This thesis concerns the analysis of simulation data generated by the agrohydrological model TNT. Our objective is to develop analytical methods for massive simulation results. We want to place the user at the heart of the decision-making process, while letting him handle and analyze large amounts of data in a very efficient way. Our first contribution is an original approach N-Catch, relying on interactive multidimensional analysis methods for archiving simulation results in a decisional database (i.e. data warehouse) adapted to the spatio-temporal nature of the simulation data. In addition, we suggest to analyze the simulation data with online analytical methods (OLAP) to provide strategic information for stakeholders to improve the decision making process. Our second contribution concern two methods for computing skyline queries in the context of data warehouses. These methods enable stakeholders to formulate new questions by combining conflicting environmental criteria, to find compromise solutions associated with their expectations, and to exploit the stakeholder preferences to identify and highlight the data of potential interest. The first method EC2Sky, focuses on how to answer efficiently and progressively skyline queries in the presence of several dynamic user preferences despite of large volume of data. The second method HSky, extends the skyline computation to hierarchical dimensions. It allows the user to navigate along the dimensions hierarchies (i.e. specialize / generalize) while ensuring the online computation of associated skylines. Finally, we present the application of our proposals for managing agricultural practices to improve water quality in agricultural watersheds. We propose a coupling between the agro-hydrological data warehouse model N-Catch and the proposed skyline computation methods.
