{"id":"0a037142368ab87f2c362bbf96c7259f3b65e256ff34bf0ad2614f0d81327216","title":"arXiv cs.AI: EarthRecord and multimodal agents analyzing satellite imagery","argument":"The authors of EarthRecord argue that existing benchmarks have not measured whether multimodal agents can analyze raw satellite archives to identify land change. Earlier benchmarks use curated images or preprocessed inputs, while EarthRecord gives agents data from 56 sites across four decades of Landsat, MODIS, VIIRS, and Sentinel-2 imagery, leaving cloud masking and cross-sensor harmonization to the agents. It poses 238 questions across ten analysis tasks and scores answers against withheld references. Controls remove the record or alter one satellite’s data while nothing changes on the ground. The authors report that leading agents complete most analyses end to end but falter when an answer requires mapping from imagery. In the alteration control, half of the right answers follow the changed data. The authors interpret this as the record moving answers from prior knowledge to the pixels, but not yet to real change on the ground. They present EarthRecord as a platform for measuring and advancing agents’ land-change analysis.","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The supplied evidence is the paper’s abstract; it does not include the full evaluation details or results by agent and task. The abstract does not specify how the benchmark’s withheld references compare in reliability or coverage.","speaker":{"name":"Zhang, Haoran, Sun, Huizhen, Wen, Hanbin, Leng, Jie, Zhu, Jiaxin, Sun, Zhaohua","roleAtSource":"Article author"},"source":{"title":"EarthRecord: From Pixels to Land Change with Multimodal Agents over Four Decades of Satellite Imagery","publisher":"arXiv cs.AI","format":"Article","url":"https://arxiv.org/abs/2611.12934","publishedAt":"2026-11-06","dateBasis":"Date basis not provided."},"evidence":[{"text":"The authors argue that existing benchmarks do not measure agents’ ability to analyze raw multi-sensor satellite archives for land change.","excerpt":"Existing benchmarks fall short: they pose questions on curated images or give agents preprocessed inputs, leaving out the raw multi-sensor archives where such analysis begins.","url":"https://arxiv.org/abs/2611.12934","relationship":"Source-stated"},{"text":"EarthRecord uses satellite data from 56 sites across four decades, leaving cloud masking and cross-sensor harmonization to the agent.","excerpt":"a benchmark built on the raw record of 56 sites: four decades of Landsat, MODIS, VIIRS, and Sentinel-2 data, with cloud masking and cross-sensor harmonization left to the agent.","url":"https://arxiv.org/abs/2611.12934","relationship":"Source-stated"},{"text":"The benchmark poses 238 land-change questions across ten analysis tasks and scores them against withheld references.","excerpt":"It poses 238 land-change questions across ten analysis tasks, scored against withheld references from global products, agreement among methods, and visual interpretation.","url":"https://arxiv.org/abs/2611.12934","relationship":"Source-stated"},{"text":"The authors report that leading agents complete most analyses end to end but falter when answers require mapping from imagery.","excerpt":"leading agents complete most analyses end to end but falter when an answer requires mapping from imagery.","url":"https://arxiv.org/abs/2611.12934","relationship":"Source-stated"},{"text":"In a control that alters one satellite’s data while the ground remains unchanged, half of the right answers follow the alteration.","excerpt":"when one satellite's data are altered, half of the right answers follow the alteration.","url":"https://arxiv.org/abs/2611.12934","relationship":"Source-stated"},{"text":"The authors interpret the results as showing that answers move from prior knowledge to the imagery, but not yet to real land change.","excerpt":"the record moves answers from prior knowledge to the pixels but not yet to the real change on the ground","url":"https://arxiv.org/abs/2611.12934","relationship":"Source-stated"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2611.12934","checkedAt":"2026-11-12T04:50:09.848Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"What would enable agents to move from answers grounded in satellite pixels to answers about real change on the ground?