# Data Hub — full documentation > A catalog and documentation hub for climate action datasets. This is the expanded companion to https://cgiar-climate-data-hub.github.io/llms.txt: docs and catalog metadata in one file. Each dataset's complete metadata record is at /catalog/.json; cite datasets per the "How to cite" section on their record page. ## Datasets ### Monthly and Seasonal Precipitation from CHIRPS v3 (Africa) - URL: https://cgiar-climate-data-hub.github.io/catalog/africa-precipitation-monthly-seasonal/ - Metadata (JSON): https://cgiar-climate-data-hub.github.io/catalog/africa-precipitation-monthly-seasonal.json - STAC collection: https://digital-atlas.s3.amazonaws.com/cdh/stac/africa-precipitation-monthly-seasonal/collection.json - Resource type: dataset - License: CC-BY-4.0 - Temporal coverage: 1981-01 to ongoing - Geography: africa - Keywords: precipitation, rainfall, CHIRPS, PTOT, monthly, seasonal, Africa, ENSO, drought - Data: https://digital-atlas.s3.amazonaws.com/domain=climate/type=observational/source=chirps-chirts-era5/region=africa/processing=monthly/ · https://digital-atlas.s3.amazonaws.com/domain=climate/type=observational/source=chirps-chirts-era5/region=africa/processing=seasonal/ Total precipitation (PTOT) aggregated from CHIRPS v3.0 and delivered as per-pixel monthly Cloud Optimized GeoTIFFs and per-year rolling 3-month (seasonal) sums over Africa, 1981 to present. This is the rainfall backbone of the KE-ENSO Explorer's seasonal/ENSO analysis: the twelve seasonal windows (JFM, FMA, ..., DJF) are the tri-month sums the notebook compares against ENSO phase and against the drought (SPEI), vegetation (NDVI) and flood layers. A derived aggregation product, not a redistribution of the CHIRPS daily archive. #### Variables - PTOT (mm) — Total precipitation over the period (calendar month, or 3-month seasonal sum). #### Dimensions - time: 1981-01 … 2026-04 (544 values) — step P1M — Month of the monthly totals (YYYY-MM); the {time} token of the monthly asset. - season: JFM … DJF (12 values) — Rolling 3-month window named by the initials of its months (JFM = Jan-Mar ... DJF = Dec-Feb). Each value covers three months; NDJ and DJF span the year boundary and are labelled by the year in which they end. - year: 1981 … 2026 (46 values) — step P1Y — Year in which a seasonal window ends; the {year} token of the seasonal asset. #### Intended use - Domains: climate - Intended: Seasonal and ENSO-phase rainfall analysis and anomaly context in the KE-ENSO Explorer. - Intended: Rainfall driver for interpreting the drought (SPEI), vegetation (NDVI) and flood layers. - Not recommended: Daily or sub-monthly rainfall analysis. — Only monthly totals and 3-month sums are published here. — use instead: The CHIRPS v3 daily archive (catalog record chirps-v3-daily). Cite: Steward, P., Funk, C., Peterson, P., Landsfeld, M. (2026) Monthly and seasonal precipitation from CHIRPS v3 (Africa), Africa Agriculture Adaptation Atlas. Alliance of Bioversity International and CIAT. Accessed through the CGIAR Climate Action Data Hub, https://cgiar-climate-data-hub.github.io/catalog/africa-precipitation-monthly-seasonal/. ### CHIRPS v3 Daily Precipitation - URL: https://cgiar-climate-data-hub.github.io/catalog/chirps-v3-daily/ - Metadata (JSON): https://cgiar-climate-data-hub.github.io/catalog/chirps-v3-daily.json - STAC collection: https://digital-atlas.s3.amazonaws.com/cdh/stac/chirps-v3-daily/collection.json - Resource type: dataset - License: CC-BY-4.0 - Access: public - DOI: 10.1038/sdata.2015.66 - Temporal coverage: 1981-01-01 to ongoing - Geography: world - Keywords: precipitation, rainfall, CHIRPS, satellite, gridded, daily, drought monitoring - Data: https://data.chc.ucsb.edu/products/CHIRPS-3.0/ CHIRPS (Climate Hazards center InfraRed Precipitation with Stations), version 3, is a quasi-global daily precipitation dataset produced by the Climate Hazards Center (UC Santa Barbara), blending 0.05° satellite infrared cold-cloud-duration imagery with in-situ station observations. It spans 1981 to near-present, covering 50°S–50°N at all longitudes, and is widely used for drought monitoring, hydrological modeling, and trend/seasonal analysis. Values represent daily accumulated precipitation in millimeters. #### Variables - precipitation (mm) — Daily accumulated precipitation #### Dimensions - time: 1981-01-01 — step P1D — Daily time step of the precipitation record #### Intended use - Domains: climate, hydrology Cite: Funk, C., Peterson, P., Landsfeld, M., Pedreros, D., Verdin, J., Shukla, S., Husak, G., Rowland, J., Harrison, L., Hoell, A., Michaelsen, J. (2015) The