Google has created a new online tool, eloquently called Dynamic World , a new dataset that could help researchers effectively track changes on the planet’s surface caused by human activity.

The service uses deep learning and satellite imagery to create a highly detailed, high-resolution land cover map that shows trees, buildings, bodies of water, and farmland.

The creation of land cover maps is usually time consuming and there are often large time gaps between the time satellite imagery is acquired and the data is published. In addition, most of these maps do not provide detailed information about objects located in a particular area – for example, a city is classified as “built-up” (designation of human-modified landscapes), even if there are large areas with parks.

“If the world wants to produce everything that is needed from the earth, protect the rest of nature and restore some of the lost resources, we need reliable monitoring of every hectare of the planet in almost real time,”

Craig Henson,

Vice President of the Food, Forests, Water and Oceans Program World Resources Institute

Dynamic World classifies the type of ground cover every 93 m2. It shows, as far as possible, that the areas are covered by one of nine types of coverage: water, flooded vegetation, built-up areas, trees, crops, bare ground, grass, shrubs, and snow/ice. Google details its system, developed in collaboration with the World Resources Institute, in an article published in Nature’s Scientific Data.

The Dynamic World model generates over 5,000 images per day, and the ground cover data is constantly updated. It enables researchers and policy makers to quickly assess the impact of various processes, including fires and storms, and respond more effectively to change.

Google has updated Timelapse — marvel in 3D, how our planet has changed in the last 37 years