Extracting Disaster-Related Location Information through Social Media to Assist Remote Sensing for Disaster Analysis: The Case of the Flood Disaster in the Yangtze River Basin in China in 2020

Round 1

Reviewer 1 Report

Although social media data are offering a wealth of information to study different geographical phenomena, study combining social media data with remote sensing for flood examination is few and far between. Thus, the study could be useful to advance hazards, risk and vulnerability literature however there are many issues in this present version of the work. Hence it requires one round of revision. It is also important to note that fail to address the issues (outline below) may ruin the applicability and practicability of this work.

[1] Abstract: Line 18: exemplify some flaws. I don’t think you are correct by saying ‘lack of’, I can show you thousands of works related to what you are trying to show here. Line 23: delete ‘the’ before information

[2] Introduction section does not reflect what is being aimed to find in this work. Thus, international significance is lacking. Though various technologies are being discussed, they are always not referred with appropriate texts. An example is smart devices info in line 47 requires attribution (https://doi.org/10.1080/17538947.2013.808277). Study related to combining social media with remote sensing data to study disaster or flooding require reviewing, some examples are https://doi.org/10.1016/j.compenvurbsys.2017.06.004; https://doi.org/10.1080/13658816.2017.1367003. You are studying flood (https://doi.org/10.1080/15230406.2016.1271356) that’s fine but other disaster (https://doi.org/10.1109/JPROC.2017.2729890) studies also used social media and remotely sensed data, they require your attention as well. English of this section [introduction] made my life difficult, hence needing proofing by a native speaker

[3] Section 2.2.1: English must be corrected here, you used Sentinel-1 SAR which is OK. But writing here is poor so clarity is needed. Should be ‘Methods’ only. Should be ‘Locational’ in line 148. The biggest issue with social media data is that they are highly unstructured, I wonder how this was solved in this work. Should be “Part of speech selection and word set construction’ [sub-section 3.1.2. What do you mean by ‘Geographical location word filtering’? Is ‘geographical’ different from ‘location’? Line 238: should be ‘locational’. Fig 4: did you construct this? If so, the caption should be’ The structure of reconstructed social network’. Change Fig 5 caption to ‘Process of flooded area extraction in this work using remotely sensed data’. Line 293-295: these statements are not correct, there are many methods these days including machine learning, AI etc. Hence these statements need revising. What type of filtering is used? In the processing and extraction of flooded area from SAR data, authors seems to have missed something. Hence section 3.3 needs further improvement. I think in-text cites 34, 35 are not entirely true since multispectral data was used with maximum likelihood classification. This work (https://doi.org/10.3178/jjshwr.19.44) could be highly useful as it is used SAR with the same algorithm, hence could assist authors to reframe texts related to remotely sensed data utilisation for flood mapping in this work. Also, SAR presents some unique problem to effectively extract flooding in diverse environment. What did you do with comprehensive analysis? What overlay technique was used, be specific in line 334. What is TF-IDF? Define and clarify for readers.

[4] I would be inclined to use works from other areas to demonstrate similarities and dissimilarities of this study with theory, thus add discussion section. Above works could be useful as well. Conclusion does not reflect your aims and missed linked with the findings. Delete ‘analysis’ from section heading 4.

[5] English of the paper needs significant improvement. Title of the work needs tweaking to reflect content of this work

Author Response

Thank you very much for your valuable comments. We have carefully considered the relevant comments and made the following modifications:

[1] Abstract: Line 18: exemplify some flaws. I don’t think you are correct by saying ‘lack of’, I can show you thousands of works related to what you are trying to show here. Line 23: delete ‘the’ before information

Response 1: The expression of content here is really poor, and we have revised it. The expression ‘lack of’ changed to “insufficient ”. We removed the corresponding word “the” before the word “information on Line 23.

