Temporal and Spatial Characteristics of Soil Salinization and Its Impact on Cultivated Land Productivity in the BOHAI Rim Region

Round 1

Reviewer 1 Report (New Reviewer)

The manuscript presents a study of the spatial and temporal characteristics of the number of cultivated land, the soil salinization and the net primary productivity of vegetation in the Shandong region as well as their relationship. Considering that the salt-affected soils are paid more and more attention, I think that the manuscript present some useful findings and most of the research objectives have been achieved. Overall, the manuscript is worthy of publication to an extent. However, I have some concerns that should be addressed before the manuscript could be published.

 1. In the Introduction part, the authors introduced soil salinization, the importance of salt-affected soils in safeguarding the global food demand, the monitoring of the soil salinization, the NPP and their relation. It is better if the authors could mention the innovation of the manuscript in the part in a clearer way. And more references should be supplemented to prove that and the research on the relationship between salinization and farm-98 land productivity has not received due attention.

2. In the Materials and Methods, the authors should explain the reasons that the five phases in 2001, 2005, 2011, 135 2015 and 2019 were adopted to inverse soil salinization. And the sampling depths (Line 155) should be explained to make the readers understand the manuscript in a clearer way.

3. In the Discussion part, the authors discussed the changes of salinized cultivated land in Shandong province. It is better that the authors could give more details on the analysis methods used in this study compared with other studies. Moreover, the implications and innovation of the study should be highlighted in a clearer way.

5. And format problems (such as superscript and subscript) occurred in this manuscript. It is better if the authors could check them carefully. Moreover, it is better if the manuscript could be revised by a native English researcher.

I t is better if the manuscript could be revised by a native English researcher.

Author Response

We thank the reviewers for their valuable comments, which have greatly helped to improve the quality of the paper, and we have revised it according to your comments as follows:

1. In the introduction section of the paper, we have added the innovation points of the paper. In addition, to the extent of our knowledge, few studies have been reported on the relationship between salinization and productivity in cultivated land. Most studies have focused mostly on the analysis of productivity of cropland in non-salinized areas, while there is less literature related to the analysis of productivity of long time series in salinized areas;
2. The timing of the five soil salinity inversion stages was determined by combining the timing of the vegetation NPP dataset with the Landsat data conditions. The vegetation NPP data can be retrieved since 2001. In addition, the 2011 dataset was selected to replace the 2010 dataset because of the poor inversion due to the large amount of clouds covered in the study area in 2010. The sampling depth has been explained in the text;
3. The research methodology and innovations are described in section 4.4 of the thesis;
4. Formatting and language issues in the manuscript have been corrected.

Author Response File: Author Response.docx

Reviewer 2 Report (New Reviewer)

The study aimed to determine the characteristics and number of cultivated land and soils in saline areas and assess the net primary productivity (NPP) of vegetation in the Shandong region around the Bohai Sea. The relationship between salinity and productivity of agricultural land was determined.

The authors showed an overall downward trend in saline soils in the Bohai Sea region between 2001 and 2019. The trend depended on years of research, location and distance from the sea and power plants. The article was written correctly. It brings interesting information that can be used to develop areas with increased salinity. Appropriate research methods were used and properly described.

However, there is no research hypothesis in the article. It should be added.

The summary is too general. They should be more closely related to research results.

After minor corrections, the article can be published in the journal Water

Author Response

We thank the reviewers for their valuable comments, which have greatly helped to improve the quality of the paper, and we have revised it according to your comments as follows:

1. The author has added research hypotheses in appropriate positions of the article based on the reviewer's comments;
2. We have made necessary modifications to the abstract to make it more closely related to the research results.

Reviewer 3 Report (New Reviewer)

In this paper, the authors first perform mapping of soil salinity in a coastal region near Beijing, China using data from Landsat and field data that were correlated using the Random Forest algorithm. Then, the data of soil salinity were compared with the NPP data to detect trends and correlations. Some steps of the methodology are not clear and need to be improved. Please find my more concrete remarks below.

