2026-02-17 スタンフォード大学
Human Development Index estimates for 2019. Gray in the grid-level estimates indicates land area believed to be unsettled. | Adapted from Sherman et al. (Nature Communications, 2026)
＜関連情報＞
- https://news.stanford.edu/stories/2026/02/satellite-imagery-ai-un-human-development-index
- https://www.nature.com/articles/s41467-026-68805-6
- https://www.nber.org/papers/w34315
衛星画像と機械学習を用いた国連人間開発指数の世界規模の高解像度推定 Global high-resolution estimates of the UN Human Development Index using satellite imagery and machine learning
Luke Sherman,Jonathan Proctor,Hannah Druckenmiller,Heriberto Tapia & Solomon Hsiang
Nature Communications Published:17 February 2026
DOI:https://doi.org/10.1038/s41467-026-68805-6
Abstract
The United Nations Human Development Index, which incorporates income, education and health, is arguably the most widely used alternative to gross domestic product. However, official country-resolution estimates (N=191) limit its use. We build on recent advances in machine learning and satellite imagery to produce and distribute global estimates of the Human Development Index for municipalities (N=61,530) and a 0. 1° × 0. 1° grid (N=819,309). To construct these estimates, we develop and validate a generalizable downscaling technique based on satellite imagery that allows for training and prediction with observations of arbitrary size and shape. We show how our estimates can improve decision-making and that more than half of the global population was previously assigned to the incorrect Human Development Index quintile within each country due to aggregation bias. We publish the satellite features necessary to increase the spatial resolution of any other administrative data that is detectable via imagery.
衛星画像と機械学習で何が測定できるか？ What Can Satellite Imagery and Machine Learning Measure?
Jonathan Proctor, Tamma Carleton, Trinetta Chong, Taryn Fransen, Simon Greenhill, Jessica Katz, Hikari Murayama, Luke Sherman, et al.
National Bureau of Economic Research Issue Date: October 2025
DOI:10.3386/w34315
Satellite imagery and machine learning (SIML) are increasingly being combined to remotely measure social and environmental outcomes, yet use of this technology has been limited by insufficient understanding of its strengths and weaknesses. Here, we undertake the most extensive effort yet to characterize the potential and limits of using a SIML technology to measure ground conditions. We conduct 115 standardized large-scale experiments using a composite high-resolution optical image of Earth and a generalizable SIML technology to evaluate what can be accurately measured and where this technology struggles. We find that SIML alone predicts roughly half the variation in ground measurements on average, and that variables describing human society (e.g. female literacy, R²=0.55) are generally as easily measured as natural variables (e.g. bird diversity, R²=0.55). Patterns of performance across measured variable type, space, income and population density indicate that SIML can likely support many new applications and decision-making use cases, although within quantifiable limits.
