Towards Global-to-Regional Forecasting with a Next-Generation Forecast Model: fvGFS

Skillful prediction of the large-scale is crucial for effective smaller-scale prediction beyond very short time scales. A global convection-permitting model has no driving boundary conditions and must maintain large-scale skill while properly handling marginally-resolved convective scales. A new global forecast model, fvGFS, has been constructed at GFDL. This new model uses the operational GFS physics coupled to the modern nonhydrostatic FV3 core, the GFDL Finite-Volume Cubed-Sphere dynamical core. FV3 has shown excellent results at all scales of motion, and supports two-way grid nesting as well as grid stretching to efficiently reach convection-permitting scales in a global model. 

We demonstrate the efficacy of the 3-km regional-to-global nested fvGFS in a number of forecasts. We show the capability to develop intense hurricanes with realistic fine-scale eyewalls and rainbands. The new model also produces skillful predictions of severe weather outbreaks and of organized mesoscale convective systems. Fine-scale orographic and boundary-layer phenomena are also simulated with excellent fidelity by fvGFS. Further expected improvements are discussed, including the introduction of more sophisticated microphysics and of scale-aware convection schemes.