15 SEPTEMBER 2009

SATO ET AL.

4809

Diurnal Cycle of Precipitation in the Tropics Simulated in a Global Cloud-Resolving Model TOMONORI SATO* Center for Climate System Research, University of Tokyo, Chiba, Japan

HIROAKI MIURA Colorado State University, Fort Collins, Colorado, and Frontier Research Center for Global Change, Japan Agency for Marine-Earth Science and Technology, Tokyo, Japan

MASAKI SATOH Center for Climate System Research, University of Tokyo, Chiba, and Frontier Research Center for Global Change, Japan Agency for Marine-Earth Science and Technology, Tokyo, Japan

YUKARI N. TAKAYABU Center for Climate System Research, University of Tokyo, Chiba, and Institute of Observational Research for Global Change, Japan Agency for Marine-Earth Science and Technology, Tokyo, Japan

YUQING WANG International Pacific Research Center, and Department of Meteorology, University of Hawaii at Manoa, Honolulu, Hawaii (Manuscript received 14 October 2008, in final form 11 March 2009) ABSTRACT This study analyzes the diurnal cycle of precipitation simulated in a global cloud-resolving model (GCRM) named the Nonhydrostatic Icosahedral Atmospheric Model (NICAM). A 30-day integration of NICAM successfully simulates the precipitation diurnal cycle associated with the land–sea breeze and the thermally induced topographic circulations as well as the horizontal propagation of diurnal cycle signals. The first harmonic of the diurnal cycle of precipitation in the 7-km run agrees well with that from satellite observations in its geographical distributions although its amplitude is slightly overestimated. The NICAM simulation revealed that the precipitation diurnal cycle over the Maritime Continent is strongly coupled with the land–sea breeze that controls the convergence/divergence pattern in the lower troposphere around the islands. The analysis also suggests that the cold pool often forms over the open ocean where the precipitation intensity is high, and the propagation of the cold pool events is related to the precipitation diurnal cycle as well as the land–sea breeze. Sensitivity experiments suggest a prominent horizontal resolution dependence of the simulated precipitation diurnal cycle. Over continental areas the 14-km run induces the diurnal peak about three hours later than the 7-km run. The 3.5-km run produces the peak time and amplitude that are very similar to those in Tropical Rainfall Measuring Mission (TRMM) precipitation radar (PR) observations. Meanwhile, the resolution dependence in phase and amplitude is negligibly small over the open oceans. This contrast sensitivity to the horizontal resolution is attributed to the differences in structure and life cycle of convective systems over land and ocean. Diurnal peaks of precipitable water vapor, precipitation, and outgoing longwave radiation (OLR) are compared over land areas using the NICAM 7-km run. The daily precipitable water vapor maximum appears around 1500 local time (LT), which is followed by the precipitation peak around 1630 LT. The diurnal

* Current affiliation: Faculty of Environmental Earth Science, Hokkaido University, Sapporo, Japan.

Corresponding author address: Tomonori Sato, Faculty of Environmental Earth Science, Hokkaido University, Kita-10, Nishi-5, Sapporo, 060-0810, Japan. E-mail: [email protected] DOI: 10.1175/2009JCLI2890.1 ! 2009 American Meteorological Society

4810

JOURNAL OF CLIMATE

VOLUME 22

cycle of high clouds tends to peak around 1930 LT, three hours later than the precipitation peak. These results from NICAM simulations can explain the cause of the phase differences among precipitation products based on several satellite observations. The authors demonstrate that the GCRM is a promising tool for realistically simulating the precipitation diurnal cycle and could be quite useful for studying the role of the diurnal cycle in the climate systems in a global context.

1. Introduction Convective heating is extremely important for the maintenance of global-scale atmospheric circulation. The diurnal cycle of precipitation is a prominent mode in the tropical convective systems; thus, it has often been a target of intensive observations in the tropics (Houze et al. 1981; Skinner and Tapper 1994; Mapes et al. 2003a,b; Zuidema 2003; Mori et al. 2004). The spaceborne observation established by the Tropical Rainfall Measuring Mission (TRMM) satellite has contributed significantly to our understanding of the structure and behavior of convective activity by direct sounding using precipitation radar on board. Additionally, the TRMM’s non-sunsynchronous orbit allows us to investigate the diurnal change of convective systems in the tropical zone with horizontally uniform coverage (e.g., Nesbitt and Zipser 2003; Hirose and Nakamura 2005). As a result, TRMM and associated products are often adopted as a reference for evaluation of general circulation models (GCMs) (Lin et al. 2000, 2002; Collier and Bowman 2004; Arakawa and Kitoh 2005; Takayabu and Kimoto 2008). Dai (2006) evaluated the diurnal cycle in GCMs using multiple observation datasets and suggested that the modeled diurnal cycle is too early in the precipitation peak and too strong in the amplitude than those observed even in the latest generation GCMs. A poor description of the diurnal cycle in the tropics in GCMs may be a matter of concern for predicting tropical climate variability (Wang et al. 2007). Previous studies (Neale and Slingo 2003; Lorenz and Jacob 2005) suggested the importance of the diurnal cycle over the Maritime Continent for simulating the global atmospheric circulation. Neale and Slingo (2003) showed that the heating anomaly over the Maritime Continent has significant impacts on the atmospheric circulation over North America and northeast Eurasia as a result of Rossby wave response. Therefore, considerable improvements are necessary for realistically simulating the diurnal cycle of precipitation by GCMs, although the diurnal cycle in the tropics is a kind of local phenomenon. Conventional GCMs generally adopt cumulus parameterization to describe subgridscale convection. As mentioned above, however, most GCMs have difficulties replicating the phase and amplitude of the diurnal cycle, particularly over land. They have limitations in simulating the propagation of the di-

urnal cycle in which mesoscale circulations play a significant role. A lot of efforts have been made to improve the cumulus parameterization for more realistic simulation of the diurnal cycles in GCMs (Zhang 2002; Wang et al. 2007; Takayabu and Kimoto 2008). The precipitation diurnal cycles are characterized by two major types according to the local time of the maximum precipitation, namely, the afternoon peak over land and the early morning peak over ocean. In the former, daytime evolution of the mixed layer as a result of surface solar heating is closely related to the establishment of both an unstable layer above the surface and thermally induced horizontal circulations, such as upslope wind and sea breeze (Zhou and Wang 2006). Gravity currents (Satomura 2000) and gravity waves (Mapes et al. 2003b) forced by convection in remote areas also contribute to triggering convection but with a phase delay. Over the ocean, the precipitation maximum tends to occur in the early morning hours since radiative cooling near the cloud top destabilizes the vertical stratification in clouds (Tao et al. 1996; Sui et al. 1998). Other mechanisms have also been suggested to explain the early morning peak over ocean, such as the diurnal change in sea surface temperature (SST) (Johnson et al. 2001). Migration of the evening convection over land also affects the morning peak of precipitation in offshore areas (Houze et al. 1981; Mori et al. 2004; Zhou and Wang 2006); thus, the mechanism responsible for the early morning peak is quite complicated over the ocean. There are three possible approaches to improve the simulation of the precipitation diurnal cycle in global models. The first method is to increase the horizontal resolution of conventional GCMs. Arakawa and Kitoh (2005) performed a 20-km mesh run using a cumulus parameterization in which the diurnal cycle over the Maritime Continent was reproduced fairly well. They revealed that the realistic land–sea breeze and the detailed topographies over land are represented well in the high-resolution GCM, which contributes to the improved behavior of the modeled diurnal cycle. In the central area of large islands, such as Borneo, however, the peak time of the diurnal cycle seems too early compared with the observations. Inland migration of convection due to the sea breeze front is essential for convective amplification in the central region of the island (Saito et al. 2001; Qian 2008). Therefore, it is

