Global mapping of GDP at 1 km2 using VIIRS nighttime satellite imagery
Description
Frequent and rapid spatially explicit assessment of socioeconomic development is critical for achieving the Sustainable Development Goals (SDGs) at both national and global levels. In past decades, scientists have proposed many methods for estimating human activity on the Earth’s surface at various spatiotemporal scales using nighttime lights (NTL) data. NTL data and the associated processing methods have been limited their reliability and applicability for systematic measuring and mapping of socioeconomic development. This study utilizes Visible Infrared Imaging Radiometer Suite (VIIRS) NTL and the Isolation Forest machine learning algorithm for more intelligent data processing to capture human activities. We use machine learning and NTL data to map gross domestic product (GDP) at 1 km2. We then use these data products to derive inequality indexes (e.g. Gini coefficients) at nationally aggregate levels. This flexible approach processes the data in an unsupervised manner at various spatial scales. Our assessments show that this method produces accurate sub-national GDP data products for mapping and monitoring human development uniformly across the globe. This repository hosts one zipped geotiff file for the global global GDP (constant 2011 US$) at 1km output of the analysis and the one tabular file (csv) produced by the aggregated results of inequality analysis - NTL-based Gini index and 20:20 ratios.