Existing data for informing energy, climate, and development policymaking in sub-Saharan Africa are scarce. Further, traditional data collection methods, such as surveys, are expensive to scale. Satellite imagery, which is rapidly becoming available at higher resolutions, combined with machine learning techniques, is increasingly used in the region to derive information for guiding decision-making where alternative solutions do not exist.

Our initial study aimed to address electricity demand constraints in the electrification planning process to ensure that electrification stimulates economic growth. We used a novel approach to identify areas with existing diesel-powered irrigation in Ethiopia by combining ground data from an agricultural survey with satellite-measured pollution, crop cover, and topography data.

Though our study showed promising results, the limited ground-truth sample size collected and the nature of the publicly available satellite data used makes it difficult to generalize our findings beyond the study region of Amhara. It is, therefore, crucial to explore innovative methods to increase the volume and diversity of ground-truth samples and improve the transferability of our models. We argue that using higher resolution satellite imagery could overcome some of the challenges encountered in our initial study and aid in realizing the potential impacts of the work.