A modeling study of self-aggregation and large-scale control of tropical deep convection

The nonhydrostatic version of the Penn State University (PSU)/National Center for Atmospheric Research (NCAR) mesoscale model MM5 is used to investigate the mechanisms responsible for the superclustering of tropical deep convection. In particular, we are interested in whether mesoscale cloud clusters and super cloud clusters represent a mode of self-aggregation of convection, or whether they require preexisting horizontal variations in the large scale flow. Here, self-aggregation refers to convective organization that occurs spontaneously under horizontally uniform boundary conditions and large-scale forcings.; First, the performance of the MM5 over the tropical warm pool region is tested by using it to simulate the convective systems over the IFA (Intensive Flux Array) during TOGA COARE (Tropical Ocean Global Atmosphere Coupled Ocean-Atmosphere Response Experiment). The model results are compared with a variety of observations. The model/observation intercomparison shows that the MM5 is able to reproduce many ensemble properties of deep convective systems in response to evolving large-scale forcings during TOGA COARE. The model-produced convective organization is also qualitatively realistic.; To study the self-aggregation of tropical deep convection, we run the MM5 with prescribed domain-averaged vertical velocity and relax domain-averaged horizontal winds to a specified reference wind profile. Two vertical profiles of mean ascent are used. One has an elevated maximum at upper troposphere and near zero net vertical velocity in the lower troposphere. The other peaks at middle-troposphere, representing vertical velocity distribution of convective towers only. The model simulations with elevated forcing develop larger cloud clusters than the runs with mid-tropospheric forcing, suggesting some degree of self-aggregation under favorable large-scale forcings. A Fourier analysis of the precipitation organization in the elevated forcing run indicates considerable variance in propagating waves of wavelength 1000–2000 km in which convective heating is positively correlated with temperature and moisture anomalies. Sensitivity tests show that the long-wavelength organization does not require horizontal variability of surface fluxes and so cannot be explained by WISHE-type (Wind-Induced Surface Heat Exchange) mechanisms. Sensitivity tests of model results to magnitude and vertical distribution of forcings, reference wind profiles, and grid resolution are also conducted.