climate hazards infrared precipitation with stations—a new environmental record for monitoring extremes. Version v3.0. Scientific Data. https://doi.org/10.1038/sdata.2015.66 Accessed through the CGIAR Climate Action Data Hub, https://cgiar-climate-data-hub.github.io/catalog/chirps-v3-daily/. ### CHIRTS-ERA5 Daily - URL: https://cgiar-climate-data-hub.github.io/catalog/chirts-era5-daily/ - Metadata (JSON): https://cgiar-climate-data-hub.github.io/catalog/chirts-era5-daily.json - STAC collection: https://digital-atlas.s3.amazonaws.com/cdh/stac/chirts-era5-daily/collection.json - Resource type: dataset - License: CC-BY-4.0 - Access: public - DOI: 10.15780/G2F08J - Temporal coverage: 1983-01-01 to ongoing - Geography: world - Keywords: temperature, daily, era5, station-blended, global, climate-hazards-center - Data: https://data.chc.ucsb.edu/experimental/CHIRTS-ERA5/ CHIRTS-ER5 daily provides global, high-resolution (0.05°) daily maximum and minimum 2-meter air temperature estimates from 1983 to present. It blends ERA5 reanalysis with station observations using the CHIRTS algorithm, producing a consistent long-term record suitable for climate trend analysis, heat-stress monitoring, and agricultural modeling. #### Intended use - Domains: climate Cite: Verdin, A., Funk, C., Peterson, P., Landsfeld, M., Tuholske, C., Grace, K. (2025) CHIRTS-ERA5 Data Repository. UCSB Climate Hazards Center. https://doi.org/10.15780/G2F08J Accessed through the CGIAR Climate Action Data Hub, https://cgiar-climate-data-hub.github.io/catalog/chirts-era5-daily/. ### Gridded livestock density for 2020 (GLW4) - URL: https://cgiar-climate-data-hub.github.io/catalog/glw4-2020/ - Metadata (JSON): https://cgiar-climate-data-hub.github.io/catalog/glw4-2020.json - STAC collection: https://digital-atlas.s3.amazonaws.com/cdh/stac/glw4-2020/collection.json - Resource type: dataset - License: CC-BY-4.0 - Temporal coverage: 2020 - Geography: world - Commodities: buffalo, cattle, chickens, goats, swine, sheep - Keywords: gridded, modelled, exposure, animal husbandry - Data: https://digital-atlas.s3.amazonaws.com/cdh/data/glw4-2020/glw4-2020.zarr · https://digital-atlas.s3.amazonaws.com/cdh/data/glw4-2020/cog/ - Link: FAO Livestock Distributions — https://www.fao.org/livestock-systems/global-distributions/en/ - Link: FAOSTAT Crop and Livestock Production — https://www.fao.org/faostat/en/#data/QCL Gridded Livestock of the World, version 4 (GLW 4) maps modelled densities of six livestock species: buffalo, cattle, sheep, goats, pigs, and chickens. This data is for the 2020 reference year. Densities are estimated with a Random Forest model that downscales harmonized subnational census counts onto a regular grid, so the values represent modelled distributions rather than direct observations. All of the values are number of animals per km^2 #### Variables - Livestock Density ({head}/km-1) — Number of livestock per km2 #### Dimensions - species: buffalo, cattle, chicken, goat, pig, sheep — The livestock species for which the data is provided #### Intended use - Domains: agricultural-production - Intended: global and regional analysis of livestock distribution - Intended: livestock exposure layers for hazard, disease, and emissions assessment - Intended: spatial targeting at national and sub-national scale - Not recommended: exact livestock counts — values are spatially modeled estimates derived from census and other data — use instead: official livestock census counts where available Cite: Food and Agriculture Organization of the United Nations (2024) GLW 4: Gridded Livestock Density (Global - 2020 - 10 km). FAO Agro-Informatic Data Catalog. https://data.apps.fao.org/catalog/dataset/9d1e149b-d63f-4213-978b-317a8eb42d02 Accessed through the CGIAR Climate Action Data Hub, https://cgiar-climate-data-hub.github.io/catalog/glw4-2020/. ### MAPSPAM 2020 - URL: https://cgiar-climate-data-hub.github.io/catalog/spam2020/ - Metadata (JSON): https://cgiar-climate-data-hub.github.io/catalog/spam2020.json - STAC collection: https://digital-atlas.s3.amazonaws.com/cdh/stac/spam2020/collection.json - Resource type: dataset - License: CC-BY-SA-4.0 - DOI: 10.7910/DVN/SWPENT - Temporal coverage: 2020 - Geography: world - Commodities: wheat, rice, maize, barley, millets, pearl-millet, sorghum, cereals, potato, sweet-potatoes, yams, cassava, root-vegetables, common-bean, chickpeas, cowpeas, pigeon-pea, lentils, pulses, soybeans, groundnuts, coconuts, oil-palms, sunflower, rapeseed, sesame, oil-crops, sugarcane, sugarbeet, cotton, fibre-crops, coffee, robusta-coffee, cocoa, tea, tobacco, banana, plantains, citrus, tropical-fruits, temperate-fruit, tomatoes, onions, vegetables, rubber, crops - Keywords: food security, crop production, spatial distribution, gridded agriculture - Data: https://digital-atlas.s3.amazonaws.com/cdh/data/mapspam2020-v2r2/spam2020-v2r2.zarr · https://digital-atlas.s3.amazonaws.com/cdh/data/mapspam2020-v2r2/cog/ - Link: MAPSPAM project website — https://www.mapspam.info/about/ — Project page describing MAPSPAM methods and releases. Global Spatially-Disaggregated Crop Production Statistics Data for 2020 Version 2.0 Release 2, represented as a gridded crop production data cube. The record describes crop area, harvested area, production, and yield by crop and production technology at 5 arc-minute resolution. #### Variables - physical_area (ha) — Physical crop area allocated to each grid cell. - harvested_area (ha) — Harvested crop area allocated to each grid cell. - production (t) — Crop production allocated to each grid cell. — Absolute quantity; regional summaries should sum values across cells. - yield (t ha-1) — Crop yield for each grid cell. — Relative quantity; do not sum yield values across grid cells. Regional summaries should use a weighted mean with harvested_area as the weight. #### Dimensions - crop: whea … rest (46 values) — MAPSPAM crop code. Full labels are provided in the dimension codes asset. - technology: all, irrigated, rainfed — MAPSPAM production system code. Full labels are provided in the dimension codes asset. #### Intended use - Domains: agricultural-production - Intended: global and regional analysis of crop area, production, and yield patterns - Intended: food security and agricultural systems assessment - Intended: spatial targeting at national and sub-national scale - Not recommended: field-scale farm management — the grid is too coarse for field-scale operational decisions. — use instead: local survey or administrative production data. Cite: International Food Policy Research Institute (IFPRI) (2026) Global Spatially-Disaggregated Crop Production Statistics Data for 2020 Version 2.0 Release 2. Version v2r2. Harvard Dataverse. https://doi.org/10.7910/DVN/SWPENT Accessed through the CGIAR Climate Action Data Hub, https://cgiar-climate-data-hub.github.io/catalog/spam2020/. ## Documentation ### Licensing and attribution https://cgiar-climate-data-hub.github.io/wikis/licensing-and-attribution/ The Hub only publishes data that can be legally redistributed and reused. This page sets out what licences are accepted and how attribution works. #### Permitted licences - **CC-BY-4.0** — preferred for new contributions - **CC-BY-SA-4.0** — accepted; note the share-alike obligation propagates - **CC0 / Public Domain** — accepted - Custom or non-commercial licences are **not** accepted; talk to the maintainers if your source data is restricted. The licence in the metadata record is an SPDX identifier and applies to the distributions listed in the record, not to third-party source data, which retains its own terms. #### Attribution Reusers must cite the dataset as given in the record's `citation` block. Every dataset page renders a copy-ready citation. When a dataset is derived from an external source (for example FAO or IFPRI data), the original providers stay listed as `licensor`/`producer` contacts and must be credited alongside the Hub. #### DOIs Datasets processed and published by the Hub get a DOI minted on publication. Datasets that already carry a DOI from their original publisher (for example Harvard Dataverse deposits) keep it — the record's `doi` field always points at the citable identifier. ## Tutorials ### Getting started with the Climate Data Hub https://cgiar-climate-data-hub.github.io/tutorials/getting-started/ The Hub is one quality-assured home for climate data across CGIAR: every dataset is harmonised to the same metadata standard, openly licensed, and published in cloud-native formats. In practice that means you find data one way, judge it one way, and access it one way — whatever the dataset, whatever your tools. This guide walks that path once, without assuming you write code. #### Find a dataset Start at the [catalog](/catalog/). Three ways in: - **Search** — names, sources, variables, and keywords all match, so "livestock", "FAO", or "zarr" each work. - **Filters** — narrow by domain, geography, commodity, file format, license, or access. - **Spatial search** — draw a box on the map and only datasets whose coverage intersects it remain. Every