• Adler, R. F., and Coauthors, 2003: The version 2 Global Precipitation Climatology Project (GPCP) monthly precipitation analysis (1979–present). J. Hydometeor., 4 , 11471167.

    • Search Google Scholar
    • Export Citation
  • Albrecht, B., 1989: Aerosols, cloud microphysics, and fractional cloudiness. Science, 245 , 12271230.

  • Chou, M-D., , and M. J. Suarez, 1994: An efficient thermal infrared radiation parameterization for use in general circulation models. NASA Tech. Memo. 104606, Vol. 3, 85 pp.

  • Del Genio, A. D., , N-S. Yao, , W. Kovari, , and K. K-W. Lo, 1996: A prognostic cloud water parameterization for general circulation models. J. Climate, 9 , 270304.

    • Search Google Scholar
    • Export Citation
  • Helfand, H. M., , and J. C. Lebraga, 1988: Design of a nonsingular level 2.5 second-order closure model for prediction of atmospheric turbulence. J. Atmos. Sci., 45 , 113132.

    • Search Google Scholar
    • Export Citation
  • Houze Jr, R. A., 1993: Cloud Dynamics. Academic Press, 573 pp.

  • Houze Jr., R. A., 1997: Stratiform precipitation in regions of convection: A meteorological paradox? Bull. Amer. Meteor. Soc., 78 , 21792196.

    • Search Google Scholar
    • Export Citation
  • Hu, Q., , and D. Randall, 1994: Low-frequency oscillations in radiative-convective systems. J. Atmos. Sci., 51 , 10891099.

  • Innes, P. M., , J. M. Slingo, , S. J. Woolnough, , R. B. Neale, , and V. D. Pope, 2001: Organization of tropical convection in a GCM with varying vertical resolution: Implication for the simulation of the Madden–Julian oscillation. Climate Dyn., 17 , 777793.

    • Search Google Scholar
    • Export Citation
  • Johnson, R., , T. M. Rickenbarch, , S. A. Rutledge, , P. E. Ciesielski, , and W. H. Schubert, 1999: Trimodal characteristics of tropical convection. J. Climate, 12 , 23972417.

    • Search Google Scholar
    • Export Citation
  • Lau, K. M., , and P. H. Chan, 1985: Aspects of the 40–50 day oscillation during northern winter as inferred from outgoing longwave radiation. Mon. Wea. Rev., 113 , 18891909.

    • Search Google Scholar
    • Export Citation
  • Lau, K. M., , and L. Peng, 1987: Origin of low-frequency (intraseasonal) oscillations in the tropical atmosphere. Part I: Basic theory. J. Atmos. Sci., 44 , 950972.

    • Search Google Scholar
    • Export Citation
  • Lau, K. M., , and H. T. Wu, 2003: Warm rain processes over tropical oceans and climate implications. Geophys. Res. Lett., 30 .2290, doi:10.1029/2003GL018567.

    • Search Google Scholar
    • Export Citation
  • Lau, K. M., , T. Nakazawa, , and C. H. Sui, 1991: Observations of cloud cluster hierarchies over the tropical western Pacific. J. Geophys. Res., 96 , 31973208.

    • Search Google Scholar
    • Export Citation
  • Lau, N. C., , and K. M. Lau, 1986: Structure and propagation of intraseasonal oscillations appearing in a GFDL GCM. J. Atmos. Sci., 43 , 20232047.

    • Search Google Scholar
    • Export Citation
  • Li, M. I., , I. S. Kang, , J. K. Kim, , and B. E. Mapes, 2001: Influence of cloud–radiation interaction on simulating tropical intraseasonal oscillation with an atmospheric general circulation model. J. Geophys. Res., 106 , 1421914233.

    • Search Google Scholar
    • Export Citation
  • Lin, X., , and R. H. Johnson, 1996: Heating, moistening, and rainfall over the western Pacific warm pool during TOGA COARE. J. Atmos. Sci., 53 , 33673384.

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

    • Search Google Scholar
    • Export Citation
  • Nakazawa, T., 1988: Tropical super clusters within intraseasonal variations over the western Pacific. J. Meteor. Soc. Japan, 66 , 823839.

    • Search Google Scholar
    • Export Citation
  • Raymond, D. J., 2001: A new model of the Madden–Julian oscillation. J. Atmos. Sci., 58 , 28072819.

  • Rosenfeld, D., 2000: Suppression of rain and snow by urban and industrial air pollution. Science, 287 , 17931796.

  • Short, D., , and K. Nakamura, 2000: TRMM radar observations of shallow precipitation over tropical oceans. J. Climate, 13 , 41074124.

  • Slingo, J. M., , and A. M. I. P. Collaborators, 1996: Intraseasonal oscillations in 15 atmospheric general circulation models: Results from an AMIP diagnostic subproject. Climate Dyn., 12 , 325357.

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

  • Sud, Y. C., , and G. K. Walker, 1999a: Microphysics of clouds with the relaxed Arakawa–Schubert scheme (McRAS). Part I: Design and evaluation with GATE phase III data. J. Atmos. Sci., 56 , 31963220.

    • Search Google Scholar
    • Export Citation
  • Sud, Y. C., , and G. K. Walker, 1999b: Microphysics of clouds with relaxed Arakawa–Schubert scheme (McRAS). Part II: Implementation and performance in GEOS 2 GCM. J. Atmos. Sci., 56 , 32213240.

    • Search Google Scholar
    • Export Citation
  • Sud, Y. C., , and G. K. Walker, 2003: Influence of ice-phase physics of hydrometeors on moist-convection. Geophys. Res. Lett., 30 .1758, doi:10.1029/2003GL017587.

    • Search Google Scholar
    • Export Citation
  • Sundqvist, H., 1978: A parameterization of non-convective condensation including prediction of cloud water content. Quart. J. Roy. Meteor. Soc., 104 , 677690.

    • Search Google Scholar
    • Export Citation
  • Sundqvist, H., 1988: Parameterization of condensation and associated clouds in models for weather prediction and general circulation simulation. Physical Based Modeling and Simulation of Climate and Climate Change, M. E. Schlesinger, Ed., Reidel, 433–461.

    • Search Google Scholar
    • Export Citation
  • Tiedtke, M., 1993: Representation of clouds in large scale models. Mon. Wea. Rev., 121 , 30403061.

  • Twomey, S. A., 1991: Aerosol, clouds and radiation. Atmos. Environ., 25A , 24352442.

  • Waliser, D. E., , K. M. Lau, , and J. H. Kim, 1999: The influence of coupled sea surface temperatures on the Madden–Julian oscillation: A model perturbation experiment. J. Atmos. Sci., 56 , 333358.

    • Search Google Scholar
    • Export Citation
  • Wang, B., , and H. Rui, 1990: Dynamics of the coupled moist Kelvin–Rossby wave on an equatorial β-plane. J. Atmos. Sci., 47 , 397413.

    • Search Google Scholar
    • Export Citation
  • Wang, W., , and M. E. Schlesinger, 1999: The dependence on convection parameterization of the tropical intraseasonal oscillation simulated by the UIUC 11-layer atmospheric GCM. J. Climate, 12 , 14241457.

    • Search Google Scholar
    • Export Citation
  • Wu, Z., 2003: A shallow CISK, deep equilibrium mechanism for the interaction between large-scale convection and circulation in the Tropics. J. Atmos. Sci., 60 , 377392.

    • Search Google Scholar
    • Export Citation
  • Yin, X., , A. Gruber, , and P. Arkin, 2004: Comparison of the GPCP and CMAP merged gauge–satellite monthly precipitation products for the period 1979–2001. J. Hydrometeor., 5 , 12071222.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    Observed and model annual mean 5-yr climatology of (a) rainfall and (b) OLR. Rain rate greater than 6 mm day−1 and OLR less than 240 Wm−2 are shaded.

  • View in gallery

    Joint PDF of rain rate (mm h−1) and cloud liquid water (mm) from (a) TRMM data, (b) E1, (c) E0, and (d) E2. The PDF is constructed for data within 20°S–20°N, 100°E–120°W.

  • View in gallery

    As in Fig. 1, except for climatology of (a) rainfall and (b) OLR of E1 and E2.

  • View in gallery

    Latitudinal distribution of zonal mean of (a) rainfall and (b) OLR for E0, E1, E2, and observations, denoted by symbols as shown.

