LiDAR-based individual tree AGB modeling of Pinus kesiya var. langbianensis by incorporating spatial structure

Abstract

Accurate and efficient estimation of individual tree aboveground biomass (AGB) is crucial for precision forestry and forest carbon stock assessment. While the influence of spatial structure on biomass is acknowledged, its integration into individual tree AGB models for enhancing accuracy remains equivocal. The UAV-LiDAR data and individual tree AGB destructively measured of natural Pinus kesiya var. langbianensis were collected from 2022 to 2023. First, Individual tree attributes and spatial structures were extracted and evaluated from LiDAR data. Then, two AGB models were developed and compared: one independent of diameter at breast height (DBH) and another incorporating stand spatial structure parameters, including the uniform angle index (W), neighborhood comparison (U), stand level rate (Si), and competition index (UCi). The results showed that AGB models can be effectively constructed based on canopy features alone, even without the inclusion of DBH. The accuracy of individual tree AGB models was improved when spatial structure parameters were incorporated. In particular, the introduction of angular scales improved the accuracy of the AGB model most significantly. On average, R2 increased by 8.668%, while RMSE, SEE, and TRE decreased by 11.262%, 10.619%, and 7.570%, respectively. Spatial structure parameters facilitated a more precise and realistic depiction of competitive interactions and spatial distribution patterns among trees, thereby enhancing model performance. It demonstrated an effective approach for leveraging UAV-LiDAR data to rapidly and precisely estimate AGB and carbon stocks for individual trees, forest stands and regional scales.

Keywords

• Pinus kesiya var. langbianensis
• Individual tree AGB model
• UAV-LiDAR
• Spatial structure
• Dummy variable