Massive 3D spatial data storage model based on file system

Because of the continue progress of spatial data acquisition technology and the rich data collection methods, spatial data growth as explosive speed, so the research on spatial data storage models has become a breakthrough. In this paper the concept model of spatial data storage is abstracted into geometry layer, attribute semantic relationship layer and core layer, the former supplies the representation of point, line, polygon, solid, and so on, the later supplies the representation of spatial relationship, semantic relationship and topology relationship, the LOD information is also included. Spatial objects are divided into different feature classes according to their semantic relationships and each feature class is stored in a separate files. The edited object is inserted into the end of feature class file rather than rewrite the whole file, which will improve the write speed. Spatial index is built for all the features in order to improve the query speed. Spatial objects are loaded into memory only when they are needed. Using the thought of classified storage, incremental addition, management by indexes and real-time scheduling, combined with the characters of file system and spatial data, This paper designed a robust physical model to satisfy the feature storage and support parallel dispatching using multi-thread. The storage model realizes the quickly dispatch of massive spatial objects. In contract with other file model, such as. gvp crated by KDGIS, it can improve 30%-60% better performance. This model has been successfully applied in the project called “research on 3D spatial data management and analysis components” in national 863 program of China.