  • View in gallery

    E1 − E2 map of (a) convective rainfall (area of values >2 mm day−1 is shaded), (b) large-scale rainfall (area of values <0 is shaded), and (c) OLR (area of values >20 Wm−2 is shaded). Contour interval for rainfall is 0.5 mm day−1 for OLR 5 Wm−2.

  • View in gallery

    Vertical profile of components of diabatic heating averaged over the warm pool region (100°–160°E, 12°S–12°N) for (a) longwave radiation, (b) shortwave radiation, (c) moist processes including convective and large-scale rain, and (d) total diabatic heating. Open circles are for E1, and solid squares for E2. Units are in K day−1.

  • View in gallery

    Joint distribution of highest detrainment level and rain rate for E0. Warm rain and cold rain, and mixed-phase regions are as shown. Contours are population counts in natural logarithm scale.

  • View in gallery

    Vertical profile of E1 − E2 difference in Q, total heating (K day−1); C1, cloudiness (%); and W, vertical velocity (negative hPa day−1) for warm rain.

  • View in gallery

    Same as in Fig. 8, except for cold rain.

  • View in gallery

    Height–longitude section of E1 − E2 difference in (a) heating due to moist processes (K day−1) and (b) cloud fraction (%).

  • View in gallery

    (a) Longitudinal distribution of E1 − E2 difference in vertical mean p velocity (negative hPa day−1) averaged over 20°S–20°N, (b) E1 − E2 difference in vertical mean p velocity, and (c) E1 − E2 difference in the baroclinic component of p velocity at 818 hPa.

  • View in gallery

    Height–longitude cross section of E1 − E2 vertical velocity (negative hPa day−1) averaged over 20°S–20°N for (a) the warm rain region and (b) the cold rain region. Vertical mean value has been subtracted.

  • View in gallery

    Height–longitude section of E1 − E2 temperature difference averaged over 20°S–20°N. Units are in K. Vertical mean has been subtracted, and warm anomalies are shaded.

  • View in gallery

    Height–time cross sections of condensation heating over the (a) Indian Ocean, (b) western Pacific, (c) central Pacific, and (d) eastern Pacific for E1. Contours with heating greater than 1 and 4 K day−1 are shaded.

  • View in gallery

    Same as in Fig. 14, except for E2.

  • View in gallery

    Height–time cross section of detrainment rate over the western Pacific for (a) E1 and (b) E2. Areas with detrainment rates greater than 2 and 4 kg s−1 are shaded.

  • View in gallery

    Time–longitude section of daily (a) precipitation for E1, (b) precipitation for E2, (c) OLR for E1, and (d) OLR for E2 averaged along the equator (8°S–8°N).

  • View in gallery

    Spatial distribution of standard deviation of 20–70-day filtered precipitation for (a) E1 and (b) E2. Rain rates greater than 3 mm day−1 are shaded from light to dark.

  • View in gallery

    Time–longitude section of 20–70-day filtered 200-hPa velocity potential for (a) E1 and (b) E2. Units are in 106 m2 s−1.

  • View in gallery

    East–west wavenumber spectrum of 20–70-day filtered 200-hPa velocity potential for (a) E1 and (b) E2.

  • View in gallery

    Schematic showing model-predicted changes in the atmospheric hydrologic cycle, and cloud–radiative–dynamic feedbacks in the Tropics due to increased autoconversion rate.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 4 4 1
PDF Downloads 3 3 1

Effects of Cloud Microphysics on Tropical Atmospheric Hydrologic Processes and Intraseasonal Variability

View More View Less
  • 1 Laboratory for Atmospheres, NASA Goddard Space Flight Center, Greenbelt, Maryland
© Get Permissions
Full access

Abstract

The sensitivity of tropical atmospheric hydrologic processes to cloud microphysics is investigated using the NASA Goddard Earth Observing System (GEOS) general circulation model (GCM). Results show that a faster autoconversion rate leads to (a) enhanced deep convection in the climatological convective zones anchored to tropical land regions; (b) more warm rain, but less cloud over oceanic regions; and (c) an increased convective-to-stratiform rain ratio over the entire Tropics. Fewer clouds enhance longwave cooling and reduce shortwave heating in the upper troposphere, while more warm rain produces more condensation heating in the lower troposphere. This vertical differential heating destabilizes the tropical atmosphere, producing a positive feedback resulting in more rain and an enhanced atmospheric water cycle over the Tropics. The feedback is maintained via secondary circulations between convective tower and anvil regions (cold rain), and adjacent middle-to-low cloud (warm rain) regions. The lower cell is capped by horizontal divergence and maximum cloud detrainment near the freezing–melting (0°C) level, with rising motion (relative to the vertical mean) in the warm rain region connected to sinking motion in the cold rain region. The upper cell is found above the 0°C level, with induced subsidence in the warm rain and dry regions, coupled to forced ascent in the deep convection region.

It is that warm rain plays an important role in regulating the time scales of convective cycles, and in altering the tropical large-scale circulation through radiative–dynamic interactions. Reduced cloud–radiation feedback due to a faster autoconversion rate results in intermittent but more energetic eastward propagating Madden–Julian oscillations (MJOs). Conversely, a slower autoconversion rate, with increased cloud radiation produces MJOs with more realistic westward-propagating transients embedded in eastward-propagating supercloud clusters. The implications of the present results on climate change and water cycle dynamics research are discussed.

Corresponding author address: Dr. William K. M. Lau, Laboratory for Atmospheres, NASA Goddard Space Flight Center, Code 613, Greenbelt, MD 20771. Email: lau@climate.gsfc.nasa.gov

Abstract

The sensitivity of tropical atmospheric hydrologic processes to cloud microphysics is investigated using the NASA Goddard Earth Observing System (GEOS) general circulation model (GCM). Results show that a faster autoconversion rate leads to (a) enhanced deep convection in the climatological convective zones anchored to tropical land regions; (b) more warm rain, but less cloud over oceanic regions; and (c) an increased convective-to-stratiform rain ratio over the entire Tropics. Fewer clouds enhance longwave cooling and reduce shortwave heating in the upper troposphere, while more warm rain produces more condensation heating in the lower troposphere. This vertical differential heating destabilizes the tropical atmosphere, producing a positive feedback resulting in more rain and an enhanced atmospheric water cycle over the Tropics. The feedback is maintained via secondary circulations between convective tower and anvil regions (cold rain), and adjacent middle-to-low cloud (warm rain) regions. The lower cell is capped by horizontal divergence and maximum cloud detrainment near the freezing–melting (0°C) level, with rising motion (relative to the vertical mean) in the warm rain region connected to sinking motion in the cold rain region. The upper cell is found above the 0°C level, with induced subsidence in the warm rain and dry regions, coupled to forced ascent in the deep convection region.

It is that warm rain plays an important role in regulating the time scales of convective cycles, and in altering the tropical large-scale circulation through radiative–dynamic interactions. Reduced cloud–radiation feedback due to a faster autoconversion rate results in intermittent but more energetic eastward propagating Madden–Julian oscillations (MJOs). Conversely, a slower autoconversion rate, with increased cloud radiation produces MJOs with more realistic westward-propagating transients embedded in eastward-propagating supercloud clusters. The implications of the present results on climate change and water cycle dynamics research are discussed.

