Earth Embeddings Reveal Diverse Urban Signals from Space

Wenjing Gong; Udbhav Srivastava; Yuchen Wang; Yuhao Jia; Qifan Wu; Weishan Bai; Yifan Yang; Xiao Huang

cs.LGApr 3, 2026

Earth Embeddings Reveal Diverse Urban Signals from Space

Authors: Wenjing Gong, Udbhav Srivastava, Yuchen Wang, Yuhao Jia, Qifan Wu, Weishan Bai, Yifan Yang, Xiao Huang,

Organizations: Texas A&M University, College Station, TX, USA. · The University of Alabama, Tuscaloosa, AL, USA. · Emory University, Atlanta, GA, USA.

Abstract

Conventional urban indicators derived from censuses, surveys, and administrative records are often costly, spatially inconsistent, and slow to update. Recent geospatial foundation models enable Earth embeddings, compact satellite image representations transferable across downstream tasks, but their utility for neighborhood-scale urban monitoring remains unclear. Here, we benchmark three Earth embedding families, AlphaEarth, Prithvi, and Clay, for urban signal prediction across six U.S. metropolitan areas from 2020 to 2023. Using a unified supervised-learning framework, we predict 14 neighborhood-level indicators spanning crime, income, health, and travel behavior, and evaluate performance under four settings: global, city-wise, year-wise, and city-year. Results show that Earth embeddings capture substantial urban variation, with the highest predictive skill for outcomes more directly tied to built-environment structure, including chronic health burdens and dominant commuting modes. By contrast, indicators shaped more strongly by fine-scale behavior and local policy, such as cycling, remain difficult to infer. Predictive performance varies markedly across cities but remains comparatively stable across years, indicating strong spatial heterogeneity alongside temporal robustness. Exploratory analysis suggests that cross-city variation in predictive performance is associated with urban form in task-specific ways. Controlled dimensionality experiments show that representation efficiency is critical: compact 64-dimensional AlphaEarth embeddings remain more informative than 64-dimensional reductions of Prithvi and Clay. This study establishes a benchmark for evaluating Earth embeddings in urban remote sensing and demonstrates their potential as scalable, low-cost features for SDG-aligned neighborhood-scale urban monitoring.

Figures & tables

Figure 1: Study area and research framework. (a) The selected six MSAs: Atlanta–Sandy Springs–Alpharetta, GA (Atlanta), Chicago–Naperville–Elgin, IL–IN–WI (Chicago), Houston–The Woodlands–Sugar Land, TX (Houston), Los Angeles–Long Beach–Anaheim, CA (Los Angeles), New York–Newark–Jersey City, NY–NJ (New York), and Seattle–Tacoma–Bellevue, WA (Seattle). (b) Land area (km²) of the six MSAs. (c) Total population from 2020 to 2023 for each MSA. (d) Conceptual framework of Earth-embedding-based urban signal prediction.

Figure 2: Benchmarking Earth embeddings for predicting urban signals. (a) Global predictive performance (test R²) of three Earth embeddings (AlphaEarth, Prithvi, and Clay) across 14 urban indicators spanning crime, income, health, and travel behavior for six US MSAs over 2020–2023 (see Supplementary Table 2 for details). (b) Global mean R² by thematic domain, obtained by averaging across indicators, cities, and years, summarizing domain-level differences in model skill. (c) Distribution of city–year experiment test R² values across all indicators, showing the variability and upper-tail behavior of each embedding.

Figure 3: Cross-city heterogeneity in the predictability of urban signals. (a) City-wise predictive performance (test R²) of three Earth embeddings (AlphaEarth, Prithvi, and Clay) across four domains for six US MSAs over 2020–2023. (b) Hierarchical clustering of MSAs based on AlphaEarth’s mean R² across domains. (c–e) Exploratory relationships between urban form indicators and domain-specific AlphaEarth predictive performance: (c) population density, (d) employment and household entropy, and (e) walkability index. Colored lines show simple domain-specific trend lines, and $\rho$ indicates Spearman’s rank correlation.

Figure 4: Consistency of urban signal prediction across years (2020–2023). Annual predictive performance (test R²) for four urban domains (crime, income, health, and travel) from 2020 to 2023, aggregated across all indicators and MSAs for (a) AlphaEarth, (b) Prithvi, and (c) Clay. Lines show the domain-wise mean R² in each year, and shaded bands indicate variability across indicators.

Figure 5: Global information density and representation efficiency of Earth embeddings. Mean predictive performance (test R²) by domain (crime, income, health, and travel) for (a) Prithvi and (b) Clay using their original high-dimensional embeddings and five 64-d compressed variants (FA-64, Isomap-64, kPCA-64, PCA-64, and RP-64). Bars show domain-wise mean R² aggregated across all indicators, MSAs, and years. The dashed line with stars denotes the original 64-d AlphaEarth performance.

