Sensitivity analysis of a wetland CH4 emission model based on temperate and arctic wetland sites

Together with water vapour and carbon dioxide (CO2), methane (CH4) is an important greenhouse gas, because of its strong global warming potential of 23×CO2 on a 100year time scale. The atmospheric mixing ratio of CH4 has increased with 151 ± 25%, since pre-industrial times. About 60% of the global CH4 emission is of antropogenous origin. From the natural sources (wetlands, termites, oceans, CH4 seeps and hydrates), the wetland environments are the major natural source of atmospheric CH4 (IPCC, 2001). Moreover, the atmospheric CH4 concentration appears to be strongly linked to climate change during the last 800 000 years (Loulergue et al., 2008). Understanding of feedbacks between climate and wetland CH4 emission, in particular in boreal/arctic regions, is a problem for predicting future climate change (Denman et al., 2007). Wetland CH4 emission is also influenced by land management (e.g., Van Huissteden et al., 2006; Hendriks et al., 2007). With the need to reduce greenhouse gas emissions, the relation between wetland CH4 emission and wetland management may become an important question in the future. Predictive models may contribute to a better understanding of feedbacks between climate and CH4 emission, or the effects of wetland management on CH4 emission (e.g. Petrescu et al., 2009; Petrescu et al., submitted 2009; Hendriks et al., 2007). However, to reduce modelling uncertainty extensive sensitivity testing and uncertainty analysis is required, in particular when models are scaled up from a local to regional or global scale. As yet, existing CH4 emission models have not been subjected to rigorous uncertainty analysis going beyond simple model-data comparisons. Here, we present an uncertainty analysis of a wetland CH4 emission model, based on the GLUE (Generalized Likelihood Uncertainty Estimation) methodology (Lamb et al., 1998; Beven, 2001 and references therein). CH4 emission from wetland soils is essentially the net result of a balance between CH4 production by methanogenic bacteria in anaerobic soil zones, and CH4 oxidation by methanotrophic bacteria in aerated soil zones and in plants. Several process models of wetland soil CH4 emission have been designed (Walter, 2000; Segers and Leffelaar, 2001; Granberg et al., 2001, Segers et al., 2001; Wania, 2007). These papers and references therein give an overview of the processes involved. CH4 is generated by methanogenic bacteria in anaerobic parts of the soil, when other electron acceptors for organic matter oxidation are exhausted or unavailable (nitrate, sulfate, Fe and Mn oxides). The substrate for methanogenesis is mainly derived from labile organic compounds, produced by the roots of the wetland vegetation. In wetlands very rapid transfer (1–2 days) of photosynthesis products to CH4 has been observed (King and Reeburgh, 2002). The two major reaction pathways for methanogenesis are CO2 reduction and acetate splitting (e.g., Bréas et