MAP PROJECTIONS FOR GLOBAL AND CONTINENTAL DATA SETS AND AN ANALYSIS OF PIXEL DISTORTION CAUSED BY REPROJECTION

With growing emphasis on global monitoring research using remotely sen sed data and geographic information systems is increasingly focused on large regions studied at small scales. These global change studies re quire the integration of data sets from several sources that are repro jected to a common map base. In small-area, large-scale studies the ch oice of a map projection has little effect on data quality. In global change studies the effects of map projection properties on data qualit y are more apparent, and the choice of projection is more significant To aid compilers of global and continental data sets, six equal-area p rojections were chosen: the interrupted Goode Homolosine, the interrup ted Mollweide, the Wagner IV, and the Wagner VII for global maps; the Lambert Azimuthal Equal-Area for hemisphere maps; and the Oblated Equa l-Area and the Lambert Azimuthal Equal-Area for continental maps. Dist ortions in small-scale maps caused by reprojection, and the additional distortions incurred when reprojecting roster images, were quantified and graphically depicted. For raster images, the errors caused by the usual resampling methods (pixel brightness level interpolation) were responsible for much of the additional error where the local resolutio n and scale change were the greatest.