Baseline Calibration of L-Band Spaceborne Bistatic SAR TwinSAR-L for DEM Generation
Osamu Isoguchi
Fengming Hu
Philip Callahan
Round 1
Reviewer 1 Report
In this paper, a new pixel-related baseline model based on the geometrical shift is proposed to reflect the position change of satellites accurately. The baseline error is then calibrated using height gradient information and a small number of point targets with a slight incidence angle difference, eliminating the need for low-frequency corner reflectors and avoiding the difficulty of selecting a calibration site. The proposed method has been successfully exploited during the initial Commissioning Phase of TwinSAR-L, demonstrating its effectiveness in evaluating precise baseline and supporting the generation of highprecision DEM. Some detailed questions should be answered as following.
1. Line 203 ‘and the threshold is set to 10−9 for a compromise between algorithm accuracy and speed…’,how did this factor is obtain, author should give a quantitative analysis or an experiment.
2. In figure 5 we can see the DEM while Fig 6 is too well, the topography effect on interferogram is not seen. The reviewer wonder if fig 6 is really the true scenario.
3. The meaning is a little bit vague, the illustration is not very self-clear, suggest author add more description on how the image is generated for reader.
4. The conventional linear model used for comparison should be cite or described.
5. Figure 9 (b)(e),the color bar is not the same, they should use same color interval.
6. Some grammar error should be corrected.
Some minor grammar and spell error should be corrected.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The authors proposed a model for estimating baseline error in DEM generation by L-band bistatic radar (TwinSAR-L) and applied it to actual observation data to demonstrate that it leads to a reduction in DEM calculation error. The baseline is corrected by using the characteristic that the calculated DEM error due to the cross-track and radial component errors of the baseline is dependent on the range direction, and by using the CRs placed over a wide range in the range distance as reference heights. In the along-track direction, the baseline is modeled nonlinearly with respect to time to reduce errors that vary in the azimuth direction. It is also well described in terms of methodology.
For scene-by-scene DEM creation, it is somewhat expected result that the error is reduced by placing a wide range of reference points as GSPs in the scene. On the other hand, if one of the objectives of TwinSAR-L is to create DEMs on a global scale, as described in the text, there seems to be a lack of analysis and evaluation for continuous DEM creation in the along-track direction. I recommend that this manuscript will be accepted for the publication after clarifying the following comments.
Regarding the error in the along-track direction, if the baseline model coefficients are determined for each scene, it is assumed that a discontinuity in the DEM between adjacent scenes will occur. It is necessary to evaluate the degree of discontinuity that would occur with the applied method and to discuss how to reduce the discontinuity in the along-track direction. Also, for errors with cross-track dependence, it is not realistic to deploy CR for each calculation scene, as the author has also discussed. Although there is an influence of penetration, it may be possible to discuss methods such as using the existing DEM as GCP and connecting the proposed method in the time direction (along-track).
Minor comments:
1) Line 118: xax and xrg are expected to be the azimuth and range positions of the master image. If so, an explanation is needed.
2) Lines 141-143: The TwinSAR-L DEM error is described as unchanged with respect to the azimuth direction (time). On the other hand, the results of Fig. 7 and Fig. 9 show that the error also changes in the azimuth direction. Does the statement here mean that the bias-like error does not change when considering a longer period?
3) Line 146: B⊥err is not shown in Fig. 2.
4) Line168:If Δ means the angle between the two sides, wouldn't the notation be ΔS'BSAP?
5) Eq. (8): Judging from Fig.3 and Eq. (10), the sign of α must be positive (+)?
6) Lines 224-225: If the fitting parameters here correspond to Eq. (3), they should be a0 to a5.
7) Line 225: How many powers of time were applied as Nk in Eq. (4)?
8) Lines 225-227: In the caption of Fig. 6, Figs 6(a-c) are the result of the linear baseline model, which is the opposite of the main text.
9) Lines 277-278: Shouldn't it be Fig. 9(a,d)?
10) Line 279: Shouldn't it be Fig. 9 (b,c,e,f)?
