An open-source AI model is using publicly available data to map building carbon emissions across entire cities – giving policymakers a new tool to design smarter, fairer decarbonisation strategies.
Developed by researchers at the National University of Singapore’s College of Design and Engineering (CDE), the machine learning framework can pinpoint urban carbon hotspots, highlight inequalities in emissions, and guide targeted interventions to cut the built environment’s footprint.
Unlike many previous tools that rely on proprietary datasets, the model is fully open-source and transferable across cities with varying levels of data availability. That makes it especially valuable for global climate action, where resources and urban contexts differ dramatically.
Applied to more than half a million buildings in Singapore, Melbourne, New York City, Seattle and Washington DC, the AI explained up to 78% of the variation in emissions. The results revealed that emissions aren’t just a matter of building size or density – they’re shaped by each city’s planning history, climate, and economic profile.
Key findings included:
- Suburban low-rise areas can be just as big a contributor to emissions as dense city centres.
- Tall buildings may benefit from economies of scale, but heat island effects drive up cooling demand in dense cores.
- Wealthier districts tend to have higher per capita building emissions, with a handful of large buildings in Manhattan responsible for more than half of its total emissions.
Crucial policy challenge
This raises a crucial policy challenge: blanket carbon pricing or uniform regulations risk penalising lower-income communities already living with outdated, inefficient stock. Instead, the researchers argue for place-based decarbonisation strategies that consider both emissions intensity and socioeconomic vulnerability.
The framework draws on a wide range of open datasets – including satellite imagery, street-level photos, population maps, road networks and climate data – processed through graph neural networks to capture spatial relationships between urban elements.
By going open-source, the team hopes to support cities worldwide in meeting net zero goals while making climate policies more equitable.
“This work demonstrates the potential of open science and AI to accelerate urban sustainability,” said Assistant Professor Filip Biljecki, who led the study. “It’s not just about understanding where emissions come from, but ensuring that climate action is both effective and fair.”
Image: ‘Department of Architecture PhD student and Lead Author Winston Yap (left) with Asst Prof Filip Biljecki (right) who led the research.‘
Read next: Unlocking the benefits of building with nature – Build in Digital
Are you a building professional? Sign up for a FREE MEMBERSHIP to upload news stories, post job vacancies, and connect with colleagues on our secure social feed.
