1. Introduction
Upper-tropospheric ice clouds play a key role in the global climate through their greenhouse and solar albedo effects. However, much uncertainty remains about their spatial distribution and quantity in general circulation models (GCMs) (Tsushima et al. 2006), which causes ambiguities in weather and climate simulations (Waliser et al. 2009, hereafter W09; Wu et al. 2009). Observationally, global distributions of the ice-water content/path (IWC/IWP) have been retrieved by satellite remote sensing, but it is known that their estimations vary among different sensors and algorithms (e.g., Fig. 4 of W09). In the GCMs of the Intergovernmental Panel on Climate Change Fourth Assessment Report (IPCC AR4); Fig. 3 of W09), the simulated IWP varied by a factor of 6 or more. It is thought that this large disagreement among models is a result of the different cloud schemes used. In particular, cumulus parameterization schemes affect ice cloud properties at low latitudes because detrainment of deep cumulus convection given in each scheme largely affects the amount of upper cloud ice.
Owing to increasing computer power in recent years, new approaches to atmospheric modeling that are superior to traditional cumulus parameterization schemes are becoming viable. It is expected that they will be able to overcome the cumulus parameterization “deadlock” (Randall et al. 2003). Two different approaches have been proposed: superparameterization (Grabowski 2001; Khairoutdinov and Randall 2001; Randall et al. 2003; Tao et al. 2009) and global cloud-system-resolving models (GCRM) (Satoh et al. 2008). Both aim to calculate explicitly deep cumulus with model mesh intervals of the order of a few kilometers by using cloud microphysics instead of cumulus parameterization schemes. In this study, we focus on the GCRM. We examine how upper-tropospheric ice clouds change with the model parameters and how the Hadley circulation intensity is related to the upper ice clouds.
Iga et al. (2007, hereafter I07) presented a first attempt to obtain climate conditions with a GCRM. They used the Nonhydrostatic Icosahedral Atmospheric Model (NICAM) (Tomita and Satoh 2004; Satoh et al. 2008) under a perpetual July condition with prescribed sea surface temperature (SST). Although they obtained climatological resultssimilar to observations, biases were still observed in the precipitation and outgoing longwave radiation (OLR). I07 also investigated the sensitivity of OLR to model parameters in the boundary layer and cloud microphysics schemes. They speculated that the differences in OLR resulted from the upper ice clouds, which were simulated differently by these sensitivity experiments. However, they did not show how or whether the OLR depended on the IWP/IWC. Other aspects of the atmospheric circulation, such as the Hadley circulation, were not discussed, as I07 was a short note. Here we extend I07 and discuss how changes in the IWP/IWC are related to the OLR, precipitation, and Hadley circulation.
The low-latitude atmospheric circulation is the Hadley circulation, driven by the diabatic heating of deep cumulus. It is theoretically well understood that the intensity of the Hadley circulation is determined to a first approximation by the radiation constraint in the subsidence region (Schneider 1977; Held and Hou 1980; Satoh 1994). Adiabatic warming due to subsidence is balanced by radiative cooling in the downward branch of the Hadley circulation. Because the thermal structure is constrained by the moist adiabat in the subtropics, the subsidence motion is approximately determined by the ratio of radiative cooling in the free atmosphere to moist adiabatic stratification. It is thus proportional to the ratio of precipitation P to surface specific humidity q0 (Sarachik 1978; Satoh and Hayashi 1992; Satoh 1994). This relation is frequently invoked in discussions of Hadley circulation change in the context of global warming. Vecchi and Soden (2007) argue that in all of the IPCC AR4 models, the overturning circulation weakens as the climate warms because q0 increases more rapidly than P with global temperature. Given this relationship, it is clear that ambiguity in upper ice clouds causes the ambiguity in the Hadley circulation intensity because the upper ice clouds greatly affect the radiation budget through regulation of longwave cooling.
This study is relevant to the future projection of global warming. It has been reported that clouds in these models behaved differently from those in conventional GCMs under global warming conditions (Miura et al. 2005; Wyant et al. 2006; Collins and Satoh 2009). Y. Tsushima et al. (2011, unpublished manuscript) and Collins and Satoh (2009) conducted idealized climate sensitivity experiments by adding 2 K to the SST and found that the upper clouds of GCRMs have sensitivities different from those in conventional GCMs. These results were taken from a specific set of model parameters, however, and it is natural to question the robustness of the results. To understand the climate sensitivity associated with upper ice clouds, we need to explore the behavior of upper clouds in models with explicit cloud convection in greater detail. Some attempts have been made in an idealized framework (Satoh and Matsuda 2009) or limited-domain experiments (Satoh et al. 2010). However, GCRM experiments are required to understand upper-cloud behavior under the condition of interaction between large-scale circulations and deep convective cloud systems.
