The preservation of water resources is of major concern in the Charente river basin, SW France. To help the local managers to reconquer a good biological status for water resources and to preserve a quality of the resource suitable for drinking water supply, we developed a method of assessing the environmental impact of alternative scenarios of agricultural practices. This integrated assessment combines typologies of agricultural systems on the area, the calculation of environmental indicators, an integrated modeling using an agro-hydrological model (SWAT) and an economical one in order to assess the cost- effectiveness of some spatialised scenarios. This method is applied to a protected area, the AAC Coulonge St Hippolyte (360 000 ha). In this context, the data warehousing is used to create decision- making tools and to evaluate the potential impacts of the evolution of agriculture on the area. The goal is that people in charge of public policies could take more informed decisions when managing sensitive areas or implementing mitigation measures. Different methods of clustering and spatial analysis, combined to surveys among farmers and experts are used to define a relevant typology of agricultural systems, using a specific approach by type of soil and rotations. Some alternative scenarios for agriculture changes are built using expertise and discussion with stakeholders. The effectiveness of mitigation measures on the reduction of pesticides transfers to freshwaters is assessed when applied on the whole area or within sub-basins. The scenarios can be combined and applied on a range of selected sub-basins or in sensitive areas. However, this approach requires an effective information system that can process for each scenario all the relevant information at embedded scales: the scale of the sub-basins for action by farmers and the scale of the whole area for public decision making. To this end, an environmental information system is created using spatial data warehouse technology. This system allowed qualifying agricultural activities, depending on each scenario, along with river basin characteristics and the indicators resulting of the assessment: values of the environmental indicators, fluxes and costs estimated by the two models. The axes of analysis allowed providing results at different levels of integration, for several dimensions e.g. time, spatial scales or pesticide type, leading to a clearer understanding of the impact of mitigation measures. Various hypotheses can be tested, helping decision-makers respond to special requests from stakeholders.