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"arXiv cs.AI: EarthRecord and multimodal agents analyzing satellite imagery","summary":"The authors of EarthRecord argue that existing benchmarks have not measured whether multimodal agents can analyze raw satellite archives to identify land change. Earlier benchmarks use curated images or preprocessed inputs, while EarthRecord gives agents data from 56 sites across four decades of Landsat, MODIS, VIIRS, and Sentinel-2 imagery, leaving cloud masking and cross-sensor harmonization to the agents. It poses 238 questions across ten analysis tasks and scores answers against withheld references. Controls remove the record or alter one satellite’s data while nothing changes on the ground. The authors report that leading agents complete most analyses end to end but falter when an answer requires mapping from imagery. In the alteration control, half of the right answers follow the changed data. The authors interpret this as the record moving answers from prior knowledge to the pixels, but not yet to real change on the ground. They present EarthRecord as a platform for measuring and advancing agents’ land-change analysis.","question":"What would enable agents to move from answers grounded in satellite pixels to answers about real change on the ground?","claims":[{"text":"The authors argue that existing benchmarks do not measure agents’ ability to analyze raw multi-sensor satellite archives for land change.","excerpt":"Existing benchmarks fall short: they pose questions on curated images or give agents preprocessed inputs, leaving out the raw multi-sensor archives where such analysis begins.","support":"source-stated"},{"text":"EarthRecord uses satellite data from 56 sites across four decades, leaving cloud masking and cross-sensor harmonization to the agent.","excerpt":"a benchmark built on the raw record of 56 sites: four decades of Landsat, MODIS, VIIRS, and Sentinel-2 data, with cloud masking and cross-sensor harmonization left to the agent.","support":"source-stated"},{"text":"The benchmark poses 238 land-change questions across ten analysis tasks and scores them against withheld references.","excerpt":"It poses 238 land-change questions across ten analysis tasks, scored against withheld references from global products, agreement among methods, and visual interpretation.","support":"source-stated"},{"text":"The authors report that leading agents complete most analyses end to end but falter when answers require mapping from imagery.","excerpt":"leading agents complete most analyses end to end but falter when an answer requires mapping from imagery.","support":"source-stated"},{"text":"In a control that alters one satellite’s data while the ground remains unchanged, half of the right answers follow the alteration.","excerpt":"when one satellite's data are altered, half of the right answers follow the alteration.","support":"source-stated"},{"text":"The authors interpret the results as showing that answers move from prior knowledge to the imagery, but not yet to real land change.","excerpt":"the record moves answers from prior knowledge to the pixels but not yet to the real change on the ground","support":"source-stated"}],"limitations":["The supplied evidence is the paper’s abstract; it does not include the full evaluation details or results by agent and task.","The abstract does not specify how the benchmark’s withheld references compare in reliability or coverage."],"mediaId":"70b6e0e37bdecc39c45b084b82b4f1d8b668649a0825bb396264ebfe352198b1","correctionReason":null,"id":"0a037142368ab87f2c362bbf96c7259f3b65e256ff34bf0ad2614f0d81327216","publicationId":"07d1ded1-486a-4481-89f5-195d16a465ef","version":1,"sourceUrl":"https://arxiv.org/abs/2611.12934","sourceTitle":"EarthRecord: From Pixels to Land Change with Multimodal Agents over Four Decades of Satellite Imagery","publisher":"arXiv cs.AI","publishedAt":null,"originalPublished":{"value":"2026-11-06","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-11-11T05:09:40.185Z","createdAt":"2026-11-12T04:18:46.482Z","historical":true,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"70b6e0e37bdecc39c45b084b82b4f1d8b668649a0825bb396264ebfe352198b1","kind":"article","url":"https://arxiv.org/abs/2611.12934","sourceUrl":"https://arxiv.org/abs/2611.12934","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-11-12T04:17:56.714Z","method":"acquired_source","contentType":"text/html"},"recommendationReason":"Recommended by Jev for understanding this source; checked by Luna."