AI Model Achieves Near-Lidar Accuracy in Forest Canopy Mapping Using Standard Satellite Imagery

Researchers have developed an advanced artificial intelligence model that produces high-resolution canopy height maps using only standard RGB imagery, achieving near-lidar accuracy for precise monitoring of forest biomass and carbon storage over large areas. This innovation addresses the critical need for cost-effective forest monitoring as forests and plantations play a vital role in carbon sequestration, yet traditional methods remain expensive and labor-intensive.

The joint research team from Beijing Forestry University, Manchester Metropolitan University, and Tsinghua University published their findings in the Journal of Remote Sensing on October 20, 2025. Their study introduces a novel framework that combines large vision foundation models with self-supervised learning to deliver sub-meter accuracy in estimating tree heights from RGB satellite images. The research addresses the long-standing problem of balancing cost, precision, and scalability in forest monitoring, offering a promising tool for managing plantations and tracking carbon sequestration under initiatives such as China's Certified Emission Reduction program.

Monitoring forest canopy structure is essential for understanding global carbon cycles, assessing tree growth, and managing plantation resources. Traditional lidar systems provide accurate height data but are limited by high costs and technical complexity, while optical remote sensing often lacks the structural precision required for small-scale plantations. The new AI-driven vision model bridges this gap by achieving a mean absolute error of only 0.09 meters and an R² of 0.78 when compared with airborne lidar measurements, outperforming traditional CNN and transformer-based methods.

The model architecture consists of three modules: a feature extractor powered by the DINOv2 large vision foundation model, a self-supervised feature enhancement unit to retain fine spatial details, and a lightweight convolutional height estimator. This approach enabled over 90% accuracy in single-tree detection and strong correlations with measured above-ground biomass. The model demonstrated strong generalization across forest types, making it suitable for both regional and national-scale carbon accounting.

Testing in Beijing's Fangshan District, an area with fragmented plantations primarily composed of Populus tomentosa, Pinus tabulaeformis, and Ginkgo biloba, showed the AI model produced canopy height maps closely matching ground truth data. Using one-meter-resolution Google Earth imagery and lidar-derived references, the model significantly outperformed global canopy height model products, capturing subtle variations in tree crown structure that existing models often missed. The generated maps supported individual-tree segmentation and plantation-level biomass estimation with R² values exceeding 0.9 for key species.

When applied to a geographically distinct forest in Saihanba, the network maintained robust accuracy, confirming its cross-regional adaptability. The ability to reconstruct annual growth trends from archived satellite imagery provides a scalable solution for long-term carbon sink monitoring and precision forestry management. The research methodology employed an end-to-end deep-learning framework combining pre-trained large vision foundation model features with a self-supervised enhancement process, with high-resolution Google Earth imagery from 2013–2020 used as input and UAV-based lidar data serving as reference for training and validation.

Dr. Xin Zhang, corresponding author at Manchester Metropolitan University, stated that their model demonstrates large vision foundation models can fundamentally transform forestry monitoring. By combining global image pretraining with local self-supervised enhancement, the team achieved lidar-level precision using ordinary RGB imagery, drastically reducing costs and expanding access to accurate forest data for carbon accounting and environmental management.

The AI-based mapping framework offers a powerful and affordable approach for tracking forest growth, optimizing plantation management, and verifying carbon credits. Its adaptability across ecosystems makes it suitable for global afforestation and reforestation monitoring programs. As detailed in the study published at https://spj.science.org/doi/10.34133/remotesensing.0880, this innovation could play a central role in achieving sustainable forestry and climate-change mitigation as the world advances toward net-zero goals. Future research will extend this method to natural and mixed forests, integrate automated species classification, and support real-time carbon monitoring platforms.