A High-Precision Remote Sensing Identification Method on Saline-Alkaline Areas Using Multi-Sources Data
, Dingkun Chang 1,2,3, Wentao Xu 1,2,3 and Boqi Yuan 1,2,3,5Round 1
Reviewer 1 Report (Previous Reviewer 2)
No comments
Author Response
Thank you very much for reviewing our manuscript.
Reviewer 2 Report (Previous Reviewer 1)
The authors have solved all of my concerns and I would recommend its acceptance for publication.
Author Response
Thank you for your appreciation on the manuscpt.
Reviewer 3 Report (New Reviewer)
Revision of Research on High-precision Remote Sensing Identification
Method of Saline-alkaline Soil based on multi-sources data
Due to Saline-alkaline soil mapping and spatiotemporal changing, the paper covers an adequate topic. Climate change may increase desertification and wind erosion. Soil water cycles may also change. This may result in increased salt accumulation in the upper layers of soils in inland waters and drainage areas due to high evaporation.
The paper still has a few flaws, mainly structurally and regarding its lengthy and wordy style. It should be significantly changed, and it is especially true for the Materials and Methods and the Conclusion.
Although the paper is Saline-alkaline soil mapping oriented, this text is not able to supported. There is not adequate soil mapping without soil excavation and soil profile descriptions. The title of the paper implies more soil science, geography, spatial analysis, and validation of spatial analysis maps. Therefore, I would expect a map for instance the positions of Minqin County at China. It is not clear to the reader whether the authors have analyzed any soil profile or only investigate surface differences. The results are not supported by cited articles in the discussion.
The greatest problem is, there are not declaration of aims of article.
The chapter Conclusion is insufficiently ‘conclusive’. Broader implementation of the results is needed there.
General comment
The title should be modified: Saline-alkaline Soil correct form is: saline-alkaline areas identifying
Keywords should be modified with more specific words (e.g.): alkali flat, salinization identifying
Introduction
The introduction has sufficient size when it is compared to the other chapters. In spite of the important citations, relevant articles missed from other countries. Based on the citations the study has it seems that the mentioned problems could be only local ones. The salt affected soil can be found in semiarid and arid climates cover more than 25 % of the earth's surface. The aims are missing.
Data and methodology, Experiment
The chapters of “Data and methodology” and “Experiment” should be reworked into one chapter, so that the sample area is presented the first. These chapters size are 5 of 15.
Main problem with material methods and experiment is that the important information and simple insufficient construction of elementary description are missing.
By following the description, it cannot be replied to the experiments of sample taking and sample preparing methods. What is the typical soil at the sample sites? What was the classification of the soil type at sample site? (WRB and USDA soil classification). What is the description of soil profiles? What was the soil textural type? What was the particle size distribution? How many was the clay% content? what is the soil class distribution at the sample sites? How did the soil structure change? It is one of the most important parts when researcher writes about soil mapping article. One table is required about physical soil properties as particle size distribution, texture, CaCO3% content and pH. The process of deposition, the saline moving and the differences between O and A horizon properties. What were the background values of saline content? This part of manuscript should determinate the understanding. The current status of manuscript can be confusing for the readers.
It is not clear which satellite family's images were analyzed? In the results Landsat is included in the methods GF2/GF6. When and how many satellite images were compared? Comparing 3 images (2010, 2015, 2020) is not enough and not relevant.
Fig 1 is not readable.
Result and discussion
Discussion is poor.
The methods and result is not supported that „2)The surface soil of Minqin contains one or more layers of clay soil. The clay soil prevents the surface water from washing salt and discharging alkali to the soil, which is not conducive to soil desalination and thus causes soil salinization.” Are there clay soils?
The all of Maps and Diagrams are not readable in this form, the font size is too small. These Figures have to be remade again.
The references are contained too few international examples.
Specific comments and questions:
Line (L.) 34: need more international references
L. 70 - 74: in This part have to mention ecosystem services.
L. 92 – 99: Have the landsat or GF images been examined/compared?
L.93: What are meaning of the following acronyms: GF-2 and GF-6?
