We managed the final ten squares on the last day completing the area I hoped we would cover. We managed 2.6ha in seven days of survey. The final four grid squares had an old excavation trench in the middle of them and a sea of mud from the spoilheap being bulldozed back into the hole.
I have left the grid on the next image (Fig. 2) which gives you some sense of scale. The grid squares are 20x20m, and site north is to the right.
In the next image (Fig. 3) I have labelled a few of the features. One can easily see many of the buildings. Because of the large differences between areas of low readings and areas of high readings I have applied a log-transform to the data to make some of the features more visible.
An alternative (and very common) method of evening-out the differences is to apply a high-pass filter (Fig. 4). The downside to this is that areas of very high readings will often end-up with a area of very low readings next to it which are purely an artefact of the filter. It is essential, therefore, to have several images and compare them so that you do not create features where there are none.
As well as the main survey with a 0.5m probe spacing (thus ‘looking’ 0.5 to 0.7m below the surface, approximately), we were simultaneously collecting 1.0m probe spacing data (which ‘looks’ about 1.0m into the ground), but at only two readings per square meter rather than four. This extra data did not add that much more to the survey (Fig. 5), but does make the north-south road show more clearly.
There is quite a bit to do in terms of interpretation and drawing-up maps of the results, but I am very happy with the new results which have greatly improved an already excellent set of results from 2002/3.
The journey home seemed to go on for ever, but at least I had some entertainingly-named beer on the way!
Many thanks to the whole team, and especially to Stefan, Wyatt, Doru and Ian.