform idealized experiments with a linear baroclinic model (Watanabe and Jin 2003) that is employed in a variety of monsoon studies (Annamalai and Sperber 2005; Annamalai 2008, manuscript submitted to J. Atmos. Sci.). Here, we use the SST difference between the 4xCO2 and 20c3m climatologies (Fig. 4a) as the forcing, and linearize the model around the CM2.1 basic state to obtain steady-state solutions.
3) IPRC RegCM
The IPRC RegCM uses hydrostatic primitive equations in spherical coordinates with sigma (pressure normalized by surface pressure) as the vertical coordinate (Wang et al. 2003). The model equations are solved with a fourth-order conservative horizontal finite-difference scheme on an unstaggered longitude–latitude grid system. The time integration is performed using a leapfrog scheme with intermittent application of an Euler backward scheme. The model physics include the cloud microphysics scheme of Wang (2001); a mass flux scheme for subgrid shallow convection, midlevel convection, and deep convection developed by Tiedtke (1989) with some modifications outlined in Wang et al. (2003); the radiation package developed by Edwards and Slingo (1996) and further improved by Sun and Rikus (1999); the Biosphere–Atmosphere Transfer Scheme developed by Dickinson et al. (1993) for land surface processes; a modified Monin–Obukhov similarity scheme for flux calculations at the ocean surface; and a nonlocal E-ε turbulence closure scheme for subgrid-scale vertical mixing (Langland and Liou 1996), which was modified to include the effect of cloud buoyancy production of turbulence kinetic energy (Wang 1999). A one-way nesting is used to update the model time integration in a buffer zone near the lateral boundaries within which the model prognostic variables are nudged to the output of the CM2.1 simulations with an exponential nudging coefficient proposed by Giorgi et al. (1993) and later modified by Liang et al. (2001). The buffer zone in this experimental setup is 5° in extent. The IPRC RegCM has been used in a variety of monsoon (Sen et al. 2004) and tropical Pacific (Wang et al. 2004) climate studies.
d. Experiments
The version of the IPRC RegCM used in this study has 28 vertical levels with high resolution in the planetary boundary layer. The lowest model level is roughly 25 m above the surface. The model domain extends from 50°S to 55°N and 5° to 170°E with a grid spacing of 0.75°, in both zonal and meridional directions. Given the size of the domain and the central location of the South Asian monsoon within the domain, the simulation of the Indian monsoon in the regional model may be determined also by the internal climate of the RegCM rather than by the nudging from the CM2.1 at the lateral boundaries. Thus, to test the dependence of the results on the grid size of the nested RegCM, we repeated the experiment and nested a smaller domain RegCM within the CM2.1 and modified the grid resolution of the RegCM to determine the resolution dependence in a systematic fashion. Two experiments were performed additionally to the reference integration with the RegCM with the model domain extending from 15°S to 35°N and 40° to 140°E with a grid spacing of 0.5° and 1.0°, respectively. Employing the IPRC RegCM, we performed three 20-member ensemble control simulations in which the driving fields, including SST, are obtained from the 6-hourly output from the 20c3m integrations (RegCM_CTL). Further three perturbation experiments were carried out similar to the control experiments but the lateral boundary conditions taken from the 4xCO2 runs (RegCM_4CO2). The six experiments are summarized in Table 1.
A bilinear interpolation method was used to regrid the CM2.1 output to the regional climate model grid to obtain the necessary lateral boundary conditions. In both cases, an ensemble of 20 integrations was performed; each started from 1 June initial conditions taken from one of the 20 consecutive years. For instance, each year 1 June output from the 20 yr 1981–2000 (271–290) of 20c3m (4xCO2) runs are incorporated as initial conditions for the respective 20 integrations of the CTL (perturbation) experiment. Each integration lasts over the entire summer monsoon season from June through September (JJAS).
3. Response of the mean monsoon
In this section, we present the response of the model monsoon to global warming first in the CM2.1 (section 3a) and then in the IPRC RegCM (section 3b). In the CM2.1, the difference plots are obtained by subtracting the 20c3m climatology from that of the 4xCO2 integrations (4xCO2 minus 20c3m), while in the regional model it is the difference between the perturbation and CTL runs (RegCM_4CO2 minus RegCM_CTL). In the difference plots, statistical significance tests (t test) are applied to assess the signal-to-noise ratio.
a. Response in the CM2.1 integrations
Figure 2 reveals that, compared to the 20c3m integrations (Fig. 2b), in the 4xCO2 runs (Fig. 2c) the SST warming extends eastward and poleward, but the overall spatial patterns in both precipitation and SST and their locations of maxima and minima remain the same in both runs. In observations (Fig. 2a), and to a larger extent in the 20c3m integrations, intense rainfall occurs in regions where SST is greater than 28°C. In sharp contrast, in the 4xCO2 simulations regions where SSTs are greater than 31°C experience high rainfall implying that deep convection occurs in areas where SSTs are higher than in the surrounding regions, supporting the moist static energy budget analysis of Neelin and Held (1987). In summary, the rainfall pattern is collocated with the SST distribution. One notable deficiency of the model is a strong cold bias in the Arabian Sea (Fig. 2b).
Figure 4 shows the time-mean differences in SST (Fig. 4a), precipitation (Fig. 4b), and evaporation (Fig. 4c). The corresponding circulation changes are shown in Fig. 5. The SST rise in the equatorial central Pacific would rather suggest permanent El Niño–like conditions in a warmer climate. Notable features in the CM2.1 during typical El Niño events are anomalous cold SST and less rainfall than usual over the west Pacific but more rainfall and low-level anomalous westerlies over central and eastern Pacific (140°E to 120°W; Fig. 4a in Annamalai et al. 2007). The absence of these signatures in the quadrupled CO2 experiments negates the notion of permanent El Niño–like conditions (Knutson and Manabe 1994, 1995; Meehl and Arblaster 2003). Yet, the simulated intensity of El Niño in this coupled model is much stronger than those observed (see Fig. 2a in Annamalai et al. 2007).