This study has two aims. The first is to examine the dependency of upper ice clouds on cloud microphysics and boundary layer parameterizations, as well as model resolution. We also hope to elucidate the relationship between IWP/IWC and OLR. We have found that the total IWP was not correlated with the OLR, indicating that the spatial distribution of IWP was more relevant. The second aim is to examine the response of the Hadley circulation to the upper ice clouds. We focused on the change in the Hadley circulation under present-day conditions, but the results will give insight into future changes at low latitudes resulting from global warming.
The structure of this paper is as follows: in section 2, the model and experimental settings are described. Following an overview of the results in section 3, we examine how upper ice clouds respond to each parameter in section 4 and discuss how the upper ice-cloud response is generated in section 4b. In section 5, the change in the Hadley circulation intensity is discussed. The summary is given in section 6.
2. Model and experimental settings
We used NICAM (Tomita and Satoh 2004; Satoh et al. 2008) for the numerical simulations. NICAM is based on nonhydrostatic governing equations and an icosahedral grid, where horizontal grid points are quasi-uniformly distributed over the sphere. It can be run in cloud-system-resolving mode with a cloud microphysics scheme. We followed the experimental design used in I07. Most experiments were performed with a horizontal mesh interval of approximately 14 km. One experiment on the sensitivity to model resolution was performed with a mesh size of approximately 7 km (referred to as CS3L200dx7). The vertical domain had 40 layers, with the model top at z = 40 km. Although the 14-km grid interval was coarser than that normally used for cloud-system-resolving simulations, the large-scale organization of tropical clouds, such as in the Madden–Julian oscillation, was similar for experiments with grid sizes of approximately 3.5, 7, and 14 km (Tomita et al. 2005; Miura et al. 2007; Nasuno et al. 2009). A series of small-planet experiments (Satoh and Matsuda 2009) also showed that the variations of upper clouds are qualitatively similar among these resolutions. The experiments with the 14-km grid interval are useful for performing sensitivity studies focused on upper ice clouds.
We used the following physical schemes in the simulations. The cloud microphysics scheme was that proposed by Grabowski (1998). It accounted for ice effects in a simplified way: airborne and precipitating hydrometeors were prognosed, and ice and liquid phases were diagnosed by a temperature-dependent function in mixed-phase clouds. The cloud ice and snow were calculated, but graupel was not included in the model. For the radiation scheme, the two-stream model simulation radiation transfer code, version 10 (MSTRNX), (Nakajima et al. 2000; Sekiguchi and Nakajima 2008) was used. Although the size and shape of the cloud particles have an important effect on the radiation budget, the cloud particle size was independently defined in the radiation scheme. In the cloud microphysics scheme of Grabowski (1998), the Marshall–Palmer distribution is assumed for snow and rain particles, whereas in the radiation scheme, we assumed effective radii ri = 40 μm and ri = 8 μm for all solid and liquid particles, respectively. Cloud particle size is independently defined in the radiation scheme and cloud physics scheme to simplify the problems, although the effect of cloud radius on radiation is an important issue in cloud and radiation interaction. Following the philosophy of cloud-resolving models, we assumed no statistical cloud-overlapping parameterizations in the radiation scheme. In other words, the hydrometeors were distributed homogeneously within each grid point, and the cloud fraction was either one or zero.
For the turbulence, we used the Mellor–Yamada level 2 closure scheme (Mellor and Yamada 1974) with a simple moisture effect for the control experiment. The moisture effect was counted using the saturated Brunt–Väisälä frequency to estimate the Richardson number in a saturated grid (Durran and Klemp 1982). The maximum mixing length L0, which determines the mixing length through L ≡ zL0/(z + L0), was externally specified and was changed in the sensitivity experiments of the boundary layer scheme. We compared the control experiment with one using a modified version of the Mellor–Yamada level 2 turbulence scheme called the Mellor–Yamada–Nakanishi–Niino model (MYNN) (Noda et al. 2010; Nakanishi and Niino 2006). This sensitivity experiment is referred to as CS4MYNN. The scheme included partial clouds and an internally determined mixing length. The resolution dependency of GCRM on the boundary layer scheme MYNN was also examined by Noda et al. (2010).
The surface flux was calculated following the procedure of Louis (1979). Distributions of SST, sea ice, and ozone were fixed to those of the July climatology, and the solar incidence was held constant at the 15 July value. In other words, the experiment was conducted under perpetual July conditions (Cess et al. 1990). These types of experiments are convenient because the equilibrium state is reached in a relatively short time compared to the case with seasonal cycles.
