The images used to analyze the data and map the results covered much of arid northern Africa, from the Atlantic to the Red Sea. Allometric equations based on a previous sample of trees allowed the researchers to convert the images into estimates of tree trunks, foliage, root size, and carbon.
New carbon assessment published in the journal Nature, was surprisingly low. While a typical estimate in a region is based on counting small areas and extrapolating the results, the technique demonstrated by NASA in this new study only counts trees that are actually there, down to a single tree.
In previous attempts, the use of satellites, arable land and terrestrial vegetation had a negative effect on optical images. If radar was used, then wetlands and irrigated areas affected the radar backscatter, predicting higher carbon stocks than NASA’s current estimates.
The researchers applied artificial intelligence (deep learning) tree mapping by applying data trained on 90,000 trees to a dataset of nearly 300,000 satellite images to measure more than 9.9 billion tree plants that cast shade and had a crown area of more than 3 square meters. meters. The researchers selected only those features that showed a clear canopy area and associated shadow, allowing the team to rule out small bushes, tufts of grass, rocks, and other misleading features.
The researchers have created an interactive map to view the results of their vast work and it is publicly available here.
Source: Compton Tucker et al, Sub-continental-scale carbon stocks of individual trees in African drylands, Nature (2023). DOI: 10.1038/s41586-022-05653-6
