Mapping informal settlements in the Global South at 10-meter resolution using multi-source satellite imagery and open spatial datasets

Over one billion people currently reside in informal settlements—those spontaneous and self-identified areas characterized by limited fundamental services with high vulnerability to climate-driven hazards. While mapping these areas is critical for achieving the Sustainable Development Goals, large-scale and high-resolution informal settlement datasets detailing their locations and boundaries are still highly needed to support in-depth quantitative analysis. To address this gap, we develop an efficient and scalable framework that integrates multi-source satellite imagery, including Sentinel-2 and Black Marble nighttime lights, with publicly available spatial datasets to map informal settlements across the Global South. Utilizing a hierarchical identification procedure, we produce a 10 m resolution map of permanent informal settlements (PIS) representing the baseline year of circa 2020. An independent validation using a stratified random-point design reveals an overall accuracy of 96.0%, with user’s and producer’s accuracy of 93.6% and 98.3%, respectively. Our results show that the PIS are predominantly clustered in West Africa, Central Africa, and South Asia, with countries such as Lebanon and India exhibiting both large PIS areas and high areal proportions. Approximately 80% of the PIS are located in vulnerable tropical and arid climate zones prone to extreme weather, while 96% belong to countries exhibiting severe economic underperformance. This fast and scalable framework fills a key data gap in resource-limited regions and provides a high-resolution geospatial baseline for researchers, urban planners, and policymakers to assess vulnerability, allocate resources, and monitor sustainable urban development in the Global South. The repository contains two GeoTIFF datasets: (1) permanent informal settlements (PIS) and (2) potential permanent informal settlements (Potential PIS). In both datasets, a pixel value of 1 indicates the presence of the corresponding settlement class, whereas 0 represents all other areas. Each dataset is distributed as a ZIP archive containing all GeoTIFF tiles. Tile filenames indicate the upper-left coordinate of each tile, and the exact spatial extent is recorded in the GeoTIFF metadata.