Comments：

Few researches have focused on the long-term changes in BC. Three-year observation data of BC in Nanjing were used to predicte the changes of BC in 2014-2021 by the machine learning method. Moreover, the Aethalometer model was used to identify the source contributions of BC, including liquid fuels and solid fuels. Results revealed that the contributions of liquid fuels combustion to BC were estimated to be 80 %,  which was responsible for 77% reduction of BC. The study develop a novel methodology of predicting long-term changes in BC. However, some issues need to be clarified. I would thus recommend a minor revision to improve this manuscript.

Minor comments:

1. Lines 19-35, the logical coherence of the abstract needs adjustment. For instance, the conclusions of the three-year observational data of BC should be summarized first, followed by the results of the long-term prediction of BC.
2. Lines 167-170, here, why are 470nm and 950nm chosen to identify source contributions of BC? Please add the related description of this in the manuscript.
3. Lines 279-280, it shows that the proportion of BC to PM_2.5 can be as high as 99%. Is there a possibility of an expression error here?

Review of egusphere-2024-2503

The study focuses on the long-term trends and sources of black carbon (BC) aerosol in Nanjing, China, using three years of observational data (2019–2021) combined with historical reconstruction (2014–2021) via a random forest model. Based on K-Z filter approach, it investigates the contributions of liquid and solid fuel combustion, the effectiveness of emission reduction measures, and the interplay between meteorology and emissions. The results highlight significant decreases in BC levels driven predominantly by reductions in liquid fuel combustion, with varying seasonal and meteorological influences. The findings are of significance, and I have some comments for the authors to consider.

1. In Figure 5, are the data presented from the training dataset or the test dataset of the random forest model? High R-square values for the training dataset could indicate overfitting if the model fails to replicate results for the test dataset, potentially compromising its generalizability. To evaluate the robustness and reliability of the random forest model, it is essential to include validation results specifically for the test dataset.
2. Line 505: The AAE values of 1 and 2 used for the Aethalometer model require justification within your study. Source apportionment outputs can vary significantly depending on the AAE values assigned for fossil fuel and biomass burning. Please provide evidence supporting the chosen values.
3. Precipitation is an essential parameter for BC scavenging, yet it appears to be absent from the input variables in your random forest model and should be included.
4. Lines 212–215. Please present figures showing the results of the 10-fold CV used for the random forest model. These figures would allow reviewers to assess the model-building process and ensure robustness. Such details could be included in the supplementary materials.
5. A figure illustrating the K-Z filter results, showing the long-term trends for emissions and meteorology, can enhance clarity. Please include. Additionally, when discussing the trends of pollutants, consider applying the Mann-Kendall test to assess the statistical significance of these trends.
6. Provide an explanation for the differing impacts of meteorology on the extent of reductions in BC liquid and BC solid by the K-Z method.
7. Extend the comparison of your findings to other regions in China and globally. In China, BC emissions exhibit notable geographic heterogeneity. Beyond the commonly studied regions like the North China Plain and East China, comparisons to other areas would be beneficial. For example, studies conducted in Nanning, Guangxi province (DOI: 10.1016/j.scitotenv.2023.166747) and in Liaoning province (DOI: 10.1016/j.envpol.2024.124470) can offer valuable context. Similarly, discussing the BC/CO and BC/PM2.5 ratios in your study relative to that in these works can enhance the scientific depth of your analysis.
8. Some typos and grammar need to be corrected, such as

line 71 “severe”

line 120-121. “was then incorporated”

Line 318, “Similarly”

Thoroughly check the manuscript for similar errors to ensure clarity and precision.