Use the canonical title MMTA: Benchmarking Multimodal Temporal Analysis with Time Series, Text, and Vision in the dataset card and all Croissant descriptions/citations. Document the paper-reported audit of 100 Terra descriptions by two independent annotators: 95% agreement, Cohen kappa 0.687 before adjudication and 93% accuracy after adjudication, with Appendix B.3 Table 6 as the source. State the sampling scope and planned-correction status. Data records, assets, licenses and author order are unchanged.
Clarify component-specific licensing in dataset metadata
Use license: other with license_name: mmta-component-specific. Make the root LICENSE a component-specific licensing overview and preserve the original CC BY 4.0 text unchanged in licenses/CC-BY-4.0.txt. Align the README and collection Croissant with the licensing overview. Dataset records, assets and source-domain Croissant descriptors are unchanged.
Publish attributed Croissant metadata and component license documentation
Add seven ordered authors, affiliations, equal-contribution and corresponding-author notes, and the final paper citation. Publish curated collection and source-domain Croissant metadata with RAI and layered licenses, including explicit Esri terms for satellite imagery. Update the dataset card and Terra image-source notice. Data manifests and assets are unchanged.