Next Article in Journal
A High-Precision Matching Method for Heterogeneous SAR Images Based on ROEWA and Angle-Weighted Gradient
Previous Article in Journal
Multi-Temporal Remote Sensing for Forest Conservation and Management: A Case Study of the Gran Chaco in Central Argentina
 
 
Article
Peer-Review Record

Landcover Change Amidst Climate Change in the Lake Tana Basin (Ethiopia): Insights from 37 Years of Earth Observation on Landcover–Rainfall Interactions

Remote Sens. 2025, 17(5), 747; https://doi.org/10.3390/rs17050747
by Sullivan Tsay Fofang 1, Erasto Benedict Mukama 1,2, Anwar Assefa Adem 3,4 and Stefaan Dondeyne 1,5,6,7,*
Reviewer 1: Anonymous
Reviewer 2:
Reviewer 3: Anonymous
Remote Sens. 2025, 17(5), 747; https://doi.org/10.3390/rs17050747
Submission received: 4 November 2024 / Revised: 24 January 2025 / Accepted: 5 February 2025 / Published: 21 February 2025
(This article belongs to the Section Biogeosciences Remote Sensing)

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Temporal changes in landcover and rainfall over nearly 40 years were analyzed using the existing datasets of the Global Land Cover Fine Classification System and the Climate Hazards Group InfraRed Precipitation with Station in this manuscript. The interaction between landcover and rainfall on soil erosion was evaluated based on RUSLE. It is valuable for analyzing long-term series data. However, I think the parameter values of the RUSLE model are not very reasonable and need further improvement. If the author can solve this problem, this article can be published. There are some issues that are to be further improved as follows:  

 

1.     L24: CHIRPS-v2, full name?

2.     L113-115:How to evaluate using these indicators of overall accuracy, producer accuracy, user accuracy, and Kappa coefficient needs to be explained in detail, and what are their evaluation criteria?

3.     L121: I don’t understand what these three levels refer to. Please explain further.

4.     L142: Check syntax.

5.     L157-159: This study assumes that the C value of the same land use type is a constant and has nothing to do with vegetation coverage. Although many studies hold this view, its rationality is questionable.

6.     Eq.(4): It is acceptable for the topography factor to remain unchanged over different periods, but it is difficult to accept unchanged soil erodibility due to the change of land use over nearly four decades. The authors should rationalize the expression of soil erodibility factors.

7.     Fig. 4b: It is necessary to explain what these three pictures represent.

8.     L277-298 and L358-396: I do not quite agree with this part of the analysis, because some factors are not assigned reasonable values in the model calculation.

9.     Figs A3 and A5: they are not clear.

10.  This manuscript does not have Figure A4, please check it.

Author Response

[Comments and Suggestions for Authors]

Temporal changes in landcover and rainfall over nearly 40 years were analyzed using the existing datasets of the Global Land Cover Fine Classification System and the Climate Hazards Group InfraRed Precipitation with Station in this manuscript. The interaction between landcover and rainfall on soil erosion was evaluated based on RUSLE. It is valuable for analyzing long-term series data.

However, I think the parameter values of the RUSLE model are not very reasonable and need further improvement. If the author can solve this problem, this article can be published. There are some issues that are to be further improved as follows:  [...]

[Response:]
Thank you for your appreciation of our work. 
We understand the reservation you formulated regarding our assumptions.  We have given some more arguments to justify our assumption and added a discussion section on the limitations of using RUSLE in our study.  More specific explanations are also given here responding to your specific queries. 

[Comment 1]
L24: CHIRPS-v2, full name?
[Response] Full name added in the abstract

[Comment 2]

L113-115:How to evaluate using these indicators of overall accuracy, producer accuracy, user accuracy, and Kappa coefficient needs to be explained in detail, and what are their evaluation criteria?

[Response]

We added a concise explanation in the text as:

Overall accuracy measures the percentage of correctly classified points out of the total reference points. Producer accuracy evaluates how well actual landcover types are identified, while user accuracy reflects the reliability of the classification. The Kappa coefficient accounts for chance agreement, with values closer to 1 indicating strong agreement and values below 0.40 suggesting poor performance. Together, these metrics provide a comprehensive assessment of classification accuracy.

But here a more extensive explanation:

To clarify the evaluation process, overall accuracy is calculated by dividing the total number of correctly classified points by the total number of reference points. This value provides a general measure of how well the classification matches the reference data. Producer accuracy, which reflects how well a given landcover type is correctly identified, is computed by dividing the number of correctly classified points for a specific class by the total number of actual reference points for that class. This indicates the probability of a reference pixel being correctly classified. Conversely, user accuracy measures the reliability of the classification by dividing the number of correctly classified points for a class by the total number of points classified as that class, indicating the likelihood that a classified pixel actually represents that class on the ground.

