The “barren” deserts on earth actually have millions of trees, surprising...

At first glance, the seemingly barren expanses of the Sahel and Sahara deserts show little green, but detailed satellite images in combination with computer deep learning have given a different picture.

In fact, around 1.8 billion trees characterize parts of the West African Sahara and the Sahel zone as well as the so-called sub-humid zone, a previously untapped premium that turns earlier assumptions about such habitats upside down, say researchers.

“We were very surprised that quite a lot of trees are growing in the Sahara,” said lead author Martin Brandt to AFP.

“Sure there are large areas with no trees, but there are still areas with a high density of trees, and even among the sand dunes there are some trees growing here and there,” added Brandt, an assistant professor of geography at the University of Copenhagen.

The survey provides researchers and conservationists with data that could serve as a guide to tackling deforestation and measuring carbon storage on land more accurately.

“For conservation, restoration, climate change, etc., such data is very important to establish a baseline,” said Jesse Meyer, programmer at NASA’s Goddard Space Flight Center who worked on the research.

“In a year or two or ten years, the study could be repeated … to see whether or not the deforestation recovery and reduction efforts are effective,” he said in a NASA press release.

Finding and counting the trees was no easy task. In areas with many trees, thick clumps of growth appear relatively clear in satellite imagery, even at low resolution, and are easy to distinguish from bare land.

However, where they are more common, satellite imagery may be too low a resolution to select individual trees or even small groups.

Higher resolution images are now available, but even then problems remain: counting individual trees, especially over large areas, is an almost impossible task.

Brandt and his team have found a solution that combines satellite images with very high resolutions with deep learning. Essentially, this involves training a computer program to do the job for them.

But that didn’t mean they could just sit back and wait for the results.

Before the deep learning program could work, it had to be trained. This was a laborious process, during which Brandt counted and labeled almost 90,000 trees individually. It took a year.

“The level of detail is very high and the model needs to know what all types of trees look like in different landscapes,” he said.

“I did not accept any misclassification and did further training when I saw misclassified trees.”

Set a maintenance baseline

The effort was worth it, he said, and made it possible to calculate the work of millions of people in just a few hours.

“Other studies are based on estimates and extrapolations. Here we see and count every tree directly. It’s the first wall-to-wall rating. ”The customer review has been automatically translated from German.

The poll was published in the magazine on Wednesday naturecovered an area of ​​1.3 million square kilometers and included the analysis of more than 11,000 images.

The technology suggests that “with certain restrictions it will soon be possible to map the location and size of every tree worldwide,” wrote Niall P. Hanan and Julius Anchang from the Department of Plant and Environmental Sciences at New Mexico State University in an overview about research.

Precise information on vegetation in deserts and other arid areas is “of fundamental importance for our understanding of ecology, biogeography and the biogeochemical cycles of carbon, water and other nutrients on a global level”, states the overview commissioned by nature.

Better information could help determine how much carbon is being stored in these locations that aren’t normally included in climate models, Brandt said.

However, it is too early to say whether an accurate count of this tree life will affect how we understand climate change and how it is accelerating, he added.

He now hopes to apply the technique elsewhere to map previously hidden trees in the 65 million square kilometers of arid regions of the world.

© Agence France-Presse