Landslide Susceptibility Mapping Using Integrated Methods: A Case Study in the Chittagong Hilly Areas, Bangladesh

Round 1

Reviewer 1 Report

Dear authors,

thank you for your study dealing with the application of integrated methods to assess the landslide susceptibility in Chittagong Hilly Areas. Your study provides significant information about the integrated use of frequency ratio (FR), analytical hierarchy process (AHP), and logistic regression (LR).

However, there are some aspects related to the aim and the layout of the manuscript not discussed and/or presented properly.

The manuscript is well elaborated, but the main limitation of the study is that the analysis did not use some essential predisposing factors like lithology due to the unavailability of data.

In my opinion, in such studies analysing the LS in a regional framework, lithology is a key factor and it is necessary to be considered as a predisposing factor.

Lithology is an important factor since different lithological units may be affected by different landslide types with variable susceptibility degrees. Moreover, superficial deposits, mostly exposed to weathering, may influence land permeability, geotechnical parameters and, therefore, the landslide type.

See these reference:

Segoni, S., Pappafico, G., Luti, T. et al. Landslide susceptibility assessment in complex geological settings: sensitivity to geological information and insights on its parameterization. Landslides 17, 2443–2453 (2020). https://0-doi-org.brum.beds.ac.uk/10.1007/s10346-019-01340-2

Bahrami, S., Rahimzadeh, B. & Khaleghi, S. Analyzing the effects of tectonic and lithology on the occurrence of landslide along Zagros ophiolitic suture: a case study of Sarv-Abad, Kurdistan, Iran. Bull Eng Geol Environ 79, 1619–1637 (2020). https://0-doi-org.brum.beds.ac.uk/10.1007/s10064-019-01639-3

Abedini, M., Tulabi, S. Assessing LNRF, FR, and AHP models in landslide susceptibility mapping index: a comparative study of Nojian watershed in Lorestan province, Iran. Environ Earth Sci 77, 405 (2018). https://0-doi-org.brum.beds.ac.uk/10.1007/s12665-018-7524-1

Yalcin, A.; Reis, S.; Aydinoglu, A.C.; Yomralioglu, T. A GIS-based comparative study of frequency ratio, analytical hierarchy process, bivariate statistics and logistics regression methods for landslide susceptibility mapping in Trabzon, NE Turkey. Catena 2011, 85, 274–287

Please note my comments and suggestions in the attached PDF document.

Comments for author File: Comments.pdf

Author Response

Reviewer #1:

Response

Thank you for the comments and suggestions. These comments and suggestions have enabled us to improve the quality of our manuscript. We have accepted most of the corrections. Please check below the answers for each of the concerns about the manuscript.

1) thank you for your study dealing with the application of integrated methods to assess the landslide susceptibility in Chittagong Hilly Areas. Your study provides significant information about the integrated use of frequency ratio (FR), analytical hierarchy process (AHP), and logistic regression (LR).

However, there are some aspects related to the aim and the layout of the manuscript not discussed and/or presented properly.

The manuscript is well elaborated, but the main limitation of the study is that the analysis did not use some essential predisposing factors like lithology due to the unavailability of data.

In my opinion, in such studies analyzing the LS in a regional framework, lithology is a key factor and it is necessary to be considered as a predisposing factor.

Lithology is an important factor since different lithological units may be affected by different landslide types with variable susceptibility degrees. Moreover, superficial deposits, mostly exposed to weathering, may influence land permeability, geotechnical parameters and, therefore, the landslide type.

Response: Thank you pointing out this issue. We have collected the lithological map of the study area and carried out the whole analysis again. It helped to improve the accuracy of the susceptibility map. The analysis indicates that lithology is an important factor. Please see the revised manuscript for detail information.

2) Also consider and add seismic activity and wildfire as triggering factors.

