Fraction cover estimation using drone-based multispectral images in six olive cultivars and different planting systems: a case study in Sicily

• Relative vegetative abundance algorithms accurately estimate fraction cover (Fc). • The NDVI depending on cultivar and planting system. • Canopy area, volume and Fc can be estimated by remote sensing images. Multispectral remote sensing in the olive orchard is expanding, with the aim of improving management for environmental sustainability as well as plant quality and yield. However, the olive tree has a discontinuous vegetative surface, depending on the planting system and cultivar. The aim of this study was to estimate the fraction cover (Fc) with different methods, characterizing the geometric and spectral features of six olive cultivars in four different planting systems. The multispectral data were acquired using a drone equipped with a multispectral camera at 70 m a.g.l. between 12:00 and 13:00 under full sun lighting conditions. Canopy area (CA) and pure Normalized Difference Vegetation Index (NDVI) of canopy and soil were extracted from Geographic Object-Based Image Analysis (GEOBIA). Fc was estimated using two methods: relative vegetation abundance (RA) algorithms and geometric ratio between canopy and available area allotted to each plant according to planting system. In the RA methods, upper (pure canopy NDVI, NDVIc) and lower (pure soil NDVI, NDVIs) limits were obtained from higher frequency class in the bimodal NDVI curve. It was found that the Fc estimated from RA algorithm and geometric ratio were strongly related (R 2 0.97 ***) suggesting that it is a reliable approach, without performing complex image analysis with segmentation and classification algorithms. The proposed model, which integrates NDVI and Fc, provided a tool to assess the health and growth status of olive orchards under various scenarios in order to improve precision management strategies. In addition, it enabled new upper and lower limits useful for the determination of Fc from satellite images.