GiESCO 2019 banner
IVES 9 IVES Conference Series 9 GiESCO 9 GiESCO 2019 9 Aroma and quality assessment for vertical vintages using machine learning modelling based on weather and management information

Aroma and quality assessment for vertical vintages using machine learning modelling based on weather and management information

Abstract

Context and purpose of the study ‐ Wine quality traits are usually given by parameters such as aroma profile, total acidity, alcohol content, colour and phenolic content, among others. These are highly dependent on the weather conditions during the growing season and management strategies. Therefore, it is important to develop predictive models using machine learning (ML) algorithms to assess and predict wine quality traits before the winemaking process.

Material and methods ‐ Samples in duplicates of Pinot Noir wines from vertical vintages (2008 to 2013) of the same winery located in Macedon Ranges, Victoria, Australia were used to assess different chemical analytics such as i) aromas using gas chromatography – mass spectrometry, ii) color density, iii) color hue, iv) degree of red pigmentation, v) total red pigments, vi) total phenolics, vii) pH, viii) total acidity (TA), and ix) alcohol content. Data from weather conditions from the specific vintages were obtained both from the bureau of meteorology (BoM) and the Australian Wine Availability Project (AWAP) climate databases. Such data consisted of: i) solar exposure from veraison to harvest (V‐H), ii) solar exposure from September to harvest (S‐H), iii) maximum January solar exposure, iv) degree days from S‐H, v) maximum January evaporation, vi) mean maximum temperature from veraison to harvest, vii) mean minimum temperature from V‐H, viii) water balance from S‐H, ix) solar exposure from V‐H, x) degree hour accumulation with base 25 – 30 °C, xi) degree hour accumulation with base 25 °C, xii) degree hour accumulation with base 30 °C, xiii) degree hour accumulation with base 35 °C, and xiv) total cumulative degree days accumulation with base 10 °C. All data were used to develop two machine learning (ML) regression models using Matlab® R2018b. The best models obtained were using artificial neural networks (ANN) with the Levenberg‐Marquardt algorithm with 5 neurons for Model 1 and 9 neurons for Model 2. Model 1 was developed using the 14 parameters from the weather data as inputs to predict 21 aromas found in the wines from the six different vinatges. Model 2 was developed using the same 14 parameters from weather data and the eight chemical parameters as targets and outputs.

Results ‐ Both models obtained presented very high accuracy to predict wine quality trait parameters. Model 1 had an overall correlation coefficient R = 0.99 with a high performance based on the mean squared error (MSE = 0.01), while Model 2 had an overall correlation coefficient R = 0.98 with a high performance (MSE = 0.03). These models would aid in the prediction of wine quality traits before its production, which would give anticipated information to winemakers about the product they would obtain to make early decisions on wine style variations.

DOI:

Publication date: June 22, 2020

Issue: GiESCO 2019

Type: Article

Authors

Sigfredo FUENTES, Claudia GONZALEZ VIEJO, Xiaoyi WANG, Damir D. TORRICO

School of Agriculture and Food, Faculty of Veterinary and Agricultural Sciences, University of Melbourne, VIC 3010, Australia

Contact the author

Keywords

wine quality, machine learning, weather, aromas

Tags

GiESCO 2019 | IVES Conference Series

Citation

Related articles…

Influence of grapevine rootstock/scion combination on rhizosphere and root endophytic microbiomes

Soil is a reservoir of microorganisms playing important roles in biogeochemical cycles and interacting with plants whether in the rhizosphere or in the root endosphere. The composition of the microbial communities thus impacts the plant health. Rhizodeposits (such as sugar, organic and amino acids, secondary metabolites, dead root cells …) are released by the roots and influence the communities of rhizospheric microorganisms, acting as signaling compounds or carbon sources for microbes. The composition of root exudates varies depending on several factors including genotypes. As most of the cultivated grapevines worldwide are grafted plants, the aim of this study was to explore the influence of rootstock and scion genotypes on the microbial communities of the rhizosphere and the root endosphere. The work was conducted in the GreffAdapt plot (55 rootstocks x 5 scions), in which the 275 combinations have been planted into 3 blocks designed according to the soil resistivity. Samples of roots and rhizosphere of 10 scion x rootstock combinations were first collected in May among the blocks 2 and 3. The quantities of bacteria, fungi and archaea have been assessed in the rhizosphere by quantitative PCR, and by cultivable methods for bacteria and fungi. The communities of bacteria, fungi and arbuscular mycorrhizal fungi (AMF) was analyzed by Illumina sequencing of 16S rRNA gene, ITS and 28S rRNA gene, respectively. The level of mycorrhization was also evaluated using black ink coloration of newly formed roots harvested in October. The level of bacteria, fungi and archaea was dependent on rootstock and scion genotypes. A block effect was observed, suggesting that the soil characteristics strongly influenced the microorganisms from the rhizosphere and root endosphere. High-throughput sequencing of the different target genes showed different communities of bacteria, fungi and AMF associated with the scion x rootstock combinations. Finally, all the combinations were naturally mycorrhized. The root mycorrhization intensity was influenced by the rootstock genotype, but not by the scion one. Altogether, these results suggest that both rootstock and scion genotypes influence the rhizosphere and root endophytic microbiomes. It would be interesting to analyze the biochemical composition of the rhizodeposition of these genotypes for a better understanding of the processes involved in the modulation of these microbiomes. Moreover, crossing our data with the plant agronomic characteristics could provide insights into their roles on plant fitness.

