Utilización de técnicas geomáticas avanzadas para la toma de decisiones en la producción vitivinícola eficiente

Abstract

Obtaining information on the water status of the vine, characterizing the soils of the vineyard, estimating quality variables of the must or predicting the weight of the pruning wood, allows knowing the physiology of the plant to make decisions about the actions that condition quality and grape yield. Generally, the water status is estimated by the water potential of the leaf using a pressure chamber, being a method that requires a lot of time and effort; characterizing vine soils through various laboratory analyzes requires complex protocols and taking samples from the soil profile; Obtaining data on grape quality variables through repetitive sampling implies allocating resources that can harm the producer in an increasingly competitive market; and weighing the pruning wood using traditional tools is a slow process that is unaffordable in large areas of vines. Using tools based on remote sensing can help reduce the time and resources spent on sample collection and analysis of quality variables, or provide information on vine vigor. Thus, in this thesis field spectroscopy techniques have been used to estimate the water status of the vine leaf and to predict the properties of the soil. In addition, taking high spatial resolution digital images with conventional cameras, together with the SfM (Structure from Motion) photogrammetric technique, has been used to estimate the characteristics of the grape and to predict the weight of the pruning wood. Using partial least squares regression (PLSR), the water status of the plant was estimated from leaf reflectance values (R2= 0.54; RMSE= 0.180), and edaphic properties were estimated from the spectral signatures of the soil, obtaining the best results for pH, electrical conductivity and phosphorus (R2 greater than 0.92). From the digital images captured with a drone, he has managed to correlate grape characteristics with vegetation indices. The best results have been obtained with RGB 2 ((�−�)/�) and RGB 3 ((�+�)/�)), two new indices resulting from this doctoral thesis that are highly correlated with the weight of 100 berries (R=0.77) and with the total polyphenol index (R=0.62), respectively. Likewise, the phenolic maturation index is also correlated with the VARI index (R=0.69). The weight of pruning wood was also estimated by linear regression between the volume of vine vegetation from dense point clouds, proving to be a fast, non-invasive and reliable method in the mencía variety, achieving an R2 of 0.71 and RMSE 224.5(g). The results of this thesis demonstrated that spectroscopy (visible and near-infrared and short-wave reflectance) is a non-destructive technique that allows characterizing vineyard soils and estimating the water status of vines, quickly and reliably) On the other hand , the applied photogrammetric methodologies have made it possible to obtain products to estimate vitivinicultural characteristics from RGB bands of digital images obtained with conventional cameras, potentially helping to improve vineyard management and increasing productivity. Likewise, the result of this work confirmed the feasibility of using SfM as a fast, non-destructive and low-cost procedure to predict the weight of pruning wood, as an indicator parameter of vegetative vigor and associated with the quantity and quality of the grape.