Data Assimilation Development and Testing for the Initial Version of the FV3-GFS

The U.S. National Weather Service (NWS) has launched a research-to-operations initiative to develop a unified, fully coupled weather-scale system, the Next Generation Global Prediction System (NGGPS). Through the NWS-led community effort, several dynamical cores were evaluated and the Finite-Volume Cubed-Sphere Dynamical Core (FV3) was chosen for the upcoming unified system. It is computationally efficient, conservative, and has hydrostatic and non-hydrostatic options, making it suitable across many spatial and temporal scales. The adaptation of the currently operational GSI-based hybrid 4DEnVar scheme for use with the new model has been a significant effort toward an initial transition to the FV3-based model.

The development and testing of an initial beta-version of the FV3-GFS system is well underway with a goal to having it run in real-time as a quasi-operational system by July 1, 2018. While many aspects of the system are designed to leverage assimilation and physics components from the current operational global forecast system (GFS), several significant changes are to be included in the initial FV3-GFS beta version.

This presentation will focus on some of the aspects of the data assimilation component of the FV3-GFS beta version, including a description of the ensemble to drive the hybrid assimilation, stochastic physics for treatment of model error, adaptation to a new bulk microphysics scheme, and technical issues related to the grid specification. Results from a low resolution prototype as well as operational-resolution retrospective and real-time experiments will be presented.