Theme	Indicator	count	mean	s.d.	min	max
Crime	Log (Violent Crime)	328,960	0.924	2.112	0.000	15.981
	Log (Petty Crime)	328,960	1.030	2.294	0.000	15.934
Income	Log (Median Income)	306,284	11.255	0.548	7.824	12.429
Health	%Obesity	287,850	32.045	7.444	10.400	64.400
	%Diabetes	287,850	11.567	3.706	0.600	50.800
	%Inactivity	287,850	25.033	8.273	6.900	75.500

Table 1: Summary statistics of urban signals for six MSAs from 2020 to 2023.

Appendix figures & tables17 assets

Supplementary material from the paper’s appendix.

Appendix

SDGs	Theme	Indicator	Definition
Goal 11: Sustainable Cities and Communities	Crime	Log (Violent Crime)	Natural logarithm of the violent crime rate.
		Log (Petty Crime)	Natural logarithm of the petty crime rate.
Goal 1: No Poverty	Income	Log (Median Income)	Natural logarithm of median household income.
Goal 3: Good Health and Well-being	Health	%Obesity	Model-based estimate for crude prevalence of obesity among adults.
		%Diabetes	Model-based estimate for crude prevalence of diagnosed diabetes among adults.
		%Inactivity	Model-based estimate for crude prevalence of no leisure-time physical activity among adults.

Appendix

Table S1: Definitions of urban signals and their corresponding SDGs.

Figure S1: Distribution of the crime indicators for six U.S. metropolitan areas, 2020–2023.

Figure S2: Distribution of the income indicators for six U.S. metropolitan areas, 2020–2023.

Figure S3: Distribution of the health indicators for six U.S. metropolitan areas, 2020–2023.

Figure S4: Distribution of the travel indicators for six U.S. metropolitan areas, 2020–2023.

Figure S5: AlphaEarth embedding maps for six U.S. metropolitan areas, aggregated to CBGs, 2020–2023. Colors represent a three-dimensional PCA projection of the embeddings, with PC1–PC3 mapped to the red, green, and blue channels. Hues are comparable across cities within a given year.

Figure S6: Prithvi embedding maps for six U.S. metropolitan areas, aggregated to CBGs, 2020–2023. Colors represent a three-dimensional PCA projection of the embeddings, with PC1–PC3 mapped to the red, green, and blue channels. Hues are comparable across cities within a given year.

Figure S7: Clay embedding maps for six U.S. metropolitan areas, aggregated to CBGs, 2020–2023. Colors represent a three-dimensional PCA projection of the embeddings, with PC1–PC3 mapped to the red, green, and blue channels. Hues are comparable across cities within a given year.

Figure S8: Supervised ML models’ performance (test R²) of Earth observation embeddings for the crime theme (global setting).

Figure S9: Supervised ML models’ performance (test R²) of Earth embeddings for the income theme (global setting).

Figure S10: Supervised ML models’ performance (test R²) of Earth embeddings for the health theme (global setting).

Figure S11: Supervised ML models’ performance (test R²) of Earth embeddings for the travel theme (global setting).

Theme	Indicator	AlphaEarth	Prithvi	Clay
Crime	Log (Violent)	0.480	0.388	0.366
Crime	Log (Petty)	0.369	0.176	0.329
Income	Log (Median Income)	0.444	0.311	0.209
Health	%Obesity	0.689	0.666	0.411
	%Diabetes	0.593	0.480	0.264
	%Inactivity	0.627	0.495	0.298

Appendix

Table S2: Indicator-level test

R^{2}

of Earth embeddings across all MSAs and years (global setting).

Figure S12: City-wise benchmark performance (test R²) of Earth embeddings across all years (2020–2023).

Figure S13: Year-wise benchmark performance (test R²) of Earth embeddings across all six MSAs.

Theme	Indicator	AlphaEarth	Prithvi	Prithvi_fa	Prithvi_isomap	Prithvi_kpca	Prithvi_pca	Prithvi_rp
Crime	Log (Violent Crime)	0.480	0.388	0.385	0.349	0.367	0.385	0.373
Crime	Log (Petty Crime)	0.369	0.176	0.164	0.153	0.161	0.167	0.167
Income	Log (Median Income)	0.444	0.311	0.299	0.278	0.290	0.300	0.301
Health	%Obesity	0.689	0.666	0.620	0.541	0.584	0.615	0.636
	%Diabetes	0.593	0.480	0.456	0.383	0.419	0.443	0.461
	%Inactivity	0.627	0.495	0.470	0.397	0.429	0.464	0.473

Appendix

Table S3: Indicator-level test

R^{2}

of Prithvi embeddings and their dimensionality-reduced variants across all MSAs and years (global setting).

