On March 20th, China Central Television reported that the China Meteorological Administration (CMA) has officially released its next-generation atmospheric numerical weather prediction (NWP) model.
The model utilizes the fully independent dynamic framework algorithm, the multi-moment constrained finite volume method (MCV), as the underlying algorithm, further enhancing the accuracy of global kilometer-scale and regional hundred-meter-scale numerical forecasts and effectively addressing the error problem of global weather-climate integrated models.
In addition, the model features a software framework designed for the latest high-performance computing systems, with 100,000-core high scalability, fully utilizing the parallel computing capability of the newest supercomputers.
In forecast experiments targeting several historical cases, such as the precipitation of the northern cold vortex during the landing of Typhoon Chanba in 2022, the global 5-kilometer resolution forecast of the next-generation model outperforms the current operational model.
Global weather forecast and long-term integration experiment results also demonstrate that the next-generation model has global weather forecast and climate simulation capabilities.
The next-generation model will also develop a global sea-land-atmosphere-ice coupled weather-climate integrated model based on a domestic coupler.
To further improve forecast accuracy and extend the prediction horizon, the next-generation NWP will focus on ultra-high resolution, multi-sphere coupling, multi-scale ensemble, and digital twin Earth simulation technology as key development directions.
Currently, the source code of the model has been shared through the CMA system and will be gradually expanded to a wider range of users.
It is expected that by 2025, the CMA Earth System Numerical Prediction Center will complete the next-generation global and regional weather forecast models. By 2030, a weather-climate integrated NWP operational system will be established, and multi-sphere coupled Earth system model forecast experiments will be carried out.
