Multi-sensor point cloud analysis for canopy volume estimation in a steep-slope Mediterranean vineyard: a comparison of LiDAR and RGB approaches

Volumetric analysis of tree canopies is increasingly important for farmers 4.0, as it enables the adoption of targeted agricultural practices. Additionally, the recent European and Mediterranean Plant Protection Organization (EPPO) guideline introduced a new labelling system for plant protection products based on biometric indices as references to express the dose. Currently, conventional manual methods used for volumetric assessments are generally time-consuming. The present study aimed to undertake a comparative analysis of canopy volumes derived from various combinations of monitoring technologies and analytical methods applied to the point cloud, in conjunction with reference volumes obtained from manual measurements, within a highly complex scenario like the Mediterranean steep-slope vineyard. The technologies tested for UAV monitoring included LiDAR, low and high cost RGB cameras, whilst terrestrial monitoring was carried out with terrestrial LiDAR (GeoSLAM TM ). The analytical methods applied were alphashape (α = 0.1–0.5), convex hull, and voxelization (s = 0.05–0.10 m). The results showed that for UAV-based surveys, the convex hull was the most appropriate analytical method (RMSE: 0.10, 0.13, and 0.35 m 3 ), while for terrestrial LiDAR, the alpha shape with α = 0.5 was the most appropriate (RMSE: 0.10 m 3 ). Overall, the results showed that volumetric data can be obtained through rapid surveys, provided the most suitable analytical method is adopted.