The terrain model employed uses data from the Shuttle Radar Topography Mission (SRTM). This is radar interferometry data, with a 30m cell size (USGS 2003). As stated previously, the underlying geology of Orkney is tilted up to the north-west, producing islands with cliffs on their north-western edges. These cliffs interfered with the radar beam, creating spurious signals and elevations and altering the shape of the coastline anywhere a cliff was present. An attempt to correct this by clipping the elevation data with the known island coastlines was successful, but also removed the sharp breaks in slope the cliffs represent, so any pathways generated near the cliff edge treated the 'cliff' like a gradual slope rather than a barrier. Some way of modifying the initial slope-based friction surface was therefore required so that the effort of scaling a cliff versus walking down onto a gently sloping sandy beach could be properly represented.
The problem of the missing cliffs was solved by reinserting a linear feature to represent the coastline, replacing the underlying slope-based values. For cliffs, a friction set to the maximum value of the topographic friction layer was assigned; in this case a value of 200. For the cut-banks a friction value of 3 was selected, and a value of 1 for the beaches. These estimates were based on personal observations over two seasons spent climbing up and down these features in the process of visiting and collecting site data for the project.
The underlying cost surface for this study is in fact a topographic one, generated using the COSTDISTANCE function in Arc Workstation 8.3/9.0, making use of the software's built-in capacity to generate anisotropic surfaces. It also makes use of Minetti's non-symmetrical energy expenditure curve, as cited by Llobera and Wheatley and Gillings, among others (Llobera 2000; Wheatley and Gillings 2002). This energy curve has a minimum point at roughly –5 degrees slope, and was approximated in a textfile table instead of using one of the built-in slope factor tables from Arc/Info. Because Minetti's energy curve is expressed in percentages, the given values were operationalised within Arc Workstation to fit the range of values assigned to the cliffs, beaches and rocky shores: where Minetti gave the energy cost of scaling a vertical surface as 100 per cent, here a slope of 90 degrees was assigned a vertical friction value of 200, exactly the same as that assigned to the similarly vertical cliff sections of the coastline. Likewise, in order to relate all energy costs to the cost of travelling a given distance over level ground, a slope of 0 degrees was assigned a vertical friction value of 1. In this way, crossing a 30m cell with a slope of 0 would represent an energy cost of 30 units.
It is reasonable to suggest that travel between islands was as much a fact of life during the Iron Age as it is up to the present day. This rather bland statement, however, belies the possibility of large changes in transportation technology, availability and cost over the two-and-a-half millennia separating the Iron Age from the present day. It must be noted that currently, with a transportation system geared around the use of the private automobile, travel between islands in Orkney is a very safe, but a time-consuming and expensive, proposition. While a few residents of the islands commute daily from a home on one island to work on another, these represent the exception rather than the rule in Orkney. As recently as the turn of the 20th century, with wider availability of small personal watercraft (and a lack of health and safety regulations), inter-island travel was likely much more common, albeit riskier than it currently is. The question then becomes, to what extent did water serve as a barrier to travel, or as a 'highway' for the same during the broch period? There is little archaeological evidence for boats, or related lines of evidence that could help us infer the importance of boats, such as bones from deep-sea fishing activities. To try and model the two possibilities (water = barrier vs water = highway) two cost surfaces were created, one with all water cells set to double the value of the land friction (i.e. 2). In the second instance all water cells were set to a friction of 0.5, rendering water travel half as costly as walking over level ground. Travelling 10km over the low-friction water would therefore be directly comparable in energy terms to travelling 5km over flat level ground, or 2.5km over the high-friction water. This permitted catchments to be defined that would be directly comparable, given the different frictions involved.
The result was two cost surfaces largely based around topography. However, my research also seeks to investigate the location of Iron Age broch sites in terms of their visibility in the landscape. This was approached by generating a cumulative viewshed map for a set of 5000 points randomly distributed on the landscape within the altitude range of known and potential broch sites (<90m ASL). The cumulative viewshed was generated by the VISIBILITY function of Arc Workstation, with an observer offset of +1.5m, to simulate a human observer of roughly average height. This cumulative viewshed was added as additional sources of friction by multiplying these with the topographic cost surface. The result was a total of six cost surfaces, divided between high and low frictional values for water rasters, and designed to produce paths of least-energy cost, lowest visibility (or more precisely lowest observability), and highest observability. The multiplication relationship was selected over addition in order to avoid confounding high visibility with high topographic friction, that is, multiplication ensures that a highly visible cliff section is even more undesirable under the low-visibility friction regime, while preserving its topographic frictional effect in the high-visibility friction regime.
© Internet Archaeology/Author(s)
University of York legal statements | Terms and Conditions
| File last updated: Tue Dec 20 2005