Assessing the 3-D Structure of Bamboo Forests Using an Advanced Pseudo-Vertical Waveform Approach Based on Airborne Full-Waveform LiDAR Data

Remote-sensing-assisted estimates of Moso bamboo forest structure are imperative for supporting sustainable forest management (SFM). Passive sensors have limited usages in characterizing continuous vertical structures of Moso bamboos since they have difficulty in penetrating dense upper canopies. Airborne full-waveform (FWF) Light Detection and Ranging (LiDAR) technology has shown significant potentials in providing an improved representation of 3-D structures in much denser canopies, while the capability of FWF to assess the structure of the unique Moso bamboo forests is less known. Many previous studies commonly assume perfectly vertical waveforms and do not account for slanted penetrations of waveforms in canopies. This study presented a framework for investigating the availability of FWF to assess subtropical Moso bamboo forest structures. We used a physical-based approach for radiometric calibration and an advanced voxel-based pseudo-vertical waveform approach to assign waveforms into voxels and reconstruct composite waveforms for extracting FWF metrics. Second, the differences in FWF-derived metrics in response to canopy structures across various management strategies were compared. Finally, the capability of FWF metrics for estimating bamboo structural parameters was evaluated by parametric and nonparametric models. The results showed that the FWF metrics varied significantly (p-values < 0.05) in response to different management strategies. In general, nonparametric approaches (locally weighted linear regression and k-nearest neighbor) outperformed parametric approaches (multiple linear regression). This study demonstrated that the pseudo-vertical waveform approach could improve the performances of FWF metrics for characterizing 3-D Bamboo forest structure and provide new insights on accurately estimating bamboo structural parameters across different management strategies.