Dynamic World

Dynamic World, built by Google and the World Resources Institute, runs a deep-learning classifier over every Sentinel-2 image as it is captured to produce 10m, 9-class land cover with a 2-5 day revisit — the first global land cover product updated in near-real-time rather than as an annual or multi-year release. Probabilistic per-pixel outputs let users threshold confidence rather than accept a single hard classification.

Its currency is the differentiator: for rapidly changing catchments — active construction, deforestation, seasonal cropping — Dynamic World tracks land-cover change as it happens, where WorldCover and CCI Land Cover only offer periodic snapshots. Runs natively in Google Earth Engine.