Enhancing Travel Behavior Simulation Through A High-Resolution Spatial And Population Disaggregation

Abstract

This paper presents a multi-stage framework to enhance activity-based travel behavior simulation by integrating high-resolution spatial and population dis-aggregation into the FEATHERS activity-based model. Using the Iterative Proportional Updating (IPU) algorithm, a synthetic population for Flan-ders was generated from the 2021 census and OVG5 mobility survey data, achieving close alignment across household and individual attributes. To address the limitations of traditional traffic analysis zones (TAZs), we introduce adaptiveZoning that combines fine-grained miniZones in the study area with larger zones (statSector s) in surrounding areas, thereby balancing spatial detail and computational efficiency. The synthetic population is dis-aggregated from statSector s to these miniZones through dasymetric mapping methods based on OpenStreetMap (OSM) residential buildings and official residential addresses. A sampling approach was introduced to assign labels to untagged OSM buildings. Comparative analysis shows tha OSM-derived residential building points yield the highest correlation with observed population distributions (R 2 = 0.94), while address-based disaggregation provides complementary benefits in mixed housing contexts. Furthermore, activity locations within daily travel plans are refined from TAZ centroids to street-level addresses using tagged building functions, supported by an artificial neural network-based schedule selector. This fine-grained integration significantly improves the spatial realism of population and travel demand modeling, reduces biases associated with aggregated zones, and enables a more accurate representation of multimodal and active transport.