15 SEPTEMBER 2009

SATO ET AL.

inferred that the mesoscale circulation associated with the coastal convection could not be properly captured even in the super-high-resolution GCMs. In another case study by Lee et al. (2007), large errors in phase and amplitude of the diurnal cycle remain in the approximately 0.58 GCM although the diurnal cycle approaches observations as the horizontal resolution increases. The second method for improving the diurnal cycle is to adopt a multiscale modeling framework (MMF or superparameterization), in which cumulus parameterization in a coarse-resolution GCM box is replaced by a two-dimensional cloud-resolving model (Khairoutdinov and Randall 2001; Grabowski 2001). The MMF has the advantage of efficiently dealing with cloud–radiation feedback and boundary layer processes in the coarseresolution GCM without too huge a computational cost. Recently, some studies have focused on the diurnal cycle in MMF (Khairoutdinov et al. 2005, 2008; Chern et al. 2006; DeMott et al. 2007; Zhang et al. 2008; Tao et al. 2009). Khairoutdinov et al. (2005) showed the improved diurnal variability of precipitation in the National Center for Atmospheric Research (NCAR) Community Atmosphere Model using superparameterization (SP-CAM) in comparison with the standard CAM. Chern et al. (2006) found that the diurnal variability and frequency in the Goddard-MMF model is superior to the Goddard finite-volume GCM (fvGCM). Zhang et al. (2008) successfully simulated the diurnal cycle of surface precipitation in their MMF. These previous studies suggest that the MMF can reduce precipitation and latent heating biases associated with the diurnal cycle in the original GCM; thus, it is a promising tool for bypassing the cumulus parameterizations (Arakawa 2004). By using a high-resolution GCM or MMF, the diurnal cycle of tropical convection is much improved. However, they have limitations for simulating the diurnal cycle signals that propagate far from the original forcing by the interaction with the mesoscale processes, such as cold pool formation and squall lines. Therefore, explicit simulation of the mesoscale circulation is a key issue for improving the diurnal cycle propagations. The third method, which we adopt in this study, is to use the global cloudresolving (or cloud-system-resolving) model (GCRM or GCSRM). The GCRM with a grid size of several kilometers directly simulates cumulus convection with global coverage using a cloud microphysics scheme without the use of any cumulus parameterization. Although the computational cost is extremely high, it is expected to reduce uncertainties from assumptions used in cumulus parameterizations. Furthermore, its global coverage allows us to examine the mechanisms of the diurnal cycle over areas under different large-scale environmental conditions, which could greatly enhance our understanding of the be-

4811

haviors of convective systems. The high resolution over coastal and inland areas enables us to resolve local circulations in association with the land–sea contrast and topographic complexity. Additionally, mesoscale circulations that originate from the convective systems are explicitly simulated in the GCRM, which should improve the propagation of the diurnal cycle signals in some regions. This is indeed the advantage of using GCRM in the diurnal cycle studies. Recent studies have shown advantages of using the GCRM for investigating tropical phenomena. Miura et al. (2007b) succeeded in reproducing the development of a tropical cyclone in GCRM with realistic land–sea and topographic distributions. Miura et al. (2007a) showed a realistic simulation of an eastward-propagating Madden–Julian oscillation (Madden and Julian 1972) event by a 7-km mesh GCRM. Currently, a 31-day-long integration with a 7-km mesh GCRM run becomes possible with state-of-the-art powerful computers; this may be long enough to study the diurnal cycle statistically. The purpose of this paper is to show the characteristics of the precipitation diurnal cycle and convective behaviors in the tropics simulated in a GCRM. The simulated diurnal cycle is evaluated using satellite observations. Additionally, we conducted some preliminary experiments to examine the sensitivity of the simulated diurnal cycle to the model horizontal resolution. We thus would discuss implications of those results to future studies using the GCRM in terms of the diurnal variability of convective systems. The model, experimental setup, and satellite data are described in section 2. Detailed comparison of the diurnal cycle in the GCRM and satellite observation is given in section 3. The behaviors of the diurnal cycle simulated with different model resolutions are discussed in section 4. Characteristics of the simulated diurnal cycle in some selected areas of special interests are given in section 5. Some possible applications to the cloud microphysics study are addressed in section 6. A summary of our findings and future perspectives are presented in section 7.

2. Method a. Model This study uses the global cloud-resolving Nonhydrostatic Icosahedral Atmospheric Model (NICAM), which is developed at the Frontier Research Center for Global Change (FRCGC) and Center for Climate System Research (CCSR) at the University of Tokyo (Tomita and Satoh 2004; Satoh et al. 2008). Sato et al. (2007) first showed the realistic simulation of the diurnal cycle of convective activity over the Tibetan Plateau by NICAM. One of the benefits of NICAM is its ability to simulate convective systems explicitly in the tropics

4812

JOURNAL OF CLIMATE

where the mesoscale convection is fundamental to the organization of multiscale convective systems. Miura et al. (2007a) discussed a realistic simulation of an eastwardpropagating Madden–Julian oscillation event that occurred in December 2006. This study analyzes Miura’s experiment with special emphasis on the diurnal cycle of convective systems in the tropics. The model framework and the experimental setting of the simulation are as follows. The NICAM adopts the icosahedral horizontal grid system that covers the globe quasi-homogeneously. Its governing equations describe the nonhydrostatic, fully compressible atmosphere and thus permit acoustic waves. The model has 40 vertical layers with a terrain-following grid system and with a model top at 38 km; the lowest layer thickness is 162 m. Nine grid points are located within the boundary layer (e.g., below 2000 m). The cloud microphysics scheme of Grabowski (2001) is used with no cumulus parameterization. Over land, we use a simple bucket model in which the initial soil moisture is given by the National Centers for Environmental Prediction (NCEP) final analysis on 1.08 3 1.08 grids. More details about the model configurations can be found in Satoh et al. (2008). The numerical experiments were performed for one month starting at 15 December 2006. Initial atmospheric conditions were obtained from the NCEP tropospheric analysis (with 1.08 by 1.08 mesh). Spatial and temporal variations of SST were interpolated from the weekly mean data of Reynolds SST (Reynolds and Smith 1994). To evaluate the resolution dependency of the diurnal cycle, we performed three experiments by changing the horizontal resolution in NICAM, namely, 3.5-, 7-, and 14-km mesh runs (hereafter NICAM-3.5km, -7km, and -14km, respectively). A 31-day integration was performed for the NICAM-7km and -14km runs, while a 7-day integration from 25 December 2006 was conducted for the NICAM-3.5km run. In this study we mainly focus on the NICAM-7km run and compare the simulated diurnal cycle with satellite observations. The sensitivity to the model horizontal resolution is discussed in section 5. Every 1.5-h output from the NICAM integration was used in the analyses.

b. Satellite data TRMM observations were used to evaluate the simulated diurnal cycle. We adopted the TRMM 3G68 (hereafter 3G68) product, which is a 0.58 3 0.58 gridded dataset based on precipitation radar (PR) observation onboard the TRMM satellite. The analysis period is December–February from 1998 through 2007 to obtain the mean diurnal cycle with 3-h intervals. Additionally, we use the TRMM 3B42 (hereafter 3B42) product, which

VOLUME 22

is 3-hourly precipitation estimates on 0.258 3 0.258 grids using microwave channel data as well as infrared data from geostationary satellites. The 3B42 product has 3-hourly global coverage so we are able to compare NICAM results for the identical period, namely from 15 December 2006 through 14 January 2007. As pointed out in the previous study, there is a time lag of the precipitation peak between the 3B42 and 3G68 products (e.g., Kikuchi and Wang 2008). Because the PR beam penetrates into the clouds and directly observes the precipitation, we consider the data from 3G68 to be more reliable than 3B42. However, we also use the 3B42 product for validation since the PR swath is very narrow and thus multiyear sampling is necessary. We mention the characteristics of diurnal phase differences among the multiyear 3G68, the multiyear 3B42, and the one-month 3B42 in the next section. Additionally, we will discuss causes of the discrepancy between the two precipitation products in section 6 based on the results from NICAM simulations.

