This repository contains the scripts used in the analysis for our paper: A Walk across Europe: Development of a high-resolution walkability index.

Code (Jupyter Notebook) Execution Order and Description

The recommended order of execution with brief descriptions:

1. 01_generate_grids.ipynb
   Generates 100m x 100m grids from the Eurostat 100km x 100km grid file.

2. 02_* notebooks
   Collect raw data from various sources and process walkability components for 100m x 100m grids:

   • 02_sidewalks.ipynb: using OSMnx.
   • 02_sidewalks_osmium.ipynb: Efficient sidewalk processing with osmium.
   • 02_street_intersections.ipynb: using OSMnx.
   • 02_street_intersections_osmium.ipynb: Efficient intersection calculation with osmium.
   • 02_green_spaces.ipynb: Extracts NDVI from Sentinel-2 (requires Google Earth Engine & Google Cloud).
   • 02_slope.ipynb: Extracts slope from NASA DEM (requires Google Earth Engine & Google Cloud).
   • 02_public_transport.ipynb: Counts OSM public transport features (requires Overpass API).
   • 02_public_transport_osmium.ipynb: Efficient version using osmium.
   • 02_landuse_mix.ipynb: Computes land use mix (entropy) using Copernicus LULC data (requires Google Earth Engine & Google Cloud).
   • 02_degree_of_urbanization.ipynb: Uses GHSL data to assign urbanization class (requires Google Earth Engine & Google Cloud).
   • 02_population.ipynb: Assigns population counts from GHSL (requires Google Earth Engine & Google Cloud).

3. 03_generate_isochrones.ipynb
   Uses Valhalla routing engine to generate isochrone polygons (also a component: 15-min walking area).

4. 04_isochrones_to_gpu.ipynb
   Converts isochrones to array format for GPU-friendly processing.

5. 05_process_isochrones_gpu.ipynb
   Runs the core GPU kernel to compute distance-decayed metrics inside isochrones (requires NVIDIA GPU).

6. 06_walk_index.ipynb
   Computes the final walkability index using all the processed components.

New vs. Old Scripts

Scripts with osmium in the filename are newly developed for efficient large-scale processing of OSM data. Earlier versions relied on OSMnx and Overpass API, which are inefficient for bulk processing.

Dependencies (in addition to libraries)

• Google Earth Engine: Requires an Earth Engine account and setup of the Python API.
• Google Cloud Storage: Temporary storage for Earth Engine exports. Requires a Google Cloud account and service account key.
• Valhalla: Isochrone generation via open-source Valhalla routing engine.
• NVIDIA GPU: Required for processing isochrones on GPU (step 05).

Required Downloads

Some scripts expect the following data to be pre-downloaded:

• europe-latest.osm.pbf – Raw OpenStreetMap data for Europe.
  Download from: Geofabrik

Interactive Web-Atlas (coming soon!)

Environment Setup

To create and activate the environment for reproducibility:

conda env create -f environment.yml
conda activate geo_env

Additional Info

If you'd like to use the already processed data for your projects, don't hesitate to reach out!
This work was done as part of the OBCT project.