National-scale methane emissions in South Korea (2010–2021): insights from multiple inversion systems

Accurate estimation of methane (CH₄) emissions is essential for assessing mitigation progress, 
yet substantial uncertainties persist at the national scale. In South Korea, CH₄ emissions are 
predominantly anthropogenic, with the waste and agricultural sectors contributing 
approximately 82% of total national emissions. This study analyzes national-scale CH₄ 
emission estimates for South Korea during 2010–2021 using multiple atmospheric inversion 
systems participating in the Methane Inversion Inter-Comparison for Asia (MICA) project. 
Results from inversions using only in situ observations indicate that prior emissions over South 
Korea were likely overestimated. Prior estimates range from 1.5 to 1.7 Tg yr⁻¹ for most years, 
whereas posterior emissions are, on average, about 15% lower than the prior estimates. A 
notable exception is the LMDZ inversion model, which yields posterior estimates that are 40
67% lower than prior values. This substantial reduction is primarily associated with the waste 
sector. Sectoral attribution reveals substantial inter-model differences. LMDZ shows a 
decreasing waste-sector emission trend in Exp. 1 but an increasing trend when only satellite 
observations are assimilated (Exp. 2), whereas the STILT-based inversion consistently 
indicates increasing waste-sector emissions. Given that the waste sector dominates national 
CH₄ emissions, these discrepancies strongly influence total emission estimates. The prior 
waste-sector emissions, derived from EDGAR v7, exceed those reported in South Korea’s 
national greenhouse gas inventory (GIR), contributing to the observed overestimation. 
Additionally, the inversion-derived posterior estimates consistently indicate an overestimation 
of prior agricultural emissions during the summer months. Model performance evaluation over 
the region of interest indicates varying levels of agreement between simulated and observed 
CH₄ mole fractions, with correlation coefficients ranging from 0.24 to 0.85 and posterior biases 
ranging from −65.6 to 0.34 ppb, highlighting the choice of transport model is important. Overall, 
this study highlights the value of multi-model inversion inter-comparisons for constraining 
national-scale CH₄ emissions, diagnosing sector-specific uncertainties, and identifying 
structural differences among inversion frameworks that can guide future improvements.