Towards Convective-Scale Forecasting in GFDL fvGFS: First Impressions and Development Prospects

FV3 and its predecessors are the leading dynamical cores for global weather and climate models in the United States. Recent advances in FV3 enable its use as a component in models capable of efficiently and skillfully simulating both global and convective-scale motions, through the use of two-way interactive grid refinement. We show results of fvGFS, the FV3 core coupled to an appropriately-modified form of the Global Forecast System (GFS) physics and initial conditions, with a global 13-km resolution and a 3-km nest over the continental Untied States. The skill of the global model is sustained, and precipitation skill is competitive with existing operational models while convective-scale motions, including severe events such as supercell thunderstorms and derechos, are skillfully predicted several days in advance. Community collaborations to improve the physics and initialization are also presented. Thoughts about effective application of global models for use in short-range applications are discussed.