3. Diurnal cycle of precipitation In this section, we present the diurnal cycle of precipitation in a one-month simulation in the NICAM-7km run. The simulated characteristics of the diurnal cycle are evaluated using the satellite observations. Figure 1 shows the local time of the precipitation peak in TRMM and the NICAM-7km run computed from the first harmonic of the composite diurnal cycle. NICAM-7km well captures the geographical distribution of the local time in the precipitation peak. In the major continental regions, such as South Africa, Australia, and South America, both the NICAM-7km and satellite observations indicate the afternoon peak of precipitation. Both the NICAM-7km and 3B42 results show the peaks later than 3G68, in particular over land. The phase difference between 3B42 and 3G68 is consistent with that reported in Kikuchi and Wang (2008). For example, 3G68 shows the diurnal peak around local noon over South America, whereas in NICAM-7km and 3B42 the diurnal peak occurs in early evening. Therefore, NICAM-7km shows too late a peak in precipitation over South America compared to 3G68. In contrast, conventional GCMs showed too early a peak over land (Yang and Slingo 2001; Dai 2006). Over other land areas, the phase difference is not so prominent between NICAM-7km and 3G68. Peak time of the convection over land areas is very similar between 11-yr (Fig. 1b) and single-year (Fig. 1c) statistics. Therefore, the studied period can be a representative in terms of the diurnal phase distributions. The NICAM-7km well reproduces the exceptional diurnal cycle that does not follow the global mean phase characteristics, that is, the

15 SEPTEMBER 2009

SATO ET AL.

4813

FIG. 1. Local time of the precipitation maximum computed as the first harmonic of the diurnal cycle in December–February over 1998– 2007 by (a) 3G68 and (b) 3B42. (c) As in (b) but for the period from 15 Dec 2006 through 14 Jan 2007. (d) As in (c) but for the NICAM-7km run. Precipitation intensity less than 0.1 (mm h21) is whitened. The NICAM-7km run is rescaled to 0.258 grids for comparison.

evening peak over land and morning peak over ocean, because of the propagation of the diurnal cycle signals. The phase of offshore propagation from the western coast of Central America and the western coast of Africa is very similar to the TRMM observations and previous studies (Mapes et al. 2003a; Gray and Jacobson 1977).

The horizontal distribution of the local time in the diurnal precipitation peak presents more complex features over oceans. Near the intertropical convergence zone (ITCZ) and the South Pacific convergence zone (SPCZ), where the total precipitation amount is large, the diurnal phase is quite different from region to region, which is a common feature both in NICAM

4814

JOURNAL OF CLIMATE

VOLUME 22

FIG. 2. As in Fig. 1 but for the diurnal amplitude.

simulations and observations. On the other hand, the NICAM-7km run shows a clear early morning peak around 0300–0600 local time (LT) over the eastern ocean basins where the midtroposphere in the subtropics is subject to large-scale subsidence, that is, over the southeast Pacific, southeast Atlantic, and southeast Indian Ocean. Since the drizzle from marine stratocumulus clouds is a major contributor to the total rainfall in these areas (Wang et al. 2004), the precipitation is

hardly detected by the TRMM PR. The stripes over the North Pacific in both 3B42 and NICAM indicate the passage of precipitation events due to synoptic-scale disturbances, and they may indicate the diurnal cycle signal. A one-month integration is too short to eliminate such influence in the extratropics. Figure 2 shows the amplitude of the precipitation diurnal cycle in the NICAM-7km run and TRMM products. The amplitude is larger in 3B42 than in 3G68.

15 SEPTEMBER 2009

SATO ET AL.

4815

averaged in the tropics (158S–158N) is about 3 times the observed from 3G68 and about twice that from 3B42. This could be partly due to the one-month simulation during which the precipitation is well above the multiyear average used in TRMM 3G68 products. In this section, we found that the NICAM simulates fairly well the diurnal cycle of precipitation in the tropics. Nevertheless, the precipitation peak tends to occur slightly later in NICAM-7km than in 3G68, especially over land. Since the 7-km mesh grid is not always fine enough to resolve cumulus convection, a short-term experiment with higher resolution is shown in the next section along with a discussion on the sensitivity of the simulated diurnal cycle to the model grid size. FIG. 3. Diurnal cycle of the precipitation intensity anomaly in 3G68 (filled triangles), 3B42 (open triangles), and NICAM-7km run (crosses) averaged over 158S–158N. Red lines indicate land grids and blue lines indicate ocean grids. The 3B42 run is based on the one-month analysis covering the simulation period.

Additionally, during the period from December 2006 to January 2007, the amplitude of the diurnal cycle (Fig. 2c) is larger than the climatological mean (Fig. 2b). This could be due to the interannual variation of the diurnal cycle and the modification of the diurnal cycle in relation to intraseasonal variation. The amplitude in NICAM is generally larger over both land and ocean than in the two observational products. Geographical distributions are, however, well captured in the NICAM-7km run. For instance, the local maxima of the diurnal amplitude over islands in the Maritime Continent and Madagascar are evident in the NICAM simulation. To examine the NICAM results more quantitatively, Fig. 3 shows the time series of the area-averaged diurnal cycle. The peak time of the diurnal precipitation is at 0600 LT over ocean in NICAM and at 0300 LT in the 3G68 product. The amplitude over ocean is in good agreement with observations. Over land areas, some systematic biases are evident in NICAM. The 3G68 indicates afternoon peak around 1500 LT, while the peak occurs at 1630 LT in NICAM, namely, about 1.5 h later than the PR observation. However, the peak in 3B42 occurs at 1800 LT, about 3 h later than the 3G68. This delay is mainly a result of the retrieval algorithm with the use of different measurements (see further discussion in section 6). The local time of the precipitation minimum also shows a phase difference between observations and the NICAM simulation. TRMM products show the minimum precipitation at 0900 LT, while NICAM shows it at 1200 LT. The precipitation begins to increase around 1200 LT in the observations; therefore, the initiation of convective systems is delayed considerably in NICAM. The amplitude of the diurnal cycle

4. Sensitivity to model grid size Previous studies using mesoscale models have shown that the life cycle and maximum intensity of convective systems could strongly depend on the model horizontal resolution (Weisman et al. 1997; Petch et al. 2002; Sato et al. 2008). It is interesting to examine how sensitive the phase and amplitude of the simulated diurnal cycle in NICAM could be to the horizontal grid size. Here, we investigate the sensitivity of the simulated diurnal cycle to the model horizontal resolution. Three runs with different grid sizes (NICAM-14km, -7km, and -3.5km) are compared. As described in section 2, the NICAM-3.5km run covers only a 7-day period, whereas NICAM-14km and -7km runs cover 31 days in total. We confirmed that the main results on the resolution sensitivity do not change significantly if the same period is analyzed for the three runs. Figure 4c shows the horizontal distribution of the time lag of the diurnal precipitation maximum between the NICAM-14km and -7km runs. Over major land areas, the peak time generally occurs earlier in the NICAM-7km run than in the NICAM-14km run. Landlocked areas in South America and Africa show more complex time lag patterns between the coarse- and fine-resolution runs because of the propagation of mesoscale convective systems (MCSs), which makes it difficult to distinguish MCSs generated locally from those propagated from other regions. Over ocean, the spatial distribution of the time lag is also very complicated. The local time of the maximum precipitation in the NICAM-7km run evidently shows the geographical contrast in the tropics (158S–equator) over Africa, the Maritime Continent, and South America (Fig. 4a). Meanwhile, the peak time in the NICAM-14km run tends to be delayed by several hours over these regions; namely, the late evening peak over land is shifted to early morning. The amplitude of the NICAM simulated diurnal cycle in the tropics is

