{"paper":{"title":"Earth Science Foundation Models: From Perception to Reasoning and Discovery","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Foundation models integrate multimodal Earth data to advance from perception to reasoning and discovery.","cross_cats":["astro-ph.EP","cs.LG"],"primary_cat":"astro-ph.IM","authors_text":"Ben Fei, Bo Liu, Fenghua Ling, Feng Liu, Fengxiang Wang, Wanghan Xu, Wangxu Wei, Wenlong Zhang, Xiangyu Zhao, Xiao-Ming Wu, Yuehan Zhang, Zelin Song","submitted_at":"2026-05-09T08:34:30Z","abstract_excerpt":"Large foundation models (FMs) are transforming Earth science by integrating heterogeneous multimodal data, such as multi-platform imagery, gridded reanalysis data, diverse geophysical and geochemical observations, and domain-specific text, to support tasks ranging from basic perception to advanced scientific discovery. This paper provides a unified review of Earth science foundation models (Earth FMs) through two complementary dimensions: depth, which traces the evolution of model capabilities from perception to multimodal reasoning and agentic scientific workflows, and breadth, which summariz"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Large foundation models (FMs) are transforming Earth science by integrating heterogeneous multimodal data, such as multi-platform imagery, gridded reanalysis data, diverse geophysical and geochemical observations, and domain-specific text, to support tasks ranging from basic perception to advanced scientific discovery. This paper provides a unified review of Earth science foundation models through two complementary dimensions: depth and breadth, compiles more than 200 datasets and benchmarks, and outlines future directions toward more integrated, trustworthy, and actionable AI Earth scientists.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the representative multimodal Earth foundation models selected for review and the more than 200 compiled datasets and benchmarks are sufficiently comprehensive and unbiased to support a reliable unified roadmap of the entire field.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"The paper delivers a unified review and roadmap of Earth science foundation models, structured by capability depth from perception to agentic reasoning and by application breadth across atmosphere, hydrosphere, lithosphere, biosphere, anthroposphere, and cryosphere, while compiling over 200 datasets","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Foundation models integrate multimodal Earth data to advance from perception to reasoning and discovery.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"c1aab9aa9438a4060f8eeb6c8fa2ced9bfc370e36b4d0d16cb72fd23b6facd17"},"source":{"id":"2605.12542","kind":"arxiv","version":1},"verdict":{"id":"439b1801-05d0-44bd-be31-657f604ccf05","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-14T22:04:45.908253Z","strongest_claim":"Large foundation models (FMs) are transforming Earth science by integrating heterogeneous multimodal data, such as multi-platform imagery, gridded reanalysis data, diverse geophysical and geochemical observations, and domain-specific text, to support tasks ranging from basic perception to advanced scientific discovery. This paper provides a unified review of Earth science foundation models through two complementary dimensions: depth and breadth, compiles more than 200 datasets and benchmarks, and outlines future directions toward more integrated, trustworthy, and actionable AI Earth scientists.","one_line_summary":"The paper delivers a unified review and roadmap of Earth science foundation models, structured by capability depth from perception to agentic reasoning and by application breadth across atmosphere, hydrosphere, lithosphere, biosphere, anthroposphere, and cryosphere, while compiling over 200 datasets","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the representative multimodal Earth foundation models selected for review and the more than 200 compiled datasets and benchmarks are sufficiently comprehensive and unbiased to support a reliable unified roadmap of the entire field.","pith_extraction_headline":"Foundation models integrate multimodal Earth data to advance from perception to reasoning and discovery."},"references":{"count":299,"sample":[{"doi":"","year":2024,"title":"Artificial intelligence for geoscience: Progress, challenges, and perspectives,","work_id":"c2509f48-3f13-4605-9c8e-a8e4029f68be","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"Aurora: A foundation model of the atmosphere,","work_id":"b8faca7b-da10-4de9-9bf9-a0046ef78cc9","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2022,"title":"arXiv:2211.02556 [physics]","work_id":"f534dc00-8a07-4e74-af68-7414725d0fc6","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"Citygpt: Towards urban iot learning, analysis and interaction with multi-agent system,","work_id":"0cdd912c-6445-4619-ba80-d6c6f9c68c83","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"Graphcast: Ai model for faster and more accurate global weather forecasting,","work_id":"f9025072-0ab2-4f15-87ab-571dcc6f122a","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":299,"snapshot_sha256":"c72b056e7ad263a35c2d6318450fc9125a9f5c1a59c007fe666d4e3870005ac7","internal_anchors":2},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}