Corresponding author address: Dr. William K. M. Lau, Laboratory for Atmospheres, NASA Goddard Space Flight Center, Code 613, Greenbelt, MD 20771. Email: lau@climate.gsfc.nasa.gov

1. Introduction

Recently, there has been a growing body of evidence indicating the importance of tropical warm rain processes in the organization of tropical convection, modulation of clouds and rain types, and possibly global warming. Using 3 yr of data from the Tropical Rainfall Measuring Mission (TRMM), Short and Nakamura (2000) found that more than 20% of the total rain from the Tropics is derived from shallow convection. Johnson et al. (1999) showed that approximately 28% of the rainfall during the Tropical Ocean Global Atmosphere Couple Ocean–Atmosphere Research Experiment (TOGA COARE) may be accounted for by warm rain from midlevel cumulus congestus, and pointed to the importance of a midtropospheric inversion layer, formed by the melting of ice-phase precipitation falling from above, in limiting the growth of penetrative deep convection. They proposed that a basic trimodal (high, middle, and low), rather than the commonly accepted bimodal (high and low), cloud distribution as a more realistic description of the tropical cloud system. They also pointed out the importance of the cumulus congestus in determining the adjustment time scale of convective cycles. Wu (2003) inferred from theoretical calculations that about 20% of the latent heating in the Tropics would be contributed by mid- to low-level condensation processes in order to maintain the observed moist static stability profile. Innes et al. (2001) demonstrated that significant improvement in the simulation of the Madden–Julian oscillation (MJO) can be achieved by increasing vertical resolution, which helps to better resolve the melting level in convection in the Met Office’s general circulation model (GCM). Li et al. (2001) showed that including cloud–radiation interaction associated with changes in the autoconversion (coalescence–collision processes in warm rain) time scale can lead to a better simulation of the MJO in a GCM. Warm rain processes are not only associated with liquid-phase condensation, but also with melting of falling snow and ice, and reevaporation of raindrops. This was further affirmed in Sud and Walker (2003), which showed that including ice-phase processes at the melt–freeze zone of falling precipitation produced a better simulation of low- and midlevel convection, and reduced the excessive development of deep convection in an early version of the microphysics of clouds with relaxed Arakawa–Schubert scheme (McRAS) used in the National Aeronautics and Space Administration (NASA) Goddard Earth Observing System (GEOS) GCM.

More recently, Lau and Wu (2003, hereafter LW) using 3 yr of TRMM data showed that warm rain may be more abundant and prevalent than all previous estimates. They estimate that warm rain contributes up to 31% of the total rain, and as much as 72% of the total rain area over the tropical oceans. Lau and Wu (2003) also found that for warm rain, mostly associated with mid- to low-level clouds, there is an increase in rainfall efficiency (defined as the rain production per unit cloud liquid water) of 8%–10% per degree Celsius increase in sea surface temperature (SST), whereas for cold rain (associated with ice physics), the rainfall efficiency is virtually independent of SST. They argued that for warm rain, where the large-scale dynamic forcing is weaker, the increasing rainfall efficiency may stem from an increased rate of conversion of water vapor to cloud droplets, and to raindrops by diffusion–deposition, and faster growth of raindrops by collision–coalescence. For deep convection, the autoconversion processes are modulated by convective cloud-scale updraft, and hence it is insensitive to surface temperature changes. Lau and Wu (2003) further hypothesized that in a warmer climate, increased warm rain efficiency may lead to faster rainout of cloud liquid water and consequently a lesser amount of water carried upward in deep convection, resulting in an overall reduction of clouds at all levels. This paper is aimed at exploring the possible physical underpinnings of the aforementioned hypothesis by examining the sensitivity of the atmospheric hydrologic cycle, and the intraseasonal variability to bulk microphysical processes governing the conversion of cloud water substance into precipitation.

2. Model description

The model used is the GEOS GCM version-2 (GEOS-2), with 4° latitude × 5° longitude resolution and 20 sigma levels, with the top of the atmosphere at 10 hPa. The model uses the Microphysics of Clouds with Relaxed Arakawa–Schubert Scheme (McRAS) developed by Sud and Walker (1999a, b). Radiative transfer calculations are based on the shortwave and longwave models of Chou and Suarez (1994), which take into account the radiative properties of cloud types and their interaction with cloud microphysics. The most important feature of McRAS germane to the model experiments presented herein is the scheme used in the conversion from cloud to rainwater, which follows Sundqvist (1978, 1988), and is adopted in many global climate models, for example, the European Centre for Medium-Range Weather Forecasts (ECMWF), the Goddard Space Flight Center (GSFC), the National Centers for Environmental Prediction (NCEP), and the Laboratoire Météorologie Dynamique (LMD), and described in Tiedtke (1993), Del Genio et al. (1996), and Sud and Walker (1999a, b) and several others:
i1520-0442-18-22-4731-e1
where m is the cloud water content, mc is the critical cloud water content, and τ the autoconversion timescale (AT), the inverse of which is referred to as the autoconversion rate or the rainfall efficiency. From Eq. (1), AT approaches infinity if m is much smaller than mc; that is, there is no conversion from cloud water to rain, but it quickly approaches its maximum value (=m/P) when m exceeds mc. To account for ice-phase microphysics, including the temperature range in which cloud ice–cloud water mixtures coexist, τ and mc are modified by
i1520-0442-18-22-4731-e2
where α is a scaling parameter for the basic autoconversion time constant τo, and mco is the basic critical water content. Here, F1, F2, and F3 are empirical functions that take into account, respectively, the dependence of cloud lifetime and critical cloud water content on the intensity of the coalescence–collision process, coexistence of ice and water droplets (the Bergeron–Findeisen process), and ice-phase physics for high cirrus clouds. In addition, McRAS includes cloud dissipation mechanisms associated with cloud-top entrainment instability (Del Genio et al. 1996) and diffusive mixing of dry and cloudy air masses within each grid cell and cloud advection. Stratiform clouds are formed by supersaturation produced by either diabatic cooling, or adiabatic cooling associated with the large-scale motion. Boundary layer clouds are formed by turbulent eddies carrying saturated water vapor into the highest detraining level below the inversion, following Helfand and Lebraga (1988). The version of McRAS used in this work also includes the latest upgrades to ice-phase physics, which allows in-cloud freezing in cumulus updrafts, and melting of falling snow (Sud and Walker 2003). The improvement in ice-phase physics has produced more realistic midlevel clouds (cumulus congestus), consistent with the observed trimodal distribution (Johnson et al. 1999).

As stated previously, LW’s results suggest that the autoconversion rate, that is, warm rain precipitation efficiency, may increase as temperature rises. On the other hand, the indirect effects of aerosols may reduce the autoconversion rate leading to increased cloud lifetime, and suppressed precipitation (Twomey 1991; Rosenfeld 2000). The suppression of drizzle, in particular, would lead to increased liquid water content in clouds that would enhance the cloud reflectivity and reduced shortwave penetration (Albrecht 1989). Furthermore, the radiative heating imbalance in the vertical and horizontal, set up by the above processes, may lead to a redistribution of convection and the large-scale circulation, which in turn modulates the radiative forcings. All these complex processes, whether in response to temperature rise or aerosol increase, appear to hinge on the sensitivity of the radiation–dynamics feedback tied to changes in the autoconversion rate. At present GCMs are incapable of simulating the detailed interaction of cloud microphysics with temperature and aerosols. Yet GCMs with prognostic cloud water use bulk cloud microphysical parameterizations that are related fundamentally to some form of autoconversion process in initiating cloud and rain formation. Thus a study of the sensitivity of cloud–precipitation interaction to the autoconversion process is important in providing some preliminary understanding of possible atmospheric hydrologic responses to global warming and aerosol forcing.

The results presented here are based on sensitivity experiments carried out with different values of the basic autoconversion rate in McRAS taken from Tiedtke (1993). We set α = 1 in the control experiment (E0), α = 0.2 for fast autoconversion (E1), and α = 5 for slow autoconversion (E2). Each experiment was carried out for five simulated years (1987–91) with identical initial conditions starting from 1 January 1987, with observed sea surface temperature and sea ice conditions similar to those used for the Atmospheric Model Intercomparison Project II. Note that while the autoconversion parameter is defined for the warm rain process, a change in the autoconversion rate will be reflected in changes in the entire convective system including warm and cold rain processes as given by Eqs. (1) and (2), as well as by induced changes in the large-scale circulation. In all the experiments, we use τo = 103 s, and mco = 10−3, and 0.3 × 10−3 kg m−3 for convective and stratiform rain, respectively. All empirical constants used in the functions Fi’s (i = 1, 2, 3) are the same as in Sud and Walker (1999a, b).

3. Results

In this section, results of the control and the anomaly experiments are compared and contrasted to determine the sensitivity of the model climate, organization of convection, and large-scale circulation regime to the microphysics of clouds and precipitation.

a. Mean climate

Figure 1 shows the comparison of the model climatological annual rainfall and outgoing longwave radiation (OLR) fields for E0 and observations of rainfall from the Global Precipitation Climatology Project (GPCP; Adler et al. 2003) and OLR from NCEP. Overall the model simulates reasonably well the spatial distribution of both climatologies. However, the model overestimates rainfall, especially over the Maritime Continent of the western Pacific, and the land regions of northwestern South America and central Africa. The model underdevelops the ITCZ over the tropical eastern Pacific and the Atlantic. The similarity in the OLR fields indicates that the model replicates the distribution of deep convection and high cloudiness reasonably well, except in the ITCZ regions, as is also the case for rainfall. The model has higher OLR values than observed in the subtropical zones especially over the southeastern tropical Pacific, due to excessive dryness over these regions. Despite the noted biases, the model circulation and rainfall climatologies provide some reassurance to the realism of the model experiments. Further comparison of the model results to other relevant observations will be discussed in later sections.

