Estimation of Regional Ground-Level PM_2.5 Concentrations Directly from Satellite Top-of-Atmosphere Reflectance Using A Hybrid Learning Model

Round 1

Reviewer 1 Report

Thank you for giving me this opportunity to read the manuscript entitled "Estimation of PM2.5 concentrations in Beijing-Tianjin-Hebei region using a hybrid-learning model". The topic of this manuscript is interesting and would be a good contribution to this field. I think it could be considered for publication in Remote Sensing once the following issues are addressed.

1. Please replace the keywords that already appear in the manuscript's title with close synonyms or other keywords, which will also facilitate your paper to be searched by potential readers.
2. Lines 26-27: References should be provided for the statements here. Besides, papers focusing on environmental exposure and environmental health are more appropriate to support the statement in Line 25, for example, the paper titled "Dynamic assessment of PM2. 5 exposure and health risk using remote sensing and geo-spatial big data".
3. Line 45: I do not think it is necessary to give a citation to the item "AOD", I suggest the author move and merge it with references 12-13.
4. Figure 1: What are the blue triangles in the right sub-figure? It looks that this information is missing in the legend. Also, the scale needs to be added to the China map.
5. Line 290: The authors used "geographically and temporally weighted regression" as the full name of GTWR in the Result section but used "Geographic and Time-Weight Regression" as the full name in the Introduction section. Please unify the full name of it.
6. Section 3.1.3: There are quite some well-designed PM2.5 concentration estimation models with good performance in various accuracy evaluation metrics, and some of these models have also been cited in this manuscript by authors. I suggest that the authors compare their proposed model with these models to prove its better performance rather than comparing it with some machine learning models.
7. Limitation should be added as a sub-section in the Discussion
8. Some grammatical errors exist in the manuscript. Therefore, a critical review of the manuscript's language will improve readability.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

Interesting paper, but some corrections are needed as folows:

• In the abstract are used to many abbreviations. If it is possible - reduce the use. 
• Equation 1, 2 – explain what is j
• Table 2. n is not definet and is something expected after „T:“
• The form of the table 2 is not clear – I would suggest to use a flow chart instead
• Same comment as previous for table 3
• Steps given in table 4 do not need a form of a table...
• Figure 4. the title must include the explanations of a-h (or relate it somehow with the table 7)
• Table 8. where are indicated which are „the same data“ for different models? Please indicate them.

Sincerely,

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

Review of “Estimation of PM2.5 concentrations in Beijing-Tianjin-Hebei region using a hybrid-learning model” by Feng et al.

This manuscript by Feng et al. suggested a hybrid-learning model to estimate surface PM2.5 concentrations over Beijing-Tianjin-Hebei region. They utilized the Top-of-Atmosphere reflectance data from MODIS and meteorological variables from ECMWF. By testing several combinations of data processing, they concluded that including wavelet decomposed TOAR and meteorological variables yield the best results in the cross-validation evaluation.

This manuscript is well written and well organized. I am confident this manuscript contribute research community, both in remote sensing and air quality studies. I recommend this manuscript for publication after minor revisions.

L10

“Top-10 of-Atmosphere (TOA) reflectance”

I suggest to use Top-of-Atmosphere reflectance (TOAR), which is the measurements of reflectance at the top of atmosphere. “R” has more meaning than TOA.

L30

“resulting in a serious decline in air quality in east-central China, “

Serious deterioration might be better expression.

L52

“However, aerosol patterns need to be determined with long-term ground-based monitoring data, which has an impact on PM2.5 estimates.”

What is aerosol pattern? Please, elaborate.

L96

“To make up for this deficiency, hybrid models created.”

were created.

L130

“Shanghai Environmental Monitoring Center”

Please, specify URL for data.

L233

“In this study, the Bayesian optimization method[39] selected”

was selected

L261

“Bayesian selected to optimize the main hyperparameters of the Random Forest and 261 LightGBM algorithms, and the optimization results are shown in Tables 5 and Table 6.”

Please, revise. Its meaning is not clear.

L322

“while the satellite TOA reflectance has a wider spatial coverage compared to AOD.”

Please, elaborate. Why does TOAR have wider spatial coverage?

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

Thank you for giving me this opportunity to read the revised version of the manuscript titled " Estimation Regional Ground-Level PM2.5 concentrations directly from Satellite Top-of-Atmosphere Reflectance using a hybrid-learning model ", and for the detailed responses to my earlier comments. I am satisfied with this revised version, and I think it is acceptable now.