A potential aid to the understanding of scale interactions is the fact that there is a certain degree of "self-similarity" in observed gross features of the dynamical structures of organized tropical convection, from the mesoscale on up to the planetary scale structure of the MJO. Convectively coupled disturbances universally exhibit strong vertical tilts in their wind, temperature, moisture, vertical velocity and diabatic heating fields. In general these disturbances display a warm lower troposphere ahead of the wave, with cooling behind, and a warm mid-troposphere within the convective region. Low level moisture and thus CAPE and moist static energy is high ahead of the waves, and drying occurs first at low levels while it is still moist aloft behind the wave. Low level diabatic heating precedes deep convective heating, followed by a signal of upper tropospheric heating over cooling. These dynamical signals are consistent with the observation that the waves show a progression from a dominance of shallow to deep convection, and then stratiform precipitation, regardless of scale or propagation direction. It is a remarkable fact that the temporal and spatial evolution of mesoscale convective complexes, which can be traced back to microphysical arguments, also exists at a certain level on the scale of the MJO.
These observations have implications for the simulation of convectively coupled waves. Some General Circulation Models appear to have peaks in their rainfall spectra corresponding to the observed spectra of tropical cloudiness. However, all of the waves identified in models thus far have corresponding equivalent depths that are universally too deep and therefore phase speeds that are too fast. These waves all scale to around the same equivalent depth, a fact which perhaps provides clues to the deficiencies of physical parameterizations involved. Simple modeling and cloud resolving studies are beginning to provide some realistic results, and will no doubt provide useful testbeds for the development of improved parameterizations in next generation GCMs.