Shanghai Academy of AI for Science (SAIS) and Fudan University have released Fuxi 2.0, a newly updated AI-powered big model for climate and weather forecasting that boasts higher accuracy, faster speed, and lower costs.

The first Fuxi model, named after one of the forefathers of Chinese civilization, was exhibited last year at the 28th session of the United Nations Framework Convention on Climate Change (COP28). It is China’s first sub-seasonal climate model, capable of forecasting for the next 15 to 60 days.

“Fuxi 2.0 has upgraded the mid-term weather forecasting model and the sub-seasonal forecasting model, making progress in industries such as new energy and aviation transportation,” said Li Hao, head of the SAIS earth science research team.

Fuxi 2.0 uses artificial intelligence large model technology to provide more accurate predictions of wind speeds, solar radiation, and power generation capabilities at a faster computation speed, which is critical for optimizing wind and solar power generation efficiency, balancing grid loads, and reducing instances of wind and solar energy curtailment.

It is also the world’s first meteorological model developed for navigation optimization. The model improves national maritime safety by delivering hourly forecasts with a 0.1-degree resolution for 15 days, according to Li.

“Fuxi will provide high-quality meteorological services for extreme weather and climate event response, water resource allocation, and energy market management,” said Xiong Shaoyuan, the China Meteorological Administration’s deputy director.

The advancement in meteorological forecasting technology not only enhances the accuracy and timeliness of weather predictions but also generates economic benefits for related industries.

“We look forward to continuous improvement from Fuxi, collectively reducing operating costs for China’s power industry,” said Feng Shuhai, director of the Electric Power Automation Department of the China Electric Power Research Institute.