Despite a uniform rise of over 2°C in SST everywhere in the tropical Indo-Pacific, the model shows significantly less rainfall over the equatorial Indian Ocean, and more rainfall over the WNPM region, the peninsular parts of India, and the equatorial west Pacific. In sharp contrast, at 850 hPa, the trades in the southern Indian Ocean and the cross-equatorial monsoon flow are weakened (Fig. 4b). This paradox between increased monsoon rainfall and weaker circulation has been noted previously in climate change experiments (e.g., Ueda et al. 2006), but the puzzling finding has not been explained to satisfaction. Here we provide a possible explanation based on 1) the equatorial wave response to regional heat sources/sinks within the ASM region, and 2) the associated changes in SST that contribute to significantly more evaporation.
First, we focus on the chain of events that leads to reduced precipitation over the equatorial Indian Ocean. Figure 4 reveals the presence of a local SST maximum over the equatorial west and central Pacific (130°E–150°W) that is conducive for a reduction in moist static stability and therefore more rainfall. Over this region, to balance the intensified low-level convergence (Fig. 5b), a deeper vertical circulation forms (Fig. 6a) that then promotes an increase in high and medium cloud amounts (Figs. 6b,c). The positive feedback between circulation and convection then maintains the rainfall over the equatorial west Pacific. The increase in troposphere water vapor (not shown) further enhances the surface warming leading to more evaporation (Ramanathan 1981). The above feedback processes cooperate in maintaining the projected features over the equatorial west Pacific. In the entire equatorial region, the vertical motion field shows the prominence of two deep cells of opposite signs (Fig. 6a), ascent over the warmest regions of the west Pacific and descent over the relatively colder eastern Indian Ocean, and the two regions being connected by an overturning mass circulation (Fig. 5a). The subsidence over the equatorial Indian Ocean is interpreted to result in a reduction in high and medium cloud amounts (Figs. 6b,c). This scenario prompts us to conclude that remote forcing determines the reduced precipitation over eastern equatorial Indian Ocean.
Another contributing factor for the projected rainfall changes is the mutual influences among the regional heat sources. Through conducting a series of sensitivity experiments with a linear atmospheric model, Annamalai and Sperber (2005) showed that, for a given negative precipitation anomaly over the equatorial Indian Ocean under steady-state conditions, the linear model simulates enhanced low-level convergence and rainfall over WNPM region and parts of India, primarily through equatorial wave dynamics (their Fig. 7). In another experiment in which warm SST anomalies are prescribed over the WNPM region for obtaining a balance in the thermodynamic energy equation, the rainfall increase over the WNPM region forces descending Rossby waves to the west, with centers over the equatorial Indian Ocean and over the Bay of Bengal extending into parts of northern India (Fig. 11 in Annamalai and Sperber 2005), a possible mechanism for the reduction in high and medium cloud amounts over both equatorial Indian Ocean and parts of northern India extending into Bay of Bengal (Figs. 6b,c).
To further confirm the above interpretation, we forced the linear model with SST differences over the tropical west Pacific (10°S–20°N, 120–180°E; Fig. 4a). The model-generated heating anomalies (Fig. 6e) promote positive heating/precipitation anomalies in the neighborhood of the imposed warm SST anomalies. As expected, negative heating anomalies over the Bay of Bengal extending into the plains of Indo-China and over the eastern equatorial Indian Ocean are reminiscent of descending Rossby waves. The vertical velocity field from the linear model solutions (Fig. 6d) also support the diagnostics presented in Fig. 6a but with a slight eastward shift in the centers of action. An examination of the temporal evolution from the linear model solutions (not shown) confirm the interactive nature between the heating and circulation variables over the tropical west Pacific region. In summary, as alluded in section 2c(1), for a meaningful interpretation for rainfall changes in a warmer climate it is necessary that models capture the three major convection centers over the ASM region in the basic state.
Now, let us examine how the precipitation changes influence the monsoon circulation and SST warming over southern Arabian Sea. The spatially well organized negative precipitation anomalies over the eastern equatorial Indian Ocean (Fig. 4b) forces a pair of anticyclones in the lower troposphere as a Rossby wave response with concentrated easterlies along the equatorial Indian Ocean, both at the surface (Fig. 5b) and at 850 hPa (Fig. 5c). The southern component of the anticyclone opposes and weakens the climatological cross-equatorial flow, thereby limiting the upwelling off Somalia. Second, the patch of surface easterly wind anomalies over the southern Arabian Sea (10°N–0°, 40°–70°E) opposes the climatological westerlies there (Fig. 3b), and is expected to deepen the thermocline. Both these factors reduce upwelling along Somalia. As a result, sea surface warms (Fig. 4a) and anomalous evaporation becomes a local maximum over the southern Arabian Sea (Fig. 4c). The interpretation offered here suggests that apart from direct and feedback processes associated with the greenhouse effect (Ramanathan 1981), changes in the atmospheric circulation also contribute to the surface warming over the southern Arabian Sea. In sharp contrast, the northern component of the low-level anticyclone strengthens the climatological westerlies in the southern Bay of Bengal and South China Sea.
Despite a weakened circulation, why does the simulated monsoon rainfall increase over parts of India? The generally promoted idea is that, in a warmer climate if the relative humidity of the surface air does not change, because of the nonlinearity embedded in the Clausius–Clapeyron equation surface evaporation will increase with increasing SST, and this additional moisture content in the troposphere results in excess rainfall. A dramatic increase in evaporation along the entire path of the climatological low-level monsoon flow (Fig. 4c) supports this idea. However, it is known that both SST and wind speed contribute to evaporation, and to quantify their relative contributions a linearized version of the bulk formula is used. A comparison of Figs. 4d,e undoubtedly confirms our interpretation that it is the change in SST that dominates the changes in evaporation over the ASM domain. This is further explored below.
