Review of Yang et al.

This manuscript offers a robust comparative analysis of six CO₂ emission inventories for China, integrating both local and global datasets. A key strength is its detailed assessment of spatial and temporal uncertainties, an often overlooked but policy-relevant aspect. The study contributes meaningfully by highlighting inventory discrepancies and emphasizing the importance of uncertainty assessments in emission reporting. However, I have the following specific comments that require clarification and revision before the manuscript can be considered for publication.

General comments

The manuscript is clearly written and well structured, with a logical flow that facilitates understanding of the main objectives and findings.

However, it is not entirely clear whether the emission inventories selected for analysis are the only relevant options available, or what criteria guided their selection. Since the manuscript references other inventories that were ultimately not included in the comparison, it would strengthen the study to provide a clearer rationale for the choices made.

The relevance of the topic is evident, especially in light of China’s pivotal role in global emissions and its commitments under the Paris Agreement. Still, the manuscript would benefit from a more explicit explanation of why comparing the latest versions of these inventories is particularly important. A clearer articulation of what distinguishes this study from previous work (beyond simply the version updates) would improve accessibility, especially for readers less familiar with the topic.

The discussion of differences between inventories and their associated uncertainties is engaging and informative. However, a clear take-home message is lacking, particularly regarding which inventories may be considered more reliable or fit for specific purposes. While it is understandable that definitive recommendations may be difficult, the current conclusions are limited, with the mainly strong guidance being to avoid the provincial CEADs inventory. Offering more concrete insights or practical recommendations, especially in the context of supporting policymaking, would significantly strengthen the manuscript.

Specific Comments

Line 35: To highlight China’s role in global emissions, please include the percentage of China’s anthropogenic emissions relative to global totals.

Line 44: The CAMS inventory should be included in this overview for completeness.

Line 48: Are there specific reasons for not including CHRED in the analysis? Please clarify.

Line 80: Consider introducing the CAMS inventory definition earlier in this section alongside the others, for consistency.

Line 80: MEIC is initially described (line 47) as a China-specific inventory, but here it is treated as a global inventory. This inconsistency may confuse readers, particularly since line 116 clarifies that the global version of MEIC is used. Please harmonize these descriptions.

Line 122: The mention of the number of species covered by CAMS is not relevant here, as the analysis focuses on a single species. Also, this level of detail is not provided for the other inventories.

Line 198: Do you have any hypotheses as to why GEMS diverges from the trends observed in other inventories, especially in the residential and commercial sectors?

Table 1: Time Resolution (GEMS column): Please change "Annually" to "Annual" to align with the other entries.

Table 1: Data Source row: Since the "last accessed" date is the same for all inventories, consider moving this note to a table footnote (e.g., marked with an asterisk) to streamline the table.

Figure 3: The growth in electricity and heat production in CAMS appears to stabilize, unlike in other inventories where growth continues. Given CAMS is based on EDGAR, a similar trend would be expected. Could this discrepancy be linked to the use of CAMS-Tempo profiles?

Line 212: It is unclear why MEIC is used as a benchmark for comparison. Please add a brief explanation of this choice.

Figure 5c: What accounts for the squared patches in this figure? A brief explanation in the caption or main text would help readers interpret the results.

Figures 4 & 5: In Figure 4, MEIC shows notable emissions over western China (green shading), while ODIAC does not. This difference should manifest as strong negative values (blue) in Figure 5, yet much of this area appears blank, which I assume represents NaN values. Did you apply any filtering? Please clarify.

Line 241: For clarity, please consider rephrasing this sentence, here is a suggestion:

"Across the spatial domain, EDGAR generally reports lower emissions than MEIC, with negative differences prevailing throughout the region."

Line 287: Could the discrepancy in Shanxi be attributed to a specific sector? A sectoral analysis, as presented in the previous section, would be valuable here.

