Monitoring vineyard canopy structure by aerial and ground-based RGB and multispectral imagery analysis
Context and purpose of the study – Unmanned Aerial Vehicles (UAVs) are increasingly used to monitor canopy structure and vineyard performance. Compared with traditional remote sensing platforms (e.g. aircraft and satellite), UAVs offer a higher operational flexibility and can acquire ultra-high resolution images in formats such as true color red, green and blue (RGB) and multispectral. Using photogrammetry, 3D vineyard models and normalized difference vegetation index (NDVI) maps can be created from UAV images and used to study the structure and health of grapevine canopies. However, there is a lack of comparison between UAV-based images and ground-based measurements, such as leaf area index (LAI) and canopy porosity. Moreover, most vineyard 3D model studies provide limited details on how they can be used to guide vineyard management. This study evaluated the accuracy of UAV-based canopy measurements, including canopy volume and NDVI and compared them with ground-based canopy measures, such as LAI and canopy porosity.
Material and methods – Throughout the 2017-18 growing season, UAV flights were performed to collect RGB and multispectral images in the research vineyard at the Waite Campus, University of Adelaide, South Australia. Using these images, canopy volume and NDVI were calculated. Ground-based measurements for LAI and canopy porosity were also carried out for comparison.
Results – LAI measured from budburst to harvest showed a peak at around veraison, before starting to decline. Similar trends were also observed in canopy volume and NDVI. Using linear regression, canopy volume of Shiraz and Semillon blocks showed a strong positive correlation with LAI (R2 = 0.75 and 0.68, respectively). NDVI was also positively correlated with LAI (R2 = 0.75 and 0.45 for Shiraz and Semillon, respectively). Canopy volume extracted from UAV-based RGB imagery could be used to monitor canopy development during the growing season. However, canopy volume has limited capacity to inform on important canopy architecture properties such as leaf density, total leaf area and porosity, known to affect yield and fruit quality. The accuracy of NDVI was also found to be strongly affected by the presence of vegetation on the vineyard floor at early development stages.
Issue: GiESCO 2019
1 The University of Adelaide, School of Agriculture, Food and Wine, Waite Research Institute, PMB 1 Glen Osmond, 5064, South Australia. Australia
2 The University of Adelaide, School of Biological Sciences, Adelaide, 5000, South Australia. Australia
Contact the author
remote sensing, unmanned aerial vehicle, leaf area index, canopy architecture, canopy volume, NDVI