AI Science Tracker
EN

AlphaEarth crop embeddings transferred less well between the United States and India

Indian Institute of Technology Kharagpur; Indian Institute of Technology Delhi; Tata Consultancy Services Research · 27 Oct 2026 · other

Discovery

AlphaEarth crop embeddings transferred less well between the United States and India

Source check: Source checked. Not independent scientific confirmation.

Finding: In a satellite crop-classification study, a classifier trained on US data scored 0.326 macro F1 in Northern India using AlphaEarth embeddings, versus 0.493 using seasonal harmonic features. The ranking also favored harmonic features in the reverse direction, even though embeddings performed better in held-out US county comparisons.

Technology: AlphaEarth Foundations geospatial embeddings; supervised crop classifiers

Publication: Preprint · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.

AI contribution
Learned AlphaEarth embeddings encoded satellite imagery for the classifiers. Comparing those embeddings with engineered seasonal features quantified the regional transfer deficit.
Test and result
The study compared satellite representations on crops represented in US field-scale data from 2017–2023 and Northern India data from 2021, with spatial validation and cross-region transfer.
Limitation / still unknown
The geographies differ in seasons, sample selection and field structure. The results do not establish transferability to other regions or to all crops.
Evidence sources and annotation context

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

Evidence sources: https://arxiv.org/abs/2610.36857v1 · https://arxiv.org/html/2610.36857v1

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review date: 21 Nov 2026. Review scope: §§2.1–2.4 and 3.6–3.8, Figures 9–11.

No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.

Editorial assessment: Impact 2/5 · Validation 2/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Open in the full catalog for filters, charts and related discoveries. Browse other discoveries.