Anthropization and Climate Change: Impact on the Discharges of Forest Watersheds in Central Africa

Round 1

Reviewer 1 Report

Please, see the enclosed file.

Comments for author File: Comments.pdf

Author Response

Manuscript ID: water-908490

Type: Article

Title: Anthropisation and Climate Change: Impact on the Discharges of Forest Watersheds in Central Africa

Authors: Valentin Brice Ebodé, Gil Mahé, Jean Guy Dzana and Joseph Armathé Amougou

 

 

Reviewers of the article water-908490 made comments (major and minor) and suggestions to improve it significantly.

On this basis, several major corrections have been made to this new version of the manuscript.

The abstract has been reorganized, as recommended by reviewer 1. A sentence introducing the thematic has been added (Lines 13-15), as well as a short paragraph at the end of the conclusion which explains the possible future implications of the findings ( lines 34-35).

The references proposed by reviewer 1 (Yira et al., 2017; Namugize et al., 2018; Gorgoglione et al., 2020; Getahun et al., 2020) have been added to justify the statement "Most of the studies ... interferences "(Lines 44-45).

The aim of the work has been better explained in the introduction, as recommended by reviewer 1 (Lines 87-89).

The definition of the terms of the correlation coefficient calculation formula has been made, as recommended by reviewer 1 (Lines 240-241).

The word "interannual" has been replaced throughout the manuscript with "whole period" and "long-term", as recommended by reviewer 2. The words "variation" and "evolution" have also been replaced with "change ". The misuse of the word "trend" has been corrected. For example, the expression "identical trends were noted" has been replaced by "the same sign of anomalies was observed" (Line 465).

The results have been modified as recommended by reviewer 2, assuming that “variability” refers to the coefficient of variation (Cv) and that the percentages of variation shown are absolute changes in Cv.

Reviewer 2 wanted to know which definition of the runoff coefficient was used in this study because having observed a significant change in this variable from the 2000s. The definition of the correlation coefficient taken into account in this study is the same as his (ratio between river discharges and rainfall, when they are expressed in the same units). Their significant increase noted in this study in the case of Nyong is due to the fact that the rainfall has decreased considerably over the past two decades, compared to previous decades. 

Reviewer 2 proposed to develop linear models based on interannual variations in rainfall and discharges, which could be used to predict (i) Spring discharges as a function of both Winter and Spring rainfall and (ii) Autumn discharges as a function of Summer and Autumn precipitation. We only developed a model based on the interannual variations of Summer rainfall (July and August) and the flows of the first month of Autumn (September), since it is only between these variables that the correlations obtained are good. For the rest, we have specified the values of the correlation coefficients obtained in the text, but we have not developed a model, taking into account that a model developed based on the data poorly correlated at the base cannot be used to make effective predictions. The complexity of the rainfall-runoff relationships in the region, due to evapotranspiration, changes in MOS, etc. could explain the poor linear relationship observed in some cases.     

 

Several minor corrections were also made to the manuscript as suggested by reviewer 2.

The countries in which the watersheds are located have been specified (Lines 97-99).

The abbreviation ORSTOM has been spelt out (Lines 142-143).

The intervals of the rainfall correction coefficients have been specified (Lines 187-188).

Additional explanations relating to the graphics in Figures 3 and 4 have been added (Lines 335-337 and 340-342).

The use of bold font for some values ​​in Table 5 has been explained (Line 480).

The period analyzed by Liénou et al (2008) has been specified (Line 393).

The section relating to the impact of changes in land-use patterns on runoff (3.3.2) has been slightly modified taking into account that no physical link was demonstrated between the two in the study.  More careful wording has been done in some places in the text (Lines 595-635).

Corrections have been made to the language errors contained in the previous version of the manuscript as suggested by reviewer 2. These corrections are indicated in the current version, with the comment "Proposal for the second reviewer". 

Reviewer 2 Report

GENERAL COMMENTS

The manuscripts analyses changes in precipitation and river discharge in two watersheds (Ntem and Nyong) in Cameroon in Central Africa from the 1950s to the 2010s, together with land use changes inferred from Landsat satellite images taken in 1973 and 2015. The major findings include (i) a significant decrease in annual discharges since the 1970s, despite no significant simultaneous decrease in annual rainfall, (ii) an apparent connection between the multidecadal variations of the wet seasons’ (spring and autumn) discharge with the preceding dry seasons’ (winter and summer) rainfall, and (iii) a strong increase in built impervious area at the expense of forest cover and water bodies, which might explain some of the discharge changes that are apparently not explained by variations in precipitation.

