Modelling distribution of archaeological settlement evidence based on heterogeneous spatial and temporal data

• We examine settlement patterns in Bohemia using large archaeological datasets.
• The proposed EDE method allows for better prediction of settlement sites.
• We explore the spatial aspect of settlement density fluctuations.
• Settlement density curves can be constructed using chrono-typological evidence.
• Sampling densities over 0.05 sites per km2 produce relevant results.

We analyse variations in prehistoric agricultural settlement behaviour both in space and time to detect main turning points and shifts in settlement patterns in Bohemia, western Czech Republic. We propose a theoretical framework to address our research question and a new evidence density estimation (EDE) method combining and extending existing approaches to produce probabilistic maps and temporal frequency distribution (TFD) curves. This method takes into account heterogeneous spatial and temporal accuracy of archaeological data, and it models settlement structure where respective sites have a specific area and a given interval of duration. We determined minimal sampling densities of archaeological data enabling the method to predict prehistoric settlement at a statistically significant level. The EDE method is universally applicable for all datasets with sampling densities of more than 0.05 archaeological actions per km2 for chrono-typologically dated evidence and 0.035 actions per km2 for radiocarbon or similar dates. The results show that changes in spatial extent, density and clustering of settlement activities occur repeatedly throughout the whole agricultural prehistory and shed new light on settlement behaviour of past populations.