}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2611.12934","checkedAt":"2026-11-12T04:50:09.848Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v6","mediaId":"70b6e0e37bdecc39c45b084b82b4f1d8b668649a0825bb396264ebfe352198b1","confidence":0.52,"provider":"typesafe","model":"jev-1.13.0","inputHash":"7bbca774f77da78505601c80124a26476ecb9bac647e660af70ca657658b7e4f","decidedAt":"2026-11-12T04:17:58.756Z"},"provenance":{"textHash":"d419becbf94da6bcd33955778b559388fe4e83decdde8f0db13447e0685b3b79","observationId":"ff728c5226104ea5f30628cb1e04325b10fdb28aba52b77b524603e54ddd1f45","retrievedAt":"2026-11-12T04:17:56.425Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"51948b46-3418-4401-8465-7a4ca540c12f"},"reviewer":{"model":"gpt-6-luna","runId":"6642609f-e16d-49cc-8786-df82454214ab"}},"audit":{"status":"passed","reviewedAt":"2026-11-12T04:18:46.482Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Zhang, Haoran","url":null,"kind":"unknown"},{"name":"Sun, Huizhen","url":null,"kind":"unknown"},{"name":"Wen, Hanbin","url":null,"kind":"unknown"},{"name":"Leng, Jie","url":null,"kind":"unknown"},{"name":"Zhu, Jiaxin","url":null,"kind":"unknown"},{"name":"Sun, Zhaohua","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Zhang, Haoran","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Zhang, Haoran","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.12934","locator":"head/meta[name=author|citation_author]"},"assertionId":"559c5a99e5edfbe2d3a697fb2cde0a5ffa9b4bb4c435e15217f985557fb9f783:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-12T04:50:09.848Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"559c5a99e5edfbe2d3a697fb2cde0a5ffa9b4bb4c435e15217f985557fb9f783","sourceUrl":"https://arxiv.org/abs/2611.12934","contentSha256":"4fda662ae5252ad97f16ea9920fdc5d1f680f26009b6db2388b1b94973fd959f","textHash":"d419becbf94da6bcd33955778b559388fe4e83decdde8f0db13447e0685b3b79","observedAt":"2026-11-12T04:50:09.848Z"}},{"actor":{"name":"Sun, Huizhen","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Sun, Huizhen","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.12934","locator":"head/meta[name=author|citation_author]"},"assertionId":"559c5a99e5edfbe2d3a697fb2cde0a5ffa9b4bb4c435e15217f985557fb9f783:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-12T04:50:09.848Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"559c5a99e5edfbe2d3a697fb2cde0a5ffa9b4bb4c435e15217f985557fb9f783","sourceUrl":"https://arxiv.org/abs/2611.12934","contentSha256":"4fda662ae5252ad97f16ea9920fdc5d1f680f26009b6db2388b1b94973fd959f","textHash":"d419becbf94da6bcd33955778b559388fe4e83decdde8f0db13447e0685b3b79","observedAt":"2026-11-12T04:50:09.848Z"}},{"actor":{"name":"Wen, Hanbin","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Wen, Hanbin","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.12934","locator":"head/meta[name=author|citation_author]"},"assertionId":"559c5a99e5edfbe2d3a697fb2cde0a5ffa9b4bb4c435e15217f985557fb9f783:participation:2","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-12T04:50:09.848Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"559c5a99e5edfbe2d3a697fb2cde0a5ffa9b4bb4c435e15217f985557fb9f783","sourceUrl":"https://arxiv.org/abs/2611.12934","contentSha256":"4fda662ae5252ad97f16ea9920fdc5d1f680f26009b6db2388b1b94973fd959f","textHash":"d419becbf94da6bcd33955778b559388fe4e83decdde8f0db13447e0685b3b79","observedAt":"2026-11-12T04:50:09.848Z"}},{"actor":{"name":"Leng, Jie","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Leng, Jie","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.12934","locator":"head/meta[name=author|citation_author]"},"assertionId":"559c5a99e5edfbe2d3a697fb2cde0a5ffa9b4bb4c435e15217f985557fb9f783:participation:3","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-12T04:50:09.848Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"559c5a99e5edfbe2d3a697fb2cde0a5ffa9b4bb4c435e15217f985557fb9f783","sourceUrl":"https://arxiv.org/abs/2611.12934","contentSha256":"4fda662ae5252ad97f16ea9920fdc5d1f680f26009b6db2388b1b94973fd959f","textHash":"d419becbf94da6bcd33955778b559388fe4e83decdde8f0db13447e0685b3b79","observedAt":"2026-11-12T04:50:09.848Z"}},{"actor":{"name":"Zhu, Jiaxin","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Zhu, Jiaxin","