Leveraging Compact Satellite Embeddings and Graph Neural Networks for Large-Scale Poverty Mapping

TL;DR

This paper introduces a graph neural network approach using compact satellite embeddings to generate detailed poverty maps in Sub-Saharan Africa, addressing data limitations and privacy-induced coordinate noise.

Contribution

It presents a novel graph-based method utilizing low-dimensional satellite embeddings and fuzzy label loss to enhance large-scale poverty prediction accuracy.

Findings

Graph structure slightly improves prediction accuracy.

Satellite embeddings effectively capture socioeconomic information.

Method demonstrates potential for large-scale socioeconomic mapping.

Abstract

Accurate, fine-grained poverty maps remain scarce across much of the Global South. While Demographic and Health Surveys (DHS) provide high-quality socioeconomic data, their spatial coverage is limited and reported coordinates are randomly displaced for privacy, further reducing their quality. We propose a graph-based approach leveraging low-dimensional AlphaEarth satellite embeddings to predict cluster-level wealth indices across Sub-Saharan Africa. By modeling spatial relations between surveyed and unlabeled locations, and by introducing a probabilistic "fuzzy label" loss to account for coordinate displacement, we improve the generalization of wealth predictions beyond existing surveys. Our experiments on 37 DHS datasets (2017-2023) show that incorporating graph structure slightly improves accuracy compared to "image-only" baselines, demonstrating the potential of compact EO embeddings for large-scale socioeconomic mapping.