Abstract: The paper aims to enhance the risk assessment and emergency management of grassland fire. The author built the risk prediction model of grassland fire in EOS based on the natural disaster risking theory by using the“smart grid” meteorological element forecast data, DGI index, NDSI index, DEM data and socioeconomic data. Then, the author interpolated all the factors into the same spatial scale data by ANUSPLIN interpolation software, calculated the grassland fire disaster risk index by GIS raster tool, divided the region into high risk, sub- high risk, medium risk, sub- low risk and low risk areas, did statistics of area and population, and developed a product of grassland fire risk prediction with business application capability. With the deepening of the supply- side reform of the meteorological industry, the improvement of refined meteorological services is enhanced continuously, and the diversification of professional meteorological application is developed, this study could provide a reference for the development of new fire insurance forecasting products.