Summary of experiments. For CS4MYNN, the level 2 turbulence scheme of Nakanishi and Niino (2006) was used, which is based on Mellor and Yamada (1974).
3. Overview of the results
Table 2 summarizes the mean simulation values. The rain is averaged in the latitude bands 30°S–30°N, which covers almost the entire region of the southern Hadley cells. The other quantities are averaged in the 30°–10°S latitude band, which corresponds to the subsidence region of the southern Hadley cell. The experimental data analyzed in this study are almost identical to those used by I07, from which some representative results are shown. Here, we focus on the upper ice clouds and vertical motions, together with the sensitivities to model parameters. I07 reported that the horizontal distributions of precipitation and cloud fraction are quite similar to those observed in the July climatology. However, biases exist, such as stronger precipitation, larger cloud fraction amounts, and stronger zonal jets in the control case (CS4L100).
One-month-averaged values for each case: QR is the column-mean radiative cooling rate and 
Here, the IWP and IWC results are compared with those in W09. Note that the seasons are different: the present results are for perpetual July, whereas W09 shows the annual-mean distributions. In Fig. 1, the horizontal distribution of IWP and the meridional distribution of IWC for CS4L100 are compared with CloudSat observations (Figs. 4e and 12 of W09, respectively). Note that both snow and ice are included in the IWP and IWC of our results (CS4L100). The global mean, tropical, and extratropical IWP values were 0.0819, 0.0936, and 0.0701 kg m−2, respectively (cf. to Figs. 1d and 18 of W09). The NICAM IWC showed slightly larger values compared to CloudSat data and other cloud-resolving model (CRM)-type GCMs [the reduced acceleration in the vertical (RAVE) approach and multiscale modeling framework (MMF)] (comparable to Figs. 5d, 12d, and 12h of W09).
(a) One-month-averaged ice water path (IWP) (g m−2) for CS4L100, (b) IWP by CloudSat (Fig. 4d of W09), (c) zonal-mean ice water content (IWC) (mg m−3), and (d) IWC by CloudSat (Fig. 5d of W09).
Citation: Journal of Climate 24, 11; 10.1175/2010JCLI3472.1
W09 reported considerable variations in IWP both in observations and GCM simulations. Although the IWP of the NICAM simulation is larger than that estimated from CloudSat, it is within the range of values shown in Figs. 1d and 3 of W09. One of the reasons the NICAM IWP is larger than that in GCMs is that most GCMs do not include snow in their IWP. In Fig. 2, the probability density functions (PDFs) of IWPs within 30°S–30°N are compared with estimated values from observations (superposed on Fig. 9d of Wu et al. 2009). Although the region and seasons were different between our simulations and observations, the simulated PDFs are mostly within observed variations, except in high IWP where they are higher than observed ones. It is probably because IWC tends to be large for the Grabowski (1998) scheme, which is used in this study, and is indicated from the analysis of the Cloud–Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO)/CloudSat simulator in NICAM (Satoh et al. 2010).
PDFs of IWP. The black line indicates the case CS4L100 averaged in 30°S–30°N. The colored lines and marks are from Fig. 9d of Wu et al. (2009): “R04,” “Hogan,” and “Sayres” were estimated from CloudSat observations; and Advanced Microwave Sounding Unit B (AMSU-B), Atmospheric Radiation Measurement tropical western Pacific (ARM TWP), and Moderate Resolution Imaging Spectroradiometer (MODIS) were also compared.
Citation: Journal of Climate 24, 11; 10.1175/2010JCLI3472.1
The zonal-mean vertical velocity is shown in Fig. 3. Consistent with the biases in precipitation and zonal jets, the meridional circulation for CS4L100 was also stronger than the reanalysis data [40-yr European Centre for Medium-Range Weather Forecasts Re-Analysis (ERA-40)]. In the following section, we explore how the meridional circulation (the Hadley circulation) changed in the sensitivity experiments. In a statistical equilibrium state, the radiative cooling in the free atmosphere is almost balanced by the diabatic heating associated with precipitation, and the radiative cooling in the subtropics drives the subsidence motion of the Hadley circulation. The outgoing longwave radiation at the top of the atmosphere varies with the radiative cooling in the free atmosphere and, hence, is greatly affected by upper ice clouds (Table 2). Thus, the key variables to understand the response of the Hadley circulation are precipitation, OLR, and upper ice clouds, whose sensitivities we examine in detail in the next section.
One-month-averaged zonal-mean ω velocity for (top) CS4L100 and (bottom) ERA-40 climatology. The color scale indicates the difference from ERA-40.
