Feng et al. performed a high-resolution inversion to quantify methane emissions from China. The authors have employed nested WRF-CMAQ simulations, with particular fine resolution (9 km) over Shanxi, the major coal mining province in China. They found that the prior inventory severely overestimates coal mining emissions in North China. The study is overall well executed and clearly presented. I am happy to recommend publication of the manuscript after the following comments are addressed, mainly regarding clarification of the methodology. 

Main comments:

1. The study discussed methane emissions from different sectors, but how the sector partitioning is done is not described. I'd suggest authors to provide further methodological details.

2. The authors have applied TROPOMI XCH4 L2 data. An earlier version of the TROPOMI data have shown substantial regional biases over East China, which may cause errors in the inversion. It would be good if the authors can have some discussion or conduct evaluation on this issue, for instance, using TCCON sites in China.  

3. I do appreciate that the authors have performed evaluation for meteorology parameters against independent data, which most of existing studies have not done. This is crucial for characterizing model transport errors and understanding the difference between inversion systems. However, the discussion is overly simple. I'd suggest the authors to expand the results on meteorology evaluation (especially wind). In particular, the evaluation over the D02 domain provides crucial information because of the complex terrain in Shanxi.

4. The authors used the optimized emissions from the D01 inversion as prior emissions for the D02 inversion. This implies that the observations over D02 are used twice in the optimization of emissions. From the Bayesian standpoint, this is problematic as it leads to over-confidence in observations. 

5. The paper in general lacks uncertainty characterization for emission flux estimates. For regions with limited observation coverage (e.g., Southern China), it is unclear to what degree the posterior estimates depend on prior estimates.

Minor suggestions:

L80: key source-> "point source scale" or "local scale"?

L95: unclear -> uncertain

L103: To my knowledge, IMI is not an operational inversion system, but more like open-source software. So it may be improper to characterize it as a US system. Similar issues may exist for other listed systems.

L188-189: Any quantitative estimates how much error it will incur for D01 and for D02 respectively, by deactivating the chemical oxidation? 

L230: How do you specify the R matrix? Also explain specifically that R is an error covariance matrix for what.

L232: Ep: Power plant sources? Seems something copied from a CO2 study.

L234: No need to capitalize O in oil

L251: Would 1 day be too short for adequate observation constraint, if you assume that prior errors are independent from one day to the next (L272-273)? 

Table 1:  What do the last two columns (building, mature) stand for? 

Table 1 and related discussion (e.g., L360): EDGAR v8.0 is used as prior information. Recent studies have shown that EDGAR has large errors in the spatial and seasonal distribution in rice emissions (Chen et al., 2025; Liang et al., 2024). I'd suggest the authors to briefly discuss the impact on emission quantification and sector attribution in Northeast and East China. 

Chen et al.: Global Rice Paddy Inventory (GRPI): a high-resolution inventory of methane emissions from rice agriculture based on Landsat satellite inundation data, Earth's Future, 2025.

Liang et al.: Satellite-based Monitoring of Methane Emissions from China's Rice Hub, Environmental Science & Technology, 2024. 

Table 2: Just a comment: The comparison with local observations, which are sensitive to emission adjustment, is valuable. 

This study presents a high-resolution CH₄ emission inversion system using the WRF-CMAQ model coupled with the ensemble Kalman filter assimilation method. It assimilated high-resolution XCH₄ concentrations from TROPOMI to estimate CH₄ emissions over China and Shanxi Province. The research highlights that the prior inventory significantly overestimates CH₄ emissions from coal and waste sectors in China, and reveals seasonal characteristics of sectoral CH₄ emissions. Additionally, the research innovatively adjusts the model's boundary conditions based on TROPOMI XCH₄ retrievals, and discusses the impact of boundary condition uncertainties on regional CH₄ emission inversion through sensitivity experiments. The scope and quality of this work are suitable for publication in ACP , and I recommend the manuscript be accepted after minor revisions based on the following comments.

1.    Lines 33–35 and 548–550: Regarding the conclusion that this study's top-down emission inversion is lower compared to others, the authors mainly attribute this to the higher resolution used here, which captures finer emission details—a point supported by comparing results from the inner and outer nested domains. While this explanation is reasonable, it may not be entirely sufficient. Other factors like differences in methods, models, and observations—including how chemical processes, soil sinks, and boundary conditions are represented—could also contribute significantly. It would be good if the authors could briefly discuss these aspects in the Discussion.

2.    Figure 1b: It shows that TROPOMI data has a high missing rate across the country, especially in the south, with some areas having coverage for only 10% of the dates. For regions with long periods of no observation, how are the posterior emissions represented? Also, what is the impact of this representation on daily，monthly and yearly CH₄ emission estimates? It would be helpful if the authors could discuss this as well.

3.    Figure 3 on page 20 shows the differences in prior and posterior emissions for different sectors. How are the sectors distinguished in the posterior emissions? It is indeed challenging to differentiate sectoral emissions in top-down emission inversion. Typically, sectoral emissions in the posterior are calculated based on the proportional grid emissions from the prior inventory. Does this study use the same method?

4.    The title of Figure 4 needs to specify the time period covered by the data. Is it the average for the entire year of 2022?

5.    On line 509, P24, 'Except for LF site...', this description is incorrect. Not only is the LF site underestimated, but the TY site is also significantly underestimated, as well as the JC site. It is recommended to correct the description."