Skip to Content
MDPIMDPI

Advanced

by
  • Dewei Yang 1,
  • Shaobo Zhong 2 and
  • Weiqi Zhong 1
  • et al.

Reviewer 1: Foyez Ahmed Prodhan Reviewer 2: Anonymous Reviewer 3: Sungwon Choi

Round 1

Reviewer 1 Report

The authors investigated several popular models for near-surface air temperature estimation. The work is largely well-written, and the methodological framework is well presented. Still, the article needs significant improvement in the introduction, methodology, and conclusion sections. Considering my observations as follows, I suggest that the paper undergo major revision before considering it for publication.

Comments

1.     Research on such a topic necessitated extensive analysis, argument, debate, and discussion, as well as the identification of some weakness in the literature, all of which are missing in the introduction section. I mean that the motivation for this research should be mentioned in the introduction section.

2.     Why did the author use the vegetation indices for air temperature prediction? Need a more clear and brief description regarding this.

3.     A table of the data, including the types of data, the sources, and how it was used in the study, is necessary.

4.     Figure 6: Left Side- The author wrote "deep learning model (6 kinds)". My question is, were any deep learning models used in this manuscript? Please clarify. Otherwise, correct it. Are the input parameters, such as Terra_day_LST, Terra_night_LST, etc., full images or extracted values? If extracted value used please mention it the image. Right Side- "Large range MODIS_LST" and "Large range MODIS_LST" means what? Is there full image used?  

5.     Line 280: Finally, the monthly satellite remote sensing data are input into the trained models to  estimate the near-surface air temperature for each month. Here, which variable was used as a response variable?

6.     Why did the author use the results of the Fang et al. study as other study to compare the predicted value. Is Fang et al. use the same data set for training the model or same input parameters? I would suggest to not use the Fang et al. study. Moreover, in Figure 9, 10, 11 and so on please use real air temperature figure instead of other study to compare the predicted value.

7.     There are other factors that also influence the air temperature that were not considered for this study. Moreover, use of a machine learning model may lead to overfitting of the data for model prediction. The author did not describe this issue. Therefore, I would recommend to briefly describe the limitations of the study.

8.     The conclusion can be improved by highlighting the innovation content of the paper, future research directions, and recommendations for policy formulation.

9.     I would strongly recommend to check the English grammar with a native English speaker.

Author Response

The authors investigated several popular models for near-surface air temperature estimation. The work is largely well-written, and the methodological framework is well presented. Still, the article needs significant improvement in the introduction, methodology, and conclusion sections. Considering my observations as follows, I suggest that the paper undergo major revision before considering it for publication.

AUTHORS’ ANSWER: We are very grateful to the reviewers for their stimulating comments and suggestions. We have gone through each section, reviewed the problems and shortcomings of the article from the reader's perspective, and responded to the issues raised by the reviewers on a point-by-point basis. Specific responses are listed below.

Comments

  1. Research on such a topic necessitated extensive analysis, argument, debate, and discussion, as well as the identification of some weakness in the literature, all of which are missing in the introduction section. I mean that the motivation for this research should be mentioned in the introduction section.

AUTHORS’ ANSWER: We revised the introductory section, re-judged the literature [see lines 159-163], and rewrote the sentences to make our purpose clearer [see lines 257-301].

  1. Why did the author use the vegetation indices for air temperature prediction? Need a more clear and brief description regarding this.

AUTHORS’ ANSWER: We have added some previous literature that provides a brief explanation of the reasons for introducing the vegetation index as an impact factor [see lines 353-356].

  1. A table of the data, including the types of data, the sources, and how it was used in the study, is necessary.

AUTHORS’ ANSWER: We added Table 2 to introduce the types of data, data sources, and how they were used in our study. [see lines 363].

  1. Figure 6: Left Side- The author wrote "deep learning model (6 kinds)". My question is, were any deep learning models used in this manuscript? Please clarify. Otherwise, correct it. Are the input parameters, such as Terra_day_LST, Terra_night_LST, etc., full images or extracted values? If extracted value used please mention it the image. Right Side- "Large range MODIS_LST" and "Large range MODIS_LST" means what? Is there full image used?

