🌍 Awesome Geospatial Embeddings

A curated list of papers that define, analyze, or evaluate geospatial embeddings — how spatial, temporal, or semantic Earth data are represented in embedding space, and how those embeddings behave or are applied.

This list intentionally excludes generic pretraining or downstream models where embeddings are incidental.

Note: In this list, the Dataset column refers to publicly released embedding datasets or products, not the datasets used for model training or inference.

---

📖 1. Surveys and Concept Papers

Reviews, taxonomies, and position papers discussing geospatial embeddings conceptually or systematically.

Abbr.	Title	Publication	Paper
LossyNeuralCompression	Lossy Neural Compression for Geospatial Analytics: A Review	GRSM 2025	Lossy Neural Compression
EarthEmbeddings	Earth Embeddings: Harnessing the Information in Earth Observation Data with Machine Learning	SIGGRAPH Frontiers 2025	Earth Embeddings
EarthEmbeddings	Earth Embeddings:Towards AI-centric Representations of our Planet	EarthArxiv 2025	Earth Embeddings
EarthEmbeddingsAsProducts	Earth Embeddings as Products: Taxonomy, Ecosystem, and Standardized Access	Arxiv 2026	Earth Embeddings as Products
EarthEmbeddings	Earth Embeddings (book chapter)	Arxiv 2026	Earth Embeddings

---

📡 2. Location Embeddings

Methods that learn embeddings for coordinates, regions, or spatial contexts.

Abbr.	Title	Publication	Paper	Code	Dataset
GeoCLIP	GeoCLIP: Clip-Inspired Alignment between Locations and Images for Effective Worldwide Geo-localization	NeurIPS 2023	GeoCLIP	Code	null
LocationEncoder	Geographic Location Encoding with Spherical Harmonics and Sinusoidal Representation Networks	ICLR 2024	LocationEncoder	Code	null
SatCLIP	SatCLIP: Global, General-Purpose Location Embeddings with Satellite Imagery	AAAI 2025	SatCLIP	Code	null
MoRA	MoRA: Mobility as the Backbone for Geospatial Representation Learning at Scale	ICLR 2026	MoRA	Code	Project Page
HybridSlepian	Localized, High-resolution Geographic Representations with Slepian Functions	Arxiv 2026	HybridSlepian	Code	null
LIANet	Location Is All You Need: Continuous Spatiotemporal Neural Representations of Earth Observation Data	Arxiv 2026	LIANet	Code	Dataset
TTE	Tessellating The Earth	ECCV 2026	TTE	Code	null

---

🖼 3. Product Embeddings

Methods that embed or represent individual data products into latent spaces suitable for geospatial analysis.

Abbr.	Title	Publication	Paper	Code	Dataset
CLAY Embeddings	Clay Model v0 Embeddings	Source Cooperative 2024	null	Code	Dataset / LGND Clay Embeddings - Sentinel2
Major TOM Embeddings	Global and Dense Embeddings of Earth: Major TOM Floating in the Latent Space	Arxiv 2024	Arxiv Paper	Code	Dataset
Earth Genome Embeddings	Embeddings for all	null	Medium 2025	null	Dataset

🧩 4. Aggregate / Multi-Observation Embeddings

Methods that aggregate multiple observations — over time, sensors, or modalities — into a unified embedding representation (e.g., temporal sequences or multimodal stacks).

Abbr.	Title	Publication	Paper	Code	Dataset
PRESTO	Lightweight, Pre-trained Transformers for Remote Sensing Timeseries	Arxiv 2023	PRESTO	Code	null
Copernicus-FM	Towards a Unified Copernicus Foundation Model for Earth Vision	ICCV 2025 Oral	Copernicus-FM	Code	Copernicus-Embed-025deg
TESSERA	TESSERA: Precomputed FAIR Global Pixel Embeddings for Earth Representation and Analysis	Arxiv 2025	TESSERA	Code	Python Library
AlphaEarth	AlphaEarth Foundations: An embedding field model for accurate and efficient global mapping from sparse label data	Arxiv 2025	AlphaEarth	null	Google Satellite Embedding
ESD	Democratizing planetary-scale analysis: An ultra-lightweight Earth embedding database for accurate and flexible global land monitoring	Arxiv 2026	ESD	Code	ESD Dataset

---

🧠 5. Evaluation and Analysis of Embeddings

Works that probe, benchmark, or visualize the behavior and quality of geospatial embeddings.

