This study presents a comprehensive overview of Aethalometer based observations from CBNET, which includes measurements from 48 stations (including 23 urban, 18 rural sites and 7 background sites) across China during the period 2008-2020.  The results are used to understand the discrepancies in chemical transport models and emission inventories.

This study attempts to take a step further and address three problems

(1) What can we learn from the comparison between BC ground observations and CTM simulations;

(2) Whether the inter-annual variations of BC can be used as an indicator of BC emissions;

(3) Which factors dominated the variations of BC in China during the past 13 years.

thereby reducing the uncertainties in BC emissions and simulations from CTMs.

The data analysis is robust and the different factors that need to be considered while handling aethalometer data were considered.

The attempt to try out season specific AAE and de-seasonalizing the data using machine learning is novel.

Minor comments

lines 113-120

The treatment of negative and following values: The manufacturer suggests retaining these values and doing a running mean before filtering the odd values. However, there is no general consensus on the methodology. Was a smoothing performed before or after removing the outliers? There would be a difference in the final values depending on the method followed. It would be good if this is also clearly stated.

lines 121-128

While a criterion of minimum of 50% data points was adopted for passing quality control for long term analysis, it is not clear whether it is 50% points for a day/month/ annually. This needs to be made clear.

Section 2.3

Station specific AAEs are determined as the 1^st and 99^th percentiles of the AAE distribution at the site. Although this avoids assuming a universal value for AAE it is still arguable that the selected values may or may not represent the actual conditions and fuel types. While the DeBC_lf (%) reduces upto 50%  DeBC_sf (%) almost triples at many stations compared to default values. These variations are huge and indicate that the eBC_lf values are overestimated while the eBC_sf underestimated by conventional methods. Are these values realistic or are the values blown up due to the use of % changes?  I suggest including figures comparing the values with and without the application of this method in the supplementary for easy comparison and understanding.  

Figures 1&2, Tables and other places

The station codes used in this study makes it too difficult to associate with individual stations. The observatories are denoted by long numbers making the figures clumsy and difficult to identify. I am curious to know if there a specific reason behind the station codes?

Figure 2

This figure is really impressive and rich in information content. On the downside we have long observatory names and randomly distributed pie charts. It is really difficult to find the pie charts corresponding to a particular location on the map. It would be way easier for the reader if they are arranged in ascending order as in Fig 3. Also, the figure caption does not say anything about the MERRA data used in the background. It is also not clear why the inset is used. Is it intended to only show the islands to the south of the mainland which got cut off?

Figure 4

It would be good to add a legend here. As I go through this section, I am curious to know if any individual location showed a positive trend.

Section 3.3

If the eBC values for China are estimated using the SS AAE direct comparison with values from existing studies does not make much sense. The older studies use the conventional methods to estimate eBC while here the corrections pull down the eBC values.

Figure and table captions.

The figures are not self-explanatory as key information is missing in the caption. It makes it difficult for the reader to easily find out key information.

Spelling errors

Table S3 Hyperparameter tunning

Concluding comments

This study has made a great effort in comparing observations with CTMs. The authors find that models and outdated inventories fail represent BC over China leading to underestimations and increasing trends. While the station specific AAE is interesting, the method remains inherently oversimplified, as it does not fully capture complexities such as the distinction between coal and biomass sources, seasonal variability in BC lifetime, and mixing state effects that influence optical properties. While the percentile-based approach is practical, the arbitrariness of AAE selection cannot be entirely ruled out. A more robust validation such as through chemical tracer analysis (e.g., OC/EC or 14C) would be ideal, although such efforts may be beyond the scope of this study and could be considered for future work. I also feel that the random forest modeling performed in this study was a bit under utilised in the end. Pictorial representation of the changes observed after using this model would make a good impact on the reader. 

Review of "Black carbon aerosols in China: Spatial-temporal variations and lessons from long-term atmospheric observations" by Zheng et al.

This study presents the characteristics of black carbon concentration variations in China using observational data from 2008 to 2020. The authors well describe the process of calculating black carbon concentrations through aethalometer measurements and explain the long-term variation characteristics of black carbon concentrations by comparing them with CTM and reanalysis data.

Numerous papers have already been published regarding the long-term characteristics of black carbon concentrations in China, and many of these studies are cited in this paper. All studies consistently report decreases in black carbon concentrations due to air quality reduction policies. Therefore, it would be beneficial to clearly explain how the scientific findings (conclusions) newly reported in this study are distinctive compared to previous works. Additionally, while this study reports measurement results up to 2020, several recent papers have explained black carbon concentration variation characteristics through various factors including emissions and meteorological conditions, albeit regionally within China (e.g., https://www.nature.com/articles/s44407-025-00010-z). From this perspective, I suggest extending the analysis period of this study to include recent years beyond 2020.

The study presents results of black carbon (eBC) concentrations observed through filter-based optical methods. The process of calculating eBC concentrations is clearly and well explained through citations of previous studies. However, for the methodology of dividing contributions into solid fuel combustion (eBCsf) and liquid fossil fuel combustion (eBClf), which represents a major finding of this study, detailed methodological presentation and sufficient validation of results are required. In particular, clear presentation of theoretical limitations and associated uncertainties in calculations based on AAE from multi-wavelength optical measurements is necessary.

Furthermore, many studies have explained the emission sources of black carbon or elemental carbon through carbon isotope analysis, determining contributions from solid fuel and liquid fossil fuel combustion. I believe the most critical aspect for publication of this study would be to validate the calculation results by comparing them with such previous research findings and demonstrating consistent results. For example, I suggest presenting comparisons between this study and carbon isotope analysis-based contribution assessments for specific locations and time periods in the main text.

Meanwhile, Sections 3.1 and 3.2 present concentrations, contributions, seasonal changes, and annual variation characteristics in a rather straightforward manner compared to the observational and analytical results, lacking detailed explanations of why such contributions and spatiotemporal variations occur. Such explanations would be necessary to establish some differentiation from existing studies.

In Section 3.3, the authors discuss results by simply presenting comparisons with observation points in other countries. For such international data, more detailed descriptions of the observational data used are needed, and the justification for why such comparisons are necessary in this study should be clearly presented. Given the purpose of explaining black carbon concentration changes in China, it appears that this entire section could be removed without detriment to the study.

The content of Section 4.1 has also been mentioned in many previous studies. The most important conclusion the authors wish to convey in this section is not clearly evident. For example, clear explanation is required regarding whether the contribution analysis results based on aethalometer in this study possess the accuracy required for CTM simulation validation and improvement, or whether the authors intend to discuss methods for improving CTM simulation accuracy.

The conclusions in Section 4.2 "Weather normalized concentrations: an indicator of BC emissions" also appear ambiguous. More detailed explanation is required regarding whether this explains the previous observational results or whether emission source accuracy improvement is needed from a modeling perspective.

The causes of black carbon concentration reduction ultimately mentioned in Section 4.3 are all results that have been discussed in previous studies. Specific description of what this study presents differentially compared to previous research is needed