AUTHORS’ ANSWER: We have modified the flow chart of our study. The deep learning models used in our study are NN, LSTM, BiLSTM, other models are machine learning models. Terra_day_LST, Terra_night_LST, etc. are full image data, need to perform value extraction to point operation before inputting into the model. Large range MODIS_LST refers to the full image data input to the trained model, which is also illustrated in the Figure 6. [see lines 547]. 

  1. Line 280: Finally, the monthly satellite remote sensing data are input into the trained models to  estimate the near-surface air temperature for each month. Here, which variable was used as a response variable?

AUTHORS’ ANSWER: The response variable here is the predicted near land surface temperature for the full map.

  1. Why did the author use the results of the Fang et al. study as other study to compare the predicted value. Is Fang et al. use the same data set for training the model or same input parameters? I would suggest to not use the Fang et al. study. Moreover, in Figure 9, 10, 11 and so on please use real air temperature figure instead of other study to compare the predicted value.

AUTHORS’ ANSWER: We have removed the study by Fang et al. from the corresponding study. The actual values of near-surface air temperature nationwide are not available, and only the near-surface air temperature at the location of meteorological stations can be accurately observed. Therefore, at the previous version of our manuscript, we considered comparing our model predictions with the results of the study by Fang et al.'s study. [see the result].

  1. There are other factors that also influence the air temperature that were not considered for this study. Moreover, use of a machine learning model may lead to overfitting of the data for model prediction. The author did not describe this issue. Therefore, I would recommend to briefly describe the limitations of the study.

AUTHORS’ ANSWER: We have added the limitations of the study in the discussion section and made our reflections on some of the factors influencing the air temperature that are not fully considered. [see lines 574-586] Among others, we have also made relevant considerations for the corresponding model over fitting phenomenon in the discussion section. [see lines 1230-1241].

  1. The conclusion can be improved by highlighting the innovation content of the paper, future research directions, and recommendations for policy formulation.

AUTHORS’ ANSWER: We have added a summary of the innovative nature of the article and the reference of the study in the field of policy making in the conclusion section, as well as an outlook on the future research directions of the article. [see lines 1329-1353].

  1. I would strongly recommend to check the English grammar with a native English speaker.

AUTHORS’ ANSWER: One of the authors has copy edited carefully the manuscript. We also ask someone, who is good at scientific paper writing and has published some papers, to review the manuscript for us. We hope the revision can be approved by you.

Reviewer 2 Report

 

The manuscript deals with the near surface air temperature estimation by several models, and truly it is helpful to the monitoring of environment and land use/cover etc. From this viewpoint, the article should be published. However, some special issues should be considered in order to improve the quality of the manuscript. The detailed notes can be found in the attachment. The main comments are as following:

 

1.     The language should be revised carefully. Many sentences in the manuscript are incomplete, and some technical terms and/or phrases are not exact enough. Furthermore, too much active voice was used in the manuscript. Lots of sentences should be rewritten clearly.

2.     The title of the manuscript named as “near-surface air temperature”, and this is what the “meteorological temperature” means. Usually the temperature estimated by remote sensing data can be divided into LST and SST which correspond to land surface and sea surface respectively. Hence should the title of the manuscript be changed to “near land surface air temperature” since all the estimation values were from land surface instead of sea surface?

3.     Generally speaking, the manuscript was well organized. However, the section of Discussion is too short and plain. The authors should try to discuss the factors which result in different results, the model? land cover since NDVI had been introduced in the manuscript? Elevation as we all know high elevation resulting in permanent frozen? Also is the temperature difference determines the accuracy of the temperature estimation since the result of winter is better than that of summer? If these were thoroughly discussed, the quality of the manuscript will be greatly improved.

4.     Most of the captions of figures and table should be revised.

5.     Figure 13, geographical division or administrative division? Usually geographical division should not be divided by administrative boundaries. Moreover, the division scheme is not suitable whether by temperature estimation or just by geographical features.

