Poni et al. - Physiology and remote-sensing precision viticulture: the ground-truthing issue

Stefano Poni1*, Matteo Gatti1 e Paolo Dosso2
1Istituto di Frutti-Viticoltura, Università Cattolica del Sacro Cuore, Piacenza
2Studio Terradat, Paderno Dugnano (MI)

nsegna_

Key words: vigor maps, leaf area, variable rate, leaf removal, fertilization, harvesting.

nsegna_

Abstract

nsegna_

Over the last decade, investigation of within-vineyard (block) variability has greatly benefitted from new tools embracing a new science called “precision farming” (PF). PF is based on the acquisition either of airborne images generated from multispectral cameras located on satellites, aircrafts or unmanned aerial vehicles (UAV) or data collected by proximal sensors mounted on the tractor which, once properly pre-processed, can yield vigor maps of the given areas characterized by a variable number of discrete levels of the Normalized Vegetation Index (NDVI). The review paper has a triple goal: a) provide a critical assessment how well-established physiological knowledge can fit and interact with the PF approach; b) evaluate the physiological background and validity of so called “variable rate technologies (VRT) which are, in turn, generated from PF and c) provide a rationale aimed at defining the best groundresolution of aerial images. The paper initially addresses the degree of correlation between NDVI, ground measured vigor and canopy function parameters. A warning is given about the generalized use of the easy-to-measure pruning weight as an index of vine vigor since its correlation with vine capacity (i.e. leaf area) is sometimes loose as well as about the physiological meaning of “high vigor” map classes which, being always expressed on a relative basis, might correspond in reality to absolute low or moderate vigor. Variable rate technologies work on the very clever principles of adding to the vineyard system input amounts calibrated towards vigor and productivity levels. VRT operations addressed in this review are fertilization, leaf removal and mechanical harvesting. Especially when tackling fertilization, criticism is expressed on experimental layouts which, besides including VRT treatment, should also embrace controls at either no or constant dose of fertilizer. Moreover, in the case of N application, doses should also take into account specific vine needs according to phenological stages as well as yield decrease and quality worsening for luxury availability. VRT applied to mechanical harvesting is indeed fascinating as available solutions include a single harvest with separation in different tanks of grapes picked from low and high vigor zones or picking in two different steps according to the maturity level. These tools are precious within a scenario where too fast and unbalanced ripening is increasingly frequent and a counteracting strategy depends upon timeliness and selectivity of the harvesting method. Lastly, this work provides a detailed rationale aimed at identifying the groundresolution of airborne images warranting best correlation with vine growth and yield parameters. With the help of numerical examples hypothesizing using different resolutions spanning from 0,2 to 20 m, it is concluded that best chance for high correlation between means collected at the ground level (elementary data represented by single vines falling within the given pixel area) and the NDVI values is for a pixel resolution varying from 2 to 5 m.

nsegna_

 Download (members only)

Download full article

02 Poni.pdf