Currently, the Central Meteorological Observatory has established a seamless intelligent digital grid forecasting technology and product system.

Based on multi-source and multi-scale numerical model forecasts, supplemented by enhanced observations across the country and the fused real-time analysis field of China, the Central Meteorological Observatory utilizes new post-processing techniques such as machine learning and artificial intelligence to construct a comprehensive product system. This system includes minute-level rolling nowcasting, hourly updated short-term forecasts, scheduled short-to-medium term forecasts, and daily extended-range forecasts at a national resolution of 5 kilometers and a time scale of 0-30 days with intervals of 10 minutes/1 hour/3 hours/12 hours. Globally, forecasts are provided at a resolution of 10 kilometers and a time scale of 0-10 days with intervals of 1 hour/3 hours. The seamless intelligent digital grid forecasting product system has been made available to more than 10,000 cities worldwide, offering accurate point-to-point forecasting services. Validated results show that the grid forecast accuracy is significantly better than mainstream raw model forecasts at home and abroad, making it advantageous compared to similar international forecast products.

Meanwhile, the Central Meteorological Observatory has developeda refined forecasting technique for importantlocationsduring major events. This technique uses machine learning and fusion to create an objective and high-resolution forecast for single points, called STNF (Single-point Targeting and Fine-grained) forecasting. It can predict temperature, wind, and precipitation at one-hour intervals for 72 hours at specific locations, even under complex terrain conditions. This advanced forecasting capability has provided strong support for the fine-grained forecasting requirements of major events, such as the 2022 Beijing Winter Olympics. During the event, the STNF objective and high-resolution forecasts outperformed similar products in various element products' scores, including first-ranked short-term forecasts of gusts and wind direction. The accuracy of these forecasts was over 10% higher than similar forecasting products.