Datasets

Continental analysis-ready satellite data platform providing cropland extent, Water Observations from Space, coastline change, fractional cover and GeoMAD products for all of Africa. Data is served as Cloud-Optimized GeoTIFF via a STAC API (explorer.digitalearth.africa/stac), OGC Web Services (ows.digitalearth.africa), an AWS S3 open bucket and a JupyterHub Sandbox. Products are derived from Landsat and Sentinel and are continuously updated.

Near-real-time remote-sensing database of evapotranspiration, biomass production and water productivity covering Africa and the Near East. Data is provided from continental to field scale (300m, 100m and 30m) as GeoTIFF and through an API, refreshed on a dekadal (10-day) cadence. Free and open access from FAO.

Machine-learning building footprint dataset first released for Africa and now covering 1.8 billion detections, each with a polygon, confidence score and Plus Code. Africa is served both as the original continent-wide product and as per-country GeoParquet/GeoJSON extracts on HDX. Data downloads as CSV by S2 cell, via gsutil, Google Earth Engine or the web explorer; latest v3 is May 2023.

[ingest-scout] Open-access harmonized subnational crop statistics for 33 Sub-Saharan African countries from 1980–2022, containing 574,204 records across 94 crop types at Admin-1 and Admin-2 levels (area, production, yield). Published in Scientific Data 2025 (doi:10.1038/s41597-025-05001-z), GitHub repo confirmed live at v1.2. Critical food-security and agriculture dataset not yet in the registry.

Imported from a community submission, needs review.

Text and speech corpus for three Kenyan languages, Swahili, Dholuo and Luhya, containing 4,442 texts (5.6 million words) and 1,152 speech files (177 hours). It also ships derived NLP sets: POS-tagged Dholuo/Luhya, 7,537 Swahili question-answer pairs and 13,400 translated sentences. Downloadable from Harvard Dataverse; released 2022.

Largest news-domain machine translation benchmark for African languages, covering 21 languages with English or French as source. It contains 142,909 parallel sentences in parquet with train, dev and test splits, hosted on HuggingFace. Licensed CC BY-NC 4.0.

Official downloadable economic time-series from South Africa's national statistics office, including CPI, GDP, PPI, employment and trade indicators. Series are provided as Excel and ASCII flat files and refreshed on monthly and quarterly release schedules. Free public access with no login.

Geocoded inventory of 98,745 public-sector health facilities assembled from national master facility lists across 50 sub-Saharan African countries and islands. Records include names, facility types, admin locations and coordinates, distributed as CSV via figshare and Harvard Dataverse. Published in Nature Scientific Data (2019) under CC BY 4.0.

Network of automatic weather stations across sub-Saharan Africa recording rainfall, temperature, humidity, wind and radiation at sub-daily intervals. Observations are accessed through an API and web portal under a data-sharing agreement, free for education and research. The station data itself is real but sits behind account credentials rather than an open download.

[ingest-scout] 49 GB document-level pretraining corpus spanning 16 African languages (Yoruba, Amharic, Egyptian Arabic, Hausa, and others) plus English/French/Arabic/Portuguese, built by auditing mC4 and crawling verified news sources; fetched HuggingFace page and confirmed 239 monthly downloads, Apache 2.0 license, collection updated June 2025, and use as the training corpus for AfriTeVa V2. High-quality complement to Wycord's existing NLP datasets.

Imported from a community submission, needs review.

First continent-scale predictive soil property maps for Africa at 30m resolution, covering organic carbon, pH, texture class, sand, clay, stone content and nutrients at 0-20cm and 20-50cm depths. Each property is published as Cloud-Optimized GeoTIFF (mean and model error layers) on Zenodo under CC BY 4.0. Released 2021 by iSDA with ISRIC.