Major/moderate

* The novelty of the paper needs to come forward in the introduction. How does this research differ from past researches? What is new here? Please explicitly mention this.
* Section 2.3.3 and 2.3.4; it's not clear if soil salinity inversion was conducted at the sampling points and then 2D-IDW was applied to produce the maps from the text. Why didn't you use the remote sensing data to produce the maps, considering that you already have the RF model?
* Line 233: Why Spearman and not Pearson's?
* Equation (9): I am a bit confused here, didn't you measure salinity in the lab? Did you measure it only for a few samples and inferred the rest using Equation 9? Please elaborate and be more precise in the manuscript.
* Table 1 is missing the number of samples; if it's the same for all depths you can include it in the caption
* You need to elaborate on how you established the RF forest; did you use cross-validation to optimize the hyperparameters? What values did you set for them? Etc. Lines 305 and 306: RMSE has units.
* Figure 4: please include the 1:1 line. I am also confused since the max value was 55.98 (Table 1) but this is missing here?
* Table 3 caption needs to improve: these values are as calculated from the RF model + 2D-IDW, correct?
* Figure 8: I know this is difficult given the amount of data plotted, but as thing stand it's quite difficult to read it. Labels etc. need to have larger font size.
* Section 4.5: Would also examining a relationship between NPP and soil salinity + vegetation indices be another aveneue future researchers could explore? Just hypothesizing here.
* Did I miss the results of the Mann-Kendall significance test? Are they reported in the manuscript?


* Line 47: to => the and then on Line 49 and => and to
* Line 114: is high => has high elevation? similarly for low
* Lines 136 to 138: Which Landsat data? L1C? Did you perform L2 analysis yourselves? Why?
* Line 143: citation for PIE?
* Line 144: What kind of interpolation did you use?
* Line 158: is this standard ISO procedure?
* Figure's 2 caption should be improved, this is not a roadmap for the future
* Line 191: Is N (uppercase) different for n (lowercase) of equation 1?
* Line 206: were determined. I would have placed these spectral indices not inline but rather in a Table.
* Line 229 and 230: Mann-Kendall and Theil-Sen need a dash between the names
* Line 257: delete extra period mark  
* Line 271: fix fontsize
* Table 2: maybe include the positive and negative signs in the last column?
* Line 392: fishing?
* Line 540: RF is not deep learning

English language is fine, some minor spelling mistakes are noted.

Author Response

We thank the reviewers for their valuable comments, which have greatly helped to improve the quality of the paper, and we have revised it according to your comments as follows:

1. The main innovation of this article is to focus on the relationship between soil salt content and NPP in long-term cultivated land series. By understanding the changing characteristics between the two, we aim to provide new ideas for the future zoning control and governance of salinization. The relevant content has been supplemented and explained in the text;
2. This article first investigated the correlation between soil salinity data and remote sensing data (vegetation index and salinity index), and then established an RF model to invert soil salinity. Finally, the IDW method was used to visualize soil salinity mapping;
3. Since the soil salinity data in the study area does not show normal distribution, considering the requirements of Pearson correlation analysis, we choose Spearman correlation coefficient;
4. Yes, we first measured the salt content of 537 soil samples collected in the laboratory, and then established a relationship equation (Equation 9) between the soil salt content of these sample points and the average conductivity values of the four layers of 2.5, 7.5, 15, and 22.5cm, which saves the tedious laboratory analysis process for subsequent research. The subsequent soil salinity data used for analysis were all converted using Equation 9;
5. The sample size for all depths is the same, and the author has added it in the article;
6. The Random Forest method is implemented by code in Matlab software. This paper does not choose to use cross validation to optimize the relevant parameters of the model, but determines the final parameters of the Random Forest model through multiple experiments with R² and RMSE as the optimal conditions;
7. The figure already contains a 1:1 line. The soil salinity data used for inversion in the paper is the tillage layer (0-20cm), with a maximum value of 37.53g/kg;
8. Table 3 in the paper is the results calculated based on the RF model and IDW;
9. Figure 8 Label has been improved;
10. In order to highlight the main theme of the article, the author deleted the Theil Sen trend analysis and Mann Kendall significance test related content that describes the trend and significance test of cultivated land NPP change. The article Technology roadmap has been corrected;
11. Line 114. “is high” means high altitude;
12. Line 136 to 138. The Satellite imagery used to extract spectral index in the paper includes Landsat 8 OLI image and Landsat 5 TM image;
13. Line 143. The 'PIE' website has been referenced in the corresponding position in the text;
14. Line 144. The relevant content related to spatial interpolation in the article is all using inverse distance weight interpolation method;
15. The title of Figure 2 has been improved;
16. Line 191. The uppercase N in equation 1 is the same as the lowercase N and has been corrected in the original text;
17. Line 392. “Fishing” is a proprietary term for building grids in ArcGIS, and in order to facilitate readers' understanding, the original text has replaced it with "grid";
18. Line 540. The original text has been corrected that the RF model is a machine learning method, not a deep learning method.

Round 2

Reviewer 3 Report (New Reviewer)

The authors have followed most of the suggestions of the reviewers. Numbering below refers to the initial review.

2. This doesn't answer my remark though. Why not use the remote sensing maps to map the area with the established model? Why use IDW from just the point data?

