HKUST's AI Breakthrough Transforms Storm Forecasting

In a critical advance for climate resilience, researchers from The Hong Kong University of Science and Technology (HKUST) have developed an AI model that can predict dangerous convective storms—including Black Rainstorms, thunderstorms and extreme heavy rainfall like those that have hit Hong Kong—up to four hours before they strike.

TL;DR
• HKUST researchers develop AI model that predicts severe thunderstorms and rainfall up to four hours in advance.
• The model improves forecast accuracy by over 15% at a resolution of 48 kilometers.
• The breakthrough strengthens the overall accuracy of the national weather forecasting system and promises to transform early warning systems for vulnerable communities across Asia.

Summary
Researchers at The Hong Kong University of Science and Technology (HKUST) have developed an AI model that can predict severe thunderstorms and rainfall up to four hours in advance, improving forecast accuracy by over 15%. The model, called the Deep Diffusion Model of Satellite Data (DDMS), uses satellite data and advanced deep diffusion technology to predict convective storms, including Black Rainstorms, thunderstorms, and extreme heavy rainfall. This breakthrough strengthens the overall accuracy of the national weather forecasting system and promises to transform early warning systems for vulnerable communities across Asia.

The DDMS model was trained using infrared brightness temperature data collected by China's FengYun 4A satellite from 2018 to 2021, and its performance was validated using samples from the spring and summer seasons of 2022 and 2023. The model's high-resolution, high-frequency forecasts are updated at approximately 15-minute intervals, covering a region of about 20 million km2, including China, Korea, Southeast Asia, and surrounding areas.

Key benefits of the DDMS model include:
• Improved forecast accuracy of over 15% at a 48-kilometer spatial scale
• Stable performance across multiple spatial scales (4 km to 48 km) and throughout different seasons
• High-resolution, high-frequency forecasts updated at approximately 15-minute intervals
• Reliable forecasts with accuracy improvements ranging from 3% to 16% and averaging 8.26% in the 2–4 hour forecast window

The research team, led by Prof. SU Hui and Dr. DAI Kuai, collaborated with national meteorological institutions, including the China Meteorological Administration's (CMA) Institute of Tropical and Marine Meteorology (ITMM) and the National Satellite Meteorological Center (NSMC). The study was published in the Proceedings of the National Academy of Sciences (PNAS).

This research has the potential to save lives and reduce economic losses from severe weather events, and it highlights the importance of continued investment in AI research and development for climate resilience.