Branch angles are an important plant morphological trait affecting light interception within forest canopies. However, studies on branch angles have been limited due to the time-consuming nature of manual measurements using a protractor. Terrestrial laser scanning (TLS), however, provides new opportunities to measure branch angles more efficiently. Despite this potential, studies validating branch angle measurements from TLS have been limited. Here, our aim is to evaluate both manual and automatic branch angle measurements of European beech from TLS data using traditional field-measurements with a protractor as a reference. We evaluated the accuracy of branch angle measurements based on four automated algorithms (aRchiQSM, TreeQSM, Laplacian, Semantic-Laplacian) from TLS data. Additionally, we assessed different ways of manual branch angle measurements in the field. Our study was based on a dataset comprising 124 branch angles measured from six European beech in a European deciduous forest. Our results show that manual branch angle measurements from TLS data are in high agreement with the reference (root-mean-squared error, RMSE: [3.57°–4.18°], concordance correlation coefficient, CCC: [0.95−0.97]) across different branch length positions. Automated algorithms also are in high agreement with the reference although RMSE is approximately twice as large compared to manual branch angle measurements from TLS (RMSE: [9.29°–10.55°], CCC: [0.83−0.86]) with manual leaf points removal. When applying the automatic wood-leaf separation algorithm, the performance of the four methods declined significantly, with only approximately 20 branch angles successfully identified. Moreover, it is important to note that there is no influence of the measurement position (branch surface versus center) for branch angle measurements. However, for curved branches, the selection of branch measurement length significantly impacts the branch angle measurement. This study provides a comprehensive understanding of branch angle measurements in forests. We show that automated measurement methods based on TLS data of branch angles are a valuable tool to quantify branch angles at larger scales.

中文翻译：

---

使用地面激光扫描评估欧洲山毛榉的树枝角度测量


分枝角是影响森林树冠内光拦截的重要植物形态特征。然而，由于使用量角器进行手动测量非常耗时，因此对分支角度的研究受到限制。然而，地面激光扫描 （TLS） 为更有效地测量分支角度提供了新的机会。尽管有这种潜力，但验证 TLS 分支角测量的研究一直很有限。在这里，我们的目标是使用传统的现场测量和量角器作为参考，从 TLS 数据中评估欧洲山毛榉的手动和自动分支角度测量。我们根据 TLS 数据的四种自动算法 （aRchiQSM、TreeQSM、Laplacian、Semantic-Laplacian） 评估了分支角度测量的准确性。此外，我们评估了现场手动分支角度测量的不同方式。我们的研究基于一个数据集，该数据集包含从欧洲落叶林中的 6 棵欧洲山毛榉测量的 124 个分支角度。我们的结果表明，TLS 数据的手动分支角度测量与参考 （均方根误差，RMSE： [3.57°–4.18°]，相关系数，CCC： [0.95-0.97]）高度一致。尽管与手动去除叶点的 TLS （RMSE： [9.29°–10.55°]，CCC： [0.83−0.86]）的手动分支角测量相比，RMSE 大约是参考算法的两倍，但自动算法也与参考算法高度一致。当应用自动木叶分离算法时，四种方法的性能显著下降，仅成功识别了大约 20 个分支角度。 此外，需要注意的是，测量位置（分支表面与中心）对分支角度测量没有影响。然而，对于弯曲的树枝，树枝测量长度的选择会显着影响树枝角度的测量。本研究提供了对森林中树枝角度测量的全面了解。我们表明，基于分支角度的 TLS 数据的自动测量方法是在更大尺度上量化分支角度的宝贵工具。