The future of archaeological predictive modelling (original) (raw)
Related papers
H. Kamermans, M. van Leusen and P. Verhagen (eds.), Archaeological prediction and risk assessment. Alternatives to current practice. Leiden University Press, Leiden (ASLU 17), pp. 19-25., 2009
In general, academic archaeologists have always been sceptical of, and sometimes even averse to, predictive modelling as practiced in archaeological heritage management (AHM) (see Van Leusen et al. 2005). The models produced and used in AHM are not considered sophisticated enough, and many of the methodological and theoretical problems associated with predictive modelling have not been taken aboard in AHM. At the same time, the production and use of predictive models has become a standard procedure in Dutch AHM (Deeben et al. 1997; 2002; Deeben 2008), and it has clearly attracted interest in other countries as well. The main reason for using predictive models in AHM is efficiency. In ‘post-Malta’ archaeology, the financial, human and technical resources allocated to archaeology have increased enormously. But at the same time, these resources have to be spent both effectively and efficiently. So why not create and use tools that will allow us to do so? Archaeological predictive models will tell us where we have the best chances of encountering archaeology. Searching for archaeology in the high probability areas will ‘pay off’, as more archaeology will be found there than in low probability zones. It is a matter of priorities: we can’t survey everything, and we don’t want to spend money and energy on finding nothing. And there is also the political dimension: the general public wants something in return for the taxpayers’ money invested in archaeology. It’s not much use telling politicians to spend money on research that will not deliver an ‘archaeological return’. But how can we be so sure that the low probability zones are really not interesting? And where do we draw the line between interesting and not interesting? These are hard choices indeed for those involved in AHM. Archaeologists who don’t have to make these choices have an easy job: they can criticize the current approach to predictive modelling from the sidelines, and don’t have to come up with an alternative. Within the BBO program we have been trying to provide such an alternative to the archaeological community (Kamermans et al. 2005). However, at the end of the project, we have to conclude that we have only been partly successful. We have done a fair amount of research, published three books and many papers, made the problems with predictive modelling internationally visible but failed to change the procedures of predictive modelling in the Netherlands. In this paper we venture to offer some explanations for the lack of success of new approaches to predictive modelling in AHM in the Netherlands up to now. And finally, we will try to sketch the future of archaeological predictive modelling, for which we can see three distinct scenarios.
T. Bloemers, H. Kars, A. van der Valk & M. Wijnen (eds.): The Cultural Landscape & Heritage Paradox. Protection and Development of the Dutch Archaeological-Historical Landscape and its European Dimension (Landscape & Heritage Studies Proceedings), pp. 431-444., 2010
In ‘post-Malta’ archaeology, the financial, human and technical resources allocated to archaeology have increased enormously. But at the same time, these resources have to be spent both effectively and efficiently. Archaeological predictive models will tell us where we have the best chances of encountering archaeology. Searching for archaeology in the high probability areas will ‘pay off’, as more archaeology will be found there than in low probability zones. It is a matter of priorities: we cannot survey everything, and we do not want to spend money and energy on finding nothing. And there is also the political dimension: the general public wants something back for the tax-payers’ money invested in archaeology. It is not much use telling politicians to spend money on research that will not deliver an ‘archaeological return’. But how can we be so sure that the low probability zones are really not interesting? And where do we draw the line between interesting and not interesting? These are difficult choices indeed for those involved in AHM. Archaeologists who do not have to make these choices can criticize the current approach to predictive modelling from the sideline, but do not have to come up with an alternative. Within the BBO-programme we have been trying to provide such an alternative to the archaeological community (see van Leusen and Kamermans, 2005; Kamermans et al., 2009). However, after five years of research, we have to conclude that we have only been partly successful. In this paper we will shortly explain the research that we have undertaken, and venture to offer some explanations for the lack of success of new approaches to predictive modelling in AHM up to now.
