AI and the World's Hard Problems · Earth & climate science · arXiv · 2026-10-22
Huayu-Global: Skillful real-time quasi-global precipitation retrieval
Satellite precipitation products face a trade-off in which real-time estimates are timely but less accurate while the most reliable products are released months later, keeping high-quality precipitation outside the decision window for flood warning and disaster response. Huayu-Global, a cross-constellation deep-learning system, combines three geostationary sensor streams from five satellites through a common retrieval framework and generates coherent quasi-global precipitation fields every thirty minutes. Across more than a million independent station records from two periods outside the training years, it improves event detection and precipitation-amount estimation over the evaluated real-time products and reaches delayed-product-level skill in key metrics.