Land Cover Change Detection and Subsistence Farming Dynamics in the Fringes of Mount Elgon National Park, Uganda from 1978–2020

Round 1

Reviewer 1 Report

About the paper “Land cover change detection and subsistence farming dynamics in the fringes of Mount Elgon National Park, Uganda from 1978-2020.” By : Hosea Opedes * , Sander Mücher , Jantiene E.M. Baartman , Shafiq Nedala , Frank Mugagga

 

Abstract:

You speak about four periods, but you made the analysis on 4 images then you have emphasized the differences between these images. So you studied 4 years, not four periods. Yes, You can expand the changes to an entire period, but you should modify “ periods” with Landcover maps for 4 years 1978,1988,2001,2010,2020.

Paper

The paper is well structured, and the authors have the scientific maturity necessary for this article, yet there are some things that must be modified.

You should explain why you have used Landsat images instead of other ones….probably because of time consistency, but you should add this info into the paper.

Your arguments were presented on 4 periods, but the land cover/land-use changes are presented in figure no:7 from 1978 directly to 2020, without presenting the intermediary results, so why did you analyze the intermediate years if you do not present those periods. Each land cover and land use change periods have some socio-economical and political causes, you should not minimize your results presenting from Start to Finish in one photo.

We believe that Figure 8 should be presented in the detailed version from 1978 to 1988-2001-2010-2020, because, if we talk about the increase of  Build up category, as an irreversible phenomenon, we presume that is correct, but we take into consideration the increase in the surface of some natural category as Shrubs or Grassland,  we can’t give this changes to a single time period, these being reversible phenomenon’s on our timeline scale it might increase in the first years and decrease in the last one.

Also, we think that you should revise the results for some categories for some years ….for example in Table 4 your data looks like there are some problems. For example, the grassland goes from surfaces of ( in KM2)  49(1978) to 40(1988)  to 74(2001)  to 32(2010)  to 27 (2020). Are you sure about this data? What happened from 1988 to 2001 and then to 2010?. Could this be a satellite image that is irrelevant? If so, please remove it. If not, please explain.

In my opinion, I understand the necessity of research in an area where are very few studies, I also understand the difficulty of finding good reliable data and I appreciate the effort, but if some years are unusable, please forget about them.

The conclusions should express the findings of the paper, if the methods are good…etc please reshape them.

 

Author Response

Reviewer 1.

Abstract:

1. You speak about four periods, but you made the analysis on 4 images then you have emphasized the differences between these images. So you studied 4 years, not four periods. Yes, You can expand the changes to an entire period, but you should modify “ periods” with Landcover maps for 4 years 1978,1988,2001,2010,2020.

• Thank you for the observation, our study indeed considered Landsat images of four years (1978, 1988, 2001, 2010 and 2020). We have updated the abstract accordingly.

2. You should explain why you have used Landsat images instead of other ones….probably because of time consistency, but you should add this info into the paper.

• Yes indeed, we have added the justification of Landsat imagery use in this study. This is now reflected in subsection 2.2 of the article as follows: ‘Particularly Landsat images were prioritized over other images in this study because of its rich historic database (10 million scenes), back dating up to 1972 [37]. The images are also medium resolution (30x30m) per pixel [55] and are easily integrated across Landsat missions (1,2,3,5,7,8, and 9). The study did not consider replacing Landsat images with sentinel or other satellite images for recent scenes to avoid temporary mis-registration error due to residual geolocation errors and differences in spectral characteristics with Landsat [56].’

 

3. Your arguments were presented on 4 periods, but the land cover/land-use changes are presented in figure no:7 from 1978 directly to 2020, without presenting the intermediary results, so why did you analyse the intermediate years if you do not present those periods. Each land cover and land use change periods have some socio-economic and political causes, you should not minimize your results presenting from Start to Finish in one photo.

• Yes, we agree. We did analyse the separate decades (1978-1988, 1988-2001, 2001-2010 and 2010-2020), but we thought that this figure would be too busy for the main text and had presented it in the supplementary information instead. However, we are happy to include the figure with the decadal flows as figure 7 now.

4. We believe that Figure 8 should be presented in the detailed version from 1978 to 1988-2001-2010-2020, because, if we talk about the increase of  Build-up category, as an irreversible phenomenon, we presume that is correct, but we take into consideration the increase in the surface of some natural category as Shrubs or Grassland,  we can’t give this changes to a single time period, these being reversible phenomenon’s on our timeline scale it might increase in the first years and decrease in the last one.

• Thank you for this comment, we have updated figure 8 in the article to present the land cover flows from 1978-1988, 1988-2001, 2001-2010 and 2010-2020. Although there are irreversible changes especially for the built-up, reversible changes are also visible especially in the park as shown in figure 8.

5. Also, we think that you should revise the results for some categories for some years ….for example in Table 4 your data looks like there are some problems. For example, the grassland goes from surfaces of ( in KM2) 49(1978) to 40(1988)  to 74(2001)  to 32(2010)  to 27 (2020). Are you sure about this data? What happened from 1988 to 2001 and then to 2010?. Could this be a satellite image that is irrelevant? If so, please remove it. If not, please explain

 

• Thank you for the observation made, we checked the output of the results and confirmed those variations. We established reasons behind those anomalies during ground truthing, expert interviews and also review of literature on previous studies conducted in the upper Manafwa watershed. We have incorporated this reason in the manuscript in sub section 3.2 as follows: ‘Results show a significant increase of 10.6% and decrease of 13.2% in grassland cover between 1988 and 2001, and 2001 and 2010 respectively. These variations indicate high conversion rates of shrubs to grassland and grassland to agriculture and other classes. Specifically, agriculture increased by 11.1% while shrubs significantly decreased by 5.9% between 2001 to 2010. These changes are attributed to conservation policy change in Uganda during the same period. The policies and legislation frameworks called for zonation and re-demarcation of the park which consequently affected enforcement [57,68].’ More details are also included in the discussion section of the article.

