A Model Study of the Relationship between Convergence and Deep Convection

Representing the initiation and development of tropical convection in weather and climate models remains one of the key challenges in atmospheric science. It is often assumed that the presence of large-scale convergence and a ‘trigger' to overcome convective inhibition are necessary for the initiation of deep convection and these factors are elements in some convection parameterisation schemes. A trigger may, for example, take the form of a convergence line, a surface hot spot (from variations in soil moisture) or originate from a region of elevated topography. Our ability to observe both large-scale convergence and triggers is limited and thus our understanding of them and their relative importance is low. In the last few years we have developed the ability to run numerical models for periods of several months with a grid-spacing small enough to begin to resolve convective processes explicitly. The UK ‘Cascade' consortium project produced 40-day simulations over the West African continent at 12 km grid-spacing with parameterised convection and at 12, 4 and 1.5 km grid-spacing where convection is allowed to evolve explicitly. The aims of this study are to use these simulations to quantify the relative contributions of large-scale convergence and local triggers to convection initiation, identify the types of small-scale triggers (linear/circular convergence features and how they are orientated in the large-scale wind) and to evaluate how the convective triggers vary with model grid-spacing and convection parameterisation. A method for distinguishing between the shapes of small-scale convergence features that cause convective initiation is presented. All three of the explicit simulations produce similar results, in which the majority of convective events are triggered by linear convergence lines. The majority of initiations occur within large-scale convergence, although a significant proportion (approximately 25-30%) occur within large-scale divergence. The simulation with parameterised convection does not show the same behaviour; initiations are equally likely within large-scale divergence and convergence and the local convergent ‘triggers' are of a different form. The explicit simulations show a larger contribution to initiation from patterns of convergence consistent with cold-pool outflows and surface hot-spots.