Hao Fu, Morgan O'Neill

How the cumulus clouds organize into a tropical cyclone remains poorly understood. The difficulty lies in that the deep convection is noisy at the kilometer scale, but follows the physical feedbacks at the mesoscale. We build a barotropic numerical model to understand the interaction of the stochastic and deterministic processes in the genesis of a tropical depression. Deep convection is represented as a multitude of isolated convergence forcing. The convection is assigned to distribute randomly at the small scale. At the mesoscale, convection is preferentially seeded to regions with a high spatially-filtered vertical vorticity. The preferential seeding mimics the physical feedbacks, and the filter implicitly represents the nonlocal convective triggering by gravity wave and cold pool. The result shows that the early-stage evolution is dominated by random vortex tube stretching. Subsequently, the regions where repetitive stretching occurs become vortex clusters, and induce more convection around them. The collision and coalescence between vortex clusters lead to a major vortex, which accelerates the growth by the preferential seeding. This physical picture agrees with a cloud-permitting simulation of spontaneous tropical cyclogenesis over uniform sea surface temperature. A theoretical model with approximate analytical solution is presented to depict the full evolution process.