Line 290: Could you comment on the provincial comparison of the two CEADs estimates beyond Shanxi? Do any provinces show consistent agreement between the two datasets, and are these primarily low-emission regions? A colored map showing the differences between the two CEADs estimates by province could be a useful addition

The paper compares six CO₂ emission inventories for China from 2000 to 2023, including global inventories (ODIAC, EDGAR, GEMS) and China-specific ones (MEIC, CHRED, CEADs). It highlights large differences between inventories, especially EDGAR vs. MEIC, and differences in spatial distributions. This is important because China has ambitious carbon reduction goals, so accurate quantification of CO2 emissions is essential for policy and climate modelling. The paper fits within the journal’s scope as it addresses atmospheric emissions and their uncertainties.

Limitations of this review: I am not an expert in CO₂ emissions inventories and the relevant literature, so my comments focus on interpretation, clarity, and presentation rather than technical accuracy of methods.

Major comments

The paper is well-structured, the argument is easy to follow, and the language is clear. However, the following aspects would need to be addressed before publication.

1. Clarify the novelty of the study

It is unclear how this work differs from previous studies. Is the novelty in using updated versions of inventories, applying new harmonisation methods, or drawing new conclusions? Please add a short paragraph in the introduction explicitly stating what is new compared to other studies mentioned (e.g., Han et al., 2020a; Liu et al., 2015; L. Zheng et al., 2025).

1. Recommendations for users

The conclusion clearly summarises findings but could be strengthened by adding actionable guidance. Readers would benefit from answers to the following questions:

• Which inventories are most reliable for specific applications?
• What are the main uncertainties that remain?
• How can inventory producers improve the next inventory versions?

A summary table of findings and recommendations could make this section more impactful.

1. Comparison to observations

The study compares inventories against each other. Without observational benchmarks, it is difficult to assess which inventory is closer to reality. Could you explain why observational comparisons were not included? If data limitations prevented this, could you state them explicitly and discuss implications for interpreting results?

Specific comments

Line 80: MEIC is described as China-specific but later implied to be global. Could you clarify?

Line 88: You mention standardising inventories to a common grid. Could this process introduce uncertainties? If so, could you quantify or acknowledge them?

Line 174: Each growth phase is described with justification based on context, except from the third phase. Could you explain why emissions increase again after 2016?

Line 215: You explain spatial gaps in ODIAC and explain that they could be due to the inventory relying on night lightning. However, you do not mention other inventories. For example, are the spatial gaps in CAMS likely to be caused by similar reasons?

Figure 4: Why was 2019 chosen as the base year? Would spatial patterns differ significantly in other years?

Figure 5: Inventories are compared to MEIC as a baseline. Could you comment on the existing uncertainties relating to MEIC, and what this means for the results?

Lines 235–240: Inventory users would benefit from specific interpretation for all inventories. For example, why does ODIAC allocate more emissions to areas? Is it related to night lighting again? Why is the CAMS pattern opposite to ODIAC?

Lines 241- 248: You explain that EDGAR has very large extremes in 0.14% of grid cells, likely due to EDGAR allocating emissions aggressively to point sources (and using outdated CARMA). The presence of such large extremes, which strongly influence averages, raises questions about the robustness of EDGAR as an inventory. Should this be a concern for users? Could you clarify how using MEIC as a baseline may influence this result?

Line 299: You find large difference between CEADS (provinces) and CEADS (sectors) for Shanxi. You conclude that sector-level estimates should be prioritised, as the sum of all provinces estimates do no match the national estimates. Could you comment on why province-level estimates are so uncertain, and different from sectors estimates? 

Section 3.3.2: In this section, you analyse timeseries for nine specific provinces, chosen based on a classification. This section currently reads as a descriptive list without a clear narrative or takeaway. Could you clarify the aim e.g., to illustrate provincial heterogeneity between inventories.

Line 361: You state that results are opposite to Han et al. (2020a). Could you give more details about these differences? Why do different versions of inventories give such different results? What does this mean for inventory users?

