GPU-Based Soil Parameter Parallel Inversion for PolSAR Data
Round 1
Reviewer 1 Report
The paper presents an interesting analysis of three PolSAR algorithms once implemented using GPUs. The core of the paper is not on the PolSAR algorithms themselves, but on how they can be implemented in an efficient way using GPU. Also, some comparisons with the CPU implementations are done. The paper is structured well. The only small point of criticism is that, overall, the authors propose separate analyses for the three different algorithms and the description of the algorithms is relatively short to be fully understood by non-expert readers. I suggest adding a final section (It could be the Conclusion section), where additional information on the interrelation among the different implementations could be highlighted. Also, is there a way to design a hybrid implementation (CPU+GPU) of the algorithms, and if so, which are the steps that better suit such an implementation? Finally, it's clear that the main purpose of such algorithms is the estimation of the polarimetric signature of the surface, but not only the soil moisture and the surface roughness are of interest. Can the author explain (better) the scope/role of PolSAR? Also giving some new insights on the future of such techniques.
Overall, the paper is well structured and deserves publication in my opinion, especially now that Polarimetric data seems to become more accessible.
Author Response
Please see the attached file.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The present work is interesting and novel while authors should rewrite the abstract to include - what this work will provide for community and how different algorithms perform for soil moisture.
2. Line 11-12 page 1 in abstract: "beside the variety of ......" does not convey any meaning kindly rewrite or change the sentence.
3. Please provide contemporary citation and also cite the below article- some of them are here-
- Assessment of evapotranspiration and soil moisture content across different scales of observation. Sensors 2008, 8, 70–117.
- A review of the methods available for estimating soil moisture and its implications for water resource management. J. Hydrol. 2012, 458, 110–117.
- Scaling of soil moisture: A hydrologic perspective. Annu. Rev. Earth Planet. Sci. 2002, 30, 149–180.
- GIS and Remote Sensing Aided Information for Soil Moisture Estimation: A Comparative Study of Interpolation Techniques. Resources 2019, 8, 70. doi: 10.3390/resources802007
- Towards improved spatio-temporal resolution soil moisture retrievals from the synergy of SMOS & MSG SEVIRI spaceborne observations. Remote Sens. Environ. 2016, 180, 403–471.
4. Kindly check the resolution of figures, prepaare and submit with 300 dpi
5. Conclusion should include future recommendation and any limitation faced during the study
6. Provide the details of test site where SAR data is used , like study area and validation sites in The imagery. Validation table is required?
7. Please check the English and grammar throughout the manuscript .
Author Response
Please see the attached file.
Author Response File: Author Response.pdf
Reviewer 3 Report
This paper reports a GPU-based calculation for various PolSAR models. The theme itself is somehow interesting for the researchers. However, it reports only for the implementation results and not from the theoretical aspects.
From current text, it seems that the authors simply implemented the existing method to GPU-computing. Implementation itself is a technical matter and not suitable for scientific journal unless the authors made approximations for GPU-friendly calculation. In such a case, the authors must describe the difference between the proposal and the existing methods and, validate the proposal. Theoretical computational cost can also be described. Current article only compares the difference in the implementation under the specific machine and thus, lacks novelty and generality.
Some minor comments:
1. Ref. 14 and 15 are not used in this article.
2. Some initials lack period (e.g., “Zhang F” in L. 60 should be “Zhang F.”).
3. The order of three methods should be aligned. Currently, "Oh, Dubois and X-Bragg" in the body text while Dubois (Sec. 2.1) Oh (2.2) and X-bragg (2.3).
Author Response
Please see the attached file.
Author Response File: Author Response.pdf
Round 2
Reviewer 3 Report
The authors tried to revise the manuscript extensively however; this reviewer could not find any good reason to recommend publication.
The accuracy doesn't depend on the implementation itself but whether it is mathematically equivalent to the original theory. As the authors developed the GPU-friendly software equivalent to the original article, there should be no difference between CPU- and GPU-based implementation.
Such a fact is widely known and agreed. Therefore, the contents of this article is rather a technical know-how than a scientific findings. Nevertheless making a fast software is important but a benchmark itself is not a science.
