Applicability Assessment of Passive Microwave LST Downscaling over Semi–Homogeneous Desert Underlying Surface Based on Machine Learning

Round 1

Reviewer 1 Report

In my view, this is a really intriguing work, and I have no doubt that its readers will find it interesting as well. Some minor problems that I see in the article are as follows:

 

1- Please provide more explanations than the introduction and statement of the problem in the beginning.

2- Insert the paper outline at the end of Introduction.

3- Page 3, Line 103; what is “?” in fine-scale LST?.

4- In section 2.2.1. Authors need to consider more explanations.

5- In Section 2.2.3, authors need to consider some refences for their statement “According to previous studies,……”.

6- In Section 2.3.3, why the authors considered "10-fold cross-validation"? , . Authors need to consider more explanations.

7- The Discussion should be more explicit and contain more quantitative and practical data.

8- Minor English grammar and spelling checking is required.

Author Response

Dear Reviewer,

 

Thank you for your valuable feedback on our work. We appreciate your positive comments and will strive to address the minor problems that you have identified in your review.

 

We also appreciate your comments on how we can improve the clarity and understanding of the paper. The specific response to the issue is as follows:

 

1. Please provide more explanations than the introduction and statement of the problem in the beginning.

We agree that more explanations could be included in the introduction and statement of the problem, and we have work to expand these sections in our manuscript. Here's what I added at the end of the first paragraph：

 

Due to the complexity of topography, soil, vegetation, weather and other factors, the LST is heterogeneous spatially [15,16] and fluctuating in time. therefore, to accurately describe these variations, high spatial and temporal resolution data are necessary. Currently, thermal infrared and optical remote sensing clear sky LST retrieval al-gorithms, such as MODIS and AVHRR, have achieved high accuracy [17-19]. However, in non-clear sky areas, such as those affected by clouds, the accuracy of these retrieval algorithms is seriously compromised [20,21]. To overcome the limitations of optical remote sensing, microwave remote sensing is used to obtain surface radiation infor-mation under complex weather conditions [22,23]. The spatial resolution of passive mi-crowave remote sensing is low, but it can penetrate clouds and is weakly affected by the atmosphere [24]. Therefore, passive microwave remote sensing is ideal for all-weather observations of LST. However, to improve the spatial resolution for regional scale re-search, other data sources must be fused with the passive microwave data. Passive mi-crowave LST spatial downscaling technique enables the acquisition of more detailed LST information, even under complex weather conditions [25,26]. The passive micro-wave-based downscaling method is a data fusion technique that relies solely on remote sensing and spectral information to enhance the spatial resolution of passive microwave remote sensing. It utilizes high spatial resolution auxiliary data to reveal the spatial heterogeneity of the land surface, and provides more accurate descriptions of the spatial distribution of LST. Unlike distance-based spatial interpolation methods, the downscaling method does not estimate unknown pixels based on the distance to known sample points. Instead, it determines pixel values from low spatial resolution passive microwave data, and then fuses them with high spatial resolution LST information to generate more accurate high-resolution LST data. It can provide more accurate and re-liable LST, which is crucial for climate change research, natural resource management, and environmental monitoring.

 

2. Insert the paper outline at the end of Introduction.

In the last paragraph of the introduction, the following content was added：

In the 2. part of our manuscript, we introduced the characteristics of the study area, the data sources used (including site data and remote sensing data), the data processing procedures, and the methods used. In the 3. part, we provided a comprehensive intro-duction to the results of the paper and explained the relevant phenomena. Specifically, 3.1 showed the correlation between the characteristic factors of semi-heterogeneous underlying surfaces and land surface temperature, which provided a preliminary evaluation of the selection of each factor and its contribution to the results. 3.2 showed the 10-fold cross-validation results of five classic models during the training process. 3.3 conducted cross-validation of the five classic downscaled models using MYD11A1's LST error <= 1K per pixel, selected the optimal model, and analyzed the contribution of each characteristic factor to the Catboost model. 3.4 analyzed the relationship between the downscaled results of passive microwave land surface temperature and multi-layer soil temperature using data from the unique measured station in the study area. Part 4 quantitatively discussed the accuracy of current downscaled results and related re-search, and provided prospects for future work. Part 5 presented the main conclusions of the paper.

