Research · Carbon Quantification

Forest Carbon Structure from Point Clouds

Deep learning segmentation of airborne LiDAR point clouds to quantify above-ground biomass and canopy structure at individual-tree scale. Delivering MRV-grade carbon stock estimates for nature markets and carbon verification schemes.

Wood-Leaf Segmentation

Drag the slider below to compare the full point cloud classification (leaf + wood) with wood-only structure. This segmentation is the foundation for biomass estimation and canopy metrics.

Wood Classification
Leaf + Wood Classification

Methodology

We use convolutional neural networks trained on terrestrial laser scanning (TLS) ground truth to segment airborne LiDAR point clouds into vegetation components. The model learns to distinguish wood from leaf material across varied forest types and canopy structures.

Key Outputs

  • •Individual tree biomass estimates — above-ground biomass quantified at single-tree resolution
  • •Canopy structure metrics — leaf area, wood volume, height profiles, and crown dimensions
  • •MRV-grade carbon verification — audit-ready outputs for carbon markets and compliance schemes

Details

Data Source

Airborne LiDAR

Model Type

Deep Learning (CNN)

Resolution

Individual tree scale

Output Format

Point cloud segmentation + metrics