[2]Introduction section does not reflect what is being aimed to find in this work. Thus, international significance is lacking. Though various technologies are being discussed, they are always not referred with appropriate texts. An example is smart devices info in line 47 requires attribution (https://doi.org/10.1080/17538947.2013.808277). Study related to combining social media with remote sensing data to study disaster or flooding require reviewing, some examples are https://doi.org/10.1016/j.compenvurbsys.2017.06.004; https://doi.org/10.1080/13658816.2017.1367003. You are studying flood (https://doi.org/10.1080/15230406.2016.1271356) that’s fine but other disaster (https://doi.org/10.1109/JPROC.2017.2729890) studies also used social media and remotely sensed data, they require your attention as well. English of this section [introduction] made my life difficult, hence needing proofing by a native speaker

Response 2: Based on your valuable comments, we have made the following modifications: 1) We revised the introduction and add the work objective of this paper. In view of the shortcomings of insufficient location information of social media data, we extracted disaster related location information from the text of social media data, and constructed a new social network together with the upload location tag of social media data. This network can be efficiently combined with remote sensing images to improve the efficiency of disaster analysis. 2)We also adjusted the structure of the introduction. After we put forward the problems to be solved in this paper, we discuss the corresponding solutions in sections, including extracting the disaster related location information contained in social media data, the construction of new social networks, and how to efficiently combine with remote sensing data to assist disaster analysis. 3) Thank you for providing many valuable references. We have effectively quoted them and optimized the references in the original text. This is very helpful for readers to better understand the content of the article.

[3]Section 2.2.1: English must be corrected here, you used Sentinel-1 SAR which is OK. But writing here is poor so clarity is needed. Should be ‘Methods’ only. Should be ‘Locational’ in line 148. The biggest issue with social media data is that they are highly unstructured, I wonder how this was solved in this work. Should be “Part of speech selection and word set construction’ [sub-section 3.1.2. What do you mean by ‘Geographical location word filtering’? Is ‘geographical’ different from ‘location’? Line 238: should be ‘locational’. Fig 4: did you construct this? If so, the caption should be’ The structure of reconstructed social network’. Change Fig 5 caption to ‘Process of flooded area extraction in this work using remotely sensed data’. Line 293-295: these statements are not correct, there are many methods these days including machine learning, AI etc. Hence these statements need revising. What type of filtering is used? In the processing and extraction of flooded area from SAR data, authors seems to have missed something. Hence section 3.3 needs further improvement. I think in-text cites 34, 35 are not entirely true since multispectral data was used with maximum likelihood classification. This work (https://doi.org/10.3178/jjshwr.19.44) could be highly useful as it is used SAR with the same algorithm, hence could assist authors to reframe texts related to remotely sensed data utilisation for flood mapping in this work. Also, SAR presents some unique problem to effectively extract flooding in diverse environment. What did you do with comprehensive analysis? What overlay technique was used, be specific in line 334. What is TF-IDF? Define and clarify for readers.

Response 3: According to your valuable suggestions, we have revised the corresponding statements in turn. Among them, 1) We changed the title of section 3 to "Methods". There is no difference between " geographical" and "location". We really don't express it well here. We collectively express it as "location information". 2) Social media is a kind of unstructured data, and we use the method of web data crawler to obtain this data. In Section 2.2.1, we just did web page parsing, so that fields contained in social media such as time, upload location tags, and text content are stored in a structured way. We describe this in our revised paper. 3) sub-section 3.1.2. The expression ‘Geographical location word filtering’ is not clear. We have changed it to “Recalling the locational words”. This section actually describes some special situations and describes how we deal with them. 4) In this article, we build a new social network. The suggestions you provided are very good. We have modified the corresponding text content. 5)We have revised the description of extracting flooded areas based on SAR images(section 3.3) to make the expression of the content more correct. The relevant extraction strategies mainly include two forms. Some methods such as machine learning, AI are used in these strategies. In addition, we have made improvements to the flow chart and overall method description. 6) We have revised and optimized the corresponding references .7) We revised Section 3.4.1. We mainly use the method of GIS-based spatial overlay analysis to help assess disaster in different areas. We describe how the analysis results are used to analyze and evaluate the disaster situation. 8) We have added a description of the "TF-IDF" method. 9)Some other suggestions, such as the title of section and figure have been modified accordingly.

[4]I would be inclined to use works from other areas to demonstrate similarities and dissimilarities of this study with theory, thus add discussion section. Above works could be useful as well. Conclusion does not reflect your aims and missed linked with the findings. Delete ‘analysis’ from section heading 4.

