--- license: cc-by-4.0 task_categories: - other viewer: false --- # BioCube: A Multimodal Dataset for Biodiversity Research [![Paper](https://img.shields.io/badge/Paper-HuggingFace-blue)](https://huggingface.co/papers/2505.11568) [![Github](https://img.shields.io/badge/Code-Github-green)](https://github.com/BioDT/bfm-data) ![Alt text](img/BioCube-description.png "BioCube Data Description") BioCube is a multimodal dataset for ecological and biodiversity modelling. For more information, please visit our [github repository](https://github.com/BioDT/bfm-data) [Link to our paper](https://arxiv.org/abs/2505.11568) ## Dataset Structure Inside the `data` folder you can find the below modalities: | **Modality** | **Source** | **Variables** | **Path** | |--------------------------------|-------------------------------------------|-------------------------------------------------------------------------------------------------|-----------------------------------| | Surface Climate | Copernicus (ERA5) | 2 m temperature, 10 m wind (u / v), mean sea-level pressure | `/data/Copernicus/ERA5/.nc` | | Atmospheric Variables | Copernicus (ERA5) | Geopotential, temperature, humidity, wind (13 pressure levels: 50–1000 hPa) | `/data/Copernicus/ERA5/.nc` | | Single-Level Variables | Copernicus (ERA5) | Land-sea mask, surface geopotential, soil type | `/data/Copernicus/ERA5/.nc` | | Species Observations | iNaturalist, GBIF, Xeno-Canto | Images, audio, coordinates, timestamp, taxonomy | `/data/Species/species_dataset.parquet` | | Descriptions | Map of Life | Text descriptions (behavior, habitat) | `/data/Species/species_dataset.parquet` | | eDNA | BOLD Systems | DNA sequences (ATCG), taxonomic identifiers | `/data/Species/species_dataset.parquet` | | Distribution Trends | Living Planet Index | Annual species occurrence and population trends (1950–2020) | `/data/Species/species_dataset.parquet` | | Red List Index (RLI) | IUCN / Map of Life | Extinction risk index (0–1), categories: EX, EW, CR, EN, VU, NT, LC | `/data/RedList/Global_red_list_index.csv` | | NDVI | Copernicus Land (SPOT, PROBA-V) | Vegetation index values (−1 to 1), 10-day temporal resolution, 1 km spatial resolution | `/data/Land/Global_ndvi.csv` | | Agriculture Indicators | World Bank | Arable land, irrigated land, cropland area, forest cover, total land area | `/data/Agriculture/Global_combined_agriculture_data.csv` | | Forest Indicators | World Bank | forest cover, total land area | `/data/Forest/Global_forest_data.csv` | All the modalities are aligned on **[WGS84](https://nl.wikipedia.org/wiki/WGS_84)** coordinate grid with $0.25^o$ resolution, for the whole world and the time horizon of the observations spans from 2000-2020. *Only the NDVI variable is provided with $0.1^o$ resolution*. Below is a list of the variable names as they appear on the dataset. For more information for the Copernicus variables see [ERA 5 documentation](https://confluence.ecmwf.int/display/CKB/ERA5%3A+data+documentation). **REMARK** Many of the variables along with the Global come with another csv file that represents only the European region. ### Download the data You can either download manually the whole or parts of the dataset, or use the cli: ```bash git lfs install git clone https://huggingface.co/datasets/BioDT/BioCube ``` ### Work with BioCube BioCube can be used for variety of applications like: - Biodiversity modelling - Ecological processes modelling - Species distribution modelling - Habitat assesment - Weather & Climate modelling Also, BioCube is a trove of information and appropriate for **large-scale Foundation Model training**. The most recent model using our dataset is **[BioAnalyst: A Foundation Model for Biodiversity](https://arxiv.org/abs/2507.09080)** **IMPORTANT:** You can find more information of how to use BioCube to produce Batches for training ML models and the whole codebase used to produce it at [BioCube - Github](https://github.com/BioDT/bfm-data) ### Acknowledgments *This study has received funding from the European Union's Horizon Europe research and innovation programme under grant agreement No 101057437 (BioDT project, https://doi.org/10.3030/101057437). Views and opinions expressed are those of the author(s) only and do not necessarily reflect those of the European Union or the European Commission. Neither the European Union nor the European Commission can be held responsible for them. This publication is part of the project Biodiversity Foundation Model of the research programme Computing Time on National Computing Facilities that is (co-) funded by the Dutch Research Council (NWO). We acknowledge NWO for providing access to Snellius, hosted by SURF through the Computing Time on National Computer Facilities call for proposals. This work used the Dutch national e-infrastructure with the support of the SURF Cooperative using grant no. EINF-10148* ### Citation If you find our work useful, please consider citing us! ``` @misc{stasinos2025biocubemultimodaldatasetbiodiversity, title={BioCube: A Multimodal Dataset for Biodiversity Research}, author={Stylianos Stasinos and Martino Mensio and Elena Lazovik and Athanasios Trantas}, year={2025}, eprint={2505.11568}, archivePrefix={arXiv}, primaryClass={q-bio.QM}, url={https://arxiv.org/abs/2505.11568}, } ```