4816

JOURNAL OF CLIMATE

VOLUME 22

FIG. 4. (a) Local time of the precipitation maximum and (b) amplitude of the diurnal cycle averaged over 158S–08. Shading, red circles, and blue crosses show one-month statistics from 3B42, NICAM-7km, and NICAM-14km, respectively. Regions where amplitude is less than 0.02 mm h21 are whitened in (a). (c) Time lag of the diurnal precipitation maximum between the NICAM-14km run and NICAM-7km run. Negative values indicate that the peak time in the NICAM-7km is earlier than in the NICAM-14km run.

larger than that in the same month statistics based on 3B42 in the continental areas and the western Pacific (Fig. 4b). The diurnal time series of the precipitation anomaly in the tropics from the three resolution runs are compared in Fig. 5 in which land and ocean domains are averaged separately. Over ocean, the phase and amplitude differences are very small, indicating that the diurnal cycle is less affected by the model grid size over ocean. In contrast, the NICAM-14km run shows a diurnal peak in precipitation at 1930 LT over land, which is 3 h later than that in the NICAM-7km run. The NICAM-3.5km run produces peak time and amplitude very similar to those in 3G68. In general, the coarser the horizontal resolution is, the later the diurnal phase will be. This property is very similar to previous findings using cloud-resolving models focusing on the mesoscale convections (Petch et al. 2002; Sato et al. 2008). Inoue et al. (2008) also revealed that the area of cloud cover becomes much

closer to the observed as the model mesh size becomes smaller. Therefore, to reproduce appropriate global distributions of the phase and amplitude of the diurnal cycle, the fine horizontal resolution with the mesh size in the order of several kilometers may be required for a GCRM. The prominent resolution dependence of the simulated diurnal cycle over land, as shown in Fig. 5, is consistent with the well-known features of the convective development in model simulations. As evident in Figs. 4 and 5, the resolution dependence is more pronounced over land than over ocean. To understand the difference between ocean and land domains, we revisit the nature of convective systems over both domains. The convective systems over land that exhibit the evident diurnal cycle are generally composed of deep convection. Takayabu (2006) found the higher flush per rain-yield ratio over land areas from 3-yr TRMM observations and indicated that the tall cumulus convection is dominant over land. In addition, the convective

15 SEPTEMBER 2009

SATO ET AL.

FIG. 5. Diurnal cycle of the precipitation intensity anomaly in the NICAM-3.5km (circles), NICAM-7km (crosses), NICAM-14km (squares), and 3G68 (filled triangles) runs averaged over 158S–158N. Red lines indicate land grids and blue lines indicate ocean grids.

systems related to the diurnal cycle over land tend to be transformed from shallow cumulus clouds near the top of the mixed layer to well-developed cumulonimbus clouds within the diurnal time scale. At the initial stage of the daytime convection, the shallow cumulus clouds and associated vertical circulation are not resolved by a coarse mesh size. As a result, vertical moisture transport by the grid-scale upward motion is not sufficient, inducing the phase delay in convective initiation (Petch et al. 2002; Sato et al. 2008). Therefore, the peak time and initiation of convection over land tend to be delayed compared to observations. On the other hand, convective systems over the open ocean, far away from the coastlines, often behave as well-organized features, such as MCSs and cloud clusters. The lifetime of such organized convective systems is generally longer than a day and influenced by large-scale atmospheric conditions. The convective systems over ocean undergo active and inactive phases because of the longwave/shortwave radiative forcing. Such organized systems can be somehow resolvable by NICAM-14km, and thus the diurnal cycle is less sensitive to the horizontal resolution. Near the coastlines, however, the resolution dependence is not negligible because the diurnal cycle is strongly affected by the land–sea breeze and the propagation of convective systems that is generated over land. To examine the phase difference along the coastal areas, longer integration is needed to collect enough samples of the convective systems.

5. Regional characteristics In the previous section, we discussed the precipitation diurnal cycle in NICAM as a global averaged view in the

4817

tropics and compared it with the TRMM observations. Here, we focus on regional characteristics of the precipitation diurnal cycle. The horizontal pattern of the simulated precipitation diurnal cycle and the associated atmospheric circulation in several regions are discussed in this section. The result from NICAM-7km run is mainly used. The diurnal cycle over the Maritime Continent in NICAM-3.5km run is available in Satoh (2008). The diurnal change of precipitation intensity over the Maritime Continent is shown in Fig. 6. The most predominant characteristics, the early morning peak over ocean and the afternoon peak over land, are simulated well in the NICAM-7km run. During 0600–0900 LT, the precipitation intensity over land becomes low, whereas, over ocean, the precipitation intensity increases along the coastlines. This pattern remains similar until 1200 LT; however, it changes dramatically at 1500 LT. Precipitation intensity over land increases during 1200–1500 LT, in particular in the coastal areas over islands. Meanwhile, precipitation intensity is decaying over oceans. During 1500–1800 LT, the rapid increase in precipitation intensity becomes evident over large islands, such as New Guinea, Sumatra, and Borneo. The exception is near the central regions of Borneo where the intensity is still weak. Precipitation intensity over the ocean is quite weak, giving rise to a remarkable land–ocean contrast. In the middle of the night around 2100–2400 LT, convective systems are only present in the central regions of large islands, such as New Guinea and Borneo, and remain intense until 0300 LT. Overall these characteristics are very similar to the TRMM observations. Hence, the NICAM-7km run well captures the diurnal cycle of precipitation over the Maritime Continent. Another marked diurnal change appears over the Cape York Peninsula in northeast Australia. The precipitation peaks around 1500 LT over the peninsula and propagates westward in the environmental easterly wind. The precipitation peaks at 0300 LT in the central regions of the Gulf of Carpentaria. Remarkable contrast of the diurnal cycle between land and ocean in the Maritime Continent is associated with the land–sea breeze (Houze et al. 1981; Saito et al. 2001) as well as the mountain slope flow (Zhou and Wang 2006). Satellite observations, such as TRMM, provide spatial distributions of precipitation features. However, the mesoscale atmospheric circulation, which is closely linked to the diurnal cycle, is difficult to observe by satellite remote sensing. Recently, Gille et al. (2005) estimated the characteristics of the land breeze using the SeaWinds scatterometers onboard Quick Scatterometer (QuikSCAT) and Advanced Earth Observing Satellite II (ADEOS-II) satellites. However, the temporal resolution is not high enough for studying the diurnal cycle, and

4818

JOURNAL OF CLIMATE

VOLUME 22

FIG. 6. Diurnal change of monthly mean precipitation intensity over the Maritime Continent simulated by the NICAM-7km run. (a) 0000, (b) 0300, (c) 0600, (d) 0900, (e) 1200, (f) 1500, (g) 1800, and (h) 2100 LT.

the technique is not applicable to land areas at all. GCRM has the advantage of providing those mesoscale flows globally because it can capture the three-dimensional flow fields with high temporal and spatial resolutions. Figure 7 depicts 2-m-height wind patterns, shown as the departure from the monthly mean wind, at two selected local times. At 1730 LT, the well-developed sea breeze is found around coastlines over islands. Inland coastal areas are characterized by prominent low-level convergence. Java is entirely covered by the low-level convergence due to northerly and southerly flows at northern and southern coastlines, respectively. On the other hand, the

offshore areas, particularly where surrounded by the islands, experience weak low-level divergence. In the morning (0530 LT), the land breeze is recognized over the offshore areas, resulting in a weak convergence zone over the coastal ocean. The convergence is locally intensified over the Java Sea and between the Borneo and Java islands because of the merging of two different land breeze systems: one from Java and the other from Borneo. This should contribute greatly to the intensification of precipitation in early morning over the ocean. From these results, the land–sea breeze system dominantly controls the diurnal phase and amplification of

15 SEPTEMBER 2009

SATO ET AL.

