Review Reports - Hyperspectral Inversion of Soil Carbon and Nutrient Contents in the Yellow River Delta Wetland

Round 1

Reviewer 1 Report

The authors have presented a detailed procedure to deduce the soil parameters such as total carbon (TC), total nitrogen (TN), and total phosphorus (TP) in wetland soils of the Yellow River Delta with the help of 240 sampling points and hyperspectral observations. The concept is well formulated but there are some issues which need to address. The specific comments are given below. Accordingly, a major revision of the manuscript has been recommended.

Comments:

1) Introduction: The introduction is not properly contextualized like literature on which bands are sensitive to what soil parameters. In L54-55, it is very limited.

2) Methods: (L135): how these models are developed and their training done

3) Fig 1, 2: Units missing.

4) Fig 2 does not give any insight on band sensitivity to TP, TN and TC. How many points are shown here? Both species are more or less the same spectral curve

5) Fig 3 to 7: Labels are not readable. here is also Unit missing for the x and y-axis

6) Fig 4 & 5: Better to introduce in the caption about OR, FD and SB

Author Response

Dear Reviewers:

Manuscript reference: diversity-1844782

We have attached a revised version of our manuscript entitled “Hyperspectral inversion of soil carbon and nutrient contents in the Yellow River Delta wetland”, which we would like to resubmit for publication as an original research article in diversity.

We would like to express our appreciation to both the Editor and the Reviewers for their constructive and insightful comments about our manuscript, which helped us improve our manuscript.

We have considerably modified the manuscript and have clarified details relating to the study as requested and recommended, but have not changed either the results or the overall framework of the study. We hope that our manuscript will now meet the standards expected for publication.

Our responses to the Reviewers’ and Editor’s comments are provided in this letter.

We hope that the revised manuscript and our accompanying responses will be sufficient for our manuscript to be suitable for publication in diversity.

We look forward to hearing from you at your earliest convenience.

Yours sincerely,

Author Response File: Author Response.pdf

Reviewer 2 Report

This paper uses hyperspectral data to measure soil carbon, nitrogen, and phosphorus in wetland soils. It details the use of modelling to predict these soil elements; Random Forest regression modelling using the first order differential of the spectral data was found to give the best predictions.

The paper is somewhat interesting, but needs rewriting to emphasise the rationale and purpose of this work. In particular, the abstract and conclusion need to explain the importance of the results for future research and applications.

I would also recommend adding more detail on the methods used, particularly as the audience for this journal are likely to be less familiar with hyperspectral data processing. It would be good to give a brief explanation of inversion modelling early in the article. There is also more detail needed on the selection of sensitive bands in Section 2.4. Significant differences are mentioned throughout the article, but there is no mention of the statistical methods used to determine these differences.

Throughout the article, care is needed to reference where appropriate. Many graph axis labels are missing, and those that are present are too small to easily read (the same applies to the graph legends).

Specific comments:

L73 – What list of wetlands is this referring to? If it is a list developed by an international organisation, the organisation should be named here.

L74 – how can it be 2-9m and also no higher than 15m? Should it be 2-15m?

Sections 2.1/2.2 – a map of the study area with sampling points would be helpful. Also, more information is needed on the random sampling method – were three samples of different depths taken at each location? L371 says sampling was stratified – which is correct?

S2.3 – two sample sets or three? Consistency needed here, or more explanation.

L91 – What traditional chemical methods?

L120-124 – I am a bit confused about the method used here. Was every band individually correlated with the chemical data? For the original and first derivative? And then the bands with significant correlations were selected as sensitive? Was there only one sensitive band or many?

L124 – reference R software

L130-133 – why the difference in methods for different banks?

L136 – full names and more details on these three models needed.

S3.1 – how were significant differences assessed? The statistics used need to be explained in the method.

Fig 1 – What units is weight shown in?

Fig 2 – My suggestion would be to show the two sample sets on the same graph, as semi-transparent ranges with an average line rather than showing the spectra of every sample in two separate graphs. This would make comparison easier.

S3.3 – how was significance tested?

L183 – In the Winter? But all samples were collected in October?

Fig. 3 – What do the axes show? This figure is unreadable, I cannot understand it.

Fig. 4-7 – axes labels needed

L300 – these ‘earlier studies’ need referencing

Table 2 needs referencing in-text

L340 – references for ‘recently published studies’ needed

Author Response

Dear Reviewers:

Manuscript reference: diversity-1844782

We would like to express our appreciation to both the Editor and the Reviewers for their constructive and insightful comments about our manuscript, which helped us improve our manuscript.

Our responses to the Reviewers’ and Editor’s comments are provided in this letter.

We hope that the revised manuscript and our accompanying responses will be sufficient for our manuscript to be suitable for publication in diversity.

We look forward to hearing from you at your earliest convenience.

