Geolocation Accuracy Validation of High-Resolution SAR Satellite Images Based on the Xianning Validation Field
, Guo Zhang 1, Qian Cheng 5 and Shuying Lv 6Round 1
Reviewer 1 Report
it seems to me a complete study, I suggest to include the metrics that resulted
Author Response
Response to Reviewer 1 Comments
We thank the reviewer for the comments and appreciate this opportunity to revise and resubmit our paper. We have systematically responded to all the reviewer’ comments below. All the changes are marked as red in the text.
Point1: it seems to me a complete study, I suggest to include the metrics that resulted
Response1: Okay, the metrics of results in abstract, result analysis and conclusion has been added, for example:
In line 26:” The results showed that with four ground control points (GCPs) laid in corners, the geolocation accuracy of each image could reach an RMSE (Root Mean Square Error) of 1.5 pixels level.”
In line 394:” By comparison, the performance of COSMO-SkyMed, TerraSAR-X, and RadarSat-2 is better than YG-3 and ALOS-PALSAR, amounting to an RMSE of 2 pixels level without GCPs.”
In line 404: “By comparison, the performance of COSMO-SkyMed, TerraSAR-X, and RadarSat-2 is better than YG-3 and ALOS-PALSAR, amounting to an RMSE of 2 pixels level without GCPs.”
Reviewer 2 Report
Comparative analysis of the list of references is weak. Authors should consider identification and scanning processes in more detail.
According to the methods used, the analysis is poorly done. Authors should describe the problem and solution in the article, and give a detailed explanation in between. The magazine is designed for a wide audience and therefore it is necessary to write as accessible as possible. Any pattern recognition contains an identification error; I did not see this study. Image overlay is not available. It is necessary to describe the resolution of pattern identification.
Expand the references. Add more professional publications and their period 2020-2023.
Author Response
Response to Reviewer 2 Comments
We thank the reviewer for the comments and appreciate this opportunity to revise and resubmit our paper. We have systematically responded to all the reviewer’ comments below. All the changes are marked as red in the text.
Point 1: Comparative analysis of the list of references is weak. Authors should consider identification and scanning processes in more detail.
Response1: Thank you for your comments and suggestions regarding our manuscript. We have carefully reviewed your feedback and taken it into account in revising our paper. Regarding the list of references, we apologize for any shortcomings in our comparative analysis. We re-examined our sources and give more attention to identifying and scanning processes in our analysis.
Point 2: According to the methods used, the analysis is poorly done. Authors should describe the problem and solution in the article, and give a detailed explanation in between. The magazine is designed for a wide audience and therefore it is necessary to write as accessible as possible. Any pattern recognition contains an identification error; I did not see this study. Image overlay is not available. It is necessary to describe the resolution of pattern identification.
Response2: Thank you for your valuable feedback on our manuscript. We appreciate your comments and have taken them into consideration in revising our paper. As per your suggestion, we have re-organized the analysis section and have made significant improvements by describing the problem and solution in detail and providing a more accessible explanation.
Point 3: Expand the references. Add more professional publications and their period 2020-2023.
Response3: Thank you for your valuable feedback. We have taken your comments into consideration and have made revisions to our list of references. We have added more professional publications and included their period from 2020 to 2023. Once again, we appreciate your suggestions and thank you for your time and effort in reviewing our manuscript.
Reviewer 3 Report
In the article, the authors present geometric performance of five representative high-resolution SAR satellites ALOS, TerraSAR-X, Cosmo-SkyMed, RadarSat-2, and Chinese YG-3. They evaluated evaluated on the same benchmark with the rational function model (RFM).
The Rational Function Model (RFM) is a commonly used approach for SAR geometric correction, which estimates the sensor geometry and corrects for distortions in the image caused by terrain relief and sensor platform motion.
Overall, the geometric accuracy of SAR images using the RFM is considered to be high. The RFM is able to model complex terrain and platform motion, which can introduce significant distortions in the image. The model is also able to correct for range and azimuth distortions, which are caused by the non-uniform antenna pattern and the motion of the platform during the SAR acquisition.
The manuscript should attract an audience in the scientific field of computer vision. The manuscript is written quite good, but the discussion must be extended. Several studies have evaluated the accuracy of SAR images corrected using the RFM, and have found that the model is able to achieve sub-pixel accuracy. For example, a study by Kim et al. (2015) evaluated the geometric accuracy of TerraSAR-X images using the RFM, and found that the RMS error in ground range and azimuth was 0.12 meters and 0.16 meters, respectively. Similarly, a study by De Michele et al. (2014) evaluated the accuracy of COSMO-SkyMed images corrected using the RFM, and found that the RMS error in ground range and azimuth was 0.3 meters and 0.2 meters, respectively.