In order to improve and support this sentence, I suggest you the following references:

Calista, M.; Miccadei, E.; Piacentini, T.; Sciarra, N. Morphostructural, Meteorological and Seismic Factors Controlling Landslides in Weak Rocks: The Case Studies of Castelnuovo and Ponzano (North East Abruzzo, Central Italy). Geosciences 2019, 9, 122.

Carabella, C.; Miccadei, E.; Paglia, G.; Sciarra, N. Post-Wildfire Landslide Hazard Assessment: The Case of The 2017 Montagna Del Morrone Fire (Central Apennines, Italy). Geosciences 2019, 9, 175.

Response: Thank you for providing these references. We have incorporated them in the revised manuscript.

3) I will be useful to see the following references to better describe the landslide susceptibility methods:

Marsala, V.; Galli, A.; Paglia, G.; Miccadei, E. Landslide Susceptibility Assessment of Mauritius Island (Indian Ocean). Geosciences 2019, 9, 493.

Carabella, C.; Miccadei, E.; Paglia, G.; Sciarra, N. Post-Wildfire Landslide Hazard Assessment: The Case of The 2017 Montagna Del Morrone Fire (Central Apennines, Italy). Geosciences 2019, 9, 175.

Response: Thank you for providing these references. We have incorporated them in the revised manuscript.

4) The Study Area section is well prepared and supported by relative references.

I would suggest to better describe the main features of the area in terms of:

Geology = please revise lines 110-119 in order to better describe the geological framework of the area (according to this layout, it seems a list of geological formations)

Geomorphology = please add some lines to highlight the landslide activity in the area, as partially done in the introduction area (lines 86-98).

Response: Thank you for your comment and advice. We have rewritten the required section.

5) Please consider to add a less elaborated north arrow.

Please revise the position of the maps, serving as location maps, in order to better depict them and to locate the study area in a national and regional framework. Maybe, put them in the same corner of the Figure and move the scale bar and the legend in lower right corner.

Response: We have revised all the figures according to the advice.

6) I would suggest the Authors to add a figure depicting this situation, also to support the field work activity

Response: We have added figures. Please check figure 2 and 3.

7) please check this paragraph in order to better explain your work in the selection of non-landslide locations.

The Authors say .."We overlay the landslide locations on the predisposing factor’s map, for example, slope map, to determine a threshold" ...

Did you do this process of overlay for all the predisposing factors?

Response: Thank you for pointing out this issue. We have revised the manuscript.

Please check the revised manuscript. We have highlighted the changes that we made according to your advice.

Author Response File: Author Response.docx

Reviewer 2 Report

The work is interesting and the topic is related to the magazine. The use / testing of integrated models for landslide vulnerability mapping consists of an interesting research area. In my opinion, this document is - in general - well organized and written. Despite these positive characteristics the manuscript need some major revision to reach the high quality standards of the "Geosciences" journal. 

To my opinion, the authors have to deal with the following substantial issues:

1. The have to add "Geology" or "Lithology" to the predisponding factors' list. This factor is critical to landslides susceptibility analyses and there is huge ammount of references to support this. In lines 534 - 536 authors mention that they have no access to soil data (soil types, permeability, depth) etc. What about lithological / geological maps ? I strongly believe that these maps /factors should be incorporated in the modeling.
2. In landslide susceptibility modelling we have to study similar types of landslides with similar failure mechanism. It is not clear if this is the case in this study. The only reference in the manuscript is in line 146-147 which is rather confusing ("304 slides, 233 flows, 74 flows (?), 31 complex, 10 topple, 78 unecognized"). I suggest to study only similar landslides.
3. To conduct LR and validation analyses the authors have to create the non-landslides dataset. In page 11 the authors describe the method for the creation of this dataset. Ths paragraph is somehow confusing: the authors stated that "selecting a safe zone ...can bring bias to the FR_LR model" and at the same time  they adopted the concept of the safe zone in their work (areas with slope < 4,5 %). I am very sceptical with the use of safe zone in LR analysis especially when it is combinde with the removal of the slope from the factors under investigation. In any case, the authors have to ensure that all the randomly selected non-landslides are non sliding areas (e.g. with RS image interpretation in these points)
4. The so called "FR_AHP" model is jut a variation of the FR model as the main advantage of the AHP approach is the use of the previous knowledge of experts for the special conditions in the study area. Thus, as expected, these two models produced similar results. To compare these models the authors have to provide data about the sensitivity and the specificity of each model and not only the AUC. 
5. In the discussion section the authors have to present the limitations and the assumptions of their work. (adequate number of predispong factors, spatial stationaruty of the relationship between predispong factors and landslide susceptibility, adequate landslides inventory etc).
6. Please remove comments about the assumptions of the paper in the conclusions section.
7. The authors have to add more comments about the significant differences between FR_AHP and FR_LR susceptibility maps. 
8. The integration of two models is not "a priori" produce bettern results than the primary models. In some cases these models can inheret the disadvantages of both models. Please comment.
9. some other issues:

