Comparison of Landsat and Sentinel-2 surface reflectance data and derived vegetation indexes: application in a rainfed vineyard

Water management in vineyards has become essential during the more frequent flash drought events. Viticulturists seek to implement appropriate soil management techniques and irrigation strategies to maintain yield and wine quality under this scenario. To do so, accurately monitoring vine evapotranspiration (ET) could indicate which management strategy is more efficient regarding vine water use. Remote sensing-based data has been included as input in agrohydrological models to estimate ET at several spatial, temporal and spectral resolutions. This study aimed to assess the influence of the different spatial resolutions of Sentinel 2A-B and Landsat 8-9 satellites data on vine ET estimation. The study was conducted in a rainfed vineyard in Chianti terroir (Tuscany, Italy), during the 2018-2019 vine growing season The comparison between the two sensors was performed at two levels: (a) the surface reflectance values of the Blue, Green, Red and Near Infra-Red bands; and (b) the vegetation indexes (specifically, NDVI and SAVI). Results showed strong correlations between spectral bands (R$^{2}$ $\gt0.78)$, with a higher correlation in the NIR band (R$^{2}$ =0.96), followed by green (R$^{2}$ =0.92), red (R$^{2}$ =0.91) and blue (R$^{2}$ =0.78) bands. Moreover, both SAVI and NDVI followed the same temporal patterns closely related to the vine canopy development along growing seasons. The SAVI index depicted a higher correlation between both sensors (R$^{2}$ =0.96) than the NDVI calculated for both sensors (R$^{2}$ =0.72), which may be related to the better performance of SAVI when the soil background effect is avoided. In conclusion, remotely sensed data from both satellites can be integrated to determine vine ET, thus discriminating soil and water management strategies based on vine water status.