This paper describes the design and development of the Refined Weather Forecast Platform based on the MICAPS 4 system. According to the requirements of fine forecasting service, the structure of the fine forecasting operation system and the flow of data are introduced. The overall system architecture design, the functional design, and the realization are explained based on the ModelViewView Model pattern. The platform adopts the highperformance distributed memory object caching technology, which improves the system operational efficiency and effectiveness, separates the data logic and user interface by using the model view separation, which enhances the expansibility and can be easily applied by the provincial meteorological services, develops the element consistency processing functions, and integrates precipitation and temperature objective forecasting algorithms, which provides a variety of interactive tools. The platform has been put into the official forecasting operation since July 2016, providing a powerful support for the national fine forecast business.