• L134: incomplete sentense
• L140: "todevelop"
• L147: a) Add details about the landslides in the study area b) add details about the non-landlsides
• L162: Examine some other classification schemes for slope parameter or at least comment why you choose the natural brakes classification
• L206-208: The annual rainfall is not always related to the incesant rainfall occured in a very short period. This is an assumption of your approach.
• L 224: What statics you implied in the 3X3 filter? Mean value? max value? frequency? Why?
• 262: by usng FR statistics for the pairwise comparison you assume that this statistics is constant and representative for the area under investigation. This is an assumption of the paper.
•  L398: "for visual representation". Only for that? what about validation analysis?
• L432 - 449: Please add comments about the "overestimations" or "undrestimations" of the models
• L491: please explain. What is a "more practial model"? 
• L507 - 509: the use of these areas for development activities is not an indicator of the FR_LR superiority
• L522: Syntax. You didnt compare the differences but the models.
• L543: Add "...and non-landslides"
• L547: ??? Did you have limited landslide locations availability in the study area?

Author Response

Reviewer #2:

Response

Thank you for the comments and suggestions. These comments and suggestions have enabled us to improve the quality of our manuscript. We have accepted most of the corrections. Please check below the answers for each of the concerns about the manuscript.

1.  The have to add "Geology" or "Lithology" to the predisponding factors' list. This factor is critical to landslides susceptibility analyses and there is huge ammount of references to support this. In lines 534 - 536 authors mention that they have no access to soil data (soil types, permeability, depth) etc. What about lithological / geological maps ? I strongly believe that these maps /factors should be incorporated in the modeling.

Response: Thank you for pointing out tis issue. We have collected lithological map form the Geological Survey of Bangladesh. Then we have carried out the whole analyses again. It has helped to improve the accuracies of the maps. Please check the revised manuscript.

2. In landslide susceptibility modelling we have to study similar types of landslides with similar failure mechanism. It is not clear if this is the case in this study. The only reference in the manuscript is in line 146-147 which is rather confusing ("304 slides, 233 flows, 74 flows (?), 31 complex, 10 topple, 78 unecognized"). I suggest to study only similar landslides.

Response: In many studies specially for regional scale susceptibility mapping types of landslides were not considered (Schiker and Moon, 2012; Marsala et al. 2019). It is also true, separate maps for separate types could have been produced and then integrated into a single map. But it is beyond the scope of this study. In the study area, among the 730 landslides 304 were slides, 233 were flows, 74 were falls, 10 were topple, and 31 were complex. Since these two types cover almost 75% of the types therefore, we did not consider the types in susceptibility mapping.

