The Next Generation Global Prediction System with the FV3

The design of the Finite-Volume dynamical core on the cubed-sphere (FV3) will be presented, with focus on FV3's key scientific and computational advantages, such as the vertically Lagrangian discretization, high-order finite-volume advection of potential temperature, air and tracer mass, and perhaps more importantly, consistent advection of absolute vorticity with afore-mentioned scalars. It is noted that FV3's C-D grid staggering of the horizontal winds with high-order finite-volume discretization, which was designed to achieve the goal of consistent PV transport, was difficult, if not impossible, to analyze by linear analyses, which led to some misleading results by some authors.

For the next Generation Global Prediction System (NGGPS), the non-hydrostatic FV3 enables the possibility of the eventual unification of regional and global models, for both weather and climate applications. This is made possible by choosing the scalable cubed-sphere grid that can easily accomodate additional regional “tiles” for two-way nesting purpose. A grid stretching capability is also built into FV3 to enable a smooth variation of resolution within the global domain. As such, FV3's computational efficiency is significantly higher than traditional variable-resolution or the Adaptive Mesh Refinement (AMR) approaches.

The FV3 has been used in production at the Geophysical Fluid Dynamics Laboratory (GFDL) and at NASA laboratories for climate simulations during the past decade. With the NGGPS project (phase-1 and phase-2), FV3 has also been systematically evaluated in Numerical Weather Prediction (NWP) mode. Due to its initial excellent performance for NWP applications, we at NOAA is setting the bar high, with the goal of being the world's best prediction system within an aggressive timeframe.