LOD-quadtree: A Spatial Index for Secondary-Storage Selective Refinement

Multi-resolution Terrain Model (MTM), which is widely used for visualizing large terrain, is initially designed as a main-memory method. Such an assumption is no longer valid with the rapidly increasing terrain size, which makes the data exchange between main memory and secondary storage the bottle-neck of multi-resolution terrain visualization. In this paper, a novel multi-resolution access method, the LOD-quadtree, is proposed to improve the efficiency of data retrieval for existing MTM based on Triangulated Irregular Network (TIN) without any modification. The proposed method integrates the spatial and resolution information to effectively reduce the I/O cost of data access. The LOD-quadtree also has a simple and flexible structure that is designed for multi-resolution terrain data. Test results demonstrate significant performance improvement introduced by these two features of the LODquadtree.