3.To conduct LR and validation analyses the authors have to create the non-landslides dataset. In page 11 the authors describe the method for the creation of this dataset. Ths paragraph is somehow confusing: the authors stated that "selecting a safe zone ...can bring bias to the FR_LR model" and at the same time  they adopted the concept of the safe zone in their work (areas with slope < 4,5 %). I am very sceptical with the use of safe zone in LR analysis especially when it is combinde with the removal of the slope from the factors under investigation. In any case, the authors have to ensure that all the randomly selected non-landslides are non sliding areas (e.g. with RS image interpretation in these points)

Response: In this for absence data sampling we used exploratory data analysis. We extracted the slope value for each of the landslide locations and sorted them in ascending order. Then divided it into ten quantiles. The first quantile includes the first 10% of the slope values of the landslide locations. It was 4.5 degree. Therefore, we considered below 4.5-degree slope as safe zone or areas where absence data can be sampled. Since we carried out exploratory data analysis based on slope, we excluded it form the logistic regression modeling. Otherwise the model would have been biased towards the slope and it would have showed the slope as the most important factor. The coefficient of the slope would have been the highest. It would have given a biased susceptibility map. Since the size of the study area is large, it is not a good way to check all the non-landslide locations using satellite images. The second law of geography which is related to spatial heterogeneity would have been violated in this regard. That is why it is better to use data exploratory analysis and determine the location for absence data sampling for regional scale mapping. Moreover, the study of Zhu et al. 2019 has proved that random sampling of absence data from the whole study area gives faulty susceptibility maps.

4. The so called "FR_AHP" model is jut a variation of the FR model as the main advantage of the AHP approach is the use of the previous knowledge of experts for the special conditions in the study area. Thus, as expected, these two models produced similar results. To compare these models the authors have to provide data about the sensitivity and the specificity of each model and not only the AUC. 

Response: Thank you for your advice. We have included statistical indices: sensitivity, specificity and over all accuracy in the revised manuscript. It helped to improve the performance assessment and the quality of the paper. Please check the manuscript. Mainly the lines: 358-364; 470-481

5. In the discussion section the authors have to present the limitations and the assumptions of their work. (adequate number of predispong factors, spatial stationaruty of the relationship between predispong factors and landslide susceptibility, adequate landslides inventory etc).

Response: We have moved the limitations section into the discussion. Please check lines: 587-596

6. Please remove comments about the assumptions of the paper in the conclusions section.

Response: thank you for your advice. We have removed.

7.The authors have to add more comments about the significant differences between FR_AHP and FR_LR susceptibility maps. 

Response: We have added more comments on the significance of the FR_AHP model. Please check lines: 470-504; 575-586

8. The integration of two models is not "a priori" produce bettern results than the primary models. In some cases these models can inheret the disadvantages of both models. Please comment.

Response: We have discussed it in the revised manuscript. Please check the lines: 575-586.

Some other issues:

• L134: incomplete sentense
• L140: "todevelop"

Response: Thank you for pointing out these issues. We have checked the whole manuscript for grammatical issues.

• L147: a) Add details about the landslides in the study area b) add details about the non-landlsides

Response: We have provided a new figure for the types of landslides and discussed about the non-landslide locations.

• L162: Examine some other classification schemes for slope parameter or at least comment why you choose the natural brakes classification

Response: We have checked different classification schemes and found that natural break gives the best performance.

• L206-208: The annual rainfall is not always related to the incesant rainfall occured in a very short period. This is an assumption of your approach.

Response: we have deleted the sentence.

• L 224: What statics you implied in the 3X3 filter? Mean value? max value? frequency? Why?

Response: We have used the low filter or the mean value. It is used to remove the noise. Generally a cell with high landslide susceptibility surrounded by cells with low susceptibility indicates the presence of noise. Therefore, mean value is recommended to use. Other filters like high pass filter is used for edge enhancement.

• 262: by usng FR statistics for the pairwise comparison you assume that this statistics is constant and representative for the area under investigation. This is an assumption of the paper.

Response: Thank you for pointing out this issue. We have revised the manuscript.

•  L398: "for visual representation". Only for that? what about validation analysis?

Response: Thank you for pointing out this issue. We have revised the manuscript.