While the relationship between wind speed and evaporation is broadly linear, the dependence of evaporation on SST is nonlinear (Zhang and McPhaden 1995). Figure 7 shows scatterplots between evaporation and SST for two key regions, the southern Indian Ocean and Arabian Sea where climatologically high evaporation occurs. While in both runs the nonlinear association between SST and evaporation exists, in the 4xCO2 runs, SST threshold for the occurrence of evaporation increases by about 3°–4°C, again consistent with the moist static analysis that depends on the SST distribution (Fig. 2c). In a warmer climate, in both regions the wind speed decreases while SST increases. Therefore, a possible interpretation is that SST warming alone leads to about 10%–15% enhancement in evaporation and atmospheric moisture over the southern Arabian Sea and ultimately leading to increase in medium cloud amount and rainfall over peninsular parts of India. In contrast, reduction in high and medium cloud amounts and associated northeasterly wind anomalies at 850 hPa (Fig. 5b) cause a rainfall reduction over central India and Bay of Bengal (Fig. 4b).
b. Response in the Regional Climate Model
Figure 8 shows the seasonal mean climatological patterns of precipitation and 850-hPa wind from the CTL integrations. The RegCM too captures the three precipitation centers over the ASM region, and the low-level wind climatology has all the prominent features but with a much stronger Somali jet whose center is shifted slightly equatorward. As regards to regional details, orographically forced rainfall maxima along the stretch of Western Ghats of India and along the Burmese mountains are realistically represented, and these narrow strips of intense rainfall are apparent in observations made at very high spatial resolution (Legates and Willmott 1990; see Fig. 3 in Annamalai et al. 1999). Another advantage of the regional model lies in capturing the southerlies in the convergence zone of South China Sea and west Pacific regions, and rainfall along the East Asian monsoon front. With regard to systematic errors in RegCM, the simulated rainfall over the Indian Subcontinent is weaker, and the excessive rainfall over the equatorial Indian Ocean is further intensified with multiple centers. This is expected since the lateral boundary conditions for the CTL experiments are taken from CM2.1 20c3m runs.
Figure 9 shows the differences (RegCM_4CO2 minus RegCM_CTL) in precipitation, low-level wind, and the lower-troposphere (850–500-hPa averaged) moisture. As in CM2.1, negative precipitation anomalies over the equatorial eastern Indian Ocean is associated with a pair of anticyclones in the low-level wind field leading to concentrated easterlies along the equatorial Indian Ocean. As expected, these changes result in the weakened trade winds in the southern Indian Ocean and also the cross-equatorial flow along the Somali coast. Despite a reduced monsoon circulation, the increase in lower-troposphere moisture content over the ASM region provides an environment favorable for stronger convection and rainfall. It should be noted here that Fig. 9b shows the moisture flux, but the wind patterns are qualitatively similar.
The salient features of the regional model solutions include 1) an intense cyclonic circulation to the north of the Philippines that is partly responsible for the increase in precipitation (and vice versa) in the western North Pacific region and 2) much stronger low-level moisture laden westerlies over the Bay of Bengal, which, under the influence of the orographic forcing, results in more precipitation along the western edges of the Burmese coast, extending into the plains of Indo-China and Bay of Bengal. Subsequently, rainfall and atmospheric heating increase is manifested in the formation of cyclonic vorticity over the head of the Bay of Bengal, a feature not present in the coarse-resolution CM2.1 (Fig. 5b). In contrast to CM2.1 rainfall projections, the regional model predicts more rainfall over the head of the Bay of Bengal. Consistent with the proposed hypothesis, in a warmer climate, the combined effects of increase in low-level westerlies (barotropic instability), troposphere moisture content, and precipitation over the central and northern Bay of Bengal are expected to aid in the growth and intensification of the synoptic systems.
4. Synoptic systems in the present and future climate
Both in the present day and in a warmer climate, three aspects of monsoon synoptic systems are investigated. First, using the tracking method their geographical distribution is examined (section 4a), followed by an assessment of their strength (section 4b), and finally a discussion on the changes in precipitation pattern due to changes in their characteristics is provided (section 4c). To the best of our knowledge, no past studies have examined the changes in monsoon synoptic systems’ characteristics in a warmer climate using a regional model.
a. Geographical distribution
To assess the genesis locations of monsoon synoptic systems, the tracking method is applied to ERA-40 products and output from selected IPCC models that showed skill in simulating the mean monsoon precipitation (Annamalai et al. 2007). Figure 10 shows the estimated number density for the recent 20 yr (1981–2000) in the reanalysis and in the last 20 yr of the respective model’s 20c3m simulations. The number of storms that are present in the domain and counting each 12-hourly time step is included in the estimation. The reanalysis confirms that most of the systems form north of 15°N in the head of the Bay of Bengal and move inland in a westward to west-northwestward direction. Over the Arabian Sea, our analysis also captures the relatively less frequent midtropospheric cyclones whose signatures are also seen at lower-troposphere levels (Sikka 2006). Overall, the results obtained from reanalysis here are consistent with those reported in Annamalai et al. (1999) using a different approach.
Although all three models represent the local maximum density over the Bay of Bengal, the west-northwestward movement into the land regions is best captured in CM2.1 runs (Fig. 10b) and so is the secondary maximum over the Arabian Sea. In the ECHAM5 coupled model studied by Bengtsson et al. (2006), systems making landfall are not represented (Fig. 10c). Compared to reanalysis, the genesis locations in CM2.1 solutions are shifted southwestward, and also the model fails to capture the systems along the southern slopes of the Himalayan Mountains.
In the RegCM_CTL solution (Fig. 11a), the broad features noted in the large-scale model are represented (Fig. 10b) but with some finer details that agree closely with the reanalysis results. They include (i) sharper inland excursion from the genesis regions, (ii) increase in number density along the Burmese topography, and (iii) meridionally aligned spatial distribution over the western Arabian Sea. Yet the number density in the RegCM is lesser than in the reanalysis, and there is no improvement in the synoptic systems’ simulation along the foothills of the Himalayas.
Figure 11 also shows the changes in the frequency of occurrences of storms in a warmer climate. In both model solutions, in the main genesis region over the Bay of Bengal, an increase (decrease) in the number of systems can be found to the north (south) of 18°. Compared to CM2.1 projections (Fig. 11d), RegCM solutions (Fig. 11c) suggest a three- to fourfold increase in the frequency. We attribute this difference to the differences in their projected basic state. In the RegCM (Fig. 9), over the Bay of Bengal north of 15°N, increase in low-level westerlies, troposphere moisture, and precipitation are particularly strong. In CM2.1, despite SST warming, dry northerlies and less rainfall prevail there (Figs. 4a,b and 5b). Similarly, in the RegCM, a slight decrease in the storm genesis in the central Bay of Bengal is possibly due to the presence of low-level anticyclonic flow to the south of 15°N (Fig. 9b).
b. Strength
In a warmer climate, will intense monsoon depressions become more frequent? To elucidate this, the frequency distribution of monsoon depressions as a function of maximum wind speed of the systems was calculated. The frequency shown in Fig. 12 is given in numbers of 12-h time steps; a synoptic system was present in the domain. The diagnostics is prepared separately for the main genesis location of the Bay of Bengal (18°–28°N, 75°–90°E), and also over a large region encompassing both the Bay of Bengal and the Arabian Sea sectors (30°N–0°, 60°–100°E).
