{"success":true,"data":{"author":"LINKS Foundation and Politecnico di Torino (Barco, Arnaudo, Bragagnolo, Rossi, Garza)","bucketAccessType":"public","bucketName":"links-ads/tmb-s3","bucketPrefix":"","category":"Earth Observation & Geospatial","cloudProvider":"huggingface","createdAt":"2026-11-17T14:12:17.494Z","description":"TMB-S3 (Temporal Modelling for Burn Scars on Sentinel-3) is a burned-area segmentation benchmark built from 246 European wildfire activations of the Copernicus Emergency Management Service (CEMS) between 2016 and 2025, split into 770 bounding boxes of about 25 x 25 km. Each box holds a Sentinel-3 OLCI time series at 300 m (21 bands Oa01 to Oa21, top-of-atmosphere reflectance, uint16) with one pre-fire acquisition followed by post-fire acquisitions (median sequence length 9, range 2 to 36), a per-frame OLCI cloud mask, the CEMS burned-area delineation and analysed-area mask at 10 m (2500 x 2500), and ESA WorldCover 10 m land cover. Splits are made at event level (146/50/50 activations for train/validation/test) and recorded in manifest.parquet. Data are one Zarr v3 store per bounding box under data/, named <EMSR id>-<AOI>-<index>.zarr, with groups s3olci/, firemask/, validity/ and landcover/ carrying affine transforms and EPSG codes; open them with zarr or xarray. Code to reproduce the paper (resampling 10 m labels to the OLCI grid) is at github.com/links-ads/tmb-s3. Licensed CC BY 4.0 by LINKS Foundation; credit Copernicus Sentinel data, CEMS and ESA WorldCover, and cite Barco et al., AICT 2026.","displayTitle":"TMB-S3: Sentinel-3 OLCI Time Series for Burned Area Segmentation of 246 European Wildfires","huggingFaceUrl":"https://huggingface.co/datasets/links-ads/tmb-s3","id":"c3c21521-3fe8-59b5-bd6b-15b265dd9391","license":"CC-BY-4.0","licenseEvidenceUrl":"https://huggingface.co/datasets/links-ads/tmb-s3","objectCountEstimate":14601,"probeStatus":"ok","region":"us","schemaCard":{"accessHints":[{"language":"python-datasets","snippet":"from datasets import load_dataset\nds = load_dataset(\"links-ads/tmb-s3\")"},{"language":"python-hub","snippet":"from huggingface_hub import snapshot_download\nsnapshot_download(repo_id=\"links-ads/tmb-s3\", repo_type=\"dataset\")"}],"caveats":["HuggingFace metadata: author=links-ads, license:cc-by-4.0, size_categories:n<1K, format:parquet, downloads=898"],"format":"mixed","generatedAt":"2026-11-17T14:23:21.185Z","layout":{"example":"manifest.parquet","pattern":"{repoId}/{filename}"},"mountPath":"/mnt/archive","sampleHeaders":[{"bytes":"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","encoding":"hex","key":"manifest.parquet","truncatedAt":512}],"sampleObjects":["data/EMSR168-1-0.zarr/firemask/data/c/0/0","data/EMSR168-1-0.zarr/firemask/data/zarr.json","data/EMSR168-1-0.zarr/firemask/zarr.json","data/EMSR168-1-0.zarr/landcover/data/c/0/0","data/EMSR168-1-0.zarr/landcover/data/zarr.json"],"sidecars":[{"bytes":"---\nlicense: cc-by-4.0\npretty_name: TMB-S3\ntask_categories:\n  - image-segmentation\ntags:\n  - remote-sensing\n  - earth-observation\n  - wildfire\n  - burned-area\n  - sentinel-3\n  - olci\n  - time-series\n  - copernicus\nsize_categories:\n  - n<1K\nconfigs:\n  - config_name: manifest\n    data_files: manifest.parquet\n---\n\n# TMB-S3: Temporal Modelling for Burn Scars on Sentinel-3\n\nTMB-S3 is a burned-area segmentation dataset of **246 European wildfire activations\n(2016–2025)** of the Copernicus Emergency Management Service (CEMS), split into **770\nbounding boxes** of about 25 × 25 km. Each bounding box contains a Sentinel-3 OLCI time series\n(one pre-fire acquisition followed by the post-fire acquisitions), the CEMS burned-area\ndelineation, the CEMS analysed area and the ESA WorldCover land cover.\n\n- **Paper:** \"Temporal Modelling for Burn Scars on Sentinel-3\", Barco L., Arnaudo E., Bragagnolo A., Rossi C., Garza P., Application of Information and Communication Technologies (AICT) Conference 2026.  \n- **Code:** [GitHub](https://github.com/links-ads/tmb-s3)\n\n## Splits\n\n| | Train | Val | Test | Total |\n|---|---|---|---|---|\n| Events (EMSR activations) | 146 | 50 | 50 | 246 |\n| Bounding boxes | 517 | 135 | 118 | 770 |\n\nSplits are made at event level: all bounding boxes of an activation are in the same split.\nThe split of each bounding box is the `split` column of `manifest.parquet`.\n\n## Structure\n\n```\nREADME.md                   this card\nmanifest.parquet            one row per bounding box\ndata/\n└── EMSR457-1-1.zarr/       one zarr (v3) store per bounding box, named <EMSR id>-<AOI>-<index>\n    ├── zarr.json               attributes: bbox_id, event_time, delineation times, UTM CRS, geometry\n    ├── s3olci/                 Sentinel-3 OLCI, 300 m\n    │   ├── data                (T, 21, H, W) uint16, bands Oa01…Oa21, top-of-atmosphere reflectance\n    │   ├── time                (T,) int64, acquisition time (Unix seconds, UTC)\n    │   ├── prefix              (T,) uint8, 0 = pre-fire, 1 = post-fire\n    │   └── cloudmask           (T, H, W) uint8, 1 = cloud (from the OLCI quality flags)\n    ├── firemask/data           (2500, 2500) uint8, 10 m, CEMS burned area (1 = burned)\n    ├── validity/data           (2500, 2500) uint8, 10 m, CEMS analysed area (1 = labelled)\n    └── landcover/data          (2500, 2500) uint8, 10 m, ESA WorldCover class index (code / 10;\n                                e.g. 1 = tree cover, 8 = permanent water)\n```\n\nEach group has its affine `transform` and `crs_epsg` in its attributes. OLCI images are about\n83–90 × 83–90 pixels.\n\nFrames within a sequence are sorted by time. Every bounding box has exactly one pre-fire\nacquisition and at least one post-fire acquisition (sequence length T: median 9, range 2–36).\nThe last post-fire acquisition is the first clear acquisition after the CEMS delineation\n(median lag +1.5 days). Acquisitions with more than 20% invalid pixels were discarded.\n\n### Labels at the OLCI resolution\n\nIn the paper, the 10 m layers are resampled to the 300 m OLCI grid: a 300 m pixel is burned if\nany 10 m pixel inside it is burned, and pixels with no analysed 10 m pixel are ignored. The\ncode repository implements this (`ResampleMask`).\n\n## Usage\n\n```bash\nhf download <org>/tmb-s3 --repo-type dataset --local-dir data/tmb-s3\n```\n\n```python\nimport zarr\n\nroot = zarr.open(\"data/tmb-s3/data/EMSR457-1-1.zarr\", mode=\"r\")\nolci = root[\"s3olci/data\"][:]      # (T, 21, H, W)\nprefix = root[\"s3olci/prefix\"][:]  # 0 = pre-fire, 1 = post-fire\nburned = root[\"firemask/data\"][:]  # (2500, 2500), 10 m\n```\n\nWith the code repository, place the dataset in `data/tmb-s3/` and follow its README to\nreproduce the paper's results.\n\n## Sources and attribution\n\n- Contains modified Copernicus Sentinel data (2016–2025), Sentinel-3 OLCI.\n- Burned-area delineations: Copernicus Emergency Management Service (© European Union),\n  rapid mapping activations listed by EMSR id in `manifest.parquet`.\n- Land cover: ESA WorldCover 10 m (© ESA WorldCover project, CC BY 4.","encoding":"utf8","key":"README.md","name":"README.md","size":4211,"truncated":true}],"sourceObjectKey":"manifest.parquet","starterCell":"from datasets import load_dataset\nds = load_dataset(\"links-ads/tmb-s3\")\nprint(ds)","starterSamplePath":"data/EMSR168-1-0.zarr","totalObjects":14601},"summary":"TMB-S3 (Temporal Modelling for Burn Scars on Sentinel-3) is a burned-area segmentation benchmark built from 246 European wildfire activations of the Copernicus Emergency…","title":"TMB-S3: Sentinel-3 OLCI Time Series for Burned Area Segmentation of 246 European Wildfires","updatedAt":"2026-11-17T20:01:31.948Z","visibility":"public","websiteUrl":"https://github.com/links-ads/tmb-s3"}}