Citation: Journal of Climate 24, 11; 10.1175/2010JCLI3472.1
4. Sensitivity of upper clouds
a. Results
Figure 4 shows the sensitivities of zonal-mean OLR, vertical velocity at 500 hPa (omega), precipitation, and cloud ice. The sensitivity of OLR is reproduced by Fig. 4a in which results from additional experiments are inserted. In the control experiment (CS4L100), the OLR is 10–20 W m−2 larger than the climatology observed by the Earth Radiation Budget Experiment (ERBE), as shown by I07. To examine the OLR bias, I07 conducted the following sensitivity experiments: turbulence mixing length L0 is doubled (CS4L100 → CS4L200); snow fall speed cs is changed from 4 to 3 (CS4L200 → CS3L200); and horizontal resolution is doubled (CS4L200 → CS3L200dx7). As the turbulence mixing increased or the snow fall speed decreased (CS4L200 and CS3L200), both the OLR and precipitation decreased, along with the OLR bias (Fig. 4a). The new experiment performed in this study with an improved boundary layer scheme (CS4MYNN) showed that the OLR is in close agreement with the ERBE data.
One-month-averaged zonal-mean values: (a) outgoing radiation (OLR), (b) precipitation, (c) omega velocity at 500 hPa, and (d) IWP (kg m−2).
Citation: Journal of Climate 24, 11; 10.1175/2010JCLI3472.1
Figure 4b shows the omega velocity at 500 hPa, and Fig. 5 shows the meridional streamfunctions. Because the stream lines are vertically aligned in the middle troposphere, the vertical velocity of one single layer at 500 hPa represents the strength of the Hadley circulation. The maximum values of the Hadley streamfunction decreased in order of the experiments: CS4L100, CS4L200, CS3L200, and CS3L200dx7. As can be seen from Figs. 4a, 4b, and 5, the OLR in the subtropics decreased with the subsidence.
One-month-averaged zonal-mean streamfunction for each case.
Citation: Journal of Climate 24, 11; 10.1175/2010JCLI3472.1
Figure 4c shows the precipitation. The change in OLR, just described, was strongly linked to the change in the upper clouds and, hence, to precipitation and the Hadley circulation, as shown in Figs. 4b and 4c. Biases in precipitation and streamfunction of the control experiment (CS4L100) were reduced in a similar way to those in the OLR, as the parameters were changed. See Fig. 3 for the bias of the circulation of CS4L200. The relationships among OLR, precipitation, and the Hadley circulation will be further discussed in section 5.
Here we examine in more detail how upper ice clouds change the OLR. The sensitivity of the latitudinal distribution of the zonal-mean IWP is shown in Fig. 4d. The change in IWP when L0 was doubled from CS4L100 to CS4L200 was remarkable: it decreased by about half around the ITCZ region and also decreased in almost all of the tropical region. As shown in Fig. 4a, the OLR decreased by 10~20 W m−2 if the L0 is doubled, if cs was reduced from 4 to 3, or if the horizontal resolution was doubled. Approximately half of the OLR change for L0 doubling was due to the increase in water vapor (i.e., the clear-sky effect, not shown here). The rest is due to the change in cloud forcing. In contrast, the change in OLR by cs or by the resolution was mainly due to the change in cloud forcing.
Figure 6a shows the relation between OLR and IWP averaged from 30°S to 30°N. As can be seen, a positive correlation was observed, except in CS4L200 → CS3L200. This result seems inconsistent with the intuitive argument that more upper ice clouds mean less OLR. In fact, this result implies that the spatial distribution of IWC, rather than IWP, is most important for the radiation budget. This is shown in Fig. 6b. It is discussed below by defining 
Relationships between values in Table 2: (a) IWP and OLR, (b)
Citation: Journal of Climate 24, 11; 10.1175/2010JCLI3472.1
A 1-h-mean IWP (kg m−2). Bright colors (yellow and red) correspond to the regions where the optical depth is more than unity.
Citation: Journal of Climate 24, 11; 10.1175/2010JCLI3472.1
Figure 8 shows weighted distributions of IWP, D(IWP) = D(IWP) ≡ IWPf(IWP), where f(IWP) is a probability density function of IWP, which satisfies ∫f(IWP)d log10 IWP = 1. The total area in the x-logscaled coordinate corresponds to an average IWP. All cases have a peak around IWP, 10 kg m−2, which occurs due to the deep convection. D(IWF = 10 kg m−2) decreased as L0 or the horizontal resolution increased, or as cs decreased. This was due to a reduction of the deep convection (Fig. 9a). However, because the infrared radiative flux was saturated for clouds deeper than τ = 1 or IWP = 0.025 kg m−2, the reduction of D at IWP = 10 kg m−2 did not greatly affect the OLR. However, in the radiatively unsaturated region IWP 0.025 kg m−2, which corresponds to thin anvil or cirrus clouds, OLR was sensitive to the value of IWP. In Fig. 8, IWP = 0.025 kg m−2 is shown by the vertical line. At this value, D increased with L0 or the horizontal resolution and decreased with cs. This is why cloud radiative forcing changed with the parameters.
