Extrapolation of Digital Soil Mapping Approaches for Soil Organic Carbon Stock Predictions in an Afromontane Environment

Round 1

Reviewer 1 Report

This research applied previously calibrated models for soil organic carbon prediction in other areas of the Maloti-Drakensberg region. In general, this work is meaningful to quantifying SOC stock in the northern Maloti-Drakensberg in South Africa and Lesotho which can reduce cost and manpower.  This manuscript has potential for publication, but a number of issues (including but not limited to the following comments) need to be addressed in advance.

Abstract: 

1. line 12: "trough" should be revised to "through"?

2. For lines 10-18, the background descriptions are too long, please cut down the length.

3. You should state the research implications at the end of the abstract. how the work contributes/adds knowledge to the topic; findings could help what kind of future research.

Introduction

4. References are required for "These carbon hotspots are being degraded, mostly due to mismanagement of livestock".

Materials and methods

5. You can move the full spelling of VDCN, TWI, MrVBF and NVDI from Table 2 to Table 1.

6. For common soil indicators e.g. BD and SOC, it is suggested that the description of the test method and calculation can be simplified, and relevant references should be supplemented.  Similar modifications can be made to NRMSE

Results & Discussion

7. Lines 454-455, what is the reason for this result or how to explain it? Some relevant research should be cited for the discussion. Throughout the article, it seems that there are only results, but no discussion content.

8. The abscissa in Figure 4 seems to be a mirror image, please modify it.

Conclusion

9. In the conclusion, it is not necessary to introduce the research background, suggesting to describe the main points/ findings, claim, and overview of future research possibilities, etc.

Author Response

First and foremost, we would like to thank you for taking the time to review our paper. We appreciate the input and hope that our revisions could make the paper better quality.

Abstract: 

1. line 12: "trough" should be revised to "through"?
   • Fixed it.
2. For lines 10-18, the background descriptions are too long, please cut down the length.
   • Reduced the background description, leave the important part.
3. You should state the research implications at the end of the abstract. how the work contributes/adds knowledge to the topic; findings could help what kind of future research.
   • Added findings and how it can contribute for future research

Introduction

4. References are required for "These carbon hotspots are being degraded, mostly due to mismanagement of livestock".
   • Added a reference.

Materials and methods

5. You can move the full spelling of VDCN, TWI, MrVBF and NVDI from Table 2 to Table 1.
   • Edited table 1 and 2
6. For common soil indicators e.g. BD and SOC, it is suggested that the description of the test method and calculation can be simplified, and relevant references should be supplemented.  Similar modifications can be made to NRMSE
   • Simplified and reduced the descriptions, also added some references.

Results & Discussion

7. Lines 454-455, what is the reason for this result or how to explain it? Some relevant research should be cited for the discussion. Throughout the article, it seems that there are only results, but no discussion content.
   • This was stated based on our own observation, however, added a reference to state that extrapolation relies on similarity of calibrated area and extrapolated area
   • Also found research that states one reason that the area ought not to differ (same levels of degradation)
   • Split the Results & Discussion section into two separate part and added more discussion
8. The abscissa in Figure 4 seems to be a mirror image, please modify it.
   • Unsure what is meant, however, redrew the figure with better resolution.
   • Just to note that the results from LASSO and UK are very similar and might look like mirror images at quick glance.

Conclusion

9. In the conclusion, it is not necessary to introduce the research background, suggesting to describe the main points/ findings, claim, and overview of future research possibilities, etc.
   • Removed the repetition of research background.

Reviewer 2 Report

Thanks for the invitation to review the paper “Extrapolation of digital soil mapping approaches for soil organic carbon stock predictions in an Afromontane environment”

The present paper is suitable for the journal

There are few suggestions from my side.

## Abstract is well written and it’s good but need to modify this section The selected models were extrapolated to four different headwater 21catchments, which included three near the Motete River (M1, M2 and M3) in Lesotho and one in the Vemvane catchment adjacent to the Tugela. Predictions were compared to measured stocks (n = 98). Results showed that the Universal Kriging, LASSO and Regression Kriging with cubist models, had the most satisfactory outcome, whereas the SoLIM models struggled to predict stocks accurately.

