Wine Cloud project is the first "Big Data" platform on the french viticulture value chain. The aim of this platform is to provide a complete traceability of the life cycle of the wine, from the wine-grower to the consumer. In particular, Wine Cloud may qualify as an agricultural decision platform that will be used for vine life cycle management in order to predict the occurrence of major risks (vine diseases, grape vine pests, physiological risks, fermentation stoppage, oxidation of vine, etc...). Also to make wine production more rational by offering winegrower a set of recommendation regarding their strategy's of production development. The proposed platform "Wine Cloud" is based on heterogeneous sensors network (agricultural machines, plant sensors and measuring stations) deployed throughout a vineyard. These sensors allow for capturing data from the agricultural process and remote monitoring vineyards in the Internet of Things (IoT) era. However, the sensors data from different source is hard to work together for lack of semantic. Therefore, the task of coherently combining heterogeneous sensors data becomes very challenging. The integration of heterogeneous data from sensors can be achieved by data mining algorithms able to build correlations. Nevertheless, the meaning and the value of these correlations is difficult to perceive without highlighting the meaning of the data and the semantic description of the measured environment. In order to bridge this gap and build causality relationships form heterogeneous sensor data, we propose an ontology-based approach, that consists in exploring heterogeneous sensor data (light, temperature, atmospheric pressure, etc) in terms of ontologies enriched with semantic meta-data describing the life cycle of the monitored environment.