• L432 - 449: Please add comments about the "overestimations" or "undrestimations" of the models

Response: We have discussed it in the revised manuscript.

• L491: please explain. What is a "more practial model"? 

Response: We have discussed it in the revised manuscript.

• L507 - 509: the use of these areas for development activities is not an indicator of the FR_LR superiority

Response: We have deleted these sentences.

• L522: Syntax. You didnt compare the differences but the models.

Response: We have revised the manuscript.

• L543: Add "...and non-landslides"
• L547: ??? Did you have limited landslide locations availability in the study area?

Response: Thank you for pointing out these issues. We have revised the manuscript.

Please check the revised manuscript. We have highlighted the changes that we made according to your advice.

Author Response File: Author Response.docx

Reviewer 3 Report

Dear Authors,

Attached is the annotated file, where I have given my comments. I am not sure whether it should be considered as an Article seeing there is hardly any novelty. These techniques are quite old now. So, if it's considered, it can be done as a Technical Note.

Thanks

Comments for author File: Comments.pdf

Author Response

Reviewer #3:

Response

Thank you for the comments and suggestions. These comments and suggestions have enabled us to improve the quality of our manuscript. We have accepted most of the corrections. Please check below the answers for each of the concerns about the manuscript.

1. Attached is the annotated file, where I have given my comments. I am not sure whether it should be considered as an Article seeing there is hardly any novelty. These techniques are quite old now. So, if it's considered, it can be done as a Technical Note.

Response: This is the first landslide inventory that covers the whole Chittagong Hilly Areas of Bangladesh. For this mapping we prepared the first inventory of the study area. Moreover, we produced the land use/land cover map of the area. We have shown that simple but integrated method can give satisfactory accuracy for regional scale susceptibility mapping. We have used two performance assessment methods and a comparison method. Therefore, we believe it is a research paper rather than a technical note.

2. You should refer the work of:

Gariano and Guzzetti, (2016) Landslides in a changing climate, https://0-doi-org.brum.beds.ac.uk/10.1016/j.earscirev.2016.08.011

Response: we have incorporated it in the revised manuscript.

3. The definition of susceptibility needs to be provided. Usually, the terms 'susceptibility', 'hazard' and 'risk' are often used interchangeably.

Response: we have defined it in the revised manuscript.

4. The above para explains in detail about quantitative methods. But, this para doesn't have more details. Please add more details. You can refer to Reichenbach et al. (2018) , Ref. No. 34

Response: we have defined it in the revised manuscript.

5. Please put these places on map. The names are native to the region or else mention that which part of the country these places are at.

Response: We have provided a new map.

6. So, your study has a detailed landslide inventory for the entire region?

Response: Yes.

Please check the revised manuscript. We have highlighted the changes that we made. We have carried out the whole analysis and included lithology in the factor. We have also included statistical indices for performance assessment.

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

The authors revised their paper substantially according to the suggestions. The paper now meets - in general -  the quality standards for publishing in IJGI after minor revision. Please check the following points:

1. In the new Figure 6 :

a) use the same fonts in the legends of the maps.

b) keep the same order in map / legend sequences 

c) use blue color for the class "water bodies" in the legend of "lithology"

d) (optional): use blue tones for the symbols of rainfall classes

2. Make a comment in the discussion section, about your assumption to analyse all the types of landslides by the same way (you assume similar sliding mechanisms)

Author Response

Reviewer #2:

Response

Thank you for the comments and suggestions. These comments and suggestions have enabled us to improve the quality of our manuscript. We have accepted most of the corrections. Please check below the answers for each of the concerns about the manuscript.

1. In the new Figure 6 :

1. a) use the same fonts in the legends of the maps.
2. b) keep the same order in map / legend sequences 
3. c) use blue color for the class "water bodies" in the legend of "lithology."
4. d) (optional): use blue tones for the symbols of rainfall classes

Response: We have accepted the suggestions and revised the manuscript.

