Spatializing global electricity consumption (2012–2024) via night-time light remote sensing

Night-time light remote sensing offers a unique proxy for human electricity consumption, but existing global datasets suffer from temporal lags, coarse resolution, and a lack of spatially explicit products. To address these gaps, this study introduces a grouped modelling strategy based on national development levels (per capita electricity consumption and economic structure). Using NPP-VIIRS night-time light data and global electricity statistics (2012–2024), we developed remote sensing estimation models at national and global scales and generated a 500-m resolution gridded electricity consumption dataset. The results reveal: (1) The night-time light–electricity relationship is nonlinear in low- and high-consumption countries but linear in medium-consumption countries; the grouped model improves national-scale estimation accuracy (adjusted R2 > 0.832). (2) A global linear time-series model (adjusted R2 = 0.918) enables robust annual electricity consumption estimation. (3) The 500-m dataset reveals an eastward shift of global consumption centres towards the Asia-Pacific and emerging economies, with Europe/North America showing ‘deep transition’ and Southeast/East Asia ‘steady expansion’. The generated dataset and modelling approach provide a reproducible remote sensing tool for large-scale energy monitoring, particularly valuable for data-scarce regions.