The imposed autoconversion rates induce fundamental changes in the relationship between cloud liquid water (CLW) and rain rate (RR) in the model climate, as shown in the joint probability distribution function (PDF) of CLW and RR for E0, E1, and E2, shown in Fig. 2. The PDF is constructed based on data for the domain 20°S–20°N, 100°E–120°W. For observations, TRMM data from 6 December 1997 to 31 August 2000 are used. The PDFs highlight the differences in each case, typical for the tropical oceanic domains. In E0 (Fig. 2c), the presence of two branches in the PDF is obvious. From the slope of the distribution, the upper branch yields a shorter AT of about 9–10 min, and the lower branch a longer AT of about 25–30 min. As will be shown later, the fast branch corresponds to precipitation driven by dynamics, including convective and large-scale (stratiform) types, and the slow branch to dynamics–radiation-driven precipitation mostly associated with anvil-type stratiform clouds. The fast branch agrees reasonably well with TRMM observation (Fig. 2a), which has a mean AT of 10–12 min, except that the model appears to underestimate the variability about the mean. Obviously, in El (Fig. 2b), the model accentuates the fast branch yielding a mean AT of about 6 min, while eliminating the slow branch completely, that is, substantially reducing the cloud–radiation effects. This, however, does not mean that all rains in E1 are convective, because large-scale rain is also a part of the convective rain system in McRAS (see discussion in the next section). Conversely, for E2 (Fig. 2d), the model tends to favor the slow branch, with a more continuous cloud–rain spectrum. A large part of the distributions in E2 has very long AT, indicated by slopes close to zero. The tropical mean convective-to-stratiform rain ratio is approximately 1.7 in E1, compared to 1.3 in E2. The plausible climate and atmospheric hydrologic cycle anomalies introduced by these different precipitation microphysics relationships are examined next.

Figure 3 shows the spatial distribution of the climatological mean precipitation and OLR for E1 and E2, respectively. In E1, rainfall is heavier than in E2 in the Maritime Continent, northwestern South America, the ITCZ, and the South Pacific convergence zone (SPCZ), where deep convection prevails. An exception is found in the western North Pacific where E1 simulates reduced rainfall relative to E2. This is due to enhanced subsidence in E1 induced by increased ascent from deep convection over the Maritime Continent. More pronounced is the change in the OLR (Fig. 3b), with E1 showing substantially less area with OLR less than 240 W m−2 compared to E2. This is due to the depletion of cloud water by the higher rainout rate in E1, which results in less cloud in the middle and upper tropospheres. The positive feedback between increasingly concentrated deep convection and large-scale subsidence in E1 contributes further to the reduction of clouds. In the zonal mean, the total rainfall increase (E1 − E2) is no more than 1 mm day−1, with the largest difference occurring mostly in the equatorial region and the southern Tropics (Fig. 4a). Overall, there is a 16% increase in total rain within the Tropics (30°S–30°N) from E2 to E1. This is much larger than the model’s internal variability, but still within the observations errors in GPCP rainfall (Yin et al. 2004). The large change in cloudiness as reflected in the zonal mean OLR is shown in Fig. 4b. There is an approximate 15–25 W m−2 OLR reduction in the Tropics in E2 compared to E1. Interestingly, E2 agrees with observations better than E1 and E0, and appears to mitigate a negative bias of deep clouds over the equatorial region and the southern tropical oceans compared to observations.

b. Convective versus stratiform precipitation

In nature, convective rain is generated by the rapid, vigorous overturning of the atmosphere required to neutralize an unstable vertical distribution of moist static energy, while stratiform clouds and rain grow by condensation–deposition in a stable environment, with relative weak vertical motions. In the Tropics, most clouds and rain systems are initially convectively driven, but in older convection, stratiform clouds and rain occur frequently (Houze 1993, 1997). In McRAS, stratiform rain is synonymous with large-scale rain. Almost all tropical rain in the model starts out as convective rain through convective (dry or moist) cloud mass flux. However, rain coming from the part of the convection wherein the cloud continues to generate rain out of cloud water longer than the typical convective lifetime (assumed to be an hour) is considered stratiform. This means that rain from old convection is counted as stratiform rain in the model. In the extratropics, or in regions of the Tropics where convection is inactive or decaying, such as in the extended anvil regions, stratiform rain is produced by large-scale uplift and supersaturation. As shown in Table 1, in E1, rainfall over the Tropics is far more convective (63% of total) compared to E2 (56%). Overall, there is a 16% increase in total rain, with a 30% increase in convective rain, and a 2% reduction in large-scale rain from E2 to E1. The distribution of rainfall increase (E1 − E2) for convective rain (Fig. 5a), shows that the largest changes are in convective towers that are anchored to the Maritime Continent, land regions of south and Southeast Asia, Africa, and central America. For large-scale rain (Fig. 5b), there is widespread reduction over the Tropics, except in regions of the deep convective cores. The reduced large-scale rain in E1 is associated with the longer AT, and a greater abundance of clouds with very light rain or drizzles in E2, compared to E1 (see discussion of Fig. 2). The similarity between the pattern of large-scale rainfall and OLR (Fig. 5c) suggests that much of the reduction in large-scale rainfall is associated with the reduction of rain from anvil or high-level clouds that contribute significantly to the OLR.

c. Vertical profiles

Figure 6 shows the vertical profiles of diabatic heating rates: longwave, shortwave, condensation heating, and the total heating averaged over the western Pacific warm pool region for E1 (open circle) and E2 (solid square), respectively. Compared to E2, E1 shows increased longwave cooling at all levels up to 250 hPa (Fig. 6a), due to the reduction in cloudiness at all levels (see later the discussion of Fig. 10). Reduced upper-level cloudiness in E1 also leads to reduced shortwave absorption by clouds, and hence relative cooling in the upper troposphere above 400 hPa (Fig. 6b). Increased condensation heating is found at all levels in E1, with contributions from both warm and cold condensation (rain out), produced, respectively, below and above the melting–freezing level at about the 5-km height (Fig. 6c). When all the heating terms are combined, increased autoconversion (E1) leads to positive total diabatic heating (cooling) below (above) the mean freezing (0°C) altitude over the warm pool region. Hence, while the direct effect of increased autoconversion is to produce more warm rain, and less cold rain because of depletion of cloud water in the upper troposphere, cloud–radiative effects introduce a differential vertical heating that will destabilize the atmospheric column, with a positive feedback leading to further enhanced condensation heating due to both warm and cold rain processes. The feedback mechanism is further explored in the following.

The responses of clouds, diabatic heating, and large-scale motions are quite different in regions of cold rain (deep clouds) and warm rain (shallow and middle clouds). As noted in LW, from TRMM data, rain systems having a daily rain rate of less than 0.2 mm h−1 with storm heights below the freezing level are considered light warm rains, whose precipitation efficiency has a strong SST dependence. Heavy rains (rain rate >2 mm h−1), with prevailing storm heights above the freezing altitude, are considered cold rain produced primarily by ice-phase precipitation processes. Mixed-phase precipitation produces rain rates between the two thresholds outlined above. The precise thresholds, however, cannot be used here because of the bias in the simulated rainfall intensity. To establish the thresholds, the joint distribution of the highest level of detrainment and daily rain rate has been constructed. The rate of detrainment is computed at model levels as the rate at which moist cloud air is mixed into the environment. The level of highest detrainment corresponds to the cloud-top level and can be identified as a proxy of the storm height.

