Project
Urban Green Space Connectivity
Spatial analysis of urban green space accessibility, ecological quality, and connectivity across Delft and Yuexiu (Guangzhou) using a reproducible R pipeline and S-MCDA framework.
Overview
Built a fully reproducible spatial analytics pipeline in R to assess urban green space quality across two contrasting cities — Delft (Netherlands) and Yuexiu District, Guangzhou (China). The framework integrates accessibility, ecological quality, spatial justice, and connectivity into a Spatial Multi-Criteria Decision Analysis (S-MCDA) to identify priority areas for nature-based interventions.
Objective
To evaluate urban green space quality through four complementary dimensions — accessibility, ecological quality, spatial justice, and connectivity — and integrate them into a spatial decision-support framework that identifies where nature-based solutions are most urgently needed.
Method
01
Accessibility assessed via green space per capita, 300 m and 500 m Euclidean buffer coverage, mean nearest-green distance, and walk-network access point intersection
02
Ecological quality evaluated through OSM green space typology classification, NDVI zonal statistics from Sentinel-2, GBIF species observation density normalised by patch area, and blue-green balance index
03
Spatial justice measured using Lorenz curve and Gini coefficient of green space per capita, bivariate choropleth of population vs. green density, and socioeconomic correlation (CBS income for Delft; VIIRS night-light as proxy for Yuexiu)
04
Connectivity quantified via fragmentation metrics (NP, MPS, ENN), BFS graph connectivity on green patch centroids, betweenness centrality, and isolated patch identification
05
S-MCDA aggregated all four dimensions into a composite urgency score (accessibility 30%, ecological quality 25%, connectivity 25%, spatial justice 20%) to produce priority tier maps and proposed green corridor overlays
06
Full pipeline orchestrated in R with all paths centralised in 00_config.R; report rendered with Quarto and published via GitHub Pages
Results
Analysis
- Delft has substantially higher green space per capita (up to 400 m²/person) vs Yuexiu (up to 10 m²/person), largely explained by a 5× population density difference rather than absence of green space
- Delft's Gini coefficient (0.605) indicates more equitable distribution than Yuexiu (0.719), though neither city shows significant income–green space correlation
- Delft operates as a dense network of many small, closely spaced patches; Yuexiu relies on fewer but substantially larger parks that remain spatially isolated from one another
- Connectivity poses the highest urgency in both cities under the MCDA, with historical urban cores scoring highest for intervention priority in both Delft and Yuexiu
- GBIF observations show inverse density–area relationships in both cities, likely reflecting sampling artefacts rather than true biodiversity differences
Reflection
- OSM completeness is lower in Chinese cities than in Western Europe, introducing potential bias in typology and connectivity metrics for Yuexiu
- MCDA weights were assigned by judgement rather than derived analytically — sensitivity analysis across weight combinations would improve robustness
- The priority tier maps represent relative ranking within each city, not absolute urgency; they identify least well-performing areas rather than genuine crisis zones
- Future work should incorporate urban cooling and biodiversity effects of blue-green infrastructure, particularly relevant given Delft's canal-rich character and Yuexiu's position within the Pearl River Delta