Expanding the Role of Computer Graphics in the Analysis of Survey Data

The Historic Landscape Project (HLP) is an innovative venture being perfonned by archaeologists in Hampshire County Council's (HCC) Planning Department. The principal aims of the project are twofold: first to obtain a representative inventory of the archaeological resources occurring on County Council-owned land, and second, to provide a description of both the negative and positive effects of current land-use practices on these resources. The basic units of infonnation are derived from artefacts gathered along systematicaUy spaced transects using a variety of coUection techniques, including quadrat provenance and shovel-testing metiiods. As tiie area of study is extensive and resources are scarce, a sophisticated sampling strategy has had to be developed. The sampling design and tiie coordination of tiie field survey are the woric of BiU Boismier and are described elsewhere (Boismier 1986, Boismier & ReiUy 1986). Activity areas are defined witii tiie computer on the basis of mapped density characteristics as discrete high-density areas of artefacts. At tiie outset of tiie project, SYMAP contour maps were produced on tiie University of Soutiiampton's mainframe to help evaluate the sampling design. These indicated tiiat Uie collection strategies employed in tiie field survey, and the subsequent mapping of artefact densities per unit area, accurately locate high density areas under a number of varying surface conditions. Experimentation convinced Boismier tiiat by changing tiie SYMAP contour-intervals it was possible to detect evidence of post-depositional patterning in artefact distributions caused, for instance, by cultivation and down-slope movement (Boismier 1986). The Historic Landscape Project had been active for some considerable time when contact was made witii IBM's UK Scientific Centre (UKSQ. After some discussion it was agreed tiiat tiie analysis of tiiese disttibutions might benefit from tiie application of more sophisticated graphical display techniques. There were two aspects to this. Firstiy, at a mundane level, uie whole process could be made much more efficient if certain key phases could be automated. For example, tiie procedure for generating the final distribution maps for tiie HCC planners was labour intensive and relatively slow. The density plots showing activity areas were produced using SYMAP, but in order to produce distribution maps these data were transferred by hand on to a base map by a draftsperson. The base maps are not held on machine and tiierefore