Advancing fine branch biomass estimation with lidar and structural models

Area of Science:

• Forestry and Remote Sensing
• Computational Biology
• Biomass Estimation

Background:

• Lidar technology offers fast and accurate tree measurements.
• Existing methods struggle with precise volume estimation of finer branches due to point dispersion.
• Accurate biomass estimation is crucial for forest management and carbon cycle studies.

Purpose of the Study:

• To develop and validate a novel structural model for accurate above-ground woody biomass estimation from lidar point clouds.
• To improve the estimation of biomass in finer tree branches, which are often neglected.
• To compare the performance of the new model against conventional cylinder fitting and pipe model theory.

Main Methods:

• Coupling point cloud-based skeletonization with multi-linear statistical modeling.
• Developing a "structural model" calibrated with manual measurements.
• Testing the model at segment, axis, and branch levels, and comparing it to existing algorithms.

Main Results:

• The structural model achieved a 1.6% normalized root mean square error (nRMSE) at the segment scale.
• Significantly lower errors (13% nRMSE) and bias (-5%) compared to cylinder fitting (92% nRMSE, 82% bias) and pipe model theory (31% nRMSE, -27% bias) at the branch level.
• Neglecting fine twigs (<5 cm diameter) can lead to a substantial underestimation of total tree biomass by up to 21%.

Conclusions:

• The structural model provides a versatile and effective method for accurate estimation of tree structure volumes, including fine branches, from lidar data.
• Manual calibration is required, but the model enables unbiased, large-scale biomass estimations.
• This approach enhances 3D tree reconstruction and standing biomass assessment.