Temporal Autocorrelation of Sentinel-1 SAR Imagery for Detecting Settlement Expansion
Round 1
Reviewer 1 Report
Comments for author File: 
Comments.pdf
Author Response
We would like to thank the anonymous reviewers for their valuable contributions to the quality of this paper. Please find below (and attached) our response to the reviewer comments, with details of how the suggestions were incorporated into the text shown in blue.
-------
Point 1.1: The authors should explain the limitations of related studies. What is specific about Sentinel-1 SAR imagery-based detection of settlement expansion compared to others such as MODIS imagery-based detection of settlement expansion
Response 1.1: Thank you for this suggestion. In the introduction, we have expanded on the contribution of SAR-based settlement detection relative to other approaches. We have more clearly pointed out the limitations of optical approaches, as well as some other supervised approaches.
Point 1.2: The authors should clearly define the problem. They should discuss why methods (such as radiometric indices) have limitations in monitoring urban expansion.
Response 1.2: Please see response 1.1, which addresses this issue as well.
Point 1.3: It is suggested to include value based results in the conclusion to explain how the proposed ACF-based approach is efficient in detecting and mapping urban expansion.
Response 1.3: Thank you. We have amended the conclusion section to comment more pertinently on the value offered by ACF-based settlement detection, with specific reference to efficiency.
Author Response File: Author Response.pdf
Reviewer 2 Report
The authors have presented a framework for detecting the settlement expansion via. Temporal autocorrelation applied on the Sentinel-1 SAR images. I have read the paper and I found some serious limitations within the paper, which need to be fixed before publication.
- The author needs to include a generalized framework in the shape of a flow diagram that presents the commonly used steps/sub-methods require/used by the authors in the literature.
- The author needs to analyze and present the works performed by the researcher in the literature in a more systematic way. The author needs to include a table summarizing the techniques used in multiple papers for each stage such as segmentation, feature extraction, feature reduction, classification, and their reported performance in the shape of a separate table for each stage. Moreover, the author should include more methods in these tables.
- The deep neural and CNN-based methods should also be included from the literature in the table section and also add the achieved performance by those methods.
- Add the benefit and limitations of each method used in the table section.
- The literature section is extremely weak, it lacks the work reported by the researcher in the last two or three years.
- The paper does not present any DNN method reported by the researcher for a similar study.
- The authors have not included any superiority in terms of accuracy/performance over the previously reported works.
- The authors have presented a very simple methodology to detect the settlement expansion over time in a specified geography and didn’t include the comparison table.
Author Response
We would like to thank the anonymous reviewers for their valuable contributions to the quality of this paper. Please find below (and attached) our response to the reviewer comments, with details of how the suggestions were incorporated into the text shown in blue.
--------
Point 2.1: The author needs to include a generalized framework in the shape of a flow diagram that presents the commonly used steps/sub-methods require/used by the authors in the literature.
Response 2.1: Thank you for the suggestion. Given the limited space available, we have included in the introduction, in narrative form, a mention of typical change detection approaches and methods used in literature.
Point 2.2: The author needs to analyze and present the works performed by the researcher in the literature in a more systematic way. The author needs to include a table summarizing the techniques used in multiple papers for each stage such as segmentation, feature extraction, feature reduction, classification, and their reported performance in the shape of a separate table for each stage. Moreover, the author should include more methods in these tables.
Response 2.2: Given that this is not a review paper, we feel that a systematic review of the field with this level of detail falls somewhat beyond the ambit of the paper. However, we agree that that a stronger review of literature and existing change detection approaches is required. We have included a more in-depth literature review of existing methods in the introduction section. We trust that this sufficiently addresses the concern.
Point 2.3: The deep neural and CNN-based methods should also be included from the literature in the table section and also add the achieved performance by those methods.
Response 2.3: Thank you - we agree with this suggestion. While not included in tabular form, we have included in our expanded literature review a wider review of deep learning and CNN-based change detection, including typical accuracies achieved.
