Assessing Raster Representation Accuracy Using a Scale Factor Model

Raster datasets of global and continental extent are subject to error resulting from projection transformation. This paper examines the error problem from a theoretical perspective and develops a model to calculate the extent of the errors. The theoretical examination indicates that error results in two forms, areal size change of pixels and categorical error resulting from loss or duplication of pixels. A scale factor model, based on the horizontal and vertical scale factors of the projection, is developed to provide a computation of the resulting error from specific projections. The model is experimentally tested with the cylindrical equal area, sinusoidal, and Mollweide projections. Results indicate that the model predicts error within one percent of actual values and that the sinusoidal projection is subject to smaller errors in projecting raster data than the other projections tested.