Yours sincerely,

Author Response File: Author Response.pdf

Reviewer 3 Report

Title:

In this study, the authors only investigated the total carbon (TC), total nitrogen (TN), and total phosphorus (TP) in wetland soils. The term “soil elements” used in the title of the manuscript is too general and has to be specific. I would suggest the authors to consider revising the title as “Hyperspectral inversion of soil carbon and nutrient contents in the Yellow River Delta wetland”.

Abstract:

The abstract of the manuscript was poorly written and it doesn’t reflect the main findings of the study. It must be re-written. Here is a good example: https://doi.org/10.1016/j.catena.2021.105222

Introduction:

The introduction did not provide sufficient background and rationale to justify the innovation of this study, given the fact that many similar studies have been carried out by others. For example, Why different models need to be evaluated? Why different vegetation types need to be considered? What are the problems of previous studies? What are the main research questions of the current study? Moreover, the three main nutrients in soils are nitrogen (N), phosphorus (P) and potassium (K). Together they make up the well-known NPK. The soil carbon (C) stock is the amount of C stored in the soil and it is not part of the soil nutrients. The authors simply mixed up this concept.

Materials and methods:

A map of the study area and the distribution of soil samples need to be presented. The authors indicate that “The sampling points were randomly distributed in the study area”. What kind of sampling strategy has been used in this study? If it is the stratified random sampling approach that has been used, then what is the strata? Vegetation types or soil types? In addition, the authors attempted to explore the impact of sample division on the model prediction performance; however, the necessity and importance of such analysis were not introduced in the introduction of the manuscript. If the authors believe that this is one of key research questions or innovations of this study, then it must be introduced and justified in the introduction section, rather than suddenly appeared here.

Results:

Based on Figure 1, the authors claim that there were significant different in these three soil elements between vegetation types as well as between soil layers. But, there is no any statistical test that has been reported. Are the soil samples normally distributed? Which statistical methods have been used to test these differences? These all needs to be explicitly reported in the results. In addition, the caption of the Figure 2 is unclear. Are these the reflectance curves of the two vegetation types or the soil of these two vegetation types? Are these reflectance curves derived from a mixed soil samples of all three layers or just derived from the top soil layer? This needs to be clearly described. Moreover, based on Figure 3, the authors claim that “The correlation between the first-order differential spectral reflectance and soil C, N, and P was significantly improved compared with the correlation between the original spectral reflectance and soil C, N, and P”. How do we know it was significantly improved? Which statistical measure has been used in this regard? P value or correlation coefficient? This again needs to be explicitly reported. Furthermore, why the authors mixed the two species for the modelling purpose in Figure 4? Why the original spectral reflectance and first order differential spectral reflectance data were analyzed? What are the reasons for such an analysis? Is this a research question that has been set up in the introduction? Last but not the least, the authors wrote “3.5. Effect of sub-species and sensitive band modelling”. What do you mean the “sub-species”? In biological classification, the term subspecies refers to one of two or more populations of a species living in different subdivisions of the species' range and varying from one another by morphological characteristics. Here, the authors are dealing with two completely different species. How it comes to a term “sub-species”?

Discussion:

Table 2 has been displayed in the discussion section, but no idea which sentence it refers to. If the data transformation is so important in this study, why the authors didn’t bring this issue up in the introduction? Moreover, the authors wrote “By combining the three modelling methods, namely PLSR, RF, and SVM, the soil nutrient contents of TC, TN, and TP were estimated”, are the authors combined these three models? Apparently NOT! The authors just applied and tested these three models separately, not combine! In addition, what do you mean “4.4. Model Variance Impact”?

Conclusions:

The authors draw three key conclusions from the current study. I would suggest authors formulating three specific research questions or objectives in the introduction section to link these three key conclusions. In the meantime, the authors need to organize three key sub-sections in the results to clearly and logically support these three key conclusions.

Author Response

Dear Reviewers:

Manuscript reference: diversity-1844782

We would like to express our appreciation to both the Editor and the Reviewers for their constructive and insightful comments about our manuscript, which helped us improve our manuscript.

Our responses to the Reviewers’ and Editor’s comments are provided in this letter.

We hope that the revised manuscript and our accompanying responses will be sufficient for our manuscript to be suitable for publication in diversity.

We look forward to hearing from you at your earliest convenience.

Yours sincerely,

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The authors have improved the overall quality of the manuscript as per the queries. They have also improved all the Figures.

Author Response

We have touched up the language of the article. Thank you very much for your recognition and help with this paper, and we wish you good health and success in your work!

Author Response File: Author Response.pdf

Reviewer 3 Report

1. The map of the study area (Figure 1) provided by the authors doesn't make any sense. If we look at Figure 1, we still have no idea where the study area is. More importantly, we still don't know the actual spatial distribution of these 80 sampling sites. I suggest authors download a high-resolution satellite image (Landsat or Sentinel-2) taken in October 2021 and display the location of the 80 sampling points.

2. Line 199, which statistical method (not software) has been used to test the difference?

3. The English of the manuscript needs to be further polished by a professional English editor.

Author Response

Your comments on the article revisions have been made as requested, and we appreciate your help.

Author Response File: Author Response.pdf