Diesel GenSat: Using Satellite Data to Detect Diesel-Powered Irrigation for Guiding Electrification in Ethiopia

Abstract

In Sub-Saharan Africa, electricity access is progressing, but electricity use for economic growth remains stagnant. Powering economies sustainably is vital to enhancing livelihoods and is particularly challenging in agriculture-led rural economies. The financial viability of electrification hinges on identifying potential sources of demand to ensure sustainable revenues for utilities, which in turn provides economic benefits to consumers. This paper presents a technique for identifying areas with diesel-powered irrigation activity in Ethiopia based on remotely sensed data. We develop and evaluate a supervised classification model based on data collected in the Western Ethiopia Highlands on irrigation practices. We find that a feature-based multivariate time series classification approach combined with a k-Nearest Neighbors model accurately predicts about 75% of areas with diesel-powered irrigation activity. Our results suggest that our technique could be valuable in identifying areas in Ethiopia with potential anchor loads for electricity grid extension by replacing existing diesel pumps for irrigation with electric pumps. Guidance on financially-viable areas to expand electricity networks, especially those with economically-vibrant activities like irrigation, is crucial for enabling electricity service providers to recover costs and expand access to more communities more quickly.