L. 176-185: What are the typical soils at study area? What were the soils physical properties?
L. 185: Figure 6. is not readable.
L. 186 - 233: How many sample points had been used to verification?
L. 252-253: How much was result of Kappa value for the Non-Saline-alkaline soil category?
L 275 - 276: The methods missed the Landsat image description. why were only 3 dates selected from Landsat picture?
L. 286 - 311: The result is not supported the Discussion of „2) The surface soil of Minqin contains one or more layers of clay soil. The clay soil prevents the surface water from washing salt and discharging alkali to the soil, which is not conducive to soil desalination and thus causes soil salinization.”
Conclusion
Unlike the other chapters of the "paper" is way too short and does not directly flow from the results and goals. Secondly, how it could be implemented and compared for other lands? What sort of areas should be analyzed with the presented method? In order to use your results, list areas where your results may be potentially implemented or employed. The prime problems are the landscape and soil physics pattern is not applied in the manuscript. When one of article write about spatial distribution it is not enough to make statistical analysis without maps. The authors have to show the environmental and soil backgrounds.
Summary
- Contrary to the title of the article, it examined the extent of salinization as a process, and the description did not include soil analysis. Therefore, the reference to soil profile and soil structure in the discussion and conclusion are not acceptable.
- The article is based only on the work of Chinese authors, except for 2 references, and therefore it would be better to publish it in a local scientific journal rather than an international journal. It has to be modified.
- The source of the data used cannot be traced: Landsat or GF-2/GF-6 images were analyzed.
- The method of verification is not described in the article; it is only mentioned as a result.
- Cohen's kappa values should have been presented for both the old and the new calculation method. The increase in area calculated for the New Method is thus not assessable.
My final opinion is that I am going to accept after throughout revision (corrections to minor methodological errors and text editing).
Author Response
请参阅附件。
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
Major comments:
The figure in this study should be further improved.
Minor comments:
1. Figure 4, In the legend, the “DEM” is modified to the “elevation”? Delete both “<VALUE>”? On the grid, remove the minutes and seconds of latitude and longitude?
2. Figure 6, On the grid, remove the minutes and seconds of latitude and longitude?
3. Figure 10, On the grid, remove the minutes and seconds of latitude and longitude?
4. Figure 12, On the grid, remove the minutes and seconds of latitude and longitude?
5. The new method proposed in this paper is based on the high-resolution series of satellites, so is it equally applicable to low-resolution satellites?
Comments for author File: Comments.pdf
Author Response
Please see the atachment.
Author Response File: Author Response.pdf
Reviewer 2 Report
Manuscript ID : remotesensing-2183496
Title : Research on High-precision Remote Sensing Extraction Method of Saline-alkaline Soil based on multi-sources data
This paper evaluated the use of multi-sources data and Decision tree classification to identify saline-alkaline soil. The topic was interesting, but the paper needs more work and clarification. The major concern is the absence of field sampling and fieldwork for the validation. Moreover, the fusion of multi-source data including various spatial resolutions is not clear.
The general comments
1. Title does not reflect work performed.
2. Line 15 : is it an Extraction or Identification?
3. Lines 15-16 : « multi-sources data » Please specify the data used in this work.
4. Decision tree model does not represent a new method.
5. What's the difference between the traditional spectral indices and the spectral indices named NDSI ? Please specify the remote sensing imagery used in this work.
6. what is the influence of the variation of the spatial resolution between the different multi-sources data on the results ?
7. Line 20 « that can distinguish saline-alkaline soil from other features » which ones?
8. Line 21 « to improve the classification accuracy ». What is the classification method? The supervised or unsupervised classification?
9. Line -23 « integrated expression of saline-alkaline soil from multisources data » The spectral indices ??
10. Line 27 « the traditional method » it's rather the traditional spectral indices ??
11. Line 28 « technological technique » it's not clear.
12. Please, spell out the novelty of the work in the last paragraph of the introduction.
13. « 2.1. Data » Describe the characteristics of the data and especially the spatial resolution. Please specify the date of the spectral data acquisition.
