Construction and Evaluation of Prediction Model of Main Soil Nutrients Based on Spectral Information
Round 1
Reviewer 1 Report
This manuscript attempts to explore the quality of fitness of four regression models in the estimation of soil nutrients from hyperspectral data. The subject could be of interest to the readers of Applied Sciences, unless the presentation was not so fuzzy and obscure. More specifically:
- The main thesis of the manuscript should be better explained, its practical value should be highlighted, and it should be more thoroughly researched and documented. Plenty of journal articles refer on this subject, for example, "Lixin Lin, Yunjia Wang, Jiyao Teng and Xiuxiu Xi: Hyperspectral Analysis of Soil Total Nitrogen in Subsided Land Using the Local Correlation Maximization Complementary Superiority (LCMCS) Method, Sensors 2015, 15, 17990-18011" so the authors' main contribution should be thoroughly shown and documented bibliographically, because it is the basis of the originality of the manuscript.
- In line with the previous remark, paragraph 3.2 which is central to the understanding of the authors work, should be completely rewritten in detail. The regression methods should be described in brief and their differences explained. The results should be analytically presented with explanations and appropriate graphs.
- The experiments conducted in the framework of this study should be better documented with photographs of experimental installations, flowcharts, etc.
- The use of figures, maps and diagrams should be considerably improved and connected to the text in order to make the manuscript more readable. The authors have to explain in detail how they produced the diagrams from the original data.
I have included detailed remarks in the attached sanitized manuscript. My suggestion is to reject the manuscript in its present form. The authors should improve the overall quality of the manuscript, prove it's originality and then resubmit it to this or another journal.
Comments for author File: 
Comments.pdf
Author Response
Please see the attachment.
Author Response File: 
Author Response.docx
Reviewer 2 Report
Simple work comparing different statistical methods to estimate the content of the most important nutrients for the soil in cotton cultivation. The care of the authors with the samples and the preparation of the experiments, which was described in a very detailed way, should be highlighted. Although the authors reached a satisfactory result about which statistical methodology is the most appropriate, they emphasize in the conclusions that everything was done in a controlled environment and, when data are collected in the field with a spectroradiometer, many other factors can influence the result, and not just the moisture that was worked on. Thus, I suggest that the conclusions of the work be improved and indicate the feasibility of using the approach tested in practice, even considering that a small amount of samples was used in a specific location, making it difficult to generalize the solution.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 3 Report
I would like to thank the authors for the work.
I read the manuscript carefully and I have just a very few comments:
Line 16: (n, P and K) nutrients need definition???
Line 56: Spectral by itself is not technology, (Field spectroscopy, Imaging Spectroscopy, Hyperspectral Remote Sensing are actually technologies)
Line 63; soil hyperspectral. Maybe soil spectral signal is more appropriate
Line 75 : based on NIRS. Write the acronym
Line 106: Table 1. This is a table. Please rewrite
Line 123: Table 2 shows the.. Please reformatting
Line 135: standard calibration was performed on the instrument?. Please better explain, you mean optimization?
Line 136: acquisition software Rs3. Software ©
Line 137: and the system fit a spectrum curve???. Please better explain
Line 145 using View SpecPro. Software ©
Line 146: Splice correction?, Please better explain
Line 152: what is the K???
Line 222: Figure 4. This is a figure. Schemes follow the same formatting. Please rewrite
Line 311: section is maybe more a conclusion than a Discussion
Line 355: references must be reformatted
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
This is a revised version of a manuscript that attempts to explore the quality of fitness of four regression models in the estimation of soil nutrients from hyperspectral data. The authors did not take seriously my previous remarks and the only substantial change they did was to add a photograph of the experimental installation. The manuscript remains fuzzy and obscure. Mainly because:
- The authors still do not explain their main contribution and the originality of the manuscript. Even the title remains general and not informative.
- Paragraph 3.2, which is central to the understanding of the authors work, remains fuzzy and confuzing. Still contains vaque expressions like: "When conducting model prediction training on soil N content, PLS needed to use the interactive verification of the RMSEC value of the built model with different principal components (Pcs), as shown in Figure 4a), when the Pcs was 5, the RMSEC value was the smallest.".
My final suggestion is to reject the manuscript.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 2 Report
Authors made changes as requested.
Author Response
Thank you for your valuable suggestions on this manuscript.
