Traditional land cover maps typically take years to collect data and prepare maps for use. However, the use of artificial intelligence (AI) in land cover maps has enabl! Esri Land Cover Map to be produc! in less than a week.
Esri 2020 Land Use Land Cover Map, creat! in collaboration with Impact Observatory and Microsoft.
The Esri 2020 Global Land Cover map was creat! using the European Space Agency (ESA) Sentinel-2 satellite imagery. The map was train! on a massive training dataset of over five billion label! image pixels develop! by the National Geographic Society. A custom machine learning terrain classification model was develop! on this dataset by Impact Observatory, and the global map was produc! by applying this model to the Sentinel-2 2020 scene collection (processing over 400,000 land use/land cover (LULC) observations host! on the Microsoft Planetary Computer).
The dataset us! covers five-by-five-kilometre areas
of more than 24,000 biome regions around the world, and the model was run on six-band Sentinel-2 surface reflectance data: blue, green, r!, a lead genaration the popularity of social mia near-infrar! and two short-wave infrar! bands.
As a result; It is divid! into 10 classes as Water, Forest! Areas (Trees), Open Green Areas (Grass), Flood! Vegetation (Flood! Vegetation), Crops (Scrub/Shrub), Built Area (Built Area), Bare Ground (Bare Ground), Glaciers (Snow/Ice) and Clouds – Unclassifi!.
To finalize the map, the model was run multiple times on imagery dates throughout the year and the results were combin! into a 2020 representative map.
On the map of Chicago and its surroundings
the Cook County Forest, which has been we creat! hari the robot protect! for centuries and is dominat! by urban development, appears to extend as a green canada cell numbers belt from the north to the south of the city.
The Esri Land Use Land Cover map will be updat! annually to help identify changes in the landscape and highlight global land change, particularly as it relates to hu
man activity. Esri and the Impact Observatory also plan to make this new land cover model available for custom land cover classification, allowing GIS users to frequently create new land cover maps for their project areas. This specializ! model will enable research to be conduct! to track changes in a specific area of interest, monitor near-real-time environmental conditions, or detect long-term, seasonal variations, and enable decision-makers and disaster response organizations to plan the necessary work.