Figure 7a shows the joint distribution of the highest detrainment level and the total rain rate for E0. Since the detrainment rate applies only to convective clouds, including those that later continue on to precipitate as stratiform rain in old convection, the joint distribution is a distribution essentially of convective rain. Any changes in rain types, convective versus stratiform, are not reflected in the distribution. It is clear that for rain rates of less than 2 mm day−1, the majority of the rain systems will have cloud tops limited to altitudes below the freezing level (near 500 hPa), and will be considered to be warm rain. For rain rates greater than 15 mm day−1, the highest detrainment levels are well above the freezing altitude. Based on Fig. 7, we use rain rates of 2 mm day−1 and 15 mm day−1 to define the thresholds for the model warm and cold rain, respectively. Under this definition, warm rain storm heights have a maximum population near 800 hPa, but also include some at storm heights reaching up to 200–300 mb. This represents contamination by cold drizzle, which amounts to less than 10%–12% of the true warm rain population. This does not affect the statistics of the warm rain in the present results. On the other hand, cold rain is relatively well defined, with the majority of storm heights above 500 hPa. Consistent with this definition, mixed-phase precipitation, as represented by moderate rain rates, has substantial storm height populations above and below the freezing level. Using these thresholds, the climatological fractional warm rain in E0 is estimated to be 9% of the total rain amount occupying 70% of the rainy area in the Tropics. For cold rain, the percentages are 52% and 7%, respectively. These numbers are in reasonable agreement with TRMM observations by LW. Table 1 shows the breakdown of the rain types in the warm–cold classification compared to the convective–stratiform classification. According to the new rain classification, the 16% increase in total rain, that is, 3.15 mm day−1 in E2 to 3.65 mm day−1 in E1, is derived from a redistribution of the rain spectrum with increases in both the warm rain (from 8.5% to 10%) and cold rain (from 50.5% to 53%), and a reduction of the mixed phase (from 42% to 37%), compared to E2. It is important to note that because of the increase in total rainfall, the absolute rainfall amount is larger in E1 compared to E2, in all three categories including the mixed phase.

To contrast the differences in the responses in the warm and cold rain regions, the vertical profiles of total diabatic heating, cloudiness, and vertical motion are computed for each rain type and shown as differences (E1 − E2) in Figs. 8 and 9. In the warm rain regime (Fig. 8), increased autoconversion leads to an overall reduction in total diabatic heating throughout the atmospheric column due to increased longwave cooling, and reduced shortwave heating (Fig. 8a). The cooling is largest, about 0.4°C day−1 near the surface and reduces toward the lower and middle tropospheres, culminating in a positive anomaly in the 900–750-hPa layer. This anomaly is due to condensation heating by the increased warm rain. The diabatic cooling shows a secondary maximum at about 300 hPa. Cloudiness is reduced at all levels, with maximum reductions of 6% near 800 hPa and 4% at 200 hPa (Fig. 8b). The lower peak coincides with the preferred level of warm clouds, and the upper peak with extended anvil clouds in the model cloud climatology (not shown). The warm rain region is dominated by strong subsidence throughout the entire troposphere with a maximum near 300 hPa (Fig. 8c), consistent with the reduction in cloudiness, and overall diabatic cooling in the region.

Conversely, in the cold rain region (Fig. 9), a faster autoconversion leads to positive net diabatic heating through most of the troposphere with a maximum value of 1°C day−1. This is accompanied by a large reduction in cloudiness below 300 hPa due to increased rain out of cloud water. As discussed in the next subsection, a secondary circulation spawned by precipitation–clouds–radiative feedback leads to increased deep convection in the cold rain region. This is evident in the increased cloud fraction at 300–200 hPa (Fig. 9b), and anomalous ascending motion maximizing at 30 hPa day−1 in the upper troposphere (Fig. 9c). The increased ascent in the upper troposphere enhances vertical moisture transport and depletes the moisture in the lower troposphere. Furthermore, the rain-laden air mass in the lower troposphere loses buoyancy. As a result, there is a large reduction in cloudiness (up to 18%) below 400 hPa (Fig. 8b), and weak net subsidence below 600 hPa (Fig. 9c).

As noted previously, almost all the rainfall increase due to fast autoconversion is in the convective type, associated with both warm, and cold rain processes. The height–longitude section of E1 − E2 condensation heating anomalies (Fig. 10a) shows that in the long-term mean, cold rain (above 500 hPa) heating is most pronounced over the tropical landmasses. Warm rain heating (below 500 hPa) is not only dominant over the tropical oceans, but also contributes substantially to the cold rain region. Indeed, the warm rain condensation heating over the Africa landmass is stronger than that of the cold rain. The effects of the increased efficiency in warm rain to deplete cloud liquid water, and to reduce mid- and low-level clouds, are obvious in Fig. 10b. In the upper troposphere, cloudiness is generally reduced, except in the vicinity of deep convection over land, where upper-layer clouds including cirrus, are increased due to the increased supply of moisture and active ice-phase processes (see Fig. 10a). The large reduction in upper-level clouds over the tropical Pacific is associated with strong sinking motion in the warm rain region (see Fig. 8c). The oceanic sinking motion reinforces the mid- to low-level cloudiness reduction due to increased autoconversion, by increasing moist stability, and suppressed penetrative convection.

d. Secondary circulations

In this section, we provide additional diagnostics to highlight the role of the warm rain and cold rain processes in response to the differential heating (both horizontal and vertical) induced by increased autoconversion. Figure 11a shows the differential (E1 − E2) vertical mean p velocity, averaged between 20°S and 20°N. The vertical mean characterizes the general sinking motion over the oceanic warm rain region due primarily to longwave cooling, and the induced rising motion concentrated over the cold rain (deep convection) region of the African continent (20°–40°E), the Maritime Continent (100°–130°E), and central South America (50°–80°W). The horizontal distribution of the mean vertical motion field (Fig. 11b) shows close resemblance to that for the convective rainfall difference pattern (Fig. 5a), indicating a large-scale overturning motion driven by the diabatic heating gradient between the vast oceanic subsidence region and the land-locked deep convection regions. The mean vertical motion field, which is nearly identical to those found at the midtroposphere (not shown), can be considered as the “equivalent barotropic” component of the vertical motion. The “baroclinic” component of the vertical motion is defined as the deviation from the vertical mean, hereafter referred to as the secondary circulation. Figure 11c shows that the secondary circulations are weaker compared to the deep convective motions, and have much smaller spatial scales, involving relative downward motions in regions of deep convection, and compensating upward and downward motions in adjacent regions. These secondary circulations are associated with the differential vertical heating gradient due to the ice-phase (cold rain) and non-ice-phase (warm rain) processes as discussed previously. They represent the model manifestation of “mesoscale convective complexes” in the observations. The geographic locations of the secondary motions associated with warm and cold rain regions are shown in Fig. 12. In these figures, the vertical motions should not be considered as connecting different parts of the warm or cold rain regions within themselves, but rather between adjacent warm and the cold rain regions, in the time-mean sense, as shown in Fig. 11c. In the warm rain region (Fig. 12a), anomalous rising motion is found below the freezing level, and sinking motion is found with comparable magnitude (3–6 hPa day−1) above it. The reverse is found in the cold rain region, but with a magnitude 5–6 times larger (Fig. 12b). The vertical motion fields suggest a two-cell secondary circulation, connecting the warm and cold rain regions. The lower cell is composed of rising motion in the lower troposphere driven by warm rain (Fig. 12a), capped by region of no vertical motion (horizontal divergence) at the freezing level, and a return sinking motion in the cold rain region (Fig. 12b). The upper cell consists of anomalous rising motion above the freezing height in the cold rain region, and forced subsidence above the freezing height in the warm rain region. Both cells contribute to large horizontal divergence at the freezing level near 500 hPa, where significant cloud detrainment is found.

