A Remote Sensing Approach for Surface Urban Heat Island Modeling in a Tropical Colombian City Using Regression Analysis and Machine Learning Algorithms

Round 1

Reviewer 1 Report

The paper uses comprehensive methods for LST retrieval, SUHI estimation, and identifying influencing factors. Please find the comments below:

Main problems:

1. Page 2/22 Line 65-66 what is the definition of SUHI in this study? and please explain why the LST range was chosen to represent SUHI. Normally, the SUHI is the difference between LST in urban and its surrounded rural LST.

2. Page 4/22 Line 163-164, the LST was retrieved from different kinds of sensors data but there might be differences caused by sensors. Data from different sensors were used in this study how did the authors make sure their consistency?

3. Page 6/22 Line 222-223, how were the standard deviations calculated? Provide the formula, please.

3. Page 7/22 please add the reference for PCA description in Line 237-238. Page 9/22 check the formulation for Sub-high temperature (SHT) in Table 2. More importantly, the range of LST intervals of such as extremely high temperature, etc. should take seasons into account, because the LST usually varies a lot when it comes to different seasons.

4. Page 13/22 Figure 11, the Pearson significance test results need to be present.

Detailed problems:

4. Page 1/22 Line 36-40: How ground meteorological stations observations are used to model UHI, please explain it.

5. Page 2/22 Line 77-79 Please add the number for each subplot and explain the specific details. The data source of Figure 1 should be removed from the captain to 2.2 data. The longitude or latitude should not be simply written on the right RGB figure, and the lines of latitude and longitude need to be drawn on the figure. Please also add the topography of the study area, because the LST is also related to altitude.

6. Page 3/22 Line 94 the resolution of each satellite product should be provided.

8. Page 11/22 Line 344-346 How the Bias was calculated Figure 8? What is the meaning of two dot lines and the solid line? Page 12/22 Line 393-394 please reorganize Table 3 to make table alignment.
9.  
10. Page 10/22 Line 337-338 The caption of Figure 7 is not consistent with the analysis in the Line 332. It seems that Figure 7 is the difference of LST between LST results and observations instead of “LST results”.

Author Response

Dear Reviewer,

We are very thankful for your valuable suggestions and comments. We have uploaded a file with our replies.

Sincerely,

Julián Garzón
Corresponding Author

Author Response File: Author Response.pdf

Reviewer 2 Report

This is a review of the article named "Surface Urban Heat Island Modeling using Principal Component Analysis, Multiple Linear Regression, and Machine Learning algorithms", written by Garzon et al.. It is an interesting well-structured article that reports on UHI modeling and forecasting techniques based on datasets retrieved from satellites. The topic is suitable for the Remote Sensing journal. However several identified shortcomings need to be addressed for rendering this article suitable for publication.

1. I suggest a reframe of the title. By making it a bit broader, rather than just mentioning the techniques applied would raise more easily reader's interest. Moreover, Principal component analysis and multiple regression are Machine learning algorithms.
2. Abstract: line 11 - "thermal variations''  and "damage" are not suitable words here.
3. Abstract: line 13 -from this sentence I understand that the authors investigate what factors generate SUHI. Is this true? Or it investigates the magnitude of the factors affecting SUHI?
4. Introduction: line32 - what do you mean by ''cooling capacity of the air"?
5. Introduction: lines 34-36 - This sentence needs to be simplified. SUHI described the highest surface temperatures within urban environments as compared to the adjusting rural areas
6. Introduction: lines 51-57: The information presented here needs to be substantially enhanced with existing literature
7. Introduction: In general it needs to be substantially enhanced. Detailed existing literature should be added with respect to studies reporting SUHI as well as other studies assessing SUHI with machine learning or other techniques. Also, the research gaps/questions that this article assesses as well as its novelty should be more clearly clarified
8. Materials and Methods: line 103 "satisfactory results" - explain better
9. Materials and Methods: lines 118-120: add references
10. Improve the syntax of the overall manuscript (e.g. lines 149-152, line 311-312)
11. line 186 - mention the "indicators"
12. line 188 - give reference to the land cover classes and some basic characteristics
13. line 212 - mention the "indicators"
14. line 217 - either take out "due to technical issues " or explain better
15. line 226 - I do not understand how the study of Guillevic is related to the current article. please explain better
16. lines 358-360. The statement here is contradictory. I suggest discussing this interesting observation and delete the second sentence.
17. line 382 - explain how did you removed outliers
18. Statistics should be reported appropriately. check here: https://www.statisticshowto.com/probability-and-statistics/reporting-statistics-apa-style/
19. line 390: I suggest providing more information concerning the assumptions. e.g. plots of residuals vs fitted etc.
20. explain better what is depicted in Figure 11 and add units

Author Response

Dear Reviewer,

We are very thankful for your valuable suggestions and comments. We have uploaded a file with our replies.

Sincerely,

Julián Garzón
Corresponding Author

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The paper has been greatly improved after major revision.

Reviewer 2 Report

I would like to thank the reviewers for carefully addressing my comments. In my opinion, the article in its current form should be accepted for publication.