4819

FIG. 7. Monthly mean diurnal anomaly of 2-m-height wind (m s21) at (a) 1730 LT and (b) 0530 LT in the NICAM-7km run. Red and blue shadings indicate the divergence and convergence of the 2-m wind, respectively (1025 s21). Red box in (a) shows the area of the cross section in Fig. 8.

precipitation near the coastlines. The physical process that controls the precipitation diurnal cycle over the offshore areas is different from those of the morning peak over the open ocean (e.g., Sui et al. 1997; Dai 2001; Nesbitt and Zipser 2003) or in the aquaplanet experiment (Tomita et al. 2005). Furthermore, Fig. 7 also implies that the diurnal cycle is influenced by terrestrial heating. The diurnal change of convergence/divergence pattern that is associated with the land–sea breeze system seems to be confined within offshore areas, several hundred kilometers in maximum from the coastlines. Areas over the ocean far away from the coastline are little influenced by the land heating and thus can be regarded as open ocean. Although this criterion may

vary depending on the environmental conditions, it is roughly consistent with observations (Houze et al. 1981; Yang and Slingo 2001; Moteki et al. 2008). Over New Guinea, the diurnal cycle of low-level wind strongly controls the precipitation activity. Figure 8 shows the transects of 2-m-height wind and precipitation along the same cross section as shown in Zhou and Wang (2006). Sea breeze circulation is evident at southwest and northeast coastal areas in the early afternoon, which well corresponds to the coastal rainfall. The morning precipitation maximum in the offshore regions is associated with the land breeze. Upslope wind makes convergent flow near the mountain ridge and initiates afternoon convection. In the late afternoon to the early morning,

4820

JOURNAL OF CLIMATE

FIG. 8. Time cross section of (a) the diurnal anomalous 2-m wind (m s21) and (b) precipitation (mm h21) in the NICAM-7km run. Positive (negative) values in (a) indicate southwesterly (northeasterly) wind. Solid lines show southwest and northeast coastal lines of New Guinea, and the dashed lines show the mountain ridge. See also Zhou and Wang (2006).

the precipitation system over the mountaintop separates into two systems; both propagate downslope to the south and north coastal areas. Figure 9 shows the diurnal change of the precipitation pattern over South America. Kikuchi and Wang (2008) classified the diurnal cycle over Amazon as an inland type. The simulated precipitation indicates that the MCSs prevail over Amazon in accordance with the development of mixed layer in early afternoon (1330 LT). The most intensive precipitation appears in the evening (1930 LT) soon after the convective systems reach their mature stage. The westward propagation is clearly seen over inland Amazon similar to that shown in some previous studies (Kikuchi and Wang 2008; Rickenbach 2004; Takayabu and Kimoto 2008). A pronounced westwardpropagating system is found over the estuary of the Amazon River. The convective systems generated along the coastline at 1330 LT move inland by 200–300 km west at 1930 LT and almost cease in the middle of the night (0130 LT). The orographic effect by the Guiana highland (around 28N, 558W) may also contribute to the generation of the convective systems near the coastline.

VOLUME 22

The NICAM-7km run also captures well the propagating precipitation events over Central America, a feature previously studied by Mapes et al. (2003a,b). Figure 10 shows an east–west cross section of the monthly mean precipitation diurnal cycle around western Colombia. The westward propagation speed of the precipitation systems is approximately 15 m s21, which agrees well with both the observed and the modeled results by Mapes et al. (2003a,b). Therefore, the NICAM reproduces well the gravity wave propagation excited by the daytime planetary boundary layer heating and the associated precipitation diurnal cycle in this region. Finally, we present preliminary results on the convectionassociated mesoscale dynamics over ocean. Cold pool formation over the ocean is an essential process that triggers the generation of new convective cells and the organization of convective systems, such as cloud clusters (Tao 2007; Nasuno et al. 2008). Recent surface meteorological observations in open ocean become available owing to the rapid increasing number of buoy stations, such as the Triangle Trans-Ocean Buoy Network (TRITON; Kuroda 2002) and the Tropical Atmosphere Ocean Project (TAO; McPhaden 1995). However, the nature and geographical distribution of the cold pool events over the open ocean are still unknown. Even with the aid of satellite remote sensing, it is still difficult to observe mesoscale meteorological elements over the open ocean. Figure 11 shows 31-day statistics of the cold pool formation. Here, we estimate the number of the cold pool formation events by a very simple method. We regard the rapid temperature drop greater than 1.5 K (1.5 h)21 as a cold pool formation event. The threshold value is subjectively determined as the largest value that does not exceed the temperature change over ocean expected from the solar insolation alone. In spite of the very simple detection method, the horizontal distribution of the cold pool formation events seems to be reasonable. The higher counts of cold pools are found over the ocean where mean precipitation is great, such as in ITCZ and SPCZ. That means the cold pools are often generated over ocean areas where the convective activity is high. The formation of cold pools is a key process that maintains the larger-scale convective organization. The diurnal change of cold pool occurrence over the Maritime Continent is depicted in Fig. 12. In the evening (1800–2400 LT), high frequency is found in the offshore areas along the coastlines, indicating that the cold pools are formed in association with the late afternoon convection over the islands. The high frequency migrates toward the open ocean in early morning (0000–0600 LT). Consequently, frequency is high over the ocean surrounding the islands in the morning (0600–1200 LT). In addition, the westward propagation of the high cold

15 SEPTEMBER 2009

SATO ET AL.

4821

FIG. 9. Diurnal change of monthly mean precipitation intensity over South America simulated by the NICAM-7km run. (a) 0130, (b) 0730, (c) 1330, and (d) 1930 LT.

pool frequency is evident over the Gulf of Carpentaria and north Australia. Thus, we regard the westward propagation of the precipitation systems shown in Fig. 6 as maintained by the gravity current propagation. Over the central Pacific, the frequency seems to be higher in early morning (0000–0600 LT) than in other local hours. This is consistent with the precipitation maximum in early morning. To obtain more detailed diurnal variation, however, longer integration is needed.

