Digital Mapping of Soil Organic Carbon Using Sentinel Series Data: A Case Study of the Ebinur Lake Watershed in Xinjiang

Round 1

Reviewer 1 Report

All comments are in the file.

Comments for author File: Comments.pdf

Author Response

Response to Reviewer 16 Comments

 

Xiaohang Li

College of Resources and Environment Science, Xinjiang University

No. 666, Shengli Road, Urumqi, Xinjiang Uygur Autonomous region, China

 

Feb. 3, 2021

Dear Editors and Reviewers:

Thank you for your letter and for the reviewers’ comments concerning our manuscript entitled “Digital mapping of soil organic carbon using Sentinel Series data: A case study of the Ebinur Lake watershed in Xinjiang” (Reference No: remotesensing-1089061). Those comments are all valuable and very helpful for revising and improving our paper, as well as the important guiding significance to our researches. We have studied comments carefully and have made correction which we hope meet with approval. Revised portion are marked in red in the paper. The main corrections in the paper and the responds to the reviewer’s comments are as flowing:

Point 1: L95: “In the past, the influence of environmental variables on soil properties has not been taken into account in soil mapping”

This is not true, see the many works from the references and Xiong’s et al. work

mentioned above.

 

Response 1: Thanks for your careful reading of our manuscript. It our mistake and we have revised this word according to your helpful comment.We changed this sentence to: The influence of environmental variables on soil properties is not considered in traditional soil mapping(Page 3, Line 68)(in red).

 

Point 2: L 191 “SOC is usually converted to SOM using the conversion coefficient of 1.724,where SOM ( g/kg ) = SOC (g/kg) × 1.724 [46]. Finally, soil organic carbon data were obtained”.

Where SOM has been used in the work? Why is this mentioned in the text?

 

Response 2: Thanks for your comment.This paper is only explaining the relationship between SOM data and SOC data, which is not related to the text, and we remove it considering the totality.

 

Point 3: Table 1. Attribute: Aspect, unit: degree.

The Aspect requires separate consideration. The fact is that the Aspect is a cyclic value. It varies from 0° to 360°, and both of these values correspond to the direction to the north. To avoid the ambiguity, it is recommended to use two local indexes: northernness (AN) and easternness (AE):

AN =cos(Aspect)

AE = sin(Aspect)

When calculating in SAGA sin and cos, the Aspect value must be expressed in radians. Northernness and easternness are dimensionless values. AN takes the values 1, -1, and 0 on the northern, southern, and eastern/western slopes, respectively. AE takes values of 1, -1, and 0 on the eastern, western, and northern/southern slopes, respectively (Florinsky I.V. Illustrated introduction to geomorphometry. // Electronic scientific publication “Almanac Space and Time” Vol. 11. Issue 1, 2016)

 

Response 3: We have revised it, thanks again(in table 1 in red).

 

Point 4: L219 Sentinel-2A data includes three different spatial resolution data 10m, 20m, 60m and contains 12 spectral bands (Table 3).

In table 3, only bands 1-8a are shown

 

Response 4:Thank you for your comments. We have revised in Table 3.

 

Point 5: L231 Table 2. Sentinel-1A data extraction information.

What terrain parameters were obtained from this data?

 

Response 5:We really appreciate your comment. Topographic parameters were extracted from the DEM data and were not extracted in the Sentinel-1A data.

 

Point 6:L251 respective characteristics of all the samples [64]

• is related to SVM

 

Response 6: We agree with your opinion. In [64] references in this paper cite the passages in the paper as：”(4) Random Forest: The random-forest-based approach is an embedded method of feature selection. The random forest consists of a collection of decision-tree classifiers  where each tree in the forest has been trained using a bootstrap sample of training data and a random subset of features sampled independently from the input features. A subset of the training data set is omitted from the training of each classifier .”. The quoted section is the random forest study section, which is consistent with the in-text section.

 

[64]Pal, M.; Foody Giles, M. Feature Selection for Classification of Hyperspectral Data by SVM. IEEE Transactions on Geoence & Remote Sensing 2010, 48, 2297–2307.

 

Point 7: L309 Table 4. Descriptive statistical analysis of SOC and environmental variables .

Units for S1÷S3 are not shown.

What are the parameters that characterize the vegetation?

Aspect maximum =6.283 means that units are in radians and not in degrees as shown in table 1.

 

Response 7: We really appreciate your comment. The concept S1÷S3 is not presented in this paper. The vegetation index is not used as a characteristic parameter in this paper, so the characteristic parameters of vegetation are not listed in Table 4.

We have modified the parameters of Aspect in table 4, thanks again(in red).

 

Point 8: L317 and Table N shows

What is N?

 

Response 8: It our mistake and we have revised the Table N has been changed to Table 5(Page 8, Line 266)(in red).

