Black Rainstorm in Hong Kong on 14 August 2025 – Observational Analysis and Forecasting/Nowcasting Aspects

The rainstorm in Hong Kong on the morning of 14 August 2025 disrupted the daily lives of the public and significantly impacted airport operations. A series of artificial intelligence (AI)-based products were prepared for integration into operations for forecasting/nowcasting purposes, including quantitative precipitation forecasts by several global AI models and an AI-driven mesoscale numerical weather prediction model, AI-based rainfall nowcasting methods, and AI-based guidance-on-guidance. In this study, we document the synoptic, mesoscale, and storm-scale processes that lead to heavy rainfall and evaluate the performance of conventional and AI-enabled models. The AI products appeared to work well in providing timely alerts for this heavy rain episode. Their performances and potential applications are presented herein, which could be useful references for other meteorological authorities facing similar flash flooding challenges in coastal areas. This is the first analysis of the persistent initialisation of convection in an isolated area and the appearance of circulating weather radar echoes, based on a three-dimensional wind field retrieved from weather radars. Directions for future studies of this rainstorm are also discussed.