The secondary circulation is manifested in a warming of the lower troposphere due to increased low-level heating by warm rain, and a cooling of the upper troposphere, principally due to longwave radiation, as shown in the height–longitude temperature difference (E1 − E2) cross section along the equator in Fig. 13. The vertical temperature distribution is quite uniform across the entire Tropics, and is similar for warm and cold rain regions, with the zero line separating the warm and the cold regions running near the freezing–melting level. The lower/warm, and upper/cold configuration, leads to an overall destabilization of the tropical atmosphere, providing a positive feedback via secondary circulations that focus and intensify deep convection over the land regions and increase warm rain, but reduce tropical cloudiness.

e. Convective recycling time

In this section, we explore the sensitivity of the convective recycling processes to rain microphysics and radiative–dynamic feedback processes. Figures 14 and 15 show the vertical cross sections of the model condensation heating in different parts of the tropical ocean over a 3-month period (1 September 1988 through 30 November 1988) for E1 and E2, respectively. During this period, the Indian Ocean is relatively free of deep convection (Figs. 14a and 15a). In E1, over the western Pacific, convective activity is more intense and convective episodes tend to occur more in clusters (Fig. 14b). Within clusters, the deep convection recycles with 2–3-day time scales. However, the time separations between clusters, or convective lulls, are of the order of 10–20 days. As shown later, this quasi-bimodal distribution is a result of the enhanced warm rain processes in E1. In E2, deep convective episodes recycle with a more uniform time scale of about 5 days (Fig. 15b). In contrast, over the central Pacific (Figs. 14c and 15c), the intensity and frequency of deep convection are reduced with more low-level heating in E1, indicating a shift to a more warm rain–dominant regime from E2 to E1. As stated previously, this shift stems from the radiation–dynamic feedback associated with increased convection in the neighborhood of land regions, and forced subsidence over the oceans, which limits the development of deep convection there. Over the eastern Pacific (Figs. 14d and 15d), both E1 and E2 are under similar climatologically large subsidence conditions. The increased rainfall efficiency moderately enhances low-level heating from more warm rains in E1.

The multiple convective recycling time scales in E1 are determined to be due to changes in the moistening processes of the lower troposphere. In E1, the lower and middle tropospheres are moistened through increased cloud detrainment by increased warm rain below the freezing height (Fig. 16a) compared to E2 (Fig. 16b). The melting of falling snow from the cold and mixed-phase rain produces cooler layers immediately beneath the freezing height, which effectively reduces the buoyancy of the rising air and prevents the rising convection to penetrate above it. Convective available potential energy (CAPE) is built up as low- and midlevel clouds, that is, cumulus congestus, grow and detrain. However, a large portion of the convection is capped by the relatively stable air at the melting–freezing zone. As a result, it takes the warm rain process more time to build up the instability required to spur deep convection (see Fig. 16a), hence, the appearance of convective lulls. Once deep convection occurs, the positive feedback via the secondary circulation sustains the faster recycling cycles (2–3 days), until convection depletes all the CAPE, and the next convective cluster cycle begins. In E2, because of the reduced rainfall efficiency, the warm rain and mixed-phase rain processes play a lesser role in regulating the convective cycles, as evident in the relative lack of detrainment at the melting–freezing level. Here, the recycling time scale for deep convection is primarily determined by the rate of moistening of the lower troposphere through low-level detrainment and convergence, and the release of CAPE in deep convection via radiative–convective adjustment (Hu and Randall 1994; Lin and Johnson 1996).

f. Intraseasonal variability

Associated with the changes in the convective recycling time scale is a fundamental shift in the circulation regime over the Tropics between E1 and E2, as illustrated in the daily time–longitude sections of rainfall and OLR (Fig. 17). Previous observations (Nakazawa 1988; Lau et al. 1991) have shown that an MJO can be identified as an eastward-propagating supercloud cluster from the Indian Ocean to the central Pacific, embedded by intermittent, transient westward-propagating cloud clusters. These features have been attributed to responses of coupled Kelvin (eastward) and Rossby (westward) components to latent heating, modified by air–sea interaction (e.g., Lau and Peng 1987; Wang and Rui 1990; Waliser et al. 1999, and many others). Such a mix of east–west propagation is reasonably well reproduced in the control E0 (not shown). Figure 17 emphasizes the differences between E1 and E2. In E1 more intense precipitation and more pronounced eastward propagation associated with the MJO are evident (Fig. 17a). In contrast, in E2 (Fig. 17b), the eastward propagation is less well defined, with quasi-stationary heavy precipitation events, accompanied by mixed eastward and westward propagations. The difference in supercloud cluster organization is clearly marked in the daily OLR results (Figs. 17c and 17d). In E1 (Fig. 17c), the OLR signal is not very well defined, and deep convection is more isolated, with contracted cloud anvils, which is consistent with the reduced upper-level cloudiness. In contrast, in E2 (Fig. 17d), the supercloud cluster structure is more pronounced, with mixed westward propagation embedded in eastward-propagating envelopes. The westward propagation of individual cloud streaks, for example, is quite pronounced during the period from June to August in E2.

Clearly, E1 and E2 represent two distinct circulation regimes excited by differences in the cloud–radiative feedback associated with changes in autoconversion. The overly energetic eastward propagation in E1 is consistent with previous simulations of the MJO in GCMs with cumulus parameterization using moisture convergence as closure, but without cloud microphysics influencing the cloud–radiative interactions (e.g., Lau and Lau 1986; Slingo et al. 1996; Sperber 2003; Wang and Schlesinger 1999). These GCM simulations tend to produce too fast and too regular eastward propagation modes, and underestimate the supercloud cluster organization and westward transients. With reduced rainfall efficiency (E2), which enables stronger precipitation–cloud–radiation interaction, the model is able to simulate a more realistic range of tropical intraseasonal variability.

Thus far, the discussion has been focused on the mean and total variability of the tropical atmospheric hydrologic processes from daily to intraseasonal time scales. The following will be focused on the MJO, traditionally defined as the collection of atmospheric features propagating eastward along the equator from the Indian Ocean to the central Pacific, as seen in broadband-filtered data (Madden and Julian 1972; Lau and Chan 1985). Figure 18 shows the 20–70-day variance of rainfall for E1 and E2. In E1, the MJO variances are noticeably enhanced over the climatological convective centers of the Maritime Continent, eastern Africa, and Central America. In E2, the centers of action of the MJO are found over the western and central Pacific around 10°–20°N, away from the Asian summer monsoon region. The reduction in MJO activity over the subtropical western North Pacific in E1 is consistent with the large-scale subsidence induced by increased convective rainfall over the continents (see Figs. 5a and 11b). Figure 19 shows the Hovmöller diagram of 20–70-day bandpassed 200-hPa velocity potential for E1 and E2, respectively. During the boreal spring in E1, MJOs, if present, tend to be more energetic and faster with shorter time intervals of 20–30 days between events. At other times, they tend to be relatively less well defined, with each event separated by longer time intervals. In E2, all MJOs are relatively less well defined overall, but they are present all year round. These features are supported by an east–west wave spectrum analysis for the entire data period shown in Fig. 20. In E2 (Fig. 20b), a single maximum near 40-day period is found for the eastward-propagating component, while in E1 a secondary peak with periodicity at 20–30 days can be found, in addition to the 40-day maximum. In both cases, no significant westward-propagating signals are detected in the 20–70-day bandpassed data.

4. Concluding discussions

From experiments with the GEOS GCM with McRAS, we find that an increased autoconversion rate can lead to a substantial change in the atmospheric water cycle through cloud–radiative–dynamical feedback processes shown in Fig. 21. For fixed sea surface temperature forcing, a direct effect of increased autoconversion is increased warm rain (non-ice phase) efficiency, associated with more rain and reduced cloudiness at all levels. The reduction in upper-level clouds causes increased longwave cooling and reduced shortwave heating, leading to overall cooling and large-scale tropical subsidence (indicated by the large downward arrow in Fig. 21). Concomitant with the radiative cooling, which is strongest at the upper levels, is increased low-level heating from warm rain, which together maintain a strong vertical heating gradient between the upper and lower tropospheres and between the regions of warm rain (low to middle clouds) and cold rain (high clouds). The heating gradient spawns a positive feedback between the region of warm and cold rain, via secondary circulations with rising motion in the lower troposphere and sinking motion in the upper troposphere in the warm rain region, connected to motions of the opposite sign in the cold rain region. In the warm rain region, the secondary circulation provides the uplift energy for buoyant plumes to rise above the lifting condensation level to form clouds, which quickly rainout. As illustrated in Fig. 21, the secondary circulation is characterized by a large-scale divergence near the melting–freezing level. Large-scale descent in the upper troposphere from radiative cooling, and ascent in the lower troposphere from increased warm rain, are found in large areas of the Tropics. The radiative and condensation heating force ascent in an increasingly concentrated region of deep convection, with a contracted anvil in the upper troposphere, and a downdraft in the lower troposphere. Even though the autoconversion processes directly affect warm rain, ice-phase condensation processes play an important role in producing the overall changes in the atmospheric water cycle. We note that the secondary circulations are very similar to those associated with stratiform rain found in older convection systems in the Tropics identified by Houze (1997).

