Assessing Street Space Quality Using Street View Imagery and Function-Driven Method: The Case of Xiamen, China
Round 1
Reviewer 1 Report
Assessing Street Space Quality Using Street View Imagery and Function-Driven Method: The Case of Xiamen, China
The research conducted assesses street space using three types of data: images from street view, point of interest, and textual data from social media, identifying the relationship between these three aspects and the street vitality and availability of service facilities.
The research is quite interesting, but I believe some important points are not being correctly nor clearly explained. It is worth proofreading the manuscript by a qualified English native/ specialist since it has several (and some of them critical) grammatical mistakes and typos throughout (abstract included).
With respect to the structure of the manuscript, I think a literature review/ previous research section should be included… Indeed, the introduction serves to outline the objective of the research and explain, in a general sense, the approach adopted in relation to existing/previous studies… but a dedicated literature section would cover in more detail previous research. My suggestion is to split the introduction and maybe introduce more references of multi-source studies that have addressed the urban environment similarly.
Moreover, I don´t think that the method should be outlined in the introduction section (paragraph 4, from the fourth line on)- please avoid repetition with the actual method section.
The contribution of the study is, to my opinion, correctly expressed and presented, assessing the vitality of the street space - a concept that includes social and physical issues, rather than only the physical aspects of the urban environment. Also, although it is not novel, the methodological approach adopted that incorporates multiple source types for assessing physical and perceptual aspects of the urban environment is quite interesting.
In section 2.1., in the first paragraph it is mentioned that “The data that this study uses for presenting and analyzing its hypotheses” … it would, therefore, be necessary to explicitly cite the hypotheses in a previous section (introduction or after the literature review).
In section 2.2.1., it would be recommended to explain a bit more clearly the process of pre-processing street view images for the broader audience. Also, cite previous work that has conducted a similar process for pre-processing images… there is a lot of work done in that direction.
In section 2.2.2. what do you mean when you say that “These data were provided in two different files”? what kind of files? What do you mean by “redundant information”? could you provide an example? How many records were deleted and why? Was the process of selecting/deleting data manually performed? I think it is worth explaining a bit more the procedure. Were these data somehow paired to the review/comment data?
In section 2.2.3 here you mention again the POIs, but instead, you are referring to the review/comment data, right? So, the POIs, in this case, is limited to “attractions, hotels, restaurants and shopping malls”. In this same section, I got lost when you mention “the buffer radius” to obtain data. Please explain more comprehensively the data collection and pre-processing.
In section 3, the fifth line, “street space environment indexes” are mentioned … what are those? – are you referring to the Individual element index that is explained later? Please clarify.
I find the method to be of great interest, and I would like to encourage the authors to reformulate the writing of the method section. Indeed, the writing and organization of sections 2 and 3 are worth revising entirely. I think that the pre-processing of data is part of the method itself since it is necessary to do it if other researchers aim to reproduce it. Therefore, maybe the whole pre-processing explanation should be moved from Section 2 to Section 3. Please make sure that the audience can follow up on the process and that the research can be reproducible.
Table 2 is a bit confusing with the G, C, and —. Wouldn’t a numerical weighting value be a bit more effective? At least for explaining the weight of the indicators. Also, the “density of scenic land” indicator… what does it refer to?
The results section (Section 4) appears to be better organized and structured. Easier to read and to follow than the two previous sections. Nevertheless, the content of it needs to be revised so that it reports exclusively the results obtained. The methodological aspects, such as using the Kernel Density Analysis tool, or the fact that experts (please explain the experts’ profile) were scoring results, should be included in the method section.
In the discussion section, it would be suggested to develop a bit further how the results compare or can be contrasted to findings from previous research and how the method proposed has truly advanced the knowledge in the field.
The limitations and future prospects section is a rather interesting reading. Again, the wording could be improved by proofreading the text.
Author Response
Please see the attachment.Author Response File: Author Response.docx
Reviewer 2 Report
This interesting research suggests a novel methodology for evaluating street space quality. The paper proposed using street view imagery (SVI), points of interest (POI), and crowdsourced data to understand the street vitality, level access to facilities and built environment characteristics. The paper employs image processing combined with spatial analysis to determine different urban functions and the quality of street space in three different Areas of Xiamen city, China. Eventually, the study validates the spatial analysis results with people’s perception through AHP analysis. While the paper provides an interesting methodology, it needs to undertake a major revision due to several limitations before being ready to publish in this journal.
The issues are itemised as below:
- Limited engagement with the literature: this paper lacks a comprehensive literature review and analysis. While the authors provided a literature review in the introduction part, I believe that the paper can be improved with further engagement with the literature about spatial analysis of urban/street space analysis quality. The literature review will help to justify selecting the urban function variables and validation criteria.
- Justification for selected variables: the chosen variables are not well explained and justified through literature review or practical reports. For example, what are the reasons for selecting scenic land? What does it mean? What are the determinant factors?
- Justification for methods used and potential flaws in selected variables: the authors need to explain further methods used to analyse variables. For example, it is unclear why the walkability and motorization indices use the same variables (sidewalk and terrain)? This obviously creates a strong correlation result in the analysis. In addition, what is the range of values expected from these indices? How to interpret them? Furthermore, there is no explanation about the AHP method for the validation of results in the methodology. Why is this method adopted? How many respondents were engaged? Who are those respondents? Residents? Tourists? How was the data collected? The spatial distribution analysis (kernel density) method was not explained and appropriately justified. Why is this method used? How about spatial correlation analysis methods such as Moran’s I or Getis Ord, which could be run globally and locally. What about the comment data? What are those comments? The authors should provide examples, and explain how they filtered, cleaned and used in the analysis. I have more comments about the methodology that can be found in the in-text feedback.
