Using AI to Measure Cooling Demand from Aerial Imagery

Air conditioning systems are ubiquitous in cities such as New York – but their actual energy consumption was previously unknown. Researchers at the Karlsruhe Institute of Technology (KIT) have now teamed up with the Swiss Federal Laboratories for Materials Science and Technology (Empa) to develop a method for determining the actual cooling demand of urban areas. For their study in the journal Energy and Buildings, the researchers used an AI model that identifies cooling units in aerial imagery, counts fans, and calculates the respective cooling capacity. This allows installed systems and their energy consumption to be determined down to the building level.

Growing energy demand

“For Manhattan, we calculated an installed cooling capacity of around 10.6 gigawatts and an annual cooling demand of ten terawatt hours,” says Florian Barth from the Institute of Applied Geosciences (AGW) at KIT, who played a key role in the calculations. The highest values were found in Midtown and the Financial District, where many high-rise buildings are concentrated, as well as for hospitals and universities.

“Cooling is one of the fastest-growing energy demands worldwide – and at the same time an invisible factor,” emphasizes Dr. Kathrin Menberg from the AGW, who was also involved in the study. “With our method, we can now determine for the first time where cooling energy is actually generated and consumed in cities – an essential foundation for better integration of heating and cooling systems in the future.” Analyses like this could support cities in efficiently utilizing waste heat, planning energy storage, and reducing CO₂ emissions.

mhe, November 12, 2025