 

Point 9: L372 Figure 2.

It is necessary to divide Figure 2 into two parts - the top (2a) and bottom (2b) and refer to these parts in the text.

Flow Accumulation is not shown in table 1.

 

Response 9: We really appreciate your comment. Figure 2 has been revised to your comments and modified in the text and is shown in highlighted form(Page 11, Line 344-347)(in red). We have added Flow Accumulation to Table 1.

 

Point 10: 10.L381 The most important DEM derivatives data were SLOP

SLOPE

 

Response 10: Due to our oversight, this conclusion is wrong, and we have removed it from the text and shown it in highlighted form.

 

Point 11: L406 The resolution keeps increasing, the image information is gradually blurred and more blurred in a wide range, and the prediction effect keeps changing.

The resolution keeps decreasing.

The same goes for the map scale: 1:10000 is a more detailed scale than 1:1000000.

 

Response 11: Indeed,We have added a scale to Figure 3 to enable accurate identification of the data in the presence of changing image resolutions.

 

Point 12: L410 Figure 3.

It is necessary to apply the same scale  (the same range) to all images to  compare them, since  SAGA adjusts the colors automatically.

 

Response 12: Indeed,We have added a scale to Figure 3 to enable accurate identification of the data in the presence of changing image resolutions.

 

Point 13: L426 Based on this, combining different types of data and using them as a basis for predicting soil organic carbon is a new research direction.

This is a deep misconception of the authors, as was shown in the first part of the review.

 

Response 13: We appreciated your comments. In this paper, this sentence is changed to: On this basis, SAR data are combined with optical data to serve as a data base for soil organic carbon prediction, and the predictive power of different types of data for digital mapping of soil organic carbon is explored(Page 13, Line 376-379)(in red).

 

Point 14: L473 that topography and slope.

The topography includes the slope.

 

Response 14: We appreciated your comments. In this paper, replace this sentence with the following: S.A.Bangroo [90] proposed in his study that topography have a direct influence on the distribution and spatial variation of SOC(Page 14, Line 423)(in red).

 

Point 15: L522 (2) Among the predictions combined with environmental variable data, the prediction accuracy in model D and model E is the best at 300m R2=0.119 and 500m R2=0.112,respectively.

It is better to give the full description of the D  and E  models in the conclusion.

In Table 5: R2=0.400 (D) and 0.383 (E, RF).

 

Response 15: We agree with your opinion. We modify the conclusion in (2) as follows: Combining all environmental variables, the best model is model G. Model G is a combination of radar data, optical data and all environmental variables. In this model, the RF method has the best modeling effect at 10m, R²=0.406, MAE =0.162, REMS=5.947, LCCC=0.266. In model E that combines SAR data with environmental variables, the prediction effect of 300m is best to reach R²=0.383. In model F that combines spectral data (S-2, S-3) with environmental variables, the 100m prediction effect is best to reach R²=0.397(Page 15, Line 470-475)(in red).

 

Point 16:  L528 (4) The spatial distribution of SOC shows that the SOC content is higher in oases and areas with more human activities, and lower in mountainous areas and areas around lake bodies.

Human activities was not studied in the manuscript.

 

Response 16: We have revised it to: (4) The spatial distribution of SOC shows that the SOC content is higher in oases, and lower in mountainous areas and areas around lake bodies(Page 15, Line 479-480)(in red).

Author Response File: Author Response.docx

Reviewer 2 Report

The authors have presented the research work titled "Digital mapping of soil organic carbon using Sentinel Series data: A case study of the Ebinur Lake". The present study lacks novelty, however, its applicability in the region is new.

Here are my few comments as regards to improving the current status of the study:

1.The manuscript is poorly formatted. 

2. I strongly suggest the authors kindly use the Journals English formatting service

3. The citation format is not according to Remote Sensing

In line 70,  I suggest you include the research "John, K., Abraham Isong, I., Michael Kebonye, N., Okon Ayito, E., Chapman Agyeman, P. and Marcus Afu, S., 2020. Using Machine Learning Algorithms to Estimate Soil Organic Carbon Variability with Environmental Variables and Soil Nutrient Indicators in an Alluvial Soil. Land, 9(12), p.487."

in line 166- "The elevation difference in the basin is large, and the terrain of Ebinur Lake is the lowest." kindly state the elevation ranges

in line 174-176, describe the properties of the soils. How are they low/high in SOC and what management practices resulted into theiir current SOC contents.

In line 178, Figure 1 is not eligible

Line 180, the number of soil sample collected for this study was not mentioned

in 197, "ASTER GDEM data from Shuttle Radar Topographic Mission (
SRTM)" where was the data sourced from? ASTER or STRM?

Line 276-281 should be removed. It is too much repetition.  

Figure 2 is not eligible

in line 410, only the map of RF prediction was shown. How about the cubist model?

