ESA worldcereal global crop type detector

The ESA WorldCereal Crop Type Detector combines Sentinel-1 (SAR) and Sentinel-2 (optical) time series with meteorological inputs and in-situ observations to produce local to global 10 m crop-type maps, processed at scale via openEO within the Copernicus Data Space Ecosystem. It generates pixel-level embeddings using the Nasa Harvest’ “Presto” geospatial foundation model and classifies them with a lightweight CatBoost model, with options for users to train or adapt the downstream classifier for their own crops and regions. It exports results as cloud-optimized GeoTIFFs for easy integration into GIS workflows.