China's AI weather agent being deployed globally

 

Shanghai Meteorological Service rolls out MAZU urban hazard warning system worldwide, sending out alerts, mitigating disasters 

A specialized artificial intelligence system developed by the Shanghai Meteorological Service is being deployed across dozens of developing nations to combat increasingly severe urban climate risks, marking a major expansion of China's role in global humanitarian tech. 

The MAZU AI Agent for Urban Multi-Hazard Early Warning operates as a critical component of China's national response to the United Nations' Early Warnings for All initiative, which aims to protect every person on Earth from hazardous climate events by the end of 2027.

According to the China Meteorological Administration, the overarching cloud-based MAZU solution — named after the ancient Chinese sea goddess who protects travelers — is already being utilized by more than 40 national meteorological agencies, with seven countries deploying deeply customized versions of the system.

"As climate change increases the frequency and intensity of extreme weather events, strengthening early warning systems has become more important than ever," said Yu Jun from the World Meteorological Organization's Regional Office for Asia and the Southwest Pacific.

Yu made the remarks on June 1 in Shanghai at the opening ceremony of the second international training course on artificial intelligence-empowered early warning systems.

"The UN's Early Warnings for All initiative calls for every person on Earth to be protected by life-saving early warning systems. Achieving this goal requires innovation, partnership, and capacity development," he added. "It offers tremendous opportunities to strengthen every component of the early warning value chain, from observations and forecasting to risk assessment, communication and early action."

The MAZU system leverages advanced satellites, radar and localized AI models to execute hyper-local functions, including minute-level disaster data collection and analysis, customized meteorological risk assessments, real-time automated weather alert broadcasting and emergency shelter navigation mapping for citizens.

The solution has seen continuously expanding international application in developing countries to help cope with extreme weather and climate risks. Since 2024, nearly 1,000 participants from over 100 developing countries and regions have received training in China on early warning technologies.

The global rollout of the AI agent for urban hazard warning is managed by the Zhang Qian Mission, an international cooperation unit formed in 2024 by the Shanghai Meteorological Service. Composed of roughly 50 young scientists and engineers, the group coordinates system delivery, installation and technical training.

The mission has established distinct, localized pipelines to handle unique geographic challenges across Asia and Africa, customizing the AI's predictive models to the specific climate threats faced by partner nations.

In Mongolia, where authorities face severe desertification, flash floods and "dzuds" — extreme winter freezes that cause catastrophic livestock mortality — the system was officially transferred in July 2025. The tool is currently being calibrated to track regional forage availability and broadcast high-priority alerts to nomadic herders, a measure aimed directly at stabilizing the country's livestock-dependent economy.

Meanwhile, the framework is being adapted in Nigeria to address severe riverine flooding and coastal high-tide inundation. Local teams are currently piloting the cloud-based infrastructure, with an operational focus on incorporating predictive flood modeling to protect vulnerable river ecosystems and facilitate early evacuations.

A similar deployment was finalized in Djibouti in late 2025. Because the East African nation faces a volatile mix of extreme arid cycles and sudden flash flooding, engineers are focusing updates on maximizing the geographic accuracy of the system to ensure rapid, neighborhood-level alerts for urban centers.

The Zhang Qian Mission takes its name from Han Dynasty (206 BC-AD 220) explorer and envoy Zhang Qian (c.164-114 BC), whose expeditions and diplomatic missions paved the way for the Silk Road.

"A big highlight of our AI agent is that it is placed locally with interfaces prepared for users. They can upload their local data to the system, which we are not able to see on the backend," said Xu Chen, a member of the Shanghai Meteorological Service's specialized international outreach team. "They can also use the tools in the system to analyze the data, and the whole process is totally safe on its own."

Joining the mission in 2024, Xu has worked closely with the meteorological team in Nigeria.

"It is very fulfilling when we see that our technology and experience can actually help the people in other countries. Our overseas counterparts may not summarize their philosophy (of meteorological service) as 'people-centered' as we do, but we do share the same goal," said Xu.

Francis Bassey Etim, technical adviser to the chairman at Nigeria's House Committee on Emergency and Disaster Preparedness, has been working with the Chinese team since last year.

"China's meteorological agencies have really advanced technologies, which we are here to learn. We hope to incorporate them into our own system to predict when floods are coming, see how we can control it, and most importantly, how we can come together to save lives and our ecosystems," said Etim during the training course in Shanghai.

"It (the AI agent) is not only a technology for today, but for the future. Hopefully when this agent comes to my country, we can have a team able to manage it. It's going to be beneficial to citizens and for a more sustainable future for us all," he added.

Last July, Shanghai donated the AI agent to meteorological authorities in Mongolia and Djibouti as part of its initial technology-sharing efforts. In October, the Zhang Qian Mission team flew to Ulan Bator, capital of Mongolia, to install the system and conduct training. Entering this year, the training program has been expanding and the two sides are continuously exploring deeper into local conditions and system upgrading.

"We already started testing the system in our country and service, and it has helped us a lot to face the increasingly intense climate hazards such as floods, lightning, drought and desertification. It gave us many useful hints, and we hope to train the system to be more accurate and helpful," said Oyunjargal Lamjav, director of the weather forecasting department at Mongolia's National Agency for Meteorology and Environmental Monitoring.

Liu Haobo, 39, team leader of the Zhang Qian Mission, has worked with the Mongolian team from the very start.

"Livestock populations support the economy and the lives of a very large number of Mongolian people. Early warnings are never only about sending alerts, but more about informing the priority groups promptly, so they can get prepared and take early actions to confront the disaster," said Liu, who hopes to leverage AI to help them come up with a customized solution to tackle the challenge of "dzuds".

According to Liu, the mission has received extensive feedback from international partners looking to tailor the platform to local infrastructure. Key engineering priorities now include leveraging AI to eliminate local language barriers, developing more efficient SMS text-alert distribution systems, and sharpening the geographic accuracy of early warning maps.

"They believe in us," Liu said. "Every time I see the confidence and trust in their eyes, I get powered up and filled with hope for the future."