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.12934","locator":"head/meta[name=author|citation_author]"},"assertionId":"559c5a99e5edfbe2d3a697fb2cde0a5ffa9b4bb4c435e15217f985557fb9f783:participation:4","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-12T04:50:09.848Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"559c5a99e5edfbe2d3a697fb2cde0a5ffa9b4bb4c435e15217f985557fb9f783","sourceUrl":"https://arxiv.org/abs/2611.12934","contentSha256":"4fda662ae5252ad97f16ea9920fdc5d1f680f26009b6db2388b1b94973fd959f","textHash":"d419becbf94da6bcd33955778b559388fe4e83decdde8f0db13447e0685b3b79","observedAt":"2026-11-12T04:50:09.848Z"}},{"actor":{"name":"Sun, Zhaohua","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Sun, Zhaohua","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.12934","locator":"head/meta[name=author|citation_author]"},"assertionId":"559c5a99e5edfbe2d3a697fb2cde0a5ffa9b4bb4c435e15217f985557fb9f783:participation:5","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-12T04:50:09.848Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"559c5a99e5edfbe2d3a697fb2cde0a5ffa9b4bb4c435e15217f985557fb9f783","sourceUrl":"https://arxiv.org/abs/2611.12934","contentSha256":"4fda662ae5252ad97f16ea9920fdc5d1f680f26009b6db2388b1b94973fd959f","textHash":"d419becbf94da6bcd33955778b559388fe4e83decdde8f0db13447e0685b3b79","observedAt":"2026-11-12T04:50:09.848Z"}}],"participationCoverage":{"total":6,"returned":6,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2611.12934","snapshotId":"559c5a99e5edfbe2d3a697fb2cde0a5ffa9b4bb4c435e15217f985557fb9f783","contentSha256":"4fda662ae5252ad97f16ea9920fdc5d1f680f26009b6db2388b1b94973fd959f","textHash":"d419becbf94da6bcd33955778b559388fe4e83decdde8f0db13447e0685b3b79","observedAt":"2026-11-12T04:50:09.848Z","basis":"publisher-page-metadata","sourceLinks":45,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/0a037142368ab87f2c362bbf96c7259f3b65e256ff34bf0ad2614f0d81327216/source"}},"canonicalUrl":"https://escapemeridian.com/readings/arxiv-cs-ai-earthrecord-and-multimodal-agents-analyzing-0a037142","formats":{"html":"/readings/arxiv-cs-ai-earthrecord-and-multimodal-agents-analyzing-0a037142","markdown":"/readings/arxiv-cs-ai-earthrecord-and-multimodal-agents-analyzing-0a037142.md","json":"/readings/arxiv-cs-ai-earthrecord-and-multimodal-agents-analyzing-0a037142.json"},"attribution":"Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims.","maintainedGuideLinks":[],"readerContext":{"version":"reader-context-v1","attribution":"Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.","terms":[{"id":"term-multimodal-model","name":"Multimodal model","explanation":"A model that works with more than one kind of input or output, such as text, images and audio.","referenceUrl":"https://developers.google.com/machine-learning/glossary#multimodal-model"},{"id":"term-benchmark","name":"Benchmark","explanation":"A defined set of tests used to compare systems. Results depend on the tasks and measures selected.","referenceUrl":"https://arxiv.org/abs/2211.09110"}],"entities":[{"id":"arxiv","name":"arXiv","kind":"organization","context":"arXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.","referenceUrls":["https://arxiv.org/","https://info.arxiv.org/help/api/index.html"],"path":"/entities/arxiv","json":"/entities/arxiv.json","markdown":"/entities/arxiv.md"}]},"researchQuestions":[],"withdrawn":false,"references":{"schemaVersion":1,"target":{"type":"reading","id":"0a037142368ab87f2c362bbf96c7259f3b65e256ff34bf0ad2614f0d81327216","revision":"07d1ded1-486a-4481-89f5-195d16a465ef","revisionKind":"publication-id","title":"arXiv cs.AI: EarthRecord and multimodal agents analyzing satellite imagery","url":"https://escapemeridian.com/readings/arxiv-cs-ai-earthrecord-and-multimodal-agents-analyzing-0a037142"},"source":{"type":"source","id":"https://arxiv.org/abs/2611.12934","revision":"d419becbf94da6bcd33955778b559388fe4e83decdde8f0db13447e0685b3b79","revisionKind":"sha256-source-text","title":"EarthRecord: From Pixels to Land Change with Multimodal Agents over Four Decades of Satellite Imagery","url":"https://arxiv.org/abs/2611.12934"},"status":"available","items":[],"next":null,"scope":"Explicit citations in public readings and public file attachments to this reading. Shared topics are not citations. Other artifact types need an explicit reference adapter."