With Dynamic World, Google wants to help researchers fight against climate change

The American giant has launched Dynamic World, a tool that provides near real-time global land cover data.

Enabling researchers and governments to develop useful solutions and minimize their effects on problems like climate change is Google’s goal with its new tool called Dynamic World. Powered by the Google Earth Engine and AI Platform user interface, it provides near real-time global land cover data. According to the company, it offers an unprecedented level of detail about what is on the land and how it is used (forest, agriculture, urban development…).

“With this information, people – like scientists and policy makers – can monitor and understand land and ecosystems so they can make more accurate estimates and effective plans to protect our planet in the future”Google explained in a statement.

Propose a better tool for researchers

For this tool, the Mountain View firm worked with the World Resources Institute (WRI), an American think tank specializing in environmental issues. This organization has identified nine types of land cover such as water, flooded vegetation, built-up areas or snow. Dynamic World uses Google’s artificial intelligence (AI) and cloud computing to detect combinations of these different types and draw conclusions about the likelihood of each being present in each pixel of a satellite image.

Google says more than 5,000 Dynamic World images are being produced using its AI model that analyzes Copernicus Sentinel-2 satellite images as they become available. The land cover data provided would therefore range from June 2015 to the day before yesterday. In other words, the information in Google’s tool would be more detailed and timely in any given day, week or month than existing datasets. “Current global land cover maps can take months to produce and typically only provide land cover data on a monthly or annual basis”explained Google.

The company also says that most existing datasets assign a single land cover type to an area of ​​land based on the most prominent feature in a satellite image combined with a cover expert’s determination. earthly. These sets can thus classify a satellite image of a city as “built up” when it also has trees or something else, in addition to many buildings. According to Google, Dynamic World will allow researchers to create their own maps based on the results of its machine learning model. They would be able to combine local information with data from the tool to produce a new map. The tool would also be useful for understanding longer-term trends in seasonal changes in the ecosystem.

Leave a Comment