card states the essentials up front: what the data is, who publishes it, its coverage and resolution, and its license. #### Read the record Click through to a dataset's page — say [GLW4 livestock density](/catalog/glw4-2020/) — and give it two minutes before committing: - **Quick facts** summarise coverage, resolution, and the time span. - **Schema** lists the variables and dimensions — what's actually in the cube. - **Appropriate use** notes, where present, tell you what the dataset should *not* be used for. Read them; they're written from experience. - **Access & use** lists every distribution with its format and a worked code example. #### Get the data into your tool - **Python or R** — every asset on a record page carries a copy-paste snippet tailored to its format, streaming just the window you need. No boilerplate to write; start from the record page, or see a full worked analysis in the [TLU notebook](/tutorials/tlu_glw4/). - **Desktop GIS** — Cloud-Optimized GeoTIFF URLs load directly in QGIS or ArcGIS as raster layers: copy the asset URL from the record page and add it as a data source. No download required. - **Just the files** — assets that are single files carry a Download button. - **AI assistants** — the Hub is machine-readable end to end; point your assistant at the [AI & agents page](/ai/) for skills and endpoints. #### Cite what you use Every record page has a copy-ready citation (APA, BibTeX, and more) and, where one exists, a DOI. Cite the original producers — for GLW4 that's FAO — alongside the Hub record; the prepared citations do this for you. #### Where next - [Tutorials](/tutorials/) — runnable, worked analyses. - [Wikis](/wikis/) — the standards and methods behind the data. - [FAQ](/faq/) — licensing, formats, and contribution questions. - Stuck? [Open an issue](/contribute/#report) and the team will pick it up. ### Tropical Livestock Units from GLW4 https://cgiar-climate-data-hub.github.io/tutorials/tlu_glw4/ Compute Tropical Livestock Units (TLU) by combining GLW4 species density layers with standard conversion factors, streaming directly from the Hub's Zarr store. ## FAQ ### What is the Climate Data Hub? A quality-assured catalog of climate and agriculture datasets across CGIAR Climate Action. Every dataset is harmonised to one [metadata standard](https://github.com/CGIAR-Climate-Data-Hub/cdh-metadata-standard), openly licensed, and distributed in cloud-native formats, so it can be discovered, cited, and used the same way. ### Do I need an account to access the data? No. All published data is openly accessible over HTTPS and S3 — stream it directly with tools like xarray or rioxarray, or download individual files. There is no registration, API key, or quota. ### How is the data licensed? CC-BY-4.0 unless a dataset page says otherwise (some sources are CC-BY-SA-4.0 or public domain). You can reuse it freely — including commercially — as long as you credit the original producers as given in the dataset's citation. See [licensing and attribution](/wikis/licensing-and-attribution/) for details. ### What formats does the Hub use? Cloud-optimised formats: Zarr for multi-dimensional data cubes and Cloud-Optimised GeoTIFF (COG) for rasters. That means you can read just the spatial or temporal window you need without downloading whole archives. ### How do I cite a dataset? Every dataset page has a copy-ready citation and, where one exists, a DOI. Cite the original producers (for example FAO or IFPRI) alongside the Hub record. ### Can I contribute a dataset? Yes. Harmonise your data to a supported format, author a metadata record against the CDH standard, and open a submission pull request. Hub maintainers and a domain reviewer check it before publication — see the [contribute page](/contribute/) for the full path. ### I found a problem with a dataset — where do I report it? Open an issue on the Hub's GitHub organisation, or email the point of contact listed on the dataset page. Records are versioned, so corrections are tracked and attributed. ## Use cases ### [Example — replace me] How [partner] used Hub data in [country] https://cgiar-climate-data-hub.github.io/in-use/example-story/ [Open with the decision or problem: who needed evidence, and why the usual path wasn't working.] #### What they did [Two or three paragraphs: which Hub datasets and tools were used, how they were combined, and what the analysis looked like. Code or maps welcome.] #### What changed [The outcome — a plan adopted, finance unlocked, a method reused. Be specific about scale and who benefits.]