{"success":true,"data":{"author":"Xiao Xiang Zhu et al., Technical University of Munich (TUM) / DLR","bucketAccessType":"public","bucketName":"zhu-xlab/So2Sat-LCZ42","bucketPrefix":"","category":"Earth Observation & Geospatial","cloudProvider":"huggingface","createdAt":"2026-11-14T14:15:15.964Z","description":"Benchmark for classifying Local Climate Zones (LCZ), the 17-class urban and natural land scheme used in urban climate research, from co-registered Sentinel-1 SAR and Sentinel-2 multispectral image patches. Patches are 32 x 32 pixels with 8 Sentinel-1 channels and 10 Sentinel-2 bands, labelled by hand. This copy of version 4 pairs the culture-10 split with geolocation: training covers 42 cities worldwide, validation the western half and testing the eastern half of 10 further cities spanning 10 cultural zones. The listing (measured 2026-11-14) shows a v4/ folder with training.h5.gz (about 16 GB), validation.h5.gz and testing.h5.gz (about 1.1 GB each), matching *_geo.h5 files with UTM coordinates, EPSG codes and world-file parameters, and read_file.py; about 18.2 GB in all. Gunzip, then read the sen1, sen2 and one-hot label arrays with h5py and numpy; the zhu-xlab GitHub repository has a notebook to write GeoTIFFs. TUM mediaTUM publishes version 4 under CC-BY-SA-4.0; cite Zhu et al., IEEE GRSM 2020.","displayTitle":"So2Sat LCZ42 v4: Local Climate Zone Classification from Sentinel-1 and Sentinel-2","id":"efc9dca2-951f-5af3-99a1-54e4c966fc07","license":"CC-BY-SA-4.0","licenseEvidenceUrl":"https://mediatum.ub.tum.de/1836598","objectCountEstimate":10,"probeStatus":"ok","referenceUrl":"https://doi.org/10.1109/MGRS.2020.2964708","region":"us","schemaCard":{"accessHints":[{"language":"python-datasets","snippet":"from datasets import load_dataset\nds = load_dataset(\"zhu-xlab/So2Sat-LCZ42\")"},{"language":"python-hub","snippet":"from huggingface_hub import snapshot_download\nsnapshot_download(repo_id=\"zhu-xlab/So2Sat-LCZ42\", repo_type=\"dataset\")"}],"caveats":["HuggingFace metadata: author=zhu-xlab, license:cc-by-4.0, downloads=868"],"format":"mixed","generatedAt":"2026-11-14T14:15:21.117Z","layout":{"example":"v4/testing_geo.h5","pattern":"{repoId}/{filename}"},"mountPath":"/mnt/archive","sampleHeaders":[{"bytes":"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","encoding":"hex","key":"v4/testing_geo.h5","truncatedAt":512}],"sampleObjects":["v4/read_file.py","v4/testing.h5.gz","v4/testing_geo.h5","v4/training.h5.gz","v4/training_geo.h5"],"sidecars":[{"bytes":"---\nlicense: cc-by-4.0\ntask_categories:\n- image-classification\ntags:\n- climate\n- remote-sensing\n---\n\n# So2Sat-LCZ42\n\nA copy of the [Fourth version](https://mediatum.ub.tum.de/1836598) of the So2Sat-LCZ42 dataset, pairing the [second version](https://mediatum.ub.tum.de/1459256?show_id=1483140) (culture-10) with geolocation:\n\n\tTraining: \t42 cities around the world\n\tValidation:\twestern half of 10 other cities covering 10 cultural zones\n\tTesting:\teastern half of the 10 other cities\t\n\n## Description of the files\n\nFirst extract `[split].h5.gz` to `[split].h5` with `gunzip [split].h5.gz`.\n\ntraining.h5:\ttraining data containing SEN1, SEN2 patches and label\n\n\tsen1:\tN*32*32*8\t\n\tsen2:\tN*32*32*10\n\tlabel:\tN*17 (one-hot coding)\n\t\nvalidation.h5:  validation data containing similar SEN1, SEN2, and label\n\n\tsen1:  \tM*32*32*8 \t\n\tsen2:  \tM*32*32*10\t\n\tlabel: \tM*17 (one-hot coding)\n\t\ntesting.h5:\ttesting data containing SEN1 and SEN2 patches and label\n\n\tsen1:  \tL*32*32*8\n\tsen2:  \tL*32*32*10\n\tlabel:  L*17 (one-hot coding)\n\ntraining_geo.h5: geolocation for the training data\n\n\tcoord:  N*6 # UTM X, UTM Y, SEN1 row number, SEN1 col number, SEN2 row number, SEN2 col number\n\tepsg:   N*1 # EPSG code\n\ttfw:    N*6 # six parameters used to generate a TFW file\n  \nvalidation_geo.h5: geolocation for the validation data\n\n\tcoord:  M*6\n\tepsg:   M*1\n\ttfw:    M*6\n\ntesting_geo.h5: geolocation for the testing data\n\n\tcoord:  L*6\n\tepsg:   L*1\n\ttfw:    L*6\n\nread_file.py:\ta demo python script to read in the files, and visualize a pair of patches\n\t\tRequired python packages: h5py, numpy, and matplotlib.\n\nPlease visit https://github.com/zhu-xlab/So2Sat-LCZ42 for more details, including a demo notebook to generate geotiffs from h5 files.