Pixel Loss and Duplication during Projection of Global Grids

Karen A. Mulcahy

 

Department of Geography

East Carolina University

Greenville, NC 27858

Email: mulcahyk@mail.ecu.edu

 

 

Little attention has been paid to the consequences of transforming global grids by map projections. The transformation of a discrete, raster structure is very different from the transformation of vector data. Vector data employs the time tested point-to-point method, is very flexible in terms of transformation, and standard cartographic theory regarding map projection transformation applies. The transformation of less flexible, global raster data sets is more complex. Given current implementations of projection transformation, data changes at the pixel level are inevitable. To track the changes at the pixel level, two metrics PL and PD, are applied to the study of global equal area projections. Projections such as the Craster Parabolic, Eckert IV and VI, Goode Homolosine, Sinusoidal, Mollweide, and Hammer-Aitoff all maintain the property of true relative area. While, one might assume that all pixels would be retained in a transformation from the Earth surface to the plane but this is not the case. In addition, the primary goal when choosing a projection is to minimize distortion. In the choice of global equal area projections, the property of angular distortion is often minimized. The results at the pixel indicate that the opposite choice may be appropriate. Greater angular distortion provides a higher retention of original pixels without duplication.