A global urban road network self-adaptive simplification workflow from traffic to spatial representation

Urban road network is crucial for understanding and revealing the spatial logic of urban organization and evolution. However, existing urban road network datasets like OpenStreetMap are designed for traffic studies, treating each lane as a distinct spatial unit of mobility, which may not align with urban studies considering each road as an integration space for social and cultural dynamics. This study established a novel workflow to self-adaptively transform the global urban road network from traffic representation to spatial representation and provides simplified urban road network data of 35 globally representative cities. Our workflow, comprising six critical stages, is anchored on the segment divergence from their surroundings to guide aggregation decisions, effectively mitigating the risks of over-aggregation and under-aggregation against the diversity of global urban backgrounds. This workflow significantly reduces the duplicated segments of roads from an average of 31.2% to 3.6% in total, performing consistently across diverse countries and continents. This dataset is expected to become a robust data layer for urban socio-economic modelling and GeoAI development.