Multi-Temporal Sentinel-1 SAR and Sentinel-2 MSI Data for Flood Mapping and Damage Assessment in Mozambique

Round 1

Reviewer 1 Report

In this work a method for a "near real time" flood mapping utilizing multi-temporal Sentinel-1 Synthetic Aperture Radar (SAR) data acquired in Beira Municipality and Macomia district (Mozambique) is proposed. A technique based on Otsu’s thresholding allowed to map automatically flooded areas that have been compared with available data of Copernicus Emergency Management Service. Furthermore, Sentinel-2 MSI images have been used to produce a land cover (LC) map of the study area and estimate the percentage of flooded areas in each LC class, and, then, the correspondent damage. The work aims to provide local authorities with tools to face and mitigate the impact of floods.

Some sections result clear and well organized but others can be improved. The arrangement of section 3 and section 4 does not improve the clarity of the manuscript: these two sections could be comprised in a more general section: “material and methods” and the description of the study area could be keep separated. 

In section 2 (Literary Review), table 1 and table 2 have to be revised. In table 1, second column, both S1 and S2 are included under “SAR Data”. In the fifth column, I suggest to use always the same syntax form. For example, table 2, line 2, column 5: 1-Good scalability; 2 - Have high accuracy (but, using the same syntax form than at point 1, it should be: High accuracy)…and so on. 

In order to improve manuscript’s reproducibility results, it should be added appropriate references to Copernicus Emergency Management Service (Copernicus EMS) data, used for the validation of results.

Figure 4, subsection 4.3: the clarity of the flowchart should be improved using a more specific language: what is the meaning of “difference of images”?

Table 5 and Table 6, Accuracy assessment: authors have to explain (also with the formula, if it is necessary) the used parameters, in particular overall accuracy and kappa. Also if authors state that a description of these parameters can be found in the references, if possible the reading of an article should be comprehensive by itself.

Finally, the last section: “Conclusions and future research” could be surely improved and treat more in deep (and more convincing) an evaluation of strong and weak points of the proposed methodology and about goals of future research.

In my opinion, the work is interesting and relevant for the journal “ISPRS”, International Journal of Geo-Information. However, several changes can improve the quality of the work. Then, I recommend to reconsider after major revision.

Author Response

Thank you for your comments. We have tried to address all of them.

Author Response File: Author Response.docx

Reviewer 2 Report

General comments:

The paper „Multi-Temporal Sentinel-1 SAR and Sentinel-2 MSI for Flood Mapping and Damage Assessment in Mozambique” presents a flood mapping method, which might be very useful for flood management, especially during extreme events. The proposed methodology employs Google Earth Engine for satellite data and computation, consequently the limitations from onsite available computational resources are significantly reduced.

The introduction and literature chapters are well written and provide adequate background for analyzed problem, regional perspective and methodical challenges. I could not find the statement of paper purpose or aim.

Other chapters lack the comprehension. Used methods and why they were chosen are not well explained and need to be improved. For example: why VH polarization was used; how pan-sharpening was done; are date sets balanced; what number of points were inundated in different LC classes? It is not entirely clear how authors validated LC Classification. 

Have you written all scripts by yourself or you have used scripts provided by other users/GEE? It would be great if you could include the link to the GEE with your script.

I am not convinced that 5.3 chapter presents Damage assessment. The title is to ambitious. The inundated area of LC classes and changes of this area are described in this chapter.

To improve the manuscript, authors should expand the clarification on used methodology and more thoroughly present the results, it would greatly increase the confidence in your proposed method and results of the manuscript.

Specific comments:

L91 “…in near real-time and LC damage assessment in Mozambique?” I am not shure that land cover damage is an applicable expression.

Figure 1. It would be much easier to understand the study area if the DEM would be presented in the manuscript?

L229 It would be much easier to understand the processing pipeline if figure 4 would be presented near the beginning of this section?

L265 “…we explored the base map in GEE during the collection process rather than solely using S2 imagery”. Please explain what do you mean and how it affects the results?

Figure 4. Would recommend to switch pre and post flood sections, because it would be more intuitive for readers (left to right).

L302 “…we repeat this process for bands B5, B6 and B7” Which process do you mean?

Figure captions are very vague for most of the figures.

Figure 9. Avoid “This image shows” or similar expressions in figure captions as well as in the manuscript text.

Author Response

Thank you for your comments. We have tried to address all of them.

Author Response File: Author Response.docx

Reviewer 3 Report

Table 1.

- SAR data should be called “Remote Sensing Data” (or similar) in order to include optical and SAR data.

- Unify the name “Sentinel-1” with “S1” ¿is it the same?

- I suggest clarifying the current status of Sentinel-1B

- The Figure 3.  It does not clearly show the information of the Copernicus service

- Figure 6, 7, 8 y 9 need expansion for further clarification

Author Response

Thank you for your comments. We have tried to address all of them.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

Authors answered to all the points of the review. Sections about methodolodogy, results and conclusions have been conveniently improved. Please correct the oversight at  line 238 page 9: "Otu’s thresholding algorithm".

I suggest a spell check of English language and style.

Author Response

Thank you for the helpful comments and suggestions. We have tried to revise the paper accordingly.

Author Response File: Author Response.pdf

Reviewer 2 Report

After revision the manuscript is improved. I still have some comments and recommendations:

Point 2: I am not convinced that 5.3 chapter presents Damage assessment. The title is to ambitious. The inundated area of LC classes and changes of this area are described in this chapter.

Response 2: Here we just followed how some other manuscript treat/call it. By finding inundeted area, mainly in Agriculture it is expected to be a damaged area

Remark 2: It is better to name as it is. For example Land cover classes in inudated area

 Figure 1. It would be much easier to understand the study area if the DEM would be presented in the manuscript?

Response : Revised

Remark. Please provide scale for height and use different palette, preferably discreet, for semiology. The palette does not have to be linear.

In general, try to explain and if possible provide the revisions you made in the response to the reviewers.

Author Response

We would like to thank you for such helpful comments and suggestions. Please, we send our  comments as attachment. 

Author Response File: Author Response.pdf