A strength of this study is the simultaneous consideration of precipitation and land use changes as drivers of river discharge change. However, an important limitation is that the analysis on the causes of the discharge changes is entirely qualitative, and the conclusions remain therefore speculative even if plausible. In addition, the statistical terminology that is used in describing the observed variations and changes is frequently unclear and unconventional, and the reading of the manuscript therefore requires some guesswork to understand what precisely is meant. The English language is reasonably good in the beginning and the end of the manuscript, but the middle parts including, in particular, the results section are plagued by a larger number of grammatical errors and cumbersome sentences.

For these reasons, major revisions seem necessary before eventual publication.

                                         

MAJOR COMMENTS

1. Section 3.3.1 makes several conclusions on how the rainfall changes in the preceding winter and summer seasons affect the discharge changes in spring and autumn. These claims would be more convincing if supported by regression models built on the interannual variations of precipitation and discharge. With several decades of data are available, it should be straightforward to build linear regression models that predict (i) spring discharge as a function of both the winter and spring rainfall and (ii) autumn discharge as a function of both the summer and autumn rainfall. With such models, the effects of the same-season and previous-season rainfall changes on the discharge changes could be objectively quantified. Furthermore, the residual time series from these regression models would have less interannual variability than the original discharge time series, making it easier to discern that part of the discharge change that cannot be explained by rainfall change.

2. Several statistical terms are used in an unclear, unusual or even erroneous way, which makes is substantially more difficult to follow the manuscript. In particular,

• Table 1 and elsewhere: the word “interannual” is used to describe the long-term means of precipitation and discharge. Please change this to “whole period” or “long-term” and explain in the text that this covers the whole period of observations. “Interannual” is misleading because it usually refers to variation from year to year.
• L297-298 and elsewhere: Please make it clear that “variability” refers to the coefficient of variation (CV) and the quoted per cent changes are absolute changes in CV. “Variability” might as well refer to the standard deviation, and the per cent change might be a relative rather than absolute difference between two periods.
• Figure 3 and elsewhere. Please explain your definition of the “runoff coefficient”. I thought it was just the ratio between river discharge and rainfall (when expressed in the same units), but the very large changes in the runoff coefficient (in Fig. 3, more than a factor of two increase from the dry period to the wet period in Nyong) suggest that some other definition has been used.
• Right column in Table 4 and elsewhere. Please replace “Variation” with “Change” or “Difference”. The word “variation” is misleading because it usually refers to shorter-term variability.
• L430 (“identical trends were noted”) and elsewhere. This is misuse of word “trend” which generally refers to systematic changes with time. Please replace by “the same sign of anomalies was observed”
• Right column in Table 7. Please replace “Evolution” with “Change”.

 

Please make these changes throughout the manuscript, not only in the specific places mentioned above.

 

MINOR COMMENTS / SUBSTANCE AND PRESENTATION, EXCLUDING PURELY LINGUISTIC COMMENTS

1. Please mention briefly the countries where the two watersheds are located.
2. What does “the latter” refer to? The Nyong and Ntem watersheds?
3. Which country? Cameroon?
4. Spell out ORSTOM.
5. L177-179. This short period of overlap for deriving the correction factors might be a major uncertainty in precipitation estimates after the year 2000. How large were the correction factors?
6. 1 to t (not i-1 to t)?
7. when is sign(x) 0?
8. L228 and 231. Why “observations” and “estimates” (and which is which)? The definition of correlation is generic and does not depend on how the time series were obtained.
9. “neo-channels are created”: what exactly do you mean by this?
10. Is the “internnual average” just the mean value for the whole period for which observations are available (cf. Major comment 2)?
11. Its coefficient of variation decreases by 2% (cf. Major comment 2).
12. L319, caption of Figure 3. Please also explain the horizontal black and grey mean value lines and the baseline against which the wet phase and dry phase mean values were determined.
13. L333-334. These changes are negligibly small (which seems fortuitous, given the substantial interannual variability). The only thing that can be said is that the average maximums have remained essentially constant.
14. 24 days earlier or 291-275 = 16 days earlier?
15. Table 4. The headings “maximum flows” and “minimum flows” have been switched.
16. L367-368. Please mention the period(s) analysed by Lienou et al., since differences in period may be important for the apparent differences in the results.
17. “never” = never earlier during the period covered by observations, i.e. since the 1950s?
18. Figure 4 suggests -9.7% rather than -8.2%
19. More precisely: … sources of long-term change and interdecadal variability
20. Table 5. Explain the use of the bold font. Does it indicate statistical significance of correlation?
21. Figure 5. What is the unit for precipitation anomalies? Are these normalized (anomaly in standard deviations) values?
22. years 1973 and 2015 for both Nyong and Ntem
23. Section 3.3.2. The reasoning in this section makes sense but the quantitative importance of land use changes remains highly speculative. As far as I see, the only way in which you could confirm your reasoning would be to run a hydrological model with and without land use changes. As far as this is not possible, more cautious wording would be appropriate in some places.
24. L648-651. Although this may well be correct, the wording would need some moderation because of the lack of quantitative analysis on the importance of land-use changes.

