Zhang et al. (2026) Fusing UAV-LiDAR, yield-table inversion, and Sentinel data for forest aboveground biomass stock and increment mapping with Monte Carlo uncertainty propagation

Identification

• Journal: Ecological Indicators
• Year: 2026
• Date: 2026-09-23
• Authors: Jiaxu Zhang, Chao Li, Alexander Ryota Keeley, Shunsuke Managi
• DOI: 10.1016/j.ecolind.2026.115530

Research Groups

• Urban Institute & Department of Civil Engineering, Kyushu University

Short Summary

This study develops a pipeline integrating individual-tree UAV-LiDAR analytics, yield-table inversion, machine learning, and multi-source satellite imagery to produce 10 m resolution maps of aboveground carbon stocks and projected carbon increments across Shikoku Island, Japan.

Objective

• Estimate individual tree age and site quality by inverting the LYCS yield tables using UAV-LiDAR-derived tree height and DBH.
• Develop an XGBoost model to predict individual tree AGB using the LYCS-derived age, site quality, topographic variables, and species identity.
• Generate 5-, 10-, and 20-year AGB increment maps through forward age projection.

Study Configuration

• Spatial Scale: Shikoku Island, Japan (approximately 18,800 km²)
• Temporal Scale: 2025 (UAV-LiDAR data acquisition)

Methodology and Data

• Models used:
  • Local Yield table Construction System (LYCS) for tree age estimation
  • XGBoost model for individual tree AGB prediction
• Data sources:
  • UAV-LiDAR point clouds
  • Sentinel-1 SAR data
  • Sentinel-2 MSI Level-2A surface reflectance data
  • Digital Elevation Model (DEM) from AW3D30

Main Results

• Individual tree age and site quality estimated with high accuracy using LYCS inversion.
• XGBoost model achieved an R² of 0.86 for individual tree AGB prediction.
• Forward age projection generated carbon increment maps at 5-, 10-, and 20-year horizons.

Contributions

• This study integrates yield-table inversion, UAV-LiDAR analytics, and multi-source satellite data fusion to produce both static and forward-projected AGB maps with traceable uncertainty estimates for Japanese plantation forests.
• The developed pipeline addresses the limitations of existing remote sensing approaches by incorporating ecological growth parameters and propagating uncertainty through the full modeling chain.

Funding

• This research was supported by the Japan Aerospace Exploration Agency (JAXA) and the Forestry Agency of Japan.

Citation

@article{Zhang2026Fusing,
  author = {Zhang, Jiaxu and Li, Chao and Keeley, Alexander Ryota and Managi, Shunsuke},
  title = {Fusing UAV-LiDAR, yield-table inversion, and Sentinel data for forest aboveground biomass stock and increment mapping with Monte Carlo uncertainty propagation},
  journal = {Ecological Indicators},
  year = {2026},
  doi = {10.1016/j.ecolind.2026.115530},
  url = {https://doi.org/10.1016/j.ecolind.2026.115530}
}