Developing an adaptive zoning system for city-scale activity-based travel demand modeling using OpenStreetMap data

Abstract

City-scale activity-based demand models (ABMs) offer detailed insights into travel behaviors; however, their accuracy is often limited by the coarse spatial zoning used in many zone-based demand models. Conventional Traffic Analysis Zones (TAZs) aggregate data at a level that masks neighborhood-specific variations and misallocates short and non-motorized trips. To enable more realistic urban mobility analysis, this paper introduces a hybrid zoning system that adapts spatial resolution to local activity density. Using Hasselt, Belgium, as a case study, we refine official statistical sectors into high-resolution miniZones. This is achieved by applying constrained k-means clustering to OpenStreetMap building data, followed by Voronoi tessellation. The resulting clusters are transformed into contiguous zones through dissolving Voronoi tessellation. This technique is applied to provide fine-grained detail within the city's scope, reflecting the density of buildings where human activity is high. To keep the refinements of the zoning com-putationally manageable for a city-scaled regional model, a gradual reduction in detail is incorporated as one moves away from the city. This is achieved by aggregating to coarser official units in the more distant regions. The final result is a hybrid zoning system. This adaptive model approach enhances the representation of trip generation and distribution within cities, providing support for more accurate activity-based travel demand modeling.