MINOR COMMENTS / LANGUAGE

25. It is hoped that the results of this study will
26. The annual precipitation oscillates
27. L103-104. On both sides of the main rivers, a topographic landscape in steps is observed which …
28. L104-106. In the middle part of each watershed, there is a major valley ( ), which …
29. during the 1990s and 2000s.
30. station rainfall
31. This choice
32. Indicators … were used
33. omit “different ranges”
34. L239-240. We can thus analyse not only the sign of change between the two periods but also the magnitude of this difference
35. L276-277. After validating the land cover maps produced, the statistical and spatial differences of each class between the two dates were studied ...
36. L295-296. a statistically significant decrease was noted in spring in both two watersheds
37. decline in them during the 2000s
38. L314-315. … discharges, although an increase in them is noted in the 2010s, after the decrease started in the 1970s.
39. their averages over the whole period
40. between the changes recorded
41. The minimums have occurred slightly later
42. L361-362. a slight decrease in the magnitude of these changes can be noted with increasing flow duration.
43. the decrease is greater for shorter-term minimums
44. L363-364. The coefficient of variation has increased slightly
45. L365-366. The minimums have appeared on the average 45 days earlier after the discontinuity
46. The change in annual rainfall (cf. Major comment 2)?
47. The 1950s and 1960s were wet on Nyong, then followed by …
48. L387-388. … opposite trends (Figure 4), with an increase in summer but a decrease in winter.
49. have led to the examination à motivate us to examine
50. This analysis
51. Opposite signs of NATL and the rainfall anomaly
52. years for which the same sign of anomalies was observed
53. the same sign of anomalies between winter rainfall indices
54. L437-438. Correlation between winter rainfall and SOI is good in both watersheds, but ..
55. years with the same sign of anomalies in winter rainfall and SOI
56. winter rainfall is correlated
57. indicates
58. L446-447. the negative correlation coefficient between
59. Frequency of years having the same/opposite signs of ocean indices and rainfall anomalies.
60. comparison of climatic indices and normalized rainfall anomalies
61. during the 1970s
62. rainfall does not decrease or decreases very slightly
63. This is all the more probable since
64. Omit ”noted”
65. rainfall does not decrease significantly, even though since 2005
66. Differences between these watersheds?
67. A few studies
68. linked to the size of the catchments, but also their large forest cover
69. until it reached 50% of its total area
70. L632-633. the proximity of the land-use changes to the outlets of the watersheds

Author Response

Manuscript ID: water-908490

Type: Article

Title: Anthropisation and Climate Change: Impact on the Discharges of Forest Watersheds in Central Africa

Authors: Valentin Brice Ebodé, Gil Mahé, Jean Guy Dzana and Joseph Armathé Amougou

 

 

Reviewers of the article water-908490 made comments (major and minor) and suggestions to improve it significantly.

On this basis, several major corrections have been made to this new version of the manuscript.

The abstract has been reorganized, as recommended by reviewer 1. A sentence introducing the thematic has been added (Lines 13-15), as well as a short paragraph at the end of the conclusion which explains the possible future implications of the findings ( lines 34-35).

The references proposed by reviewer 1 (Yira et al., 2017; Namugize et al., 2018; Gorgoglione et al., 2020; Getahun et al., 2020) have been added to justify the statement "Most of the studies ... interferences "(Lines 44-45).

