The aim of this article has been to propose a standard method for generating viewsheds that overcome a number of limitations inherent in binary viewsheds. The basic approach advocated is to integrate multiple viewer point viewsheds with probable viewsheds. The resultant most probable visible area is then selected and employed as input to carry out Higuchi and cumulative viewsheds.
This multi-technique visibility analysis has been applied in an attempt to reach robust archaeological conclusions as regards visual control of the landscape area from Iberian hillforts of central Laietania. In order to assess the archaeological relevance of the multi-technique visibility analysis proposed, a series of comparisons between different kinds of viewsheds have been presented. Altogether, visible areas of binary viewsheds show few differences from those of multiple viewer point and probable viewsheds generated from a single viewer point. In turn, the two latter cases showed remarkable similarities. When five viewer points are taken into account, there is no significant change between the visible areas of the probable and ordinary viewsheds. Essentially, the contribution made by the probable viewshed is to indicate that the edges of the viewshed should be considered less visible.
The overall similarity of the results can be attributed to the low quoted error (1.39 RMSE) of the DEM employed. Even when employing a DEM with low error, bearing in mind that relatively small divergence in visible areas can affect zones of special archaeological interest, I would strongly advocate implementing either multiple viewer point or probable viewsheds. Calculating probable viewsheds would appear relevant when the viewer location covers a discrete point. Otherwise, multiple viewer point viewsheds seem a more straightforward technique with the potential to reduce errors and changes in the results of binary viewsheds that may stem from varying the location of the viewer point. Given a low DEM error, it would not seem essential to combine multiple viewer point with probable viewsheds. However, by doing so, a greater degree of accuracy is achieved.
The most probable visible area indicated by these combined visibility analyses has been used to carry out Higuchi and cumulative viewsheds. These have been crucial for exploring the degree of visual control that hillforts might have had over the landscape. Higuchi viewsheds have revealed areas that, despite being seen by a number of settlements, could have been effectively visually controlled only by those nearby, since others were too distant to distinguish events taking place there with clarity (e.g. an army approaching). In turn, a cumulative viewshed has indicated which zones might have been watched over exclusively by one hillfort and which by a number of them.
Through this multi-technique visibility analysis it can be suggested that just before the arrival of the Romans, the hillforts of the central Laietania area formed a highly structured community as regards visual surveillance. While different hillforts would have first and foremost exerted visual control over discrete zones of the landscape, overall visual control of the area would have been carried out collectively through a visual communication network. This network in turn appears to have been based upon two visibility sub-networks, one between the settlements on the mountains and the other between those on the coast, the two communicating exclusively through Les Maleses hillfort. Thus, taking into account the close chronology for the urbanisation of the hillforts, their location could reflect a conscious design largely guided according to functional and defensive criteria.
In a subsequent stage of the research, the protocol proposed here could be employed to repeat the viewsheds using more accurate data and to generate additional analysis with the aim of further refining archaeological conclusions.
Consideration should be given to repeating the multi-technique visibility analysis using a 10 x 10m resolution DEM, which can be created out of digital cartography at 1:5000 scale of the ICC. This could overcome inaccuracies generated by a too large cell resolution. The RMSE for the study area could be more thoroughly tested in that DEM by comparing spot-heights included in the digital cartography with their equivalent locations in the DEM. It would also be relevant to compare the viewsheds obtained with additional analysis including vegetation. The latter could be performed not in order to consider lack of visibility caused by vegetation as evidence but as a possibility (Zamora 2006, 47). That is to say, a lower degree of certainty could be given to those currently visible areas potentially obstructed by vegetation.
© Internet Archaeology/Author(s) URL: http://intarch.ac.uk/journal/issue23/4/5.html
Last updated: Tues Mar 04 2008