Researchers at the Technical University of Munich have released the Global Building Atlas, the first comprehensive 3D map encompassing every building on the planet, tallying an estimated 2.75 billion structures. Published this week in Earth System Science Data, the project used high-resolution satellite imagery and advanced machine learning algorithms to generate Level of Detail 1 models for the entire globe. Unlike previous datasets limited to 2D footprints or focused primarily on Western nations, the atlas includes height and volume data for structures ranging from Manhattan skyscrapers to rural dwellings in the Global South. The data is openly available, allowing users to zoom from a continental view down to individual neighborhoods.
This development fills a critical gap in earth science and urban planning: the lack of consistent, global data on the built environment. By moving from 2D footprints to 3D volumes, scientists can now model energy demand, population density, and disaster risk with greater accuracy. The inclusion of previously under-mapped regions in Africa and Asia is particularly significant for humanitarian aid and infrastructure development. For the 3D modeling community, the atlas represents a major democratization of data, providing a base layer that can be refined with local terrestrial or airborne lidar without starting from scratch.
The project also highlights the growing intersection of AI and remote sensing, with algorithms trained to distinguish rooftops from roads and vegetation across diverse biomes. The researchers note that buildings account for nearly 40% of global CO₂ emissions; understanding the volume of the world’s building stock is a prerequisite for effective climate action. The atlas introduces a new global indicator—building volume per capita—revealing deep economic disparities and establishing a baseline for tracking urbanization in the decades ahead.
Written by Adam Clark. Adam has spent the past 13 years exploring the world from above by using drones, satellites, and mapping tools to better understand our landscapes. Connect with him on LinkedIn: Adam Clark
