DET is here explored in a 3-layer model of the troposphere that includes a boundary layer over an ocean of constant temperature, a middle layer in which cumulus updrafts can entrain mass, and an upper outflow layer. The model is rigid in that it requires both convergence of boundary layer air and a wind-induced surface flux of moist entropy to sustain deep convection [cf. K. Ooyama, J. Atmos. Sci., 26, 3-40 (1969)]. The initial turbulence is confined to the lower two layers of the troposphere, whereas the upper layer starts at rest. After an incubation period, the influence of deep convection can supercede ideal 2D mechanisms of self-organization, such as vortex merger. A strong cyclone-anticyclone asymmetry can develop, with relatively intense cyclones dominating the system. Hurricanes generally form at "high'' values of the sea-surface temperature (SST), the Coriolis parameter f, and the surface-exchange coefficient for moist entropy CE. However, it is shown that there exist regions in parameter space where random noise can evolve into one of two distinct structures: a tropical cyclone or a metastable synoptic scale gyre. In general, specifying the initial energy spectrum is insufficient for predicting the end state.
The hurricanes that emerge in our simulations of DET are realistic in several ways. During rapid intensification, they typically develop polygonal eyewalls and mesovortices. Subsequently, they exhibit moderate intensity oscillations that resemble eyewall breakdown and regeneration cycles. Finally, the time-averaged intensity of a DET hurricane increases with the SST and the ratio of CE over the momentum exchange coefficient CD.