The aim of the work has been better explained in the introduction, as recommended by reviewer 1 (Lines 87-89).

The definition of the terms of the correlation coefficient calculation formula has been made, as recommended by reviewer 1 (Lines 240-241).

The word "interannual" has been replaced throughout the manuscript with "whole period" and "long-term", as recommended by reviewer 2. The words "variation" and "evolution" have also been replaced with "change ". The misuse of the word "trend" has been corrected. For example, the expression "identical trends were noted" has been replaced by "the same sign of anomalies was observed" (Line 465).

The results have been modified as recommended by reviewer 2, assuming that “variability” refers to the coefficient of variation (Cv) and that the percentages of variation shown are absolute changes in Cv.

Reviewer 2 wanted to know which definition of the runoff coefficient was used in this study because having observed a significant change in this variable from the 2000s. The definition of the correlation coefficient taken into account in this study is the same as his (ratio between river discharges and rainfall, when they are expressed in the same units). Their significant increase noted in this study in the case of Nyong is due to the fact that the rainfall has decreased considerably over the past two decades, compared to previous decades. 

Reviewer 2 proposed to develop linear models based on interannual variations in rainfall and discharges, which could be used to predict (i) Spring discharges as a function of both Winter and Spring rainfall and (ii) Autumn discharges as a function of Summer and Autumn precipitation. We only developed a model based on the interannual variations of Summer rainfall (July and August) and the flows of the first month of Autumn (September), since it is only between these variables that the correlations obtained are good. For the rest, we have specified the values of the correlation coefficients obtained in the text, but we have not developed a model, taking into account that a model developed based on the data poorly correlated at the base cannot be used to make effective predictions. The complexity of the rainfall-runoff relationships in the region, due to evapotranspiration, changes in MOS, etc. could explain the poor linear relationship observed in some cases.     

 

Several minor corrections were also made to the manuscript as suggested by reviewer 2.

The countries in which the watersheds are located have been specified (Lines 97-99).

The abbreviation ORSTOM has been spelt out (Lines 142-143).

The intervals of the rainfall correction coefficients have been specified (Lines 187-188).

Additional explanations relating to the graphics in Figures 3 and 4 have been added (Lines 335-337 and 340-342).

The use of a bold font for some values ​​in Table 5 has been explained (Line 480).

The period analyzed by Liénou et al (2008) has been specified (Line 393).

The section relating to the impact of changes in land-use patterns on runoff (3.3.2) has been slightly modified taking into account that no physical link was demonstrated between the two in the study.  More careful wording has been done in some places in the text (Lines 595-635).

Corrections have been made to the language errors contained in the previous version of the manuscript as suggested by reviewer 2. These corrections are indicated in the current version, with the comment "Proposal for the second reviewer". 

Round 2

Reviewer 1 Report

I am glad to see that the authors followed most of my suggestions. The manuscript was substantially improved.

I only recommend the authors to consider to re-organize the Conclusions. This section has to stand-alone. Therefore, the authors should re-state their thesis at the beginning and summarize their main points of evidence for the readers.

After this revision, in my opinion, the manuscript is ready to be published.

Author Response

After reviewing it twice, the reviewers of article water-908490 made a few minor comments to improve it.

Reviewer 1 asked to re-state the thesis at the beginning of the conclusion, before summarizing the main points of evidence for the readers. This was done in the text, between lines 666-668.

Reviewer 2 proposed to replace certain terms in the text for greater clarity. These corrections are indicated in the current version, with the comment "Proposal for the second reviewer".

Figure 3 has been corrected as recommended by reviewer 2. The calculation error it displayed has been revised.

The reviewer 2 had a concern relating to the values ​​of the runoff coefficients in Table 3. The calculations were verified as he suggested. The misunderstanding would come from the fact that it seems that the reviewer has divided the deviations of discharges by those of rainfall, which explains the difference between our results and his. In reality, the runoff coefficient we used is the ratio between the depth of runoff (in mm) and precipitation (in mm). And yet, to calculate depth of runoff, we divide volumes of runoff by the surface of the catchment area over a specific period (annual, seasonal, etc.). The difference of formula explains the difference of results. 