The Kappa coefficient (κ) quantifies the agreement between the classified landcover and the reference data while accounting for agreement occurring by chance. It ranges from -1 to 1, where values closer to 1 indicate strong agreement, values around 0 suggest no better than random classification, and negative values imply disagreement. Typically, Kappa values above 0.80 are considered excellent, 0.60-0.80 as good, 0.40-0.60 as moderate, and below 0.40 as poor (Congalton, 2007). These metrics together provide a comprehensive assessment of classification performance, allowing us to identify both strengths and weaknesses in the landcover data.

[Comment 3]

L121: I don’t understand what these three levels refer to. Please explain further.

[Response]

Thank you for this remark; indeed that was not clear. 
We added the following explanation. The three levels of the Intensity Analysis – interval, category, and transition – provide a structured framework for understanding landcover change dynamics:

  • Interval Level – This evaluates the intensity of overall landcover change across different time intervals (e.g., 1985–1995, 1995–2005). It compares the observed change to a uniform distribution, identifying whether changes were faster or slower during specific periods.
  • Category Level – This level focuses on the gain and loss of individual landcover categories, revealing which classes are experiencing disproportionate changes. It highlights whether certain landcover types are more dynamic compared to others.
  • Transition Level – This examines how landcover transitions occur between categories, assessing the intensity of change from one class to another. It identifies dominant transitions (e.g., forest to agriculture) and pinpoints where changes are concentrated.

By analyzing these three levels, the method uncovers patterns of change that might be missed through simpler assessments, offering a detailed understanding of landcover dynamics over time.

[Comment 4]

L142: Check syntax.

[Response] Thanks, well spotted. We corrected the mistake.

[Comment 5]
 L157-159: This study assumes that the C value of the same land use type is constant and has nothing to do with vegetation coverage. Although many studies hold this view, its rationality is questionable.

[Response]
This observation is indeed valid. The C-factor is original intend for particular crop types and using this for a landcover type implies that it can vary significantly depending on the specific crop (e.g., maize versus teff), the variety of the same crop, and the landscape context in which it is grown. For example, it has been shown that the median C-factor for maize (Zea mays) is 0.7 on the Makonde plateau, while it is 0.2 on the inland plains of South Eastern Tanzania (Kabanza et al., 2013); similarly, it has been reported that the C-factor for teff (Eragrostis tef) is 0.07 in Northern Ethiopia, while it is 0.25 in Central Ethiopia (Nyssen et al., 2009). However, the landcover data available for the Lake Tana Basin does not distinguish between different crops, and there is no data on how the C-factor may vary over time or across different landscape settings within the basin.

Given these limitations, a practical approach was adopted where the C-factors are treated as average values representative of the landcover units within the Lake Tana Basin, based on reported values from the literature for the region. While there is variation within a landcover type (e.g., different crops within cropland), the differences between the major landcover types typically differ by an order of magnitude or more (e.g. cropland 0.15 vs shrubland 0.014, tree-cover 0.001). Therefore, although the C-factor for a landcover type, can vary per crop and vegetation type, it seems reasonable to assume to be lower than the variation between the landcover types.  Thus, we would argue that this simplification is reasonable for the purpose of studying the impacts of landcover changes at the basin scale and over a time span of nearly 40 years.

We have explained this now in the discussion section "4.4 Limitations of using the RUSLE".  As the reviewer rightly points out; this approach is often taken, but rarely (if ever justified).  We have done now so, highlighting that while there are finer-scale variations in C-factors, the approach used here remains representative of the broader landcover change effects that we aim to study.

[Comment 6]

Eq. (4): It is acceptable for the topography factor to remain unchanged over different periods, but it is difficult to accept unchanged soil erodibility due to the change of land use over nearly four decades. The authors should rationalize the expression of soil erodibility factors.

[Response]

This observation is also correct. The erodibility of soils, particularly of heavily eroded areas, will have changed. However, as explained, we are focusing on the interaction of the temporal variations in landcover (C) and rainfall erosivity (R); we, therefore, consider the factors K, LS and P to remain constant. In the discussion section, we also reflect on this assumption.

[Comment 7]

Figure 4b: It is necessary to explain what these three pictures represent.

[Response]

Thanks for the remarks. We added more details to the caption to clarify what the individual maps represent.

[Comment 8]

L277-298 and L358-396: I do not quite agree with this part of the analysis, because some factors are not assigned reasonable values in the model calculation.

[Response]

The reviewer expressed concerns regarding our assumptions about the C-factors and the constancy of the K-factor in the model calculations.

Regarding the C-factor, we believe we have provided a clear rationale, emphasizing that while variability exists within landcover types, the differences we considered between the major landcover types are orders of magnitude. The C-factors we apply are intended to represent average values for each landcover type, reflecting broader patterns rather than specific, fine-scale variations.

