ABSTRACT
Many of today’s GIS provide tools that perform terrain analysis. One such tool is the lineof-sight (LOS) function, which calculates the intervisibility between an observer and a target location. There has been much previous work carried out in this area (Fisher, 1991; Franklin and Ray 1994; De Floriani and Magillo, 1994; Wang et al., 1996), leading to a variety of LOS strategies and algorithms. Multiple LOS operations can be performed from a fixed observer to multiple target locations to form a visibility map or viewshed (Yoeli, 1985). These intervisibility functions make use of representations of the Earth’s surface, termed digital terrain models (DTMs), which are inevitably subject to error. The quality of a particular DTM will have an obvious effect on the accuracy of any LOS analysis performed on it. It has been shown that small elevation errors can propagate through to large application errors, particularly for intervisibility and viewshed analysis (Huss and Pumer, 1997). It follows that for many applications, very accurate DTMs are required in order to obtain meaningful results. This chapter sets out to raise awareness of some of the error-related problems that can arise when performing intervisibility analysis, and suggests ways in which these can be addressed.
