Global methane emissions and their variability over two decades: insights from atmospheric inversion modelling

Methane (CH₄) is the second most important greenhouse gas, with its atmospheric levels having more than doubled since pre-industrial times, and exhibiting highly variable growth rates in recent decades. Significant uncertainties remain in the global methane budget, despite ongoing efforts to quantify methane sources and sinks using bottom-up and top-down approaches. In particular, the relative contributions of natural and anthropogenic sources to the accelerated increase in recent years remain unclear. Wetlands are major natural sources of methane and are expected to respond to climate variability and the long-term increase in global temperature. Accurately capturing their temporal variability and long-term trend is a significant challenge to global methane budget assessments. In addition, atmospheric methane loss, which is primarily caused by the oxidation of hydroxyl radicals (OH), is a significant source of uncertainty in atmospheric inverse modelling. Uncertainties in the magnitude and temporal variability of OH directly impact the estimation of methane emissions. Therefore, improving estimates of methane emissions at global to regional scales advances our understanding of methane and its climate feedback.

In this context, we have applied the Jena CarboScope Global Inversion System to estimate global methane surface fluxes from 2000 to 2024. This analysis uses a Bayesian atmospheric inversion framework to optimize surface emissions by combining atmospheric transport modelling with long-term atmospheric methane observations.  Inversions were carried out at a horizontal resolution of approximately 3.8° latitude and 5° longitude. The prior emissions included wetland fluxes (ORCHIDEE model), anthropogenic emissions (EDGAR database), biomass burning emissions (GFEDv4s), as well as other minor source categories (termites, freshwater, geological and the ocean). To evaluate the sensitivity of emission estimates to assumptions about atmospheric loss, we conducted inversions varying the OH fields, such as a climatological OH distribution and an interannually varying OH field.

We present the temporal evolution of global methane emissions over the last two decades, analyze their spatial distribution and regional emission patterns, and evaluate the consistency of the inversion results using independent observational datasets across different regions. Using two representations of atmospheric methane loss, we found that differences exist not only in whether interannual variability is included, but also in mean magnitude and spatial distribution. These differences lead to variations in inferred emission strengths. The largest variations in flux magnitude occur in  North America temperate, South America temperate, and Eurasia temperate regions. Overall, global posterior fluxes tend to be larger than prior fluxes, with model adjustments that are spatially heterogeneous. While there are differences in absolute flux estimates, posterior global methane emissions derived from various inversion setups show consistent interannual variability, with minor differences in variability during the latter part of the period. Overall, this work provides a multi-decadal, top-down perspective on global methane emissions, emphasizing the importance of accounting for uncertainties in atmospheric loss when interpreting methane budgets.