The frequency distribution in CM2.1 20c3m simulations (blue bar) examined either over the larger domain (Fig. 12a) or only over the Bay of Bengal (Fig. 12b) exhibits asymmetric shapes with a peak around 12 m s−1, after which the distribution tails off rapidly. In the coarse-resolution CM2.1, few intense storms (wind speed > 15 m s−1) are simulated. For comparison, the frequency distribution in the ERA-40 data is also given in Fig. 12 (green bars) for the larger domain. The distribution looks very similar to CM2.1 result, with slightly more storms in the ERA-40 data.
Since the sharp-peaked distribution found in CM2.1 is typical for coarse-resolution models compared to observed distributions with broader peaks and heavier populated tails (Sugi et al. 2002), the RegCM_CTL solutions (blue bars in Figs. 12c) suggest a more realistic representation of the storm strength probability density function with a much broader maximum and significantly more storms with wind speeds exceeding 15 m s−1. We must stress here that, even in perfect simulations, a model employing high resolution will not produce observed intensities, and the peak surface winds are expected to be lower than observed (Walsh et al. 2007).
The frequency distribution in CM2.1 simulations with 4xCO2 concentration, either over the larger domain (red bars in Fig. 12a) or only over the Bay of Bengal (red bars in Fig. 12b) is similar to the corresponding 20c3m results showing only a marginal increase in the number of storms with wind speeds greater than 12 m s−1. On the other hand, we calculated the number of storm steps, 979 in CTL to 1031 in 4xCO2 runs, illustrating an increase in the total number of storms in a warmer climate.
Considering the Indian Ocean region between 60° and 100°E and between the equator and 30°N, the RegCM_4xCO2 experiment shows also little changes to the CTL simulations (Fig. 12c) although the increase in storm systems is more pronounced than in CM2.1 simulations. However, if the calculation is conducted with restriction to the Bay of Bengal (18°–28°N, 75°–90°E) the increase in number and intensity of the systems is much more pronounced (Fig. 12d).
c. Precipitation pattern
One notable impact of the depressions is that the river basins of Tapti and Narmada in central and western India are situated along their track, and hence receive copious rainfall (50%–100% above normal). Since these rivers flow in the same direction as the depressions’ movement, runoff results in floods that subsequently submerge villages and crop fields for several days (Sikka 2006). Having recognized that a greater number of intense storms are expected in a warmer climate (e.g., Webster et al. 2005) and there is a sharp increase in their westward excursion toward the Indian Subcontinent from the regional model projections, the next natural question to ask is: Which regions of India will experience more precipitation and hence are prone to more number of flood days? To address this question, we identify 12-hourly rainfall associated only with the storms both in the CTL and perturbation experiments of the RegCM. Figure 13 shows the difference between the CTL and 4xCO2 simulations in the spatial distribution of precipitation for days when synoptic systems were present in the domain. As expected, the increase in storm activity over the Bay of Bengal leads to a pronounced increase of rainfall over northern bay. Of significance here is the rainfall increase over central India, the coastal belts of West Bengal, and Orissa. One direct implication of this model result is that a greater number of flood days can be expected in the regions that lie along the depressions’ preferred track.
d. Sensitivity to threshold, resolutions, and domain
To test the dependency of the results on the choice of the vorticity threshold and the search radius given in section 2b, these parameters were varied and the analysis of the observed number of storms repeated. This was done for all six experiments given in Table 1 with resolutions of 0.5° × 0.5°, 0.75° × 0.75°, and 1.0° × 1.0°. An examination of the mean monsoon precipitation differences among the solutions reveal that convective rainfall increases over central and northern Bay of Bengal, while large-scale rainfall is increased along the Western Ghats of India (figure not shown). Irrespective of the choice of the domain or resolution, the robustness in capturing the mean monsoon changes imply that any changes in the projections of synoptic systems need not be significantly influenced by changes in domain or resolutions.
Figure 14a shows the dependency of the number of tracked storms on the employed vorticity threshold. It is obvious that the number of storms increases when the vorticity threshold is decreased and vice versa. However, the functional dependence is weaker around the threshold used for the analysis of the integrations (characterized by the filled symbols). The overall number of detected storms is slightly higher in the runs with the finer resolution.
Figure 14b shows that there is no functional dependence of the number of tracked storms on the distance of the vorticity maximum and the pressure minimum around the chosen 250-km radius.
Given the size of the domain and the central location of the South Asian monsoon within the domain, the simulation of the Indian monsoon in the regional model may be determined also by the internal climate of the RegCM rather than by the nudging from the CM2.1 at the lateral boundaries. The dependence of the results on the grid size and resolution of the nested RegCM is highlighted in Figs. 14c–f.
Figures 14c,d show the frequency distribution for the Bay of Bengal and Arabian Sea sector (30°N–0°, 60°–100°E). The results are quite similar in all three experiments both for the control and the 4xCO2 integrations. However, there is a slight shift to higher wind speeds in the integration with the highest resolution (0.5° × 0.5°) compared to the experiment conducted with the coarsest resolution (1.0° × 1.0°). Very similar results are obtained when the analysis is done for the main genesis location of the Bay of Bengal (Figs. 14e,f). These sensitivity analyses suggest that the resolution increase helps to reproduce slightly stronger storm events.
5. Summary and discussion
Here, first we summarize the major results with a schematic (section 5a), followed by a discussion focused on the current understanding of regional climate response to global warming, and close the section with some future plans (section 5b).
a. Summary
Diagnostics performed with the GFDL CM2.1 climate model that has the “best” representation of current mean monsoon climate among all the IPCC AR4 coupled models suggests that, in the global warming scenario, the mean monsoon rainfall over southern India and tropical western Pacific increases by 10%–20% while the cross-equatorial monsoon flow weakens. While previous studies have noted a similar paradox, we interpret the results within the framework of known equatorial dynamics and the major points are illustrated in a schematic (Fig. 15).