P. Verhagen (2005). Prospection Strategies and Archaeological Predictive Modelling.
M. van Leusen and H. Kamermans (eds.), Predictive Modelling for Archaeological Heritage Management: A research agenda. (NAR 29). Rijksdienst voor het Oudheidkundig Bodemonderzoek, Amersfoort, pp. 109-122., 2005
A key problem in predictive modelling is the availability of representative archaeological input data that can be used either as input to an inductive predictive model, or as a test set for an independent check of the model. Almost all available archaeological data sets are biased in one way or another to specific site types or regions. Some of this bias originates as a result of the archaeological prospection techniques used for discovering sites. The aim of archaeological survey is to establish without any doubt the presence of archaeological sites. Subsidiary goals might be defined for a survey, like the determination of the exact location of the site, its type and dating, the layout of the site and even the conditions of the buried artefacts (see e.g. Hey and Lacey, 2001). For predictive modelling however, it suffices to obtain evidence of the presence or absence of an archaeological site, at a location that is as precise as possible. Dating and typology of a site are desirable properties to be known, but if they are not available, a non-specific predictive model might still be constructed. The definition of an archaeological site on the basis of survey data however is problematic in itself. Tainter (1983) provides a useful working definition, that is cited by Zeidler (1995), in which the criterion for defining an archaeological site is the presence of at least two different artefacts in close proximity, or other evidence of purposive behaviour, such as archaeological features or architectural remains. Two different objects is the minimal archaeological manifestation which will consistently reflect purposive behaviour, whereas a single object cannot differentiate accidental loss. This definition does not take into account the possibility that the artefacts may be encountered ex situ, but it serves well as a minimal standard. However, if only one or even no artefacts are found during survey, one cannot be certain that there is no site. The degree of confidence for establishing the absence of an archaeological site is highly dependent on the survey method chosen, and surveys may therefore underestimate the actual number of sites in an area by varying degrees. This has severe consequences, both for the curators who want to be as certain as possible that all sites in a region have been found during survey, as well as for predictive modellers, who depend on representative site samples to develop their models.
5. New developments in archaeological predictive modelling
Amsterdam University Press eBooks, 2011
In this paper the authors present an overview of their research on improving predictive modelling into true risk assessment tools. Predictive modelling as it is used in archaeological heritage management today is often considered to be a rather crude way of predicting the distribution of archaeological remains. This is partly because of its lack of consideration of archaeological theory but also because of a neglect of the effect of the quality of archaeological data sets on the models. Furthermore, it seems that more appropriate statistical methods are available for predictive modelling than are currently used. There is also the issue of quality control, a large number of predictive maps have been made but how do we know how good they are? The authors have experimented with two novel techniques that can include measures of uncertainty in the models and thus specify model quality in a more sophisticated way, namely Bayesian statistics and Dempster-Shafer modelling. The results of the experiments show that there is room for considerable improvement of current modelling practice but that this will come at a price because more investment is needed for model building and data analysis than is currently allowed for. It
earsel.org
Most archaeological predictive models lack significance because fuzziness of data and uncertainty in knowledge about human behaviour and natural processes are hardly ever considered. One approach to cope with such uncertainties is utilization of methods, which base on approaches of the probability theory like Bayes Theorem or Dempster-Shafer-Theory. In our case study we analyze an area of 50 km² in southern Rhineland Palatinate (Germany) near the Celtic oppidum "Hunnenring" by use of Dempster-Shafer's theory of evidence for predicting spatial probability distribution of archaeological sites. This technique incorporates uncertainty by assigning various weights of evidence to defined variables, in that way estimating the probability for supporting a specific hypothesis (in our case the hypothesis presence or absence of a site). Selection of variables for our model relies both on assumptions about settlement patterns and on statistically tested relationships between known archaeological sites and environmental factors. The modelling process is conducted in a Geographic Information System (GIS) by generating raster-based likelihood surfaces with a cell resolution of 10 m for the six selected variables 'distance to water', 'distance to road network', 'distance to graves', 'slope', 'landforms' and 'geology'. The corresponding likelihood surfaces are aggregated to a final weight of evidence surface, which results in a likelihood value for every single cell of being a site or a non-site. Finally the result is tested against a database of known archaeological sites for evaluating the gain of the model. To address the high potential of soil erosion processes of the low mountain parts of our study area a model was developed which allocates erosion and deposition zones to those areas. The combination of this model with the predictive model yields a more differentiated estimation, especially with regard to the suspected potential of archaeological remains being conserved until present.
2014
The identification of areas that are insignificant for archaeological research can be used for guidance and support in projects that involve decision-making about the use of land and modern development activities. On the other hand, the identification of areas significant for archaeological research can contribute to archaeological knowledge and mini-mise the risk of unsuccessful excavations. This paper presents a review of the most recent and representative applications of pre-dictive modelling in Archaeology, which demonstrate that predictive models can be successfully exploited by archaeological research and Cultural Heritage Management (CHM).