6. Applications to the cloud microphysics study Studies of the diurnal cycle using a GCRM have potential to help improve the retrieval algorithms based on satellite observations (Shige et al. 2004). This study uses both TRMM 3B42 and 3G68 products to evaluate the NICAM simulations. As shown in Fig. 1 and in previous

studies (e.g., Rasch et al. 2006; Dai et al. 2007; Kikuchi and Wang 2008), 3B42 tends to show a later peak time than 3G68 or ground observations. Despite the discrepancies, some studies were using 3B42 to evaluate the modeled precipitation diurnal cycle (Dai 2006; Rasch et al. 2006) because of its merit in global coverage. Since the 3B42 uses an infrared channel from geostationary satellites to increase the spatial coverage, the peak time is strongly affected by the cloud cover rather than the precipitation; thus, it differs from the peak time of precipitation intensity. Figure 13 shows diurnal cycles of precipitable water vapor, precipitation, and outgoing longwave radiation (OLR; shown as dOLR) in the NICAM-7km run averaged over land in the tropics. The dOLR is computed as the departure of the clear-sky OLR from the total OLR. Larger dOLR corresponds to higher clouds and wider coverage. There are remarkable phase

4822

JOURNAL OF CLIMATE

FIG. 10. Longitude–time cross section of the monthly mean precipitation intensity averaged over Central America (38–78N) simulated in the NICAM-7km run. See also Mapes et al. (2003a).

differences among variables over land. First, precipitable water vapor increases in the morning as evaporation from the surface increases, and it reaches the daily maximum at 1500 LT. Precipitation peaks about 1.5 h later than precipitable water vapor (1630 LT). The dOLR peaks at 1930 LT, 3 h later than the precipitation peak. The daily peaks of precipitation and dOLR roughly correspond to the mature stage and decay stage characterized by the spreading cirrus associated with deep convection, respectively. Yamamoto et al. (2008) also pointed out the phase difference among three precipitation products. The different phase in the first harmonic of the precipitation diurnal cycle over Amazon in 3G68 and 3B42 (Figs. 1a,b) is consistent with the phase difference between precipitation intensity and OLR simulated in NICAM. The GCRM experiment thus could explain the cause of the phase difference among precipitation products. In recent years, new satellite observations, such as by the CloudSat and the Cloud-Aerosol Lidar and Infrared

VOLUME 22

Pathfinder Satellite Observation (CALIPSO), started to sound cloud microphysical properties. Further collaboration between GCRM and satellite remote sensing communities is necessary, particularly, in terms of cloud properties that are an inevitable aspect of the climate change issues. Some groups have already made available satellite data simulators such as the Cloud Feedback Model Intercomparison Project (CFMIP) Observational Simulator Package (COSP; http://cfmip.metoffice.com/ COSP.html) and the Satellite Data Simulator Unit (SDSU; http://precip.hyarc.nagoya-u.ac.jp/sdsu/sdsu-main. html) to verify the microphysical parameters in numerical models. For example, Masunaga et al. (2008) proposed a methodology to diagnose a GCRM based on a forward radiative transfer model that is able to calculate the TRMM and CloudSat reflectivity from the modeled hydrometeor variables. Model validations using such new observations should be quite beneficial for ameliorating the cloud physics scheme in GCRMs. In turn, the GCRMs are able to provide related atmospheric variables that are difficult to observe from spaceborne measurements and thus are supplementary to the satellite observations. The remarkable resolution dependence is found over land in NICAM. This suggests that even when a global nonhydrostatic model is used higher resolution is necessary when cumulus parameterizations are not used. Figure 5 indicates that the phase difference is evident even in the morning hours, which correspond to the initial stage of convection. Therefore, the resolution sensitivity is strongly coupled to the boundary layer processes in continental areas. In the future, the detailed analysis of the boundary layer processes, such as mass flux and entrainment/detrainment ratio, can be a focus of investigation. Such information could be useful for improving cumulus parameterization schemes used in conventional GCMs. Actually, statistics of NICAM

FIG. 11. Counts of the cold pool formation events over the ocean during the 31-day integration of the NICAM-7km run. Land areas are masked out.

15 SEPTEMBER 2009

SATO ET AL.

4823

FIG. 12. Diurnal change of the cold pool formation event over the ocean during the 31-day integration of the NICAM-7km run. (a) 0000–0600, (b) 0600–1200, (c) 1200–1800, (d) 1800–2400 LT.

results are used to evaluate a new cloud scheme proposed by Watanabe et al. (2008). The GCRM is undoubtedly innovative in the study of our climate system. It requires, however, huge computational costs to perform long-term simulations targeting climate change and variability. Therefore, another possible way to make use of the merit of GCRM, from a viewpoint of the diurnal cycle, is to reduce uncertainties in GCMs by improving cumulus parameterizations and enhancing our understanding of the cloud–radiation interaction processes.

7. Summary and discussion This study presents the diurnal cycle of precipitation in the tropics simulated in a global cloud-resolving model. Here, we statistically analyzed the diurnal cycle of tropical deep convection using a one-month simulation from NICAM. The simulation captures the precipitation diurnal cycle fairly well compared with the TRMM products, although the amplitude is overestimated. Too large amplitude of the precipitation diurnal cycle in Figs. 4b and 5 may be attributed partly to the simplified description of the land surface, that is, the bucket model,

which may be also causing a phase delay in the diurnal cycle of precipitation, and partly to the one-month simulation, which may present an average well above the multiyear average used in the 3G68. The NICAM experiments suggest that there is a remarkable resolution dependence in the phase and amplitude of the diurnal cycle of precipitation over land. The local time of the precipitation maximum becomes later as the horizontal grid size becomes coarser. The NICAM-3.5km run exhibits a realistic peak time; however, the NICAM-7km run displays a 1.5-h delay in the diurnal phase, and the NICAM-14km run causes another 3-h delay, namely a 4.5-h phase delay compared with both the NICAM-3.5km run and the TRMM 3G68. The amplitude has a similar tendency; namely, the coarser-resolution NICAM-7km and NICAM-14km runs cause a larger-amplitude precipitation diurnal cycle than the high-resolution NICAM-3.5km run. In contrast, over ocean, the precipitation diurnal cycle is not sensitive to the horizontal resolution. Even the NICAM-14km run shows an early morning peak over the ocean similar to the NICAM-3.5km run and observations. Such a difference in the sensitivity to horizontal resolution is

4824

JOURNAL OF CLIMATE

FIG. 13. Monthly mean diurnal cycle of the precipitable water vapor (PWV; circles), precipitation intensity (crosses), and dOLR (squares) anomalies in the NICAM-7km run averaged over 158S–158N. Red lines indicate land grids and blue lines indicate ocean grids. dOLR is shown as the departure of the clear-sky OLR from the total OLR. Larger dOLR corresponds to higher and wider clouds.

attributed to the different structure and life cycle of convective systems over land and over ocean. Over land, precipitation systems that are relevant to the diurnal cycle generally experience generation, development, and decay stages within one day. Over ocean, however, precipitation systems are generally well organized and long lived and may have a lifetime longer than one day. Therefore, in a crude view, only convective activity may change diurnally over ocean. The strong sensitivity to model horizontal resolution over land seems to have its roots in the reproducibility of the boundary layer process that is crucially important for the initiation of convection. More detailed analyses of the boundary layer processes and cloud properties are the theme for future studies. High performance in simulating regional characteristics of the diurnal cycle proves that the GCRM is capable of resolving the key mechanisms controlling the precipitation diurnal cycle like the land–sea breeze and gravity wave/current propagations. Therefore, GCRM has the advantage of simulating the diurnal cycle features that propagate far from the original disturbance in association with mesoscale circulation features, such as cold pools and squall lines, which is the major difference between the GCRM and other GCMs. By using the GCRM, classification of the diurnal cycle may be possible using a GCRM based on the controlling mechanisms, which would improve our understanding of the diurnal cycle of clouds and precipitation processes. Since the TRMM observations became available, studies on the precipitation diurnal cycle have progressed significantly in terms of understanding real features of the