Through the cloud–radiation–dynamics feedback process, an increase in autocoversion rate will lead to a shorter recycling time for deep convection, which plays a key role in regulating the time scale and intensity of the model MJO. Increased cloud–radiative feedback produces a supercloud cluster structure with westward-propagating transients, embedded in eastward-migrating convection in agreement with observations. The experiments also illustrate the importance of low- to midlevel heating and moistening processes, due to shallow clouds and cumulus congestus in determining the slow time scale of the MJO. We find that increased rainfall efficiency and the resulting diminished cloud–radiative feedback in E1 cause more intermittent MJOs, with preferred eastward propagation, but a lack of the supercloud cluster structure. These inferences are in agreement with recent observations and theories of the MJO (Johnson et al. 1999; Raymond 2001; Wu 2003). How the intermittent nature of the MJO relates to the multiple time scales in the convective recycling associated with increased rainfall efficiency is not known, and is a subject of further investigation.

An important implication of the present results is that atmospheric hydrologic and dynamical processes may be fundamentally controlled by microphysical processes of clouds and precipitation. Autoconversion is an essential process by which cloud drops grow into raindrops in the initial process of rain formation. Warm rain systems are generally associated with low and middle clouds, even though some eventually grow into deep convection, and precipitate as cold rain through ice-phase precipitation. As suggested by LW, a warmer climate may lead to increased autoconversion and, hence, affect the redistribution of rain and cloud types, in ways similar to those inferred in this study. Alternatively, autoconversion may be suppressed and cloud lifetime prolonged by increasing the level of anthropogenic aerosols, which act as cloud condensation nuclei, favoring more abundant but smaller cloud droplets (Twomey 1991; Albrecht 1989; Rosenfeld 2000). Hence, our model experiments, E1 and E2, may also be viewed as surrogate simulations, respectively, for less (higher rainfall efficiency) and more (lower rainfall efficiency) aerosols in the environment. Most importantly, our results imply that, regardless of whether it is aerosol effects or global warming, microphysical processes of precipitation and clouds are the critical pathways by which these effects may regulate the atmospheric water cycle and climate. Given the coarse resolution and crude microphysical parameterization in the GCM, the present work represents a best-effort attempt to address aspects of these pathways. Increasing model resolution will help, but will not be sufficient by itself. Organized and sustained efforts in improving the representation of microphysical cloud–precipitation processes in global climate models is paramount to producing more robust results.

Acknowledgments

This work is jointly supported by the Tropical Rainfall Measuring Mission (TRMM) project, and the Modeling, Analysis and Prediction Program of the NASA Earth Science Enterprise. We thank the two anonymous reviewers who provided insightful comments and suggestions.

REFERENCES

  • Adler, R. F., and Coauthors, 2003: The version 2 Global Precipitation Climatology Project (GPCP) monthly precipitation analysis (1979–present). J. Hydometeor., 4 , 11471167.

    • Search Google Scholar
    • Export Citation
  • Albrecht, B., 1989: Aerosols, cloud microphysics, and fractional cloudiness. Science, 245 , 12271230.

  • Chou, M-D., , and M. J. Suarez, 1994: An efficient thermal infrared radiation parameterization for use in general circulation models. NASA Tech. Memo. 104606, Vol. 3, 85 pp.

  • Del Genio, A. D., , N-S. Yao, , W. Kovari, , and K. K-W. Lo, 1996: A prognostic cloud water parameterization for general circulation models. J. Climate, 9 , 270304.

    • Search Google Scholar
    • Export Citation
  • Helfand, H. M., , and J. C. Lebraga, 1988: Design of a nonsingular level 2.5 second-order closure model for prediction of atmospheric turbulence. J. Atmos. Sci., 45 , 113132.

    • Search Google Scholar
    • Export Citation
  • Houze Jr, R. A., 1993: Cloud Dynamics. Academic Press, 573 pp.

  • Houze Jr., R. A., 1997: Stratiform precipitation in regions of convection: A meteorological paradox? Bull. Amer. Meteor. Soc., 78 , 21792196.

    • Search Google Scholar
    • Export Citation
  • Hu, Q., , and D. Randall, 1994: Low-frequency oscillations in radiative-convective systems. J. Atmos. Sci., 51 , 10891099.

  • Innes, P. M., , J. M. Slingo, , S. J. Woolnough, , R. B. Neale, , and V. D. Pope, 2001: Organization of tropical convection in a GCM with varying vertical resolution: Implication for the simulation of the Madden–Julian oscillation. Climate Dyn., 17 , 777793.

    • Search Google Scholar
    • Export Citation
  • Johnson, R., , T. M. Rickenbarch, , S. A. Rutledge, , P. E. Ciesielski, , and W. H. Schubert, 1999: Trimodal characteristics of tropical convection. J. Climate, 12 , 23972417.

    • Search Google Scholar
    • Export Citation
  • Lau, K. M., , and P. H. Chan, 1985: Aspects of the 40–50 day oscillation during northern winter as inferred from outgoing longwave radiation. Mon. Wea. Rev., 113 , 18891909.

    • Search Google Scholar
    • Export Citation
  • Lau, K. M., , and L. Peng, 1987: Origin of low-frequency (intraseasonal) oscillations in the tropical atmosphere. Part I: Basic theory. J. Atmos. Sci., 44 , 950972.

    • Search Google Scholar
    • Export Citation
  • Lau, K. M., , and H. T. Wu, 2003: Warm rain processes over tropical oceans and climate implications. Geophys. Res. Lett., 30 .2290, doi:10.1029/2003GL018567.

    • Search Google Scholar
    • Export Citation
  • Lau, K. M., , T. Nakazawa, , and C. H. Sui, 1991: Observations of cloud cluster hierarchies over the tropical western Pacific. J. Geophys. Res., 96 , 31973208.

    • Search Google Scholar
    • Export Citation
  • Lau, N. C., , and K. M. Lau, 1986: Structure and propagation of intraseasonal oscillations appearing in a GFDL GCM. J. Atmos. Sci., 43 , 20232047.

    • Search Google Scholar
    • Export Citation
  • Li, M. I., , I. S. Kang, , J. K. Kim, , and B. E. Mapes, 2001: Influence of cloud–radiation interaction on simulating tropical intraseasonal oscillation with an atmospheric general circulation model. J. Geophys. Res., 106 , 1421914233.

    • Search Google Scholar
    • Export Citation
  • Lin, X., , and R. H. Johnson, 1996: Heating, moistening, and rainfall over the western Pacific warm pool during TOGA COARE. J. Atmos. Sci., 53 , 33673384.

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

    • Search Google Scholar
    • Export Citation
  • Nakazawa, T., 1988: Tropical super clusters within intraseasonal variations over the western Pacific. J. Meteor. Soc. Japan, 66 , 823839.

    • Search Google Scholar
    • Export Citation
  • Raymond, D. J., 2001: A new model of the Madden–Julian oscillation. J. Atmos. Sci., 58 , 28072819.

  • Rosenfeld, D., 2000: Suppression of rain and snow by urban and industrial air pollution. Science, 287 , 17931796.

  • Short, D., , and K. Nakamura, 2000: TRMM radar observations of shallow precipitation over tropical oceans. J. Climate, 13 , 41074124.

  • Slingo, J. M., , and A. M. I. P. Collaborators, 1996: Intraseasonal oscillations in 15 atmospheric general circulation models: Results from an AMIP diagnostic subproject. Climate Dyn., 12 , 325357.

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

  • Sud, Y. C., , and G. K. Walker, 1999a: Microphysics of clouds with the relaxed Arakawa–Schubert scheme (McRAS). Part I: Design and evaluation with GATE phase III data. J. Atmos. Sci., 56 , 31963220.

    • Search Google Scholar
    • Export Citation
  • Sud, Y. C., , and G. K. Walker, 1999b: Microphysics of clouds with relaxed Arakawa–Schubert scheme (McRAS). Part II: Implementation and performance in GEOS 2 GCM. J. Atmos. Sci., 56 , 32213240.

    • Search Google Scholar
    • Export Citation
  • Sud, Y. C., , and G. K. Walker, 2003: Influence of ice-phase physics of hydrometeors on moist-convection. Geophys. Res. Lett., 30 .1758, doi:10.1029/2003GL017587.

