Drouet, Jean-Louis, Duretz, Sylvia, Fiorelli, Jean-Louis, Blanfort, Vincent, Capitaine, Mathieu, Capian, Nicolas, Gabrielle, Benoit, Martin, Raphaël, Lardy, Romain, Cellier, Pierre, Soussana, Jean-François, Environnement et Grandes Cultures (EGC), Institut National de la Recherche Agronomique (INRA)-AgroParisTech, Agro-Systèmes Territoires Ressources Mirecourt (ASTER Mirecourt), Institut National de la Recherche Agronomique (INRA), UR 0874 Unité de recherche sur l'Ecosystème Prairial, Institut National de la Recherche Agronomique (INRA)-Unité de recherche sur l'Ecosystème Prairial (UREP)-Ecologie des Forêts, Prairies et milieux Aquatiques (EFPA), Centre de Coopération Internationale en Recherche Agronomique pour le Développement (Cirad), Agronomie et Fertilité Organique des Sols, Ecole Nationale d'Ingénieurs des Travaux Agricoles de Clermont-Ferrand (ENITAC), Collège de Direction (CODIR), AgroParisTech-Institut National de la Recherche Agronomique (INRA), UAR 0233 Collège de Direction, and Institut National de la Recherche Agronomique (INRA)-Direction Collégiale (DCOLL)-Collège de Direction (CODIR)
Agricultural systems, especially livestock systems, are a large source of emissions of greenhouse gas (N2O, CH4n CO2) and reactive nitrogen (NH3, NOx, Nitrates). Mitigation options suggested in the literature mainly focused on a signle gas and dealt with isolated processes. An intagrated modelling tool was developed to assess the effect of farm practices on reactive nitrogen and greenhouse gas emissions at the whole farm level. It integrated the farm model FARMSIM developed during the GreenGrass project (2002-2004) and using the IPCC methodology to simulate emissions from farm components (housing, manure management, livestock energy consumption). The FARMSIM model was coupled with two process-based models of ecosystems : the cropland model CERES-EGC and the grassland model PASIM. Prior to model use to assess mitigation options, a sensitivity analysis was carried out to reduce the number of input factors to be measured or collected for model calibration. The model inputs were biophysical parameters (e.g., soil and vegetation parameters, emissions factors), farming practices (e.g., N management, grassland management, herd management, farm structure,land use change) and meteorological conditions. Simulated output variables were N2O, CH4, CO2, NH3, emissions. The reference farm was reconstructed from farming practices within an intensive dairy farm located in NOrth-Eastern France (INRA Mirecourt, Lorraine). Several statistical methods were implemented to calssify input factors (Morris method), identify interactions between input factors (rank-based methods) and analyse the uncertainty of simulated variables (Fourier-based method). First results showed the relevance of such mehtods to assess the effect of input factors e.g., soil biophysical properties, farm N management, herd structure on gas emissions. That forms a preliminary study to the assessment of mitigation options integrating national and European policies as well as farmers decisions