Quantifying terrestrial and aquatic ecosystem methane emissions with process-based biogeochemistry and atmospheric transport and chemistry models

To improve the quantification of methane emissions from Arctic wetlands and lakes, an integrated modeling framework was developed. It includes a newly developed process-based lake biogeochemical model and a widely used 4-D VAR inversion algorithm implemented with the nested grid high-resolution GEOS-Chem Adjoint model. The new process-based lake biogeochemical model includes the processes of methane production, oxidation and transport within a one-dimensional sediment and water column. The model is validated using observational data from five lakes located in Siberia and Alaska, representing a large variety of environmental conditions in the Arctic. The modeled lake temperature, dissolved methane concentration and methane fluxes agree well with the observations. It is found that bubbling-rate-controlling nitrogen stripping is the most important factor in determining methane fraction in bubbles. Lake depth and ice cover thickness in shallow waters are also controlling factors. It demonstrates that the thawing of Pleistocene-aged organic-rich ice complex (yedoma) fuels sediment methanogenesis by supplying a large amount of labile organic carbon, resulting in high methane effluxes at thermokarst margins of yedoma lakes. By using the developed process-based lake biogeochemical model with geographical datasets, it is estimated that the annual mean methane emissions from Arctic lakes are on average 11.86 Tg yr-1 during 2004-2009, which is in the range of the recent estimates of 7.1-17.3 Tg yr-1 and is on the same order of the methane emissions from northern high-latitude wetlands. The methane emission rate varies spatially over high latitudes from 110.8 mg CH4 m-2 day-1 in Alaska to 12.7 mg CH4 m-2 day-1 in northern Europe. Under Representative Concentration Pathways (RCP) 2.6 and 8.5 future climate scenarios, when assuming the distribution of lakes unchanged, the methane emissions from Arctic lakes will increase by 10.3 and 16.2 Tg CH4 yr-1, respectively, by the end of the 21st century. By adapting a region-specific landscape evolution model to a pan-Arctic scale, the evolution of thaw lakes in the Arctic can be simulated. The simulations show that the extent of thaw lakes expands throughout the century in the northern areas of the pan-Arctic where the reworking of epigenetic ice in drained lake basins will continue. Coupling with the developed process-based lake biogeochemical model, it is projected that the methane emissions from Arctic lakes by 2100 are 28.3±4.5 Tg CH4 yr-1 under RCP 2.6 and 32.7±5.2 Tg CH4 yr-1 under RCP 8.5, which are about 2.5 and 2.9 times of the simulated present-day emissions. Most of the emitted CH4 originates from non-permafrost carbon stock. For permafrost carbon, the cumulative amount mineralized via methanogenesis is projected to be 3.4±0.8 Pg C under the weak warming condition and 3.9±0.9 Pg C under the strong warming condition. Although the lost permafrost carbon represents a small fraction of the global soil carbon pool, the increased CH4 emissions from pan-Arctic lakes could raise global atmospheric CH4 concentrations as large as 69 ppb. To constrain Arctic methane fluxes, a nested-grid high-resolution inverse model in the Arctic domain is used to assimilate both high-precision surface measurements and high-volume satellite measurements. The global inversions indicate that the total methane fluxes and individual wetland source are in the range of 496.4–511.5 Tg yr-1 and of 130.0–203.3 Tg yr-1, respectively, which are consistent with the other estimates. The estimated Arctic methane fluxes are in the range of 8.8–20.4 Tg yr-1. The optimized methane fluxes from Arctic lakes are ∼7.6 Tg yr-1, a significant amount to the Arctic methane cycle. The global and Arctic inversions of methane mixing ratio in boundary layer and free troposphere are compared well with the observed data