Methane emissions from rice paddies : experiments and modelling

This thesis describes model development and experimentation on the comprehension and prediction of methane (CH ₄ ) emissions from rice paddies. The large spatial and temporal variability in CH ₄ emissions and the dynamic non-linear relationships between processes underlying CH ₄ emissions impairs the applicability of empirical relations. Mechanistic concepts are therefore starting point of analysis throughout the thesis.The process of CH ₄ production was investigated by soil slurry incubation experiments at different temperatures and with additions of different electron donors and acceptors. Temperature influenced conversion rates and the competitiveness of microorganisms. The experiments were used to calibrate and validate a mechanistic model on CH ₄ production that describes competition for acetate and H ₂ /CO ₂ , inhibition effects and chemolithotrophic reactions. The redox sequence leading eventually to CH ₄ production was well predicted by the model, calibrating only the maximum conversion rates.Gas transport through paddy soil and rice plants was quantified by experiments in which the transport of SF ₆ was monitored continuously by photoacoustics. A mechanistic model on gas transport in a flooded rice system based on diffusion equations was validated by these experiments and could explain why most gases are released via plant mediated transport. Variability in root distribution led to highly variable gas transport.Experiments showed that CH ₄ oxidation in the rice rhizosphere was oxygen (O ₂ ) limited. Rice rhizospheric O ₂ consumption was dominated by chemical iron oxidation, and heterotrophic and methanotrophic respiration. The most abundant methanotrophs and heterotrophs were isolated and kinetically characterised. Based upon these experiments it was hypothesised that CH ₄ oxidation mainly occurred at microaerophilic, low acetate conditions not very close to the root surface. A mechanistic rhizosphere model that combined production and consumption of O ₂ , carbon and iron compounds with iron adsorption kinetics and diffusive transport in a rice plant and rhizosphere, confirmed this hypothesis. Oxidation of CH ₄ depended on acetate and O ₂ concentrations and on variables influencing competition between methanotrophs and chemical iron oxidation. Oxidation of CH ₄ also depended on root growth dynamics and was intrinsically dynamic.The process-based concepts were simplified in a field scale model on CH ₄ emissions by dividing a rice paddy into a rhizosphere compartment and a bulk soil compartment. The field scale model was validated by independent CH ₄ emission measurements from fields in the Philippines, China and Indonesia in different seasons and with different inorganic and organic fertiliser additions. The model predicted CH ₄ emissions well with only few generally available site-specific input parameters. A sensitivity analysis showed that the model was very sensitive to the description of substrate supply.The field scale model was coupled to a Geographic Information System to scale up regional CH ₄ emissions from rice paddies, as was the aim of the overall project. Regional CH ₄ emission estimates were however affected by the applied interpolation technique and by data resolution effects in a case study for the island of Java, Indonesia. The scaling effects were induced by the combination of a loss of information on heterogeneities and by non-linear model responses. Data availability and not model uncertainty, which was small for the field scale model developed in this thesis, limits upscaling of CH ₄ emissions from rice paddies to regions.