The SWAT model was used to model the impacts of climate, soils and agricultural practices on nitrate flows in a 1310 km 2 catchment in western France. Spatialized data were used for natural features (climate, soil, topography), while agricultural activities, finely represented by crop sequences over 3 years, and their associated cultural practices were mapped by remote sensing. The model was calibrated and validated for discharge and nitrate flows at a gauging station. Results are analyzed with respect to leaching for each crop sequence and for each soil type, as nitrate leaching is highly sensitive to the soil and the crop sequence. The lowest risks were found in clayey soils and the highest in sandy soils and/or in sequences including maize. In collaboration with local stakeholders, five scenarios of alternative practices were simulated to evaluate their consequences for nitrogen flows: reduced fertilization, catch crops, shallow cultivation, no-till with catch crops and filter strips. The impacts of the conversion of a pasture into wheat and rapeseed were also assessed. At the catchment gauging station, our 9-year simulations showed a reduction in nitrate flow of 8% with filters strips, 11% with catch crops, 12% with no-till with catch crops, and 15% with reduced fertilization. Shallow cultivation had no impact on nitrate flow. Inversely, the conversion of temporary pastures, which accounts for 32% of the catchment area, to cereals and rapeseed increased nitrate flow by 18%. The impacts of each scenario varied in accordance with leaching at the parcel scale and with the proportion of area affected by the practice. The results show that modelling can improve our understanding of the impacts of agricultural practices on water quality at different scales.