3. Great, please include this in the manuscript.

6. Then this needs to be reported more rigorously. So you tested various parameters for number of trees and minimum number of leaves? How did you pick the optimal values? By checking RMSE in the train or the test set? If you selected based on the performance on the test set, then this is not considered optimal as one needs to have a validation set which is used for this purpose.

7. I am not sure this is a 1:1 line, it looks like a least squares line! If you follow the values, you will see that e.g. a sample observed as 33 is predicted as 22 and this falls on the 1:1 line!

8. Thank you, please include this in the manuscript.

 

 

Author Response

  Thank you for the reviewer's suggestions. All of your suggestions are very important and have significant guiding significance for our paper writing and research work. We have made the following modifications based on your valuable opinion:

2. Referring to previous studies^[1-2], the general idea of our modeling is to establish the relationship between different spectral indexes and soil salt content through remote sensing inversion, and then extract the spectral index inversion of the point in previous years to obtain its soil salt content data. Considering the different land cover and geomorphic types in the study area, directly using established remote sensing maps to map regional soil salinity may lead to amplification errors and reduced accuracy. It seems that most previous studies have also been conducted using the methods presented in the current paper. However, your opinion is very constructive and we will try it out in future research. Thank you very much!
3. We have added the correlation diagram between soil salt content and spectral index in the study area in the paper. See the text for details.
4. In Matlab software, we test different numbers of decision trees and leaf nodes to determine the relatively optimal parameter settings that meet the research needs based on the R² and RMSE of the training and validation sets. In relevant research^[3-4], they also obtained the relatively optimal parameters through this method.
5. We have added a 1:1 line to the graph, as detailed in the paper.

 

  We sincerely hope that this revised manuscript has addressed all your comments and suggestions. Thank you again for the enthusiastic work and valuable suggestions of the reviewer！

 

 

Reference

• Li, Y.S., Chang, C.Y., Wang, Z.R., Zhao, G.X., 2022. Remote sensing prediction and characteristic analysis of cultivated land salinization in different seasons and multiple soil layers in the coastal area, Int. J. Appl. Earth Obs. 111, 102838.
• Wang, L.P., Wang, X., Wang, D.Y., Qi, B.S., Zheng, S.F., Liu, H., Liu, H.J., Luo, C., Li, H.X., Meng, L.H., Meng, X.T., Wang, Y.H., 2021. Spatiotemporal changes and driving factors of cultivated soil organic carbon in northern China’s typical agro-pastoral ecotone in the last 30 years. Remote sens. 13(18), 3607.
• Zhao, W.J., Zhou, C., Zhou, C.Q., Ma, H., Wang, Z.J., 2022. Soil salinity inversion model oasis in Arid Area based on UAV Multispectral remote sensing, Remote Sensing, 2022, 14(8), 1804.
• Cui, X., Han, W.T., Zhang, H.H., Cui, J.W., Ma, W.T., Zhang, L.Y., Li, G., 2022. Estimating soil salinity under sunflower cover in the Hetao irrigation district based on unmanned aerial vehicle remote sensing, Land Degrad. Dev. 34(1), 84-97.

Author Response File: Author Response.docx

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

The authors present a detailed report on the distribution of soil salinity and average NPP in a coastal region in five non-consecutive years (2001, 2005, 2011, 2015 and 2019). The salt status of soils was determined by field sampling and laboratory analysis, and NPP values were derived from NASA data. After mapping both sets of data, statistical comparisons were made between the level of soil salinization and NPP (as an indicator of land productivity) on cultivated land.

 

Comments (general and specific)

The paper is unnecessarily long – try to condense points and avoid repetition. Phrases like ‘It can be seen from Figure xyz…’ can be better expressed by ‘Figure xyz shows…’.

Aims need to be clarified – objective (1), lines 100-101, is broadly descriptive and should be removed.

NPP is not uniform across different crop types. What are the crops whose NPP is being assessed in the cultivated land areas?

Tracing temporal changes in specific variables requires that the time-tested variables remain constant over the period of analysis. Did the crop types that were cultivated remain the same over the period? Did annual rainfall differ between years and affect NPP for rainfed crops? Was there a contribution from irrigated cropping?

Definition of the categories used to classify degrees of soil salinization should appear at the commencement of this section in the results.

Line 239 – Spearman Rank correlation coefficient: There is no need to include sentences outlining the details of this test – it is a standard non-parametric test.

The Results section is long and contains detailed descriptions which would be suitable for a regional technical report but need to be condensed for an international research publication. It would be sensible to highlight just two or three key points under each of the ‘Results’ sub-headings, omitting detailed descriptions relating to nominated Counties/Districts.