    • Search Google Scholar
    • Export Citation
  • Sundqvist, H., 1978: A parameterization of non-convective condensation including prediction of cloud water content. Quart. J. Roy. Meteor. Soc., 104 , 677690.

    • Search Google Scholar
    • Export Citation
  • Sundqvist, H., 1988: Parameterization of condensation and associated clouds in models for weather prediction and general circulation simulation. Physical Based Modeling and Simulation of Climate and Climate Change, M. E. Schlesinger, Ed., Reidel, 433–461.

    • Search Google Scholar
    • Export Citation
  • Tiedtke, M., 1993: Representation of clouds in large scale models. Mon. Wea. Rev., 121 , 30403061.

  • Twomey, S. A., 1991: Aerosol, clouds and radiation. Atmos. Environ., 25A , 24352442.

  • Waliser, D. E., , K. M. Lau, , and J. H. Kim, 1999: The influence of coupled sea surface temperatures on the Madden–Julian oscillation: A model perturbation experiment. J. Atmos. Sci., 56 , 333358.

    • Search Google Scholar
    • Export Citation
  • Wang, B., , and H. Rui, 1990: Dynamics of the coupled moist Kelvin–Rossby wave on an equatorial β-plane. J. Atmos. Sci., 47 , 397413.

    • Search Google Scholar
    • Export Citation
  • Wang, W., , and M. E. Schlesinger, 1999: The dependence on convection parameterization of the tropical intraseasonal oscillation simulated by the UIUC 11-layer atmospheric GCM. J. Climate, 12 , 14241457.

    • Search Google Scholar
    • Export Citation
  • Wu, Z., 2003: A shallow CISK, deep equilibrium mechanism for the interaction between large-scale convection and circulation in the Tropics. J. Atmos. Sci., 60 , 377392.

    • Search Google Scholar
    • Export Citation
  • Yin, X., , A. Gruber, , and P. Arkin, 2004: Comparison of the GPCP and CMAP merged gauge–satellite monthly precipitation products for the period 1979–2001. J. Hydrometeor., 5 , 12071222.

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

Observed and model annual mean 5-yr climatology of (a) rainfall and (b) OLR. Rain rate greater than 6 mm day−1 and OLR less than 240 Wm−2 are shaded.

Citation: Journal of Climate 18, 22; 10.1175/JCLI3561.1

Fig. 2.
Fig. 2.

Joint PDF of rain rate (mm h−1) and cloud liquid water (mm) from (a) TRMM data, (b) E1, (c) E0, and (d) E2. The PDF is constructed for data within 20°S–20°N, 100°E–120°W.

Citation: Journal of Climate 18, 22; 10.1175/JCLI3561.1

Fig. 3.
Fig. 3.

As in Fig. 1, except for climatology of (a) rainfall and (b) OLR of E1 and E2.

Citation: Journal of Climate 18, 22; 10.1175/JCLI3561.1

Fig. 4.
Fig. 4.

Latitudinal distribution of zonal mean of (a) rainfall and (b) OLR for E0, E1, E2, and observations, denoted by symbols as shown.

Citation: Journal of Climate 18, 22; 10.1175/JCLI3561.1

Fig. 5.
Fig. 5.

E1 − E2 map of (a) convective rainfall (area of values >2 mm day−1 is shaded), (b) large-scale rainfall (area of values <0 is shaded), and (c) OLR (area of values >20 Wm−2 is shaded). Contour interval for rainfall is 0.5 mm day−1 for OLR 5 Wm−2.

Citation: Journal of Climate 18, 22; 10.1175/JCLI3561.1

Fig. 6.
Fig. 6.

Vertical profile of components of diabatic heating averaged over the warm pool region (100°–160°E, 12°S–12°N) for (a) longwave radiation, (b) shortwave radiation, (c) moist processes including convective and large-scale rain, and (d) total diabatic heating. Open circles are for E1, and solid squares for E2. Units are in K day−1.

Citation: Journal of Climate 18, 22; 10.1175/JCLI3561.1

Fig. 7.
Fig. 7.

Joint distribution of highest detrainment level and rain rate for E0. Warm rain and cold rain, and mixed-phase regions are as shown. Contours are population counts in natural logarithm scale.

Citation: Journal of Climate 18, 22; 10.1175/JCLI3561.1

Fig. 8.
Fig. 8.

Vertical profile of E1 − E2 difference in Q, total heating (K day−1); C1, cloudiness (%); and W, vertical velocity (negative hPa day−1) for warm rain.

Citation: Journal of Climate 18, 22; 10.1175/JCLI3561.1

Fig. 9.
Fig. 9.

Same as in Fig. 8, except for cold rain.

Citation: Journal of Climate 18, 22; 10.1175/JCLI3561.1

Fig. 10.
Fig. 10.

Height–longitude section of E1 − E2 difference in (a) heating due to moist processes (K day−1) and (b) cloud fraction (%).

Citation: Journal of Climate 18, 22; 10.1175/JCLI3561.1

Fig. 11.
Fig. 11.

(a) Longitudinal distribution of E1 − E2 difference in vertical mean p velocity (negative hPa day−1) averaged over 20°S–20°N, (b) E1 − E2 difference in vertical mean p velocity, and (c) E1 − E2 difference in the baroclinic component of p velocity at 818 hPa.

Citation: Journal of Climate 18, 22; 10.1175/JCLI3561.1

Fig. 12.
Fig. 12.

Height–longitude cross section of E1 − E2 vertical velocity (negative hPa day−1) averaged over 20°S–20°N for (a) the warm rain region and (b) the cold rain region. Vertical mean value has been subtracted.

Citation: Journal of Climate 18, 22; 10.1175/JCLI3561.1

Fig. 13.
Fig. 13.

Height–longitude section of E1 − E2 temperature difference averaged over 20°S–20°N. Units are in K. Vertical mean has been subtracted, and warm anomalies are shaded.

Citation: Journal of Climate 18, 22; 10.1175/JCLI3561.1

Fig. 14.
Fig. 14.

Height–time cross sections of condensation heating over the (a) Indian Ocean, (b) western Pacific, (c) central Pacific, and (d) eastern Pacific for E1. Contours with heating greater than 1 and 4 K day−1 are shaded.

Citation: Journal of Climate 18, 22; 10.1175/JCLI3561.1

Fig. 15.
Fig. 15.

Same as in Fig. 14, except for E2.

Citation: Journal of Climate 18, 22; 10.1175/JCLI3561.1

Fig. 16.
Fig. 16.

Height–time cross section of detrainment rate over the western Pacific for (a) E1 and (b) E2. Areas with detrainment rates greater than 2 and 4 kg s−1 are shaded.

Citation: Journal of Climate 18, 22; 10.1175/JCLI3561.1

Fig. 17.
Fig. 17.

Time–longitude section of daily (a) precipitation for E1, (b) precipitation for E2, (c) OLR for E1, and (d) OLR for E2 averaged along the equator (8°S–8°N).

Citation: Journal of Climate 18, 22; 10.1175/JCLI3561.1

Fig. 18.
Fig. 18.

Spatial distribution of standard deviation of 20–70-day filtered precipitation for (a) E1 and (b) E2. Rain rates greater than 3 mm day−1 are shaded from light to dark.

Citation: Journal of Climate 18, 22; 10.1175/JCLI3561.1

Fig. 19.
Fig. 19.

Time–longitude section of 20–70-day filtered 200-hPa velocity potential for (a) E1 and (b) E2. Units are in 106 m2 s−1.

Citation: Journal of Climate 18, 22; 10.1175/JCLI3561.1

Fig. 20.
Fig. 20.

East–west wavenumber spectrum of 20–70-day filtered 200-hPa velocity potential for (a) E1 and (b) E2.

Citation: Journal of Climate 18, 22; 10.1175/JCLI3561.1

Fig. 21.
Fig. 21.

Schematic showing model-predicted changes in the atmospheric hydrologic cycle, and cloud–radiative–dynamic feedbacks in the Tropics due to increased autoconversion rate.

Citation: Journal of Climate 18, 22; 10.1175/JCLI3561.1

Table 1.

Tropical mean accumulated rainfall amount and percent contribution in each rain category.

Table 1.
Save