Sentinel satellites and Google Earth Engine produced the first building-height map covering an entire nation at 10-meter resolution — and the world's researchers can't stop downloading it.
From ground level, China's urbanization is a blur of construction cranes and expanding skylines. From orbit, it becomes data. Wanben Wu turned two European Space Agency satellites into a nationwide tape measure, processing petabytes of Sentinel-1 radar and Sentinel-2 optical imagery through Google Earth Engine to estimate the height of every building structure across mainland China at 10-meter resolution. The result — CNBH-10m — is the first dataset of its kind for any country at this scale and precision.
The numbers are staggering. More than 1.3 million researchers, planners, and analysts have downloaded the dataset since its 2023 release, making it one of the most accessed geospatial products on Zenodo. The demand speaks to a fundamental gap in urban science: satellite imagery can show where buildings are, but knowing how tall they are transforms two-dimensional land-use maps into three-dimensional models of human settlement. Building height is the missing variable in energy consumption models, urban heat island studies, population density estimates, and disaster risk assessments.
What makes CNBH-10m especially powerful is its consistency. Previous building height estimates for China relied on patchy lidar surveys or coarse statistical models that varied wildly between provinces. By processing the entire country through a single pipeline — random forest regression trained on ICESat-2 laser altimetry — Wu produced a dataset where a 15-meter building in Shenzhen is measured the same way as a 15-meter building in Urumqi. For researchers studying urbanization patterns across China's 600+ cities, that consistency is transformative.
Median estimated building height for China's largest urban areas, showing distinct vertical profiles
Percentage of mapped building footprints falling into each height category
City governments can use CNBH-10m to model population density, plan infrastructure, and enforce zoning regulations with three-dimensional precision. The dataset turns urban planning from a 2D map exercise into volumetric analysis of where people actually live and work.
Building height directly affects urban heat islands, wind corridors, and energy demand. Climate modelers now have the vertical dimension needed to produce accurate city-scale heat and energy forecasts across all of China — not just the handful of cities with lidar data.
Because CNBH-10m relies entirely on freely available Sentinel data and Google Earth Engine, the methodology can be replicated for any country on Earth. The 1.3 million downloads suggest the world is waiting for equivalent datasets beyond China's borders.
Share this story