Lush vegetation is not the first picture that comes to mind when thinking about a desert. And yet, a new study utilising detailed satellite imagery and deep learning algorithms has found that the West African Sahara and Sahel deserts are actually home to approximately 1.8 billion trees, all told.
The researchers, who were themselves surprised by the results, explained that while there certainly are many areas where trees can be spotted only intermittently, there are also vast stretches of desert with a high tree density.
While current satellite imaging technology does offer resolutions high enough to spot trees not only in clumps, but also individually, counting them one by one over vast territories would be a near impossibility.
To overcome this issue, the research team deployed a deep learning algorithm that required Marin Brandt, lead author on the study, to individually count and label nearly 90,000 trees – a process which took him a full year to complete.
“The level of detail is very high and the model needs to know how all kind of different trees in different landscapes look,” he said. “I did not accept misclassifications and further added training when I saw wrongly classified trees”.
Whereas previous studies of similar type relied on estimations and extrapolations, this is the first one to directly count each individual tree within a vast area of 1.3 million square kilometres. Without deep learning, a similar study, reliant only on high resolution satellite imagery, would take millions of people and several years to conduct.
According to co-author on the study Jesse Meyer from NASA’s Goddard Space Flight Center, these findings could be used as a baseline for tracking deforestation, carbon storage on land, and a variety of other climate change-related efforts.
“In a year or two or ten, the study could be repeated… to see if efforts to revitalise and reduce deforestation are effective or not,” he said in a NASA press release.
With further improvements, the technology could soon be used to map the location and size of every tree worldwide, although likely with some limitations. Before that becomes a live option, however, Brandt plans to use the technique to count trees in other arid lands across the world that take up around 65 million square kilometres in total.