New AI-Driven Method Enhances Carbon Stock Mapping Precision

Researchers have developed a novel method for estimating above-ground carbon (AGC) in individual trees, potentially revolutionizing carbon stock mapping and climate change mitigation efforts. The study, published in the Journal of Remote Sensing, combines very high-resolution (VHR) satellite imagery with machine learning algorithms to achieve unprecedented precision in carbon sequestration tracking.

Led by Martí Perpinyana-Vallès, the research team at Lobelia Earth S.L. created an Artificial Neural Network (ANN) model trained on over 400 individual tree crowns. By integrating spectral signatures and crown area data from Pléiades high-resolution satellite imagery, the model achieved an R² of 0.66 and a relative RMSE of 78.6% in AGC estimates. This approach significantly reduces biases seen in previous technologies, particularly in underestimating carbon stocks in dryland regions.

The study's methodology involved constructing a comprehensive AGC reference database from on-the-ground tree measurements, which were then converted into biomass using species-specific allometric equations. Deep learning models were employed to segment individual tree crowns and extract spectral information from VHR imagery, resulting in a highly accurate model with a tree-level RMSE of just 373.85 kg.

This innovation addresses longstanding limitations in carbon stock estimation by allowing for the accurate geolocation of individual trees. The use of Pléiades Neo satellite imagery, known for its exceptional 30cm native resolution, enabled this unprecedented level of precision in Earth observation.

The implications of this research are far-reaching. It has the potential to improve global carbon cycle assessments, optimize land use strategies, and enhance reforestation initiatives. For policymakers and environmental scientists, this method could provide essential data to support climate change mitigation strategies and international climate agreements.

As the world grapples with the urgent need to address climate change, this AI-driven approach to carbon stock mapping offers a powerful tool for more effective environmental management. By providing more accurate data on carbon sequestration, it could help guide targeted interventions and measure the success of climate mitigation efforts with greater precision.