Prediction of the Tropospheric NO2 Column Concentration and Distribution Using the Time Sequence-Based versus Influencing Factor-Based Random Forest Regression Model
Round 1
Reviewer 1 Report
Manuscript has been reviewed.
1. Why is Beijing selected for this study?
2. In figure 3, it is shown that distribution of NO2 through out the year is same. Is it average value of all twelve months. Please write, in which month it is maximum / minimum.
3. Figure 4 is missing.
4. Figure 5 are mentioned at two places (page 6 and 7-8). Please make necessary corrections.
5. Some experimental measurement is also required to measure NO2 concentration for proper validation of results.
6. Conclusion is to be re-written in short.
Author Response
Thank you very much for your work, your comments have been very helpful and we have amended them accordingly.
In response to question 1, we have added the reason for the choice of study area on page 4.
For question 2, we have added the appropriate explanation and values to the figure.
In response to question 3, Figure 4 is present in our manuscript, the pdf appears to be missing.
In response to question 4, we have made changes after checking.
In response to question 5, due to time and budget constraints we were unable to measure the NO2 concentration experimentally and instead we have added a comparison of predicted and measured values on page 18.
In response to question 6, we have streamlined the conclusions.
Translated with www.DeepL.com/Translator (free version)
Reviewer 2 Report
see attached pdf
Comments for author File: 
Comments.pdf
Author Response
Thank you very much for your work, your comments have been very helpful and we have made the appropriate changes in response to your comments.
Reviewer 3 Report
The contributions are limited. Please find my comments as follows:
1. The authors should highlight the research gaps behind the motivation for this study?
2. Can the study of one city make some global conclusions?
3. A more extensive literature review is needed.
4. Why only Random Forest Regression? Why not others?
5. A good block diagram can be helpful to explain the methodology used in the manuscript.
6. What are the novel contributions of the present study?
7. The state-of-the-art comparisons to earlier atmospheric-based analyses are missing in the manuscript.
8. References are limited and old.
9. Some of the earlier and recent works need to be cited and used for the comparison of the current study:
a) A Novel Method for Regional NO2 Concentration Prediction Using Discrete Wavelet Transform and an LSTM Network (10.1155/2021/6631614)
b) Estimating monthly global ground-level NO2 concentrations using geographically weighted panel regression (C Li, S Managi - Remote Sensing of Environment, 2022 – Elsevier)
c) Machine learning-based estimation of ground-level NO2 concentrations over China (Y Chi, M Fan, C Zhao, Y Yang, H Fan, X Yang… - Science of The Total …, 2022 – Elsevier)
d) Estimating daily ground-level NO2 concentrations over China based on TROPOMI observations and machine learning approach (S Long, X Wei, F Zhang, R Zhang, J Xu, K Wu… - Atmospheric …, 2022 – Elsevier)
e) Impacts of Certain Meteorological Factors on Atmospheric NO2 Concentrations during COVID-19 Lockdown in 2020 in Wuhan, China (T Ju, T Geng, B Li, B An, R Huang, J Fan, Z Liang… - Sustainability, 2022 - mdpi.com)
f) GIS and Time Series Modelling Approach to Predict Tropospheric Nitrogen Dioxide (C ROUT, G SHUKLA, V BENIWAL… - Annals of Agri-Bio …, 2022 - agribiop.com)
g) A machine learning model to estimate ground-level ozone concentrations in California using TROPOMI data and high-resolution meteorology (W Wang, X Liu, J Bi, Y Liu - Environment International, 2022 – Elsevier)
h) Assessment of NO2 population exposure from 2005 to 2020 in China (Z Huang, X Xu, M Ma, J Shen - Environmental Science and Pollution …, 2022 – Springer)
i) Tracing out the effect of transportation infrastructure on NO2 concentration levels with Kernel Density Estimation by investigating successive COVID-19-induced … (KD Kovács, I Haidu - Environmental Pollution, 2022 – Elsevier)
j) Spatiotemporal neural network for estimating surface NO2 concentrations over north China and their human health impact (C Zhang, C Liu, B Li, F Zhao, C Zhao - Environmental Pollution, 2022 – Elsevier)
k) Ground-Level NO2 Surveillance from Space Across China for High Resolution Using Interpretable Spatiotemporally Weighted Artificial Intelligence
Author Response
Thank you very much for your work, your comments have been very helpful and we have amended them accordingly.
For questions 1, 2, 3, 4 and 6 we have added a description and explanation in the introduction.
For question 5, we have drawn up a flow chart.
For question 7, we have added a reference on page 9.
For questions 8,9 we have added references and citations.
Round 2
Reviewer 3 Report
Earlier comments are not addressed properly. Please find the comments:
1. Figure 3 is very basic. It can be improved with more details so that reader can get a good idea about the process flow.
2. “RFR” needs to be defined before.
3. Figure 3: Training data should be given to random forest regressor (RFR) and optimization and tuning of parameters done on given data. Then, performance should be accessed on the test set.
4. As mentioned, “Considering that the random forest model has a faster computing speed and lower technical requirements…………………”.. Authors should illustrate with a comparative table or diagram in that context.
5. A comparison with earlier and recent works needs to be shown for the comparison with the current study.
6. Please take care of language and typo mistakes to enhance the readability of the paper.
Author Response
Thank you very much for your work, your comments have been very positive for us and we have made changes in response to your comments.
In response to Figure 3, we have made changes.
On page 3 ''In contrast to previous studies, we discuss two commonly used methods for prediction and analysis based on random forest regression models (RFR). RFR is defined.''
On pages 15 and 16, comparisons with published studies using traditional methods are added.
On page 3, as we did not carry out studies using other methods, the relevant tables or images are not available and we cite previous studies for this section instead.