In the tropics, model projections indicate that the equatorial western Pacific experiences the highest warming of SST and hence more precipitation while the eastern equatorial Indian Ocean warms the least and also receive less rainfall (Fig. 4). Consistent with rainfall changes, a deeper circulation with ascent (descent) over the equatorial western Pacific (Indian Ocean) is projected. Subsequently, the reduced rainfall anomalies over the eastern equatorial Indian Ocean force twin anticyclones as a Rossby wave response (Figs. 5 and 9), and the southern component opposes and weakens the cross-equatorial monsoon flow. The patch of easterly anomalies (Fig. 5) centered in the southern Arabian Sea is expected to deepen the thermocline north of the equator. Both these factors limit the coastal upwelling along Somalia, resulting in local sea surface warming and eventually leading to a local maximum in evaporation over the southern Arabian Sea (Fig. 4). Our diagnostics imply that, in addition to the increased CO2-induced rise in temperature, evaporation, and atmospheric moisture, changes in local circulation further increases SST, evaporation, and atmospheric moisture, leading to increased rainfall over the peninsular parts of India. The dynamical linkage between the tropical west Pacific and the Indian monsoon region is also shown by performing idealized experiments with a linear atmospheric model (Fig. 6).
Owing to regional details in precipitation changes, the IPRC RegCM was integrated for 20 yr with lateral boundary conditions taken from CM2.1, both for 20c3m and 4xCO2 integrations. A series of integrations to examine the sensitivity of the regional model solutions to resolutions and changes in the domain are performed (Table 1). In all the solutions and in a warmer climate, the regional model solutions confirm the increase (decrease) in monsoon rainfall (circulation). In addition, the orographic nature in monsoon precipitation along the slopes of Burma and increased rainfall over northern Bay of Bengal and the associated circulation features are better captured in the regional model (Fig. 9).
Our results indicate that both CM2.1 and RegCM models simulate the genesis location and west-northwest movement of synoptic systems into the Indian Subcontinent (Figs. 10 and 11). Possibly because of a better representation of mesoscale convective activities and their interaction with dynamics, the RegCM solutions project a significant increase in the number of monsoon depressions in a warmer climate, in particular those with wind speeds of 15–20 m s−1 (Fig. 12) or greater depending on resolutions (Fig. 14). Based on this model results, a possible implication is that an increase in the number of flood days over central India can be expected in a CO2-richer climate. Since the physical package used in the RegCM differs significantly from that employed in CM2.1, it is unclear whether the higher resolution alone is responsible for the projected changes in the storm frequency or not. For instance, in the three coupled general circulation models examined here (Fig. 10), CM2.1 has a horizontal resolution (∼2° × 2.5°) that is coarser than the other two models; yet, the simulation of synoptic systems are relatively better in CM2.1. Therefore, efforts are underway to verify the present results with a higher-resolution CM2.1 model itself.
b. Discussion
It is accepted that the CO2-induced water vapor is the most significant greenhouse gas several times more powerful than CO2 itself at directly warming the surface. The largest contribution to the surface warming stems from positive feedback among H2O evaporation, latent heat release, and infrared emission (Ramanathan 1981). In this feedback loop, our diagnostics reveal an additional contribution from changes in local circulation features in enriching SST warming and evaporation over southern Arabian Sea. Needless to say, to quantify any future monsoon projections, climate models need to represent the Indian Ocean processes realistically, and observational efforts are needed to monitor all aspects of this tropical ocean.
Over the Bay of Bengal, while the coarse-resolution CM2.1 does not predict more rainfall, the IPRC RegCM integrated at various higher horizontal resolutions predicts a significant rainfall increase. In a CO2-richer climate, this rise in rainfall and associated latent heat release may have aided in the intensification of monsoon depressions, and therefore an increase in the number of flood days over central India can be anticipated. In addition, compared to CM2.1, the regional model predicts a three- to fourfold increase in the occurrence of synoptic systems over the northern Bay of Bengal (Fig. 11c). These results, however, do not guarantee that the unsuccessful simulation of precipitation over Bay of Bengal in CM2.1 is due to coarser horizontal resolution. In ongoing research, we note that the GFDL atmospheric general circulation model (AM2.1) forced by SST taken from CM2.1 captures much stronger monsoon precipitation climatology over the Bay of Bengal (not shown) than does the coupled model analyzed here (Fig. 2). Since our approach with the regional model is uncoupled, more systematic work is needed to pinpoint the relative roles of resolution, regional air–sea interaction, and physical packages implemented in the models.
Unfortunately, our confidence in climate change predictions on regional scales will likely remain low because of model systematic errors. For instance, in the GFDL model climatology, excess (deficient) precipitation over the equatorial Indian Ocean (Bay of Bengal extending into the Indian Subcontinent), and the severe cold tongue problem of the eastern equatorial Pacific may have influenced our results.
Our future works will address the following questions: (i) In a warmer climate, over the tropical Indo-Pacific region, what factors led to the relatively cold SST over the eastern equatorial Indian Ocean (EEIO), and how does this time-mean condition influence the future state of the Indian Ocean climate variability? (ii) In the RegCM solutions, why do the storms in a preferred window (wind speed of 15–20 m s−1) intensify in a warmer climate, and does the frequency of monsoon onset vortex increase because of an increase in SST and evaporation in the southern Arabian Sea? (iii) In both CM2.1 and RegCM solutions, how realistically are the three-dimensional characteristics of the storms simulated? (iv) How would the monsoon respond in experiments which also include the effect of black carbon aerosols, apart from quadrupling of CO2 concentrations?
Acknowledgments
The authors thank the international modeling groups for providing their data for analysis and the Program for Climate Model Diagnostics and Intercomparison (PCMDI) for collecting and archiving the model data. This work was supported by the Office of Science (BER) U.S. Department of Energy, Grant DE-FG02-07ER6445. This work was also supported by the three institutional grants (JAMSTEC, NOAA, and NASA) of the IPRC. The authors acknowledge the helpful comments of the reviewers, in particular Reviewer A whose suggestions led to sensitivity tests reported in section 4d. We also thank Dr. Yuqing Wang for suggestions on the use of the IPRC Regional Climate Model.
REFERENCES
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: South Asian summer monsoon and its relationship with ENSO in the IPCC-AR4 simulations. J. Climate, 20 , 1071–1092.
Ashrit, R. G., A. Kitoh, and S. Yukimoto, 2005: Transient response of ENSO-monsoon teleconnection in MRI-CGCM2.2 climate change simulations. J. Meteor. Soc. Japan, 83 , 273–291.
