To understand how much carbon our earth can store, NASA is using AI and supercomputers to count a baffling number of trees.
Recently, scientists from NASA’s Goddard Space Flight Center in Maryland partnered with an international team of researchers to count the world’s trees and keep track of their growth over time. To do this they are using high-resolution satellite images.
Using one of the fastest supercomputers on earth (Blue Waters at the University of Illinois), the team performed a - as Good News Network put it - “deep-learning” analysis on terrain images from across great areas of west Africa. Not only could they detect and count trees previous satellites had failed to see, but they could also commence estimating the carbon-storing potential of those trees as well.
Earlier, much of the world’s attempt to assess a big number of trees has focused on well-forested sections. To get a fuller picture, the NASA team desired to focus on isolated trees in semi-arid and drylands in west Africa.
“These dry areas are white on maps—they are basically masked out because normal satellites just don’t see the trees. They see a forest, but if the tree is isolated, they can’t see it. Now we’re on the way to filling these white spots on the maps. And that’s quite exciting,” says Martin Brandt, lead author and assistant professor of geography at the University of Copenhagen, in a statement.
To train the machine-learning algorithms, Brandt personally marked nearly 90 000 trees stretching over different terrain. By doing so the software gets different shapes and shadows to learn the difference. The team has also trained the algorithms to identify both individual and small clusters of trees in different types of terrain. A job that may have taken a trained eye several years to complete only took a few weeks for artificial intelligence with the right training in place.
From this point forward, with assessments like this, conservationists will have an easier time tracking deforestation around the world. The data from up above will be compared to older data. This way scientists can assess whether conservation efforts are working or not.
Lastly, by improving the ability of scientists and researchers to count trees where they couldn’t count before, and to appraise the carbon storage of those trees, climate scientists have the help they need to make global measurements of carbon storage in the future. In a world where storing our excess carbon is becoming more crucial, this will be a vital tool.