资源环境科技发展态势分析平台

The observed rise in atmospheric methane (CH4) from 375 ppbv during the Last Glacial Maximum (LGM: 21,000 years ago) to 680 ppbv during the late preindustrial era is not well understood. Atmospheric chemistry considerations implicate an increase in CH4 sources, but process-based estimates fail to reproduce the required amplitude. CH4 stable isotopes provide complementary information that can help constrain the underlying causes of the increase. We combine Earth System model simulations of the late preindustrial and LGM CH4 cycles, including process-based estimates of the isotopic discrimination of vegetation, in a box model of atmospheric CH4 and its isotopes. Using a Bayesian approach, we show how model-based constraints and ice core observations may be combined in a consistent probabilistic framework. The resultant posterior distributions point to a strong reduction in wetland and other biogenic CH4 emissions during the LGM, with a modest increase in the geological source, or potentially natural or anthropogenic fires, accounting for the observed enrichment of delta(CH4)-C-13.

Plain Language Summary Methane is the next most important greenhouse gas in the atmosphere after carbon dioxide. Since industrialization, methane has risen from around 680 ppbv (parts per billion volume) to 1,800 ppbv today. Before industrialization, methane levels were dominated by natural processes. The largest recent changes occurred during ice age to interglacial transitions. Measurements on gas bubbles preserved in ice cores show that methane rose with global temperatures, from around 375 ppbv during the ice age, to 680 ppbv prior to the Industrial Revolution. Explaining this amplitude remains a challenge, because there are no measurements of past sources or sinks of methane. Stable isotopes of methane provide additional information, because different sources and sinks impart unique signatures to the methane measured in ice cores. We use these measurements with a computer model of the global methane cycle. We then employ a statistical approach to learn how sources changed, taking account of uncertainties. We find that a near doubling of the wetland methane source is required. This is much greater than the change simulated in this and other methane models. This potentially indicates that methane models are undersensitive, with implications for understanding how the methane cycle will evolve in the near future.