In-Memory Distributed Mosaicking for Large-Scale Remote Sensing Applications with Geo-Gridded Data Staging on Alluxio
Round 1
Reviewer 1 Report
The authors presented an in-memory Spark-enabled distributed data mosaicking at a large scale with geo-gridded data staging accelerated by Alluxio. As for me, the presented method sounds good but has some limitations which should be addressed.
1) "mosaicking is practically suffered 56 from a staggering number of “messy“ scenes when scaling to a large region". Another limitation of mosaicking is radiometric differences between scenes that should be addressed. I know that the focus of this paper is to produce a method for image stitching. However, the radiometric differences between patches should be addressed well in the introduction using these citations [x][xx][xxx]. After that, the author should mention that the paper focuses on developing image stitches.
[x] https://doi.org/10.1109/MGRS.2019.2921780
[xx] https://doi.org/10.1109/TGRS.2021.3063151
[xxx] https://doi.org/10.1109/LGRS.2020.3031398
The authors can see the radiometric differences between scenes in the final mosaicking results in Figure 7.
2) please show five zoom areas between scenes in Figure 7 to show the performance of the proposed method.
3) In the dataset description, I see the three optical and 1 SAR images, Landsat 7, 8, Sentinel 2, and Sentinel 1 (SAR). Do the authors generate the mosaic from all of these images? Optic and SAR !!!! Or generate a mosaic for each sensor separately?
Please show the results of mosaicking in RGB format for Optical images and show the Sentinel 1 image, band by band.
Which polarization was used for Sentinal 1, and which mode (ascending or descending)?
Please show the part of mosaicking results from ENVI soft beside the presented mosaic and compare these to each other in terms of quality.
4) It is recommended to include a discussion section to analyze the reasons for the experimental results, the limitations of the proposed method, and possible application scenarios.
Author Response
Please see the attachment
Author Response File: 
Author Response.docx
Reviewer 2 Report
In this study authors proposes an in-memory Spark-enabled distributed data mosaicking at a large scale with geo-gridded data staging accelerated by Alluxio,
Topic of the study is original and important in this field
1- I would like to recommend to improve introduction section
Yilmaz, M. & Uysal, M. (2017). Comparing uniform and random data reduction methods for DTM accuracy . International Journal of Engineering and Geosciences , 2 (1) , 9-16 . DOI: 10.26833/ijeg.28600
2- Image mosaicking is widely used in the remote sensing field. suggested method is very good. However,is there a mosaicing problem ? ıs there a color distortion or radiometric distortion radiation balance, seam line extraction, and image blending. The authors can explain mosaicing results .
3- Please explain GeoMesa what is algorithm of GeoMesa. What is the contribution of suggested method . is there any disadvazntages.? Especially in a big data
4- You are using three type of images, Landsat 7, 8, Sentinel 2, and Sentinel 1 where is the obtained mosaiced images. Please add the images and explain advantege and disadvantage of all mosaics and please axplain more detaişled all experimental results
5- Please add new images and explain advantege and disadvantage of all mosaics and please axplain more detailed all experimental results
6- Please add more referances (add up-to-date references that have been published on this topic recently)
Author Response
Please see the attachment
Author Response File: Author Response.docx
Reviewer 3 Report
The article is devoted to methods of processing large amounts of data. One of the options that allows you to spend less time processing the received satellite information is presented. 80% is a review article. Even when describing their method, the authors of the article pay a lot of attention to describing the results obtained by other researchers. If it is customary to publish such articles (review articles) in this section of this journal, then it should be published with minor editorial edits. But there is very little research part in this article.
Author Response
Please see the attachment
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
The authors well addressed most of my comments. However, the first comment is not well addressed. The author should be informed that Image mosaicing is an effective means of constructing a seamless image by aligning multiple partially overlapped images. So radiometric diffrences caused by different factors should be reduced by different methods. I suggested the citations for a deeper insight, but the reader cite the refernces that are more related to image enhancement (except one related to color balancing), not image mosaiking. Once again, if the authors developed a mosaicking method, please cite the related works such as the following:
https://doi.org/10.1016/j.imavis.2019.07.002
https://doi.org/10.1109/MGRS.2019.2921780
https://doi.org/10.1109/TGRS.2021.3063151
https://doi.org/10.1109/LGRS.2020.3031398
Moreover, Celik et al.[38] proposed a multi-dimensional histogram equalization method to achieve color equalization. Color equalization is not a kind of color balancing. Please fix the comment for more readability. Once again, the paper was well structured after corrections, and the author presented an excellent research work that the related works should fulfill for more readability.
Author Response
Please see the attachment
Author Response File: Author Response.docx
Reviewer 2 Report
it can be accepted after minos revision
only I would like to recommend to improve introduction section
Ahady, A. B. & Kaplan, G. (2022). Classification comparison of Landsat-8 and Sentinel-2 data in Google Earth Engine, study case of the city of Kabul . DOI: 10.26833/ijeg.860077
Benbahrıa, Z. , Sebari, İ. , Hajji, H. & Smiej, M. F. (2021). Intelligent mapping of irrigated areas from Landsat 8 images using transfer learning . DOI: 10.26833/ijeg.681312
hady, A. B. & Kaplan, G. (2022). Classification comparison of Landsat-8 and Sentinel-2 data in Google Earth Engine, study case of the city of Kabul . DOI: 10.26833/ijeg.860077
Author Response
Please see the attachment
Author Response File: Author Response.docx