3. Page 3, Line 103; what is “?” in fine-scale LST ?

The concept of "fine-scale LST" refers to surface temperature data with high spatial resolution. The "?" is a punctuation mark used at the end of a "How can we...." statement to indicate a question.

4. In section 2.2.1. Authors need to consider more explanations.

Based on your suggestion, we have supplemented the information for AMSR-2 passive microwave data and MODIS related data. Please refer to the resubmitted manuscript for detailed information.

5. In Section 2.2.3, authors need to consider some refences for their statement “According to previous studies,……”.

Thank you for your reminder. Based on your suggestion, I have included the following four references. Please refer to the resubmitted manuscript for detailed information.

24. Sun, D.; Li, Y.; Zhan, X.; Houser, P.; Yang, C.; Chiu, L.; Yang, R. Land Surface Temperature Derivation under All Sky Conditions through Integrating AMSR-E/AMSR-2 and MODIS/GOES Observations. Remote Sensing 2019, 11, doi:10.3390/rs11141704.
25. Li, Y.; Wang, X.; Ma, Y.; Hu, G.; Gui, H.; Zhang, G. Downscaling land surface temperature through AMSR-2 passive microwave observations by Catboost semiempirical algorithms. Arid Zone Research 2021, 38, 1637-1649, doi:10.13866/j.azr.2021. 06. 15.
26. Mao, K.; Shi, J.; Li, Z.; Qin, Z.; Li, M.; Xu, B. A Physical Statistical Algorithm for Inverting LST from Passive Microwave AMSR-E Data. Scientia Sinica(Terrae) 2006, 1170-1176.
27. Tan, J.; Esmaeel, N.; Mao, K.; Shi, J.; Li, Z.; Xu, T.; Yuan, Z. Deep Learning Convolutional Neural Network for the Retrieval of Land Surface Temperature from AMSR2 Data in China. Sensors 2019, 19, 2987, doi:10.3390/s19132987.

 

6. In Section 2.3.3, why the authors considered "10-fold cross-validation"? , . Authors need to consider more explanations.

The authors considered "10-fold cross-validation" as it is a widely used technique in machine learning to evaluate the performance of a model on a given dataset. It involves partitioning the data into 10 subsets, using 9 subsets for training and 1 subset for testing, and repeating this process 10 times with each subset being used for testing exactly once. This helps to reduce the risk of overfitting and provides a more accurate estimate of the model's performance.

However, we acknowledge that further explanations are needed to clarify the authors' decision to use this technique. We will work on providing additional details in the revised manuscript to better explain the rationale behind our approach. Thank you for bringing this to our attention.

 

7. The Discussion should be more explicit and contain more quantitative and practical data.

Thank you for your feedback. In order to more clearly indicate the validation based on practical data and the accuracy of the results in the discussion, we have made revisions in the resubmitted manuscript. Please refer to the resubmitted manuscript for details.

The main purpose of this study is to explore the applicability of five classic machine learning passive microwave downscaled methods on semi-homogeneous underlying surfaces. Unfortunately, there is a limitation in the paper that only one measured station was available to provide temperature information during satellite overpass in the selected typical semi-homogeneous underlying surfaces. In future research, we will expand the study area so that more stations can be used to quantitatively verify the results.

8. Minor English grammar and spelling checking is required.

Thank you for bringing up the issues with English grammar and spelling. We have had the manuscript edited by MPDI and we will also attach the certificate of English editing in the attachment. Thank you again for your suggestions.

 

Once again, thank you for your valuable feedback. We will make the necessary revisions and resubmit the paper.

 

Sincerely,

Yongkang Li

 

Author Response File: Author Response.pdf

Reviewer 2 Report

Thank you for inviting me to review the entitled manuscript "Applicability assessment of passive microwave LST downscaling over semi-homogeneous desert underlying surface based on machine learning ". This study utilized Catboost model to downscale the mesoscale surface temperature using microwave data and obtained all-weather surface temperature data, which is innovative to some extent but still has some issues:

1. The citation content is incorrect. Please check if the cited references and their expressions in the text match. For example, citation 5 (‘Mapping regional turbulent heat fluxes via variational assimilation of land surface temperature data from polar-orbiting satellites’) mainly studies heat flux, which is not appropriate to be placed under the urban heat island effect.

2. The use of Catboost in the article is a highlight, but it is not mentioned in the abstract and introduction. It is only introduced in line 258 of the text. It is recommended to modify the logic.