Theme	Indicator	AlphaEarth	Clay	Clay_fa	Clay_isomap	Clay_kpca	Clay_pca	Clay_rp
Crime	Log (Violent Crime)	0.480	0.366	0.166	0.133	0.171	0.162	0.324
Crime	Log (Petty Crime)	0.369	0.329	0.119	0.096	0.127	0.117	0.257
Income	Log (Median Income)	0.444	0.208	0.087	0.081	0.091	0.087	0.171
Health	%Obesity	0.689	0.411	0.300	0.112	0.156	0.119	0.292
	%Diabetes	0.593	0.263	0.097	0.046	0.068	0.053	0.187
	%Inactivity	0.627	0.298	0.142	0.074	0.102	0.070	0.221

Appendix

Table S4: Indicator-level test

R^{2}

of Clay embeddings and their dimensionality-reduced variants across all MSAs and years (global setting).

Geospatial Representation Learning

Geospatial representation learning builds reusable embeddings of places, imagery, maps, mobility, and Earth data for spatial prediction and reasoning.

See all 133 papers

Sep 8, 2026

Applying foundation model embeddings towards urban livability evaluation

While accurate measurement of socioeconomic indicators remains challenging in data-scarce regions, which limits policy interventions and resource allocation, high-resolution geospatial data is widely available and can contain information on various livability statistics. We investigate which physical features are encoded within foundation model embeddings, such as AlphaEarth, AnySat, and TerraMind, and provide a systematic framework for identifying the most predictive geospatial indicators. By analyzing how different types of geospatial data influence urban livability predictions, our approach enables researchers to prioritize the most informative features for their specific applications. Additionally, we demonstrate how to leverage foundation model embeddings to enhance prediction performance for these outcomes. This work contributes a principled methodology for extracting actionable information from satellite imagery while accounting for complex spatial dependencies, with applications in predicting urban livability in regions with limited observation data.

Ayush Khot, Wen Zhou, Shaowen Wang

Siebel School of Computing and Data Science, University of Illinois at Urbana-Champaign, Urbana, IL, USA · Department of Geography and Geographic Information Science, University of Illinois at Urbana-Champaign, Urbana, IL, USA · Department of Geography and Geographic Information Science & Siebel School of Computing and Data Science, University of Illinois at Urbana-Champaign, Urbana, IL, USA

Feature Attribution Geospatial Foundation Models Geospatial Representation Learning Remote Sensing

Ask PDF

Aug 4, 2026

Earth Embeddings

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.

Adam J. Stewart, Heng Fang, Isaac A. Corley +1

Chair of Data Science in Earth Observation, Technical University of Munich, Munich, Germany · Munich Center for Machine Learning, Munich, Germany · KTH Royal Institute of Technology, Stockholm, Sweden +1

Geospatial Representation Learning Land Cover Classification Remote Sensing Earth Science

Ask PDF

May 11, 2026

Slum Detection and Density Mapping with AlphaEarth Foundations: A Representation Learning Evaluation Across 12 Global Cities

Pixel-level slum mapping has long been constrained by limited cross-city generalisation, the absence of continuous density estimation, and weak global comparability. AlphaEarth Foundations (AEF), a globally consistent 64-dimensional annual surface embedding at 10 m, offers a new analysis-ready basis for lightweight slum monitoring, but its applicability to slum detection - an indirectly coupled task shaped by both built form and socio-economic processes - remains untested. We evaluate AEF on slum classification and sub-pixel density estimation across 12 cities and 69 city-year pairs (2017-2024), using GRAM pseudo-masks as supervisory labels. The evaluation spans four training strategies, two protocols (random split and 3x3 spatial block cross-validation), six auxiliary feature configurations, and five baseline models, complemented by representation-level analyses (PCA, SHAP) and full-AOI mapping. Five findings emerge. (1) Same-city cross-year training is optimal under both protocols (median spatial F1 = 0.616, R^2 = 0.466); temporal expansion outperforms cross-city transfer, indicating city-scale representational drift. (2) Regression R^2 is driven primarily by zero/non-zero boundary discrimination: positive-pixel R^2 is consistently negative across all cities, revealing limited capacity to model intra-pixel density gradients at 10 m. (3) PC36 is consistently top-ranked across tasks; classification saturates at k = 32 while regression remains unsaturated at k = 64. (4) POI features yield the largest density gain (Delta R^2 = +0.064). (5) For six cities meeting dual-task usability thresholds, full-AOI inference across 2017-2024 preserves slum cluster structure (mean SSIM = 0.926). The study delineates the capabilities and complementarity needs of foundation-model embeddings for slum monitoring.

Shuyang Hou, Ziqi Liu, Haoyue Jiao +7

State Key Laboratory of Information Engineering in Surveying, Mapping, and Remote Sensing, Wuhan University, Wuhan, China

Geospatial Foundation Models Geospatial Representation Learning Spatial Generalization Semantic Mapping Density Estimation Urban Representation Learning

Ask PDF

Abstract

Geospatial Representation Learning

Explore similar work

Applying foundation model embeddings towards urban livability evaluation

Earth Embeddings

Slum Detection and Density Mapping with AlphaEarth Foundations: A Representation Learning Evaluation Across 12 Global Cities