14. « orthorectification correction » how was the orthorectification done?
15. « FLAASH atmospheric correction module » added the input parameters of the model.
16. How were the surface reflectance spectra evaluated?
17. Please, Improve the resolution of the Figures.
18. Why did you choose the « Decision tree classification »
19. The process of multi-source data fusion (with various spatial resolutions) should be detailed. Author's can consult article's fusion method part in : https://doi.org/10.1080/19479830903561035
20. Line 125 « to select different types of sample elements on the image » why you didn't use field data with GPS Coordinates ?
21. Explain the choice of these different inputs? What is the relationship between these data and the Salinization or alkalinization of soils ? I will request the authors to apply requite statistics like ANOVA or Kryskal-Walis to understand changes in parameters in relation to soil characteristics. Author's can consult article's Statistical method part in https://doi.org/10.1007/s11356-022-21890-8
22. The uncertainty of the predicted models must be quantified in order to confirm these results. However, I suggest adding a Models uncertainty analysis. Pls, See the section Models uncertainty analysis: https://doi.org/10.3390/rs14164080
Regards, reviewer
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
this paper proposed a high-precision method for saline-alkaline soil extraction by 15 multi-sources data.
Remote sensing data and workload are sufficient, which conform to extraction of salinization information. It has certain value in extraction of remote sensing soil information and is suitable for publication. Words and sentences need further revision.
(1) What are the advantages of the four spectral indices proposed in this paper?
(2) Whether the paper highlights the proposed spectral index in comparison with the traditional spectral index.
(3) How to select the band information corresponding to the proposed spectral index?
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 4 Report
GENERAL COMMENTS
Introduction – clarity in the descriptions of some results from previous studies mentioned in the introduction are needed to be able to understand what they are in the context of the study being made.
1. Lines 51-52: “…while the accuracy of saline-alkaline soil information extraction was not proportional to the size of information in remote sensing data.” What do you mean by this. Clarity of this part of the sentence.
2. Lines 54-57: “With the help of hyperspectral data, Pang Guojin[17] et al. analyzed the relationship between soil salinity content and spectrum by constructing a model, which contributes to the quantitative remote sensing monitoring of saline-alkaline soil.” – what did this study show in terms of the relationship between salinity content and spectrum?
3. Lines 60-62: “Accurate distribution data are required for saline-alkaline soil management, so a high-performance inversion model is needed to obtain the distribution data.” – what is meant by accurate distribution data here? What is inversion model that is being mentioned?
4. Line 66: “Researches using indirect features…” – what is meant by indirect features here? Describe.
Data and methodology
5. Description of the images and data used.
5.1 What are GF-6/WFV images, GF-2/PMS images? When where they acquired? What resolution/s do they have? What are their spectral characteristics? What were they used for?
5.2 DEM – which DEM data and resolution
6. Vector boundary – what is this data
7. Line 89: “….the periods of the remote sensing images are selected from June to July” – which specific year/years?
8. Line 104: ”…to obtain 4-band fused image data with a spatial resolution of 1m for the study area” – do you mean resampled to 1 metre?
9. Figure 4 – write the units used for the slope and DEM in the legend.
10. Lines 124-125: “…we use GF-6 /WVF image data, combining GF-2 image and GoogleEarth high- resolution image data to select different types of sample elements on the image, and find 125 the best spectral index of the band combination of saline-alkaline soil.” – when you said combine, how did you do this. Clear description of what was done in this part.
11. Line 206: “8” - write as “eight”
12. Lines 206-208: “There are 8 main feature quantities commonly used for texture information extraction in remote sensing images: mean, variance, homogeneity, contrast, dissimilarity, entropy, angular second moment, and correlation.” – what do these different indicators show, and what will be their relevance in the extraction of saline-alkaline soil? For instance, the mean, what will it indicate and how will the greyscale image showing this can be interpreted? How about the variance? Homogeneity? Contrast? Dissimilarity? Entropy?? Angular Second moment? Correlation?
