Towards a Unified Prediction System from Weather to Climate Scale (Invited Presentation)

The NOAA/Geophysical Fluid Dynamics Laboratory has been developing a unified regional-global modeling system with variable resolution capabilities that can be used for severe weather predictions (from tornado outbreaks to tropical cyclones) and ultra-high-resolution (1-km) regional climate simulations within a unified global modeling framework for both weather and climate predictions.

The foundation of this flexible modeling system is the nonhydrostatic Finite-Volume Dynamical Core on the Cubed-Sphere (FV3). Owning to its accuracy, adaptability, and computational efficiency, the FV3 has been selected, in July 2016, as the “engine” for NOAA's Next Generation Global Prediction System (NGGPS), which represents a great opportunity for the unification of weather and climate prediction systems.

We have built into the modeling system a stretched grid capability, a two-way regional-global nested grid capability, and an optimal combination of the stretched and two-way nests, making kilometer-scale regional simulations within a global modeling system feasible with today's High Performance Computing System. One of the main scientific goals is to enable simulations of high impact weather phenomena (such as tornadoes, thunderstorms, category-5 hurricanes) within an IPCC-class climate modeling system previously regarded as impossible. In this presentation I will demonstrate that, with the FV3, it is computationally feasible to simulate not only super-cell thunderstorms, but also the subsequent genesis of tornado-like vortices using a variable-resolution global model.

The development and tuning (calibration) strategy between traditional weather and climate models are understandably different due to different metrics, and very different predictability requirements.. Applications of traditional “climate” metrics or standards, such as angular momentum conservation, energy conservation, and flux balance at top of the atmosphere, can lead to insight into problems of traditional weather prediction model for medium-range prediction, and vice versa. The merger of weather and a compresensive climate modeling system can also help in the study of the various sources of extended-range predictability. Therefore, the unification in weather and climate models can happen not just at the algorithm or parameterization level, but also in the metric and tuning strategy used for both applications, and ultimetely, with benefits to both weather and climate applications.