On March 5, a research team led by experts from the Chinese Academy of Meteorological Sciences (CAMS) of the China Meteorological Administration (CMA), in collaboration with multiple domestic and international research institutions, published the results of the world's first Artificial intelligence-driven Global Aerosol-Meteorology Forecasting System (AI-GAMFS) in the international academic journal Nature.

Available on Nature

GUI Ke, a core team member and associate researcher at CAMS, introduced that the system's development commenced in May 2024. Trained on 42 years' worth of global aerosol reanalysis data spanning 120,000 time instances, it is capable of generating operational forecasts covering a 5-day period globally with a three-hourly resolution in under one minute. It also boasts a spatial resolution of 50 kilometers and can produce 8 rolling forecasts per day.

System Architecture and Forecasting Strategy Credits to GUI Ke

The forecasting outputs encompass 54 variables, covering the optical properties, surface concentrations, and relevant meteorological elements of 5 key aerosol components, including dust, sulfate, black carbon, organic carbon, and sea salt.

China's Aerosol Networked Remote Sensing Observation Network Credits to GUI Ke

CHE Huizheng, stated that currently, AI-GAMFS has successfully completed the entire closed loop from scientific research breakthroughs to operational implementation.

Domestically, the system has been put into operation at the National Meteorological Centre (NMC) of CMA and meteorological departments in more than 10 provinces (autonomous regions, municipalities) such as Shaanxi and Ningxia. It has also played a supporting role in the post-analysis of sand and dust storm weather forecasts and environmental meteorology operational consultations.

In addition, meteorological departments in Gansu and others have also conducted downscaling optimization based on the system's products, achieving refined forecasting applications with a local spatial resolution of 5 kilometers.

Globally, the system has been available on the International Meteorological Early Warning Operational Support Platform at the World Meteorological Centre Beijing (WMC-BJ), delivering forecasting services globally.

Meanwhile, the system has been fully opened up in accordance with international open-source requirements, enabling localized deployment and operation with just a single server. This provides developing countries with a low-cost, high-precision solution for aerosol forecasting.

Going forward, the research team will continue to advance the iterative upgrading of the system. Meanwhile, it will develop a regional refined environmental-meteorological AI-driven system, which will also be promoted for applications in fields such as ecological conservation and environmental protection and transportation, contributing Chinese solutions to the global response to climate change.

Author: LUO Lan

Editor: LIU Shuqiao