13. Line 209: “8” - write as “eight”
14. Line 210: “4” - write as “four”
15. Lines 215-217: “Height and slope information from DEM are introduced to carry out reclassification 215 in ArcGIS. As shown in Figure 9, there are some differences in elevation among features. 216 For example, vegetation and urban are generally flatter.” – How was this carried out? How did you choose the values for reclassification for elevation and slope for the respective classes used? Describe and explain the method for clarity.
16. After you have extracted the texture, spectral characteristics and elevation and slope information, how did you carried out, using your data the decision tree classification. The section in 2.3 is not enough description. The entire method there gives only an overview of the method.
17. How was this exactly applied with your data using the results from the spectral indices that you applied and computed for, the texture qualities using the different stats indicators that you mentioned and the elevation and slope information?
17.1 How did you produce the map in figure 10 based on this method?
18. How about the accuracy assessment. The same comment as the previous one. In 2.4, this gives an overview of the evaluation method in general. But how did you perform this in the case that you used? Some that you need to answer here are:
18.1 What ground truth data did you use and how did you derive them?
18.2 How many ground truth data were used?
18.3 How many samples did you get for each class?
Results and discussion
19. Make sure to describe your results properly and comprehensively
20. How will you describe the results and show trends here?
21. How will you characterise these areas?
22. How will you realte your results to geographic locations (as to where they are)
23. Figure 10 – based on my question earlier, if you will be reading the mtehods, it is difficult to relate how this map was produced.
24. Lines 236-237. You need to write here as well the total.
25. Lines 227-229: Some parts of this should be lifted in the description of how you performed the evaluation method under the Methods section.
26. Figure 12 and Lines 250-252 – which, among equations 1 to 3 pertains to the NDSI used for comparison?
27. Lines 250-252: “…we can see that the accuracy of the new salinity index NDSInew was improved by 8.7% compared with the traditional salinity index NDSIold , indicating the effectiveness of the new spectral index in the extraction of saline-alkaline soil.” – what is the basis of this statement that the new spectral index is better or more effective? Which part of the method section described the evaluation of this?
28. Section 4.3 (Analysis of saline -alkaline soil change) – how was this exactly performed? If presented in the results, a corresponding method section should be presented on how you did this, including the descriptions of the data that you used for this step.
29. Discussion – so far, what you presented are results of what you have done, but no discussion putting your methods and results into perspective of earlier or similar studies.
Others
30. write full names of acronyms used the first time used: ENVI, FLAASH
31. properly and even the contents of the tables.
32. References are incorrectly formatted- Check both in-text references and the reference list.
33. Choices of references – how will you put your research from an international context in terms of the references that you use? Important with diversity of perspective and methods when it comes to research.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
The authors did not answer all the questions. Analysis and measurement of model uncertainty have not been carried out, which influences the accuracy and precision of the obtained results.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 4 Report
Thank you for answering the previous comments. However, the reason why they are asked is that the answers to these that you provided should be evident in the text for clarity. Therefore, it is expected that your answers should be incorporated to the revision. Thus, I'll write them again. Discussion part is also missing.
GENERAL COMMENTS
Introduction – clarity in the descriptions of some results from previous studies mentioned in the introduction are needed to be able to understand what they are in the context of the study being made.
1. Lines 51-52: “…while the accuracy of saline-alkaline soil information extraction was not proportional to the size of information in remote sensing data.” What do you mean by this. Clarity of this part of the sentence.
2.
3. Lines 54-57: “With the help of hyperspectral data, Pang Guojin[17] et al. analyzed the relationship between soil salinity content and spectrum by constructing a model, which contributes to the quantitative remote sensing monitoring of saline-alkaline soil.” – what did this study show in terms of the relationship between salinity content and spectrum?