Bengtsson, L., K. Hodges, and R. Roeckner, 2006: Storms tracks and climate change. J. Climate, 19 , 3518–3543.
Charney, J. G., and M. E. Stern, 1962: On the stability of internal baroclinic jest in a rotating atmosphere. J. Atmos. Sci., 19 , 159–172.
Dash, S. K., J. R. Kumar, and M. S. Shekhar, 2004: On the decreasing frequency of monsoon depressions over the Indian region. Curr. Sci., 86 , 1404–1411.
Delworth, T. L., and Coauthors, 2006: GFDL’s CM2 global coupled climate models. Part I: Formulation and simulation characteristics. J. Climate, 19 , 643–674.
Dhar, O. N., and S. S. Nandargi, 1995: Some aspects of seasonal rainstorms of India. Theor. Appl. Climatol., 50 , 205–212.
Dickinson, R. E., A. Henderson-Sellers, and P. J. Kennedy, 1993: Biosphere-atmosphere transfer scheme (BATS) version 1e as coupled to the NCAR Community Climate Model. NCAR Tech. Note, NCAR/TN-387+STR, National Center for Atmospheric Research, Boulder, CO, 72 pp.
Edwards, J. M., and A. Slingo, 1996: Studies with a flexible new radiation code. I: Choosing a configuration for a large-scale model. Quart. J. Roy. Meteor. Soc., 122 , 689–719.
Fennessy, M. J., and Coauthors, 1994: The simulated Indian monsoon: A GCM sensitivity study. J. Climate, 7 , 33–43.
GFDL Global Atmospheric Model Development Team, 2004: The new GFDL global atmosphere and land model AM2–LM2: Evaluation with prescribed SST simulations. J. Climate, 17 , 4641–4673.
Giorgi, F., M. R. Marinucci, G. T. Gates, and G. DeCanio, 1993: Development of a second generation region climate model (RegCM2). Part II: Convective processes and assimilation of lateral boundary conditions. Mon. Wea. Rev., 121 , 2814–2832.
Gnanadesikan, A., and Coauthors, 2006: GFDL’s CM2 global coupled climate models. Part II: The baseline ocean simulation. J. Climate, 19 , 675–697.
Goswami, B. N., V. Venugopal, D. Sengupta, M. Madhusoodanan, and P. K. Xavier, 2006: Increasing trend of extreme rainfall events over India in a warming environment. Science, 314 , 1442–1445.
Griffies, S. M., M. J. Harrison, R. C. Pacanowski, and A. Rosati, 2003: A technical guide to MOM4. GFDL Ocean Group Tech. Rep. 5, NOAA/GFDL, Princeton, NJ, 295 pp.
Hu, Z. Z., M. Latif, E. Roeckner, and L. Bengtsson, 2000: Intensified Asian summer monsoon and its variability in a coupled model forced by increasing greenhouse gas concentrations. Geophys. Res. Lett., 27 , 2681–2684.
Joseph, P. V., and R. Simon, 2005: Weakening trend of monsoon low-level jet stream through India 1950 to 2003. CLIVAR Exchanges, No. 10, International CLIVAR Project Office, Southampton, United Kingdom, 27–30.
Kitoh, A., S. Yukimoto, A. Noda, and T. Motoi, 1997: Simulated changes in the Asian summer monsoon at times of increased atmospheric CO2. J. Meteor. Soc. Japan, 75 , 1019–1031.
Knutson, T. R., and S. Manabe, 1994: Impact of increased CO2 on simulated ENSO-like phenomena. Geophys. Res. Lett., 21 , 2295–2298.
Knutson, T. R., and S. Manabe, 1995: Time-mean response over the tropical Pacific to increased CO2 in a coupled ocean–atmosphere model. J. Climate, 8 , 2181–2199.
Langland, R. H., and C-S. Liou, 1996: Implementation of an E-e parameterization of vertical subgrid-scale mixing in a regional model. Mon. Wea. Rev., 124 , 905–918.
Legates, D. R., and C. J. Willmott, 1990: Mean, seasonal, and spatial variability in gauge corrected global precipitation. Int. J. Climatol., 10 , 111–127.
Liang, X-Z., K. E. Kunkel, and A. N. Samel, 2001: Development of a regional climate model for U.S. Midwest applications. Part I: Sensitivity to buffer zone treatment. J. Climate, 14 , 4363–4378.
Lin, S. J., 2004: A “vertically Lagrangian” finite-volume dynamical core for global models. Mon. Wea. Rev., 132 , 2293–2307.
Mahfouf, J. F., D. Cariolle, J. F. Royer, J. F. Geleyn, and B. Timbal, 1994: Response of the Météo-France climate model to changes in CO2 and SST. Climate Dyn., 9 , 345–362.
Mandke, S. K., and U. V. Bhide, 2003: A study of decreasing storm frequency over Bay of Bengal. J. Indian Geophys. Union, 7 , 53–58.
May, W., 2002: Simulated changes of the Indian summer monsoon under enhanced greenhouse gas conditions in a global time-slice experiment. Geophys. Res. Lett., 29 , 1118. doi:10.1029/2001GL013808.
Meehl, G. A., and W. M. Washington, 1993: South Asian summer monsoon variability in a model with doubled atmospheric carbon dioxide concentration. Science, 260 , 1101–1104.
Meehl, G. A., and J. M. Arblaster, 2003: Mechanisms for projected future changes in South Asian monsoon precipitation. Climate Dyn., 21 , 659–675.
Milly, P. C. D., and A. B. Shmakin, 2002: Global modeling of land water and energy balances. Part I: The land dynamics (LaD) model. J. Hydrometeor., 3 , 283–299.
Moorthi, S., and A. Arakawa, 1985: Baroclinic instability with cumulus heating. J. Atmos. Sci., 42 , 2007–2031.
Neelin, J. D., and I. M. Held, 1987: Modeling tropical convergence based on moist static energy budget. Mon. Wea. Rev., 115 , 3–12.
Ramanathan, V., 1981: The role of ocean–atmosphere interactions in the CO2 climate problem. J. Atmos. Sci., 38 , 918–930.
Reynolds, R. W., and T. M. Smith, 1994: Improved global sea surface temperature analyses using optimum interpolation. J. Climate, 7 , 929–948.
Sabre, M., K. Hodges, K. Laval, J. Polcher, and F. Desalmand, 2000: Simulation of monsoon disturbances in the LMD GCM. Mon. Wea. Rev., 128 , 3752–3771.
