GlobalBuildingMap -- Unveiling the Mystery of Global Buildings

Abstract

Understanding how buildings are distributed globally is crucial to revealing the human footprint on our home planet. This built environment affects local climate, land surface albedo, resource distribution, and many other key factors that influence well-being and human health. Despite this, quantitative and comprehensive data on the distribution and properties of buildings worldwide is lacking. To this end, by using a big data analytics approach and nearly 800,000 satellite images, we generated the highest resolution and highest accuracy building map ever created: the GlobalBuildingMap (GBM). A joint analysis of building maps and solar potentials indicates that rooftop solar energy can supply the global energy consumption need at a reasonable cost. Specifically, if solar panels were placed on the roofs of all buildings, they could supply 1.1-3.3 times -- depending on the efficiency of the solar device -- the global energy consumption in 2020, which is the year with the highest consumption on record. We also identified a clear geospatial correlation between building areas and key socioeconomic variables, which indicates our global building map can serve as an important input to modeling global socioeconomic needs and drivers.

Figures (8)

• Figure 1: The big data analytics approach used to generate building footprints at a global scale. The workflow can be briefly summarized into five steps: data acquisition, analysis-ready data preparation, machine learning, inferencing, and post-processing. We use the Global Urban Footprint (GUF) to detect built-up areas on a 0.2° $\times$ 0.2° grid. Planet APIs were employed to acquire either Surface Reflectance or Basemap images based on the cloud coverage. Those products were calibrated and mosaicked for each 0.2° $\times$ 0.2° urban cell. The inferencing step takes those mosaics and four pretrained models as input, and produces the raw global building maps. Post-processing includes filtering false positives in non-urban areas with land cover layers and visualization.
• Figure 2: A glance at the GlobalBuildingMap. The black area in the upper left subfigure shows the coverage of the GBM. Two areas, (A) Cairo, Egypt and (B) Lagos, Nigeria, are exemplified by showing their building footprint density as a percentage in a 250 m $\times$ 250 m patch, and their detailed individual building footprints at the city block level. It demonstrates that GBM not only shows the clustering of human settlements, but also clearly indicates individual buildings.
• Figure 3: Global building density from three sources: our GBM (green), Google (red), and OSM (blue). Pure RGB color indicates only one source is available. Blended colors, including cyan, yellow, magenta, and white, indicate multiple sources are available. It can be seen that our footprint is the only source covering the entire globe. Google is only available in southern regions, and has a slight advantage in a few spots in eastern Africa and India, as red dominates those regions. OSM aligns well with our building footprint in Europe, as well as some Asian and North American cities, but is not available elsewhere. Only GBM covers some of the most populous regions in the world in East Asia.
• Figure 4: Among various data sources, our GBM shows globally consistent building maps on a building instance level. From the top to bottom are results (in white) obtained by OSM, Google, Microsoft, GHSL, WSF, and GBM.
• Figure 5: Rooftop solar potential analysis of global buildings. Color indicates the yearly solar potential per pixel in the range of 0--10 GWh/year with a spatial resolution of 250 m $\times$ 250 m. Two cities that are ideal for placing rooftop solar panels, A. Cairo, Egypt and B. Delhi, India, are zoomed in.