Weighted distribution of IWP, D(IWP) ≡ IWPp(IWP) in 30°S–30°N, where p(IWP) is the PDF of log10 IWP. Total area in these logscaled coordinates corresponds to the average IWP. Black vertical lines indicate IWP = 0.025 kg m−2, corresponding to unit optical depth.
Citation: Journal of Climate 24, 11; 10.1175/2010JCLI3472.1
One-month-averaged mass flux averaged in 30°S–30°N. The threshold vertical velocity is (a) 1 and (b) 0.1 m s−1.
Citation: Journal of Climate 24, 11; 10.1175/2010JCLI3472.1
To estimate the influence of frozen particles on radiation, we used
b. Mechanism
In this subsection, we consider the mechanism behind these upper-ice-cloud sensitivities. Figure 10 shows the zonal-mean relative humidity. For the change CS4L100 → CS4L200, the relative humidity increased by more than 10% in the upper troposphere of the ITCZ region and by 30% in the lower stratosphere at low latitudes. The high relative humidity in CS4L200 was caused by the enhanced vertical diffusion coefficient KV owing to the change in mixing length L0 from 100 to 200 m. The vertical mixing was supposed to occur in the lower boundary layer; however, it was also present in conditionally unstable layers. The coefficient KV became larger at saturated grid points, both in the lower boundary and the deep convection, which we explicitly resolved. Thus, in our simulations, the vertical transport of water was caused by explicit grid-scale advection and vertical mixing by the turbulence scheme, for both shallow and deep convection. Although KV was largest in the boundary layers, the vertical mixing was not negligible at any altitude. In the stable region outside the moist convective grids, it took the minimum threshold value KV = 0.15 m2 s−1. In the case of CS4L200, doubling L0 enhanced the vertical transport of water content in deep convective columns. This wetness increase prohibited the evaporation of cirrus and anvil. This is why the upper layer was much cloudier in CS4L200 than in CS4L100.
One-month-averaged zonal-mean relative humidity for each case.
Citation: Journal of Climate 24, 11; 10.1175/2010JCLI3472.1
For the change CS4L200 → CS3L200, the decrease in snow fall speed causes the increase in cloud fraction and consequently the OLR decreased. The increase in cloud fraction causes more evaporation to occur and wet upper troposphere in the ITCZ region. The effect of snow fall speed on cloud fraction was examined by Jakob (2002), Tanaka et al. (2008), and Satoh and Matsuda (2009).
The dependency on horizontal resolution can be understood in terms of small-scale vertical convection. Figure 9 shows the convective mass flux between 30°S and 30°N defined by two different threshold values of vertical velocity; 1 and 0.1 m s−1. The former threshold extracts only deep convection, whereas the latter also extracts weaker convection. Figure 9b shows that small-scale convection increased with the horizontal resolution (CS3L200 → CS3L200dx7), while Fig. 9a shows unchanged deep convection. The increase in small-scale convection probably occurred because it was better represented in the higher-resolution experiment (CS3L200dx7).
5. Sensitivity of the Hadley circulation
Here, we examine how the upper ice clouds affected the intensity of the Hadley circulation. We particularly focus on the southern Hadley cell because we used the perpetual July condition. As shown by Table 2 and Figs. 4 and 11, the subsidence motion of Hadley circulation was closely related to the upper ice clouds. In Table 2, the rain is averaged in the latitude bands 30°S–30°N, which almost covers the entire region of the southern Hadley cell. Other quantities are averaged in 30°–10°S, which represents the latitudes of the subsidence region of the southern Hadley cell.
