This work proposes a methodology using temporal data and domain knowledge in ord er to analyze a complex agronomical feature, namely the influence of vine water deficit on grape quality. Raw temporal data are available but they are not direc tly usable to estimate vine water deficit. The methodology associates advanced t echniques in computer science and statistics. A preliminary step is required to determine if the amount of water effectively used by the vine is sufficient or n ot. This step necessitates an ecophysiological model, based on expertise. The ex pertise is first formalized in an ontology, under the form of concepts and relat ionships between them, and then used in conjunction with raw data and mathematic al models to design a software sensor. Next the software sensor outputs are put in relation to product quality, assessed by quantitative measurements. This rela tion is analyzed by regression trees and advanced data analysis methods, such as functional data regression. The methodology is applied to a case study involving an experimental design in F rench vineyards. The temporal data consist of sap flow measurements, and the go al is to explain fruit quality parameters (sugar concentration and weight), usin g vine's water variations at key stages of vine phenological development. The results are discussed, as well as the method genericity and robustness.
