Dynamic World

Near real-time 10m global land use land cover data

Free & Open Available in Google Earth Engine

mid-2015 to present

Understanding how land is being used on Earth is fundamental to addressing three pressing issues now facing us: climate change, food insecurity and loss of biodiversity. Dynamic World provides an open, continuous feed of 10-meter land use land cover (LULC) data for 9 classes across the globe to advance land use and land change research.

Dynamic World is produced in near real time from EU Copernicus / ESA Sentinel-2 acquisitions. Dynamic World was produced using a purpose-built convolutional neural net model, inferencing in near real-time on AI Platform.

The continuous nature of the product's outputs enables refinement, extension, and even redefinition of the LULC classification. Users developing analyses with our product are not constrained to a prescriptive LULC taxonomy nor a fixed annualized time-scale and are enabled to extend the estimated probabilities to entirely new labels or reduce the data to an arbitrarily precise time range.

Dynamic World is a joint initiative between the World Resources Institute and Google, using training data collected by the National Geographic Society.

Check it out! Dynamic World Explorer and in the Google Earth Engine Data Catalog.
Try it for yourself! Introduction to Dynamic World in 3 Parts

A paper describing the methodology is in press.
Learn about the custom open source model at https://github.com/google/dynamicworld.

Near Real-time

Continuously-updating LULC data allows study of interannual processes in production of derivative maps

Per-pixel Probabilities

Probabilities across 9 LULC classes for every cloud‑free S2 TOA image back to 2015 and ongoing

Derivative Map Products

Earth Engine users can reduce the ImageCollection or train classifiers to create derivatives specific to a region

Explore Example Scripts

Dynamic World Explorer (Full Screen)
Script: https://earthoutreach.users.earthengine.app/view/dynamicworld
Author: Valerie Pasquarella, Google Visiting Researcher from Boston University

Dynamic World Near Realtime
Script: https://earthoutreach.users.earthengine.app/view/dynamicworldnrt
Author: Valerie Pasquarella, Google Visiting Researcher from Boston University

Estimating Crop Production in Ukraine
Script: https://allredbw.users.earthengine.app/view/ukraine-production
Author: Brady Allred, Google

Dynamic World Annual Mode Composites
Script: https://davidtryse.users.earthengine.app/view/dynamic-world-composites
Author: David Tryse

Learn More

Google I/O 2022
Apply Responsible AI Principles when Building Remote Sensing Datasets, feat. Dynamic World with codelab
Speakers: Unni Nair, Darryl-Mason Robinson

Geo for Good Summit 2021
Dynamic World: https://earthoutreachonair.withgoogle.com/events/geoforgood21?talk=forest-dynamicworld
Speakers: Tanya Birch, Oliver Guinan, Valerie Pasquarella, Fred Stolle