The influence of spatial configuration on e-scooter traffic flows

• First spatial configuration-based model of e-scooter traffic flows. • High explanatory power of spatial configuration across all three study sites. • First validation of the theory of natural movement using city-scale GPS tracking datasets. • Models are transferable with high predictive power at new locations. The theory of natural movement posits that the configuration of space is the primary determinant of movement patterns. Existing research in this domain, however, often relies on manual counting, fails to capture temporal variations in movement, and lacks evidence of transferability across locations. This study addresses these limitations by leveraging passively generated GPS data produced by the globally growing mode of e-scooters. We model e-scooter traffic flows using GPS data of ∼80,000 trips over three months in 2021 in three areas: Salford (United Kingdom), Rochdale (United Kingdom), and Mannheim/Ludwigshafen (Germany). Spatio-temporally cross-validated spatial lag models demonstrate the high predictive power of spatial configuration-based parameters, while additional attractor parameters contribute minimally to model performance. Our findings provide actionable insights for infrastructure planning and policy interventions, particularly for the effective planning, operation, and management of e-scooter schemes. Moreover, the results highlight that spatial configuration-based modelling methodology can be augmented with a temporal dimension, with broad potential application in transport planning.