Earth Embeddings

2026-08-04

AI Summary

This study addresses the high computational and storage costs incurred by repeatedly processing raw remote sensing imagery in Earth observation by proposing the use of foundation model–generated embeddings as reusable representations for efficient analysis of locations, image patches, or pixels. The work systematically reviews and evaluates various Earth embeddings across dimensions including spatial coverage, resolution, embedding dimensionality, storage overhead, and reproducibility, establishing a unified evaluation framework and guidelines for application deployment. Through strategies such as feature pooling, multi-source fusion, and spatial transfer, the embeddings are adapted to diverse downstream tasks—including land cover mapping, crop classification, ecological hazard modeling, socioeconomic prediction, and semantic retrieval—demonstrating their effectiveness and clarifying the conditions and performance boundaries under which they outperform traditional handcrafted features.

Abstract

Earth observation is moving from foundation models that users must run themselves toward embedding products that package model feature outputs as reusable data without needing to download and process the imagery used to generate them. Earth embeddings are vectors that summarize locations, image patches, or pixels, letting users analyze compact features instead of repeatedly training or running large models on raw satellite imagery. This chapter explains the main types of Earth embeddings, from implicit location encoders to explicit patch and pixel products, and compares their coverage, resolution, dimensionality, storage cost, licenses, and reproducibility. We review their use in land cover and crop mapping, ecological and hazard modeling, socioeconomic prediction, and semantic search, with evidence on when embeddings improve on conventional features and when pooling, fusion, or spatial transfer limit performance. Two case studies show practical workflows for similarity search and land cover mapping. We close with guidance for choosing, evaluating, storing, compressing, and publishing embeddings, and with open problems in oceanic and atmospheric coverage, uncertainty, and benchmarking.

Problem

• Earth embeddings
• satellite imagery
• feature representation
• reproducibility
• benchmarking

Innovation

• Earth embeddings
• foundation models
• remote sensing
• feature reuse
• satellite imagery analysis