Dynamic World
One of the most commonly used foundational datasets for geospatial analyses is land use land cover (LULC) data, which are used in climate, energy, water, agriculture, conservation, and many other sectors. Dynamic World V1 is built by training a deep learning model on densely annotated training labels for 9 land cover classes, and is generated using Google Earth Engine and AI Platform.
Dynamic World is producing land cover probabilities per pixel for the Sentinel-2 1C: Multispectral TOA mission.
Nature Scientific Data - data descriptor
Brown, C.F., Brumby, S.P., Guzder-Williams, B. et al. Dynamic World, Near real-time global 10 m land use land cover mapping. Sci Data 9, 251 (2022). https://doi.org/10.1038/s41597-022-01307-4
Explore Dynamic World for examples and an EE App with a before/after slider. Read Google's announcement "Land cover data just got real-time."
Use the tool below to view Dynamic World over custom date range with single-scene predictions on individual Sentinel-2 images.
This map was generated by running the Dynamic World neural network model over single Sentinel-2 TOA scenes. In the view above, you can select to view a mode composite for a desired date range. The map above visualizes the Top-1 (highest probability) label using the palette in the legend and the Top-1 confidence via hill-shade.
The nature of the data allows us to better reflect the true nature of landscapes, which at any scale can be a combination of different land cover classes.
The data are produced for the Dynamic World Project by Google, in partnership with the World Resources Institute and National Geographic Society. Dynamic World data are available under a CC-BY-4.0 Attribution license.
Resources: