Propagating effects of database generalization on the viewshed

Few studies have systematically examined the effects of different possible generalization methods on the products of GIS operations. In this paper the effects of generalizing a Digital Elevation Model (DEM) on the area which is determined to be visible (the viewshed) is examined. Among the many different methods of generalization which are possible, a limited set of operators is examined here. First, they all fit geometrically from one resolution to another, and they can be grouped into two types: regular sampling and statistical summaries. In the latter class four different operations are used: for any cell at the target resolution, elevations are determined, from the arithmetic mean, the maximum, the minimum, and the maximum difference from the mean of the cells within the kernel at the original scale. Changes in resolution of 0.5, 0.33, 0.25, and 0.2 of the original study area are studied at 100 viewing points in each of two study areas. At the original resolution a 120 × 120 pixel area was examined, giving generalizations to 60 × 60, 40 × 40, 30 × 30, and 24 × 24 pixels. The viewsheds determined over these different resolution DEMs are compared with a number of possible viewsheds derived by generalization of the viewshed over the original DEM. Of those tested, the maximum deviation from the mean within kernel emerges as the most reliable estimate of the pattern and area of the viewshed at all resolutions. The importance of this conclusion is that different generalization operators yield more or less faithful versions of the ungeneralized product (the viewshed), and it seems indicative that a similar variety in stability of the product with generalization operator will result in many, if not all, complex products of spatial analysis.