In Introduction, authors explained the proposed used technology. Introduction is written very well.
Abstract does not contain information about the results obtained.
Add a statistical significance test to assess if the differences between the compared models are really significant. If your results data are not parametric, you could use the Sign Test or Wilcoxon or Friedman tests.
I have objections to the discussion section. The authors need to re-organize ,the results and discussion therein to better highlight to the reader what was done and what is relevant. The gain of the presented technique for the addressed application should be made more explicit in the form: What do the findings allow what was possible before. Authors should discuss the results and how they can be interpreted in perspective of previous studies and of the working hypotheses.
Conclusions are correct.
Author Response
Please see the attachment
Author Response File: 
Author Response.pdf
Reviewer 4 Report
The work is useful for the users to know the geometric quality of the SAR image products from the five SAR satellites. Some suggestions for improving the text are as follows.
1. Line 73, PAD? It should be APD.
2. Line 64, Line 79, YG-13, should be YG-3?
3. Eq(3), the parameters a0,a1,a2,b0,b1,b2, are confused with that in eq(2). While x’,y’ are of the same meaning with rn,cn in eq(2), please use the same symbols.
4. Which model was fitted with GCPs? It is not so clear. I know it is equation (3), but it is better to classify it. You have eq (2) with parameters that can also be fitted if you have many GCPs.
5. “Therefore, a possible factor remains to be the atmosphere propagation delay, which was not taken into consideration in the simulation procedure”, your simulation method is based on RFM which can be thought as one copy of RD used to generate the SAR image product, so as you have realized through analysis in line 340, APD had been considered by RD thus by RFM and by simulation. This fact is conflict with this comment.
Author Response
Response to Reviewer 4 Comments
We thank the reviewer for the comments and appreciate this opportunity to revise and resubmit our paper. We have systematically responded to all the reviewer’ comments below. All the changes are marked as red in the text.
Point 1. Line 73, PAD? It should be APD.
Response1: Yes, it has been corrected:
Line 73:” Experimental results indicated that an un-documented APD correction might have been incorporated in delivered timing annotations, for APD compensation leads unexpectedly to results worsening.”
Point 2. Line 64, Line 79, YG-13, should be YG-3?
Response2: Okay, in line 64 and line 79, it is correct, the YG-13(Yaogan-13). The used image in this paper is YG-3, they are from a same series.
Point 3. Eq(3), the parameters a0,a1,a2,b0,b1,b2, are confused with that in eq(2). While x’,y’ are of the same meaning with rn,cn in eq(2), please use the same symbols.
Response3: Yes, the parameters in eq(2) have been changed :
Line 128:“”
Point 4. Which model was fitted with GCPs? It is not so clear. I know it is equation (3), but it is better to classify it. You have eq (2) with parameters that can also be fitted if you have many GCPs.
Response4: Okay, the used model was mentioned, and the changes are:
Line 131:” In this paper, the affine transformation model was used in bias compensation for the RFM [21].”
Line 280:” The geolocation accuracy of each image was analyzed and verified using the affine transform model based on different schemes of laid GCPs, as follows in Table 2:”
Line 349:” The used CPs for verifying were obtained in the same way with GCPs, and the affine transform model was used to fitted the GCPs.”
Point 5. “Therefore, a possible factor remains to be the atmosphere propagation delay, which was not taken into consideration in the simulation procedure”, your simulation method is based on RFM which can be thought as one copy of RD used to generate the SAR image product, so as you have realized through analysis in line 340, APD had been considered by RD thus by RFM and by simulation. This fact is conflict with this comment.
Response5: We apologize for any confusion caused by the statement that the atmosphere propagation delay was not taken into consideration in the simulation procedure. Upon further review, we realized that this statement was incorrect. In fact, our simulation method is based on RFM, which is a copy of RD used to generate the SAR image product. As mentioned in line 340 of the paper, APD had been considered by RD, RFM, and the simulation process. Therefore, we have taken into consideration the effect of the atmosphere propagation delay during the simulation process. We hope this clarification addresses your concern.
Author Response File: 
Author Response.docx
Round 2
Reviewer 3 Report
The authors took into account all my comments.