Relationships between values in Table 2: (a) QR and OLR, (b) QR and precipitation, (c) QR and 
Citation: Journal of Climate 24, 11; 10.1175/2010JCLI3472.1
First, we note from Table 2 and Fig. 11a that OLR and the net radiative cooling of the atmosphere QR are closely correlated. This reflects that short wave radiation does not effect the variation in radiative cooling. One exception was found for CS4MYNN (not plotted in Fig. 11), where low-level cloud amounts and outgoing shortwave radiation (OSR) were larger than the other values. In this case, where an improved turbulence model (Noda et al. 2010; Nakanishi and Niino 2006) was used, the lower cloud amounts were larger than those of the observation, resulting in a lower OLR (by ~10 W m−2) and higher OSR (by ~24 W m−2, not shown) than that for CS4L100. Note that there is also a heat source (sink) at the surface because SST is fixed in these simulations and the heat source for CS4MYNN is ~10 W m−2 larger. As a result, the column radiative cooling rate QR was coincidentally close to that of CS4L100. In these two cases, the strength and the width of the Hadley circulation and the precipitation intensity were very close. The net QR was also strongly correlated to P (Table 2 and Fig. 11b). This occurred because the radiative cooling was almost balanced by latent heat release due to precipitation in the global equilibrium atmospheric energy balance. Here QR and P were not balanced when the integral range was limited to the Hadley cell region in Table 2. This was due to the eddy heat flux between the Hadley cell region and the extratropics. Nevertheless, QR and P were closely correlated in our results. Because OLR varies with the area of the upper ice clouds as upper ice clouds increase, radiative cooling and precipitation both decrease (Fig. 6b). The subsidence flow of the Hadley cell is strongest in the subtropic region. Its magnitude there is determined by the local radiation balance.
In the above discussion, small eddy components are ignored. Here, we distinguish between the Hadley cell component and the other component: 
To summarize, the strength of the Hadley circulation is tightly linked to the coverage of the upper ice clouds. Figure 11e shows the relation between the subsidence and
6. Summary
We studied the relation between upper ice clouds and the Hadley circulation in global cloud-system-resolving simulations under a perpetual July condition. A global nonhydrostatic model was used to examine the dependency on the parameters of the cloud microphysics and the boundary layer mixing scheme, together with the horizontal resolution. We performed several sensitivity studies with horizontal mesh interval Δx = 14 km and one study with Δx = 7 km. The global cloud-system-resolving simulations allow the deep convection and associated upper-cloud systems (anvil clouds) to be represented by a cloud microphysics scheme instead of cumulus parameterizations. The latter caused major ambiguities in conventionally used general circulation models (GCMs). Although simulations with Δx = 14 km are relatively coarse for deep convective core resolution, they are fine enough to capture key properties of upper clouds and mesoscale organization (Tomita et al. 2005; Miura et al. 2007; I07; Sato et al. 2009; Satoh and Matsuda 2009; Miura et al. 2009). This makes them useful for studying the impacts of physics schemes on cloud-system-resolving simulations.
For the control case (CS4L100), the simulated climatology was close to observations, as reported in I07. The horizontal distribution of ice water path (IWP) and meridional distribution of ice water content (IWC) were within the ranges of the other numerical models and were slightly larger than the values retrieved from CloudSat data. In this case, precipitation, OLR, and the strength of the Hadley circulation were slightly larger. Understanding the sensitivity of these variables to model parameters was the key aim of this study. We examined sensitivity to the turbulent mixing length in the boundary layer scheme L0, the snow fall speed in the cloud microphysics scheme cs, and the horizontal resolution Δx. As L0 increased, cs decreased, or as Δx decreased, the OLR decreased. The OLR was reduced by 10–20 W m−2, yielding results that were closer to the observations in I07. In all cases, the change in OSR was small (several W m−2).
We found that the IWP and OLR were positively correlated. This apparently contradicts the view that OLR decreases as upper ice clouds increase. IWP decreased when L0 increased, cs decreased, or Δx decreased. The fraction of the area where the optical depth of ice clouds was more than unity,
We examined the relationship between the Hadley circulation and upper ice clouds. To a first approximation, the subsidence motion of the Hadley circulation may be derived from radiative cooling in the free atmosphere of the subtropics. Because the radiative cooling in the subsidence area is correlated with the OLR in the Hadley circulation region, the strength of the Hadley circulation is strongly linked to the behaviors of upper ice clouds. In fact, the Hadley circulation weakens when radiative cooling decreases in most cases. An exception was found in the case where the snow fall speed was reduced. There, the Hadley circulation became stronger when the radiative cooling was decreased. We speculate that the frozen particles, which were falling more slowly, tended to evaporate in upper layers and that this additional evaporative cooling enhanced the circulation. Our sensitivity studies showed that, in general, as the coverage of the upper ice clouds became larger, the Hadley circulation became stronger. The coverage of the upper ice clouds was more closely related to optically active thin clouds.
This study gives a new perspective on circulation change due to global warming. Some studies emphasize the relative importance of the boundary layer humidity and moist adiabatic lapse rate on the change of the circulation due to global warming (Satoh and Hayashi 1992; Sugi et al. 2002; Held and Soden 2006; Vecchi and Soden 2007). However, the precise behavior of upper ice clouds also has impacts on the Hadley circulation. Because clouds are the most ambiguous factors in climate sensitivity (Soden and Held 2006), new models with cloud microphysics schemes should give different or improved sensitivity over current GCMs (Collins and Satoh 2009). To improve climate sensitivity further, we need to study upper ice clouds in greater detail through sensitivity studies like that conducted here.