Point 2.4: Add the benefit and limitations of each method used in the table section.
Response 2.4: While not in tabular form, we have made an effort to evaluate other change detection methods. This is found in the expanded literature review in the introduction section, as well as in the discussion section (Section 4.3).
Point 2.5: The literature section is extremely weak, it lacks the work reported by the researcher in the last two or three years.
Response 2.5: We have addressed this issue by expanding the literature review in the introduction section.
Point 2.6: The paper does not present any DNN method reported by the researcher for a similar study.
Response 2.6: We have addressed this issue by expanding the literature review in the introduction section, as mentioned in Response 2.3.
Point 2.7: The authors have not included any superiority in terms of accuracy/performance over the previously reported works.
Response 2.7: Section 4.3 of the manuscript addresses the relative performance of the ACF-based method. We have added additional comparisons with literature to strengthen this section. Please note that the original article did not claim superiority in terms of accuracy.
Point 2.8: The authors have presented a very simple methodology to detect the settlement expansion over time in a specified geography and didn’t include the comparison table.
Response 2.8: We present this methodology as an alternative, unsupervised change detection approach which has simplicity as a relative strength, compared to supervised deep learning approaches. We hope that future work can expand on the idea we propose, and apply it to other geographies. We have addressed the issue of the comparison table by expanding the literature review in the introduction section. We hope that this amendment sufficiently addresses the legitimate concerns expressed regarding the literature.
Author Response File: Author Response.pdf
Reviewer 3 Report
This paper “Temporal autocorrelation of Sentinel-1 SAR imagery for detecting settlement expansion” presents the Temporal autocorrelation function (ACF)-based methods for detecting settlement expansion. Their results indicate that ACF is capable of detecting and mapping settlement expansion in an accurate manner and can be proven time and cost-effective. This is really a scientifically sound paper with exceptionally good written scientific English. Overall pre-processing steps and statistical analysis were presented in a very accurate manner. I have not found any big major issues with their findings. After the following corrections, I think this paper is ready to be published.
I have only one issue with this paper
The only minor addition I think authors should include in this paper is regarding Figures 5, 6 and 7. While comparing sentinel 1 imagery with the optical imagery they used google earth imagery as ground truth data. I will suggest including the date of the acquired image. For example, Figure 5 and Figure 6 do not have any date mentioned (Or at least the year of the image).
Another important point is that Google earth did not capture these images (Figure 7). Kindly refer to google earth again and see which satellite captured these images and it is preferable to write the name of the satellite in the scientific paper. (It’s most probably GEO Eye or Quick bird imagery). They should check and make amendments.
Author Response
We would like to thank the anonymous reviewers for their valuable contributions to the quality of this paper. Please find below (and attached) our response to the reviewer comments, with details of how the suggestions were incorporated into the text shown in blue.
---------------------
Point 3.1: The only minor addition I think authors should include in this paper is regarding Figures 5, 6 and 7. While comparing sentinel 1 imagery with the optical imagery they used google earth imagery as ground truth data. I will suggest including the date of the acquired image. For example, Figure 5 and Figure 6 do not have any date mentioned (Or at least the year of the image).
Response 3.1: Thank you for this sensible comment. We have added the image dates to the figure captions and to the text.
Point 3.2: Another important point is that Google earth did not capture these images (Figure 7). Kindly refer to google earth again and see which satellite captured these images and it is preferable to write the name of the satellite in the scientific paper. (It’s most probably GEO Eye or Quick bird imagery). They should check and make amendments.
Response 3.2: Thank you. While Google Earth does not specify the exact sensor used, we have attributed Maxar Technologies for providing the Google Earth data in Figure 5, 6 and 7. It is likely their WorldView-3 imagery, but we could not directly confirm this.
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
The author has fixed all my reviews by improving the quality of figures and text, i am going to accept this pape for publication.
