We analyzed the potential of non-destructive optical sensing of grape skin anthocyanins for selective harvesting in precision viticulture. We measured anthocyanins by a hand-held fluorescence optical sensor on a 7 ha Sangiovese vineyard plot in central Italy. Optical indices obtained by the sensor were calibrated for the transformation in units of anthocyanins per berry mass, i.e., milligrams per gram of berry fresh weight. A full protocol for optimal data filtration, interpolation, and homogeneous zone delineation based on a very large number of optical measurements is proposed. Both the single signal-based fluorescence index (ANTHR) and the two signal ratio-based index (ANTHRG) can be used for Sangiovese grapes. Significant separations of grape-quality batches were obtained by several methods of data classification and zone delineations. Basic statistical criteria were as efficient as the K-means clustering. The best separations were obtained for three classes of grape skin anthocyanin.
Keywords: Sangiovese; Vitis vinifera L.; anthocyanin mapping; grapevine; management zones; optical sensors; precision agriculture.