VOLUME 22

convective systems. Additionally, by the aid of mesoscale models, the regional characteristics of the precipitation diurnal cycle and controlling mechanisms have been studied intensively (e.g., Qian 2008; Saito et al. 2001; Sato and Kimura 2005; Satomura 2000; Warner et al. 2003; Zhou and Wang 2006). In the present study, we successfully simulate the precipitation diurnal cycle in the tropics using the GCRM, although some shortcomings remain. The GCRM is also able to simulate the diurnal cycles over supercomplex terrains, such as the Tibetan Plateau and the Himalaya Range, as that was shown to be the important process controlling downstream regional climate (Sato et al. 2007; Shi et al. 2008). In the near future, the capability of realistically simulating the phase and amplitude of the precipitation diurnal cycle in GCRMs could mitigate uncertainties in cumulus parameterizations in climate simulations. The advantage of global coverage of GCRMs enables us to study the impact of the diurnal cycle on global atmospheric circulation. For example, a sensitivity experiment in which the diurnal cycle is artificially suppressed in a certain area may help identify the role of the diurnal cycle in regional and global climate systems and in climate variability. Modulation of the large-scale atmospheric condition by the Madden–Julian oscillation is affecting the diurnal cycle over the Maritime Continent (Ichikawa and Yasunari 2006). Furthermore, the modulation of intraseasonal oscillation by the regional diurnal cycle is also an important topic for GCRMs. Acknowledgments. We thank all members of the NICAM group for their efforts in developing and improving the model. The NICAM simulations were conducted on the Earth Simulator at Japan Agency for Marine-Earth Science and Technology. This study was supported by the VL project (Virtual Laboratory for the earth’s climate system diagnostics) and Grant-in-Aid for Scientific Research (20840013) funded by Ministry of Education, Culture, Sports, Science, and Technology, Japan and JST/CREST. YW has been supported in part by JAMSTEC, NOAA, and NASA through their sponsorships of the International Pacific Research Center (IPRC) in the School of Ocean and Earth Science and Technology (SOEST) at the University of Hawaii at Manoa.

REFERENCES Arakawa, A., 2004: The cumulus parameterization problem: Past, present, and future. J. Climate, 17, 2493–2525. Arakawa, O., and A. Kitoh, 2005: Rainfall diurnal variation over the Indonesian maritime continent simulated by 20 km-mesh GCM. SOLA, 1, 109–112, doi:10.2151/sola.2005-029.

15 SEPTEMBER 2009

SATO ET AL.

Chern, J., W.-K. Tao, and X. Lin, 2006: Diurnal variations of the water cycle in the Goddard Multi-scale Modeling Framework. Proc. 27th Conf. on Hurricanes and Tropical Meteorology, Monterey, CA, Amer. Meteor. Soc., 16D.4. Collier, J. C., and K. P. Bowman, 2004: Diurnal cycle of tropical precipitation in a general circulation model. J. Geophys. Res., 109, D17105, doi:10.1029/2004JD004818. Dai, A., 2001: Global precipitation and thunderstorm frequencies. Part II: Diurnal variations. J. Climate, 14, 1112–1128. ——, 2006: Precipitation characteristics in eighteen coupled climate models. J. Climate, 19, 4605–4630. ——, X. Lin, and K. L. Hsu, 2007: The frequency, intensity, and diurnal cycle of precipitation in surface and satellite observations over low- and mid-latitudes. Climate Dyn., 29, 727–744. DeMott, C. A., D. A. Randall, and M. Khairoutdinov, 2007: Convective precipitation variability as a tool for general circulation model analysis. J. Climate, 20, 91–112. Gille, S. T., S. G. Llewellyn Smith, and N. M. Statom, 2005: Global observations of the land breeze. Geophys. Res. Lett., 32, L05605, doi:10.1029/2004GL022139. Grabowski, W. W., 2001: Coupling cloud processes with the largescale dynamics using the Cloud-Resolving Convection Parameterization (CRCP). J. Atmos. Sci., 58, 978–997. Gray, W. M., and R. W. Jacobson Jr., 1977: Diurnal variation of deep cumulus convection. Mon. Wea. Rev., 105, 1171–1188. Hirose, M., and K. Nakamura, 2005: Spatial and diurnal variation of precipitation systems over Asia observed by the TRMM Precipitation Radar. J. Geophys. Res., 110, D05106, doi:10.1029/ 2004JD004815. Houze, R. A., S. G. Geotis, F. D. Marks, and A. K. West, 1981: Winter monsoon convection in the vicinity of North Borneo. Part I: Structure and time variation of the clouds and precipitation. Mon. Wea. Rev., 109, 1595–1614. Ichikawa, H., and T. Yasunari, 2006: Time–space characteristics of diurnal rainfall over Borneo and surrounding oceans as observed by TRMM-PR. J. Climate, 19, 1238–1260. Inoue, T., M. Satoh, H. Miura, and B. E. Mapes, 2008: Characteristics of cloud size of deep convection simulated by a global cloud resolving model. J. Meteor. Soc. Japan, 86A, 1–15. Johnson, R. H., P. E. Ciesielski, and J. A. Cotturone, 2001: Multiscale variability of the atmospheric mixed layer over the western Pacific warm pool. J. Atmos. Sci., 58, 2729–2750. 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. ——, ——, and C. DeMotte, 2005: Simulations of the atmospheric general circulation using a cloud-resolving model as a superparameterization of physical processes. J. Atmos. Sci., 62, 2136–2154. ——, C. DeMotte, and D. A. Randall, 2008: Evaluation of the simulated interannual and subseasonal variability in an AMIPstyle simulation using the CSU multiscale modeling framework. J. Climate, 21, 413–431. Kikuchi, K., and B. Wang, 2008: Diurnal precipitation regimes in the global tropics. J. Climate, 21, 2680–2696. Kuroda, Y., 2002: TRITON: Present status and future plan. Report for the International Workshop for Review of the Tropical Moored Buoy Network, JAMSTEC, 77 pp. [Available online at http://www.jamstec.go.jp/jamstec/TRITON/future/pdf/ Status.pdf.] Lee, M.-I., and Coauthors, 2007: Sensitivity to horizontal resolution in the AGCM simulations of warm season diurnal cycle of

