ABSTRACT
Intervisibility and viewshed analysis are amongst the most common functions of those GIS that support digital terrain modelling. In the main, they are used to aid planning decisions, such as the siting of contentious developments with the aim of minimising visual intrusion, or alternatively for identifying locations which maximise the field-of-view, such as for broadcast coverages, scenic viewpoints, watchtowers, or missile defences (Franklin and Ray, 1994). Whilst the viewshed quantifies visibility for a limited set of test locations, its major shortcoming is that it does not identify optimal locations. An alternative is the reverse viewshed, which quantifies the visibility at each possible location within the planning zone, thus identifying both intrusive and discreet sites. However, whilst viewshed analysis is widely acknowledged as being processor-intensive, the calculation of the reverse viewshed can increase this workload by many more orders-ofmagnitude. This chapter describes the implementation of the reverse viewshed algorithm on a distributed cluster of networked machines. The practicality of parallelising many GIS operations is now within the reach of many users, such as corporate industry, academia, local and national government, and environmental agencies, without the need for any further investment. The key to unlocking this opportunity is to demonstrate how existing resources (i.e. networked computers) can be used for popular, but processor-intensive tasks.
