Chinese Journal of Computers   Full Text
Title	A Method for Organizing Map Dataset to Support Multi-scale
Authors	YE Chang-Chun ZHOU Xing-Ming
Address	(School of Computer, National University of Defense Technology, Changsha 410073)
Year	2004
Issue	No.7(964—970)
Abstract &   Background	GIS applications have a property called multi-scale in displaying map. Multi-scale means that the scale of map display determines the detail level. The conventional spatial indexes such as Hilbert-R-Tree don’t adapt to this property. The access methods based on them have two disadvantages: The data records of features of same rank are non-clustered, which are always used togethe; Because only a data record of a feature is read for each I/O operation, I/O granularity is too small. These two disadvantages make the data access inefficient. So, this paper presents a new spatial index called Hierarchical Hilbert-R-Tree (HHRT). Unlike Hilbert-R-Tree, which displaces all index records in the lowest layer of tree, HHRT distributes index records of different ranks among different layers of tree. The ranks increase from bottom layer to top one (if any). All data records are ordered by the width-first sequence of index records. Then, The data records of features of same rank are clustered. Furthermore, the data records referenced by index records of same leaf nodes are sequentially stored. HHRT call these set of data records cluster. The data size of clusters are close to the size of a disk page. For each I/O operation, a cluster instead of a data record is read from disk. So I/O granularity is optimized. Together, the two disadvantages of the access method based on Hilbert-R-tree are banished. Experiments based on real map dataset prove the data access method is more efficient based on HHRT than based on Hilbert-R-Tree. keywords map dataset; spatial index; data organization; data access; hierarchical Hilbert-R-Tree