Abstract: According to the computer resources available in Shanghai Regional Meteorological Center and numerical forecast products of National Meteorological Center, the schemes of lateral boundary nesting, cumulus parameterization and planetary boundary layer parameterization are selected. On the basis of the dynamical frame of NCAR-PSU/MM5V2, the methods of data analysis and assimilation, post-processing and vision software of forecast products are studied. In addition, the fifth-generation operational numerical prediction system of Shanghai Regional Meteorological Center is set up and the surface element predictions per 3 h for 138 cities over East China can be provided. Through half an year quasi-operation, it is shown that the prediction performance is good, and it is valuable for the operational forecast of meteorological observatory.