Sen, O. L., Y. Wang, and B. Wang, 2004: Impact of Indochina deforestation on the East Asian summer monsoon. J. Climate, 17 , 1366–1380.
Shukla, J., 1978: CISK, barotropic and baroclinic instability and the growth of monsoon depressions. J. Atmos. Sci., 35 , 495–500.
Shukla, J., 1984: Predictability of time averages: Part II. The influence of the boundary forcing. Problems and Prospects in Long and Medium Range Weather Forecasting, D. M. Burridge and E. Kallen, Eds., Springer-Verlag, 155–206.
Sikka, D. R., 2000: Monsoon floods. Joint COLA/CARE Rep. 4, 154 pp.
Sikka, D. R., 2006: A study on the monsoon low pressure systems over the Indian region and their relationship with drought and excess monsoon seasonal rainfall. COLA Rep. 217, 145 pp.
Simmons, A. J., and J. K. Gibson, 2000: The ERA-40 project plan. ERA-40 Project Rep. Series 1, European Centre for Medium-Range Weather Forecasts, Reading, United Kingdom, 63 pp.
Sperber, K. R., and T. N. Palmer, 1996: Interannual tropical rainfall variability in general circulation model simulations associated with the Atmospheric Model Intercomparison Project. J. Climate, 9 , 2727–2750.
Sperber, K. R., and H. Annamalai, 2008: Coupled model simulations of boreal summer intraseasonal (30–50 day) variability, Part I: Systematic errors and caution on use of metrics. Climate Dyn., 31 , 345–372. doi:10.1007/s00382-008-0367-9.
Stowasser, M., Y. Wang, and K. Hamilton, 2007: Tropical cyclone changes in the western North Pacific in a global warming scenario. J. Climate, 20 , 2378–2396.
Sugi, M., A. Noda, and N. Sato, 2002: Influence of the global warming on tropical cyclone climatology: An experiment with the JMA global model. J. Meteor. Soc. Japan, 80 , 249–272.
Sun, Z., and L. Rikus, 1999: Improved application of exponential sum fitting transmissions to inhomogeneous atmosphere. J. Geophys. Res., 104 , 6291–6303.
Tiedtke, M., 1989: A comprehensive mass flux scheme for cumulus parameterization in large-scale models. Mon. Wea. Rev., 117 , 1779–1800.
Turner, A. G., P. M. Ness, and J. M. Slingo, 2005: The role of the basic state in the ENSO-monsoon relationship and implications for predictability. Quart. J. Roy. Meteor. Soc., 131 , 781–804.
Ueda, H., A. Iwai, K. Kuwako, and M. E. Hori, 2006: Impact of anthropogenic forcing on the Asian summer monsoon as simulated by eight GCMs. Geophys. Res. Lett., 33 , L06703. doi:10.1029/2005GL025336.
Walsh, K. J. E., M. Fiorino, C. W. Landsea, and K. L. McInnes, 2007: Objectively determined resolution-dependent threshold criteria for the detection of tropical cyclones in climate models and reanalyses. J. Climate, 20 , 2307–2314.
Wang, Y., 1999: A triply nested movable mesh tropical cyclone model with explicit cloud microphysics—TCM3. BMRC Research Rep. 74, Bureau of Meteorology Research Centre, 81 pp.
Wang, Y., 2001: An explicit simulation of tropical cyclones with a triply nested movable mesh primitive equation model—TCM3 Part I: Model description and control experiment. Mon. Wea. Rev., 129 , 1370–1394.
Wang, Y., O. L. Sen, and B. Wang, 2003: A highly resolved regional climate model and its simulation of the 1998 severe precipitation events over China. Part I: Model description and verification of simulations. J. Climate, 16 , 1721–1738.
Wang, Y., S-P. Xie, H. Xu, and B. Wang, 2004: Regional model simulations of boundary layer clouds over the southeast Pacific off South America. Part I: Control experiment. Mon. Wea. Rev., 132 , 274–296.
Watanabe, M., and F-F. Jin, 2003: A moist linear baroclinic model: Coupled dynamical–convective response to El Niño. J. Climate, 16 , 1121–1139.
Webster, P. J., G. J. Holland, J. A. Curry, and H. R. Chang, 2005: Changes in tropical cyclone number, duration and intensity in a warming environment. Science, 309 , 1844–1846.
Wilks, D. S., 1995: Statistical Methods in the Atmospheric Sciences. Academic Press, 467 pp.
Xavier, P. K., and P. V. Joseph, 2000: Vertical wind shear in relation to frequency of monsoon depressions and tropical cyclones of Indian Seas. Proc. TROPMET-2000, National Symp. on Ocean and Atmosphere, Cochin, India, Indian Meteorological Society, 232–245.
Xie, P., and P. Arkin, 1996: Analyses of global monthly precipitation using gauge observations, satellite estimates, and numerical model predictions. J. Climate, 9 , 840–858.
Zhang, G. J., and M. J. McPhaden, 1995: The relationship between sea surface temperature and latent heat flux in the equatorial Pacific. J. Climate, 8 , 589–605.
Yearwise frequency of occurrences of storm activity over the Indian monsoon region during June–September for the period 1958–2001 derived from ERA-40 products. To illustrate the recent declining trend in storm activity, a 11-yr running mean (shown in thick black) of the number of storms is also plotted. Also shown (dotted lines) is the observed number of depressions and deep depressions over the northern Indian Ocean.
Citation: Journal of Climate 22, 4; 10.1175/2008JCLI2218.1
Seasonal average (JJAS) precipitation (color shading; mm day−1) and SST (contours, °C): (a) observations, (b) CM2.1 twentieth-century (20c3m) integrations, and (c) same as (b) but for quadrupling CO2 (4xCO2) integrations. The 28°C isotherm is shown in dotted lines.
Citation: Journal of Climate 22, 4; 10.1175/2008JCLI2218.1
Seasonal average (JJAS) climatology constructed from the GFDL twentieth-century integrations: (a) 850-hPa winds (m s−1), (b) 10-m winds (m s−1), (c) evaporation (mm), and (d) easterly shear (U200 minus U850). In (a) and (b) westerlies are shaded progressively and the reference vector is also shown.
