Sulfur hexafluoride (SF6) is an extremely potent greenhouse gas with a global warming potential (GWP) of ~25,000 over 100 years and a long atmospheric lifetime (1000–3200 years). Its impact on the global climate led to its inclusion in the Kyoto Protocol and the Paris Agreement. Anthropogenic emissions, primarily from high-voltage electrical switchgear, are the main source. Global SF6 mole fractions and emissions have increased rapidly since the 2000s, with reductions in Annex-I countries seemingly offset by increases in non-Annex-I countries, including China. China's high electrical power demand makes it a major contributor. While bottom-up emission inventories exist for China, large discrepancies exist between estimates. Top-down estimates, derived from atmospheric observations, are crucial for validating and improving national inventories. However, previous top-down estimates had limited sensitivity to emissions from regions like western China. This study addresses this gap by using atmospheric observations from a Chinese measurement network and a top-down inverse modeling framework to derive SF6 emissions in China from 2011–2021, comparing them to previous studies and investigating regional variations.

Several bottom-up SF6 emission inventories have been reported for China, based on energy and industrial activity data and emission factors. Officially reported Chinese SF6 emissions to the UNFCCC exist for several years this century, but discrepancies exist between these and other bottom-up estimates (e.g., the US EPA estimate being much lower than other studies). Top-down estimates from atmospheric observations, recommended by IPCC guidelines, have also been conducted, but these used measurements outside China (in South Korea or Japan), limiting their sensitivity to emissions from western China. Previous top-down studies often focused on eastern China and extrapolated to national totals using population density as a proxy, leading to potential inaccuracies.

This study derived SF6 emissions in China (mainland only) from 2011–2021 using atmospheric observations from nine sites within a Chinese measurement network and a top-down inverse modeling framework. The framework comprised atmospheric observations, sensitivities of observations to emissions and boundary conditions (simulated by the UK Met Office NAME model), and a hierarchical Bayesian inference algorithm using prior information from EDGAR v7.0. The NAME model used meteorological fields from the UK Met Office Unified Model, simulating particle back-trajectories to calculate sensitivities. Data filtering excluded observations with sensitivities dominated by stagnant conditions. A hierarchical Bayesian inference algorithm, utilizing an MCMC method (NUTS and slice sampler), solved for emissions, boundary conditions, and uncertainties. Regional inversion performance was assessed using uncertainty reduction, improvements in RMSE, and correlations between simulated and observed mole fractions. Global SF6 emissions were estimated using the AGAGE 12-box model and AGAGE global background observations.

SF6 emissions in China increased substantially from 2.6 (2.3–2.7) Gg yr−1 in 2011 to 5.1 (4.8–5.4) Gg yr−1 in 2021. Top-down estimates from this study are consistent with the EDGAR v7.0 inventory and a recent bottom-up estimate by Guo et al., but substantially larger than the US EPA estimate and earlier Chinese national communications to the UNFCCC. The discrepancy with official reports decreased after 2014, suggesting improvement in China's national inventory methodology. Significant SF6 emissions were identified in western China, likely due to expanding power generation and transmission. The power industry (electricity consumption/production) is a better proxy for provincial SF6 emissions than population or GDP. China's contribution to global SF6 emissions increased from 34% in 2011 to 57% in 2021. The increase in China's emissions nearly doubled the global increase, effectively offsetting emission reductions elsewhere in the world. In 2021, China's SF6 emissions reached 125 (117–132) million tonnes CO2-eq yr−1, comparable to several countries' total CO2 emissions.

The study's findings highlight the significant and increasing contribution of China to global SF6 emissions, largely driven by the expansion of its power industry, particularly in the western regions. The top-down estimates provide valuable validation for bottom-up inventories and reveal previously under-quantified emissions from less-populated regions. The substantial increase in China's emissions offsets reductions achieved in other countries, underscoring the importance of focusing on emission control measures in non-Annex-I countries. The strong correlation between SF6 emissions and the power industry suggests that targeting this sector is key for emission reduction strategies. The comparison with previous top-down studies illustrates the importance of within-country measurements for capturing regional variations and improving accuracy. The high CO2-eq emissions of SF6 in China emphasize its potential to counteract efforts to achieve carbon neutrality.

This study demonstrates a substantial and sustained increase in SF6 emissions from China, significantly impacting global emissions and offsetting reductions elsewhere. The identification of significant emissions from western China highlights the need for comprehensive monitoring across the country. The strong correlation between emissions and the power sector underscores the importance of targeted emission control measures in this sector. Continued monitoring, employing denser networks of atmospheric measurement sites and advanced methodologies, is vital for evaluating the efficacy of future emission control policies and ensuring progress toward carbon neutrality goals.

The study's reliance on atmospheric observations and inverse modeling introduces uncertainties. The accuracy of the model's representation of atmospheric transport and mixing processes affects the precision of the emission estimates. The study focuses on mainland China, excluding Hong Kong and Macau, which might introduce slight biases. The use of prior information from bottom-up inventories could potentially influence the results, although sensitivity tests were performed to mitigate this influence. The assessment of global emissions relies on a global model which may have its own inaccuracies.