},"revisions":[{"id":"0a037142368ab87f2c362bbf96c7259f3b65e256ff34bf0ad2614f0d81327216","title":"arXiv cs.AI: EarthRecord and multimodal agents analyzing satellite imagery","argument":"The authors of EarthRecord argue that existing benchmarks have not measured whether multimodal agents can analyze raw satellite archives to identify land change. Earlier benchmarks use curated images or preprocessed inputs, while EarthRecord gives agents data from 56 sites across four decades of Landsat, MODIS, VIIRS, and Sentinel-2 imagery, leaving cloud masking and cross-sensor harmonization to the agents. It poses 238 questions across ten analysis tasks and scores answers against withheld references. Controls remove the record or alter one satellite’s data while nothing changes on the ground. The authors report that leading agents complete most analyses end to end but falter when an answer requires mapping from imagery. In the alteration control, half of the right answers follow the changed data. The authors interpret this as the record moving answers from prior knowledge to the pixels, but not yet to real change on the ground. They present EarthRecord as a platform for measuring and advancing agents’ land-change analysis.","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The supplied evidence is the paper’s abstract; it does not include the full evaluation details or results by agent and task. The abstract does not specify how the benchmark’s withheld references compare in reliability or coverage.","speaker":{"name":"Zhang, Haoran, Sun, Huizhen, Wen, Hanbin, Leng, Jie, Zhu, Jiaxin, Sun, Zhaohua","roleAtSource":"Article author"},"source":{"title":"EarthRecord: From Pixels to Land Change with Multimodal Agents over Four Decades of Satellite Imagery","publisher":"arXiv cs.AI","format":"Article","url":"https://arxiv.org/abs/2611.12934","publishedAt":"2026-11-06","dateBasis":"Date basis not provided."},"evidence":[{"text":"The authors argue that existing benchmarks do not measure agents’ ability to analyze raw multi-sensor satellite archives for land change.","excerpt":"Existing benchmarks fall short: they pose questions on curated images or give agents preprocessed inputs, leaving out the raw multi-sensor archives where such analysis begins.","url":"https://arxiv.org/abs/2611.12934","relationship":"Source-stated"},{"text":"EarthRecord uses satellite data from 56 sites across four decades, leaving cloud masking and cross-sensor harmonization to the agent.","excerpt":"a benchmark built on the raw record of 56 sites: four decades of Landsat, MODIS, VIIRS, and Sentinel-2 data, with cloud masking and cross-sensor harmonization left to the agent.","url":"https://arxiv.org/abs/2611.12934","relationship":"Source-stated"},{"text":"The benchmark poses 238 land-change questions across ten analysis tasks and scores them against withheld references.","excerpt":"It poses 238 land-change questions across ten analysis tasks, scored against withheld references from global products, agreement among methods, and visual interpretation.","url":"https://arxiv.org/abs/2611.12934","relationship":"Source-stated"},{"text":"The authors report that leading agents complete most analyses end to end but falter when answers require mapping from imagery.","excerpt":"leading agents complete most analyses end to end but falter when an answer requires mapping from imagery.","url":"https://arxiv.org/abs/2611.12934","relationship":"Source-stated"},{"text":"In a control that alters one satellite’s data while the ground remains unchanged, half of the right answers follow the alteration.","excerpt":"when one satellite's data are altered, half of the right answers follow the alteration.","url":"https://arxiv.org/abs/2611.12934","relationship":"Source-stated"},{"text":"The authors interpret the results as showing that answers move from prior knowledge to the imagery, but not yet to real land change.","excerpt":"the record moves answers from prior knowledge to the pixels but not yet to the real change on the ground","url":"https://arxiv.org/abs/2611.12934","relationship":"Source-stated"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2611.12934","checkedAt":"2026-11-12T04:50:09.848Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"What would enable agents to move from answers grounded in satellite pixels to answers about real change on the ground?