3. It would be helpful to introduce the "classical models" mentioned in the abstract and the conclusion in the introduction.

4. MODIS land surface temperature product has certain limitations, and when using it as the ground truth, it is necessary to ensure that high-quality data is used. It is important to provide information on the specific data used in the experiment, such as the image date, quantity, and local temperature level. Additionally, it would be helpful to provide information on the dataset splitting ratio and whether any optimization was used. This would allow readers to better understand the quality and reliability of the results obtained from the experiment.

5. Adding a technical flowchart in the "Methods" section is necessary, and it should integrate section 2.2.3.

6. Parameter is written incorrectly in line 238.

7. It would be helpful to explain the meaning of the red solid line and dashed line in Figure 2.

8. The color bar in Figure 3 should be adjusted as it currently appears to belong only to Figure 3b.

9. The discussion and experimental results section can be adjusted. The results related to the main experiment can be placed in Chapter 3, while supplementary comparative experiments and analysis of results related to the main experiment can be placed in Chapter 4.

10. The article lacks downscaled result images and analysis, and only made comparisons based on accuracy indicators. It is necessary to add images of downscaled results and provide a visual comparison. As the current downscaled product result in Figure 8 is not very ideal, the article model's results have a significant pixelated feeling compared with MODIS, similar to directly resampled results.

11. The numbering of Chapter 5 is incorrect, please correct it.

Author Response

Dear Reviewer,

 

Thank you for your valuable feedback on our work. We appreciate your positive comments and will strive to address the minor problems that you have identified in your review.

 

We also appreciate your comments on how we can improve the clarity and understanding of the paper. The specific response to the issue is as follows:

 

1. The citation content is incorrect. Please check if the cited references and their expressions in the text match. For example, citation 5 (‘Mapping regional turbulent heat fluxes via variational assimilation of land surface temperature data from polar-orbiting satellites’) mainly studies heat flux, which is not appropriate to be placed under the urban heat island effect.

Thank you for bringing this to our attention. We apologize for the error in citation content. After reviewing our references, we have found that citation 5 is not relevant to the urban heat island effect and we will remove it from the paper. We appreciate your thorough review of our work and will ensure that all references are appropriate and accurately cited in resubmitted revisions.

 

2. The use of Catboost in the article is a highlight, but it is not mentioned in the abstract and introduction. It is only introduced in line 258 of the text. It is recommended to modify the logic.

Thank you for your valuable feedback. The Catboost algorithm was a carefully selected algorithm. Previous studies have shown that Catboost outperforms other algorithms such as Random Forest and Deep Networks in terms of computational efficiency and model accuracy. We agree that the use of Catboost is an important aspect of our research and should be highlighted in the abstract and introduction. We will modify the logic of the paper to ensure that the use of Catboost is mentioned earlier in the text, and we will update the abstract and introduction accordingly. We appreciate your suggestions and will make the necessary changes to improve the clarity and coherence of our paper.

3. It would be helpful to introduce the "classical models" mentioned in the abstract and the conclusion in the introduction.?

Thank you for your suggestion. We will revise the introduction to include a brief introduction of the classical models mentioned in the abstract and conclusion.

4. MODIS land surface temperature product has certain limitations, and when using it as the ground truth, it is necessary to ensure that high-quality data is used. It is important to provide information on the specific data used in the experiment, such as the image date, quantity, and local temperature level. Additionally, it would be helpful to provide information on the dataset splitting ratio and whether any optimization was used. This would allow readers to better understand the quality and reliability of the results obtained from the experiment.

Thank you for your valuable suggestions.

Firstly, there are certain limitations to MODIS surface temperature. We specifically used an average LST error <1K pixel for cross-validation in downscaled results, which indicates a certain level of evaluation value for MODIS. Additionally, to further evaluate the downscaled results, we also applied for evaluation of multi-layer soil temperature at the Fukang Desert Ecological Station, which is the only station in the study area. This study evaluated the applicability of five microwave downscaling methods on a semi-homogeneous surface in 2019. The remote sensing data used were passive microwave brightness temperature data from AMSR-2, MODIS MYD11A1 surface temperature data, and MYD13A2 vegetation index data. We did not partition the data by time to train the model, but rather used all data throughout the year to train the machine learning model.