Response 4: Your suggestion is very nice. We add a summary and theoretical analysis in Section 4, and cite references to indicate the thesis of this paper. The title of section 4 is revised as " Results ". We also revised section 5 , added the findings of this paper, and improved the text expression at the same time.

[5]English of the paper needs significant improvement. Title of the work needs tweaking to reflect content of this work

Response 5:We invited professionals to help improve the English expression of this article. And we also changed the title to better fit the content of the paper.

Thank you again for reviewing this paper and for your valuable comments.

Author Response File: Author Response.docx

Reviewer 2 Report

The paper is well-written and addresses an interesting problem. It can be published after some minor concerns.

1. The title should be revised in the second part particularly. Maybe something like “the case of flood disaster in Yangtze River Basin in China in 2020.”
2. For the several first sentences of the introduction, please use appropriate references.
3. In figure 1, it is not clear the location of these cities. Which part of China?
4. Section 3.3, “Flood area extraction” is a bit confusing. This section should be revised. Also, please revise figure 5.
5. There are some grammatical errors. Please revise the paper to remove these problems.

Author Response

The title should be revised in the second part particularly. Maybe something like “the

Thank you very much for your recognition of the article. And your valuable comments are very helpful to the article. We have carefully considered the relevant comments and made the following modifications:

[1]The title should be revised in the second part particularly. Maybe something like “the case of flood disaster in Yangtze River Basin in China in 2020.”

Response 1: We have revised the title to better fit the content of the article. The title is “Extracting disaster-related location information through social media to assist remote sensing for disaster analysis: the case of the flood disaster in the Yangtze River Basin in China in 2020”

[2]For the several first sentences of the introduction, please use appropriate references.

Response 2:In the introduction, we have added and optimized some references, which are helpful for readers to better understand the content of the article. At the same time, we have adjusted the narrative structure of the introduction to make it more reasonable.

[3]In figure 1, it is not clear the location of these cities. Which part of China?

Response 3:We have modified Figure 1 and added some descriptions. This is useful for interpreting the orientation of the study area in China.

[4]Section 3.3, “Flood area extraction” is a bit confusing. This section should be revised. Also, please revise figure 5.

Response 4: We described the flow of the method in more detail and updated the flow chart.  

[5]There are some grammatical errors. Please revise the paper to remove these problems

Response 5:We invited professionals to help improve the English expression of this article.

Thank you again for reviewing this article and for your valuable comments.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

The revision is a much-improved version relative to the original submission. I am generally happy with the revision however there are a few issues that must be clarified and explained to improve its readership.

[1] Line 440, 522: must define what sort of overlay analysis was done here since there are many overlay methods in GIS

[2] Line 411: the citation is shown in other language in reference section and full detail is not captured, I recommend to use full reference (https://www.jstage.jst.go.jp/article/jjshwr/19/1/19_1_44/_pdf/-char/ja), if possible use DOI https://doi.org/10.3178/jjshwr.19.44

[3] Line 565: delete ‘the’ before ‘….disaster-related…’

[4] Line 613: delete ‘the’ before ‘…temporal..’

Author Response

Thank you very much for your recognition. The comments you gave are very detailed and valuable. They are very useful for perfecting this article.

[1] Line 440, 522: must define what sort of overlay analysis was done here since there are many overlay methods in GIS

Response 1: There is indeed a problem with my formulation here. In fact, I just superimposed the multi-source data on one space. It is incorrect to use "the GIS-based spatial overlay analysis" to state this. A little more rigor is needed here, and we've revised that.

[2] Line 411: the citation is shown in other language in reference section and full detail is not captured, I recommend to use full reference (https://www.jstage.jst.go.jp/article/jjshwr/19/1/19_1_44/_pdf/-char/ja), if possible use DOI https://doi.org/10.3178/jjshwr.19.44

Response 2: Your comment is very nice, thanks for providing the corresponding references. We also noticed the references for this citation before, but didn't think about it too much. We have revised this.

[3] Line 565: delete ‘the’ before ‘….disaster-related…’

Response 3: We have revised this.

[4] Line 613: delete ‘the’ before ‘…temporal..’

Response 4: We have revised this.

Thank you again for your hard work in revising this article. Your comments and suggestions will be very helpful to our future work.

Author Response File: Author Response.docx