Search

Aim: This short communication aims at providing insights to verify whether common yield sampling protocols (i.e., one round trip within the fields over two representative rows) are optimal whatever the considered fields. In addition, it aims to show how factors like the spatial organisation of the within-field yield variability, the length of the rows, the erratic variance, etc. may affect the optimal sampling route and the error of the yield estimation.Methods and Material: A new algorithm based on constraint programming and stochastic approaches was used to provide optimal sampling routes for vineyards. This algorithm guarantees the representativeness of the measurement sites and a minimization of the walking distance. Practical constraints (trellised structure, starting point, etc.) are considered by the algorithm to optimise the walking distance and the resulting sampling route. The algorithm has been applied to 60 simulated vineyards with known yield variability. Characteristics like yield spatial structure, row length and proportion of erratic variance were controlled during the simulation process and were used to study how they affect the optimal sampling route derived from the algorithm.Results: The row length as well as the spatial organization of the within-field yield variability are the main factors that determine the optimality of a sampling route. Spatial organisation of the yield happens to have a strong incidence; fields with small yield patterns (Range of the semi-variogram = 25 m) showed a yield estimation error of less than 2 % with an optimal sampling route of three minutes with 7 sampling sites, whereas it takes more than 5 minutes (with 9 sampling sites) to achieve the same estimation error for fields with larger spatial patterns (range > 50 m). Results also highlight the relevance of original sampling routes which intend to sample only the beginnings of rows or mixed approaches based on a round trip in two inter-rows and complementary samples on the beginnings of one or more rows.Conclusions: This study shows that an optimal sampling route strongly depends on the field characteristics. The optimal sampling route should therefore be tailored to each field. This approach is a first step which shows how this methodology could be used to identify other factors of influence. It could also apply to real fields to optimise other logistic operations in viticulture.Significance and Impact of the Study: This short communication demonstrates the necessity to tailor sampling strategy to characteristics of each field to provide both an optimised sampling route (minimum walking distance with minimum samples) and the best possible estimate. It also proposes an original approach based on field simulations and an optimal sampling route generation algorithm. This approach makes it possible to produce new insights (and also to validate empirical practices) that can help the wine industry to better manage the logistics at harvest. This paper also gives considerations when it comes to the choice of a sampling route for a given field.

Um die Ernte und das Keltern der Trauben besser vorzubereiten zu konnen, ist eine Abschatzung des Ertrags pro Parzelle sehr wichtig. Dies wird in der Regel durch Probennahmen an einer begrenzten Anzahl von Messpunkten kurz vor der Ernte erreicht. Die Durchfuhrung dieser Stichproben wird durch bestimmte praktische Aspekte stark eingeschrankt. Dies betrifft zum Beispiel die Zeit, die diese Operation in Anspruch nimmt. Sie wird von der Anzahl der Messungen und der jeweiligen Route bestimmt, die gewahlt wurde, um die Messstellen zu erreichen.

Le développement de l'agriculture de précision nécessite de collecter des données par échantillonnage de manière plus rationnelle. Cet échantillonnage s'appuie sur des outils statistiques qui n'intègrent pas toujours les contraintes opérationnelles. L'objet de cette étude est d'utiliser le raisonnement par contraintes pour optimiser l'échantillonnage dans une parcelle de vigne. Il s'agit à la fois de choisir, dans la parcelle, les points qui respectent les contraintes d'échantillonnage tout en optimisant le déplacement du technicien. C'est donc un problème de tournée, pour lequel la Programmation par Contrainte a développé de nombreux algorithmes de filtrage ces dernières années, notamment la contrainte WeightedSubCircuit [Briot et al. 2017] bien adaptée aux tournées ne passant que par un sous-ensemble de points. Nous proposons ici un modèle pour une version simplifiée du problème et les premiers résultats obtenus avec le solveur Choco., The development of precision agriculture pushes growers to collect sampling data more efficiently. Sampling methods rely on statistical tools that do not always consider operational constraints. The purpose of this study is to use Constraint Programming to optimize sampling in vineyard block. It is both about choosing sampling points that meet operational constraints in the vineyard and optimizing the path taken by the practionner. It is thus a routing problem for which Constraint Programing developed many filtering algorithms in the last few years, in particular the constraint WeightedSub- Circuit [Briot et al. 2017], well suited for routes connecting a small subset of points. We expose in this paper a first model for a simplified approach of the problem as well as the first results that we get from the Choco solver.