Citation: Journal of Climate 22, 4; 10.1175/2008JCLI2218.1
Seasonal mean (JJAS) climatology differences between the 4xCO2 and 20c3m integrations: (a) SST (°C), (b) precipitation (mm day−1), (c) evaporation (mm day−1), (d) evaporation (mm day−1) due to changes in SST, and (e) evaporation (mm day−1) due to changes in surface winds. Only statistically significant values are shown.
Citation: Journal of Climate 22, 4; 10.1175/2008JCLI2218.1
Same as Fig. 4 but for (a) 925-hPa velocity potential (m2 s−1), (b) 10-m winds (m s−1), (c) 850-hPa wind (m s−1), and (d) vertical easterly shear. In (a) negative values are shown in contours with an interval of 0.5e06, and the reference wind vector in (b) is shown.
Citation: Journal of Climate 22, 4; 10.1175/2008JCLI2218.1
Same as Fig. 4 but for (a) vertical velocity (Pa s−1) averaged over 11°S–5°N, (b) high cloud amount (%), and (c) medium cloud amount (%). Also shown are steady-state solutions obtained from the linear baroclinic model for SST forcing (10°S–20°N, 120°–280°E), (d) vertical velocity averaged over 12°S–5°N, and (e) model-generated vertically integrated heating anomalies (K day−1). In (a)–(d), positive values are shaded progressively and negative values are shown as contours. The contour interval is (a) 0.005, (b) 3, (c) 2, and (d) 0.005. In (e), positive (negative) values are shown in thick (dotted) contours with an interval of 2 K day−1.
Citation: Journal of Climate 22, 4; 10.1175/2008JCLI2218.1
Scatterplot between SST and evaporation (mm): (a) Arabian Sea (5°–15°N, 55°–75°E), and (b) southern Indian Ocean (7°–22°S, 60°–100°E). The monthly means during JJAS averaged over the regions of interest for the last 30 yr of the respective integrations are binned together. The values for the 20c3m (4xCO2) runs are shown in open circles (closed squares).
Citation: Journal of Climate 22, 4; 10.1175/2008JCLI2218.1
Regional model climatology averaged for July–September and constructed from 20 yr of integration: (a) precipitation (mm day−1) and (b) 850-hPa wind (m s−1). In (b) westerlies are shaded progressively, and the reference vector is also shown.
Citation: Journal of Climate 22, 4; 10.1175/2008JCLI2218.1
Seasonal mean (JJAS) differences between RegCM_4xCO2 and RegCM_CTL experiments: (a) precipitation (mm day−1), (b) moisture flux at 850 hPa (m s−1), (c) specific humidity (g kg−1). In (a) positive values are shaded progressively while negative values are shown as contours with an interval of 0.5 mm day−1.
Citation: Journal of Climate 22, 4; 10.1175/2008JCLI2218.1
Occurrence of synoptic systems from (a) ERA-40 data (1981–2000), the suite of twentieth-century integrations of coupled models, (b) GFDL CM_2.1, (c) MPI, and (d) MRI. The last 20 yr of the 20c3m simulations (1981–2000) are used for the calculation. The units are numbers per 2.5° square box per 4-month (JJAS) period.
Citation: Journal of Climate 22, 4; 10.1175/2008JCLI2218.1
Occurrence of synoptic systems in the Indian Ocean region: (a) RegCM_CTL integrations, (b) RegCM_4CO2 runs, and (c) difference between the 4xCO2 experiment and the control experiment [(b) minus (a)]. In (d), the difference but from the CM2.1 simulations is shown. The units are numbers per 2.5° square box per 4-month (JJAS) period. In (c) and (d) the positive (negative) values are shaded progressively (in contours). Note that the scaling in (c) is twice compared to that shown in (d).
Citation: Journal of Climate 22, 4; 10.1175/2008JCLI2218.1
Frequency distribution as a function of maximum wind speed of each synoptic system estimated for (a), (c) the entire northern Indian Ocean, and (b), (d) only over the Bay of Bengal: (a) and (b) result from CM2.1; (c) and (d) result from RegCM. The blue bars indicate the control runs and the red bar the 4xCO2 integrations. The frequency is given in numbers of 12-h time steps; a system was present in the domain.
Citation: Journal of Climate 22, 4; 10.1175/2008JCLI2218.1
Difference between (RegCM_4xCO2 − RegCM_CTL) simulations in the spatial distribution of precipitation (mm day−1) for days when synoptic systems were present. Positive values are shaded progressively, while negative values are shown in contours with an interval of 1 mm day−1.
Citation: Journal of Climate 22, 4; 10.1175/2008JCLI2218.1
(a) Dependency of the detected number of storms on the vorticity threshold for three different resolutions of the RegCM model (control integrations), and (b) dependency of the detected number of storms on the search radius distance between vorticity maximum and closed pressure minimum for the 0.75° × 0.75° RegCM control integration. (c)–(f) Frequency distribution as a function of maximum wind speed of each synoptic system estimated for (c), (d) the entire Indian Ocean, and (e), (f) only over the Bay of Bengal for the control and 4xCO2 integrations, respectively. All distributions are shown for the three employed resolutions of 0.5°x0.5° (red bars), 0.75°x0.75°(green bars), and 1.0° × 1.0° (blue bars), respectively.
Citation: Journal of Climate 22, 4; 10.1175/2008JCLI2218.1
A schematic illustrating the response of the mean monsoon in a global warming scenario, as projected by the CM2.1 coupled model integrations. The diagnostics show that the pronounced SST warming over the equatorial western Pacific region promotes enhanced rainfall and deeper ascent, while the relatively colder SST over the equatorial eastern Indian Ocean favors descent and reduced rainfall. The negative precipitation anomalies over the equatorial Indian Ocean force twin anticyclones in the lower troposphere as a Rossby wave response. The southern component opposes the climatological monsoon flow and weakens the upwelling off Somalia resulting in a local SST warming and enhanced evaporation over the Arabian Sea. That is, in addition to the increased CO2-induced rise in temperature, evaporation, and atmospheric moisture, the local circulation changes, further increasing SST, evaporation, and atmospheric moisture leading to increased rainfall over parts of India.
Citation: Journal of Climate 22, 4; 10.1175/2008JCLI2218.1
Model domain size and horizontal resolutions of the performed experiments with the IPRC RegCM.
* International Pacific Research Center Contribution Number 545 and School of Ocean and Earth Science and Technology Contribution Number 7581.