Acknowledgments
We are grateful to the NICAM development members for helpful discussions throughout this study. We also thank the anonymous reviewers for useful comments. This research was supported by the Core Research for Evolutional Science and Technology (CREST) program of the Japan Science and Technology Corporation (JST) and by the Innovation Program of Climate Projection for the 21st Century of the Ministry of Education, Culture, Sports, Science, and Technology (MEXT). The Earth Simulator at the Japan Agency for Marine-Earth Science and Technology was used for the simulations.
REFERENCES
Cess, R. D., and Coauthors, 1990: Intercomparison and interpretation of climate feedback processes in 19 atmospheric general circulation models. J. Geophys. Res., 95, 16 601–16 615.
Collins, W. D., and M. Satoh, 2009: Simulating global clouds—Past, present, and future. Clouds in the Perturbed Climate System: Their Relationship to Energy Balance, Atmospheric Dynamics, and Precipitation, J. Heintzenberg and R. J. Charlson, Eds., Strüngmann Forum Reports, Vol. 2, MIT Press, 469–486.
Durran, D. R., and J. B. Klemp, 1982: On the effects of moisture on the Brunt–Väisälä frequency. J. Atmos. Sci., 39, 2152–2158.
Grabowski, W. W., 1998: Toward cloud resolving modeling of large-scale tropical circulations: A simple cloud microphysics parameterization. J. Atmos. Sci., 55, 3283–3298.
Grabowski, W. W., 2001: Coupling cloud processes with the large-scale dynamics using the cloud-resolving convection parameterization (CRCP). J. Atmos. Sci., 58, 978–997.
Hasumi, H., and S. Emori, Eds., 2004: K-1 coupled GCM (MIROC) description. CCSR, University of Tokyo Tech. Rep. 1, 34 pp.
Held, I. M., and A. Y. Hou, 1980: Nonlinear axially symmetric circulations in a nearly inviscid atmosphere. J. Atmos. Sci., 37, 515–533.
Held, I. M., and B. J. Soden, 2006: Robust responses of the hydrological cycle to global warming. J. Climate, 19, 5686–5699.
Iga, S., H. Tomita, Y. Tsushima, and M. Satoh, 2007: Climatology of a nonhydrostatic global model with explicit cloud processes. Geophys. Res. Lett., 34, L22814, doi:10.1029/2007GL031048.
Jakob, C., 2002: Ice clouds in numerical weather prediction models: Progress, problems, and prospects. Cirrus, D. K. Lynch et al., Eds., Oxford University Press, 327–345.
Khairoutdinov, M. F., and D. A. Randall, 2001: A cloud resolving model as a cloud parameterization in the NCAR Community Climate System Model: Preliminary results. Geophys. Res. Lett., 28, 3617–3620.
Louis, J., 1979: A parametric model of vertical eddy fluxes in the atmosphere. Bound.-Layer Meteor., 17, 187–202.
Mellor, G. L., and T. Yamada, 1974: A hierarch of turbulence closure models for planetary boundary layers. J. Atmos. Sci., 31, 1791–1806.
Miura, H., H. Tomita, T. Nasuno, S. Iga, M. Satoh, and T. Matsuno, 2005: A climate sensitivity test using a global cloud resolving model under an aqua planet condition. Geophys. Res. Lett., 32, L19717, doi:10.1029/2005GL023672.
Miura, H., M. Satoh, H. Tomita, A. T. Noda, T. Nasuno, and S. Iga, 2007: A short-duration global cloud-resolving simulation with a realistic land and sea distribution. Geophys. Res. Lett., 34, L02804, doi:10.1029/2006GL027448.
Miura, H., M. Satoh, and M. Katsumata, 2009: Spontaneous onset of a Madden-Julian oscillation event in a cloud-system-resolving simulation. Geophys. Res. Lett., 36, L13802, doi:10.1029/2009GL039056.
Nakajima, T., M. Tsukamoto, Y. Tsushima, A. Numaguti, and T. Kimura, 2000: Modeling of the radiative process in an atmospheric general circulation model. Appl. Opt., 39, 4869–4878.
Nakanishi, M., and H. Niino, 2006: An improved Mellor-Yamada level-3 model: Its numerical stability and application to a regional prediction of advection fog. Bound.-Layer Meteor., 119, 397–407.