\n\n\n## Citation\n```bibtex\n@ARTICLE{Zhu2020So2Sat,\n  author={Zhu, Xiao Xiang and Hu, Jingliang and Qiu, Chunping and Shi, Yilei and Kang, Jian and Mou, Lichao and Bagheri, Hossein and Haberle, Matthias and Hua, Yuansheng and Huang, Rong and Hughes, Lloyd and Li, Hao and Sun, Yao and Zhang, Guichen and Han, Shiyao and Schmitt, Michael and Wang, Yuanyuan},\n  journal={IEEE Geoscience and Remote Sensing Magazine}, \n  title={So2Sat LCZ42: A Benchmark Data Set for the Classification of Global Local Climate Zones [Software and Data Sets]}, \n  year={2020},\n  volume={8},\n  number={3},\n  pages={76-89},\n  doi={10.1109/MGRS.2020.2964708}}\n```\n\n","encoding":"utf8","key":"README.md","name":"README.md","size":2353,"truncated":false},{"bytes":"# \\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\n\n# \n\n# ============================================\n\n# So2Sat LCZ42 dataset\n\n# ============================================\n\n# \n\n# Authors: Xiaoxiang Zhu, Jingliang Hu, Chunping Qiu, Yilei Shi, Jian Kang, Lichao Mou,\n\n# Hossein Bagheri, Matthias Haeberle, Yuansheng Hua, Rong Huang, Lloyd Hughes, Hao Li,\n\n# Yao Sun, Guichen Zhang, Shiyao Han, Michael Schmitt, Yuanyuan Wang\n\n# \n\n# Signal Processing in Earth Observation, Technical University of Munich, and\n\n# Remote Sensing Technology Institute, German Aerospace Center.\n\n# \n\n# This work is funded by European Research Council starting Grant:\n\n# So2Sat: Big Data for 4D Global Urban Mapping - 10^16 Bytes from Social Media to Earth\n\n# Observation Satellites\n\n# Project website: http://www.so2sat.eu/\n\n# \n\n# Team website: http://www.sipeo.bgu.tum.de/\n\n# \n\n# \\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\\*\n\n# Description of the files\n\nPlease extract [split].h5.gz to [split].h5 first using `gunzip [split].h5.gz`.\n\ntraining.h5:\ttraining data containing SEN1, SEN2 patches and label\nsen1:\tN*32*32*8\nsen2:\tN*32*32*10\nlabel:\tN\\*17 (one-hot coding)\n\nvalidation.h5:  validation data containing similar SEN1, SEN2, and label\nsen1:  \tM*32*32*8\nsen2:  \tM*32*32*10\nlabel: \tM\\*17 (one-hot coding)\n\ntesting.h5:\ttesting data containing only SEN1 and SEN2 patches,\nsen1:  \tL*32*32*8\nsen2:  \tL*32*32*10\nlabel:  L\\*17 (one-hot coding)\n\n\n\ntraining\\_geo.h5:  geolocation of the training patches\n\ncoord: coordinates of the patch center\n\ntfw: six parameters used to generate a TFW file\n\nepsg: epsg code, coordinate reference system of the patch\n\n\n\nvalidation\\_geo.h5:  geolocation of the training patches\n\ncoord: coordinates of the patch center\n\ntfw: six parameters used to generate a TFW file\n\nepsg: epsg code, coordinate reference system of the patch\n\n\n\ntesting\\_geo.h5:  geolocation of the training patches\n\ncoord: coordinates of the patch center\n\ntfw: six parameters used to generate a TFW file\n\nepsg: epsg code, coordinate reference system of the patch\n\n\n\nread\\_file.py:\ta demo python script to read in the files, and visualize a pair of patches\nRequired python packages: h5py, numpy, and matplotlib.\n\n\n\n# Description of the content of sen1\n\nSentinel-1 data bands (the 4th dimension of data):\n1st band: Real part of original VH complex signal\n2nd band: Imaginary part of original VH complex signal\n3rd band: Real part of original VV complex signal\n4th band: Imaginary part of original VV complex signal\n5th band: Intensity of lee filtered VH signal\n6th band: Intensity of lee filtered VV signal\n7th band: Real part of lee filtered PolSAR covariance matrix off-diagonal element\n8th band: Imaginary part of lee filtered PolSAR covariance matrix off-diagonal element\n\nPixel size: 10m by 10m\n\n\n\n# Description of the content of sen2\n\nSentinel-2 data bands (the 4th dimension of data):\n1st band: B2\n2nd band: B3\n3rd band: B4\n4th band: B5\n5th band: B6\n6th band: B7\n7th band: B8\n8th band: B8A\n9th band: B11 SWIR\n10th band: B12 SWIR\n\nPixel size: 10m by 10m\n\nDetails about the bands can be found: https://sentinels.copernicus.eu/web/sentinel/user-guides/sentinel-2-msi/overview\n\n","encoding":"utf8","key":"v4/README.md","name":"README.md","size":3422,"truncated":false}],"sourceObjectKey":"v4/testing_geo.h5","starterCell":"from huggingface_hub import hf_hub_download\npath = hf_hub_download(repo_id=\"zhu-xlab/So2Sat-LCZ42\", filename=\"v4/testing_geo.h5\", repo_type=\"dataset\")\nprint(path)","starterSamplePath":"v4/testing.h5.gz","totalObjects":10},"summary":"Benchmark for classifying Local Climate Zones (LCZ), the 17-class urban and natural land scheme used in urban climate research, from co-registered Sentinel-1 SAR and Sentinel-2…","title":"So2Sat LCZ42 v4: Local Climate Zone Classification from Sentinel-1 and Sentinel-2","updatedAt":"2026-11-14T17:26:55.309Z","visibility":"public","websiteUrl":"https://github.com/zhu-xlab/So2Sat-LCZ42"}}