At the equator, the amplitude of the semidiurnal pressure perturbations associated with atmospheric tides is largest. The spatial gradients of the pressure tides can induce low-level divergence and convergence. Therefore, the semidiurnal atmospheric tides have been discussed as a modulator of rainfall variations, especially over the open ocean where there is no dominant large-scale forcing with a regular period (e.g., Malkus 1964; Brier and Simpson 1969; Dai 2001; Yasunaga et al. 2007). Lindzen (1978), however, doubts the importance of the tides, because the convergence associated with the semidiurnal tide is too small. It is generally difficult to obtain enough observational data over the open ocean to clarify the relationship between the atmospheric tides and precipitation. Therefore, the importance of atmospheric tides to diurnal variations of precipitation is a matter of controversy.
A cloud-resolving model (CRM) is a powerful and convenient tool to investigate the development of cumulus convection, and CRMs have been used in order to discuss the early morning peak of rainfall (e.g., Liu and Moncrieff, 1998; Sui et al. 1998; Kubota et al. 2004). However, the models cover only an area of several hundred kilometers and are insufficient to discuss the effects of atmospheric tides. On the other hand, a general circulation model (GCM) is employed in some investigations in order to examine the relationship between the atmospheric tides and precipitation (e.g., Woolnough et al. 2004). However, the GCM includes a cumulus parameterization to represent effects of sub-grid scale cumulus convection. Cumulus parameterization contains several prescribed parameters (e.g. entrainment rate, equilibrium time scale of convective activity, convective suppression based on large scale relative humidity, and so on), and the timing or sensitivity of precipitation to atmospheric tides can be controlled to some extend through tuning the parameters.
Recent progresses of computational resources enable us to conduct global cloud-resolving simulations with a resolution of several kilometers. Tomita and Sato (2004) developed the Nonhydrostatic ICosahedral Atmospheric Model (NICAM), and Tomita et al. (2005) reported preliminary results of global cloud-resolving simulations for an aqua planet condition with the resolutions of 14km, 7km and 3.5 km. These simulations are suitable to investigate the relationship between the precipitation cycle and the atmospheric tides, because the model domain covers the whole globe and the continental influence is excluded. Therefore, the present study explores the atmospheric tides and precipitation cycle, making use of the simulation results by the aqua-planet NICAM. Even the resolution of 3.5 km is not enough fine to represent shallow cumulus, which can be regarded as a precursor of deeper convection (e.g., Kubota et al. 2004; Mapes et al. 2006), and the timing of the deep convection development might be influenced by the coarse resolution. However, it is the very first attempt to simulate atmospheric general circulation with a three dimensional cloud-resolving model, and such global CRM simulation only contains the least uncertainties at the present moment. Therefore, it will be quite useful to describe the precipitation cycle and the role of atmospheric tides.
2. Model and Experimental Setup
The model used in the present study is NICAM, as described above. The model equations are based on a nonhydrostatic framework (Tomita and Satoh 2004), and guarantee conservation of total mass and total energy (Satoh 2002, 2003). The conservation property is suitable for long-term simulations.
The horizontal grid intervals are about 14, 7, and 3.5km (hereafter referred to as Exp-14km, Exp-7km, and Exp-3.5km, respectively). The model has 54 levels in the vertical with the model top at 40 km, and a fine resolution (75 m) at the lowest level and a coarser grid spacing (750 m) in the upper levels. Time intervals are 30 sec for Exp-14km and Exp-7km, and 15sec for Exp-3.5km. Moist processes are explicitly represented, using the simple cloud microphysics scheme proposed by Grabowski (1998), and no cumulus parameterization is employed, even in the Exp-14km. The level-2 closure model (Mellor and Yamada, 1974) is applied to represent turbulent diffusion. The radiation and surface flux schemes are based on Nakajima et al. (2000) and Louis (1979), respectively. Radiation is calculated every 10 min for Exp-14km and Exp-7km, and 5 min for Exp-3.5km, and other physical processes are updated at each time step.
SST is fixed by a zonally uniform value with a peak on the equator, and the aqua planet setup is based on the method proposed by Neale and Hoskins (2000). Initial conditions for the Exp-14km are obtained from a 3.5-year integration with a conventional AGCM with T42L59 resolution. The NICAM with 14-km resolution (Exp-14km) is integrated for 90 days. The results of Exp-14km on the 60th day are interpolated onto Exp-7km gridpoints, and a 30-day integration is performed for Exp-7km. Results of Exp-7km on the 20th day are also utilized as initial condition for a 10-day integration for Exp-3.5km.
Variations in surface pressure show clear semidiurnal cycle in all simulations (Exp-14km, Exp-7km, and Exp-3.5km), and the amplitudes are comparable to the observed value over the tropical open ocean (about 1.1 hPa). Semidiurnal variations in zonal wind are also simulated, while variations in meridional wind show diurnal cycle.
The precipitation peak in the early morning, which is commonly observed over the tropical open ocean, is well reproduced in all experiments (Exp-14km, Exp-7km, and Exp-3.5km), although the peak time in the Exp-14km delays several hours from that in the Exp-7km or Exp-3.5km. Moreover, the minor precipitation peak in the afternoon is found in Exp-7km and Exp-3.5km, and the finer resolution experiments simulate semidiurnal cycles of precipitation. Precipitation anomaly has zonal wavenumber 2, and girdles the earth in a day to the west direction. The precipitation peak time coincides with that of surface pressure minimum. The SST is constant during the integration. Therefore, it can be considered that the atmospheric tides play a role in the semidiurnal cycle of precipitation.