Reviewer 2 Report

GENERAL COMMENTS

I appreciate the improvements that the authors have made to this manuscript based on the comments of myself and another reviewer. Overall, the manuscript is in clearly better shape than the first version, including but not limited to the clarification of the terminology. However, I still have a few mostly minor comments. The most important concerns the runoff coefficients which still seem to have pecularities (comments 7 and 8).                                            

DETAILED COMMENTS  

1. discharges of two watersheads of Central Africa (Nyong and Ntem)
2. They can
3. It is particularly useful in the case of Ntem?
4. observed changes? The study is about the past, not future.
5. Should this not be “time series (Xi) 1 to t and t + 1 to N” (cf. the equation on L231)
6. sign (x) = 1 if x > 0, 0 if x = 0 and -1 if x < 0?
7. Figure 3c. I am confused by the printed numerical values. As far as I can estimate, the mean runoff coefficients in Nyong are ~49% for the whole period, ~39% for the dry phase and ~61% for the wet phase. Thus, the mean before the discontinuity would be 20% below, and that after the discontinuity 24% above the long-term mean. This is much less than the values (-35.7% and +42.6%) printed in the figure.
8. Table 3. I still cannot make sense of the deviations in the runoff coefficients. Take the annual values for Nyong in 1960-69 as an example. Precipitation was much less above the whole period mean (+2.2%) than discharge (+17.7%). Simple algebra, neglecting interannual variability, would suggest a deviation of (117.7/102.2-1)*100% = +15.2% in the runoff coefficient, but the reported value is -34.2%. Many other values seem equally peculiar. Please check your calculations or, if they are correct, provide an explanation for these apparent discrepancies.
9. L509-510. The sentence “This is visible during the 2010s …” seems to contradict the two previous sentences. They argue for a connection between winter rainfall and spring runoff coefficients and discharges, but then this sentence indicates that a change in winter rainfall was not reflected in the spring runoff coefficients and discharges. I cannot follow the reasoning – what do you mean by “This is visible”?
10. “Between the 1970s and 1990s, the flows and runoff coefficients increased …”. Confusing, as there is no data for the discharge and runoff coefficients in the 1990s in Table 3 (due to lack of observations).

Author Response

After reviewing it twice, the reviewers of article water-908490 made a few minor comments to improve it.

Reviewer 1 asked to re-state the thesis at the beginning of the conclusion, before summarizing the main points of evidence for the readers. This was done in the text, between lines 666-668.

Reviewer 2 proposed to replace certain terms in the text for greater clarity. These corrections are indicated in the current version, with the comment "Proposal for the second reviewer".

Figure 3 has been corrected as recommended by reviewer 2. The calculation error it displayed has been revised.

The reviewer 2 had a concern relating to the values ​​of the runoff coefficients in Table 3. The calculations were verified as he suggested. The misunderstanding would come from the fact that it seems that the reviewer has divided the deviations of discharges by those of rainfall, which explains the difference between our results and his. In reality, the runoff coefficient we used is the ratio between the depth of runoff (in mm) and precipitation (in mm). And yet, to calculate depth of runoff, we divide volumes of runoff by the surface of the catchment area over a specific period (annual, seasonal, etc.). The difference of formula explains the difference of results. 

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 paper focuses on the seasonal changes of the outflow of 2 rivers in Central Africa. It is interesting as there are few works from this region. However, the paper is not methodologically compiled. Authors use incomparable time data. They evaluate the landscape structure in 2019, but in one river basin they compare it with 1973 and in the other with 2000. Urbanization and population in one river basin is assessed from 2013. The state of the population in 2019 and its impact on the landscape is certainly different. The evaluation of hydrological characteristics is after decades. However, it is incorrect to use a value that is missing 7 out of 10 years (1993-1999). Similarly, crop production ends with the situation from 2008 (the paper evaluates the situation in 2019). After this period, there was a crisis, which certainly affected the agriculture, but it was not reflected in the contribution. It is necessary to add current and comparable data to the article.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

