A Comprehensive Comparison of Individual Tree Crown Delineation of Plantations Using UAV-LiDAR Data: A Case Study for Larch (<italic>Larix Olgensis</italic>) Forests in Northeast China

Abstract

Individual tree crown delineation (ITCD) employing unmanned aerial vehicle light detection and ranging data can directly obtain high-precision tree-level structural information within a block, with this information being the foundation for monitoring and management of the forest, thus reducing time-consuming labor. Despite the fact that numerous ITCD algorithms have been proposed, there has not yet been a robust and comprehensive comparison of these algorithms in plantations. In this article, we evaluated the performance of seven classic ITCD methods under various stand densities and crown classes and analyzed the parameter sensitivity as well as the correlation of segmentation accuracy with optimal parameters and stand metrics. The results demonstrate that the segmentation and crown description accuracy, stability, and adaptability of the algorithm should be comprehensively considered when choosing an algorithm. The forest characteristics impact the accuracy of the algorithms, and the complexity of the forest canopy structure and omission error of suppressed trees are the key factors impacting ITCD accuracy. Furthermore, this study shows that it is feasible to control the parameters of the algorithm through stand measurement. These results will be helpful in guiding the selection of ITCD methods and will provide support for improving the ITCD algorithm in the future.

Keywords

• Crown class
• individual tree crown delineation (ITCD)
• light detection and ranging (LiDAR)
• sensitivity analysis
• unmanned aerial vehicles (UAVs)