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

2025/11/03 by Pettersson, Markus B., Daoud, Adel
#FOS: Computer and information sciences #Machine Learning (cs.LG)

paper · doi:10.48550/arxiv.2511.01408

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.

Citations

• AlphaEarth Foundations: An embedding field model for accurate and efficient global mapping from sparse label data
• TerraMind: Large-Scale Generative Multimodality for Earth Observation
• Statistical modeling: the three cultures
• Fast Graph Representation Learning with PyTorch Geometric
• Inductive Representation Learning on Large Graphs
• Semi-Supervised Classification with Graph Convolutional Networks

Related

• Predicting Poverty Level from Satellite Imagery using Deep Neural Networks
• A satellite foundation model for improved wealth monitoring
• Platonic Representations for Poverty Mapping: Unified Vision-Language Codes or Agent-Induced Novelty?
• Poverty Prediction with Public Landsat 7 Satellite Imagery and Machine Learning
• SATELLITE AND ARTIFICIAL INTELLIGENCE IN MAPPING MULTIDIMENSIONAL POVERTY IN AFRICA