A quantitative scale-setting approach for building multi-scale spatial databases

Finding a balance between data redundancy and manipulation efficiency can sometimes be problematic when building multi-scale spatial databases (MSDBs). In order to render MSDBs real-timely, coarser representations with fewer vertices should be invoked once the response time exceeds the tolerable limitation. Of course, this would result in faster response speeds. With the vertex histogram making it possible to simulate the change in the number of vertices queried (NVQ) for the most complex area when the display scale shrinks, a NVQ-based approach is set forward to fix the representations by comparing the NVQ of the most complex area with the tolerable NVQ. This causes the number of levels and definite scales to be determined not only by the dataset itself but also by the generalization method. As a result, the arbitrary appointment of levels or scales when building multi-scale spatial datasets is avoided. It is argued that this approach facilitates the use of fewer representation levels to build a multi-scale database with a reasonable scale-setting scheme. Furthermore, the response time is restricted within the limit set in advance. A case study with a real spatial dataset verifies the effect of the proposed approach. © 2009 Elsevier Ltd. All rights reserved