Sentinel-2 data analysis and comparison with uav multispectral images for precision viticulture

Abstract

Precision viticulture (PV) requires the use of technologies that can detect the spatial and temporal variability of vineyards and, at the same time, allow useful information to be obtained at sustainable costs. In order to develop a cheap and easy-to-handle operational monitoring scheme for PV, the aim of this work was to evaluate the possibility of using Sentinel-2 multispectral images for long-term vineyard monitoring through the Normalized Difference Vegetation Index (NDVI). Vigour maps of two vineyards located in northeastern Italy were computed from satellite imagery and compared with those derived from UAV multispectral images; their correspondence was evaluated from qualitative and statistical points of view. To achieve this, the UAV images were roughly resampled to 10 m pixel size in order to match the spatial resolution of the satellite imagery. Preliminary results show the potential use of open source Sentinel-2 platforms for monitoring vineyards, highlighting links with the information given in the agronomic bulletins and identifying critical areas for crop production. Despite the large differences in spatial resolution, the results of the comparison between the UAV and Sentinel-2 data were promising. However, for long-term vineyard monitoring at territory scale, further studies using multispectral sensor calibration and groundtruth data are required.