One of the most challenging meteorological problems in tropical regions is the prediction of severe weather and precipitation resulting from thunderstorms. Forecasters in tropical regions have few conceptual models at their disposal and there is a notable lack of useful theory that they can call upon. Numerical weather prediction in the tropics is often of limited value for forecasting weather involving deep convection, which is due in part to the paucity of appropriate storm-scale observations. The aim of this proposal is to better understand and better predict where tropical convection will be initiated and how it will be organized.