Abbr.	Title	Publication	Paper	Code / Benchmark
DCVA	Unsupervised Deep Change Vector Analysis for Multiple-Change Detection in VHR Images	TGRS 2019	DCVA	null
NeuCo-Bench	NeuCo-Bench: A Novel Benchmark Framework for Neural Embeddings in Earth Observation	Arxiv 2025	NeuCo-Bench	Code
GeoINRID	Measuring the Intrinsic Dimension of Earth Representations	Arxiv 2025	GeoINRID	Code
NeurEO	NeurEO: dissecting Earth observation embeddings with computational neuroscience	Eurips REO 2025	NeurEO	null
Embed2ScaleWinner	Fused Foundation Model Embeddings for Earth Observation Compression: A Winning Solution to the Embed2Scale Challenge	Eurips REO 2025	Embed2ScaleWinner	Code
PatchPooling	From Pixels to Patches: Pooling Strategies for Earth Embeddings	Arxiv 2026	PatchPooling	Code
EmbedEO	How To Embed Matters: Evaluation of EO Embedding Design Choices	Arxiv 2026	EmbedEO	null
AlphaEarthInterpret	What on Earth is AlphaEarth? Hierarchical structure and functional interpretability for global land cover	Arxiv 2026	AlphaEarthInterpret	null
EmbedUrban	Earth Embeddings Reveal Diverse Urban Signals from Space	Arxiv 2026	EmbedUrban	null
EmbedComplementarity	Better Together: Evaluating the Complementarity of Earth Embedding Models	Arxiv 2026	EmbedComplementarity	Code
LocEnc-XAI	What's in an Earth Embedding? An Explainability Analysis of Location Encoders	Arxiv 2026	LocEnc-XAI	Code
TSA-Tessera	Temporal Sensitivity Analysis of Tessera Embeddings	Arxiv 2026	TSA-Tessera	null

---

🏆 6. Open Challenges and Competitions

Active or past public challenges focused on geospatial embedding learning or evaluation.

Year	Challenge	Host / Platform	Link
2025	CVPR EARTHVISION: Embed2Scale Challenge	CVPR EarthVision 2025	Embed2Scale
2025	TerraMind Blue-Sky Challenge Round 2: Team Urban Embeddings	IBM & ESA Φ-lab	TerraMind Challenge
2026	Reaching new heights with GeoFM	KTH & ITU & AI4Good & ESA Φ-lab	GeoFM

---

🌎 7. Applications of Geospatial Embeddings

Papers that use existing embeddings for mapping, retrieval, similarity search, or reasoning tasks.

Abbr.	Title	Publication	Paper	Code / Dataset
-	Cropland Mapping using Geospatial Embeddings	Arxiv 2025	Paper	Code
-	Leveraging Compact Satellite Embeddings and Graph Neural Networks for Large-Scale Poverty Mapping	Eurips REO 2025	Paper	null
-	Harvesting AlphaEarth: Benchmarking the Geospatial Foundation Model for Agricultural Downstream Tasks	Arxiv 2025	Paper	Code
-	From Landslide Conditioning Factors to Satellite Embeddings: Evaluating the Utilisation of Google AlphaEarth for Landslide Susceptibility Mapping using Deep Learning	Arxiv 2026	Paper	null
EarthEmbeddingExplorer	Sentinel-2 Image Retrieval with Global, Multi-modal Embeddings	EGU26/ICLR26 ML4RS	Paper	WebAPP
-	Inferring Height from Earth Embeddings: First insights using Google AlphaEarth	Arxiv 2026	Paper	null
RS-Embed	Any Model, Any Place, Any Time: Get Remote Sensing Foundation Model Embeddings On Demand	SIGSPATIAL 2026	RS-Embed	Code
-	Assessing the Effectiveness of Deep Embeddings for Tree Species Classification in the Dutch Forest Inventory	Arxiv 2026	Paper	Code
SSDM	Structure-Semantic Decoupled Modulation of Global Geospatial Embeddings for High-Resolution Remote Sensing Mapping	Arxiv 2026	SSDM	null
-	Characterizing Brazilian Atlantic Forest Restoration Outcomes with Geospatial AlphaEarth Embeddings	ICLR ML4RS 2026	Paper	null
-	Do Foundation Model Embeddings Improve Cross-Country Crop Yield Generalisation? A Leave-One-Country-Out Evaluation in Sub-Saharan Africa	Arxiv 2026	Paper	Code
-	Geospatial foundation-model embeddings improve population estimation unevenly across space and scale	Arxiv 2026	Paper	Code
DFR-Gemma	Enabling Intrinsic Reasoning over Dense Geospatial Embeddings with DFR-Gemma	Arxiv 2026	Paper	null
-	Characterizing AlphaEarth Embedding Geometry for Agentic Environmental Reasoning	Arxiv 2026	Paper	null
-	Continuous biome representations from Earth observation embeddings	Arxiv 2026	Paper	null
-	Embeddings based Anomaly Detection for Cleaning Global Crop Type Reference Datasets	ECCV Terrabytes II 2026	Paper	Code
-	Above-ground Biomass Estimation with Geospatial Foundation Models	Arxiv 2026	Paper	Code
-	Earth observation embeddings are effective sub-grid descriptors for probabilistic weather downscaling	Arxiv 2026	Paper	null
-	Planetary Prediction Engine: Autonomous Geospatial Prediction via Intelligent Data Selection and Foundation Model Embeddings	Arxiv 2026	Paper	Blog

---

🌐 8. Related Works (for context)

Broader studies on representation learning and multimodal embeddings that provide conceptual context for geospatial embedding research.