4. Lines 60-62: “Accurate distribution data are required for saline-alkaline soil management, so a high-performance inversion model is needed to obtain the distribution data.” – what is meant by accurate distribution data here? What is inversion model that is being mentioned?
5. Line 66: “Researches using indirect features…” – what is meant by indirect features here? Describe.
Data and methodology
6. Description of the images and data used.
5.1 What are GF-6/WFV images, GF-2/PMS images? When where they acquired? What resolution/s do they have? What are their spectral characteristics? What were they used for?
5.2 DEM – which DEM data and resolution
7. Vector boundary – what is this data
8. Line 89: “….the periods of the remote sensing images are selected from June to July” – which specific year/years?
9. Line 104: ”…to obtain 4-band fused image data with a spatial resolution of 1m for the study area” – do you mean resampled to 1 metre?
10. Figure 4 – write the units used for the slope and DEM in the legend.
11. Lines 124-125: “…we use GF-6 /WVF image data, combining GF-2 image and GoogleEarth high- resolution image data to select different types of sample elements on the image, and find 125 the best spectral index of the band combination of saline-alkaline soil.” – when you said combine, how did you do this. Clear description of what was done in this part.
12. Line 206: “8” - write as “eight”
13. Lines 206-208: “There are 8 main feature quantities commonly used for texture information extraction in remote sensing images: mean, variance, homogeneity, contrast, dissimilarity, entropy, angular second moment, and correlation.” – what do these different indicators show, and what will be their relevance in the extraction of saline-alkaline soil? For instance, the mean, what will it indicate and how will the greyscale image showing this can be interpreted? How about the variance? Homogeneity? Contrast? Dissimilarity? Entropy?? Angular Second moment? Correlation?
14. Lines 215-217: “Height and slope information from DEM are introduced to carry out reclassification 215 in ArcGIS. As shown in Figure 9, there are some differences in elevation among features. 216 For example, vegetation and urban are generally flatter.” – How was this carried out? How did you choose the values for reclassification for elevation and slope for the respective classes used? Describe and explain the method for clarity.
15. After you have extracted the texture, spectral characteristics and elevation and slope information, how did you carried out, using your data the decision tree classification. The section in 2.3 is not enough description. The entire method there gives only an overview of the method.
16. How was this exactly applied with your data using the results from the spectral indices that you applied and computed for, the texture qualities using the different stats indicators that you mentioned and the elevation and slope information?
17.1 How did you produce the map in figure 10 based on this method?
17. How about the accuracy assessment. The same comment as the previous one. In 2.4, this gives an overview of the evaluation method in general. But how did you perform this in the case that you used? Some that you need to answer here are:
18.1 What ground truth data did you use and how did you derive them?
18.2 How many ground truth data were used?
18.3 How many samples did you get for each class?
Results and discussion
18. Make sure to describe your results properly and comprehensively
19. How will you describe the results and show trends here?
20. How will you characterise these areas?
21. How will you realte your results to geographic locations (as to where they are)
22. Figure 10 – based on my question earlier, if you will be reading the mtehods, it is difficult to relate how this map was produced.
23. Lines 236-237. You need to write here as well the total.
24. Lines 227-229: Some parts of this should be lifted in the description of how you performed the evaluation method under the Methods section.
25. Figure 12 and Lines 250-252 – which, among equations 1 to 3 pertains to the NDSI used for comparison?
26. Lines 250-252: “…we can see that the accuracy of the new salinity index NDSInew was improved by 8.7% compared with the traditional salinity index NDSIold , indicating the effectiveness of the new spectral index in the extraction of saline-alkaline soil.” – what is the basis of this statement that the new spectral index is better or more effective? Which part of the method section described the evaluation of this?
27. Section 4.3 (Analysis of saline -alkaline soil change) – how was this exactly performed? If presented in the results, a corresponding method section should be presented on how you did this, including the descriptions of the data that you used for this step.
28. Discussion – so far, what you presented are results of what you have done, but no discussion putting your methods and results into perspective of earlier or similar studies.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