##  Need to update with latest reference The core method was used to determine soil bulk density. In short this method required a metal core with a fixed volume for all cores used, that had to be inserted into the  soil in order to extract an undisturbed sample. The sample had to be undisturbed to en sure that the soil particles do not rearrange as with the case of an auger extraction, where  compaction within the bucket can occur when the soil is drilled. The cores used in this  study all had a volume of 100 cm3. The samples were put into an oven to 105˚C for 48
hours and weighed after cooling down [32]. The bulk density was then calculated by Equation (1):

## Update the fig Figure 3. The relative covariate importance plot gained from the Random Forest (RF) model. The 348
relevance is assigned based on the percentage increase in MSE that would occur from removing a 349
certain covariate.

Rest things are quite good 

Author Response

1. Abstract is well written and it’s good but need to modify this section The selected models were extrapolated to four different headwater 21catchments, which included three near the Motete River (M1, M2 and M3) in Lesotho and one in the Vemvane catchment adjacent to the Tugela. Predictions were compared to measured stocks (n = 98). Results showed that the Universal Kriging, LASSO and Regression Kriging with cubist models, had the most satisfactory outcome, whereas the SoLIM models struggled to predict stocks accurately.
   1. Added clarification that n = 98 was soil samples collected
   2. Added the results of the models

2. Need to update with latest reference The core method was used to determine soil bulk density. In short this method required a metal core with a fixed volume for all cores used, that had to be inserted into the  soil in order to extract an undisturbed sample. The sample had to be undisturbed to ensure that the soil particles do not rearrange as with the case of an auger extraction, where  compaction within the bucket can occur when the soil is drilled. The cores used in this  study all had a volume of 100 cm3. The samples were put into an oven to 105˚C for 48 hours and weighed after cooling down. The bulk density was then calculated by Equation (1):
   1. Simplified the description and added another reference

3. Update the fig Figure 3. The relative covariate importance plot gained from the Random Forest (RF) model. The relevance is assigned based on the percentage increase in MSE that would occur from removing a certain covariate.
   1. Updated the figure description and resized the figure

Reviewer 3 Report

Dear authors,

After my detail evaluation, the paper is well organized and wrotten. However, it needs some revision. You shoul consider them to improve the paper. You can find my detailed revision suggession,

Best wishes,

Comments for author File: Comments.pdf

Author Response

First and foremost, we would like to thank you for taking the time to review our paper. We appreciate the input and hope that our revisions could make the paper better quality.

Abstract:

1. You should add model results, which model did achieve the best predicition performance?
   • Added the results
2. It is very long and confusing! You should simply rewrite the intro!
   • Reduced/rewrote the intro of the abstract
3. This is soil samples?
   • Added some clarification
4. Firstly you used full name
   • Changed it
5. Firstly you used full name
   • Changed it
6. The conclusion is very confusing!
   • Rewrote the conclusion of the abstract to provide clarification

Introduction:

7. Reference!
   • Added a reference

Materials and methods:

8. You should add some detailed information about soil type, moisture and temperature regimes.
   • Added soil types observed, as well as describing the saturated state of the soils in the valley bottom. The rest of the moisture regime is implied by the classification where the organic soils are saturated with water for long periods of time and the Humic soils are freely drained. Unfortunately, the temperature regimes are not part of the South African classification, however we added more information regarding the temperature of the study area.
9. It is not necessary in method section. You can use this discussion in the discussion part!
   • Moved it to discussion
10. it is not necessery in method section. You can use this discussion in the discussion part!
    • Moved it to discussion
11. it is not necessery in method section. You can use this discussion in the discussion part!
    • Moved it to discussion
12. it is not necessery in method section. You can use this discussion in the discussion part!
    • Moved it to discussion
13. which depth?
    • Provided clarification that sampling was done from the surface to the bedrock (due to nature of shallow profiles), resulting in Total SOC of each sampling site.
14. Why do you have many references? are they methods'refeneces ?
    • All these references used the “dry combustion method” for SOC, however, left the references that also used the LECO machine for their dry combustion.

Results:

15. The figure ise very resolution, you should redraw it!
    • Changed the resolution from 300dpi to 600dpi and increased the layout size

Round 2

Reviewer 1 Report

Please double check spelling, language, such as "and is stock (SOCs)SOC is crucial for" in line 11. wrong spellings?