Line 388: “Ensemble approaches” please define how this method would be used and explain why they would help mitigate biases

General comments:

This manuscript analyzes and compares six bottom-up inventories and assesses their uncertainty. This work compares different inventories from international and domestic teams, and it will be useful to the global stocktake and accurately assess China’s CO₂ emissions. The topic is interesting and meaningful, but many statements and explanations in the manuscripts are not rigorous enough. I suggest more modifications and improvements before acceptance.

Special comments:

• Is it reasonable to use the mean and SD to assess the uncertainty of these emission inventories?
• Activity data and emission factors are the two important factors that influence the emission inventory. I also suggest adding this important information to Table 1, although point, line, and area source proxies are listed.
• The Chinese government also reports national greenhouse gas emissions to the UNFCCC. I think it is better to compare the national CO₂ emissions between government-reported data and the six bottom-up inventory data mentioned in this study.
• Line 168-174. Many studies report that China’s emissions peaked in 2013 or 2014, so the first phase is better set as 2000-2013 or 2014. Also, the second phase is mainly due to the air control policy, besides the adjustment of energy and industrial structure.
• Line 180-185. Although the global stocktake is held every five years, the stocktake assesses the achievement of NDCs of each country. Also, the baseline year of the Chinese 2020 and 2030 carbon reduction targets is 2005. I suggest the authors rewrite these sentences.
• Figure 4. Point and line sources of CAMS originated from EDGAR (Table 1). Why is the line source information lost in Figure 4d, especially in western China? Furthermore, the map of means (Figure 4f), most of the line and area information was lost.
• Figure 5c. Why are there some squares with high values in the west and northeast China?
• Figure 7. Why is the CEADs province data nearly ten times higher than other inventories in Shanxi Province?
• Figure 8. EAGAR and MEIC are the highest and lowest inventories for national CO₂ emissions, but these values varied at the provincial level. What are the key factors that affected these results? For example, CAMS had the highest values in Liaoning, Hubei provinces and Shanghai but the lowest in Hebei and Shandong provinces.
• Figures S1&S2, Why do SD and CV for Hubei and Guangdong decrease sharply in 2023?
• Table 1. Why can CAMS report the data in 2026 when it was published in 2023?
• Line 104. What does “BP plc” mean?
• Table S1. Please add a footnote for “1” mentioned in the transport sector.

Please find attached my comments in the Supplement.

The manuscript by Yang et al. 2025 provides a comparative analysis of China’s anthropogenic CO₂ emissions over the period 2000–2023, based on six widely used bottom-up inventories. The topic is highly relevant given the importance of accurate CO₂ accounting for climate change mitigation and policy verification. The dataset selection is comprehensive, and the study offers valuableinsights into temporal trends, sectoral contributions, and spatial differences across inventories. Overall, the manuscript is well written and scientifically sound.

Comments:

Although comparing multiple inventories is valuable, similar studies have been conducted for China in the past. The authors should make it clearer what distinguishes this work. Does the novelty lie mainly in the inclusion of the most recent versions of the inventories? Or is it the extension to 2023 and more detailed assessment?

The conclusion is very long and contains many technical details. The authors discuss discrepancies between the inventories and ultimately recommend one, the CEADs for sectoral data, but what is the direct impact of this on emissions control in China? The text could end with a strong recommendation toimprove regional inventories aimed at supporting mitigation policies.

Specific comments:

Line 183: What is the reason for this stable period between 2012 and 2017?

Figure 2: Why are the results presented as a five-year average? The CEAD values for the period between 2012 and 2017 are occasionally negative. What factors could explain this phenomenon?

Figure 3: has not been cited within the main text

Line 211: Why was the year 2019 chosen as the reference year for comparing spatial patterns in Figure 4? Could this choice be clarified?

Line 234: Could the authors clarify why the MEIC inventory was chosen as the benchmark for Figure 5?

Figure 7: Could the authors clarify whether the pattern observed for Shanxi Province is unique to this province or if it occurs in other regions as well?