A probabilistic model for sustainable wine growing

Objectively evaluating the quality of a vineyard in the context of climate change is not always simple. Bayesian networks are widely used for knowledge representation and reasoning under uncertainty in natural resource management. There is a rising interest for this methodology as tools for ecological and agronomic modelling. We designed a probabilistic model that takes into account the parameters defining the status of a vineyard with their associated interactions. No such model has been developed before. It includes an inference engine and software. Data were collected from vine-growing experts. The model includes a database with more than 660 grape varieties. For climate, our model uses a classification method (Tonietto and Carbonneau, 2004) involving multivariate measurements of climate on the basis of three indices: heliothermal index (HI), cool night index (CI), and dryness index (DI). Our model should ease assessments of the likely impact of the choices and decisions of vine growers on the quality of new vineyards to be planted. Thanks to this mathematical model, any kind of simulation of climate change based on the international indexes can be performed. Some examples will be presented. Same thing concerns a primary evaluation of models of sustainable Viticulture. The general frame of the GiESCO charter of sustainable Vitiviniculture is reminded on that occasion.