Process-based modelling of N losses from terrestrial ecosystems

Abstract

Nitrogen (N) loss to atmosphere and ground water can contribute to environmental destruction, by N emission as the greenhouse gas nitrous oxide (N2O) and by leaching of N to contamination and eutrophication of water. Dynamic models are important keys to increase the knowledge of ecosystems N processes. This thesis studied N fluxes, particularly soil N dynamics, in different ecosystems in relation to interactions between physical and biological processes using a process-based model called the CoupModel. A sub-model implementation of two important processes contributing to nitrogen oxide (NO), N2O, and dinitrogen gas (N2) emissions; nitrification and denitrification, from another process-based model (PnET-N-DNDC) was done. The models were applied on an old spruce forest in Germany and both model outputs of NO and N2O emissions were compared. After further development of the nitrification and denitrification sub-model, the CoupModel successfully described the general pattern of N dynamics. The CoupModel was thereafter applied to an organic crop rotation system. The model generated a complete N budget with small uncertainty ranges for simulated N inputs and outputs, but with a trade-off on estimated differences of N storage in the soil; why it could not be conclusively determined whether the system acted as a source or sink for N. As an outcome of the simulation, almost one fifth of the parameters that were calibrated could be identified as being either system- or field-dependent and had a significant impact on the simulation for nitrate leaching and N2O emission data. We used the CoupModel to evaluate N2O emission sensitivity to soil pH on an organic soil in a birch forest. By generating a decrease of the N2:N2O ratio in the CoupModel, we successfully described the increased N2O emission with decreasing soil pH. The simulated dynamic of N2O emission generated in a time delay compared to measurements, may be related to uncertainties in the model description of the groundwater dynamics and other processes involved. However, the CoupModel robustly managed to quantify total carbon and nitrogen budgets. Finally, we applied the CoupModel on data from ten different locations in Sweden with different N deposition and meteorological data. These were investigated for different drainage depth and initial soil N conditions. The soil differences had large impacts on both N mineral leaching and N denitrification. An application of selected parameter sets was tested; revealing that the uncertainty range of parameter values had impact on simulated N budgets, most notable so for the denitrification pool.