Analysis of the uncertainty associated with the estimation of nitrogen losses from farming systems

This work describes the analysis of the uncertainty linked to the annual direct and indirect losses of different nitrogenous compounds at the scale of a group of farms. The nitrogen (N) forms taken into account are: ammonia (NH3), nitric oxide (NO), nitrous oxide (N2O), dinitrogen (N2) and nitrate (NO3). The gaseous N emissions for the different components of the farms are estimated with a selection of adapted emission factors. The NO3 losses at the farm scale are calculated as the difference between the surplus of the farm-gate N balance and the gaseous N emissions.The feasibility of this method and the validity of the estimated NO3 losses have been demonstrated for a group of farms situated in the Naizin catchment area (Brittany, France) [Payraudeau, S., van der Werf, H.M.G., Vertès. F., 2006. Evaluation of an operational method for the estimation of emissions of nitrogen compounds for a group of farms. IJARGE 5 (2–3), 224–246]. Two sources of uncertainty were studied: (i) that associated with the terms of the farm-gate N balance and (ii) that associated with the different N emission factors. A Monte Carlo (MC) approach was used to assess the uncertainty associated with the calculated losses according to the uncertainty in the terms of the farm-gate N balance and in the emission factors used. The nitrogen losses are assessed, amongst others, to contribute to the establishment of an emissions inventory to be used in Life Cycle Assessment (LCA) of farming systems. Therefore, an analysis of the main problems limiting the implementation of the MC approach in LCA studies of agricultural production is proposed. The different stages in the implementation of the MC approach are described. The results reveal a greater sensitivity of emissions of NH3, N2O and N2 to the uncertainty associated with the emission factors than to that associated with the terms of the farm-gate N balance. The two sources of uncertainty contribute similarly to the uncertainty of NO3 losses. In spite of the occasionally large confidence intervals in the factors, the confidence intervals for the different direct losses are reasonably small.