Temporal variation of ecosystem scale methane emission from a boreal fen in relation to common model drivers

Models for calculating methane emission from wetland ecosystems typically relate the methane emission to carbon dioxide assimilation. Other parameters that control emission in these models are e.g. peat temperature and water table position. Many of these relations are derived from spatial variation between chamber measurements by space-for-time approach. Continuous longer term ecosystem scale methane emission measurements by eddy covariance method provide us independent data to assess the validity of the relations derived by space-for-time approach.We have analyzed eleven-year methane flux data-set, measured at a boreal fen, together with data on environmental parameters and carbon dioxide exchange to assess the relations to typical model drivers. The data was obtained by the eddy covariance method at Siikaneva mire complex, Southern Finland, during 2005-2015. The methane flux showed seasonal cycles in methane emission, with strongest correlation with peat temperature at 35 cm depth. The temperature relation was exponential throughout the whole peat temperature range of 0-16°C. The methane emission normalized to remove temperature dependence showed a non-monotonous relation on water table and positive correlation with gross primary production (GPP). However, inclusion of these as explaining variables improved algorithm-measurement correlation only slightly, with r2=0.74 for exponential temperature dependent algorithm, r2=0.76 for temperature - water table algorithm, and r2=0.79 for temperature - GPP algorithm. The methane emission lagged behind net ecosystem exchange (NEE) and GPP by two to three weeks. Annual methane emission ranged from 8.3 to 14 gC m-2, and was 20 % of NEE and 2.8 % of GPP. The inter-annual variation of methane emission was of similar magnitude as that of GPP and ecosystem respiration (Reco), but much smaller than that of NEE. The interannual variability of June-September average methane emission correlated significantly with that of GPP indicating a close link between these two processes in boreal fen ecosystems.