Concerning the K-factor, we acknowledge that it is likely to have changed in the field over time. However, the focus of our analysis is on the interaction between landcover changes and rainfall patterns. The “potential soil erosion” serves as an indicator to explore this interaction, rather than as a prediction of actual soil erosion outcomes. Our goal is to assess how landcover dynamics and rainfall erosivity jointly influence erosion potential, rather than to simulate precise erosion rates.

To make this clear, we have made minor language adjustments to emphasize the interaction between landcover and rainfall, shifting the focus away from soil erosion itself. We hope these clarifications address the reviewer’s concerns.


[Comment 9]

Figs A3 and A5: they are not clear.

[Response]

We are afraid that the Figures may not be sharp on the ms you obtained; but that’s a software issue of MDPI. In our version are they are sharp and of high resolution.

[Comment 10]
This manuscript does not have Figure A4, please check it.

[Response]

Thank you, well spotted; it is now corrected by adjusting the numbering from A5 to A4

 

 

 

Reviewer 2 Report

Comments and Suggestions for Authors

Dear Authors,

The manuscript ‘Landcover change amidst climate change and implications for soil erosion: Insights from 37 years of Earth Observation data for the Lake Tana Basin (Ethiopia)’ focuses on 37 years of land cover change using earth observation data.  

This research study was conducted in an important research area, and the authors used remote sensing data and CHIRPS-v2 data to assess land cover and rainfall trends. The title and abstract of this manuscript reflect its content. The introduction presents the purpose of the research investigation. However, the introduction can be improved by adding more relevant experiments conducted in Ethiopia. In addition, novelty of this study can be highlighted. The materials and method section and results are well explained.  The figures and graphs are clear and well-explained. Discussion can be improved by providing the limitations of this study. In the conclusion, major findings should be included. 

Author Response

Comments and Suggestions for Authors

Dear Authors,

The manuscript ‘Landcover change amidst climate change and implications for soil erosion: Insights from 37 years of Earth Observation data for the Lake Tana Basin (Ethiopia)’ focuses on 37 years of land cover change using earth observation data.  

This research study was conducted in an important research area, and the authors used remote sensing data and CHIRPS-v2 data to assess land cover and rainfall trends. The title and abstract of this manuscript reflect its content. The introduction presents the purpose of the research investigation. However, the introduction can be improved by adding more relevant experiments conducted in Ethiopia. In addition, novelty of this study can be highlighted. The materials and method section and results are well explained.  The figures and graphs are clear and well-explained. Discussion can be improved by providing the limitations of this study. In the conclusion, major findings should be included. 

[Response]

Thank you for this valuable and constructive comment.
In the introduction, we have elaborated a paragraph giving more information on previous, and recent research on the topic, particularly on the Lake Tana Basin. Against that backdrop, we then also highlight the novelty of our research which are essentially (i) that we emphasis analysing the observation data, including the interactive effect of LULC change x Rainfall, rather than modelling the effect; (ii) that we analyse a wider timespan than most other studies.  These aspects are now better explained in the discussion section.

Reviewer 3 Report

Comments and Suggestions for Authors

This manuscript reported a case study on the soil erosion change driven by land cover change and rainfall change. Frankly, the study have noting new in academy or technology. However, it provide interesting information on the environment change in the study area. The methodology is sound and clear. However, I have some concerns on the manuscript, mainly the writing work. 

1. The introduction needs to be improved by review impact of land cover change and rainfall change on soil erosion, as well related methods. The research gap should be clearly described in the introduction.

2. objective “(3) How has the interaction between changes in landcover and rainfall affected potential soil erosion 79 in the Lake Tana Basin?” , replace the "interaction" with "combination". The interaction between land cover change and rainfall was not explored in this paper, but the impact of rainfall change and land cover change on soil erosion.

3. Merge 3.1 and 3.2. The new title may be "Land cover change" 

4. Only the trend of soil erosion was given. The contribution of rainfall and land cover change to soil erosion should be analyzed further.

5. In 4.1 and 4.2 , remarks on results repeat the results section. Discussion section should focus on the divers of land cover change, rainfall change, and the regime of impacts on soil erosion.

6. Language quality need to be improved. Though there are no significant grammar errors, but the dividing of paragraph and some expressions should be modified. I suggest a language polish before accept.

Author Response

[Comment 1]
This manuscript reported a case study on the soil erosion change driven by land cover change and rainfall change. Frankly, the study have nothing new in academy or technology. However, it provides interesting information on the environment change in the study area. The methodology is sound and clear. However, I have some concerns on the manuscript, mainly the writing work.