In order to provide readers with a clearer understanding of the data used in the study, we have modified the following sentence in the abstract section: Focusing on the semi-homogeneous underlying surface of Gurbantunggut Desert, Taking the 365 days of AMSR-2 and MODIS data (which can be scanned once during the day and night) in 2019 as an example, we evaluated the applicability of five classical, passive microwave, downscaling methods based machine learning of Catboost.

5. Adding a technical flowchart in the "Methods" section is necessary, and it should integrate section 2.2.3.

Regarding the data processing part, I have added a flowchart(Figure 2) based on your suggestion. The flowchart is shown below.

Figure 2 Flowchart of ata processing steps.

 

6. Parameter is written incorrectly in line 238.

Thank you for bringing this to our attention. We have reviewed line 238 and have corrected the parameter. We apologize for any confusion this may have caused.

7. It would be helpful to explain the meaning of the red solid line and dashed line in Figure 2.

Thank you for your valuable feedback. We apologize for the oversight in not providing an explanation for the red solid line and dashed line in Figure 2. The red solid lines represent the 25th and 75th percentiles of the correlation, while the dashed line represents the median of the correlation statistics. We have updated the resubmitted manuscript to reflect this information.

8. The color bar in Figure 3 should be adjusted as it currently appears to belong only to Figure 3b.

Thank you for your valuable feedback. We apologize for the confusion caused by the color bar in Figure 4. We will adjust the color bar to make it clear that it applies to both Figure 4a and 4b.

The figure number for Figure 3 has been changed to Figure 4 due to the addition of the flow chart in Figure 2.

9. The discussion and experimental results section can be adjusted. The results related to the main experiment can be placed in Chapter 3, while supplementary comparative experiments and analysis of results related to the main experiment can be placed in Chapter 4.

Thank you for your valuable feedback. We apAs per your suggestion, there are indeed some materials that can be added as supplementary discussion, such as the statistical section on relevance. However, if we were to make significant changes to the content of the chapters, it may disrupt the existing logic. Therefore, we did not make significant changes to the content of the chapters in the submitted manuscript. We appreciate your suggestion to adjust the structure of our manuscript.

Thank you once again for your valuable feedback. We appreciate your guidance and will certainly remember your suggestions in our future writing. We look forward to submitting improved manuscripts in the future and working with you again.

The logic is “3.1 showed the correlation between the characteristic factors of semi-heterogeneous underlying surfaces and land surface temperature, which provided a preliminary evaluation of the selection of each factor and its contribution to the results. 3.2 showed the 10-fold cross-validation results of five classic models during the training process. 3.3 conducted cross-validation of the five classic downscaled models using MYD11A1's LST error <= 1K per pixel, selected the optimal model, and analyzed the contribution of each characteristic factor to the Catboost model. 3.4 analyzed the relationship between the downscaled results of passive microwave land surface tempera-ture and multi-layer soil temperature using data from the unique measured station in the study area. Part 4 quantitatively discussed the accuracy of current downscaled re-sults and related research, and provided prospects for future work. Part 5 presented the main conclusions of the paper”.

10. The article lacks downscaled result images and analysis, and only made comparisons based on accuracy indicators. It is necessary to add images of downscaled results and provide a visual comparison. As the current downscaled product result in Figure 8 is not very ideal, the article model's results have a significant pixelated feeling compared with MODIS, similar to directly resampled results.

Thank you for your valuable feedback. We acknowledge that the article lacks visual representation of the downscaled results, and we understand the importance of providing a visual comparison.

As you mentioned, "the article model's results have a significant pixelated feeling compared with MODIS". This phenomenon does exist in our study area. In this study, we used vegetation indices as auxiliary data for downsampling. The images presented in the article are for the month of August, during which desert vegetation withers due to high temperatures and lack of water. Therefore, the heterogeneity information that can be provided by vegetation indices is greatly weakened. The feature vectors selected in this article are all spectral characteristics or spectral index data provided by remote sensing, which completely rely on remote sensing data. In the future, we will further add non-remote sensing data, such as elevation, slope, surface thermodynamics, and thermal roughness (currently ongoing work), to participate in downsampling research. In addition to the impact of vegetation on roughness, I believe that the undulating terrain of the desert will also affect the reflection characteristics of heat to some extent.

Thank you again for your valuable feedback. We will work hard to address these issues in our future research.

