AI Shows that the Sahara has Millions of Trees

A Desert Full of Life

Satellite images of the Sahara desert show a dry expanse, the endless rolling dunes we know from movies. Except typical satellite images do not reveal individual trees; however, that doesn’t always mean they’re not there. Scientists from the University of Copenhagen and NASA trained artificial intelligence to recognize trees and have them take another look. There is a great deal of vegetation in the Western Sahara: an approximated 1.8 billion trees and shrubs.

Lead author Martin Brandt of the university’s Department of Geo-sciences and Natural Resource Management claims that he and his team were quite shocked to see that many trees grow in the Sahara Desert, given that up until now, most individuals assumed that virtually none existed. Brandt added that he and his team counted hundreds of millions of trees just in the desert. Accomplishing this wouldn’t have been achievable without this technology (artificial intelligence).

The research is published in the journal Nature.

As essential as trees are in combating climate change, researchers need to understand what trees there are and where. The research’s finding contributes to the international tree inventory.

The vegetation Brandt and his coworkers have found is in the Western Sahara, an area of roughly 1.3 million square kilometers that includes the desert, the Sahel, and the sub-humid areas of West Africa.

These trees and bushes were overlooked in previous inventories of carbon-processing of worldwide forests. Trees outside of forested regions are typically not considered in climate models, and incredibly little concerning their carbon stocks is known, stated Brandt. They are essentially a white spot on maps and an unidentified element in the worldwide carbon cycle Brandt added.

Along with being important climate-change information, the research can assist in the region’s strategic development in which the plant life grows because of a higher understanding of regional ecosystems.

Factoring in the White Spots

There’s been a belief that there’s hardly enough vegetation beyond forested regions to be worth counting. Consequently, the research represents the first time a considerable number of trees (In the hundreds of millions when not considering bushes) have been registered in the drylands region.

Members of the university’s Department of Computer Science taught a machine-learning module to identify trees by supplying countless images. This training left the AI capable of identifying trees in the tiny details of satellite images provided by NASA. The task took the Artificial intelligence only hours (it would require a human years to execute a comparable evaluation).

Co-author Christian Igel stated that this technology has massive potential when it involves documenting changes worldwide and eventually contributing towards worldwide climate goals. Igel added that it is an inspiration for him and his team to design this type of artificial intelligence.

Brandt said that without a doubt, he believes it marks the beginning of a new scientific period.

The scientists intend to fine-tune their AI to offer a more in-depth accounting of the trees it recognizes in satellite images.

The study’s senior author, Rasmus Fensholt, claims that he and his team are additionally considering using satellites to identify tree species, as tree types are important relative to their value to regional communities that use wood as part of their livelihoods. Both livestock and humans eat trees and their fruit, and when maintained in the fields, trees have a positive impact on crop returns since they enhance the balance of water and nutrients.

In the future, we can expect an expansion of the team’s tree hunt to a greater area of Africa. The long-term objective is to develop a more detailed and precise global database of trees that grow past the limits of forests.

Originally published by: bigthink.com