Reconstructing Late Iron Age Transport Network from Site Positions

International audienceThe end of the second Iron Age saw the explosion of trade leading to a boom in medium and long-distance traffic. However, the understanding of these traffic flows is hampered by the lack of knowledge of the road network in use during the last decades of Gallic independence. The objective of this communication is to propose a method for estimating realistic transport costs between settlements, in a context where the road network is unknown.Reconstructing a transport network from the sole location of archaeological sites remains a difficult problem despite the ongoing development of spatial analyses and Geographic Information Systems. The classic solution consists in approximating the transport network by the set of least cost paths (LCP) between sites based on the constraints of geographical space (relief, wetlands, etc.). The challenge lies in the difficulty of translating these geographical constraints into transport costs, which is still a very topical problem (Herzog 2013; Verhagen et al. 2019). A common way to evaluate the plausibility of the LCP is to compare them with additional data, such as the position of sites that were not included in their computation. Our work extends the state-of-the-art in three directions:1) Our method is based on the articulation of partial historical data (cost ratios between different means of transport) with multiple simulations to estimate unknown values (transhipment costs, river crossing costs, etc.). All these simulations are then compared using systematic sensitivity analyses, including the use of novel aggregated metrics that characterise a LCP collection and emphasize global effects of some parameters. 2) Few studies have attempted to take into account the multimodal nature of old transport (land and river), and the many scientific and technical problems posed by modelling changes in means of transport. In this context, this work proposes a new explicit implementation of the bimodal (terrestrial and fluvial) nature of these old transports.3) The last challenge is the application of these methods to extremely voluminous data, using a high-resolution numerical terrain model at the scale of central-eastern Gaul (260 000 km²). The numerous simulations needed to assess the sensitivity of LCP calculations to the cost parameters are associated with a very large consumption of computational resources (time and memory). A dedicated implementation is needed to complement the tools currently in use, such as GIS softwares and R packages traditionally used for LCP.In the end, this method and its ongoing refinement bring new elements to the understanding of the profound economic upheavals occurring at the dawn of antiquity.Herzog, Irmela 2013The potential and limits of optimal path analysis. In Computational Approaches to Archaeological Spaces, pp. 179–211. Left Coast Press. Walnut Creek.Verhagen, Philip, Laure Nuninger, and Mark R. Groenhuijzen 2019Modelling of Pathways and Movement Networks in Archaeology: An Overview of Current Approaches. In Finding the Limits of the Limes: Modelling Demography, Economy and Transport on the Edge of the Roman Empire, edited by Philip Verhagen, Jamie Joyce, and Mark R. Groenhuijzen, pp. 217–249. Springer International Publishing, Cham.