During the opening ceremony of the World Artificial Intelligence Conference (WAIC) on July 26, the China Meteorological Administration (CMA) launched MAZU, a Joint Action for Early Warnings for All, to build an Early Warning Service Network and to share practical experience and technological achievements with global partners.

On this occasion, the AI Agent for Multi-Hazard Early Warning, MAZU-Urban was donated to representatives from Djibouti and Mongolia. This marks the official launch of the System, which deeply integrates artificial intelligence (AI) technology with meteorological science.

Under MAZU's framework, the Shanghai Meteorological Service has taken the lead in jointly developing MAZU-Urban in collaboration with the National Meteorological Centre of CMA, China Unicom Shanghai Branch, Shanghai Academy for Science and Intelligence, Shanghai Artificial Intelligence Laboratory, and other institutions. Powered by AI technology as its core engine, MAZU-Urban deeply integrates advanced algorithms with multi-source data, and is equipped with cloud-based early warning system products, meteorological open-source models, as well as the Africa-region model developed in Shanghai.

During the interview

MAZU-Urban integrates Shanghai's independently innovated AI-powered meteorological models. By leveraging knowledge enhancement, multi-modal data fusion, and deep reasoning techniques, it constructs an early warning corpus and develops custom-made forecasting and early warning generation and distribution functions for specific regions based on the global forecasting model framework. This makes for a comprehensive early warning system that covers the entire chain from "risk knowledge - monitoring and early warning - warning dissemination - emergency response."

The interface of MAZU-Urban

In response to the practical needs of regions such as Macao, China, and Mongolia, the R&D team has developed role-specific and hazard-specific workflow prompts for disaster prevention and mitigation guidelines. These prompts guide large language models to engage in in-depth reasoning and dynamically generate preventive strategies tailored to local conditions.

Geared to the complexity of global application scenarios, MAZU-Urban innovative adopts a "three-terminal integrated" architecture. The all-in-one machine terminal serves as the intelligent hub, housing a complete set of AI knowledge pairs and prompt workflows to support precise decision-making by professional departments. The tablet terminal focuses on key industries such as port and shipping, providing customized meteorological risk assessments through AI models. The mobile phone terminal directly reaches the public, offering intelligent Q&A services powered by AI models and delivering location-based disaster alerts and evacuation guidance.

Since January in 2025, MAZU-Urban has been in trial operation in 35 countries and territories across Asia, Africa, and Oceania, receiving widespread acclaim.

MAZU-Urban of Shanghai Meteorological Service

And on the occasion of 2025 SDG Cities Global Conference, the Shanghai Meteorological Service unveiled the co-construction action plan of MAZU-Urban.

In active response to the advocate for "Healthy Shanghai", meteorological departments of Shanghai have collaborated with health departments to deliver health meteorological services and effect assessment in Fengxian District, with the aim of developing and utilizing climate resources, enhance the self-management capacity of key population groups regarding health, and achieve the transformation of public health services from "disease treatment" to "risk prevention".

In August, the Shanghai Climate Centre, the Fengxian District Meteorological Office, and the Fengxian District Center for Disease Control and Prevention signed a framework agreement on meteorological and health research cooperation. According to agreement, the three parties will establish a mechanism for sharing meteorological data and public health information, facilitate joint assessments of the impacts of climate change on population health, and enable targeted strengthening of health risk warnings and preventive measures for key population groups, along with conducting quantitative evaluations of their effectiveness.

On October 10, the Fengxian District Meteorological Office, in collaboration with the District Center for Disease Control and Prevention, officially launched the "Health + Meteorology" project. The first batch of pilot communities selected were Nanqiao (representing urban areas) and Tairi (representing rural areas) in Fengxian. The project aims to deliver risk forecasting services for cardiovascular and cerebrovascular diseases during autumn and winter, and complete an evaluation of the service effectiveness. Additionally, through family doctors, it precisely disseminates health protection guidelines for extreme weather conditions to key population groups, helping individuals with chronic diseases and others to take effective self-protection measures.

Currently, the Fengxian District Meteorological Office has set up a climate change health risk monitoring and assessment platform—a large-screen display for health meteorology scenarios. By presenting real-time disease risk levels, dynamic changes in high-risk populations, distributions of high-risk locations, and future trends, the platform provides references for government and industry regulatory departments to assess and respond to future health risks.

Editor: LIU Shuqiao