Automated versus Hybrid Street Network Modelling for Centrality and Accessibility Analysis

Manual techniques in creating or editing street network modelling face challenges in terms of time demands, resources, and reproducibility. Automated open-source modelling workflows could address these challenges. This paper assesses an automated workflow by comparing it to a hybrid (manually edited, utilising OSM data) model using centrality and accessibility metrics. Using the street network of Nicosia, Cyprus, as a case study, we analyse the two resulting networks (automated vs. hybrid) through angular centrality and accessibility metrics across multiple radii. The models’ outputs are statistically compared via Spearman rank correlations and Bland–Altman agreement analysis, respectively, to assess agreement and spatial pattern differences. Results indicate a strong positive correlation between automated and hybrid models overall (minimum ρ ≈ 0.70), with higher agreement for closeness at larger scales. Betweenness agreement is lower at small scales but improves when segment length is considered. Accessibility metrics also show close agreement (∼95% of values within a similar range), though discrepancies increase with larger catchment distances. These findings demonstrate that automated open-source workflows can efficiently produce street network models that can reveal citywide accessibility patterns, while local divergences highlight the need for preprocessing refinements to align with the manual audit. While the study employed a single workflow (Cityseer) and focused on a single city, the findings provide generalizable insights into the validity and applicability of automated street network models in urban design and planning.