Spatial Justice through Urban Green Space Connectivity

Project report of Group E — Applied Spatial Analytics 2026, TU Delft.

1. Roxanne Vuijk
2. Belina Aileen
3. Daman Dogra

---

Overview

This project analyses urban green space accessibility, biodiversity, and spatial justice across two cities — Delft, Netherlands and Yuexiu, Guangzhou, China — using open spatial datasets. The analysis addresses five sub-questions (SQ1–SQ5) spanning accessibility, typology, equity, connectivity, and multi-criteria prioritisation.

---

Prerequisites

• R ≥ 4.3
• RStudio or VS Code with the Quarto extension
• Quarto CLI ≥ 1.4

Required R packages (install once):

install.packages(c(
  "sf", "terra", "exactextractr", "dplyr", "tidyr",
  "ggplot2", "patchwork", "scales", "ggrepel",
  "ggspatial", "ineq", "here"
))

---

Data

Data files are not tracked in this repo. Place them at the paths expected by R/00_config.R.

Download the data folder from: Google Drive

Expected structure after download:

data/
├── delft/
│   ├── raster/    # delft_worldpop_proj.tif, ndvi_delft_proj.tif, …
│   └── vector/    # delft_boundary_proj.gpkg, delft_osm_green_proj.gpkg, …
│   └── README.md
└── Yuexiu/
    ├── raster/    # Yuexiu_viirs_proj.tif, …
    └── vector/    # yuexiu_boundary.gpkg, guangzhou_osm_green_proj.gpkg, …
    └── README.md

All paths are centralised in R/00_config.R — edit that file if your data lives elsewhere.

---

How to run

1. Clone and open

git clone https://github.com/Applied-Spatial-Analytics/create-your-report-groupe.git

Open asa2025-report.Rproj in RStudio, or open the folder in VS Code.

2. Run the full pipeline

source("run_pipeline.R")

This sources all scripts in order (R/01_load_data.R → R/07_label_reference_mcda_map.R). Intermediate .rds objects and all fig_*.png figures are saved automatically to outputs/ (path controlled by OUT_ROOT in R/00_config.R).

3. Render the report

In RStudio: open report.qmd and click Render. In VS Code / terminal:

quarto render

Output goes to docs/index.html (set via output-dir: docs in _quarto.yml). This is the folder GitHub Pages serves the live report from — no manual copying needed.

---

File structure

create-your-report-groupe/
│
├── data/                               # not tracked — see Data section above
│   ├── delft/
│   │   ├── raster/                     # worldpop, ndvi, worldcover, viirs
│   │   └── vector/                     # boundary, wijken, buurten, income, OSM green/roads/water, GBIF
│   └── Yuexiu/
│       ├── raster/                     # worldpop, ndvi, worldcover, viirs
│       └── vector/                     # boundary, subdistricts, OSM green/roads/water, GBIF, VIIRS points
│
├── R/
│   ├── 00_config.R                     # All paths, CRS constants, buffer distances, MCDA weights, colour palettes
│   ├── 01_load_data.R                  # Load & validate all vector + raster layers; saves yuexiu_data.rds / delft_data.rds
│   ├── 02_accessibility.R              # SQ1 — green space per capita, 300/500 m buffers, nearest-green distance
│   ├── 03_typology_biodiversity.R      # SQ2 — OSM green typology, NDVI zonal stats, GBIF species density
│   ├── 04_spatial_justice.R            # SQ3 — Lorenz curve, Gini coefficient, bivariate choropleth, income/VIIRS correlation
│   ├── 05_connectivity.R               # SQ4 — fragmentation metrics (NP, MPS, ENN), graph connectivity, betweenness
│   ├── 06_mcda_nbs.R                   # SQ5 — MCDA scoring, NbS corridor prioritisation
│   ├── 07_label_reference_mcda_map.R   # Adds transliterated district labels to MCDA maps (Yuexiu)
│   └── 08_context.R                    # Context/overview maps for report introduction
│
├── outputs/                            # Auto-generated: .rds intermediates + fig_*.png figures
│
├── docs/                               # Rendered report served by GitHub Pages
│   ├── index.html                      # Rendered Quarto report
│   ├── outputs/                        # Figure PNGs referenced by the HTML report
│   └── report_files/libs/              # JS/CSS assets (Bootstrap, Quarto HTML)
│
├── pages/                              # Split pages for future multi-page website version
│
├── run_pipeline.R                      # Sources all R/ scripts in order
├── _quarto.yml                         # Quarto project config (output-dir: docs)
├── report.qmd                          # Main report narrative — pulls figures from outputs/
├── report.qmd.bak                      # Backup of original single-page report
├── asa2025-report.Rproj                # RStudio project file
├── references.bib                      # BibTeX bibliography
├── methodology.md                      # Extended methodology notes
└── README.md

---

Outputs

All figures are written to outputs/ by the pipeline scripts and copied to docs/outputs/ on render:

Figure	Script	Content
fig_context_gz_dl.png	08	Study site context map — Guangzhou and Delft
fig_context_yx_dl.png	08	Study site context map — Yuexiu and Delft
fig_ugs_yx_dl.png	08	Urban green space overview map
fig_population_density.png	02	Population density by subdistrict/wijk
fig_access_per_capita.png	02	Green space m² per capita by subdistrict/wijk
fig_buffer_coverage.png	02	Population within 300/500 m of green space
fig_nearest_green_distance.png	02	Distance to nearest green space
fig_network_walk_access.png	02	Walk-network accessibility map
fig_green_typology.png	03	OSM green space typology breakdown
fig_ndvi_zonal.png	03	NDVI zonal statistics by admin unit
fig_ndvi_violin.png	03	NDVI distribution violin plots by typology
fig_gbif_density.png	03	Biodiversity (GBIF) observation density per hectare
fig_blue_green_ratio.png	03	Blue vs. green space balance index
fig_lorenz_gini.png	04	Lorenz curve and Gini coefficient
fig_bivariate_choropleth.png	04	Bivariate map: green density × population density
fig_equity_correlations.png	04	Scatter plots: green access vs. socioeconomic proxy
fig_enn_distribution.png	05	Euclidean nearest-neighbour distance distribution
fig_fragmentation_metrics.png	05	NP, MPS, ENN fragmentation metrics
fig_connectivity_maps.png	05	Green space connectivity graph maps
fig_betweenness_vs_area.png	05	Patch betweenness centrality vs. area
fig_mcda_maps.png	06	MCDA composite urgency scores by subdistrict/wijk
fig_mcda_radar.png	06	City-mean MCDA sub-scores by criterion
fig_priority_tiers.png	06	NbS intervention priority tiers (High/Medium/Low)
fig_nbs_corridors.png	06	Proposed green corridors overlaid on priority zones
fig_mcda_distribution.png	06	Distribution of MCDA composite scores

---

Reproducibility notes

• All file paths live in R/00_config.R — no hardcoded paths elsewhere.
• Uses only open, globally available datasets (OSM, WorldPop, GBIF, VIIRS, NDVI / WorldCover).
• The workflow is adaptable to any city/district by updating R/00_config.R.
• The report is published via GitHub Pages from the docs/ folder on main. Every render-and-push automatically updates the live version.
• Population floor artifact filter (pop_count < 100) is applied before all per-capita calculations to exclude WorldPop placeholder values on non-residential zones.