A numerical study of the parametrization of deep tropical convection

A two-dimensional mesoscale cloud model is used to simulate an ensemble of deep tropical convective clouds using GATE data. Random heating perturbations near the sea surface initiate the development of a population of clouds which grow under the influence of specified large-scale, radiative and surface forcing. Budgets of convective heating and moistening are presented and the model found to be a reasonable representation of the atmosphere. The model is used to investigate the parametrization of cumulus convection in large-scale models. Equations governing the effect of convection upon the large-scale atmosphere are derived in full and compared with approximate forms used by previous workers. These forms are the basis of many parametrization schemes. Simple representations of cumulus heating (Q1), cumulus moistening (Q2), and dynamical transports for use in parametrization schemes in large-scale atmospheric models are studied. The importance of shallow convection in determining the vertical structure of Q2 is recognized. The dependence of convective rainfall upon atmospheric structure and large-scale processes (the closure problem) is investigated. The importance of downdraughts is emphasized and difficulties in the calculation of Q2 identified. Study of parametric forms for cumulus momentum transports indicates that the ‘cumulus friction’ hypothesis is inadequate. It is suggested that present closure schemes are too restrictive.