2. Comment in the discussion section about your assumption to analyze all the types of landslides in the same way (you assume similar sliding mechanisms)

Response; Thank you for mentioning that. We have added new sentences at the end part of the discussion.

Reviewer 3 Report

Dear Authors,

Thank you for revising the manuscript. Unfortunately, you haven't responded to all my queries. It seems that you have deliberately chosen only those comments with which you were comfortable. This is very a unfortunate situation, as a reviewer, we spent time enhancing the manuscript and examining it from a scientific lens. But, it seems you do not care or want to provide a formidable response to the questions. 

The few responses which you provided are very generic. Your response to the Technical Note question is " This is the first landslide inventory that covers the whole Chittagong Hilly Areas of Bangladesh. For this mapping, we prepared the first inventory of the study area. Moreover, we produced the land use/land cover map of the area. We have shown that a simple but integrated method can give satisfactory accuracy for regional-scale susceptibility mapping. We have used two performance assessment methods and a comparison method. Therefore, we believe it is a research paper rather than a technical note."

I hope you understand the difference between an original article and a Technical Note. Doing an already proven study for some other region, does not suffice for an original study. It has no novelty and contribution. That's why a Technical Note is prepared. Secondly, the integration of ML techniques is a decade-old problem. Read the papers of Pradhan, Pourghasemi, Pham, Shirzadi, there is an abundance of literature. 

I am by no means against getting accepted as Original Article, but for that you need to provide strong scientific reasoning, which is currently missing.

Anyways, article type is the last concern. But, the minimum expectation from authors is to respond to the comments of a reviewer, which you have not done properly.

Thanks  

Author Response

Reviewer #3:

Response

Thank you for the comments and suggestions. These comments and suggestions have enabled us to improve the quality of our manuscript. We have accepted most of the corrections. Please check below the answers for each of the concerns about the manuscript.

1. Not sure if the climate is a predisposing factor? If rainfall-induced landslides are considered, then rainfall will be predisposing or triggering?

I guess climatic variation/climatic change would be better.

Response: Thank you for pointing out this issue.  Yes, the climate can not be a predisposing factor. We have deleted it from the text.

2. You should refer to the work of:

Gariano and Guzzetti, (2016) Landslides in a changing climate, https://0-doi-org.brum.beds.ac.uk/10.1016/j.earscirev.2016.08.011

Response: We have incorporated it in the revised manuscript.

3. The definition of susceptibility needs to be provided. Usually, the terms 'susceptibility', 'hazard' and 'risk' are often used interchangeably.

Response: We have defined it in the revised manuscript. Please check line 48-51.

4. The above para explains in detail quantitative methods. But, this para doesn't have more details. Please add more details. You can refer to Reichenbach et al. (2018), Ref. No. 34

Response: Thank you for pointing out this. We have rewritten this part and added the reference of Reichenbach et al. (2018) and Aleotti and Chowdhury (1999). Please check lines 70-81 in the revised manuscript.

5. Please put these places on the map. The names are native to the region, or else mention which part of the country they are in.

Response: We have provided a new map.

6. What about the data inventory?

Response: We have rewritten the sentence as

The method selection for landslide susceptibility mapping depends on the scale of analysis, cost, and timeline of the project and the inventory's data [19].

7. Here, you can refer to the recent review articles:
8. Merghadi et al. (2020) Machine learning methods for landslide susceptibility studies: A comparative overview of algorithm performance. https://0-doi-org.brum.beds.ac.uk/10.1016/j.earscirev.2020.103225
9. Dikshit et al. (2020) Pathways and challenges of the application of artificial intelligence to geohazards modeling. https://0-doi-org.brum.beds.ac.uk/10.1016/j.gr.2020.08.007

Response; Thank you for providing these useful articles. We have incorporated them in the revised manuscript.

8. So, there is no specific range of deaths due to landslides? You can check Froude and Petley (2018) or NASA GLC.

check this work on Indian Himalayas:

Dikshit, A.; Sarkar, R.; Pradhan, B.; Segoni, S.; Alamri, A.M. Rainfall Induced Landslide Studies in Indian Himalayan Region: A Critical Review. Appl. Sci. 2020, 10, 2466.