Nasuno, T., H. Miura, M. Satoh, A. T. Noda, and K. Oouchi, 2009: Multi-scale organization of convection in a global numerical simulation of the December 2006 MJO event using explicit moist processes. J. Meteor. Soc. Japan, 87, 335–345.
Noda, A., K. Oouchi, M. Satoh, H. Tomita, S. Iga, and Y. Tsushima, 2010: Importance of the subgrid-scale turbulent moist process: Cloud distribution in global cloud-resolving simulations. Atmos. Res., 96, 208–217.
Randall, D. A., M. F. Khairoutdinov, A. Arakawa, and W. W. Grabowski, 2003: Breaking the cloud parameterization deadlock. Bull. Amer. Meteor. Soc., 84, 1547–1564.
Sarachik, E. S., 1978: Tropical sea surface temperature: An interactive one-dimensional atmosphere-ocean model. Dyn. Atmos. Oceans, 2, 455–469.
Sato, T., H. Miura, M. Satoh, Y. N. Takayabu, and Y. Wang, 2009: Diurnal cycle of precipitation in the tropics simulated in a global cloud-resolving model. J. Climate, 22, 4809–4826.
Satoh, M., 1994: Hadley circulations in radiative–convective equilibrium in an axially symmetric atmosphere. J. Atmos. Sci., 51, 1947–1968.
Satoh, M., and Y.-Y. Hayashi, 1992: Simple cumulus models in one-dimensional radiative convective equilibrium problems. J. Atmos. Sci., 49, 1202–1220.
Satoh, M., and Y. Matsuda, 2009: Statistics on high-cloud areas and their sensitivities to cloud microphysics using single-cloud experiments. J. Atmos. Sci., 66, 2659–2677.
Satoh, M., T. Matsuno, H. Tomita, H. Miura, T. Nasuno, and S. Iga, 2008: Nonhydrostatic Icosahedral Atmospheric Model (NICAM) for global cloud resolving simulations. J. Comput. Phys., 227, 3486–3514.
Satoh, M., T. Inoue, and H. Miura, 2010: Evaluations of cloud properties of global and local cloud system resolving models using CALIPSO/CloudSat simulators. J. Geophys. Res., 115, D00H14, doi:10.1029/2009JD012247.
Schneider, E. K., 1977: Axially symmetric steady-state models of the basic state for instability and climate studies. Part II: Nonlinear calculations. J. Atmos. Sci., 34, 280–297.
Sekiguchi, M., and T. Nakajima, 2008: A k-distribution-based radiation code and its computational optimization for an atmospheric general circulation model. J. Quant. Spectrosc. Radiat. Transfer, 109, 2779–2793.
Soden, B. J., and I. M. Held, 2006: An assessment of climate feedbacks in coupled ocean–atmosphere models. J. Climate, 19, 3354–3360.
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.
Tanaka, K. K., T. Yamamoto, S. Watanabe, and K. Nakajima, 2008: Analytic model of upper tropospheric clouds in tropical Hadley cell. Earth Planets Space, 60, 219–228.
Tao, W.-K., and Coauthors, 2009: A multiscale modeling system: Developments, applications, and critical issues. Bull. Amer. Meteor. Soc., 90, 515–534.
Tomita, H., and M. Satoh, 2004: A new dynamical framework of nonhydrostatic global model using the icosahedral grid. Fluid Dyn. Res., 34, 357–400.
Tomita, H., H. Miura, S. Iga, T. Nasuno, and M. Satoh, 2005: A global cloud-resolving simulation: Preliminary results from an aqua planet experiment. Geophys. Res. Lett., 32, L08805, doi:10.1029/2005GL022459.
Tsushima, Y., and Coauthors, 2006: Importance of the mixed-phase cloud distribution in the control climate for assessing the response of clouds to carbon dioxide increase: A multi-model study. Climate Dyn., 27, 113–126, doi:10.1007/s00382-006-0127-7.
Vecchi, G. A., and B. J. Soden, 2007: Global warming and the weakening of the tropical circulation. J. Climate, 20, 4316–4340.
Waliser, D. E., and Coauthors, 2009: Cloud ice: A climate model challenge with signs and expectations of progress. J. Geophys. Res., 114, D00A21, doi:10.1029/2008JD010015.
Wu, D. L., and Coauthors, 2009: Comparisons of global cloud ice from MLS, CloudSat, and correlative data sets. J. Geophys. Res., 114, D00A24, doi:10.1029/2008JD009946.
Wyant, M. C., M. Khairoutdinov, and C. S. Bretherton, 2006: Climate sensitivity and cloud response of a GCM with a superparameterization. Geophys. Res. Lett., 33, L06714, doi:10.1029/2005GL025464.
