An Approach of Identifying and Extracting Urban Commercial Areas Using the Nighttime Lights Satellite Imagery
Round 1
Reviewer 1 Report
Although the article provides an interesting methodological contribution, it would be interesting to provide a better explanation of its applicability for different disciplines (economy, urban planning, sociology etc.).
The elements explaining the aims and objectives of the conducted research, as well as the description of local context (i.e. the selected cases/cities and the analyzed activity), would contribute to the better understanding of the provided data, especially for international readers.
Author Response
Please see the attachment.
Author Response File: 
Author Response.docx
Reviewer 2 Report
The manuscript introduces an automated way of detecting urban commercial areas from a novel nighttimelight sensor. The manuscript is interesting and worth publishing.
In general, I think language editing would significantly increase the readability. It seems that the authors did include parts of subjective discussions in the results section, despite having a dedicated discussions section. The discussion needs to be improved and the authors should critically reflect their on work. That needs either needs to be separated or combined, not both. The figures display a mix of font families and sizes, detail and replications. Effort needs to be put into generating a consistent impression to achieve a professional look of the manuscript.
In detail, I have some remarks that need to be addressed. The location in the manuscript is indicated by line numbers.
21-22: "we believe" does not indicate confidence in the results.
23-24: The conclusion in the abstract is very generic. Provide more specific concluding paragraph in the abstract to raise interest in the reader.
29-36: The first paragraph of the introduction lacks to outline the objective of the work presented. However, I think it is necessary to understand the rest of the introduction. The transition to the second paragraph has to be improved.
38: Here, and in the introduction in general, [2] is cited heavily. It is an unpublished thesis, though. Include more references to increase scientific foundation and acknowledge the work in the field.
53: "and so on" appears very colloquial.
63: The social networks are not mentioned before, only VGI. Yet they seem to be used synonymously.
65-66: There is no logical transition between the two paragraphs. It is difficult to follow the narrative.
85: "we take four good ones" seems very subjective. Please elaborate how the decision was made.
119-122: This paragraph is very difficult to follow. Also in line 121 "useless" should be rephrased. It seems colloquial, subjective, and not thought out.
Figure 4 is not acceptable for publication. Split it into a table, if the information in the left needs to be included. The plot on the right is not readable at all.
249: The green coverage comes out of nowhere. Was it mentioned before, what purpose is it necessary for?
Table 1: does this serve as the reference? What do you compare the results of your method to in order to assess its reliability?
271-284: Seems to be written from locals who know the area well. An international, non-local, audience of the manuscript could appreciate more abstract descriptions.
Figure 7: What is the point of this figure? It seems that the only difference are the indices I, z, and p. This should better be a table.
297-299: This is an example where subjective, discussion-like evaluation takes place in the results section. Should be moved to discussion.
Table 2: How do you evaluate the error of commission?
313: Why discussion if there is a dedicated discussion chapter?
335-341 Is highly evaluative and therefore is not suited for a results section.
359 "we can conclude" indicates that this should be moved to conclusions section.
365-367: I have a hard time understanding this sentence.
384-385 The part in quotes appears for the first time. Quotes can indicated no well thought out text. If it is a direct citation indicate accordingly.
388: "skillfully" is a very subjective evaluation.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 3 Report
The paper proposes an automatic detection of urban commercial areas exploiting nighttime satellite imagery analyzed by means of exploratory spatial statistics.
The satellite platform considered (Luojia) is highly interesting due to its recent launch and the few pieces of research currently available in the literature exploiting its imagery. This type of analysis is of general interest and it is promising to investigate patterns of human activities in highly urbanized areas.
English is good, paragraphing is fine.
However, the paper is poorly technically sound due to the very strong hypothesis which is behind the analysis that is the clustering of nighttime lighting is straightly correlated to the commercial areas only. This fact is also discussed by the authors which identify biases due to other land-use features, such as transport hubs, that affect the detection accuracy of commercial areas. I believe that additional investigations are required in order to state that the method is innovative as well as that it can substitute traditional methods for commercial area identification. This because the use of ancillary data on commercial areas location seems mandatory in order to distinguish among clusters that actually point out the target land-use. The above emerges also during the proposed detection accuracy assessment which lacks correlation analysis e.g. between cluster intensities and false/true positives.
