AlphaEarth Foundations: An embedding field model for accurate and efficient global mapping from sparse label data

TL;DR

AlphaEarth Foundations introduces a novel embedding field model that integrates diverse geospatial data to produce accurate, scalable maps from sparse labels, outperforming existing methods without re-training.

Contribution

The paper presents a new embedding field model for global mapping that outperforms existing featurization approaches and is capable of assimilating multi-source spatial and temporal data.

Findings

Embeddings outperform other featurization methods across diverse mapping tasks.

The model enables accurate global mapping from sparse labels.

Released a comprehensive dataset of global embedding layers from 2017-2024.

Abstract

Unprecedented volumes of Earth observation data are continually collected around the world, but high-quality labels remain scarce given the effort required to make physical measurements and observations. This has led to considerable investment in bespoke modeling efforts translating sparse labels into maps. Here we introduce AlphaEarth Foundations, an embedding field model yielding a highly general, geospatial representation that assimilates spatial, temporal, and measurement contexts across multiple sources, enabling accurate and efficient production of maps and monitoring systems from local to global scales. The embeddings generated by AlphaEarth Foundations are the only to consistently outperform a suite of other well-known/widely accepted featurization approaches tested on a diverse set of mapping evaluations without re-training. We have released a dataset of global, annual, analysis-ready embedding field layers from 2017 through 2024.