Application of Getis-Ord Correlation Index (Gi) for Burned Area Detection Improvement in Mediterranean Ecosystems (Southern Italy and Sardinia) Using Sentinel-2 Data
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThe author takes Sentinel-2 satellite data with Getis Ord local spatial autocorrelation index as the research object. This study proposes an adaptive thresholding method that combines Getis Ord local spatial autocorrelation index (Gi), Sorensen Dice similarity index, and NBR to improve the false positive problem of dNBR. The novelty of this work is high, but there are still very few areas in the article that need to be changed. The detailed information of the changes is listed in the PDF file details.
Comments for author File: 
Comments.pdf
Thank you for your submission. Although your research is thorough and insightful, I suggest you make some modifications to the English to improve the clarity and readability of your paper.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsThe research “Application of Getis-Ord correlation index (Gi) for burned areas detection improvement in Mediterranean ecosystems (Southern Italy and Sardinia) using Sentinel-2 data”, presented a methodology for identifying burned areas based on the application of thresholds and spatial autocorrelation analysis in Sentinel2 images.
The research was well-based and included several statistical tests to validate the results found. The proposed methodology, which used the widely known dNBR index, managed to significantly reduce false positives, which were a major weakness in its application.
The text has great relevance and quality in presentation. I bring below some minor points that were not so clear:
Minor Points
I suggest that information on how spatial autocorrelation indices can help improve dNBR data for mapping burned areas be added in the introduction. NOTE: During reading it became clear that the authors were concerned with adding this context to the methodology topic, yet I believe that some comment in the introduction would be useful, for example moving L274-279 to the beginning.
L50-51: This statement is not real, there are some biomes in which the aforementioned index does not have satisfactory results, such as in tropical forest areas.
L301-305: In the future practical application of the indexes, could this step be done in a simpler way? Given that basically all previously identified areas were used to generate the thresholds, losing meaning in the mapping itself.
Table 5. Where did the “cropland pixels” data come from?
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
Comments and Suggestions for AuthorsTitle: Application of Getis-Ord correlation index (Gi) for burned areas detection improvement in Mediterranean ecosystems (Southern Italy and Sardinia) using Sentinel-2 data
By: Antonio Lanorte, Gabriele Nolè, Giuseppe Cillis
Review comments
The authors of this paper innovatively constructed two new indices, dGINBR and dNBRGI, by performing the Getis-Ord local spatial autocorrelation which is widely used in landscape ecology. These two indices greatly reduce the false positives generated by dNBR of the three burned areas. The proposed approach could be used in other regions for accurate burned area mapping and post-fire management activities. In general, the manuscripts were well written. And therefore, I suggest a minor revision.
Specific comments
1. Line 18: delete obtained.
2. Line 49: insert “on” after “based”.
3. Line 50 replace “region” by “band”.
4. Line 51-52: delete “in the quantification”.
5. Line 54-56: need cite a reference to support you description, “Therefore, a threshold of +0.1 is considered the most appropriate to distinguish burned from un- burned areas”.
6. Line 88: change “of” to “on”.
7. Line 108: delete “and subsequent spatial analysis”.
8. Line 111: delete “a” and “tool”.
9. Line 114: delete “tools”.
10. Combine table 2 with table 1.
11. Line 149: delete “.” After “autocorrelation”, and change “are” to “were”.
12. Line 311-314: The determination of the threshold of dNBRGi (and 311 dGiNBR) was not clearly stated. This needs more detailed descriptions.
13. Combine table 4 with table 3. Why there lack lag distance of 4 and 5 for Tanca dNBRGI and dGINBR?
14. Why did the dNBRGi and dGiNBR for Brienza and San Fili-Rende refer to the lag 5 distance, while for Tanca-Altara to the lag 3 distance? Why did you choose the lag distance with highest SDS values, that should be lag distance 1?
15. Combine table 7 to 9 together.
16. Line 467: Cite references to support your statement, “which can be determined mainly by the drying out of green vegetation, but also by modest changes in ground cover or moisture, as well as the presence of shadows or smoke”.
17. Line 467-469: Cite references to support your statement, “which can be determined mainly by the drying out of green vegetation, but also by modest changes in ground cover or moisture, as well as the presence of shadows or smoke”.
18. Line 482-483: how could you guarantee burned area at least equal to that provided by dNBR in terms of number of burned pixels?
19. Line 488-489: Cite references to support your statement, “This is mainly the case in burned areas where the vegetation was already very dry and sparse before the fire”.
20. Line 489-491: Cite references to support your statement, “when the state of the vegetation is different (less dry or more humid), as in the case of Brienza and San Fili, the number of false negatives is much lower. Or they are low-severity burned areas”.
21. Line 492-502: add a table to show these numbers clearly.
Comments for author File: Comments.pdf
the English is well.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