6.     Section 3.2, the estimation values were evaluated by different seasons. How the average values of different seasons calculated? How the seasons be classified?

7.     The Discussion part of the manuscript seems like “future prospects/outlook”. It is suggested that two sections: result and discussion, should be delineated clearly.

8.     Pay attention to the format and standard of the references.

 

Some special comments can be seen in the attachment.

 

 

Comments for author File: Comments.pdf

Author Response

Comments and Suggestions for Authors

The manuscript deals with the near surface air temperature estimation by several models, and truly it is helpful to the monitoring of environment and land use/cover etc. From this viewpoint, the article should be published. However, some special issues should be considered in order to improve the quality of the manuscript. The detailed notes can be found in the attachment. The main comments are as following:

 AUTHORS’ ANSWER: We are very grateful to the reviewers for their stimulating comments and suggestions. We have gone through each section, reviewed the problems and shortcomings of the article from the reader's perspective, and responded to the issues raised by the reviewers on a point-by-point basis. Specific responses are listed below.

  1. The language should be revised carefully. Many sentences in the manuscript are incomplete, and some technical terms and/or phrases are not exact enough. Furthermore, too much active voice was used in the manuscript. Lots of sentences should be rewritten clearly.

AUTHORS’ ANSWER: One of the authors has copy edited carefully the manuscript. We also ask someone, who is good at scientific paper writing and has published some papers, to review the manuscript for us. We hope the revision can be approved by you.

  1. The title of the manuscript named as “near-surface air temperature”, and this is what the “meteorological temperature” Usually the temperature estimated by remote sensing data can be divided into LST and SST which correspond to land surface and sea surface respectively. Hence should the title of the manuscript be changed to “near land surface air temperature” since all the estimation values were from land surface instead of sea surface? 

AUTHORS’ ANSWER: We have fixed the issue in the full text.

  1. Generally speaking, the manuscript was well organized. However, the section of Discussion is too short and plain. The authors should try to discuss the factors which result in different results, the model? land cover since NDVI had been introduced in the manuscript? Elevation as we all know high elevation resulting in permanent frozen? Also is the temperature difference determines the accuracy of the temperature estimation since the result of winter is better than that of summer? If these were thoroughly discussed, the quality of the manuscript will be greatly improved.

AUTHORS’ ANSWER: We have added the limitations of the study in the discussion section and made our reflections on some of the factors influencing the air temperature that are not fully considered. [see lines 1230-1241] Among others, we have also made relevant considerations for the corresponding model over fitting phenomenon in the discussion section. [see lines 1175-1184]. The discussion on the effect of temperature difference on the results is also added [see lines 1211-1219].

  1. Most of the captions of figures and table should be revised.

AUTHORS’ ANSWER: We have revisited and renamed all the chart names.

  1. Figure 13, geographical division or administrative division? Usually geographical division should not be divided by administrative boundaries. Moreover, the division scheme is not suitable whether by temperature estimation or just by geographical features.

AUTHORS’ ANSWER: In our study we are trying to examine the heterogeneity of the estimation effect of the model in different regions, and the partitioning according to the commonly used Chinese integrated regional partitioning, with some descriptions of the partitioning. [see lines 938-949]

  1. Section 3.2, the estimation values were evaluated by different seasons. How the average values of different seasons calculated? How the seasons be classified? 

AUTHORS’ ANSWER:We have added a description of how to distinguish the seasons in the text.  [see lines 750-755]

  1. The Discussion part of the manuscript seems like “future prospects/outlook”. It is suggested that two sections: result and discussion, should be delineated clearly.

AUTHORS’ ANSWER:We have revised the discussion section accordingly, placing the outlook for future research in the conclusion section.

  1. Pay attention to the format and standard of the references.

AUTHORS’ ANSWER: We have made changes to the format of the references.

