Abstract:

Taking into account both the spatial and temporal dimensions of raster data, spatiotemporal aggregation plays an important role in analyzing and predicting the space phenomena. Because the current spatiotemporal aggregation algorithms oriented to the geographical raster images are short of accuracy, two precise algorithms called "Tile Pasting" and "Window Segmentation"  are proposed. We present basic principles and detailed implementations for them, analyze their properties and applicable conditions respectively, and test their effects and efficiencies in our spatial database named "BeyonDB". The test result indicates that, both of the two algorithms can  solve the problem of inaccuracy in the traditional methods with a slight loss of efficiency.

Key words: spatiotemporal aggregation;geographical raster image;“tile pasting&rdquo, algorithm;“window segmentation&rdquo, algorithm