- The facts and figures to explain the study area: most of the values throughout the manuscript are provided in a general term, and it is hard to understand the magnitude of variables. For instance, in page 12, lines 360-361: “Amoy YatSen Rd is one of Xiamen’s prosperous commercial streets, with a large population flow.” There should be facts and figures to understand prosperity and population flow. I have indicated other examples inside the text; please check my in-text comments.
- The discussion and conclusion sections need to be improved and extended. The authors should compare their results with those of others’ findings in the discussion section. It is also essential to explain the implications of the results. The conclusion section needs to be improved by explaining the contribution of this research in the theory and practice
- There are several typos throughout the text. For example, see page 15, line 420: replace “tapes” with “types”.
As mentioned above, my further detailed comments can be found in the text, which is attached.
Comments for author File: Comments.docx
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 3 Report
I find the submission interesting and very socially relevant. However, I have a few comments which are described below, and refer mainly to clarity issues:
- Please write a paragraph explaining the process exhibited in figure 2.
- Are the images in figure 3 originals or cropped? Maybe you should show one or two images before and after the preprocessing.
- Are the POI data referred in 2.2.3 a subset of those referred in 2.2.2?
- The procedure described in 2.2.3 in not clear, especially the buffer part. Please rewrite.
- In line 150 what to you mean by "and so on"? Is there any other type of data?
- The description of figure 4 provided from line 158 to line 169 is not clear. Readers not familiar with your methods (such as space environment indices or entropy TOPSIS) will not understand your method.
- You seem to use the term index (plural is indices) and indicator interchangeably. These are different concepts. Please clarify and use the same term throughout the article.
- In figure 4, do all the data sources lead to all the indices? If not, please make the figure more explicit.
- Please provide a reference for the DeepLabv3.
- The explanation you give for the interface heterozygosity is not clear. What do you mean by a restriction of buildings, etc?
- Same comment for motorisation.
- Exactly how do you match the street view collection points with the street segments? What is the distance between each image on a street?
- in line 230, the maximum distance of effective influence: what is this maximum distance referring to? maximum distance to what? effective influence on what?
- In line 260: what is the particularity of Area II?
- How do you obtain the scores mentioned in lines 254-5? For example for scenic land?
- Since I could not understand how you compute the scores I could not follow your reasoning in the last paragraph of 3.3 (lines 263-269).
- When describing figure 6 you mention several toponyms (e.g. Huaqiao university or Gulangyu island) which have no meaning for the reader not acquainted with the area, unless you place those toponyms on the maps. Otherwise just use west, east, top, bottom, etc
- Same comments for figure 12. E.g Where is Amoy YatSen Rd, Railway station?
- In figure 14 sometimes it is quite difficult to distinguish the colours of the different levels of street quality. Please try to find a better color scheme and maybe also make the lines thicker.
- In 4.3 it would be important to provide some statistic showing for each area the amount of each quality level (e.g. length of road) for each urban function. Maybe a table (matrix) would help. Otherwise it is very difficult to interpret the maps.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 4 Report
The article is made on the current topic of the development of public spaces of city streets is well structured and meets the requirements of IMRAD
The presented assessment system can be used to assess the land use of urban streets. Its strong point is the use of GIS technology to obtain reasonable results.
As a comment, I want to note the lack of information on the selected classes of streets. The article presents only the rationale for the allocation of classes, but not their description. You should add a few paragraphs and maybe pictures that will show the content of each class of streets. This will improve the perception of the results of the study.
The article is recommended for publication
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 5 Report
I received for review a very interesting and useful study with an international dimension. The research was logically thought out. The proof was done correctly, but the presentation of the data is not entirely clear. I have three main comments.
1) Was preprocessing of street view data done comprehensively? Or maybe each image was analyzed individually and as soon as discrepancies were identified its parameters were subjected to the process of unification?
2) Results of spatial analyses presented in Fig.14 are poorly legible. Changing the gradation of green color saturation does not give a clear presentation of different functionalities of space, which merges with the color of water on the orthophoto. I propose to change e.g. to a shade of purple, which stands out on the orthophotomap. The same applies to contours of street quality in shades of red.
3) Please better describe the select group of experts engaged to validate the data. What was the representation of this group? How was the expert survey conducted?
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Round 2
Reviewer 2 Report
Dear Authors,
Thank you for submitting the improved manuscript, which addressed many issues. However, there are comments that have not been addressed thoroughly. For instance, I am still not convinced by your response to my comment on the justification of using the same variables for walkability and motorization indices. Adding the number of pixels belonging to the person does not address the collinearity issue. Furthermore, in using the AHP method, I could not understand how a pairwise comparison was conducted. Table 5 shows the evaluation criteria and rankings, but a further explanation of the hierarchy of criteria and sub-criteria, the consistency index, and sensitivity analysis should be provided in the methodology section. In point 21, please explain why the infrastructures and services are robust? What determines the robustness? The number of incidents? The extent of coverage? Service satisfaction by users?
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
I don't see the changes with toponyms that you claimed you made to figures 6, 11 and 13.
Except for figure 3, all figures have lost quality and are now difficult to read.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 3
Reviewer 2 Report
Thanks to the author(s) for clarifying the points raised and improving the manuscript. I believe that the paper is ready to be published in IJGI