Premium
On the use of integrated process models to reconstruct prehistoric occupation, with examples from Sandy Flanders, Belgium
Author(s) -
Zwertvaegher Ann,
Werbrouck Ilke,
Finke Peter A.,
De Reu Jeroen,
Crombé Philippe,
Bats Machteld,
Antrop Marc,
Bourgeois Jean,
CourtPicon Mona,
De Maeyer Philippe,
De Smedt Philippe,
Sergant Joris,
Van Meirvenne Marc,
Verniers Jacques
Publication year - 2010
Publication title -
geoarchaeology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.696
H-Index - 44
eISSN - 1520-6548
pISSN - 0883-6353
DOI - 10.1002/gea.20332
Subject(s) - process (computing) , computer science , digital elevation model , scale (ratio) , calibration , data science , archaeology , data mining , geology , geography , remote sensing , cartography , statistics , mathematics , operating system
Intensive archaeological investigations in Sandy Flanders (Belgium) revealed sites dating from the Final Paleolithic to the Neolithic, showing a discontinuous spatial and temporal distribution. To improve the understanding of these occupational patterns, a paleolandscape reconstruction is proposed. A major problem in paleolandscape reconstruction is that basic data are scattered in the temporal and spatial sense. Therefore, we propose an interdisciplinary approach to the application of different process models to soil–water–landscape reconstruction. The process models used include a digital elevation model, a hydrological, a pedogenetic, and a land‐evaluation model. The result is a modeling framework in which these discipline‐specific models, which provide input to each other, are integrated. Outcomes of the different models are still preliminary, because of the ongoing calibration and application of the models. The paper focuses on the methodological aspects of constructing the modeling framework and the questions one needs to answer in advance to facilitate the integration of the model results. Furthermore, errors within each individual model need to be accounted for and ideally are propagated into the next modeling step. Because of model complexity and runtime this is presently unfeasible. Alternatively, we propose to repeat the last step of the model framework (the land‐evaluation procedure) for perturbations of the parameters reflecting the estimated model errors. We emphasize the difficulties occurring when integrating these different models, such as those relating to scale differences and error propagation. © 2010 Wiley Periodicals, Inc.