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"arXiv cs.AI: EarthRecord and multimodal agents analyzing satellite imagery","summary":"The authors of EarthRecord argue that existing benchmarks have not measured whether multimodal agents can analyze raw satellite archives to identify land change. Earlier benchmarks use curated images or preprocessed inputs, while EarthRecord gives agents data from 56 sites across four decades of Landsat, MODIS, VIIRS, and Sentinel-2 imagery, leaving cloud masking and cross-sensor harmonization to the agents. It poses 238 questions across ten analysis tasks and scores answers against withheld references. Controls remove the record or alter one satellite’s data while nothing changes on the ground. The authors report that leading agents complete most analyses end to end but falter when an answer requires mapping from imagery. In the alteration control, half of the right answers follow the changed data. The authors interpret this as the record moving answers from prior knowledge to the pixels, but not yet to real change on the ground. They present EarthRecord as a platform for measuring and advancing agents’ land-change analysis.","question":"What would enable agents to move from answers grounded in satellite pixels to answers about real change on the ground?","claims":[{"text":"The authors argue that existing benchmarks do not measure agents’ ability to analyze raw multi-sensor satellite archives for land change.","excerpt":"Existing benchmarks fall short: they pose questions on curated images or give agents preprocessed inputs, leaving out the raw multi-sensor archives where such analysis begins.","support":"source-stated"},{"text":"EarthRecord uses satellite data from 56 sites across four decades, leaving cloud masking and cross-sensor harmonization to the agent.","excerpt":"a benchmark built on the raw record of 56 sites: four decades of Landsat, MODIS, VIIRS, and Sentinel-2 data, with cloud masking and cross-sensor harmonization left to the agent.","support":"source-stated"},{"text":"The benchmark poses 238 land-change questions across ten analysis tasks and scores them against withheld references.","excerpt":"It poses 238 land-change questions across ten analysis tasks, scored against withheld references from global products, agreement among methods, and visual interpretation.","support":"source-stated"},{"text":"The authors report that leading agents complete most analyses end to end but falter when answers require mapping from imagery.","excerpt":"leading agents complete most analyses end to end but falter when an answer requires mapping from imagery.","support":"source-stated"},{"text":"In a control that alters one satellite’s data while the ground remains unchanged, half of the right answers follow the alteration.","excerpt":"when one satellite's data are altered, half of the right answers follow the alteration.","support":"source-stated"},{"text":"The authors interpret the results as showing that answers move from prior knowledge to the imagery, but not yet to real land change.","excerpt":"the record moves answers from prior knowledge to the pixels but not yet to the real change on the ground","support":"source-stated"}],"limitations":["The supplied evidence is the paper’s abstract; it does not include the full evaluation details or results by agent and task.","The abstract does not specify how the benchmark’s withheld references compare in reliability or coverage."],"mediaId":"70b6e0e37bdecc39c45b084b82b4f1d8b668649a0825bb396264ebfe352198b1","correctionReason":null,"id":"0a037142368ab87f2c362bbf96c7259f3b65e256ff34bf0ad2614f0d81327216","publicationId":"07d1ded1-486a-4481-89f5-195d16a465ef","version":1,"sourceUrl":"https://arxiv.org/abs/2611.12934","sourceTitle":"EarthRecord: From Pixels to Land Change with Multimodal Agents over Four Decades of Satellite Imagery","publisher":"arXiv cs.AI","publishedAt":null,"originalPublished":{"value":"2026-11-06","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-11-11T05:09:40.185Z","createdAt":"2026-11-12T04:18:46.482Z","historical":true,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"70b6e0e37bdecc39c45b084b82b4f1d8b668649a0825bb396264ebfe352198b1","kind":"article","url":"https://arxiv.org/abs/2611.12934","sourceUrl":"https://arxiv.org/abs/2611.12934","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-11-12T04:17:56.714Z","method":"acquired_source","contentType":"text/html"},"recommendationReason":"Recommended by Jev for understanding this source; checked by Luna."