Hans Kamermans / Martijn van Leusen / Philip Verhagen

(University of Leiden / University of Groningen / RAAP Archeologisch Adviesbureau, Amsterdam, Netherlands)

The Netherlands are unique in Europe in having adopted predictive modelling as an important tool for archaeological heritage management. The Netherlands are probably the only country in the world boasting a national predictive map. Predictive maps are used both for influencing spatial planning decisions, as well as for deciding on the best archaeological research strategy in the case of planned urban developments. However, after more than fifteen years of practice in developing and applying predictive models in the Netherlands, we have to conclude that the scientific and methodological basis of predictive modelling is not yet secure. This can lead to predictions that may be completely wrong, and at best cannot be controlled for their quality. In this paper we will briefly outline the development of predictive modelling in Dutch archaeology, its relation to the archaeological heritage management process in the Netherlands, and the main problems associated with its application. Finally, and most importantly, we will try to point to way forward: in what ways can predictive modelling be improved, and are these improvements useful additions to the decision making process in spatial planning? We will focus on two aspects that have recently been studied in more detail: the inclusion of uncertainty measures in predictive models, and the potential of testing programs for improving model quality.

For more information on the subject, see the BBO Predictive Modelling project website at http://www.archeologie.leidenuniv.nl/index.php3?m=39&c=92