Building NOAA’s Seasonal Forecasting System (SFS): Process-Based Advances

Abstract

NOAA is advancing the development of the Unified Forecast System (UFS)-based Seasonal Forecast System (SFS), which is intended to replace the current operational Climate Forecast System version 2 (CFSv2). The new system aims to address long-standing forecast deficiencies and improve overall prediction skill, with particular emphasis on enhancing prediction skill for the Niño-3.4 index and the 2-meter temperature and precipitation forecasts over North America. SFS shares the same UFS-based Earth-system infrastructure and components as the Global Forecast System version 17 but with further development for seasonal applications. This manuscript details the updates made to the SFS atmospheric component dynamics and physics and the ocean component to maximize SFS seasonal prediction skills. The updates include the adoption of the hydrostatic Finite-Volume Cubed-Sphere dynamical core, implementation of a critical correction to the ozone parameterization to better represent the Antarctic ozone hole, a well-tuned ECMWF non-stationary gravity wave drag scheme to reduce stratospheric biases, the introduction of a constant background ocean gustiness parameterization to mitigate warm sea surface temperature biases, the application of a total turbulent energy-based eddy-diffusivity mass-flux vertical turbulent mixing scheme, and tuned convective cloud water contribution in the cloud-radiation interaction. A near-real-time beta version (SFS Beta 1.0) was formally launched in March 2026 to support NOAA’s operational seasonal outlooks, marking a major milestone in system development, with full operational implementation targeted for calendar year 2027. A preliminary verification has shown consistent improvement compared with CFSv2.