}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2611.12934","checkedAt":"2026-11-12T04:50:09.848Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v6","mediaId":"70b6e0e37bdecc39c45b084b82b4f1d8b668649a0825bb396264ebfe352198b1","confidence":0.52,"provider":"typesafe","model":"jev-1.13.0","inputHash":"7bbca774f77da78505601c80124a26476ecb9bac647e660af70ca657658b7e4f","decidedAt":"2026-11-12T04:17:58.756Z"},"provenance":{"textHash":"d419becbf94da6bcd33955778b559388fe4e83decdde8f0db13447e0685b3b79","observationId":"ff728c5226104ea5f30628cb1e04325b10fdb28aba52b77b524603e54ddd1f45","retrievedAt":"2026-11-12T04:17:56.425Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"51948b46-3418-4401-8465-7a4ca540c12f"},"reviewer":{"model":"gpt-6-luna","runId":"6642609f-e16d-49cc-8786-df82454214ab"}},"audit":{"status":"passed","reviewedAt":"2026-11-12T04:18:46.482Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Zhang, Haoran","url":null,"kind":"unknown"},{"name":"Sun, Huizhen","url":null,"kind":"unknown"},{"name":"Wen, Hanbin","url":null,"kind":"unknown"},{"name":"Leng, Jie","url":null,"kind":"unknown"},{"name":"Zhu, Jiaxin","url":null,"kind":"unknown"},{"name":"Sun, Zhaohua","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Zhang, Haoran","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Zhang, Haoran","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.12934","locator":"head/meta[name=author|citation_author]"},"assertionId":"559c5a99e5edfbe2d3a697fb2cde0a5ffa9b4bb4c435e15217f985557fb9f783:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-12T04:50:09.848Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"559c5a99e5edfbe2d3a697fb2cde0a5ffa9b4bb4c435e15217f985557fb9f783","sourceUrl":"https://arxiv.org/abs/2611.12934","contentSha256":"4fda662ae5252ad97f16ea9920fdc5d1f680f26009b6db2388b1b94973fd959f","textHash":"d419becbf94da6bcd33955778b559388fe4e83decdde8f0db13447e0685b3b79","observedAt":"2026-11-12T04:50:09.848Z"}},{"actor":{"name":"Sun, Huizhen","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Sun, Huizhen","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.12934","locator":"head/meta[name=author|citation_author]"},"assertionId":"559c5a99e5edfbe2d3a697fb2cde0a5ffa9b4bb4c435e15217f985557fb9f783:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-12T04:50:09.848Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"559c5a99e5edfbe2d3a697fb2cde0a5ffa9b4bb4c435e15217f985557fb9f783","sourceUrl":"https://arxiv.org/abs/2611.12934","contentSha256":"4fda662ae5252ad97f16ea9920fdc5d1f680f26009b6db2388b1b94973fd959f","textHash":"d419becbf94da6bcd33955778b559388fe4e83decdde8f0db13447e0685b3b79","observedAt":"2026-11-12T04:50:09.848Z"}},{"actor":{"name":"Wen, Hanbin","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Wen, Hanbin","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.12934","locator":"head/meta[name=author|citation_author]"},"assertionId":"559c5a99e5edfbe2d3a697fb2cde0a5ffa9b4bb4c435e15217f985557fb9f783:participation:2","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-12T04:50:09.848Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"559c5a99e5edfbe2d3a697fb2cde0a5ffa9b4bb4c435e15217f985557fb9f783","sourceUrl":"https://arxiv.org/abs/2611.12934","contentSha256":"4fda662ae5252ad97f16ea9920fdc5d1f680f26009b6db2388b1b94973fd959f","textHash":"d419becbf94da6bcd33955778b559388fe4e83decdde8f0db13447e0685b3b79","observedAt":"2026-11-12T04:50:09.848Z"}},{"actor":{"name":"Leng, Jie","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Leng, Jie","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.12934","locator":"head/meta[name=author|citation_author]"},"assertionId":"559c5a99e5edfbe2d3a697fb2cde0a5ffa9b4bb4c435e15217f985557fb9f783:participation:3","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-12T04:50:09.848Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"559c5a99e5edfbe2d3a697fb2cde0a5ffa9b4bb4c435e15217f985557fb9f783","sourceUrl":"https://arxiv.org/abs/2611.12934","contentSha256":"4fda662ae5252ad97f16ea9920fdc5d1f680f26009b6db2388b1b94973fd959f","textHash":"d419becbf94da6bcd33955778b559388fe4e83decdde8f0db13447e0685b3b79","observedAt":"2026-11-12T04:50:09.848Z"}},{"actor":{"name":"Zhu, Jiaxin","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Zhu, Jiaxin","