China's AI weather warning system Mazu expanding across Global South

Seven developing countries are now using a Chinese AI-powered weather warning system on the ground to help them better prepare for natural disasters such as typhoons and rainstorms. Pan Jinjun, chief engineer at the China Meteorological Administration, described the system – named Mazu – as an international public good that provided “tailored early warning services for developing nations”. Speaking to several media outlets ahead of the World Artificial Intelligence (AI) Conference in Shanghai, Pan said it was the first national project responding to the UN Early Warnings for All initiative, which aims to ensure everywhere on Earth is protected from extreme weather. The public cloud version of the Mazu (Multi-hazard Alert, Zero-gap and Universal) system, which offers quicker access to meteorological data, has been tested by 40 Global South countries, according to People’s Daily. The full system has also been deployed on the ground in seven countries – Pakistan, Djibouti, Ethiopia, Jordan, Mongolia, the Solomon Islands and Sri Lanka – where it has been physically woven into daily operations. The system, named for the sea goddess Mazu, was intended from the outset for global use and first unveiled at last year’s World AI Conference in Shanghai. According to Pan, it integrates China’s own AI weather forecasting models and data from meteorological satellites with observation data from partner countries and provides “menu-style, customised services” based on the user’s needs.

Pan said the system had already “won broad recognition from international organisations”. Last week’s conference heard details of how Mazu was being used. For example, it is used to issue warnings about multiple hazards in Pakistan and provides detailed weather forecasts to Djibouti’s port authorities. Meanwhile, in Nigeria, the cloud version is used for urban flooding forecasts. The system has gone through a number of upgrades this year, with Djibouti, which was given the 1.0 version last year, formally receiving the 2.0 system at this year’s conference. A representative from the meteorological administration told China News Weekly that the updated system had improved the accuracy of forecasts by cutting its spatial resolution from 9km to 3km (5.6 miles to 1.8 miles), and could now provide updated forecasts every six hours rather than once every three days. They added that Mazu would also be made available to industry bodies that needed detailed weather services, such as shipping firms or port authorities. Pan said AI was fast becoming a core engine for improving extreme weather warnings by providing rapid forecasts that covered a wide area. He added that it was also able to use satellite, radar and other data to better spot and track typhoons, cyclones and heavy rain. The development of such systems highlights China’s growing importance in the global development of AI, but Pan warned that AI forecasting models were still being held back by the limited amount of real-world data available. “Building AI capabilities in the Global South is becoming a new focus of governance,” Song Haitao, a member of the UN’s Independent International Scientific Panel on AI and director of the Shanghai AI Research Institute, told the conference on Friday.

“What is more important is to genuinely help more countries and institutions participate together and to foster the global, wise development of AI.” Other speakers told the event that China’s early exploration of AI governance – together with its active role in international standard-setting and capacity-building – offered valuable support for global efforts to develop responsible and inclusive AI, according to state-backed broadcaster CGTN.