The use of rootstock as a lever in the face of climate change and dieback of vineyard

As viticulture faces challenges such as climate change or vineyard dieback, the choice of the variety and rootstock becomes more and more crucial. To study rootstock levers in the Bordeaux region, a parcel of Cabernet Sauvignon (CS) was planted with four rootstocks in 2014. Twenty repetitions of each of the following four rootstocks were set up: 101-14 MGt, Nemadex AB, 420A MGt and Gravesac. The number of bunches, yields and pruning weights of the vine shoots were measured individually on 240 vines from 2017 to 2021. Since 2020, nitrogen status assessed by assimilable nitrogen level, hydric status assessed by δ13C and berry maturity were measured on 80 samples taken from 20 repetitions of the four rootstocks. A lower yield was measured for CS grafted onto Nemadex AB due to the lower number of bunches and the lower weight of berries. The differences between the other three rootstocks are small, but CS grafted onto 420A MGt was the most productive. The CS grafted onto Nemadex AB had the lowest pruning weight while 101-14 MGt had the highest. In 2020, δ13C showed a more moderate water stress with 101-14 MGt and 420A MGt than with Nemadex AB. Surprisingly, the Gravesac was under more stress than the 101-14 MGt. The nitrogen status in the berries was better for Nemadex AB but this was perhaps due to the significantly lower weight of the berries.Rootstock 101-14 MGt attained the highest accumulation of sugars in the berries while 420A MGt allows to preserve higher acidity. The parcel is still young which may explain some of the results. These measures must therefore be continued over the next several years to fully assess the effects of these rootstocks on the development of the vines and the quality of the production under new climatic conditions.

Mesoclimate impact on Tannat in the Atlantic terroir of Uruguay

The study of climate is relevant as an element conditioning the typicity of a product, its quality and sustainability over the years. The grapevine development and growth and the final grape and wine composition are closely related to temperature, while climate components vary at mesoscale according to topography and/or proximity to large bodies of water. The objective of this work is to assess the mesoclimate of the Atlantic region of Uruguay and to determine the effect of topography and the ocean on temperature and consequently on Tannat grapevine behavior.

Delaying irrigation initiation linearly reduces yield with little impact on maturity in Pinot noir

When to initiate irrigation is a critical annual management decision that has cascading effects on grapevine productivity and wine quality in the context of climate change. A multi-site trial was begun in 2021 to optimize irrigation initiation timing using midday stem water potential (ψstem) thresholds characterized as departures from non-stressed baseline ψstemvalues (Δψstem). Plant material, vine and row spacing, and trellising systems were concomitant among sites, while vine age, soil type, and pruning systems varied. Five target Δψstem thresholds were arranged in an RCBD and replicated eight times at each site: 0.2, 0.4, 0.6, 0.8, and 1.0 MPa (T1, T2, T3, T4, and T5, respectively). When thresholds were reached, plots were irrigated weekly at 70% ETc. Yield components and berry composition were quantified at harvest. To better generalize inferences across sites, data were analyzed by ANOVA using a mixed model including site as a random factor. Across sites, irrigation was initiated at Δψstem = 0.24, 0.50, 0.65, 0.93, and 0.98 MPa for T1, T2, T3, T4, and T5, respectively. Consistent significant negative linear trends were found for several key yield and berry composition variables. Yield decreased by 12.9, 15.9, 19.5, and 27.4% for T2, T3, T4, and T5, respectively, compared to T1 (p < 0.0001) across sites that were driven by similarly linear reductions in berry weight (p < 0.0001). Comparatively, berry composition varied little among treatments. Juice total soluble solids decreased linearly from T1 to T5 – though only ranged 0.9 Brix (p = 0.012). Because producers are paid by the ton, and contracts simply stipulate a target maturity level, first-year results suggest that there is no economic incentive to induce moderate water deficits before irrigation initiation, regardless of vineyard site. Subsequent years will further elucidate the carryover effects of delaying irrigation initiation on productivity over the long term.

Optimizing stomatal traits for future climates

Stomatal traits determine grapevine water use, carbon supply, and water stress, which directly impact yield and berry chemistry. Breeding for stomatal traits has the strong potential to improve grapevine performance under future, drier conditions, but the trait values that breeders should target are unknown. We used a functional-structural plant model developed for grapevine (HydroShoot) to determine how stomatal traits impact canopy gas exchange, water potential, and temperature under historical and future conditions in high-quality and hot-climate California wine regions (Napa and the Central Valley). Historical climate (1990-2010) was collected from weather stations and future climate (2079-99) was projected from 4 representative climate models for California, assuming medium- and high-emissions (RCP 4.5 and 8.5). Five trait parameterizations, representing mean and extreme values for the maximum stomatal conductance (gmax) and leaf water potential threshold for stomatal closure (Ψsc), were defined from meta-analyses. Compared to mean trait values, the water-spending extremes (highest gmax or most negative Ysc) had negligible benefits for carbon gain and canopy cooling, but exacerbated vine water use and stress, for both sites and climate scenarios. These traits increased cumulative transpiration by 8 – 17%, changed cumulative carbon gain by -4 – 3%, and reduced minimum water potentials by 10 – 18%. Conversely, the water-saving extremes (lowest gmax or least negative Ψsc) strongly reduced water use and stress, but potentially compromised the carbon supply for ripening. Under RCP 8.5 conditions, these traits reduced transpiration by 22 – 35% and carbon gain by 9 – 16% and increased minimum water potentials by 20 – 28%, compared to mean values. Overall, selecting for more water-saving stomatal traits could improve water-use efficiency and avoid the detrimental effects of highly negative canopy water potentials on yield and quality, but more work is needed to evaluate whether these benefits outweigh the consequences of minor declines in carbon gain for fruit production.