4825

precipitation over the United States and northern Mexico. J. Climate, 20, 1862–1881. Lin, X., D. A. Randall, and L. D. Fowler, 2000: Diurnal variability of the hydrologic cycle and radiative fluxes: Comparisons between observations and a GCM. J. Climate, 13, 4159–4179. ——, L. D. Fowler, and D. A. Randall, 2002: Flying the TRMM satellite in a general circulation model. J. Geophys. Res., 107, 4281, doi:10.1029/2001JD000619. Lorenz, P., and D. Jacob, 2005: Influence of regional scale information on the global circulation: A two-way nesting climate simulation. Geophys. Res. Lett., 32, L18706, doi:10.1029/2005GL023351. Madden, R. A., and P. R. Julian, 1972: Description of global-scale circulation cells in the tropics with a 40–50 day period. J. Atmos. Sci., 29, 1109–1123. Mapes, B. E., T. T. Warner, M. Xu, and A. J. Negri, 2003a: Diurnal patterns of rainfall in northwestern South America. Part I: Observations and context. Mon. Wea. Rev., 131, 799–812. ——, ——, and ——, 2003b: Diurnal patterns of rainfall in northwestern South America. Part III: Diurnal gravity waves and nocturnal convection offshore. Mon. Wea. Rev., 131, 830–844. Masunaga, H., M. Satoh, and H. Miura, 2008: A joint satellite and global cloud-resolving model analysis of a Madden-Julian Oscillation event: Model diagnosis. J. Geophys. Res., 113, D17210, doi:10.1029/2008JD009986. McPhaden, M. J., 1995: The Tropical Atmosphere Ocean (TAO) array is completed. Bull. Amer. Meteor. Soc., 76, 739–741. Miura, H., M. Satoh, T. Nasuno, A. T. Noda, and K. Oouchi, 2007a: A Madden-Julian Oscillation event simulated using a global cloud-resolving model. Science, 318, 1763–1765. ——, ——, H. Tomita, A. T. Noda, T. Nasuno, and S. Iga, 2007b: A short-duration global cloud-resolving simulation with a realistic land and sea distribution. Geophys. Res. Lett., 34, L02804, doi:10.1029/2006GL027448. Mori, S., and Coauthors, 2004: Diurnal land–sea rainfall peak migration over Sumatera Island, Indonesian maritime continent, observed by TRMM satellite and intensive rawinsonde soundings. Mon. Wea. Rev., 132, 2021–2039. Moteki, Q., and Coauthors, 2008: Mechanism of the northward propagation of mesoscale convective systems observed on 15 June 2005 during PALAU2005. J. Geophys. Res., 113, D14126, doi:10.1029/2008JD009793. Nasuno, T., H. Tomita, S. Iga, H. Miura, and M. Satoh, 2008: Convectively coupled equatorial waves simulated by a global nonhydrostatic experiment on an aqua planet. J. Atmos. Sci., 65, 1246–1265. Neale, R., and J. Slingo, 2003: The maritime continent and its role in the global climate: A GCM study. J. Climate, 16, 834–848. Nesbitt, S. W., and E. J. Zipser, 2003: The diurnal cycle of rainfall and convective intensity according to three years of TRMM measurements. J. Climate, 16, 1456–1475. Petch, J. C., A. R. Brown, and M. E. B. Gray, 2002: The impact of horizontal resolution on the simulations of convective development over land. Quart. J. Roy. Meteor. Soc., 128, 2031–2044. Qian, J. H., 2008: Why precipitation is mostly concentrated over islands in the maritime continent. J. Atmos. Sci., 65, 1428– 1441. Rasch, P. J., and Coauthors, 2006: A characterization of tropical transient activity in the CAM3 atmospheric hydrologic cycle. J. Climate, 19, 2222–2242. Reynolds, R. W., and T. M. Smith, 1994: Improved global sea surface temperature analyses using optimal interpolation. J. Climate, 7, 929–948.

4826

JOURNAL OF CLIMATE

Rickenbach, T. M., 2004: Nocturnal cloud systems and the diurnal variation of clouds and rainfall in southwestern Amazonia. Mon. Wea. Rev., 132, 1201–1219. Saito, K., T. Keenan, G. Holland, and K. Puri, 2001: Numerical simulation of the diurnal evolution of tropical island convection over the maritime continent. Mon. Wea. Rev., 129, 378–400. Sato, T., and F. Kimura, 2005: Diurnal cycle of convective instability around the central mountains in Japan during the warm season. J. Atmos. Sci., 62, 1626–1636. ——, H. Miura, and M. Satoh, 2007: Spring diurnal cycle of clouds over Tibetan Plateau: Global cloud-resolving simulations and satellite observations. Geophys. Res. Lett., 34, L18816, doi:10.1029/2007GL030782. ——, T. Yoshikane, M. Satoh, H. Miura, and H. Fujinami, 2008: Resolution dependency of the diurnal cycle of convective clouds over the Tibetan Plateau in a mesoscale model. J. Meteor. Soc. Japan, 86A, 17–31. Satoh, M., 2008: Numerical simulations of heavy rainfalls by a global cloud-resolving model. J. Disaster Res., 3, 33–38. ——, 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, doi:10.1016/j.jcp.2007.02.006. Satomura, T., 2000: Diurnal variation of precipitation over the Indo-China Peninsula: Two-dimensional numerical simulation. J. Meteor. Soc. Japan, 78, 461–475. Shi, X.-Y., Y. Wang, and X.-D. Xu, 2008: Effect of mesoscale topography over the Tibetan Plateau on summer precipitation in China: A regional model study. Geophys. Res. Lett., 35, L19707, doi:10.1029/2008GL034740. Shige, S., Y. N. Takayabu, W. K. Tao, and D. E. Johnson, 2004: Spectral retrieval of latent heating profiles from TRMM PR data. Part I: Development of a model-based algorithm. J. Appl. Meteor., 43, 1095–1113. Skinner, T., and N. Tapper, 1994: Preliminary sea breeze studies over Bathurst and Melville Islands, northern Australia, as part of the island thunderstorm experiment (ITEX). Meteor. Atmos. Phys., 53, 77–94. Sui, C.-H., K. M. Lau, Y. N. Takayabu, and D. A. Short, 1997: Diurnal variations in tropical oceanic cumulus convection during TOGA COARE. J. Atmos. Sci., 54, 639–655. ——, X. Li, and K.-M. Lau, 1998: Radiative–convective processes in simulated diurnal variations of tropical oceanic convection. J. Atmos. Sci., 55, 2345–2357. Takayabu, Y. N., 2006: Rain-yield per flash calculated from TRMM PR and LIS data and its relationship to the contribution of tall convective rain. Geophys. Res. Lett., 33, L18705, doi:10.1029/2006GL027531. ——, and M. Kimoto, 2008: Diurnal march of rainfall simulated in a T106 AGCM and dependence on cumulus schemes. J. Meteor. Soc. Japan, 86A, 163–173.

VOLUME 22

Tao, W. K., 2007: Cloud resolving modeling. J. Meteor. Soc. Japan, 85B, 305–330. ——, S. Lang, J. Simpson, C. H. Sui, B. Ferrier, and M.-D. Chou, 1996: Mechanisms of cloud-radiation interaction in the tropics and midlatitudes. J. Atmos. Sci., 53, 2624–2651. ——, 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. ——, 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. Wang, Y., S.-P. Xie, H. Xu, and B. Wang, 2004: Regional model simulations of marine boundary layer clouds over the southeast Pacific off South America. Part I: Control experiment. Mon. Wea. Rev., 132, 274–296. ——, L. Zhou, and K. Hamilton, 2007: Effect of convective entrainment/detrainment on simulation of tropical precipitation diurnal cycle. Mon. Wea. Rev., 135, 567–585. Warner, T. T., B. E. Mapes, and M. Xu, 2003: Diurnal patterns of rainfall in northwestern South America. Part II: Model simulations. Mon. Wea. Rev., 131, 813–829. Watanabe, M., S. Emori, M. Satoh, and H. Miura, 2008: A PDFbased hybrid prognostic cloud scheme for general circulation models. Climate Dyn., doi:10.1007/s00382-008-0489-0. Weisman, M. L., W. C. Skamarock, and J. B. Klemp, 1997: The resolution dependence of explicitly modeled convective systems. Mon. Wea. Rev., 125, 527–548. Yamamoto, M. K., F. A. Furuzawa, A. Higuchi, and K. Nakamura, 2008: Comparison of diurnal variations in precipitation systems observed by TRMM PR, TMI, and VIRS. J. Climate, 21, 4011–4028. Yang, G. Y., and J. Slingo, 2001: The diurnal cycle in the tropics. Mon. Wea. Rev., 129, 784–801. Zhang, G. J., 2002: Convective quasi-equilibrium in midlatitude continental environment and its effect on convective parameterization. J. Geophys. Res., 107, 4220, doi:10.1029/ 2001JD001005. Zhang, Y., and Coauthors, 2008: On the diurnal cycle of deep convection, high-level cloud, and upper troposphere water vapor in the Multiscale Modeling Framework. J. Geophys. Res., 113, D16105, doi:10.1029/2008JD009905. Zhou, L., and Y. Wang, 2006: Tropical Rainfall Measuring Mission observation and regional model study of precipitation diurnal cycle in the New Guinean region. J. Geophys. Res., 111, D17104, doi:10.1029/2006JD007243. Zuidema, P., 2003: Convective clouds over the Bay of Bengal. Mon. Wea. Rev., 131, 780–798.