In conclusion, the study is poorly structured, the english translation is also poor. 

And judging by the standard of Remote Sensing Journal, I cannot recommend the work be published as it is.

The authors should properly format the work, source for English translation service.

 

 

Author Response

Response to Reviewer 11 Comments

 

Xiaohang Li

College of Resources and Environment Science, Xinjiang University

No. 666, Shengli Road, Urumqi, Xinjiang Uygur Autonomous region, China

 

Feb. 2, 2021

Dear Editors and Reviewers:

Thank you for your letter and for the reviewers’ comments concerning our manuscript entitled “Digital mapping of soil organic carbon using Sentinel Series data: A case study of the Ebinur Lake watershed in Xinjiang” (Reference No: remotesensing-1089061). Those comments are all valuable and very helpful for revising and improving our paper, as well as the important guiding significance to our researches. We have studied comments carefully and have made correction which we hope meet with approval. Revised portion are marked in red in the paper. The main corrections in the paper and the responds to the reviewer’s comments are as flowing:

Point 1:The manuscript is poorly formatted.

 

Response 1: Thank you for your comments. We have revised.

 

Point 2:I strongly suggest the authors kindly use the Journals English formatting service

 

Response 2: Thanks for your comment.We have revised.

 

Point 3: The citation format is not according to Remote Sensing

In line 70, I suggest you include the research "John, K., Abraham Isong, I., Michael Kebonye, N., Okon Ayito, E., Chapman Agyeman, P. and Marcus Afu, S., 2020. Using Machine Learning Algorithms to Estimate Soil Organic Carbon Variability with Environmental Variables and Soil Nutrient Indicators in an Alluvial Soil. Land, 9(12), p.487."

 

Response 3: Considering your valuable comment, we have revised(Page 3, Line 71)(in red).

 

Point 4: in line 166- "The elevation difference in the basin is large, and the terrain of Ebinur Lake is the lowest." kindly state the elevation ranges

 

Response 4: Thank you for your rigorous comment. We have revised(Page 3, Line 117)(in red).

 

Point 5: in line 174-176, describe the properties of the soils. How are they low/high in SOC and what management practices resulted into theiir current SOC contents.

 

Response 5: Your suggestion really means a lot to us. We add this sentence：With the change of climate and hydrothermal conditions, SOC has the highest correlation with saline soils, which increases with the increase of soil salinization. In contrast, SOC correlates less with gray-brown desert soils and wind-sanded soils in deserts. Since the study area is located in an arid zone, there is a clear correlation between the change of SOC and the management of soil salinization in the arid zone, thus making the SOC distribution in the area have a certain regional distribution status(Page 4, Line 126-132)(in red).

 

Point 6: In line 178, Figure 1 is not eligible

 

Response 6: Thank you for your comments. We have revised.

 

Point 7: Line 180, the number of soil sample collected for this study was not mentioned

 

Response 7: Thank you for your comments. We have revised(Page 4, Line 138)(in red).

 

Point 8: in 197, "ASTER GDEM data from Shuttle Radar Topographic Mission (

SRTM)" where was the data sourced from? ASTER or STRM?

 

Response 8: Thank you for your comments. It our mistake and we have revised this. The DEM data are from SRTM (ShuttleRadarTopographyMission), not ASTER GDEM.We have revised(Page 4, Line 151).

 

Point 9: Line 276-281 should be removed. It is too much repetition.  

 

Response 9: Thank you for your comments. We have revised(Page 7, Line 235-236)(in red).

 

Point 10: Figure 2 is not eligible

 

Response 10: Thank you for your comments. We have revised.

 

Point 11: in line 410, only the map of RF prediction was shown. How about the cubist model?

 

Response 11: Thanks for your careful reading of our manuscript. We mentioned in “3.4 Spatial prediction results of SOC” that the RF model has higher accuracy and can better predict the distribution of SOC, so we show the prediction plots of different resolutions in the RF model. Since the Cubist model has a lower prediction accuracy and cannot show the distribution of SOC and the detail map, we do not show the Cubist model prediction map.

 

 

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

I believe that the manuscript has not been significantly improved.

 

My main comments, which require reworking of the introduction and justification of the chosen experimental scheme, were not commented on.

In means the answer to the question “Does the introduction provide sufficient background and include all relevant references?” is negative.

The fundamental article  (Xiong et al., 2014 ) is not included in the introduction and is absent in references.

No response was received to my comments

1. Soils
2. Biota / Human
3. Relief (Topography).

The revision of the manuscript by the authors cannot be considered as major, but only as minor. The authors responded only to minor comments on errors in the text.

I think their responses are insufficient.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

The quality of the paper has generally been improved.

Good luck!

 

 

Author Response

Please see the attachment.

Author Response File: Author Response.docx