Response: Thank you for pointing out this issue. We have rewritten the sentence.

9. Yes, it's correct. But the question is, at what spatial scale? If there is a significant region where land degradation is affecting landslides, then the spatial scale becomes important.

Response: Thank you for pointing out this issue. We have added the following sentence.

A landslide susceptibility map of the whole area would provide useful guidance for land use and regional planning in this area, specifically in urban areas and areas where cultivation agriculture is practiced.

10. Please put these places on the map. The names are native to the region, or else mention which part of the country they are in.

Response: Thank you for pointing out this issue. We have provided a new map.

11. So, your study has a detailed landslide inventory for the entire region?

Response: We have rewritten this line and deleted the sentence partially because of the absence of a landslide inventory for the entire CHA.

12. Please mention in terms of Koppen classification. Read this paper's description of the study area:

Gariano et al. Automatic calculation of rainfall thresholds for landslide occurrence in Chukha Dzongkhag, Bhutan. Bull Eng Geol Environ 78, 4325–4332 (2019). https://0-doi-org.brum.beds.ac.uk/10.1007/s10064-018-1415-2

Response: Thank you for suggesting this. We have incorporated it into the revised manuscript.

The added sentences are

According to the Koppen classification, CHA's climate is characterized by the tropical monsoon climate with annual rainfall from 2540-3777 mm [33]. About 80% of the landslides occur during the monsoon season (June-September) due to excessive monthly rainfall (about 480 mm) [36]. 15% of the rainfall occurs during pre-monsoon (March-May) and post-monsoon (October-November) seasons. The Winter season (December-February) is the driest season in the study area.

13. Not only rock, it's also because of soil and other geotechnical parameters.

Response: Thank you for pointing out this issue. We have revised the sentence. Please check the revised manuscript.

Please check the revised manuscript. We have highlighted the changes that we made. We have carried out the whole analysis and included lithology in the factor. We have also included statistical indices for performance assessment.

14. This is a very coarse resolution. Higher-resolution DEMs are available publicly? Why not use that

Response: In this article, Evaluating the Effects of Digital Elevation Models in Landslide Susceptibility Mapping in Rangamati District, Bangladesh, authors have found that in the study area, the accuracy of the susceptibility maps does not vary with the use of different global DEMs like SRTM, ALOS PALSAR, and SRTM. The study area has a local DEM, and the resolution was the highest, but the quality is very poor. It is also mentioned in the article.

https://0-www-mdpi-com.brum.beds.ac.uk/2072-4292/12/17/2718

Therefore, we used ASTER DEM in our study.

15. Refer:

Huete et al. (2002) Overview of the radiometric and biophysical performance of the MODIS vegetation indices. https://0-doi-org.brum.beds.ac.uk/10.1016/S0034-4257(02)00096-2

Response: Thank you for providing this article. We have incorporated it into the revised manuscript.

16. The use of NDVI is specific on a certain day? If you take the cumulative rolling average for the entire landslide period, will it not make a difference?

Response: Yes, it can make a difference. But if we want to take the cumulative average, then we must use a series of Landsat images. In the study area, during the monsoon, most of the images are filled with the cloud. The use of certain day gives an idea about the condition of NDVI in the study area and its relationship with the landslide. In the study, we have used 13 causal factors, and the impact of one factor will not be that high on the landslide susceptibility map.

17. Isn't the number of weather stations less? What is the buffer distance of landslide locations? Mention how did you assign the rainfall values with respect to weather stations? Ideally, for mountainous regions, the buffer is 5 km, but 15 km is also accepted. Refer to Gariano's work, which I mentioned earlier

Response: We have the data of 32 weather stations and the Kriging method for spatial interpolation.

