Synthetic Aperture Radar, Multi-Source Data Fusion, and Machine Learning for Aboveground Biomass and Carbon Stock Estimation: A Systematic Review and Bibliometric Analysis
This systematic review and bibliometric analysis synthesizes 68 studies from 2004 to 2026 to demonstrate how integrating Synthetic Aperture Radar, multi-source remote sensing, and machine learning has advanced aboveground biomass and carbon stock estimation, while highlighting a critical research gap in applying these robust, scalable frameworks to heterogeneous urban tree environments.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
Trees are the planet's lungs, breathing in carbon dioxide and storing it as wood, leaves, and roots. This stored carbon is a vital counterweight to the greenhouse gases warming our atmosphere. To understand how much carbon the world's forests hold, or how much a city's street trees might save, scientists must measure the "aboveground biomass," which is simply the total weight of all the living plant material above the soil. For decades, the only way to get an accurate count was to walk into a forest, measure every tree with a tape measure, and do the math by hand. While this method is precise, it is slow, expensive, and impossible to do everywhere at once. Today, we have a faster way: looking at the Earth from space. Satellites can now take pictures of forests and cities, using different types of "eyes" to see what the human eye cannot. Some satellites see color, others see the shape of the canopy, and some use radar, which bounces invisible waves off trees to reveal their structure even through clouds.
A new review of scientific research brings together the latest findings on how we are using these space-based tools to count trees and estimate their carbon. The researchers behind this review, Temitope Emmanuel Paul, Adedayo Adeleke, and Abel Ramoelo, did not go out into the field themselves. Instead, they gathered and analyzed 68 different studies published between 2004 and 2026. Their goal was to see how scientists are combining data from different satellites and using smart computer programs to figure out how much carbon is stored in trees, both in wild forests and in our cities. They found that the best results come not from using just one type of satellite picture, but from mixing them together. By combining the color information from optical cameras, the structural details from laser scanners, and the all-weather view from radar, scientists can build a much clearer picture of a forest's health and size.
The story of this research is one of rapid evolution. In the early years of the period they studied, scientists mostly tried to use single tools to solve the problem. They might rely on a satellite that sees color, or one that uses radar, but rarely both. The review shows that this approach has changed. Today, the most successful studies mix data from multiple sources. They take the spectral data, which tells us about the greenness and health of the leaves, and combine it with structural data that reveals the height and density of the branches. One of the most powerful tools in this mix is radar, which sends out microwave signals that can penetrate clouds and see the shape of the trees even when the sky is overcast. Another key tool is LiDAR, which uses laser pulses to create a precise 3D map of the forest floor and the canopy above it. When researchers fuse these different types of information, the accuracy of their carbon estimates improves significantly compared to using any single source alone.
To make sense of all this mixed data, scientists are turning to machine learning. These are computer programs that learn from examples rather than following a fixed set of rules. The review found that the most reliable and widely used programs are "ensemble" methods, which work by combining the predictions of many smaller models to reach a final answer. Among these, a method called Random Forest has been the most popular and robust choice. It handles the messy, complex data from satellites very well and can tell us which pieces of information were most important for the final calculation. Other powerful tools, like Extreme Gradient Boosting, are also being used to squeeze out even more accuracy. More recently, deep learning, a type of artificial intelligence that mimics the human brain, has begun to appear in the research. These advanced systems are particularly good at spotting individual trees in crowded city streets, but they require vast amounts of training data and powerful computers to work effectively.
Despite these impressive technological leaps, the review highlights a significant gap in our knowledge. While we have learned a great deal about measuring carbon in vast, continuous forests, we know much less about how to do it in cities. Urban environments are incredibly complex; trees are scattered among buildings, roads, and concrete, and their shapes are often distorted by the surrounding structures. The researchers found that very few studies have successfully applied the powerful combination of radar and machine learning to urban trees. Most city-focused studies still rely heavily on optical images and laser scans, often leaving out the radar data that could help see through clouds and understand the tree structure in difficult conditions. This is a missed opportunity, because cities are becoming increasingly important for climate solutions, and we need accurate ways to count the carbon in our street trees to plan for a cooler future.
The path forward, according to the authors, involves bridging this gap. They suggest that future research must focus on creating models that can work across different types of landscapes, from dense rainforests to busy downtowns. This means developing better ways to handle the unique challenges of cities, where radar signals can bounce off buildings and create confusing echoes. It also means building standardized ways to test these models so that a method developed in one city can be trusted in another. The review concludes that while we have made tremendous progress in combining satellite data and artificial intelligence, the real work lies in making these tools robust enough to handle the messy reality of the world. By refining these methods, we can move from estimating carbon in broad strokes to counting it with precision, giving us the clear data we need to manage our forests and cities for a stable climate.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.