Source data for "Experienced walkability and the paradox of urban density in Chinese cities"

Overview This record provides the shareable derived data supporting the study “Experienced walkability and the paradox of urban density in Chinese cities.” It covers 82 Chinese metropolitan areas and includes street and neighborhood walkability indices, aggregate pedestrian experience results, urban form classifications and sensitivity results, and policy simulation summaries for four representative cities. The release contains 188 files: 164 GeoJSON files, 20 CSV files, and four Jupyter notebooks. The notebooks include plotting code and saved outputs from the authors’ executed analysis, allowing the code and corresponding figure results to be inspected together. Data structure 1. Street level walkability Directory: street-level walkability/ Item Description Files {city_name}_streets.json for 82 cities Format GeoJSON FeatureCollection, CRS84 longitude/latitude Spatial unit Street segment (LineString) Coverage 2,266,484 street segments Attributes id, SWI, DWI SWI is the final Static Walkability Index. DWI is the daily averaged Dynamic Walkability Index derived from time varying vehicular exposure. Higher values indicate more favorable conditions. Both indices are expressed as within city relative positions rather than physical units. 2. Neighborhood level walkability Directory: neighborhood-level walkability/ Item Description Files {city_name}_neighborhoods.json for 82 cities Format GeoJSON FeatureCollection, CRS84 longitude/latitude Spatial unit Fixed 0.005° × 0.005° grid cell (Polygon) Coverage 159,494 grid cells Attributes id, pop_norm, dist_to_c, SWI, DWI The neighborhood SWI and daily averaged DWI values are calculated by length weighted aggregation of the street segments within each grid cell. pop_norm is the normalized population value used in the urban form analysis, and dist_to_c is the distance to center variable supplied for that analysis. These fixed grid neighborhoods are distinct from the resident specific 15-minute catchments used in the individual and policy analyses. 3. Individual level experience Directory: individual-level experience/ This directory contains six CSV files summarizing results across all 82 cities. Files Contents Neighborhood_SWI v.s PE_SWI (82 cities).csv City specific fitted relationships between neighborhood SWI and experienced static walkability Neighborhood_SWI v.s PE_DWI (82 cities).csv City specific fitted relationships between neighborhood SWI and experienced dynamic walkability Neighborhood_DWI v.s PE_SWI (82 cities).csv City specific fitted relationships between neighborhood DWI and experienced static walkability Neighborhood_DWI v.s PE_DWI (82 cities).csv City specific fitted relationships between neighborhood DWI and experienced dynamic walkability Built environment correlates with PE_SWI (82 cities).csv City specific correlations between built environment attributes and PE_SWI Built environment correlates with PE_DWI (82 cities).csv City specific correlations between built environment attributes and PE_DWI The four fitted relationship files contain 101 index positions from 0 to 1. Each city column reports the fitted estimate followed by its 95% confidence interval in the form estimate (lower, upper). The built environment files contain one row per city and cover amenity categories, street density, green and industrial land, land use mix, and building footprints. 4. Policy simulation summaries Directory: model simulation/ The directory contains 12 CSV files: three modeled scenarios for Beijing, Shijiazhuang, Suzhou, and Dongguan. Scenario File pattern Amenity localization Amenity localization ({city_name}).csv Built environment improvement Built environment improvements ({city_name}).csv Traffic calming index perturbation Traffic calming interventions ({city_name}).csv Each file contains 11 scenario intensities from 0 to 1 in increments of 0.1 and reports percentage changes in PE_SWI, PE_DWI, and PE_Composite relative to baseline for dense and sparse neighborhoods. The first column retains the file schema name delta_SWI; depending on the scenario, it represents the modeled intensity alpha_POI, alpha_SWI, or alpha_DWI. The SWI and DWI scenarios perturb normalized indices and should not be interpreted as percentages of physical infrastructure improvement or traffic reduction. In the amenity localization scenario, intensity denotes the proportion of eligible home based extra neighborhood trips redirected to local destinations. 5. City summaries and urban form File Contents City_Morphology_Correlation.csv City names and coordinates, resident population, study area, city level walkability correlations, pedestrian experience summaries, and the baseline urban form classification sensitivity_city_classification_wide.csv Baseline and alternative urban form classifications under eight parameter sensitivity settings Urban form is classified into four types: Radial Monocentric, Concentric Monocentric, Clustered Polycentric, and Dispersed Polycentric. 6. Figure notebooks Files: Figure 1.ipynb through Figure 4.ipynb The notebooks provide the analysis and plotting workflows for the corresponding main text figures. They contain executable code and saved outputs from the authors’ validation run, so the displayed results can be inspected alongside the code that generated them. Some original workstation paths are retained as execution provenance. Key definitions Term Definition SWI Static Walkability Index, representing relative built environment quality for walking. DWI Dynamic Walkability Index, representing relative favorability of vehicular exposure conditions. The released street and neighborhood values are daily averages. PE_SWI Time weighted static walkability encountered along inferred walking routes. PE_DWI Time weighted dynamic walkability encountered along inferred walking routes, including time varying street exposure and modeled intersection dwell where applicable. PE_Composite Modeled combination of PE_SWI and PE_DWI used in the route choice and policy analyses. Neighborhood index Walkability associated with a resident’s home based, city specific, circuity adjusted 15-minute catchment in the individual and policy analyses. This differs from the fixed grid cells used in the urban form analysis. “Experienced” refers to objectively measurable conditions encountered along inferred walking routes, not subjective perceptions of safety, comfort, stress, or satisfaction. Reproducibility boundary This record contains shareable derived data rather than the restricted production inputs. It supports inspection of the reported results and regeneration of figures where the released fields suffice. It does not include raw mobile phone records, proprietary heavy truck trajectories, the complete five-minute segment-time exposure fields, all reconstructed individual routes, or the full individual simulation inputs. Some original cartographic boundary layers are also not included. These exclusions reflect privacy protections, provider agreements, and data licensing restrictions. The released daily DWI values should not be interpreted as the complete five-minute exposure data used in the production analysis. Citation When using these data, please cite both this Zenodo record and the associated article.