Zonage vitivinicole: recherches et considérations initiales sur une proposition de “nouvelle” méthodologie d'”évaluation de la qualité” du produit tel qu’élément base pour le zonage aussi

Grapevine yield is a key indicator to assess the impacts of climate change and the relevance of adaptation strategies in a vineyard landscape. At this scale, a yield model should use a number of parameters and input data in relation to the information available and be able to reproduce vineyard management decisions (e.g. soil and canopy management, irrigation). In this study, we used data from six experimental sites in Southern France (cv. Syrah) to calibrate a model of grapevine yield limited by water constraint (GraY). Each yield component (bud fertility, number of berries per bunch, berry weight) was calculated as a function of the soil water availability simulated by the WaLIS water balance model at critical phenological phases. The model was then evaluated in 10 grapegrowers’ plots, covering a diversity of biophysical and technical contexts (soil type, canopy size, irrigation, cover crop). We identified three critical periods for yield formation: after flowering on the previous year for the number of bunches and berries, around pre-veraison and post-veraison of the same year for mean berry weight. Yields were simulated with a model efficiency (EF) of 0.62 (NRMSE = 0.28). Bud fertility and number of berries per bunch were more accurately simulated (EF = 0.90 and 0.77, NRMSE = 0.06 and 0.10, respectively) than berry weight (EF = -0.31, NRMSE = 0.17). Model efficiency on the on-farm plots reached 0.71 (NRMSE = 0.37) simulating yields from 1 to 8 kg/plant. The GraY model is an original model estimating grapevine yield evolution on the basis of water availability under future climatic conditions.  It allows to evaluate the effects of various adaptation levers such as planting density, cover crop management, fruit/leaf ratio, shading and irrigation, in various production contexts.

Currently, the main goal of agriculture is to promote the resilience of agricultural systems in a sustainable way through the improvement of use efficiency of farm resources, increasing crop yield and quality under climate change conditions. This last is expected to drastically modify plant growth, with possible negative effects, especially in arid and semi-arid regions of Europe on the viticultural sector. In this context, the monitoring of spatial behavior of grapevine during the growing season represents an opportunity to improve the plant management, winegrowers’ incomes, and to preserve the environmental health, but it has additional costs for the farmer. Nowadays, UAS equipped with a VIS-NIR multispectral camera (blue, green, red, red-edge, and NIR) represents a good and relatively cheap solution to assess plant status spatial information (by means of a limited set of spectral vegetation indices), representing important support in precision agriculture management during the growing season. While differences between UAS-based multispectral imagery and point-based spectroscopy are well discussed in the literature, their impact on plant status estimation by vegetation indices is not completely investigated in depth. The aim of this study was to assess the performance level of UAS-based multispectral (5 bands across 450-800nm spectral region with a spatial resolution of 5cm) imagery, reconstructed high-resolution satellite (Sentinel-2A) multispectral imagery (13 bands across 400-2500 nm with spatial resolution of <2 m) through Convolutional Neural Network (CNN) approach, and point-based field spectroscopy (collecting 600 wavelengths across 400-1000 nm spectral region with a surface footprint of 1-2 cm) in a plant status estimation application, and then, using Bayesian regularization artificial neural network for leaf chlorophyll content (LCC) and plant water status (LWP) prediction. The test site is a Greco vineyard of southern Italy, where detailed and precise records on soil and atmosphere systems, in-vivo plant monitoring of eco-physiological parameters have been conducted.

In viticulture, the challenges of local climate modelling are multiple: taking into account the local environment, fine temporal and spatial scales, reliable time series of climate data, ease of implementation and reproducibility of the method. At the local scale, recent studies have demonstrated the contribution of spatialization methods for ground-based climate observation data considering topographic factors such as altitude, slope, aspect, and geographic coordinates (Le Roux et al, 2017; De Rességuier et al, 2020). However, these studies have shown questions in terms of the reproducibility and sustainability of this type of climate study. In this context, we evaluated the potential of MODIS thermal satellite images validated with ground-based climate data (Morin et al, 2020). Previous studies have been encouraging, but questions remain to be explored at the regional scale, particularly in the dynamics of the massive use of bioclimatic indices to classify the climate of wine regions. The results at the local scale were encouraging, but this approach was tested in the current study at the regional scale. Several objectives were set: 1) to evaluate the downscaling method for land surface temperature time series, 2) to identify regional thermal structure variations. We used weekly minimum and maximum surface temperature time series acquired by MODIS satellites at a spatial resolution of 1000 m and downscaled at 500 m using topographical variables. Two types of analyses were performed:

homogeneous zones by intersecting several partial zonings of major factors that influence vineyard growth. Each of them follows specific process from their corresponding disciplines. Soil zoning specifically refers to a Soil Resource Inventory map that has traditionally been generated by conventional soil mapping methods. These methods have shortcomings in reaching fine cartographic and categorical details and involve significant expenses, which undermines their applicability. A new framework named Digital Soil Mapping has introduced quantitative models by statistical techniques to establish soil-landscape relationships and is able to provide intensive scale cartography.

In the present study, a microzoning at 1:10.000 scale is generated from an initial zoning, where the conventional soil map with polytaxic map units is replaced by a new one from digital techniques that disaggregates them. The comparison between the zonings considers a quantitative evaluation of capability for each Homogeneous Terroir Unit by means of the Viticultural Quality Index and its categorization based on its distribution by map. The spatial intersection of both maps gives rise to a confusion matrix in which the flows of class variations after the substitution are assessed.

The results show a five-fold increase in the number of Homogeneous Terroir Units identified and a larger differentiation among them, evidenced by a wider range in the capability index distribution. Both elements are accompanied by an increase in the detection of areas of higher potential within previously undervalued uniform zones.These features are a direct effect of the improvements brought by Digital Soil Mapping techniques and would verify the advantages of their implementation in the Integrated Terroir zoning. Eventually, such new highly detailed terroir units would benefit precision viticulture and sustainable management practices.

A large varietal collection including over 1700 varieties was maintained in Conegliano, ITA, since the 1950s. Phenological data on a subset of 400 grape varieties including wine grapes, table grapes, and raisins were acquired at bud break, flowering, veraison, and ripening since 1964. Despite the efforts in maintaining and acquiring data over such an extensive collection, the data set has varying degrees of missing cases depending on the variety and the year. This is ubiquitous in phenology datasets with significant size and length. In this work, we evaluated four state-of-the-art methods to estimate missing values in this phenological series: k-Nearest Neighbour (kNN), Multivariate Imputation by Chained Equations (mice), MissForest, and Bidirectional Recurrent Imputation for Time Series (BRITS). For each phenological stage, we evaluated the performance of the methods in two ways. 1) On the full dataset, we randomly hold-out 10% of the true values for use as a test set and repeated the process 1000 times (Monte Carlo cross-validation). 2) On a reduced and almost complete subset of varieties, we varied the percentage of missing values from 10% to 70% by random deletion. In all cases, we evaluated the performance on the original values using normalized root mean squared error. For the full dataset we also obtained performance statistics by variety and by year. MissForest provided average errors of 17% (3 days) at budbreak, 14% (4 days) at flowering, 14.5% (7 days) at veraison, and 17% (3 days) at maturity. We completed the imputations of the Conegliano dataset, one of the world’s most extensive and varied phenological time series and a steppingstone for future climate change studies in grapes. The dataset is now ready for further analysis, and a rigorous evaluation of imputation errors is included.