Learning the Indicative Patterns of Simulated Force Changes in Soil Moisture by BP Neural Networks and Finding Differences with SMAP Observations

Round 1

Reviewer 1 Report

From the perspective of the full text, this paper tries to do two things. The first one is to train the soil moisture of global map with BNPP, establish the relationship between soil moisture and year, and analyse which regions are the main features through DTD. The second thing is to analyse some of the forcing factors that affect climate change. However, these two things do not work well together, leading to the following problems:

1. The topic of the paper is not clear. It is suggested to modify the structure and content of the paper after clarifying this problem.

2. The expression of Abstract is not clear, such as the paper for what problem, mainly carried out what research, how to carry out, what research results are, what conclusions, etc.

3. What is the main idea of the paper? Is it to study the response relationship between climate change and its influencing factors, or the relationship between soil moisture and them? The Introduction should be clear what the problem is to solve for this paper. The structure and thinking of the Introduction is not clear enough.

4. SMAP observation data (2006-2020) cannot be found in Table1.

5. What does atm/ocn mean in Table 1?

6. What is the format of the dataset? Text or pictures? If it is in text format, how is it converted to the input image format?

7. For BNPP, the size of data sets is an important factor affecting its performance. According to the general rule, there is only one global map per year, which cannot meet the requirements of BNPP learning. How does the paper solve the problem of insufficient data?

8. In Section 3.1, why the details of the BPNN have been put into the Appendix A, which should be described in the main text.

9. Descriptions of BNPP parameters should be given in the main text, not Appendix A.

10. Some important parameters of BNPP are unclear, such as learning rate and iteration times.

11. When the DTD analyses the significant area features (Figs. 4-6), it is recommended to be highlighted, and how these significant features are represented, which should be explained in the figure.

12. BNPP studies the areas that affect soil moisture, and how are these linked with forcing factors?

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Dear Author(s),

Please find below my concerns and recommendations regarding your manuscript proposal entitled "Learning the Indicative Patterns of Simulated Forced Changes in Soil Moisture by BP Neural Networks and Finding Differences with SMAP Observations".

 

1. In the Introduction, please clearly define and describe the following important aspects:

- the research gap: based on the previous results from the literature, define the gap covered by your article;

- the research question(s): based on the research gap, define and describe the research question(s) your article is answering to;

- the research goal: the readers should know from the very beginning of the manuscript which is the goal of your article.

 

2. The acronym "SMAP" appears within the Abstract on the first page, but its complete description ("Soil Moisture Active and Passive") is revealed on page 3 at row 124.

Please describe the acronym earlier, at the first appearance.

 

3. At the end of the Introduction section, you should define the research hypotheses.

I tried to find at least one research hypotheses in your manuscript, but I couldn't.

 

4. The references you used are quite old and quite few. In order to improve the general context of your research, I recommend you to include in your work the following valuable resources:

https://doi.org/10.1016/j.jhydrol.2021.126104 (these articles offer important details about BPNN Training), https://doi.org/10.1080/00949655.2018.1463376, https://doi.org/10.3390/su12104142 (here you will find useful details about MLR-Multiple Linear Regression), 

https://doi.org/10.1007/s10639-018-9820-5 (details about fuzzy classification). 

 

5. In the section Results I recommend you to add a new distinct chapter entitled "Discussions".

Here you should discuss about your results and the other results from the literature.

Make comparisons, discussions and analyses.

 

6. The Conclusions sections must be improved by presenting the following important aspects:

- the research limitations: be honest and shortly present the limitations of your approach;

- the managerial implications (here is the place where you can "sell" your results to the readers);

- the future research directions. 

 

Dear Author(s),

Please consider all the above remarks as being constructive recommendations in order to improve the general quality of your manuscript proposal.

 

Kind Regards!

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

After revision, the paper is clearer. However, there are still some problems to be confirmed:

1. According to the description of the paper, soil moisture is affected by many factors, and existing models for predicting soil moistures has some problems, and the author tries to provide a new model method with BP Neural Networks, is that right? If this, this point should be made clear in abstract and introduction.

2. The author pointed that the results are: By comparing the standard multiple linear regression method, results demonstrated ANN model can gain reliable indicators.”. Then the abstract should have data or explanatory language to prove the result, rather than a simple description

3. line20: Since the ANN Model can obtain reliable indicators, what do these indicators exactly refer to?

4. The author pointed that the conclusions are: “This study shows that using ANN as well as visualization tools can clearly discern the differences between model simulations and observed data.”. But this is more like the second result. It is suggested that the author first make clear the difference and connection between the results and conclusions, and then make corresponding modifications.

5. Line25-26：what does model simulation mean? The SMAP data or predicting data from ANN model? Why should it be distinguished from observations?

Author Response

Thanks for your advice. We have addressed the Reviewer’s comment in the following attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

Dear Author(s),

I have read the revised version of your manuscript proposal and I appreciate you efforts to improve the article.

However, certain aspects should be improved.

Please see below my concerns regarding the new version of the manuscript.

 

1. According to my previous constructive remarks, I recommended you to clearly define the research hypothesis at the end of the Introduction section.

Now, between lines 66-69 you say: "“The purpose of this paper is to learn the indicator of the variability of anthropogenic forcing based on natural forcing and internal variability as well as model differences.”

But this is not really a research hypothesis. Please revise and improve this aspect.

 

2. In my previous review, at point #4, I recommended you some valuable references. You said that you used them, but in the final reference list they are not syncronyzed. For example, the doi of the reference #16 is correct, but its description is not correct. Please carefully re-synchronize the description so that it fits the doi.

Also, the reference https://doi.org/10.1007/s10639-018-9820-5 was not included in your manuscript, although it is very relevant.

Please revise all the recommended references from the previous round of review and implement them carefully in the text and in the final bibliography.

 

3. In the new version of the manuscript, the title of the section 6 is "6. DiscussionsResults".

Please revise and keep only the title "6. Discussions", because the results were presented in the previous chapter of the article.

 

4. The English should be revised by a professional proof-reader.

 

Dear Author(s),

I really hope these constructive recommendations will be useful for you to improve the quality of your article.

 

Kind Regards!

Author Response

Thanks for this advice. We have addressed the Reviewer’s comment in the following attachment.

Author Response File: Author Response.docx

Round 3

Reviewer 1 Report

The problems I mentioned have been modified.