A novel approach in archaeological predictive modelling: Case study from western Türkiye Hellenistic settlements

2023-3
Er, Melek
The archaeological predictive models (APMs) have been developed since 1980s to locate unknown archaeological sites. However, the major elements for the development of APMs are still criticized: the archaeological input data, the variables used in prediction, the statistical analysis and the testing of resulting model. This study aims to address each of these criticisms through a case study focused on the poleis of the Hellenistic period in western Türkiye. A high quality of archaeological data and relevant variables for prediction are accepted into the analysis. Elevation, ruggedness, slope, aspect, arable land, access to water and rock types used in the city walls were assessed as possible predictive variables. The study found that variables such as ruggedness, slope, aspect, and arable land were highly predictive. Additionally, the study introduced a new way of use for the road network as a socio-cultural variable, along with APMs. Unlike previous studies, this study developed a statistical method that allows for polygonal representation of archaeological settlements while deriving landscape characteristics of the site. At each step, the suggested statistical approaches were tested for their repeatability and sensitivity to the sample size. The resulting variables were selected based on their performance, repeatability, and sensitivity to proceed the unification of them as a final predictive map.
Citation Formats
M. Er, “A novel approach in archaeological predictive modelling: Case study from western Türkiye Hellenistic settlements,” Ph.D. - Doctoral Program, Middle East Technical University, 2023.