The flood early warning and forecast platform needs to be supported by many hydrological data, and to deal with the absence and anomaly of hydrometeorological data in small and medium-sized rivers, this study proposes the quality control technology and method of multi-source hydrometeorological data. The model encapsulation service technology with state and interaction characteristics is studied and developed in cloud environment to provide multi-level on-demand services of data, algorithms, and models. The vertical channel inundation algorithm of small and medium-sized rivers is proposed, and the visualization of flood inundation based on UE4 virtual engine is realized. Based on above key technologies, an intelligent flood forecasting and dispatching platform for small and medium-sized rivers is developed. At present, the platform has been applied in the Tunxi demonstration basin in Anhui Province, and extended to the Qinhuai River in Nanjing and Zhangzhou of Fujian Province, providing important support for the flood prediction and early warning of small and medium-sized rivers.