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…

Inhibition of Oenococcus oeni during alcoholic fermentation by a selected Lactiplantibacillus plantarum strain

The use of selected cultures of the species Lactiplantibacillus plantarum in Oenology has grown in prominence in recent years. While initial applications of this species centred very much around malolactic fermentation (MLF), there is strong evidence to show that certain strains can be harnessed for their bio-protective effects. Unwanted spontaneous MLF during alcoholic fermentation (AF), driven by rogue Oenococcus oeni, is a winemaking deviation that is very difficult to manage when it occurs. This work set out to determine the efficacy of one particular strain of Lactiplantibacillus plantarum(Viniflora® NoVA™ Protect), against this problem in Cabernet Sauvignon must. The work was carried out at commercial scale and in a winery environment and compared the bio-protective culture with the more traditional approach of reducing must pH by the addition of tartaric acid. The combination of both was also investigated. The concentration of both Oenococcus oeni and Lactiplantibacillus plantarum was determined using qPCR. The adventitious Oenococcus oeni showed the most growth during AF in the control wine, whereas in the wines treated with Lactiplantibacillus plantarum a bacteriostatic effect against this species was observed. This effect was comparable to the wines treated with tartaric acid. This has particular commercial relevance for controlling the flora in musts with high pH, or when the addition of tartaric acid is either not permitted or is prohibitive for other reasons.

Effect of the commercial inoculum of arbuscular mycorrhiza in the establishment of a commercial vineyard of the cultivar “Manto negro

The favorable effect of symbiosis with arbuscular mycorrhizal fungi (AMF) has been known and studied since the 60s. Nowadays, many companies took the chance to start promoting and selling commercial inoculants of AMF, in order to be used as biofertilizers and encourage sustainable biological agriculture. However, the positive effect of these commercial biofertilizers on plant growth is not always demonstrated, especially under field conditions. In this study, we used a commercial inoculum on newly planted grapevines of a local cultivar grafted on a common rootstock R110. We followed the physiological status of vines, growth and productivity and functional biodiversity of soil bacteria during the first and second years of 20 inoculated with commercial inoculum bases on Rhizophagus irregularis and Funeliformis mosseaeAMF at field planting time and 20 non-inoculated control plants. All the parameters measured showed a neutral to negative effect on plant growth and production. The inoculated plants always presented lower values of photosynthesis, growth and grape production, although in some cases the differences did not reach statistical significance. On the contrary, the inoculation supposed an increase of the bacterial functional diversity, although the differences were not statistically significant either. Several studies show that the effect of inoculation with AMF is context-dependent. The non-favorable effects are probably due to inoculation ineffectiveness under complex field conditions and/or that, under certain conditions, AMF presence may be a parasitic association. This puts into question the effectiveness of its application in the field. Therefore, it is recommended to only resort to this type of biofertilizer when the cultivation conditions require it (e.g., very low previous microbial diversity, foreseeable stress due to drought, salinity, or lack of nutrients) and not as a general fertilization practice.

A better understanding of the climate effect on anthocyanin accumulation in grapes using a machine learning approach

The current climate changes are directly threatening the balance of the vineyard at harvest time. The maturation period of the grapes is shifted to the middle of the summer, at a time when radiation and air temperature are at their maximum. In this context, the implementation of corrective practices becomes problematic. Unfortunately, our knowledge of the climate effect on the quality of different grape varieties remains very incomplete to guide these choices. During the Innovine project, original experiments were carried out on Syrah to study the combined effects of normal or high air temperature and varying degrees of exposure of the berries to the sun. Berries subjected to these different conditions were sampled and analyzed throughout the maturation period. Several quality characteristics were determined, including anthocyanin content. The objective of the experiments was to investigate which climatic determinants were most important for anthocyanin accumulation in the berries. Temperature and irradiance data, observed over time with a very thin discretization step, are called functional data in statistics. We developed the procedure SpiceFP (Sparse and Structured Procedure to Identify Combined Effects of Functional Predictors) to explain the variations of a scalar response variable (a grape berry quality variable for example) by two or three functional predictors (as temperature and irradiance) in a context of joint influence of these predictors. Particular attention was paid to the interpretability of the results. Analysis of the data using SpiceFP identified a negative impact of morning combinations of low irradiance (lower than about 100 μmol m−2 s−1 or 45 μmol m−2 s−1 depending on the advanced-delayed state of the berries) and high temperature (higher than 25oC). A slight difference associated with overnight temperature occurred between these effects identified in the morning.

Local ancient grapevine cultivars to face future viticulture

Among the different strategies to cope with the negative impacts of climate change on viticulture, the exploitation of genetic diversity is one of the most promising to adapt to new conditions and maintain wine production and quality. One of the biggest concerns in the context of climate change is to improve water use efficiency (WUE). In this way, the use of genotypes that present a better response to drought and high WUE is a key issue. In this work, physiological performance analysis was conducted to compare the water deficit stress (WDS) responses of local and widespread grapevines cultivars. Leaf gas exchange, water use efficiency (WUE) at different levels (leaf and long-term WUE (∆13C)), leaf osmotic adjustment and other water relations parameters were determined in plants under well-watered and WDS conditions alongside assessment of the levels of foliar hormones concentrations. Results denote that local cultivars displayed better physiological performance under WDS as compared to the widely-distributed ones. he results corroborate the hypothesis that better stomatal control allows increasing leaf WUE under drought as occurred in the local Callet cv.; but the minority local cultivar Escursac cv. showed high WUE under both treatments. In this case, high WUE can be related to maintaining higher photosynthetic activity under drought. The different mechanisms underlying the better performance under WDS and high WUE of minority local cultivars are discussed.

Geospatial trends of bioclimatic indexes in the topographically complex region of Barolo DOCG

Barolo DOCG is an economically important wine producing region in Northwest Italy. It is a small region of approximately 70 km2 gross area. The topography is very complex with steep sloped hills ranging in elevation from below 200 m to 550 m. Barolo DOCG wine is made exclusively from the Nebbiolo grape. Bioclimatic indexes are often used in viticulture to gain a better understanding of broader climate trends which can be compared temporally and geographically. These indexes are also used for identifying potential phenological timing, growing region suitability, and potential risks associated with expected climatic changes. Understanding how topography influences bioclimatic indexes can help with understanding of mesoscale climate behaviour leading to improved decision making and risk management strategies. The average monthly maximum and minimum temperatures, the Cool Night Index, the Huglin Index, and the monthly diurnal range (from July to October) were calculated using data from 45 weather stations within a 40 km radius of the Barolo DOCG growing area between the years 1996 and 2019. Linear and multiple regression models were developed using independent variables (elevation, aspect, slope) extracted from a digital elevation model to identify significant relationships. Bioclimatic indexes were then kriged with external drift using independent variables that showed significant relationships with the bioclimatic index using a 100 m resolution grid. The maximum monthly temperatures and the Huglin Index showed consistent significant negative relationships with elevation in all years. The minimum monthly temperatures showed no relationship with elevation but in some months a small but significant relationship was observed with aspect. Due to the lack of a relationship between minimum monthly temperatures and elevation compared to the significant relationship between maximum monthly temperatures and elevation, monthly diurnal range had a negative relationship with elevation.