On May 6, 2025, a research team led by Chenguang Zhou introduced FengShun-CSM, an artificial intelligence-based climate system model capable of delivering 60-day global daily forecasts across 29 critical environmental variables. This development marks a significant advancement in predictive meteorology, offering enhanced accuracy over traditional models.

FengShun-CSM integrates interactions among the atmosphere, ocean, land, and cryosphere, providing a comprehensive approach to climate modeling. Traditional models have often struggled to accurately represent these complex interactions over extended periods. By leveraging machine learning techniques, FengShun-CSM addresses these challenges, resulting in more reliable forecasts.

In comparative analyses, FengShun-CSM demonstrated superior performance over the European Centre for Medium-Range Weather Forecasts (ECMWF) subseasonal-to-seasonal (S2S) model, particularly in predicting precipitation patterns, land surface variables, and oceanic components. This enhanced capability is largely attributed to its improved representation of intra-seasonal variability modes, notably the Madden-Julian Oscillation (MJO), a key driver of tropical weather patterns.

The implications of FengShun-CSM's advancements are far-reaching. More accurate long-term forecasts can enhance disaster preparedness, potentially reducing loss of life and property. Improved oceanic predictions support marine conservation efforts and sustainable fishing practices. Additionally, reliable land surface and precipitation forecasts assist farmers in optimizing planting schedules and resource allocation, leading to increased agricultural productivity and food security.

FengShun-CSM builds upon previous AI-driven climate models. In December 2023, FuXi-S2S provided global daily mean forecasts up to 42 days, outperforming ECMWF's S2S model in predicting total precipitation and outgoing longwave radiation. Earlier, in June 2023, FuXi delivered 15-day global forecasts with a temporal resolution of six hours, matching ECMWF's ensemble mean performance. By September 2024, FuXi-2.0 advanced machine learning weather forecasting by providing hourly global forecasts and integrating both atmospheric and oceanic components. FengShun-CSM distinguishes itself by offering extended 60-day forecasts and encompassing a comprehensive set of 29 critical variables across multiple domains, marking a significant step forward in AI-based climate modeling.

The introduction of FengShun-CSM represents a pivotal advancement in climate system modeling, leveraging artificial intelligence to provide more accurate and extended forecasts. Its superior performance compared to traditional models underscores the potential of AI in enhancing our understanding and prediction of complex climate systems, with profound implications for disaster preparedness, environmental conservation, and agricultural productivity.