Space Lasers, AI and Satellite Data Combine to Map Forest Biomass

The Problem

Forests store around 80 percent of the world’s terrestrial carbon, making them critical in regulating Earth’s climate. Traditional methods to measure forest aboveground biomass, which is key to understanding a forest’s carbon cycle, including retention and release, are slow, costly, and limited to small areas. Accurate, large-scale mapping of forest aboveground biomass informs policy decisions related to carbon accounting and forest management.

The Work

Hamdi Zurqani, an Assistant Professor of Geospatial Science with the Arkansas Forest Resources Center, which is part of the Arkansas Agricultural Experiment Station and the University of Arkansas at Monticello, used open-access satellite data to build a fast, scalable system for mapping aboveground biomass. His approach integrated NASA’s Global Ecosystem Dynamics Investigation (GEDI) LiDAR data with optical and radar data from the European Space Agency’s Sentinel-1 and Sentinel-2 satellites. The data were analyzed through four machine learning algorithms: Gradient Tree Boosting, Random Forest, Classification and Regression Trees (CART), and Support Vector Machine.

The Results

Among the tested algorithms, Gradient Tree Boosting achieved the highest accuracy and lowest error rates in biomass estimation. Random Forest also performed well but was slightly less precise. CART yielded moderate results, while the Support Vector Machine struggled with accuracy. The study found that the most reliable biomass predictions came from combining Sentinel-2 optical imagery, vegetation indices, topographic data, and canopy height with GEDI LiDAR measurements. This demonstrated that multi-source data integration is essential for dependable large-scale forest aboveground biomass mapping.

The Value

Zurqani’s method provides a faster, more accurate way to measure aboveground biomass in forests. Improved biomass mapping supports better tracking of carbon sequestration and emissions from deforestation, informing climate change policies and sustainable forest management strategies. While challenges remain, like weather effects on satellite data and limited LiDAR coverage, the approach offers a promising approach for biomass mapping. Future research will explore advanced AI models, like neural networks, to further refine biomass estimates and assist in safeguarding forests, Zurqani said.

About the Researchers