UAV LiDAR for accurate quantification of shelterbelt biomass: a reliable approach

Deng, Rongxin, Wu, Yuzong, Wang, Wenjuan, Li, Shiming, Niu, Xiaotao, Yang, Gao, Yao, Zhihong
Agroforestry Systems · Springer Nature · 2026-11-20 · DOI: 10.1007/s10457-026-01747-6

Аннотация

Farmland shelterbelts are a vital component of greening projects in plain regions of China. Accurate estimation of their biomass is essential for the scientific assessment of the carbon sequestration capacity. This paper developed a biomass estimation method for farmland shelterbelts using UAV LiDAR data. Firstly, tree height was extracted by individual tree segmentation algorithms considering the characteristics of farmland shelterbelts. Then, individual tree diameter at breast height (DBH) was predicted by machine learning algorithms. Finally, the shelterbelt biomass was estimated by a universal biomass model and the accumulation of individual tree biomass. The results showed that, (1) the segmentation method incorporating shelterbelt age and configuration characteristics enabled precise extraction of tree parameters (tree height: R ^2 = 0.928, RMSE = 0.908 m; crown width: R ^2 = 0.873, RMSE = 0.647 m); (2) the random forest algorithm showed better performance in predicting DBH compared with other machine learning algorithms ( R ^2 = 0.951, RMSE = 0.031 m, rRMSE = 10.8%); (3) the estimated biomass achieved a relatively high accuracy ( R ^2 = 0.906, RMSE = 7.415 t/belt, rRMSE = 27.6%), demonstrating the reliability of this method for farmland shelterbelt biomass estimation. This study established a reliable remote sensing method for precise estimation of the farmland shelterbelt biomass, which could provide reliable data for farmland shelterbelt resource surveys and carbon sink capacity assessments.

Читать у издателя →

Похожие статьи