The experiments are conducted using well-known techniques which are fully mathematically described in the text whereas sensitivity analysis and input parameters are not well mentioned. For instance, which is the spatial weighting schema adopted to compute the Moran’s I? Which is the significance level that you used as a threshold to distinguish between clusters and random patterns? These analysis choices and must be stated in the text to allow replicability. Please, mention also the software used at each step of the analysis as well as information about data (Luojia imagery) licensing and accessibility.
Finally, the figures are not appropriated. In details:
- Figures 2, 5, 6, 8: Background image canvas has the same colour (white) of pixels considered in the analysis making hard their distinction.
- Figures 5,6,8: Brown clusters are really difficult to be seen from the black background pixels. Please, use a more proper colour schema to allow readers understanding the maps.
- Figures 3, 7: The schematics of Moran’s I interpretation are clearly copied by the ArcGIS documentation. You must cite the source of not original figures or better you should provide a more compact and original description of the spatial association analysis results you obtained.
To sum up, the paper requires a very extensive revision to be reconsidered for publication.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 4 Report
In the introduction, it would be necessary to specify in more detail the aim and benefits of the paper. It would also be desirable to provide a more detailed description of the baseline situation and to highlight what the new contribution will bring. It would be appropriate to include also the main hypotheses.
The notion of „urban commercial areas“ is not quite clear. It would be desirable to specify what all falls into this category.
The method is very interesting but from my point of view there are many "uncertainty" factors.
Discussion is not related to results of the study in most cases and is too general. Discussion do not present relevant major findings. It is a brief summary of the authors' results and experiences. It does not provide comparisons with similar research. In the discussion it would be necessary to compare the results and the approaches with the results of other authors. It would be desirable to evaluate what new this paper brings, what are the positives, what risks, etc. This part has not a character of a discussion, it would have to be revised.
Conclusion are not convincing and convey no new information. The conclusion would need to be expanded, it would be appropriate to indicate how the results could be used both in science and in practice. It would also be necessary to specify in more detail the research questions and problems that need to be further investigated. The importance of research for practice should be added.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
The latest version of the article provides sufficient contextual clarification and does not need further upgrading.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 2 Report
I acknowledge the work the authors put into some of my previous comments. However, some were not addressed satisfactory.
My general comment concerning the inconsistency in figure design was not addressed at all. "The figures display a mix of font families and sizes, detail and replications. Effort needs to be put into generating a consistent impression to achieve a professional look of the manuscript"
Figure 4.: I do not agree with the argumentation of the authors.The reasoning for the choice of design of this figure is that the used software generated it this way. In my opinion, that is unacceptable. If the content of the figure is obscured by design, it is best to leave it out completely.
New Figure 7: you should add the credits to the base map.
Former Figure 7 (Morans I, now Figure 8): This figure is pointless. It provides no new information to figure 3 except the values for Moran's I, z-score and p-value. Put the values in a table and you do not distract the reader with this figure. Thus, you help the reader to identify the important information at first glance. Later in 4.1 you refer to only the values anyways.
Comment 16: "Table 2: How do you evaluate the error of ommision?" addressed whether or not the method is able to assess overclassification errors. Could you please explain that?
Figure 9 and all alike are not readable at all. Neither on screen nor printed. Please rethink the figure design and font size of the legend.
Comment 20. The added sentence does not add any value. Please split up the sentence as it is hard to understand what you mean.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 3 Report
The paper quality has improved in its general form. However, the explanation provided for the definition of parameters in the statistical analysis highlights a naive usage of those tools by lacking solid motivations as well as sensitivity analysis that justify the choices (need for extending motivation on the use of few permutations for the CSR inference, the spatial weighting schema and its possible alternatives or effects on the results, a possible quantitative or semi-quantitative description of the false positive detections, etc.).
I believe that the paper would better fit a lower impact factor journal.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 4 Report
Many of my comments have been accepted. The discussion and conclusions could be richer. Comparisons could also be made with authors from other countries. But in principle, I agree with the publication of paper. Best regards ZitaAuthor Response
Please see the attachment.
Author Response File: Author Response.docx