11) Lines 306-308: Why do the DEM errors due to the 1m baseline error and derivative values differ except for the y component? For example, for the x component, they are listed as 278.56 m and 274.58 m/m, respectively. The value of the z-component (Bz) also appears to be different from the value read from the 44.33 degree incidence angle in Fig. 11a.
12) Lines 349-350: Although the DEM calculated from the TwinSAR-L is assumed to be lower than the SRTM DEM due to penetration, the difference in Fig. 9 is positive. Is this because the difference is calculated based on the DEM calculated from the TwinSAR-L?
13) Line 353-355: The difference between the SRTM DEM and that generated by TwinSAR-L is interpreted as the difference in scattering points caused by the difference in wavelengths. Therefore, the expression "penetration error" may not be appropriate.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
The paper entitled “Baseline Estimation and Calibration of L-band Spaceborne Bistatic SAR TwinSAR-L for DEM Generation” presents a novel baseline calibration for L-band Spaceborne Bistatic SAR, achieving a good performance of DEM generation. The paper is well written and valuable. I only have two minor comments.
1. I think the title should be slightly changed. The main topic of the paper is baseline calibration. The method of baseline estimation is not innovative. I think it is better to remove the ‘Estimation’.
2. The authors propose a pixel-related baseline model to estimate the baseline error. In the practical application, how to select the pixels for estimation?
Author Response
Dear Reviewer,
We’d like to express our most sincere gratitude for your effort and patience in reviewing our manuscript. We deeply appreciate your constructive comments that greatly help improve the technical quality and the presentation of this manuscript.
We have studied the comments carefully. The revised portion has been marked in our revised manuscript. Appended to this letter are our point-by-point responses to the comments and suggestions raised by the Reviewer.
Thank you in advance for your time.
Sincerely,
Jingwen Mou
E-Mail: [email protected]
Tel.: +86 18801070903
Response to the Reviewer’s comments:
Reviewer 3
Comments and Suggestions for Authors:
Comment 1: I think the title should be slightly changed. The main topic of the paper is baseline calibration. The method of baseline estimation is not innovative. I think it is better to remove the ‘Estimation’.
Response: We appreciate your valuable time and feedback in helping us improve the quality of our work. We agree that the highlight of our paper is the baseline calibration, and we appreciate your suggestion to change the title to reflect this focus.
Comment 2: The authors propose a pixel-related baseline model to estimate the baseline error. In the practical application, how to select the pixels for estimation?
Response: Thanks for your great comment. In this experiment, two steps required pixel selection: ① Fitting the baseline model polynomial, as described in Equation (4); ② Estimating the initial baseline matrix, as described in Equation (14).
For the first step, we selected approximately 90,000 points with coherence coefficients higher than 0.9. This selection criterion was based on the principle that the points with high coherence coefficients generally receive less disruption, which leads to a better fit.
For the second step, we selected the pixels where the corner reflectors were located and used them to construct the initial baseline matrix using the baseline model. The pixels chosen for this step had a high signal-to-noise ratio and a coherence coefficient exceeding 0.95, thereby reducing the impact of random errors on baseline error extraction.
Thanks again for your comments.
Reviewer 4 Report
Aiming to generate the high-accuracy DEM of the bistatic system TwinSAR-L, this paper proposes a pixel-related baseline model and a calibration method to derive the precise baseline. The technical soundness of the methods is confirmed, and the real data results are presented. However, there are still some problems that need to be solved.
1) Using appropriate abbreviations to keep the paper as concise as possible. For example, “Line-of-sight” can be abbreviated as “LOS”.
2) Explain the arrangement of corner reflectors and their influence on the calibration.
None.
Author Response
Dear Reviewer,
We'd like to express our most sincere gratitude for your effort and patience in reviewing our manuscript. We deeply appreciate your constructive comments that greatly help improve the technical quality and the presentation of this manuscript.
We have studied the comments carefully. The revised portion has been marked in our revised manuscript. Appended to this letter are our point-by-point responses to the comments and suggestions raised by the Reviewer.
Thank you in advance for your time.
Sincerely,
Jingwen Mou
E-Mail: [email protected]
Tel.: +86 18801070903
Response to the Reviewer's comments:
Reviewer 4
Comment 1: Using appropriate abbreviations to keep the paper as concise as possible. For example, "Line-of-sight" can be abbreviated as "LOS".
