OENO One special issue
Currently, the greatest challenge for vine growers is to improve the yield and quality of grapes by minimizing costs and environmental impacts. This goal can be achieved through a better knowledge of vineyard spatial variability. Traditional platforms such as airborne, satellite and unmanned aerial vehicles (UAVs) solutions are useful investigation tools for vineyard site specific management. These remote sensing techniques are mainly exploited to get the Normalized Difference Vegetation Index (NDVI), which is useful for describing the morpho-vegetational characteristics of vineyards. This study was conducted in a vineyard in Tuscany (Italy) during the 2017, 2018 and 2019 seasons. Ground data were acquired to detect some agronomic variables such as yield (kg/vine), total soluble solids (TSS), and pruning weight (kg/vine). Remote sensed multispectral images acquired by UAV and Sentinel-2 (S2) satellite platform were used to assess the analysis of the vegetative variability. The UAV NDVI was extracted using both a mixed pixels approach (both vine and inter-row) and from pure canopy pixels. In addition to these UAV layers, the vine thickness was extracted. The aim of this study was to evaluate both classical Ordinary Least Square (OLS) and spatial statistical methods (Moran Index-MI and BILISA) to assess their performance in a multi-temporal comparison between satellite and ground data with UAV information. Good correlations were detected between S2 NDVI and UAV NDVI mixed pixels through both methods (R2 = 0.80 and MI = 0.75). Regarding ground data, UAV layers showed low and negative association with TSS (MI = – 0.34 was the lowest value) whereas better spatial autocorrelations with positive values were detected between UAV layers and both yield (MI ranged from 0.42 to 0.52) and pruning weight (MI ranged from 0.45 to 0.64). The spatial analysis made by MI and BILISA methodologies added more information to this study, clearly showing that both UAV and Sentinel-2 satellite allowed the vigour spatial variability within the vineyard to be detected correctly, overcoming the classical comparison methods by adding the spatial effect. MI and BILISA play a key role in identifying spatial patterns and could be successfully exploited by agricultural stakeholders.
Authors: Laura Pastonchi, Salvatore Filippo Di Gennaro*, Piero Toscano and Alessandro Matese
Institute of BioEconomy, National Research Council (CNR-IBE), Via G. Caproni, 8, 50145 6 Florence, Italy
Keywords: Unmanned Aerial Vehicle (UAV), Sentinel-2 data precision viticulture, Moran’s index (MI), Local indicators of spatial autocorrelation (LISA), vineyard variability