 

6. Also, we think The conclusions should express the findings of the paper, if the methods are good…etc please reshape them.

 

• Thank you for the observation made, we thought the conclusion sufficiently summarized the findings of the study. We have improved the conclusion of the article to also highlight the importance of the methods used as follows: ‘Despite emergence of new image classification methods such as SVM, maximum likelihood method still provides reliable results.’

Author Response File: Author Response.pdf

Reviewer 2 Report

This manuscript presents an analysis of spatial-temporal land cover changes from 1917 to 2020 using Landsat data. The study area is the Mount Elgon region in Uganda. The paper is well organized. The result is thoroughly analyzed.

My main concern is the collection of remote sensing data. Did the authors only use five Landsat images from 1978 to 2020? If so, it is too few for a time series analysis. To get a more reliable result, I suggest the authors to adopt at least one image from each observation year (~40 images total).

Detailed remarks:

There are no line numbers throughout the paper. It is difficult for reviewers to mark detailed comments.

Author Response

Reviewer 2

1. This manuscript presents an analysis of spatial-temporal land cover changes from 1917 to 2020 using Landsat data. The study area is the Mount Elgon region in Uganda. The paper is well organized. The result is thoroughly analysed.

• Thank you for the compliments, we appreciate them.

2. My main concern is the collection of remote sensing data. Did the authors only use five Landsat images from 1978 to 2020? If so, it is too few for a time series analysis. To get a more reliable result, I suggest the authors to adopt at least one image from each observation year (~40 images total).

• Thank you for the comment. Indeed we used five images (1978, 1988, 2001, 2010, 2020). Our study focused on major and significant changes over several decades. It was not the objective of our study to analyse minor changes (from year to year). These changes are harder to find when a high classification accuracy of >83% is considered. Furthermore, historical Landsat satellite yearly images are difficult to find when considering the search criteria we followed as mentioned in sub section 2.2 of the article. We think that the decadal and major changes are well reflected in the images we analysed.     

Author Response File: Author Response.pdf

Reviewer 3 Report

This research on Mt. Elgon land cover change using Landsat images is comprehensive and provides valuable knowledge for the regional research in agriculture, forest management research etc. I only have a few minor issues to be revised or clarified before considering publishing on remote sensing.

 

1. The authors mentioned aerial images in the data section. I am interested in that and suggest adding a figure to show some of the imagery
2. Table 4 & Figure 5, interestingly, during 1988~2001, grassland and shrubs experienced significant increases in this period but decreases afterwards. Do you have any clues on this?
3. A significant degradation trend was found in the tropical high forest stocks. Was that caused mainly by human factors or natural (e.g. climate change) or both?

 

Author Response

Reviewer 3

1. This research on Mt. Elgon land cover change using Landsat images is comprehensive and provides valuable knowledge for the regional research in agriculture, forest management research etc.

• Thank you for these positive comments.

2. The authors mentioned aerial images in the data section. I am interested in that and suggest adding a figure to show some of the imagery

• The aerial photographs were part of the ancillary data that was used in validation of the old images. We have added one of the photographs in the supplementary material.

 

3. Table 4 & Figure 5, interestingly, during 1988~2001, grassland and shrubs experienced significant increases in this period but decreases afterwards. Do you have any clues on this?

• Yes, this is true. We doubled checked these variations of land cover for grasslands, shrubs and agriculture and confirmed them. The reasons for these significant increases and decreases in 1980s and early 2000 was attributed to the impact of the enactment of conservation policies, re-demarcation and enforcement. Whenever there was strict monitoring, the park area revegetated and before that time, there was a lot of laxity on conservation area management. Most laws and policies on conservation were enacted in the late 1990s and early 2000. We have added these reasons in subsection 3.2 and section 4 of the paper. The paragraph has been updated to reflect the following information: ‘Results show a significant increase of 10.6% and decrease of 13.2% in grassland cover between 1988 and 2001, and 2001 and 2010 respectively. These variations indicate high conversion rates of shrubs to grassland and grassland to agriculture and other classes. Specifically, agriculture increased by 11.1% while shrubs significantly decreased by 5.9% between 2001 to 2010. These changes are attributed to conservation policy change in Uganda during the same period. The policies and legislation frameworks called for zonation and re-demarcation of the park which consequently affected enforcement [57,68].’  More details are also included in the discussion section of the article.

 

4. A significant degradation trend was found in the tropical high forest stocks. Was that caused mainly by human factors or natural (e.g. climate change) or both?

• Thank you very much for this comment, significant degradation was prevalent in the park area as seen in figures 7 and 8. The causes were mainly arising from human factors, especially population pressure, high demand for arable land, human-induced fires during the dry season and limited monitoring and enforcement of the laws. These factors have been explained further in Section 4.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

I think the authors made the requested changes and I think it can be published.

Reviewer 2 Report

The point-to-point responses are adequate. Thanks for your revision.