1. Introduction

Precipitation activity in the earth's equatorial atmosphere is organized as a hierarchical structure which comprises Madden-Julian oscillation, super cloud clusters, and cloud clusters (Nakazawa, 1988; Lindzen, 2003). Hayashi and Sumi (1986) (referred to as HS86 hereafter) invented, as an idealization of the earth's atmosphere, the concept of an "aqua-planet" general circulation model (GCM). In the experiment with an idealized specification where the entire planet is covered only by the ocean whose sea surface temperature is zonally uniform and north-south symmetric and with a simplified GCM from the present-day standard which contains a simplified water cycle and radiative processes with the resolution of 128x64 horizontally and 12 vertical levels, they suggested an existence of hierarchical structure in the tropical atmosphere; precipitation areas of the scale of grid-scale of the model that propagate eastward coherently, and a planetary scale modulation of their intensity that propagates also eastward with the period of about 30 days and the wavelength of 40,000km. HS86 named the former as the super-cluster and related the latter as Madden-Julian oscillation, and argued that they are the structures intrinsic in the internal dynamics of the equatorial region of a moist atmosphere.

A large number of numerical studies have been performed with aqua-planet set-ups since HS86. These results show that precipitation patterns in GCMs depend on the spatial resolution, numerical schemes, and implementation of physical processes (e.g., Lee et al., 2001, 2003; Yamada et al. 2005). It is now recognized that the gridscale precipitation structure propagating eastward coherently and the variability of 40,000 [km] scale, both of which are present in HS86, do not necessarily appear in aqua planet GCMs. However, systematic examination of the dependence of precipitation structures on these factors have not been attempted except for Aqua-Planet Experiment Project APE) now being carried out as an international cooperative project, partly because the necessary computational resources are large. Due to the lack of such systematic studies, no agreement has been reached on how the precipitation patterns should be in GCMs, and we are still short of understanding on the dynamics that results in the difference of precipitation patterns in different GCMs. In those situations in mind, we consider that it is useful to reexamine systematically the variety of precipitation patterns in GCMs with simplified physical processes like those used in HS86 rather than sophisticated physical processes of the present day realistic GCMs, but with a broader parameter space than was tractable in the era of HS86.

One of the possible factors that can control the structure of the precipitation is the vertical distribution of condensation heating. Based on an aqua-planet GCM experiment, Numaguti and Hayashi (1991) concluded that the maintenance of the super-clusters in HS86, i.e., the grid-scale eastward propagating precipitation structures are governed by the dynamics of wave-CISK (the conditional instability of the second kind; Hayashi, 1970; Lindzen, 1974; the detailed explanation is given in Appendix Brief review of wave-CISK ). The vertical profile of condensation heating crucially affects the embodiment of wave-CISK dynamics (e.g., Lau and Peng, 1987; Numaguti and Hayashi, 2000. see also a criticism against wave consideration by Lindzen, 2003). If the grid-scale eastward propagating precipitation structures in GCMs are the manifestations of the wave-CISK dynamics, they will be less prominent when the vertical profile of condensation heating represented in a GCM is not consistent with that required by the wave-CISK dynamics. In this study, we perform a series of aqua-planet GCM experiments where the vertical structure of condensation heating are varied, and present some of the results, including animation of time series, of our examination on how the behaviors of grid-scale precipitation structures respond to the change of profile of heating.