11. The numbering of Chapter 5 is incorrect, please correct it.

Thank you for bringing this to our attention. We apologize for the error in numbering and have corrected it immediately. We appreciate your attention to detail and your valuable feedback.

Once again, thank you for your valuable feedback. We will make the necessary revisions and resubmit the paper.

 

Sincerely,

Yongkang Li

 

Author Response File: Author Response.pdf

Reviewer 3 Report

This manuscript studied five classical LST downscaling methods on the semi-homogeneous underlying surface in the Gurbantunggut desert. The paper also describes the studied area, input data (AMSR-2, MODIS, and ground stations), data processing, and validation methods. Accordingly, This contribution has downscaled the LST to 1 km spatial resolution using brightness temperature from a microwave sensor (AMSR-2) using the Catboost algorithm and machine learning. This contribution is a case study useful for various remote sensing, physical and environmental communities. Generally, the paper’s architecture is good; however, some parts should be improved by adding details.

    The introduction is straightforward, providing background on LST downscaling methods. Many related works have been cited to give readers a clear idea and link this work with other papers downscaling the LST or other variables using different techniques and models. Practical methods and information were introduced in the methodology, notably the studied zone, the used data sources, the processing steps, validation, and correlation. The case study has been explained as well. The result section is clear and rich in correlation graphs and maps. The conclusion is also consistent.

I would suggest a few minor revisions:

Why have you strictly used the polar satellites in the “2.2.1 Remote Sensing Data” section? Himawari and FenYung GEO satellites also cover the study region, providing LST and NDVI with high temporal and spatial resolution. Could you justify the selected data sources in the paragraph?

In the “2.2.3 Data Processing” section, please try to add a visible flowchart illustrating the data processing steps. It would be best if you also used a small algorithm for processing. For example, have you used the cloud mask to filter out the cloudy pixel or only the LST_Error?

In brief, I suggest accepting the paper after correcting the manuscript based on these suggested comments.

Author Response

Dear Reviewer,

 

Thank you for taking the time to review our paper entitled " Applicability assessment of passive microwave LST downscaling over semi-homogeneous desert underlying surface based on machine learning". We appreciate your positive feedback and are glad to hear that you found the paper well written.

 

We also appreciate your comments on how we can improve the clarity and understanding of the paper. The specific response to the issue is as follows:

 

1. Why have you strictly used the polar satellites in the “2.2.1 Remote Sensing Data” section? Himawari and FenYung GEO satellites also cover the study region, providing LST and NDVI with high temporal and spatial resolution. Could you justify the selected data sources in the paragraph?

Our study area is located in the Gurbantunggut Desert in western China. The coverage area is at the outermost edge of Himawari's full disk image. Due to the unique nature of the semi-homogeneous underlying surface in the study area, other areas were not chosen despite being farther away from the center of Himawari. In addition, for the 1km medium resolution land surface temperature data obtained through remote sensing inversion, MODIS data has been widely recognized for its quality on a global scale by most scholars. Finally, because AMSR-2 inherited the observation mission of AMSR-E, the time difference between the daily transmission of GCOM-W satellite's AMSR-2 and Aqua satellite's MODIS is less than 30 minutes during a single scanning period (16 days). In the study area, the observation times of the two are very close.

In future studies on downscaling surface temperature over heterogeneous underlying surface types such as grasslands and mountains, I will take into account the valuable advice provided by you. Thank you once again for your valuable advice.

 

2. In the “2.2.3 Data Processing” section, please try to add a visible flowchart illustrating the data processing steps. It would be best if you also used a small algorithm for processing. For example, have you used the cloud mask to filter out the cloudy pixel or only the LST_Error?

Regarding the data processing part, I have added a flowchart(Figure 2) based on your suggestion. The flowchart is shown below.

Figure 2 Flowchart of ata processing steps.

       Regarding the question "have you used the cloud mask to filter out the cloudy pixel or only the LST_Error?", my answer is "Yes, I have." In the "2.2.3 Data Processing" section of the manuscript, it is stated that "MYD11A1 pixels with quality average LST error <= 1K were selected to match AMSR-2 microwave BT data and MYD13A2 vegetation index data."

 

Once again, thank you for your valuable feedback. We will make the necessary revisions and resubmit the paper.

 

Sincerely,

Yongkang Li

Author Response File: Author Response.pdf

Reviewer 4 Report

Dear authors

Your paper is about assessment of applicability of passive microwave LST downscaling, based on ML.

