Artificial intelligence reveals millions of Sahara trees

Newswise – If you think that the Sahara is covered only with golden dunes and scorched rocks, then you are not alone. Perhaps it is time to put an end to that notion. In an area of ​​West Africa, 30 times larger than Denmark, an international team led by researchers from the University of Copenhagen and NASA has counted more than 1.8 billion trees and shrubs. The 1.3 million km2 area covers the Sahara Desert, the western part known as the sub-humid regions of Western and Western Africa.

“We were very surprised to see that quite a few trees actually grow in the Sahara Desert, because until now, most people thought that none actually existed. We numbered hundreds of millions of trees in the desert alone. Doing so Will not done.” This may be possible without technology. In fact, I think it marks the beginning of a new scientific era, ”said Martin Brandt, assistant professor in the Department of Geology and Natural Resources Management of the University of Copenhagen, now the lead author of the study’s scientific article. Nature.

This work was achieved through a combination of detailed satellite imagery, and intensive learning provided by NASA – an advanced artificial intelligence method. Normal satellite imagery is unable to identify individual trees, they are literally invisible. In addition, a limited interest in counting trees outside the forested areas led to the prevailing view that there were almost no trees in this particular area. This is the first time that trees in large arid areas have been counted.

Role of Trees in Global Carbon Budget

According to Martin Brandt, new knowledge about trees in such arid regions is important for several reasons. For example, they represent an unknown factor when it comes to the global carbon budget:

Martin Brandt explains, “Trees outside of forested areas are usually not included in climate models, and we know little about their carbon stocks. They are basically a white spot on the map and in the global carbon cycle. There is an unknown component. ”

In addition, the new study may contribute to better understanding the importance of trees for biodiversity and ecosystems and for people living in these areas. In particular, enhanced knowledge about trees is also important for developing programs promoting agroforestry, which play a major environmental and socio-economic role in arid regions.

“Thus, we are also interested in using satellites to determine tree species, as tree types are important with respect to their value to local populations who use wood resources as part of their livelihoods Trees and their fruits are consumed by both livestock and humans. And when preserved in fields, trees have a positive impact on crop yields as they improve the balance of water and nutrients, “Explains Rasmus Fensholt, professor in the Department of Geology and Natural Resources Management.

High capacity technology

The research was conducted in collaboration with the Department of Computer Science of the University of Copenhagen, where researchers developed intensive learning algorithms that made it possible to count trees over such a large area.

Researchers show intensive learning models of what a tree looks like: they do this by feeding thousands of images of different trees. Depending on the tree shape recognition, the model can then automatically identify trees over large areas and thousands of images. The model only requires hours that would take thousands of humans many years to achieve.

“When it comes to global changes and ultimately contributing to global climate goals, this technology has great potential,” says Christian Egel, professor and co-author of the department. Of computer science.

The next step is to expand the count to a much larger area in Africa. And over the long term, it aims to create a global database of all trees growing outside forest areas.



  • Researchers counted 1.8 billion trees and shrubs with crowns larger than 3 m2. Thus, the actual number of trees in the area is even greater.
  • Deep learning can be referred to as an advanced artificial intelligence method where an algorithm is trained to recognize specific patterns in large amounts of data. The algorithm used in this research was trained using approximately 90,000 images of different trees in different scenarios.
  • The scientific article of this study has been published in the famous journal Nature.
  • The research was conducted by researchers at the University of Copenhagen; NASA Goddard Space Flight Center, USA; HCI Group, University of Bremen, Germany; Universe Paul Sabatier, France; Rusticism Conseil, France; Center de needle Ecology, Senegal; Geosciences Enforcement Toulouse (GET), France; Ecole Normale Superior, France; Universe Catholic de Louvain, Belgium.
  • Research is supported by, among others, the AXA Research Fund (Postdoctoral Program); Independent Research Fund Denmark – Sapere Aude; The Willem Foundation and the European Research Council (ERC) under the European Union’s Horizon 2020 program.


Leave a Reply

Your email address will not be published.