Agricultural Yield, Soil & Climate Dataset (West Africa)

This is a synthetic dataset generated from high-quality expert-labelled seed data. All records are algorithmically derived — statistical distributions, inter-field correlations, and annotation characteristics faithfully replicate real-world patterns from the source data, while ensuring no real individual, organisation, or transaction can be identified or reconstructed.

The Agricultural Yield, Soil & Climate Dataset covers 300K+ geo-referenced agricultural plot observations across Nigeria, Ghana, and Senegal — three of West Africa's most agriculturally significant economies. Each record links plot-level yield outcomes (kg/ha) for major crops (maize, cassava, sorghum, groundnut, cowpea) to matched soil-chemical profiles, seasonal climate variables, and satellite-derived vegetation indices (NDVI, EVI) sourced from Sentinel-2 and MODIS imagery.

Data originates from a combination of national agricultural extension services, geo-tagged field surveys conducted by partner agronomists, and publicly available satellite archives post-processed into 250 m grid cells. Yield records are validated against sub-national production statistics from FAO and national bureaux of statistics, with anomaly flags for suspected mis-reporting or environmental shocks (drought, flood, pest outbreak).

The dataset supports the full spectrum of agricultural AI applications — from field-level yield prediction and input optimisation to district-scale early-warning systems and parametric agri-insurance product design. It is structured to interoperate with GeoJSON boundary files and common remote-sensing Python libraries (rasterio, xarray, geopandas).