Abstract: Under global warming, the increasing frequency of extreme rainstorm-induced floods poses severe challenges to traditional rainfall and water regime monitoring and forecasting systems. Focusing on the Lushui River Basin, a key flood control area in the middle reaches of the Yangtze River, this study established an integrated "sky-space-earth-water-project" monitoring and perception system in the basin, coupling hydrometeorological radar retrieval, rainfall fusion forecasting, rainwater retention capacity assessment, distributed hydrological modeling, hydrodynamic coupling computation, and forward-reverse-forward optimization. The resulting Three-Lines-of-Defense framework spans "cloud precipitation-ground rainfall-riverine floods", enabling end-to-end flood forecasting and proactive control. During the 2025 extreme rainfall event, this approach successfully limited the forecast error for the Lushui Reservoir's inflow flood peak to within 3%, achieved a deterministic coefficient of 0.94 for the flood process forecast, and accurately predicted the end time of the heavy precipitation event three hours in advance. By comparing and recommending optimized reservoir operation schemes, it ensured the safe operation of the Lushui Reservoir. This achievement marks a shift in operational decision-making from passive response to active judgment, providing robust support for scientific and early decision-making in basin flood control operations.