Projecting Global Raster Databases

The need for regional and global models for spatially distributed phenomena and processes has driven the development of large raster databases of terrain elevation, population, land cover, vegetation, soils, temperature, and rainfall. The use of commercial geographic information system software to process these databases requires projection from spherical coordinates to plane coordinate systems and transformation from one plane system to another. Recent research has demonstrated that such projection and transformation of raster databases is error-prone and may generate data that corrupt the results from spatial models. This paper presents two approaches to handle the problems of raster database projection and transformation. The first is the development of a Web-based decision support system to aid users with the selection of appropriate projections based on data type, areal extent, location, and properties to be preserved. The second is a dynamic approach to projecting raster databases that adjusts the output pixel size according to the exact area of the corresponding pixel in the spherical coordinate system.