The paragraph from about line 471 is explanation/interpretation and therefore should appear in the Discussion, not the Results section.

Lines 538-545 – this sentence is too long and should be broken into shorter sentences or dot points.

Line 559 – ‘offshore’ – this is probably a language misinterpretation: ‘offshore’ in English refers to usually near-coastal areas currently beneath the ocean. If the areas you are referring to have been reclaimed from the ocean (i.e., barriers built to prevent sea incursion over previously submerged land), then an explanation would be helpful. Conversely, if sea level rise/storm surges are encroaching on previously cultivated land then this could be explained.

Lines 595-598 – this seems to be repetition.

Table 2 needs to be re-formatted – it is unclear. Why is the ‘2001’ notation appearing on the LHS? The unit of measurement (km2) must be indicated for the table. A proportion is not a ratio – the heading for column 10 should be ‘Change (% of total area)’

Figure 5 – all map legends must be in English for an English language publication. The legend must also indicate what is being represented in each of the boxes (a) through to (g).

Figure captions must be informative – e.g., Figure 7: ‘Change trend and significance test of cultivated land NPP in Shandong area around Bohai Sea from 2001 to 2019. (a) Theil-Sun change trend; (b) ‘Mann-Kendall significance test’.

Language expression issues – what is ‘dynamic monitoring of soil salinization’ in line 106?

There is frequent use of the term ‘inversed’ (e.g., lines 214-215 – ‘…then the soil salt content information ….was inversed.’). Please establish exactly what you mean here and in other places – should this be ‘inferred’? or ‘applied’ or ‘estimated’? Once the intended meaning for the context has been found, then replace all references to ‘inverse’ with the appropriate word/s. (note: ‘inverse’ in English means ‘opposite’, ‘reverse’).

Lines 105-109 form a single sentence which is unclear and needs to be re-written. It is also not apparent that this objective is achieved.

Use ‘ha’ for hectare (line 31) and tonnes (t) in preference to billions of kg (line 538)

Reviewer 2 Report

Modern phenomena causing the problem of sustainable development are, in addition to reducing the area of agricultural land, salinization in some areas. The paper is focused on monitoring the temporal and spatial characteristics of soil salinization by the surrounding environment. The authors observed the reduction of the area of farmed land over 20 years. It follows from the available methodological part that the area of interest was quite extensive. The authors described in detail the study area, targeted soils, used methods of evaluation of the achieved results. Statistical methods in the SPSS program were used for statistical and graphical evaluation of the results, as well as inverse interpolation in Arcgis 10.7. The achieved results were evaluated in detail with not only the results of coefficients of variation, average values, trend analysis and the like, but also supported by graphic displays. The results show that there were changes in the areas of cultivated and salted land during the monitoring years. The aim of similar studies should be data for reducing the acreage of built-up land and increasing the acreage with the possibilities of salinity regulation.

 

Strengths side : In my opinion, the article fulfills the character of a scientific article in a scientific journal (current content). Introduction, methodology, results and discussion are processed at the required level.

 

Weaknesses side: - none

 

Other comments: - values in table 1 – coefficient of variation is not in %, in % the result is e.g. 188%

It is at the editor's discretion whether it would be more suitable for the magazine, e.g. Sustainability, since the primary area does not concern water, but soil salinity.

Reviewer 3 Report

This paper is a spatio-temporal analysis of the saline-alkali properties of cultivated soil in Bohai Bay area of Shandong Province. Personally, I think this may be a good research report, but there is still a big gap between it being regarded as a research paper. This paper does not show innovation. It is just a large amount of data processing. The selected methods are also used by most scholars, and the research topic is not very novel and lacks of new ideas. Therefore, in my opinion, this paper is not enough to be published in this journal, and my opinion is to reject the paper. The main comments are as follows:

1. The article involves a lot of content, and the description of the research results of each content is also very long, so I cannot accurately grasp the topic of the article;

2. I can't understand why the author tested the salt content of different layers of soil, especially the maximum distance between layers was only 7.5cm, so it was difficult to completely separate each layer of soil in the sampling process, leading to great uncertainty in the final result.

3. How to understand the surface layer? What is its thickness? Is it the surface of the soil?

4. Why does the author still use the conductance instrument? The author has collected so many soil samples, and the distribution of the samples is very uniform in space. It is completely possible to directly use the relationship between the salt content of soil samples and the spectral index to achieve the inversion. Why do we need an additional conductivity in the middle?

5. The author's explanation of the research results is too tedious and detailed, and many figures are written, which is unnecessary and will distract the attention of readers.