DATA QUALITY IMPLICATIONS OF RASTER GENERALIZATION

This study is concerned with the data quality implications of raster generalization. The study focuses specifically on the effects of neighborhood-based generalization (categorical filtering) on thematic accuracy. These effects are examined empirically using raster land cover maps. Accuracy is defined in terms of changes in class membership between original and generalized maps. Results indicate that changes are concentrated in those portions of the map and for those classes that exhibit high levels of spatial variability