1.8 billion individual trees found in West Africa's drylands
Add Axios as your preferred source to
see more of our stories on Google.
Satellite image of trees in Senegal (coordinates: 14.723, -16.303). Photo: ©2020 Maxar Technologies
The deserts and drylands of West Africa appear treeless, but researchers have found more than 1.8 billion individual trees and shrubs there, according to a new paper.
Why it matters: Non-forest trees support flora and fauna, provide sources of food and shelter for animals and people, and help moderate climate change by absorbing carbon.
- But their exact role in the planet's cycling of carbon, water and other nutrients isn't well-understood, in part because their numbers are unknown. Satellite images typically used to measure and monitor forests haven't been high enough resolution to pick up isolated trees.
- Methods in the new study could also be used to manage tree planting efforts, like the Great Green Wall of Africa, an attempt to fight desertification that is behind schedule, in part because of challenges in monitoring the project, per The Guardian.
How they did it: Martin Brandt, of the University of Copenhagen, and his colleagues analyzed 11,128 satellite images with a 0.5 meter resolution using deep learning algorithms to map single trees in a roughly 500,000-square-mile area spanning the West African Sahara desert, the semi-arid Sahel region and a sub-humid area to the south.
- They report finding more than 1.8 billion individual trees with a median crown size of 12 square meters.
- The finding "challenges prevailing narratives about dryland desertification, and even the desert shows a surprisingly high tree density," researchers write in the journal Nature this week.
- "We provide here the technique and evidence that it is possible to map and measure each single tree," says Brandt.
What's next: The next step is to analyze larger areas with an aim of creating a global database of all non-forest trees, ultimately including those in other ecosystems, says Brandt.
Yes, but: Training the deep learning algorithm required manually labeling 90,000 tree canopies on sample images.
- "This approach becomes untenable for work on a global scale, and more-automated (unsupervised) methods for extracting information from satellite imagery would be necessary," Niall Hanan and Julius Anchang of New Mexico State University wrote in an accompanying article.
