Internet Archaeol. 16. Rajala. The characteristics of GIS data

3.0 Ontology: What archaeological entities can be studied with GIS?

3.1 The characteristics of GIS data

The most extensive discussions regarding the nature of archaeological GIS data are within the volume Interpreting space (Allen et al. 1990). This collection of articles characterises the way archaeological GIS data has been traditionally viewed – first and foremost as geographically referenced data (cf. Savage 1990, 23; Wheatley and Gillings 2002, 23-87).

In his article, Savage (1990) defined the three main research areas of archaeological GIS. The first two, site location studies and GIS procedures, were directly related to the definition of data. On the one hand, site location studies defined archaeological data as find spots. Although GIS restricts the presentation of features, points were seen ultimately archaeological (Zubrow 1990a, 68). The developments in survey methodology (e.g. Thomas 1975) and the wider use of raster model (cf. van Leusen 1993) have moved the research away from the point model, but on the level of applications it is still prevalent (e.g. Richards 1996; Wheatley and Gillings 2002). On the other hand, studies related to GIS procedures have revealed the inconsistency of digitised data. Kvamme (1989; 1990) introduced the effect different algorithms have on the reliability of data and on the validity of results. Much of the discussion ended up concentrating on the quality of digitised maps and elevation data; not specifically archaeological data, but something archaeologists shared with other GIS practitioners. However, the awareness of this problem has raised the standards (cf. Gillings et al. 1999a).

Archaeological GIS studies use two types of digital data: archaeological and non-archaeological. The ways archaeological archives are transformed into databases are discussed in detail by those people who design databases (e.g. Arrayo-Bishop and Lantada Zarzosa 1992; 1995). These can be relatively technical (e.g. Ryan 2004) and do not necessarily discuss the transformation from data to information (although see Lohse et al. in press). Nevertheless, the types of data and information that can be created have been discussed (cf. Warren 1990a, 90). Warren defined data as empirical observations measured using different scales of measurement. Warren approached his subject from a statistical point of view and discussed the effect of scales of measurement on the usability of data. Although the actual attributes can be measured on interval or ratio scales, Warren concluded that the power of predictions – he was discussing predictive modelling – should be limited to the ordinal scale. This confirms that the GIS data, if not ultimately qualitative, can mostly be used only in a qualitative environment.