Abstract
The canopy volume of fruit trees is an important basis for precise orchard management. However, current methods for predicting canopy volume cannot accurately identify and exclude canopy porosity, resulting in a larger prediction than the actual volume. To address this issue, this study proposes a calculation method of canopy effective volume (EV) for fruit tree based on LiDAR point cloud data. In this method, the fruit tree canopy model is first reconstructed using an improved alpha-shape algorithm, and its volume is calculated. Then, the canopy effective volume coefficient was constructed, and the product of the two was used as the canopy effective volume. To evaluate the accuracy and applicability of the proposed method, both simulated fruit tree and orchard experiments were conducted and compared with the prediction results of alpha-shape by slices (ASBS), convex hull by slices (CHBS), and voxel-based (VB) methods. The results show that the best model prediction performance is achieved when the voxel size is the average nearest neighbor distance of the point cloud and the partition size is five times the voxel size. The method achieved an R² of 0.9720, an RMSE of 0.0203 m(3), and an MAE of 0.0192. Compared with the prediction results of the ASBS, CHBS, and VB methods, the volume reduction rates were 0.5101, 0.6953, and 0.6213, respectively. The EV method can accurately quantify the canopy effective volume after removal of canopy porosity and provide decision support for precise orchard management.