Abstract: In order to share and make better use of geographic information in agricultural applications, a spatial grids division and coding method based on Morton code was introduced in this paper. Taking the land utilization layer of Ningxia province as the example, the entire sample region was divided into three spatial grids layers and the final spatial grid’s size of third layer was refined to 1 m2. Encoded with the two dimension running encode (2DRE), the number of spatial grids could be highly compressed. At the same time, by summing up the general requirements in the application of spatial grids, a two-layered index structure was designed within relational database. In the experiment, a specific area with about 6×104 polygons was selected to validate SQL Server 2008 and VS2010.net 4.0 software. It was the conclusion that the spatial grids division and coding method based on Morton code could make the grid data shrink to one-tenth of its original number, the method had also overcome the technical difficulties that every figure had a different number of 2D run- length compressing coding and the retrieval speed of spatial data was improved to tens of thousands of records every second. This paper provides an effective method for the organization and management of the agricultural information resource.