W = 6 × exp(−3.5 × |S + 0.05|)
Tobler’s hiking function. This function estimates walking speed as a function of slope. Here, W is walking speed in kilometres per hour, and S is slope, expressed as elevation difference divided by distance (dh/dx).
Landforms suitable for movement in the broader Lasithi plateau region (channels, saddles, and planar areas). This data will be utilised within a Point Process Modelling framework to investigate whether site location can be attributed to these flatter areas of the landscape across Crete. Study conducted by Andriana Xenaki:
In ARNAC, this part of the project focuses on the computational study of movement across the Cretan landscape. We examine how different GIS tools, algorithms, and slope-dependent functions model routes, travel time, and effort in a terrain that is large, rugged, and highly varied. A central aim of the project is to test Least Cost Path analysis in relation to the specific morphology of Crete, where steep slopes, gorges, changing coastlines, and modern alterations to the terrain all affect how ancient movement can be reconstructed. Rather than accepting existing models uncritically, ARNAC compares different methods and software packages in order to assess which ones perform best under Cretan conditions. In parallel, additional spatial approaches, including Point Process Modelling and Focal Mobility Network, will also be tested as part of the project’s broader methodological framework.
This work is not only technical but also evaluative. We compare digital outputs with archaeological evidence, with documented routes, and with travel times recorded by members of the research team during fieldwork. In this way, the project investigates whether existing functions are sufficient, whether they need to be adjusted, and whether new coefficients or modelling approaches should be proposed. ARNAC therefore aims to produce not just route predictions, but also a best-practice methodological guide for using Least Cost Path analysis in Crete and in other similarly challenging landscapes.
Comparison of multiple cost-function algorithms using a case study from the Lasithi Plateau. The results demonstrate substantial convergence in least-cost routes between Psychro and surrounding Neopalatial sites. Despite differing assumptions regarding slope, energy expenditure, and walking velocity, route predictions consistently follow a limited number of topographically constrained corridors. The study was conducted by Andriana Xenaki in the R statistical environment using the movecost package.
Comparison of the distribution of path lengths produced by different cost-function algorithms in the Lasithi Plateau case study (Tobler, Tobler off-path, Irmischer–Clarke, Alberti, Herzog, Rees, and Pandolf). The boxplots display a high degree of similarity across all models: median route lengths are nearly identical, variability remains consistent, and no algorithm produces substantially longer or shorter routes. The study was conducted by Andriana Xenaki in the R statistical environment using the movecost package.