Response: Thanks for pointing this out. We have revised related content in our new manuscript. Thank you for your valuable advice to improve our paper.
Comment 2: Explain the arrangement of corner reflectors and their influence on the calibration.
Response: Thanks for your great comment. In the manuscript, we conducted an analysis of how baseline error impacts height error and found that the introduction of baseline error results in height error that varies with the look angle. Therefore, to mitigate this effect, we need to deploy calibrators in the SAR image's near, middle, and far view regions. Additionally, every view requires at least two corner reflectors to avoid outliers. As a result, a minimum of six corner reflectors must be deployed. The corner reflectors should be deployed across a wide range of incidence angles and distributed across the entire swath to prevent space-varying interferometric phase errors to ensure the effectiveness of the calibration. The arrangement of corner reflectors is shown in Figure 5.
Thanks again for your comments.
Reviewer 5 Report
The methods described in Sec 2 are incomprehensible. There seems to be a mixing of SAR image to interferogram to height that is not explained (see specific comment on Fig 4).Perhaps if one had fully internalized the methods used by Tandem-X they would be understandable, but even that seems a stretch. It is impossible to judge the Novelty and Significance under these conditions.
A few specific items that will only help a small amount with the above problem(s):
L 103 (end): conclusion -> conclusions.
L 108: What is “poly” mean on delta-t?
L 115: There is a jump from “image” as in Fig 1 to “interferogram” that is not explained.
L 128: There is no reference or explanation for translating attitude error into baseline error. Is this based on motion of antenna phase center relative to the CM of the satellite from changing attitude?
L 166: Missing “is” (?).
L 183: What is GCP? Spell out at first use.
Fig 4: It seems that there is at least one missing step in the flow: going from coregistered images to Initial Baseline.
L 190: Why/what is the N dimension on B?
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
the comments are all answered
Author Response
Dear Reviewer,
We'd like to express our most sincere gratitude for your effort and patience in reviewing our manuscript.
Thanks for your time.
Sincerely,
Jingwen Mou
E-Mail: [email protected]
Tel.: +86 18801070903
Reviewer 5 Report
L 51: “changed” -> “changing”
L 63: Apparently the “reference interferogram” is computed from the existing DEM. If this is so, say it; otherwise, give the source of the reference interferogram.
L 119: fs,az has opposite sub/superscripts from eq 3.
L 122-128: This paragraph seems to imply a connection between eq 2 and 3 that is not explained or shown. It also seems that the x’s in eq 3 are not positions, but position differences.
Eq 3 is not immediately understandable, and its relation to the rest of the analysis, other than removing along-track error, is not clear.
L 144-145: Does the SAR antenna really have multiple beams? Even (or perhaps, especially) if so, there will surely be differences between the two satellites. This error/change cannot be so easily dismissed.
Pp 6-7: This analysis is too complex, messy to be evaluated in a reasonable amount of time. It would be much clearer with vectors and some explanation. Computers can then do the final math.
Fig 5: Do triangles on DEM in (b) show locations of corner reflectors?
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 3
Reviewer 5 Report
I have no further comments.
Author Response
Dear Reviewer,
We sincerely thank you for your dedicated time and effort in reviewing my work and providing valuable suggestions. Your expertise and thoughtful feedback have been invaluable in shaping our manuscript, and we truly appreciate your contributions.
Your comprehensive and constructive critique has played a vital role in improving the quality and effectiveness of our manuscript. We are grateful for your specific suggestions, as they have helped us address our weaknesses and challenged us to think deeper and explore alternative perspectives. Your insightful comments have undoubtedly elevated our work's overall quality and coherence.
In response to your previous suggestions, we have diligently modified the article to incorporate your valuable feedback. We have taken your recommendations to heart and have carefully revised the content to enhance its clarity, coherence, and overall impact. To ensure ease of review, we marked the areas where the modifications were made, allowing you to identify the revisions made following your suggestions quickly.
Thank you in advance for your time.
Sincerely,
Jingwen Mou
E-Mail: [email protected]
Tel.: +86 18801070903
