Parallel Landscape Visibility Analysis: A Case Study in Archaeology

Abstract

Viewshed analysis is one of geographic information system (GIS) applications commonly used in archaeology. With the availability of large and high-resolution terrain data, viewshed analysis offers a significant computational opportunity while also posing a challenge in GIS and landscape archaeology. Although a number of studies adopted high-performance and parallel computing (HPC) for handling the compute- and data-challenge of viewshed analysis, there are few archaeological studies involving HPC to address such a challenge. It could be because of the complexity of HPC techniques that are difficult for archaeologists to apply. Therefore, this study presents a simple solution to accelerate viewshed analysis for archaeological studies. It includes a parallel computing approach with shared-nothing parallelism (each computing node accesses some specific pieces of datasets, and it has own memory and storage). Moreover, it is powerful for handling compute- and data-intensive research in landscape archaeology. In addition, the unique features of visibility patterns (irregular, fragmented, and discontinuous) may introduce useful information for landscape archaeologists. Thus, we added fragmentation calculation following viewshed analysis to further examine the influence of visibility patterns. We draw our case study from the metropolitan area of Oyo Empire, West Africa (1600–1830 AD). This parallel computing approach used an equal-viewpoint decomposition strategy on a Windows-based computing cluster. Our results showed that our parallel computing approach significantly improve computing performance of viewshed fragmentation analysis. Also, the results of viewshed fragmentation analysis demonstrated that there exists a relationship between visibility patterns and terrain information (elevation).