Later prehistoric land-use intensities: base data and tables

This data-sets comprises the base GIS layers and tables to the publication: Arnoldussen, S., Verschoof-Van der Vaart, W., Kaptijn, E., & Bourgeois, Q. P. J. (2022). Field systems and later prehistoric land use: New insights into land use detectability and palaeodemography in the Netherlands through LiDAR, automatic detection and traditional field data. Archaeological Prospection, <a href="https://doi.org/10.1002/arp.1891">doi:10.1002/arp.1891</a>.<br><br>

The paper to which these data pertain explores how AI-assisted mapping of prehistoric field systems can improve reconstructions of past land use and population in the Netherlands. Traditional palaeodemographic models have largely relied on limited “nodal” data—such as settlements or barrows—which underrepresent the true extent of past human activity. In contrast, the authors argue that Celtic fields—prehistoric agricultural field systems bordered by earthen banks—provide a more stable and extensive proxy for reconstructing land use and estimating population sizes.<br>

Using LiDAR data and automated detection algorithms based on the YOLOv4 framework, four case-study regions across different geological zones were analysed. The automated detections were then verified and refined through expert manual mapping. This “human–computer” strategy significantly increased the mapped area of Celtic fields—on average by a factor of 1.84—adding between 40 and 113 hectares of identified prehistoric fields per study area.<br>

Population estimates derived from these expanded field system areas ranged from 34 to 147 individuals per region, aligning closely with estimates based on traditional settlement data (80–137). Regional variation was largely explained by modern land use and research intensity, with forest and heathland areas offering better preservation and detectability.<br>

The study concludes that Celtic field mapping provides a robust and spatially representative method for palaeodemographic reconstruction, especially in areas where settlement data is scarce. AI-assisted LiDAR detection is shown to be fast and reliable, and when combined with expert interpretation, it offers a scalable approach for reconstructing later prehistoric land use. The authors advocate a complementary use of field system and settlement data, with field system coverage offering a stronger basis for cross-regional comparison and demographic modelling.