Quantifying and attributing CO emissions over Central Asia using TROPOMI and explicit observational uncertainty

Abstract. Carbon monoxide (CO) is a crucial atmospheric constituent influencing both air quality and climate. Using TROPOMI CO and HCHO column retrievals within the Model-Free Inversion Estimation Framework (MFIEF), this study quantified daily, gridded CO emissions in Central Asia (Xinjiang-China, Kazakhstan, Kyrgyzstan, and Uzbekistan) for 2019–2024. Results reveal a marked interannual decline of ~ 38 % in mean emissions, accompanied by a weakening of emission hotspots. Seasonal peaks in winter and early spring highlight the roles of heating and industrial demand. Importantly, explicit perturbation-based uncertainty analysis showed that ~ 69 % of grid-level estimates are unreliable if observational uncertainties are ignored or using an overly simplified emissions estimation approach, underscoring the nonlinear propagation of retrieval errors. By integrating coal consumption data, we confirm the consistency between satellite-inferred emissions and bottom-up activity estimates, while also identifying missing sources such as underground coal fires. This study demonstrates the effectiveness of MFIEF in data-scarce regions, provides actionable insights for inventory improvement and mitigation strategies, and highlights the framework's potential extension to CH4 and CO2 retrieval-based emission estimation.