🛰️ 8.1 Remote Sensing Representation Learning

Representative works on large-scale or foundational representation learning in Earth observation — including early frameworks and recent foundation models.

For a more comprehensive collection of remote sensing foundation models, see Awesome Remote Sensing Foundation Models.

Abbr.	Title	Publication	Paper	Code
MOSAIKS	A generalizable and accessible approach to machine learning with global satellite imagery	Nature Comm. 2021	MOSAIKS	Code
SeCo	Seasonal Contrast: Unsupervised Pre-Training from Uncurated Remote Sensing Data	ICCV 2021	SeCo	Code
SatMAE	SatMAE: Pre-training Transformers for Temporal and Multi-Spectral Satellite Imagery	NeurIPS 2022	SatMAE	Code
Galileo	Galileo: Learning Global & Local Features of Many Remote Sensing Modalities	ICML 2025	Galileo	Code
GAIR	GAIR: Improving Multimodal Geo-Foundation Model with Geo-Aligned Implicit Representations	Arxiv 2025	GAIR	null
OlmoEarth	OlmoEarth	Arxiv 2025	OlmoEarth / Project Page	Code / Python Library
-	How Much of a Model Do We Need? Redundancy and Slimmability in Remote Sensing Foundation Models	Arxiv 2026	Paper	null
-	LEPA: Learning Geometric Equivariance in Satellite Remote Sensing Data with a Predictive Architecture	Arxiv 2026	Paper	Code
TerraCodec	TerraCodec: Compressing Optical Earth Observation Data	Arxiv 2025	Paper	Code
HideAndSeek	Hide and Seek: Investigating Redundancy in Earth Observation Imagery	Arxiv 2026	Paper	null
-	No One Knows the State of the Art in Geospatial Foundation Models	Arxiv 2026	Paper	Code
-	Pretrain Where? Investigating How Pretraining Data Diversity Impacts Geospatial Foundation Model Performance	CVPR EarthVision 2026	Paper	Code
Tempov	A satellite foundation model for improved wealth monitoring	Arxiv 2026	Paper	null
EarthShift	EarthShift: a benchmark for measuring robustness to real-world distribution shifts in Earth observation	Arxiv 2026	Paper	Code

🔄 8.2 General Multimodal Embedding and Modality Gap Studies

Works that investigate or analyze the latent space gap between visual and textual modalities — relevant for understanding cross-modal geospatial embeddings.

Abbr.	Title	Publication	Paper	Code
ModalityGap	Mind the Gap: Understanding the Modality Gap in Multi-modal Contrastive Representation Learning	NeurIPS 2022	ModalityGap	Code
DinoVision	Into the Rabbit Hull: From Task-Relevant Concepts in DINO to Minkowski Geometry	Arxiv 2025	DinoVision	Code

🌋 8.3 Potentially Interesting Works

Applied or thematic studies that are not embedding-centric but highlight promising directions or use cases for geospatial representation learning (e.g., disaster response, environmental monitoring, planetary mapping).

Abbr.	Title	Publication	Paper	Code
-	The Potential of Copernicus Satellites for Disaster Response: Retrieving Building Damage from Sentinel-1 and Sentinel-2	Arxiv 2025	Paper	Code
-	Landslide Hazard Mapping with Geospatial Foundation Models: Geographical Generalizability, Data Scarcity, and Band Adaptability	Arxiv 2025	Paper	null
-	On the Generalizability of Foundation Models for Crop Type Mapping	IGARSS 2025	Paper	Code
-	Deploying Geospatial Foundation Models in the Real World: Lessons from WorldCereal	ICML Terrabytes 2025	Paper	null

---

🤝 Contributing

Pull requests welcome!

Please add only works that explicitly discuss or apply geospatial embeddings — their definition, analysis, evaluation, or use.

🔖 Citation

If you use this list in your research, please consider citing our paper:

@unpublished{fang2026earth,
    author = {Heng Fang and Adam J. Stewart and Isaac Corley and Xiao Xiang Zhu and Hossein Azizpour},
    title = {Earth Embeddings as Products: Taxonomy, Ecosystem, and Standardized Access},
    note = {arXiv preprint arXiv:2601.13134},
    month = jan,
    year = {2026}
}

@unpublished{stewart2026earth,
  author = {Adam J. Stewart and Heng Fang and Isaac Corley and Xiao Xiang Zhu},
  title = {Earth Embeddings},
  note = {arXiv preprint arXiv:2608.03410},
  month = aug,
  year = {2026}
}