Confidence Intervals and Regions for Proportions under Various Three-Endmember Linear Mixture Models

Round 1

Reviewer 1 Report (Previous Reviewer 4)

Thanks for addressing my comments. I have no more comments.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report (Previous Reviewer 2)

Dear Author,

The paper is very theoretical. According to me it could be valuable to give (in the conclusions) some examples of practical usage of this methodology, for example agriculture, cities landscape etc. Another remarks are: to write some words whether the elaborated methodology can be used for higher resolution images (for example Sentinel); to write which model is better to use in practice (for instance in agricultural drought analysis). In methodology, I advise to write what areas in Australia and what years these large group of remote sensing data covered. It should also be written, what plants were covered by PV images (e.g. grasses, crops, orchards, forests, others?) what areas and what periods were covered by the NPV images (plants dormancy?, or just after harvesting?) and any similar example concerning bare soils. I understand, that water bodies were excluded from the analysis.

Kind regards

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report (New Reviewer)

Abstract:

It would be nice to add a sentence at the beginning to explain why “Many papers in recent years have been devoted to estimating the per pixel proportions of three broad classes of materials”. The ultimate goal that they are trying to achieve by doing so.

 

Introduction:

The manuscript provides a good introduction.

I find that your manuscript fit most appropriately with the journal’s scope to “publish novel/improved methods/approaches and/or algorithms of remote sensing to benefit the community, open to everyone in need of them”. So, it would be nice to directly and explicitly explain how the model is used in the context of remote sensing and its benefit to the wider community.

If would be nice if you could expand on why it is important to use linear mixture model and how it is better/more efficient/differ from previous methods.

Could you provide more details on the current state of research in this area and the significance of your new research.

 

Methodology:

The manuscript discusses several different models and methods used in linear mixture modeling, it would be great if you could tie each one back to the main focus of the paper.

Are there any limitations of data set or the number of samples in your methodology?

Could add a little bit more details on how the data were collected, such as the field-estimate location and timing of the measurements. Optionally, I would not mind seeing some explanation on how the exploratory data analysis was conducted, including any statistical techniques used to analyze the data.

You research contains a lot of data, not saying that you need more, but could you give support on why 1169 Landsat TM spectra and associated field measurements collected at 913 sites around Australia is sufficient?

 

Discussion:

The discussion section of the manuscript provides a clear overview of the models analysed and the results obtained.

It would be great if you could provide a more detailed interpretation of the results such as how much more efficient/accurate it is with your method to produce CI and JCRs.

Could add more on the implications of the findings. How can future research benefit from this. What are some of the potential limitations.

 

Conclusion:

The conclusion concise and well presented.

 

Overall:

 

The scope of this manuscript could be improved by bring more common remote sensing elements (ex. How to apply this) into discussion and compress some mathematical formulas.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

General:

This paper focuses on generating proportional confidence intervals and joint confidence regions under PL and NNL models to assess the accuracy of estimators. The manuscript should undergo extensive revisions.

 

Major comments:

There are some major issues in the manuscript.

1. γ specific expression should be introduced in Equation 15, and is the γ the same as the formula 20?

2. Equation 20, which I think may be ambiguous, suggests that  “” can be omitted or subscript ””

3. The annotation in Figure 3 is not very clear, what does its horizontal and ordinate represent?

4. There are many fonts in the text that are italicized, such as “either” in line 246, is it the focus or a mistake? It is recommended to change to a uniform font.

 

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Title of the article: Confidence intervals and regions for proportions under various three-endmember linear mixture models

Comments and Suggestions:

The article presents an attempt to solve the important problem of the accuracy of environmental information estimated on the basis of remote sensing. It is especially important, when it concerns the analysis of smaller areas, e.g. several or several hectares. Higher resolution images are needed for this, but these are expensive. Therefore, data from in situ measurements are an important factor, which can be helpful in assessing the accuracy of the results obtained from medium resolution remote sensing.

The article presents a detailed description of the mathematical solution to the problem of the confidence interval concerning the analysis of hyperspectral imagery. The Author shows how to create confidence intervals (CIs) and common confidence regions (JCRs) for ratios (proportions p) associated with various linear mixture models.

The method was elaborated on the analysis of dataset of 1169 six-band Landsat TM spectra and associated field measurements from 913 sites in Australia.

The author's very detailed reasoning is presented in the article. There are links to articles that we would need to read/see to understand the exact algorithms presented.

Please, explain the criterion of choosing the spectra analysed in the paper. Could You please, give some examples of photosynthetic vegetation (PV), non-photosynthetic vegetation (NPV) and bare soil (BS) taken account in your research? I think it would be good to validate your methodology on one area in Australia covered by either PV, NPV or BS. Then, your methodology will be better understood for non-mathematical auditorium but also researchers dealing with RS, geography and agriculture.

In the last Chapter the Author discusses his own results. It is worth to add few research articles to this discussion.

I recommend the article for publishing after considering the comments described above.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

The paper assess the accuracy of their estimators and produce confidence intervals (CIs) and joint confidence regions (JCRs) for the proportions associated with various linear mixture models. I read this article as a whole but in my opinion current version of the manuscript was not suitable for publication.

The detail comments can be found below.

1. The paper focus on an interesting topic but the structure of the paper was not organized reasonable. Such as the results and methods were mixed together and discussion and conclusion were mixed, too.

2. The Introduction should be more focus and have a logical expression.  

3. The methods should not stack formulas and should focus on your topic. Readers always want to something more than common sense.

4. Results is an important part of your research. But author mixed it with the methods.

5. Discussion section should be rewritten and deeply discussed.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 4 Report

The font size in some figures are too small.

Figure 1: is it the average value of the dataset? It would be better to show the variance.

Line 161: How is the nominally purest defined?

Section 3: It would be better to add some discussions regarding the practical applications of JCRs.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 3 Report

Thanks for author to revise his manuscript. Although author do a lot of work, but the results part only has a few figures to support the whole manuscript. And the manuscript also has a big language representation problem. I still think the current version was not suitable to publish in Remote Sensing.

 

Special comments:

1. Line 23， [1-12], It is a citation train, you needn’t to add an input to the manuscript of all used references. So, please highlight the most important elements of the cited papers.

2. Line 25,’In [15], shade is a fourth endmember, added to PV, NPV and BS. In [16],.....’. Authors usually used the sentence such as ‘XX defined shade as the fourth end-member[15] ’ or ‘Shade was defined as fourth end-member in XX’ research[15]’, but not use ‘in[15],......’ .

3. Line 33,  AutoMCU method, the abbreviations should be given the full name, even though most people know what it means.

4. Line 37, I show how......, it is usually stated in the third-person narrative form in scientific paper such as this research...., the author....., and so on.

5. Express logically. Take introduction section as example, line 37 to 49, in these paragraph, the author want to express how to produce CIs and related, and explain the reason why to select three endmember models, and this part is the work of your studies. And the next paragraph, author started talking about other previous research. And a better logically expression, the previous research comparison and advantage or disadvantage described first, then introduce your research aim and related methods, especially for the shortcomings of previous studies.

6. Line 72, AutoMCU was not explained in the first scene but here.

7. ‘The lead author of that paper asked me how to construct CIs for his proportion estimators.’ This is a research paper but not Q&A in twitter.