Natural wetland methane emissions simulated by ICON-XPP

Natural wetlands emit roughly one third of the global methane emissions, yet their contribution remains highly uncertain. To reduce this uncertainty, we incorporated a fully coupled wetland-hydrology and wetland-methane module into the ICON-XPP Earth-system model and performed a suite of coupled model experiments with prescribed sea surface temperatures spanning 1855–2014. Wetland extent is diagnosed online with a TOPMODEL-based wetland scheme that exploits high-resolution topographic indices, while methane production follows a temperature-dependent anaerobic decomposition formulation calibrated to recent global budgets. The baseline simulation yields a time-averaged natural wetland methane flux of 182 (154–205) Tg(CH 4 ) yr −1 for the recent period 2000–2012, in line with the multi-model mean of the global methane budget. Sensitivity experiments reveal that: CO 2 fertilisation is the dominant driver of the observed 12 % increase in emissions since the late 19th century; fixing terrestrial CO 2 at preindustrial levels reverses the trend, producing a net decline in methane flux. Anthropogenic drainage of croplands curtails potential wetland area, especially in the northern extratropics, and offsets part of the CO 2 -driven rise in emissions. Experiments that relax drainage or freeze historical land-use increase wetland area by 0.2–0.6 Mm 2 and raise emissions. The storage of surface water strongly modulates regional precipitation recycling in a coupled model setup, thus significantly altering precipitation in comparison to uncoupled offline experiments. Overall, the coupled model reproduces observed wetland extents and methane emissions, but suggests that current Earth-system models likely underestimate human-induced hydrological alterations, particularly drainage and water-management practices. Our results underscore the necessity of interactive land-atmosphere coupling and refined hydrological representations for robust projections of future wetland methane feedbacks.