What is building footprint data?
A building footprint is a polygon that outlines a building's shape as seen from above, usually extracted from satellite imagery using machine learning. It allows analysis of individual buildings, which is valuable in countries like Pakistan where detailed building surveys are often unavailable.
Data source
I used Microsoft's open building footprint dataset, which at the time of my study covered about 1.2 billion buildings detected from Bing Maps imagery (2014 to 2023). It is free to use under the Open Database License (ODbL). I downloaded the Pakistan tiles as GeoJSON using a Jupyter Notebook and combined them with the SRTM 30 m DEM, WorldPop 2020 population data, and administrative boundaries from HDX.
Case study: rooftop solar potential in Khyber Pakhtunkhwa
My BS thesis estimated solar potential for every building in nine tourism districts of Khyber Pakhtunkhwa. I calculated solar radiation from the DEM, added 5 m under footprints to approximate building height, and assigned radiation to each building using an area-weighted fishnet approach. Of 1,501,035 buildings, 570,634 were eligible for rooftop solar, and the results were shared through an interactive dashboard.
Other uses
The same data can support flood and GLOF exposure analysis, evacuation planning, urban growth monitoring, access to services, and post-disaster damage assessment.
Limitations
Connected buildings are sometimes merged into one polygon, footprints lack height information, and imagery dates vary, so recent construction may be missing. Cross-checking with OpenStreetMap or Google Open Buildings helps.