Details of 10 billion trees help monitor carbon stocks

The drylands of Africa’s Sahel region are not exactly deserts. They support numerous free-standing trees that represent an important store of carbon, and provide ecosystem services such as wood for cooking, food for livestock and protection against soil erosion. It is important to monitor the distribution, carbon stocks and density of these trees as the climate changes, but current monitoring techniques do not have sufficient resolution to do this. 

Knowledge of tree size and distribution is also important for carbon credit trading, which is likely to become a common means of offsetting the production of greenhouse gases in future. Carbon offset credits allow companies to emit a given amount of CO₂ in exchange for the greenhouse gas being neutralized elsewhere, in the form of trees planted or left to grow. But keeping track of how many new trees are planted as a result of trades, globally, and how many are disappearing is extremely challenging.

Researchers from the University of Copenhagen’s Department of Geosciences and Natural Resource Management and Department of Computer Science, in collaboration with the American space agency NASA, have now pioneered a method that makes it possible to count the number of trees across vast swaths of land and calculate how much carbon is sequestered within each tree. The new method, along with details about its application in the Sahel region, is published in the journal Nature.

The researchers made use of field data, artificial intelligence (AI) associated with high performance computing, and high spatial resolution imagery to develop their model for mapping several billion trees and their carbon uptake in the Sahel. They analyzed more than 300,000 satellite images of the semi-arid sub-Saharan region north of the equator, and identified 9.9 billion trees. The experts then used AI to train one of NASA’s supercomputers to identify all individual trees from the images, using their characteristic appearance – for example, acacias have flat crowns.

On the ground, the researchers characterized the structural parameters of trees from 30 different species (e.g., their height, crown area, density and biomass) as well as measuring the carbon stock stored in each tree’s different components – leaves, wood and roots. 

“We have done a lot of fieldwork, weighing trees when they are cut down to collect data on the amount of carbon they contain. We combined this with the satellite images of the number of trees and the size of individual tree crowns and fed it all into a larger model,” explains Professor Rasmus Fensholt, who played a key role in the research. 

The researchers used the field data, satellite images and computer model to calculate that the 9.9 billion trees across the Sahel region currently store 840,000,000 tons of carbon. This is lower than carbon stocks in forests on mainland France, for example, which are estimated to be in the region of 2.4 billion tons. However, it is likely that most models currently used in climate simulations underestimate the density and carbon stocks of isolated trees in drylands, and this new method is likely to be far more accurate. 

For Jean-Pierre Wigneron, INRAE researcher at the Soil-Plant-Atmosphere Interactions Joint Research Unit and co-author of the study, “this study is exceptional because it is pioneering a type of approach that will revolutionize the monitoring of trees and forests on a global scale: in the short term, it will be possible to map the world’s trees from the center of the Amazon to our schoolyards. Many fields of application will become much more efficient and accurate, such as tracking carbon stocks, biodiversity, monitoring logging and protection against illegal forest degradation, in near real time.”

According to Professor Fensholt, the new method can also be an important tool for companies when they and others want to pay for their emissions – either in the form of newly planted trees or by paying farmers not to cut down trees on their land.

“There are many indications that carbon credit trading will become more and more extensive in the future – not just for earth’s forests, but for billions of trees beyond forests. Therefore, it is critical for us to be able to monitor whether carbon trading reflects the actual number of trees in nature and whether it has a positive effect on climate. This is what we are coming with a solution to,” he says.

Besides helping to control the effect of climate credits, the method can also be used to investigate whether various nature restoration projects with trees in regions such as the semi-arid Sahel, are actually going as they should.

“Over the past 10-15 years, considerable resources have been spent on large-scale tree planting projects in arid regions of the world, including those financed by the World Bank. Have they worked? Have the trees survived? Our method can be used to help map this,” says Fensholt, adding that the next research project in the pipeline will be to look at how the number of trees has evolved over past decades.

In the current study, the researchers applied their new method to assessing changes in tree density and carbon stocks in an agroforestry region in Senegal, north of Khombole. When comparing the data on tree density in 2002 and 2021, the new method identified that carbon stocks in the region have almost doubled. 

According to the researchers, the database of the nearly 10 billion Sahelian trees is now publicly available. For each tree it includes the wood mass, foliage mass, root mass and carbon stock. This information is essential for scientists, decision-makers, agronomists and foresters working on dryland restoration, but also for farmers, who can use these data to estimate and value the carbon stocks of the trees on the land they farm. In addition, the groundwork for the new method is done and it is ready to be deployed in the near future by public agencies, NGOs and others interested in monitoring tree tallies and their carbon content.

“Our study demonstrates that deep learning techniques can revolutionize the global mapping of individual trees and their biomass. Our artificial neural networks learn to extract complex patterns from large amounts of satellite images, allowing for more accurate and efficient identification of individual trees and subsequent estimation of their biomass,” says Professor Christian Igel from the Department of Computer Science at the University of Copenhagen.

—

By Alison Bosman, Earth.com Staff Writer

Check us out on EarthSnap, a free app brought to you by Eric Ralls and Earth.com.