 

Reviewer 3 Report

Dear authors,

I congratulate the authors that they show the potential of using various machine leaning to retrieve air temperature (Ta). The topic of the manuscript is highly relevant in the actual context of Remote Sensing. The objective of this paper is clearly mentioned. The procedure followed and experiments designs are good and those are clearly presented in the manuscript.
Even though this paper is good it needs some improvement from the current form. I recommend accepting this article after these revision.

Major comments

1. Please add the temporal resolution and spatial resolution of each data used in this study and the method to resolve the difference between the two resolutions of the data. In addition, information on the spatial resolution and temporal resolution of the final result is required.

2. Please add the total number of data used to apply machine learning, and add a comment on the percentage of data used for training, validation, and testing.

3. NDVI and EVI basically mean vegetation index, so multicollinearity needs to be evaluated in order to use the two data as input data.

4. Excluding LST as input data, only DEM and VI were entered. Please add considerations for other input data.

5. Please add considerations about the physical properties of Ta.

 

Author Response

Please see the attachment

Author Response File: Author Response.doc

Round 2

Reviewer 1 Report

The authors answered my all questions sincerely. So I don't have any further comments.

Author Response

Thank you very much for your comments and suggestions on this study.

Reviewer 2 Report

 The quality of revised manuscript was greatly improved by the authors’ hard works, however the readability of the paper is still a bit poor, and somewhat improvement of the English writing is appreciated and expected.

  Some specific suggestions are as follows:

1.       Better an abbr. for “near land surface air temperature”, since the term appears too frequently in the manuscript.

2.       The beginning of the last paragraph of page-1 should be rewritten. Furthermore, pay attention to the usage of past tense, past participle, and present participle etc. in the paper.

3.       Make it clear that figure 1 illustrates both the location and elevation of stations thus the caption of the figure should be rewritten. 

4.       Caption of table 2, table 3 and figure 6 are not suitable. Make them clear.

5.       Unit of legend in figure 9, figure 10, figure 11 and figure 12 should be added.

6.       The captions of figure 9-12 in first version were noted “in 2015”. Surely it is necessary if they are truly temperature(s) in 2015.

 

Some minor spelling error or writing suggestions can be seen in attachment.

Comments for author File: Comments.pdf

Author Response

The quality of revised manuscript was greatly improved by the authors’ hard works, however the readability of the paper is still a bit poor, and somewhat improvement of the English writing is appreciated and expected.

AUTHORS’ ANSWER: Thank you for your renewed encouragement and suggestions for this study. In response to these comments, we have made changes where appropriate and provided individual responses to the issues raised. The specific responses are as follows.

  1. Better an abbr. for “near land surface air temperature”, since the term appears too frequently in the manuscript.

AUTHORS’ ANSWER: Near land surface air temperature is abbreviated as NLSAT according to the reviewer’s suggestion.

  1. The beginning of the last paragraph of page-1 should be rewritten. Furthermore, pay attention to the usage of past tense, past participle, and present participle etc. in the paper.

AUTHORS’ ANSWER: The sentences have been rewritten. We have checked and corrected the tenses all over the paper according to the common suggestions for an academic paper. We carefully checked the full text to avoid expression and grammar problems as much as possible.

  1. Make it clear that figure 1 illustrates both the location and elevation of stations thus the caption of the figure should be rewritten.

AUTHORS’ ANSWER: The caption has been added to make it clear.

  1. Caption of table 2, table 3 and figure 6 are not suitable. Make them clear.

AUTHORS’ ANSWER: We have checked the captions of all figures and tables and made revision if considered improper.

  1. Unit of legend in figure 9, figure 10, figure 11 and figure 12 should be added.

AUTHORS’ ANSWER: The notation degree Celsius (℃) has been added in the legend of each figure.

  1. The captions of figure 9-12 in first version were noted “in 2015”. Surely it is necessary if they are truly temperature(s) in 2015.

AUTHORS’ ANSWER: This is a typo in the revision. They are temperatures in 2015. We have added the year of 2015 in the latest revision.

Reviewer 3 Report

Dear authors,

I think this article can be published.

sincerely.

Author Response

Thank you very much for your comments and suggestions on this study.