Unmanned Aerial Vehicle (UAV) Multispectral Imagery and GIS-based Approaches for Vineyard Row Identification and Segmentation in Precision Viticulture

This study used multispectral data obtained from an Unmanned Aerial Vehicle (UAV) and Geographic Information System (GIS) - based spatial analysis approaches to identify vineyard rows, segment vineyard canopies, and quantify vine vigour in the Okanagan region. High-resolution multispectral imagery with a ground pixel size of 3cm was processed to vegetation indices, including Normalized Differential Red Edge (NDRE) and Green Normalized Difference Vegetation Index (GNDVI).

A canopy height model was generated to isolate vine structures using a 0.5m threshold, and a rule-based geometric clustering approach was applied to delineate vineyard rows. NDRE values ranged from 0.13 to 0.65 in the test block and 0.15 to 0.64 in the applied block, while GNDVI ranged from −0.01 to 0.64 and −0.29 to 0.64, respectively. Spatial patterns were consistent across both indices, with higher vigour concentrated in the central rows and lower vigour along the block margins.