The paper is well written. You showed that it is possible to use the Catboost algorithm for downscaling of the LST data based on the BT of multichannel passive microwaves.

I have a few comments that go in the direction of a better understanding of your paper:

L64-65: rephrase the sentence to make it clearer.

Figure 1: I think that it will be better to rearrange this figure, study area to be bigger part of this Figure 1, and surrounding area smaller.

L190-L224: It will be easier to understand if you show this part of the text through a graph with processing steps

…..

All my other comments are provided in the accompanying document.

Thank you.

Comments for author File: Comments.pdf

Author Response

Dear Reviewer,

 

Thank you for taking the time to review our paper entitled "Assessment of Applicability of Passive Microwave LST Downscaling, Based on ML." We appreciate your positive feedback and are glad to hear that you found the paper well written.

 

We also appreciate your comments on how we can improve the clarity and understanding of the paper. The specific response to the issue is as follows:

 

1. L64-65: rephrase the sentence to make it clearer.

This is the result after rephrasing: The fusion and downscaling of passive microwaves to obtain continuous LST with high spatio-temporal resolution is a promising field of research.

2. Figure 1: I think that it will be better to rearrange this figure, study area to be bigger part of this Figure 1, and surrounding area smaller.

Figure 1 Overview of study area.

 

3. L190-L224: It will be easier to understand if you show this part of the text through a graph with processing steps

Figure 2 Flowchart of ata processing steps.

4. Please explain this sentence more detail. “Interestingly, some phenomena did appear that were not found in the downscaling work carried out on the mixed underlying surface.”

As the concluding paragraph of our discussion, this sentence summarizes our finding of a phenomenon not observed in downscaled studies on other underlying surfaces. The phenomenon we found specifically refers to the third paragraph of our discussion.

Correlation coefficient characteristics of factors of microwave BT and LST: Within the rectangular study area range (including the semi-homogeneous underlying surface of Gurbantunggut Desert and its surrounding oases), the statistical results of correlation (Figure 3) showed that the correlation coefficients between BT at 89 GHz and LST were higher than those of BT at lower frequencies, such as 23.8 GHz and 36.5 GHz. This result is consistent with the study by Mao Kebiao [32,33], whose findings represent the mainstream views [24,42]. However, regarding the semi-heterogeneous underlying surface of Gurbantunggut Desert, the correlation coefficients of BT at relatively lower frequencies was higher than that of BT at 89 GHz (Figure 3). It may be that vegetation, surface roughness, or other factors impact the correlation between BT and LST. This phenomenon further indicates that it is important for us to assess LST downscaling methods using passive microwaves on particular underlying surfaces of different complexity.

5. Other problems.

Other minor issues have been corrected in the resubmitted manuscript. For example, we have replaced low-resolution images with higher-resolution ones to address issues with image clarity.

 

Once again, thank you for your valuable feedback. We will make the necessary revisions and resubmit the paper.

 

Sincerely,

Yongkang Li

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

The author has made significant efforts in revising the manuscript, highlighting the core content and the validity of the experiment. Most of the previously raised issues have been satisfactorily addressed. However, there are still some remaining issues as follows:

 

1. The issue regarding the red solid and dashed lines in Figure 3 remains unresolved in the new manuscript. It is recommended to include an explanation of these lines in the figure caption to assist readers in understanding.

Author Response

Dear Reviewer,

 

Thank you for your Positive feedback of the other problems related to our previous version of the manuscript. At the same time, we deeply apologize for the unresolved problem caused by our carelessness. The specific response to the unresolved problem is as follows:

 

1. The issue regarding the red solid and dashed lines in Figure 3 remains unresolved in the new manuscript. It is recommended to include an explanation of these lines in the figure caption to assist readers in understanding.

Thank you for your valuable feedback. The red solid lines represent the 25th and 75th percentiles of the correlation, while the dashed line represents the median of the correlation statistics.

We have added a clear annotation on line 358 of the resubmitted manuscript that reads as follows: "In Figure 3, the red solid lines represent the 25th and 75th percentiles of the correlation, while the dashed line represents the median of the correlation statistics."

 

Once again, thank you for your valuable feedback. We will make the necessary revisions and resubmit the paper.

 

Sincerely,

Yongkang Li

 

Author Response File: Author Response.docx