A Framework for Individual Tree Growth Reconstruction Using Multi-Platform Laser Scanning

2026-07-27 · Source: cs.CV updates on arXiv.org · Field: Agriculture & Food Systems — Forestry & Silviculture, Precision Agriculture & Smart Farming · Depth: Advanced, quick

Summary

A new framework has been developed for reconstructing individual tree growth, specifically focusing on diameter at breast height (DBH) and stem volume, using multi-platform laser scanning data. This study utilized 136 point clouds acquired between 2014 and 2025 from 11 different airborne (ALS), mobile (MLS), and terrestrial (TLS) laser scanning platforms across boreal forest test sites. The framework employs deep learning-based segmentation on MLS data for tree delineation, ensuring consistent multitemporal tree correspondence across all point clouds. Stem curves were derived from MLS/TLS data, while ALS data provided height estimations. A height growth-based scaling model was then applied to reconstruct stem attributes over time and estimate growth. Results indicate that this modeled growth achieved higher agreement with manual growth estimates compared to differencing independently estimated attributes. The framework demonstrated temporal robustness, with errors stabilizing after 5-6 years and reaching maximum RMSEs of 8-12% for DBH and 12-23% for volume after 12 years.

Key takeaway

For forest scientists or remote sensing engineers focused on long-term individual tree growth monitoring, you should consider integrating multi-platform laser scanning data with a height growth-based scaling model. This approach provides more robust and accurate growth estimates than traditional differencing methods, especially over extended periods. Employ deep learning for consistent tree delineation across multitemporal datasets to improve the reliability of your growth reconstructions.

Key insights

Combining multi-platform laser scanning with a height growth-based scaling model accurately reconstructs individual tree growth over long periods.

Principles

• Multitemporal tree correspondence is crucial for growth analysis.
• Modeled growth can outperform direct attribute differencing.
• Combining under-canopy and airborne data enhances robustness.

Method

Delineate trees using deep learning on MLS data, transfer to other point clouds. Derive stem curves from MLS/TLS, use ALS for height. Apply a height growth-based scaling model to reconstruct attributes and estimate growth.

In practice

• Use deep learning for consistent tree delineation across time.
• Integrate MLS/TLS for stem data with ALS for height.
• Employ scaling models for robust long-term growth reconstruction.

Topics

• Laser Scanning
• Forest Monitoring
• Tree Growth
• Deep Learning
• Point Clouds
• Remote Sensing

Best for: AI Scientist, Research Scientist, Computer Vision Engineer

Related on AIssential

• Self-Supervised Tree-level Biomass Estimation in Urban Environments From Airborne LiDAR and Optical Observations — Self-supervised deep learning combined with LiDAR and orthophotography enables accurate, scalable urban tree biomass estimation.
• Neural Tree Reconstruction for the Open Forest Observatory — Neural Radiance Fields (NeRF) offer superior 3D forest reconstruction, overcoming classical method limitations for critical climate applications.
• SegmentAnyTreeV2: Scaling Transformer-Based Tree Instance Segmentation Across Sensors, Platforms, and Forests — SegmentAnyTreeV2 offers robust, scalable tree instance segmentation for forest point clouds using a Transformer-based architecture.
• Global-Local Monte Carlo Tree Search in Vision-Language Models for Text-to-3D Indoor Scene Generation — PRM-guided MCTS with hierarchical scene representation enables LVLMs to generate realistic 3D indoor scenes by correcting errors and balancing search.
• Contrasting Local and Global Modeling with Machine Learning and Satellite Data: A Case Study Estimating Tree Canopy Height in African Savannas — Local SatML models can outperform global or fine-tuned global models for specific regional tasks.
• Scalable Deep Learning Framework for Global High-Resolution Land Use Reconstruction — AI4Land uses U-Net on HPC to reconstruct global 1 km land use from coarse data, improving climate model inputs.

See Counsel's argued verdicts on the open AI decisions leaders are weighing →

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by cs.CV updates on arXiv.org.