Abstract: In geoscience field,the massive volume data has become a bottleneck problem for the development of true 3D GIS.Because of the low performance of database access techniques,the traditional massive volume data organization relies heavily on file system.But this method has great limitation,such as the insecurity and poor share of data,concurrent access abnormality,data redundancy,and lack of managment.To solve these problems,a new space division method,the homologous octree,is proposed in this paper.Furthermore,a massive volume data organization method based on relational database with its strategy for space search,multiresolution,data swapping between main memory and external storage is also demonstrated.The practice shows that the performance of this method will meet the need of the actual applied requirements，such as 3D visualization.Since the relational database is the most practical and popular kind of database,this method has a better practical value and application prospect.

Key words: true 3D GIS, relational database, massive volume data, octree

摘要： 在地学领域，体数据的海量性已经成为制约真三维GIS发展的瓶颈之一。由于受到目前数据库访问技术的限制，传统海量体数据的组织还主要依靠文件系统，从而导致安全性和共享性差、并发访问异常、数据冗余以及无法统一管理等诸多问题。针对以上缺陷，提出一种类八叉树空间划分思想，在此基础之上，给出了基于关系数据库的海量地学真三维体数据组织方法，并实现了相关的空间查找、多分辨率和海量数据内外存交换策略。实践证明，该方法在性能上可以满足海量数据的组织及三维可视化等实际应用的需要。鉴于关系数据库是目前实用性、普及性最高的数据库，因此该方法具有较高的推广性和实用价值。

关键词: 真三维GIS, 关系数据库, 海量体数据, 八叉树