It has been common practice within archaeological research to perform viewsheds from a single viewer point. Nevertheless, as Mitcham (2002) has shown, this may lead to only partial visible areas from a given site being identified, particularly when the viewer location covers an area rather than a discrete point (Figure 4).
Figure 4: Viewsheds from the same site (St Catherine's Hill) using one and multiple viewer points. The visible area is shaded grey. On the left, the viewshed from the highest point in the centre of the hillfort, and on the right this first coverage has been combined with the viewsheds from four other viewer positions around the circumference of the fort. It is clear that a viewshed from the highest point within this fort does not give an accurate representation of the visible area from the site (Mitcham 2002, 77, fig. 10.3).
Since the hillforts encompass large areas of a given hill-top, a single viewpoint viewshed would not seem appropriate. A multiple viewer point approach has, therefore, been implemented.
Archaeological studies of Iron Age hillforts have applied this method following different criteria. For example, Mandry and Rakos (1996, 111) calculated viewsheds from the four corners of hillfort ramparts, selecting an eye height of 5m above the terrain to simulate a 'looker' located upon the rampart. Similarly, Mitcham (2002, 76) chose five viewer positions for each hillfort: one in the middle or highest point of the site and the others in the inner ramparts, roughly corresponding to the four major compass directions. In the latter case, the viewer height was 1.5m plus an unspecified extra height to account for the rampart. Both papers assume that there would have been lookouts controlling the landscape from different positions at the top of ramparts or towers. Zamora (2006, 42), in contrast, considered a range of observer locations around the hill of the site and nearby hills with higher visibility. This is presumably to obtain the maximum possible visibility from the area (the precise rationale is not specified). Two offsets were taken into account in Zamora's work: 10m, assuming visibility from ramparts or towers, and 2m, i.e. a person on a slight elevation.
3.1.2 General criteria
In the current study, the approach advocated by Mitcham has been adopted, with five viewer points selected: one in the highest point in the centre of the hillfort and the others around the hill, roughly corresponding to north, south, east and west (Figure 5: centre). Ramparts, some with towers, are documented in the majority of hillforts of the study area [note 1]. As a result, an eye height of 6m above the terrain has been selected. The possibility of running viewsheds from numerous points around the hill (as advocated by Zamora 2006) with the aim of obtaining the maximum visible area was considered. However, one trial for every hillfort indicated that while this would have considerably increased the time to run the analyses (which would have been specially relevant when calculating probable viewsheds) overall the results would have been virtually identical (compare Figure 5: centre and bottom).
3.1.3 Discussion
In contrast to Mitcham's results (Figure 4), the multiple and single viewer point viewsheds (the latter based on a centre point) of our Iberian hillforts are similar, as can be seen in Figure 5 (top and centre) [note 2]. The reasons for this divergence from the results gained by Mitcham are unknown. It could be attributed to differences in offsets, viewer point locations, terrain around the hillforts or the precise algorithm implemented.
Despite their similarity to simple binary viewsheds, in this case, multiple viewer point viewsheds are relevant. On the one hand, relatively small divergences in visible areas could be archaeologically significant. On the other, multiple viewer points can mitigate against relevant changes in the results of binary viewsheds that may stem from varying the location of the viewer point. For example, a given viewer point in a slightly different location may result in visibility over inland and coast (Figure 5: top) or, by contrast, only over inland areas (Figure 6: top). Additional viewer points can help to overcome any misplacement (compare Figures 5 (centre) and 6 (bottom).
Figure 5 (left): Comparative viewsheds from Puig Castellar hillfort. From top to bottom: binary viewshed, multiple viewshed from five viewer points, and multiple viewshed from many points around the hill Figure 6 (right): Comparative viewsheds from Puig Castellar hillfort. From top to bottom: binary and multiple viewsheds displacing the same viewer point
Or compare these figures side by side.
