The use of parallel and distributed computing techniques for the implementation of visibility algorithms

A necessary application of GIS is the ability to determine the most suitable sites for a particular development, particularly if visibility, or more specifically, visual intrusion is likely to be a key factor in gaining planning approval. To aid the site selection process in these instances, it is essential to have an indication of the nature of the terrain and, in particular, be able to determine the visibility index of either an area or set of points. The visibility index of a point is defined as the number of occurrences of a particular entity within an unobstructed line-of-sight (LOS) from that point. Entities might include buildings, postcode centroids, or most commonly, other surveyed heights within the terrain model. As the complexity of the terrain model or the number of entities increases, so does the processing overhead. This inevitably means that conventional uni-processor systems provide poor response times when calculating the visibility index. In these circumstances parallel processing techniques can be used to enhance the benefits delivered by the GIS. In the past, the authors have focused their research around a PC-based Transputer network for parallelising this problem (Ware et at, 1996). Whilst this has provided some benefits, the processors are nonetheless slow in comparison to modern processors and moreover are a specialised resource. The increasing availability of computer networks, combined with advances in modern PC operating systems, means that many organisations already have an existing multipurpose parallel processing resource which could be utilised. In this paper, the authors present algorithms and techniques for parallelising the visibility index operation on a cluster of Pentium-based workstations. In this environment the main area of concern is the division of the workload between all the available processors, such that only subsets of the regular grid digital elevation model (DEM) are mapped to each processor. Each DEM partition is processed separately and later collated to provide die overall solution. Promising results for an inter-visibility study of the South Wales valleys are presented, with emphasis on an application for determining the most suitable locations for siting wind turbine generators (WTGs) within a pre-defined area.