Land Use and Land Cover Change Modulates Hydrological Flows and Water Supply to Gaborone Dam Catchment, Botswana

Round 1

Reviewer 1 Report

The paper deals with applying satellite imagery to analyze the river flows in Botswana. The article has scientific and practical contributions to solving the problem when gauged stations and the unavailability of measured results occur. Literature insight is at a reasonable level, and the structure of the methodology and paper is well-defined. 

I am proposing a major revision. Here are my suggestions.

-time series modeling methods should be applied to the obtained time series. Such should be provided to see discordances, regulations, and similar. The final acquired time series should be tested in this way.

-authors should elaborate on why they have applied particular methods in their research. 

-some pictures are blurry, and it is hard to see anything from them. 

Authors should turn on the spell checker. Many of the mistakes will be avoided. 

Author Response

Response to Reviewer 1 Comments

 

Point 1 time series modeling methods should be applied to the obtained time series. Such should be provided to see discordances, regulations, and similar. The final acquired time series should be tested in this way.

 

Response

We accept the suggestion and we would like thank the reviewer for this.  We have applied the time series modeling methods to see how SWAT compares with data-based Time series modeling (TS). The changes as a result of inclusion of TS are now available under Section 2.3.2 on lines 393-408, under Section 3.2 on lines 516-532, Table 6, Figure 6a and section 5 on line 872-873.

 

 

Point 2: authors should elaborate on why they have applied particular methods in their research. 

 

Response

The comment is accepted and we have now elaborated why we have chosen specific methods with supporting evidences from literature. The explaination are given under methodology section  on line 134 to 144 and 516-528.

 

Point 3: some pictures are blurry, and it is hard to see anything from them. 

The comment is well taken and we improved the clarty of Figure 4, Figure 5, Figure 7 and  Figure 10 .

 

Just for information purpose, we have made a number rewording based on similarity report from mdpi publisher. However, we would like to indicate that similarity of words and phrases of technical nature can not be considered as similarity for concern. The similarity report that we have received includes just isolated words or phrases which may account for about 90% of similarity. Nevertheless we have not ignored all of them and reword wherever possible. There is also a similarity on acknowledgement of funding with our own previous published paper. We can not alter this part of the manuscript as it is part of the funding agreement with the funders and should remain unchanged.

 

Finally, we appreciate the comments of the reviewer as we strongly believe the comments have improved the clarity and readability of the manuscript.  We also appreciate the editor for giving us the chance to review and improve the manuscript.

 

Author Response File: Author Response.docx

Reviewer 2 Report

The study presents an evaluation of the effect of changes in land cover on streamflow and state variables using the SWAT model. The use of random forest to classify the land cover classes might be of interest to the readers as well as the application of the SWAT model to a semi-arid basin. Here are some main comments:

Introduction:  I would suggest adding a paragraph about the use of random forest to analyse satellite images.

Aim: Please see suggestion in the attached annotated manuscript.

Methods: An improvement might be to compare the results for soil moisture, evapotranspiration and groundwater to local or global datasets if available. 

Discussion: It would be useful to add a section on the level of significance of the hydrological alteration index. It seems that some of the changes are not significant.

Attached is also a version of the manuscript with additional specific comments and suggestions.

 

Comments for author File: Comments.pdf

The use of terminology can be improved.

Author Response

Response to Reviewer 2 Comments

 

Point 1 Introduction:  I would suggest adding a paragraph about the use of random forest to analyse satellite images.

 

Response

We thank the reviewer for the above suggestion.  We have now included the use of Random Forest (RF) classifier in LULC classification of satellite images on line 119-149, 176-178 of the revised manuscript.

 

 

Point 2: from the annotated manuscript

• Please define LULC
• What is blue and green water. Please could you explain?
• I would suggest rewriting to emphasise the effect of land use changes on streamflow and other variables.
• Please give more details
• I would suggest to replace with hyperparameters
• How was this number determined?
• What is the objective function?
• I would suggest adding this section to the methods
• Is there a cutoff value such as in p-values?
• I don't think that this sentence in needed. Please delete.

Response

• The comment is accepted and the abstract includes LULC definition on line 12 of the revised maniscript.
• blue and green water are explained on line 79-80 of the revised manuscript.
• The effect of land use change is now rewritten to emphasis the impact on stream flow and other variables on line 119-149.
• More details on how the boundaries of the catchment is delinated now given on line 175-177 .
• The word “essential input” is repaced by the word “hyperparameters” in the revised maniscript on line 176-178.
• The reviewer comment on how the 120 tree was determined? is now explained in the revised maniscript on lines 238-239. In brief, the number of trees are determined by increasing the number of trees and checking whether training and validation accuracies have improved with optimal computational time.
• The reviewer’s inquiry regarding the objctive function, it is a simple cost function that minimizes the sum of square of difference between observed stream flow and model (SWAT) stream flow. This can be found on lins 344-354.  
• The text in the older version of the maniscript on line 402-404 is now moved to methodology section on line 315-320.
• Is there a cutoff value such as in p-values?
• Land use change impact on stream flow is highly significant when the p-values is near to 0 and the impact of LULC change is little between the pre- and post-impact periods when p-values is near to 1. The cutoff value depends on the choice of significance level which is commonly referred to as 75%, 80%, 90%, 95%, 99%. In this study, only 90-day minimum flow is significantly altered at 82%. The amendment made is given on lines 753-759 .
• The comment is acepted and change is effected accordingly.

 

Point 3: Methods: An improvement might be to compare the results for soil moisture, evapotranspiration and groundwater to local or global datasets if available. 

Response

We appreciate the reviewer recommendation to include additional data sets to assess SWAT model. Although the local and global data availability and accuracy of available data are challenges, we decided to implement the suggestion using reanalysis global data which contain assimilated observations. We used evapotransipiration (ET) from GLDAS-2.1 and calculated soil moisture (SW) from soil moisture inflitration rate obtained from MERRA-2 reanalysis model. Figure 6 is included and corresponding description of these data and results are found on page 4-5 lines 176-192 and page 14-15 on lines 537-570 of the revised manuscript.

 

Point 4: Discussion  It would be useful to add a section on the level of significance of the hydrological alteration index. It seems that some of the changes are not significant.

 

Response

As indicated under point 2 above from commentes in the annotated manuscript, there are various significant levels used in the literature depending on the required accuracies. We have included some texts reflecting this idea on lines 753-759, 845-846 .

 

Just for information purpose, we have made a number rewording based on similarity report from mdpi publisher. However, we would like to indicate that similarity of words and phrases of technical nature can not be considered as similarity for concern. The similarity report that we have received includes just isolated words or phrases which may account for about 90% of similarity. Nevertheless we have not ignored all of them and reword wherever possible. There is also a similarity on acknowledgement of funding with our own previous published paper. We can not alter this part of the manuscript as it is part of the funding agreement with the funders and should remain unchanged.

 

Finally, we appreciate the comments of the reviewer as we strongly believe the comments have improved the claity and readability of the manuscript. We also appreciate the editor for giving us the chance to review and improve the manuscript.

 

Round 2

Reviewer 1 Report

This time, the authors improved the manuscript in accordance with my comments. 

Reviewer 2 Report

The authors have now included a lot more information on the methods. The introduction and discussion sections are also improved in this new version. 

I am not sure if I can comment but I think that it just requires minor/moderate edits.