Electricity demand mapping from open-source data for low- and middle-income countries

Abstract

Spatially resolved energy systems modelling is increasingly used to provide more accurate insights into electrification planning and infrastructure development, yet spatially resolved electricity demand data is often unavailable in low- and middle-income countries (LMICs). This study presents a novel, open-source methodology to build a high-resolution electricity demand map covering the buildings and industry sectors, and applies it to Zambia as a case study. Our approach integrates publicly available GIS data, national surveys (DHS), and official statistics. For the buildings sector, machine learning is used to map residential demand and a top-down model for services; industrial demand is assessed with a separate bottom-up process model. Our bottom-up estimates are validated against national statistics, capturing 70 % of residential and 80 % of industrial demand before final scaling. The results reveal a stark geographic concentration of consumption, with the Lusaka and Copperbelt provinces alone accounting for nearly 60 % of building demand and the vast majority of industrial demand. This granular dataset can underpin the development of spatially explicit energy system models, facilitating informed decisions on grid infrastructure expansion, optimising electrification for off-grid areas, and supporting more equitable energy access in line with Sustainable Development Goals. The methodology is designed for replicability in other countries, offering a valuable tool for researchers and policymakers across other LMICs.