Numerical experiments using a general circulation model (GCM) were performed for the purpose of investigating the possibility of the generation of a long period (30-40 days) oscillation as a collective motion of cumulus activity (so-called wave-CISK) along the equator of an ocean covered globe ("aqua planet"). In our model, the SST distribution was symmetric about the equator and uniform in the longitudinal direction.
The results of a 90 day integration exhibited the spontaneous appearance of a collective motion of convective activity together with an east-west wavenumber one circulation (the "30 day oscillation"). The characteristics of this east-west circulation resemble those of the observed 30-60 day oscillation in the actual atmosphere. The 30 day oscillation is characterized by a superposition of two different scales: the scale of precipitation patterns ("super clusters") which is nearly equal to the equatorial radius of deformation, and the scale observed as the modulation of precipitation patterns and the east-west circulation with a wavenumber one. The whole structure moves eastward at a slow phase speed (15m/s).
The results also exhibited the spontaneous formation of double ITCZs around the equator. Thus the production of the double ITCZ structure does not necessarily require a minimum SST at the equator. In the equatorial region between the north and south ITCZs, active convection forms super clusters, while in the regions poleward of the ITCZs, active convection forms tropical cyclones.
Another experiment without the moist processes resulted in the abrupt disintegration of the 30 day oscillation into Kelvin and Rossby waves. This indicates that strong mode coupling between the equatorial free waves is required in order to maintain the 30 day oscillation. The slow phase speed, the strong mode coupling and the double structure in scale indicate that the collective motionof convective activity along the equator cannot be explained simply as a Kelvin wave mode of the linear equatorial wave-CISK theory presented so far. The 30 day oscillation should be regarded as a new type solution of the equatorial wave-CISK problem.