At first I would like to thank Authors for hard work and interesting article. The presented paper deals very important issues as is the impact of climate change and raising humans activities to discharges from forest catchments. Below is a few comments, which I believe can improve a little bit of work.
1. In my opinion, the Introduction is too general. The Authors can complete this part of the article e.g. by paragraph when the world’s trends of changing precipitation are presented. For example please see to work: Młyński D., Wałęga A., Cebulska M. Trends, Variability, and Seasonality of Maximum Annual Daily Precipitation in the Upper Vistula Basin, Poland. Atmosphere, 9(8), 313.
2. Without a doubt, the undertaken studies are very important in hydrology. But at the end of the Introduction, the novelty or Author’s contribution should be clearly emphasized.
3. At the beginning of materials of methods should provide what type of hydroclimatic data were used in work. I think that better term than ‘hydroclimatic’ is ‘hydrometeorological’.
4. In lines 60-62 the Authors stated that for Nyong the data comes from two sources. Were in both sources use the same methodology to flow measure? In this very important in point of homogeneity data view.
5. The same question for rainfall data. What was the resolution for rainfall data? Daily or others?
6. What kind of methods were used to average rainfall data for particular catchments?
7. What about for verification hydrometeorological data for independences and homogeneity? Did the Authors compare to conduct such an analysis? Taking into opportunity the changes in hydrometeorological phenomena, the trends in time series should be verified. For example, the Mann-Kendall or modified Mann-Kendall test can be implicated.
8. Also, the figure and description of land use in analyzed catchments should be added. The land use plays an important role in runoff shaping.
9. Generally, the Materials and Methods chapter must be orderly. First, the Authors should provide information about catchments: location, physiographic and meteorological factors. Next, the used data must be provided. In my opinion, the current description is too large. There is not necessary to provide such detail information. Just provide the sources, kind of data, period. At the end of the chapter the Authors should precise to describe how the assumed work aim was realized (what kind of methods were used). Actually there is a lack of it.
10. In my opinion, the Results chapter should be started by initial analysis for data. Initial analysis should include the descriptive statistics calculations: min, mean, max, standard deviation and coefficient of variability. The Authors should focus mostly on the last one which describes the changing dynamics for rainfall data. The calculations should be conducted for annual and seasonal data.
11. Line 217: the Authors stated that Petitt test was used. In the Methodology, they did not describe the test. They must provide what is the kind of test? For what significant level the calculations were made? Why they selected this test and not seasonal (modified) Mann-Kendall which is recommended by WMO for trend detecting in hydrometeorological data.
12. Lines 228-247: I would like to see the Authors comments on what is the reason in trend changing in rainfall data.
13. In my opinion, the Results chapter has significant shortcomings. Namely, the aim of work is the impact of human activities for discharges in analyzed catchments. The part of results which describe the discharges is very poor. The Authors show the only table with basic factors. I would like to see the hydrographs for catchments, characteristic flows in the analyzed period, etc. But the most important is trend analysis. Actually the analysis is connected with rainfall data. This causes a slight confusion in article reading.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 3 Report

Please, see the enclosed file.

Comments for author File: Comments.pdf

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The authors positively modified the publication, but my comments were not supplemented.
Input data are incomparable:
„The demographic data used in this work are those of the population censuses of the years 1973, 1987, 2005 and 2018 for Cameroon; 1983, 2001 and 2012 for Equatorial Guinea; 1993 and 2013 for Gabon. These data have been collected from local institutions (Central Bureau of Censuses and Population Studies in Cameroon, Census of Population and Housing in Equatorial Guinea and General Census of Population and Housing in Gabon). Agricultural statistics are from agricultural campaigns carried out in Cameroon by the Ministry of Agriculture and Rural Development in 2000 and 2008. „
For example, population numbers at country level for each year are recorded at the United Nations.
It is not possible to talk about the state of agriculture in 2018 from data from 2008.
It is not possible to compare river basins with significantly different input data (line 295: images). Nevertheless, the situation in 1973 is different from the situation in 2000. Based on other initial data, the result is clearly incompatible and incomparable.
I do not recommend publishing.

Reviewer 2 Report

I have read clearly the revised manuscript. Almost all my concerns were solved. But I still do not see the novelty of work.

Reviewer 3 Report

I am glad to see that the Authors followed some of my suggestions to improve their manuscript. However, I should reject the paper for two main reasons:

• As the Authors mentioned in lines 142-143: "In such a study, the ideal would be to have data covering the same period approximatively, but these data cover different periods." This aspect threatens the entire methodology on which this work is based on. It is not scientifically correct comparing land use, hydrology, the population of different years. This means that the results of this work do not scientific soundness.
• The Authors did not follow my suggestions about the references that were out of date. I believe a scientific paper does not have only local references since it has to be universally accepted. Therefore, the introduction (and sometimes the methodology used too) has to refer to the last scientific publications that tackled similar problems with the aim of highlighting the scientific gap that this work is filling. This part is missing in this work.