Abstract: Under the dual pressures of climate change and human activities, the energy balance between the terrestrial and atmospheric spheres has been disrupted, leading to significant anomalies in hydrological cycles. Consequently, extreme disasters such as rainstorm floods have shown increasing frequency. In order to effectively cope with the increasingly frequent extreme floods and droughts in the Changjiang River Basin, this paper elucidates that the seasonal advance and retreat of the Western Pacific Subtropical High serves as the critical driver of rain belt migration. It analyzes temporal variation patterns of the Meiyu front and autumn rainfall in West China, investigates the thermodynamic response mechanisms of extreme precipitation and surface runoff to climate change, and elaborates on the fundamental characteristics of Hook structure migration under global warming. Projections indicate a widespread increase of 1~3℃ in peak temperature points by the end of this century, thereby triggering more severe extreme hydrological disasters. Additionally, recent advances in AI-based meteorological models and deep learning algorithms for ensemble probabilistic hydro-meteorological forecasting are reviewed. Building on these analyses, a novel paradigm integrating engineering and non-engineering measures is proposed. Engineering measures focus on enhancing basin-scale flood storage and drainage capacity, while non-engineering measures emphasize refined flood-season staging, dynamic control of operational water level and multi-objective joint optimization of reservoir groups using flood forecasts. Supported by cutting-edge AI technologies, this paradigm effectively addresses the growing extremity, complexity, and uncertainty of rainstorm floods under climate warming. It provides robust support for improving the Changjiang River Basin′s resilience against water disasters and safeguarding water security.