We used the annual rainfall data from thirty-two weather stations of the Bangladesh Meteorological Department (BMD) to produce the annual rainfall map (Figure 6a) of the study area. We used the Kriging tool in ArcGIS to produce a rainfall prediction surface map of the study area using these weather stations' data

Round 3

Reviewer 3 Report

Dear Authors,

These are my comments:

Not sure whether it was difficult to find my annotated pdf in the first version as you have still missed some of my comments.

1. Kindly refrain from using the word "We" in a scientific manuscript. I read your latest version, it also has several sentences containing "we", "our". Remove it
2. Figure 6 (a,b) needs improvement. This cannot be accepted in an article
3. The article needs serious English revision from a native speaker.

Now, your responses:

1. So, there is no specific range of deaths due to landslides? You can check Froude and Petley (2018) or NASA GLC.

Response: Thank you for pointing out this issue. We have rewritten the sentence.

Your changes in the manuscript are incomplete. 

2. The Koppen classification doesn't have the reference of Koppen. You should add it.

3. Here, you can refer to the recent review articles:

Response; Thank you for providing these useful articles. We have incorporated them in the revised manuscript.

These changes are not reflected in the manuscript.

The title in your manuscript has " in the end.

L576- "In our study we did not prepare...?? It seems there was a hurry to submit the manuscript. 

I would again advise the authors to check all the comments and suggestions, incorporate it. The English needs to be revised. The graphs needs to be up to the standard. The references are incomplete, with no uniformity

Author Response

Reviewer #3:

Response

Thank you for the comments and suggestions. These comments and suggestions have enabled us to improve the quality of our manuscript. We have accepted most of the corrections. Please check below the answers for each of the concerns about the manuscript.

1. Kindly refrain from using the word "We" in a scientific manuscript. I read your latest version, it also has several sentences containing "we", "our". Remove it

Response: Thank you for pointing out these issues. We have removed all we and our from the article.

2. The article needs serious English revision from a native speaker.

Response; Thank you for pointing out this issue. After finishing an article, we send it to the writing center of our university. The writing center checked the first draft. During the revision, b some problems occurred that we did not review. We have revised the whole manuscript again and removed all the grammatical and spelling mistakes.

For scientific manuscripts, it was recommended to use passive voice. In recent days authors are using us and our words as well. Our writing senders recommend mixing up so that the manuscript becomes more readable, and readers do not feel bored. That is why in our paper, we followed this format. Nevertheless, we accepted the advice.

3. The title in your manuscript has " in the end.

L576- "In our study we did not prepare...?? It seems there was a hurry to submit the manuscript. 

I would again advise the authors to check all the comments and suggestions, incorporate it.

Response: We have revised the whole manuscript tried to get rid of these issues.

4. Figure 6 (a,b) needs improvement. This cannot be accepted in an article

Response: We have provided a new Figure.

5. 1. So, there is no specific range of deaths due to landslides? You can check Froude and Petley (2018) or NASA GLC.

Response: Thank you for providing this reference. The data of NASA is not updated and based on the whole landslide casualties of the study area. We carried out a field investigation under several projects. We have used the data of the Department of Disaster of Bangladesh, which is not freely available. They have accurate data, and we have also used the data from newspapers. Therefore, this is the most precise number, and we do not want to refer to the NASA GLC. We have provided the reference of three articles.  

6. The Koppen classification doesn't have the reference of Koppen. You should add it.

Response; Thank you for pointing out this issue. We have provided the reference in the revised manuscript.

7. Here, you can refer to the recent review articles

Response: We have incorporated them in the revised manuscript.

8.  The English needs to be revised. The graphs needs to be up to the standard. The references are incomplete, with no uniformity

Response: We have revised the manuscript and improved the quality of the English. We have checked the references and ensured uniformity. We have revised Figure 6a and b. Otherwise, it seems to us a very generic comment. May be reviewer should have been more specific, like Figures 6 and b. We have provided 600 dpi images, and we believe this is the standard.