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.12934","locator":"head/meta[name=author|citation_author]"},"assertionId":"559c5a99e5edfbe2d3a697fb2cde0a5ffa9b4bb4c435e15217f985557fb9f783:participation:4","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-12T04:50:09.848Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"559c5a99e5edfbe2d3a697fb2cde0a5ffa9b4bb4c435e15217f985557fb9f783","sourceUrl":"https://arxiv.org/abs/2611.12934","contentSha256":"4fda662ae5252ad97f16ea9920fdc5d1f680f26009b6db2388b1b94973fd959f","textHash":"d419becbf94da6bcd33955778b559388fe4e83decdde8f0db13447e0685b3b79","observedAt":"2026-11-12T04:50:09.848Z"}},{"actor":{"name":"Sun, Zhaohua","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Sun, Zhaohua","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.12934","locator":"head/meta[name=author|citation_author]"},"assertionId":"559c5a99e5edfbe2d3a697fb2cde0a5ffa9b4bb4c435e15217f985557fb9f783:participation:5","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-12T04:50:09.848Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"559c5a99e5edfbe2d3a697fb2cde0a5ffa9b4bb4c435e15217f985557fb9f783","sourceUrl":"https://arxiv.org/abs/2611.12934","contentSha256":"4fda662ae5252ad97f16ea9920fdc5d1f680f26009b6db2388b1b94973fd959f","textHash":"d419becbf94da6bcd33955778b559388fe4e83decdde8f0db13447e0685b3b79","observedAt":"2026-11-12T04:50:09.848Z"}}],"participationCoverage":{"total":6,"returned":6,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2611.12934","snapshotId":"559c5a99e5edfbe2d3a697fb2cde0a5ffa9b4bb4c435e15217f985557fb9f783","contentSha256":"4fda662ae5252ad97f16ea9920fdc5d1f680f26009b6db2388b1b94973fd959f","textHash":"d419becbf94da6bcd33955778b559388fe4e83decdde8f0db13447e0685b3b79","observedAt":"2026-11-12T04:50:09.848Z","basis":"publisher-page-metadata","sourceLinks":45,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/0a037142368ab87f2c362bbf96c7259f3b65e256ff34bf0ad2614f0d81327216/source"}}}],"audits":[{"publicationId":"07d1ded1-486a-4481-89f5-195d16a465ef","verdict":"passed","findings":[{"dimension":"attribution","status":"pass","explanation":"The benchmark, its rationale, its results and its interpretation are consistently attributed to the EarthRecord authors on arXiv. Phrases such as 'argue', 'report' and 'interpret' keep the source's assertions distinct from established fact or Escape Meridian opinion.","excerpt":"We introduce EarthRecord, a benchmark built on the raw record of 56 sites"},{"dimension":"summary","status":"pass","explanation":"The summary faithfully conveys the gap the authors identify, the benchmark design, both controls, the main findings and the authors' interpretation. It preserves the paper's sharp point that half of the right answers follow altered data. It omits the count of twelve agents evaluated and the 'closed book' label; neither omission distorts meaning. Terms such as cloud masking, cross-sensor harmonization and multimodal agents are kept exactly and suit term-library tooltips. The title and question follow from the evidence.","excerpt":"Evaluations of twelve multimodal agents reveal that leading agents complete most analyses end to end but falter when an answer requires mapping from imagery."},{"dimension":"date","status":"pass","explanation":"The originalPublished date of 2026-11-06 matches the v1 submission date in the source. The feed date of 2026-11-11 is correctly not used as the publication date.","excerpt":"[Submitted on 6 Nov 2026]"},{"dimension":"support","status":"pass","explanation":"Each claim's excerpt is verbatim and supports the claim, and the numbers 56, 238, ten and four decades are accurate. The 'half of the right answers' result is stated with its condition intact. The question follows from the authors' 'not yet' finding without asserting an answer. The limitations correctly note that the evidence is only the abstract. No unsupported benefits or costs are implied.","excerpt":"when one satellite's data are altered, half of the right answers follow the alteration."},{"dimension":"media","status":"pass","explanation":"The selected media is the verified direct arXiv abstract page for this paper, which is the most direct access to the source. No other media candidates were found.","excerpt":null}],"createdAt":"2026-11-12T04:18:46.482Z","model":"anthropic/claude-opus-5.5"}],"correctionUrl":"https://escapemeridian.com/corrections/"}