[Response]

Thank you for your valuable feedback. We appreciate your recognition of the interesting information provided by the study and the soundness of the methodology. We would like to clarify that the primary focus of the manuscript is not primarily on soil erosion, but on examining the combined and interactive effects of land-use/land-cover changes and rainfall variations. This integrated perspective aims to highlight the interplay between these two factors, which we believe adds meaningful insight to the field.

To address this, we have revised the title and adjusted relevant sections of the manuscript to better emphasize this focus. This point was also noted by Reviewer 1, and we have taken care to ensure the manuscript reflects this emphasis more clearly.

[Comment 2]
The introduction needs to be improved by review impact of land cover change and rainfall change on soil erosion, as well related methods. The research gap should be clearly described in the introduction

[Response]

This concern overlaps with the major concern raised by reviewer-2. Therefore, to address, this we have added in the introduction more information on past research carried in the Lake Tana Basin, and so clarified the novelty of our research.

[Comment 3]

Objective “(3) How has the interaction between changes in landcover and rainfall affected potential soil erosion 79 in the Lake Tana Basin?”, replace the "interaction" with "combination". The interaction between land cover change and rainfall was not explored in this paper, but the impact of rainfall change and land cover change on soil erosion.

[Response]

We respectfully disagree with the suggestion to replace "interaction" with "combination." Our analysis indeed focuses on the interaction between landcover change and changes in rainfall patterns, specifically by examining the "R × C" term (Eq. 5 and Eq. 7).

Mathematically, the multiplicative nature of the RUSLE model inherently accounts for the interactions between the various factors, including rainfall and landcover. This interaction is essential for understanding how the combined effects of these changes influence potential soil erosion, which is the core focus of our study.

[Comment 4]
Merge 3.1 and 3.2. The new title may be "Land cover change" .

[Response]
Thank you for the excellent suggestion. We implemented this.

[Comment 5]
Only the trend of soil erosion was given. The contribution of rainfall and land cover change to soil erosion should be analyzed further.

[Response]

We are sorry, but we may not have understood this comment. We don’t agree that “only the trend in soil erosion is given”.

As part of the results, in section 3.1 we report on landcover changes, in section 3.2 on changes in rainfall pattern; in section 3.3. on changes in rainfall erosivity, and in section 3.4 on the effect of the interaction of R × C, albeit framed as "potential soil erosion".

Based on these results we then discuss drivers and implications of the LULC changes (section 4.1), the implications of changes in rainfall (section 4.2), the effect of the interactions between changes in LULC and rainfall (section 4.3), and the limitations of using the RUSLE (a new section 4.4).

[Comment 6]
In 4.1 and 4.2 , remarks on results repeat the results section. Discussion section should focus on the divers of land cover change, rainfall change, and the regime of impacts on soil erosion.

[Response]

Thanks for the observation, that was indeed the case.  We have reduced this to some cross-references to key results, to put them in perspective with findings in literature, and to indicate how they allow us to address our research questions.

[Comment 7]
Language quality need to be improved. Though there are no significant grammar errors, but the dividing of paragraph and some expressions should be modified. I suggest a language polish before accept..

[Response]
This remark is much appreciated; we strived to improve the language where and as we could.

 

 

 



 

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Authors have responded to my comments well and made corrections. I recommend it could be accepted for publishing.

Author Response

Dear Editor,

1. About the accuracy of the GLC_FCS30D data

The accuracy we found for GLC_FCS30D aligns very well with the accuracy reported by the providers of the data; and, we believe this to be remarkably good given that it concerns a global data set with a high spatial (30 m) resolution.  

Having said that, you are right that the data has a substantial error margin, and as we now also point out in the text, for the earlier dates (before ~2005) practically it cannot even be quantified.

To acknowledge these limitations, we have added a paragraph in the discussion (lines: 454-468), as part of the section on "Limitations of using RUSLE" where we make the point that
- the accuracy we found, aligns with what's been reported globally, and also for the USA and the EU
- we are the first to report such figures for multiple time intervals, and for Africa
- the accuracy for this landcover data prior to 2005 cannot be confirmed
- inherently, there is some unquantifiable uncertainty in the landcover change data; but our findings align well with what we know about the socio-economic dynamics in the area (and actually also with the terrain knowledge of two of the authors)

2. About Fenta's method (2017)
We have now provided a longer explanation justifying the use of the Modified Fournier Index in the method section, such that it is also clear why we used Fenta's equation (lines:172-180, in the methods section);

We also added a reflection on this in the discussion, under the section "Limitations of using RUSLE" (lines 445-453)

By re-reading our paper; we made some minor language edits; particularly in the abstract and the conclusion.

We are grateful for the remarks which allowed us to improve the paper and sincerely hope that it will be acceptable for publication.

Sincerely,

Stefaan Dondeyne 

Back to TopTop