At first glance the seemingly desolate expanse of the Sahel and Sahara deserts looks a bit greener, but the detailed satellite image attached to the computer deep learning reveals a different picture.
In fact, the 1.8 billion trees in some parts of the West African Sahara and the Sahel Desert, and the so-called sub-humid region, are already such a number of rewards that overturn previous assumptions about such habitats.
“We were very surprised that so many (so many) trees are grown in the Sahara Desert,” lead author Martin Brande told AFP.
Branden, an assistant professor of geography at the University of Copenhagen, added, “Certainly there is a large area without any trees, but there are still areas of tree-lined density, and even here and some trees are growing between the sandy hills.” .
The survey provides data to researchers and conservationists that can help guide efforts to combat deforestation and help measure carbon deposits on land more accurately.
“Data like this is very important for establishing a baseline for maintenance, restoration, climate change and so on,” said Jesse Meyer, a programmer at NASA’s Goddard Space Flight Center who works on the research.
“In one or two or ten years, this study could be repeated to see if repetition and deforestation efforts are effective,” he said in a NASA press release.
Finding and counting trees was no easy task. In areas with abundant shrubs, thick clumps of growth can be seen relatively clearly in satellite images even at relatively low altitudes and are easily recognizable from open ground.
But where they are more scattered, satellite images may be too low resolution to select individual trees or small groups.
High-resolution images are now available, but even then problems remain: counting individual trees, especially in large areas of the region, is an almost impossible task.
Brand and his team came up with a solution that combined learning images with very high resolution satellite images – essentially training a computer program to work for them.
But that doesn’t mean they will just sit and wait for the results.
Before the Learning Danda study program could be operationalized, it had to be trained, an intensive process that individually counted and labeled 90,000 trees for the brand. It took him a year.
“The level of detail is very high and the model needs to know what all the different types of trees in different landscapes look like,” he said.
“I didn’t add the wrong classification and more training when I saw the wrongly classified trees.”
Setting the defense baseline
It was worth the effort, he said, that millions of people would count years of work in just hours.
“Other studies are based on conjecture and extrapolation. Here we look at each tree directly and count, it is the first evaluation from wall to wall.”
The survey was published in the journal Wednesday Nature, Covers an area of 1.3 million square kilometers (approximately 500,000 square miles) and analyzes more than 11,000 images.
The technology suggests that “it will soon be possible to map the location and size of each tree worldwide,” said Niall P., of the Department of Plant and Environmental Sciences at New Mexico State University. Henan and Julius Anchang wrote in a review. Research.
And accurate information about vegetation in deserts and other arid zones is “fundamental to our understanding of world-class ecology, biogeography and the biochemical cycles of carbon, water and other nutrients,” they wrote in the review. Nature.
Better information can help determine how much carbon is stored at these sites, which are not typically included in climate models, the brand said.
But, it is too early to say whether accurately calculating the life of these trees will affect how we understand climate change and its acceleration.
He hopes the technology will now be used elsewhere to map more previously hidden trees in the world’s arid regions of 65 million square kilometers (25 million square miles).
ન્સી Agency France-Press
.
