Treffer: 3D walking accessibility in practice: exploring the imperfections from data, method, and assumptions of human-space interaction.

Current accessibility measures predominantly focus on motorized transport and 2D space. Limited studies have explored the operationalization and errors of 3D accessibility. This study contributes to the ontology of imperfection in modeling morphology-sensitive walking accessibility. We computed the accessibility to healthy food, considering different combinations of node-snapping processes (2D vs. 3D), transport infrastructures, cost functions (slope-unaware vs. least-time vs. least-effort), and trip orders (outbound first vs. inbound first). Findings showed that conventional 2D routing underestimates travel time by at least ∼1.5 minutes and ∼2 minutes for 50% of the population compared to the least-time and least-effort 3D routings. Besides, the process of connecting the location (2D vs. 3D) to the network can generate unrealistic trips to unrealistic places that can over or underestimate travel costs. Last, with least-effort routing, the difference between inbound and outbound round trips is greater than one minute for at least 8.37% of the population, reflecting the deficiency of using outbound trips or their round trip to accurately assess the access situation. Our findings identified spatially varied errors modulated by different combinations and interplay of data, methods, and the assumption of human-space interaction, highlighting the importance of a context-and-place-specific methodology to avoid the one-size-fits-all analytical framework. HIGHLIGHTS: The ontology of the imperfection in modeling walking accessibility was examined. The process of snapping the location to the network can generate errors. Large discrepancies in travel costs exist between modeling inbound first and outbound first. Spatiality leads to errors in accessibility measurement. [ABSTRACT FROM AUTHOR]