HUDI Lahore: Urban Quality Index Pipeline

This repository builds a grid-based Urban Quality Index for Lahore from remote sensing, OSM roads/POIs, AQI, and built-form features.

What this repo does

• Extracts and prepares core spatial layers (NDVI, LST, NL, AQI, buildings, roads, POIs).
• Builds 250m and 100m Lahore grids.
• Computes EQ, UD, and Overall index scores in notebooks/index.ipynb.
• Runs two ablation families:
  • metric drop ablations (notebooks/ablations/)
  • weight-scenario ablations (notebooks/ablations_weight_scenarios/)

Core formula

• EQ = weighted average of environmental indicators
• UD = weighted average of urban development indicators
• Overall = 0.4 * EQ + 0.6 * UD
• Visualization maps use robust scaling to [0,10] (Overall_vis).

100m base EQ weights:

• NDVI_sc 0.35, LST_sc 0.35, AQI_sc 0.20, POI_access_sc 0.10

UD weights:

• BFP_presence_sc 0.10, Highrise_decay_sc 0.30, Height_sc 0.20, BFP_density_sc 0.20, Road_density_sc 0.20

Repository structure

• notebooks/index.ipynb main integration + index + ablations
• notebooks/POI/ POI fetch + accessibility surface
• notebooks/OSM/ roads/OSM preparation
• notebooks/AQI/, notebooks/NDVI/, notebooks/LST/, notebooks/NL/ source-specific processing
• figs/ static figures for documentation/paper
• DATA_INVENTORY.md dataset-level file list, sources, and resolutions

Main outputs

• notebooks/lahore_index_250m.geojson
• notebooks/lahore_index_100m.geojson
• notebooks/lahore_index_250m.csv
• notebooks/lahore_index_100m.csv
• notebooks/lahore_index_map_final.html

Quick run order

1. Generate or verify source layers in domain notebooks (AQI, NDVI, LST, NL, POI, OSM).
2. Open notebooks/index.ipynb.
3. Run all cells top-to-bottom.
4. Check ablations:
   • notebooks/ablations/*.png
   • notebooks/ablations_weight_scenarios/*.png

Figures

Data/Method snapshots

Metric-drop ablations (100m)

Weight-scenario ablations (100m)