Pruning Weight Estimation Using Multispectral Sensors in a Vineyard in Southern Italy

Precision viticulture demands characterization of the spatial variability of vineyard vegetative status to plan appropriate management. Pruning weight is an important indicator of vegetative growth and vine vigour. This study aims to provide a new approach for pruning weight estimation using multispectral sensors.The study was conducted in 2021 growing season, in a traditional Mediterranean vine-growing area in western Sicily (Camporeale, Italy). The methodology is based on analyzing multispectral images acquired using an Unmanned Aerial Vehicle (UAV) in a vineyard of Catarratto cultivar. Data acquired by multispectral surveys were got during the summer period. Pruning weight was measured in appropriate sampling points during the winter season. Using NRG (Near Infrared-Red-Green) imagery two vegetation indices were calculated, precisely Normalised Difference Vegetation Index (NDVI) and Green Normalized Difference Vegetation Index (GNDVI). Object-Based Image Analysis Approach (OBIA) was used with the purpose of canopy pixels segmentation from the soil and shadows. Correlation coefficients between vegetation indices and pruning weight were calculated. Pruning weight was estimated with a linear analysis model. Among the vegetation indices taken into consideration, NDVI showed higher correlations with pruning weight compared to the GNDVI index, particularly in the early phenological stages. In fact, the Pearson correlation coefficient values obtained at berries pea size stage, were equal to 0.84. This rapid and accurate methodology demonstrated high accuracy in vineyard pruning weight estimation, providing interesting potential to support grape growers to improve vineyard pruning management.