4. The Methodology of Viewshed Construction

As the WF1415 towers are not preserved to their original heights (see section 4.3), viewshed analysis provides a way to explore missing structures and how views extended both to and from WF1415. Viewshed analysis involves a calculation carried out upon a digital elevation model (DEM) with the aim of mapping areas in and out of view from a given viewpoint. In this way a calculation of what can theoretically be seen from that point can be made from any point on the DEM (Wheatley and Gillings 2002, 204).

The first step is to gather information about the Faynan and discuss the particularities of using viewsheds in this environment and with this population. One of the main criticisms with viewshed analysis is its uncritical application (see Wheatley and Gillings 2000). The methodology here incorporates steps to take into account the unique factors of the Faynan region, rather than creating a generic viewshed. What can be viewed can be generated by a computer; however, what is possible to see and at what distance influences the effectiveness of surveillance.

A Digital Elevation Model (DEM) was created from paper maps of the Levant at the scale of 1:50,000km that were scanned and geo-referenced in WGS84 projection at the National Resource Authority of Jordan in a collaborative project with the National Endowment for the Humanities, the American Centre for Oriental Research and the Department of Antiquities, Jordan. These are available as part of JADIS (The Jordan Archaeological Database & Information System) (Savage 2008). The contour data of the scanned map were digitised by hand.

The DEM was created from these contours with ESRI's ArcInfo Topogrid Tool. This program mimics hydrological effects when creating a DEM, leading to greater accuracy (Conolly and Lake 2006, 110). The resulting DEM was of a 10 metre resolution.

As with all DEMs created from paper maps, there are inaccuracies. Ideally paper maps should not be over-handled or exposed to changes in humidity or temperature. These conditions and ill treatment can cause the paper to expand or wrinkle, warping the contour lines. This is unavoidable, but utmost care was taken in handling the paper maps to limit these effects (Savage 2008).

In order to test the general accuracy of the DEM it was compared to the elevations of sites collected during surveys of the area (Levy et al. 2001). This functioned as a check that the recorded elevations of the sites fell within the corresponding contour lines. The elevations were found to be within ± 10m of the recorded heights, with no great discrepancies between them. Although survey points were not collected throughout the entire landscape, this test provided a basic check of the robustness of the DEM.

DEMs can be generated in other ways – for example using radar data; this admittedly would have created a more accurate terrain model (Connolly and Lake 2006, 77). Conducting initial searches at the start of this project, it was difficult to find a high-resolution DEM created from radar data at a reasonable price. Moreover, given the limited period available to conduct research it was decided that it was not effective time-management to become proficient at programs that analyse remote sensing data. Over the course of these investigations the accessibility of these datasets has increased tremendously. A future avenue of research would be to incorporate radar-generated DEM into the project. However, the DEM created here is robust enough for the purposes it is used for, i.e. modelling potential views.


© Internet Archaeology/Author(s) URL:
Last updated: Tue Nov 3 2009