The Geophysical Survey of an Iron Age City in Central Anatolia: Kerkenes Dag 
519 
4403000 
4402800 
4402600 
4402400 
676600 676800 677000 677200 677400 
Figure 8: Density map of sub-surface features prepared with 50m grid spacing. 
Two rose diagrams are produced, namely, frequency-weighed and length-weighed. The frequency-weighed diagram was based on the 
frequencies that were calculated in Microsoft Excel, which is at 10 degrees intervals. The linear-weighed rose diagram is based on 
length, which is produced by dividing the length of the features into lm intervals. In the frequency-weighed rose diagram, two main 
directions are observed at NE-SW and NW-SE directions with densities of 12 % and 12.5 %, respectively (Fig. 5b left). These 
directions are almost similar to the directions observed in the length-weighted rose diagram (Fig. 5b right). 
The lengths of digitised lines are calculated using coordinates of beginning and ending points (Fig. 7). The density map of subsurface 
features is produced in Surfer 6.0 with 50m grid spacing (Fig. 8). The features are mainly concentrated on the central northern and 
western parts of the study area. Those areas are very close to water resources. 
From the raw data, it is known that the sub-surface features have a box-shape pattern. However, it is not clear, whether these boxes 
are squares or rectangles. Spatial distributions of sub-surface features for two different length ranges are shown (Figs. 9a and 9b). 
The patterns in these figures suggest that the features do not form squares. The lengths in one direction are greater than those in the 
other direction. 
Length interval: 7 to 9.99 m 
Length interval: 10 to 12.99 m ■<■' ' 
, ' J ', 
' 
'An" V À 
\\ X 
N 
N 
_ ^ ' 
' - 
1 ' * V S ' N V 
f S' \ o/ ' •* 
% " " \ < \ *■ 
s'- 1 L 
" y \ N 1 - v s 
' " \ " x ' ' ' * ♦ ' % 
/ Vt " S A 
" x r 
"s' V - ' ' ° 
. s f 
' ' ' 
Figures 9a and 9b: An example for the spatial distribution of sub-surface features for two length ranges. 
To derive “decision rules” that present the relationship between the sub-surface features and the landscape, difference histograms 
were produced for elevation, slope and aspect (Fig. 10). Those histograms were produced by subtracting corresponding histograms of 
sub-surface features and landscape. This resulted in three “decision rules” in relation with elevation, slope and aspect. For the first 
two, the difference presents sharp frequencies, from which the properties of the preferred areas, in terms of elevation and slope, may 
be determined. Considering the elevation histogram, the preferred areas for sub-surface features have an elevation between 1290-