Precise Orbit Determination for Climate Applications of GNSS Radio Occultation including Uncertainty Estimation
, Gottfried Kirchengast 1,2,3, Marc Schwärz 1,2, Christian Pock 1,†, Adrian Jäggi 4, Yago Andres 5 and Christian Marquardt 5Round 1
Reviewer 1 Report
Review of the manuscript „Precise Orbit Determination for Climate Applications of GNSS Radio Occultation including Uncertainty Estimation” submitted to Remote Sensing.
The manuscript describes selected issues of the precise orbit determination of LEO satellites and their impact on the total errors budget in the determined orbits. The methodology employed in the manuscript is correct and provides sufficiently new ideas to merit a publication in Remote Sensing.
Below, I provide a list of issues that should be resolved to improve the quality and readability of the manuscript.
Issues
- The full name of the „Bernese” is “Bernese GNSS Software”
- For RO, not only high-accurate orbits are needed but also high-accurate clocks
- Please use a consistent tense in the abstract: preferably simple present.
- Line 138 “are not yet available to the community by early 2020, however.” – it is very uncommon to write however at the end of the sentence
- Line 237: “Due to the measurement geometry the SLR residuals ∆rSLR are most sensitive to the radial component of the orbit solution.”- this holds for high-orbiting GNSS satellites. For low-orbing satellites, such as GRACE, GOCE, CHAMP, this is not true. LEOs are tracked by SLR from the very beginning of the pass till the very end of the pass. The south-north LEO passes are sensitive to the along-track component, whereas the passes in the east or west direction from the station are very sensitive to the cross-track component, see: https://doi.org/10.1016/j.asr.2018.08.033 For LEOs, we cannot assume that SLR validates only the radial orbit component. SLR is sensitive to all orbital components and to the time biases (at both SLR stations and in satellites orbits).
- Table 1. There is one fundamental difference between CODE repro2 and repro2015. In repro2015, the extended Empirical CODE Orbit Model (ECOM2) has been used for the very first time. This model includes the estimation of twice-per-revolution empirical orbit parameters in the satellite-Sun direction D. The model significantly changes the GLONASS orbits, however, GPS orbits are also substantially changed (see: https://doi.org/10.1007/s00190-015-0814-4 ). Therefore, when you see differences between solutions based on CODE/IGS repro2 and repro2015, they should also be assigned to different models used for precise orbit determination of GNSS satellites, ECOM1 versus ECOM2.
- The “standard” solution is named here “baseline solution”. Personally, from the name I would expect that the baselines between satellites (GRACE-A and GRACE-B) are calculated. I think that “benchmark solution” would be a better name for this kind of solutions. The baselines between only ground stations are employed here or did I misunderstand something?
- Figure 6 – some subtitles are covered by figures. Please correct it.
- The paper mentions many times “RO”. However, the analysis and the focus is on orbit determination and quality assessment of the orbits. I do not see a need to address the issue of RO in some many sections of the paper, even if this is the major goal of this study (to obtain high-quality orbits for RO). However, you do not show any results from RO in this paper.
- Table 1: “pseudo-stochastic accelerations” are estimated here for Bernese solutions. Do you estimate accelerations as piece-wise constant or piece-wise linear accelerations? What sigmas are employed for radial, along-track and cross-track components?
- Table 4: Standard deviations of SLR residuals for GRACE are twice larger for WEGC-BC than for WEGC-NC. What is the reason? The AIUB solution also looks much better than WEGC-BC which is in fact also AIUB-like solution. Do you use the same background models for SLR validation in all cases?
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Thank you for submitting this interesting paper. "Precise Orbit Determination for Climate Applications of GNSS Radio Occultation" is very important and has become a lot of leverage due to so-called remote sensing techniques for probing the Earth’s atmosphere.
This article describes a novel setup for precise orbit determination (POD), that enables mutual consistency checks of the calculated orbit solutions and it is used for position and velocity uncertainty estimation, including estimated systematic and random uncertainties.
The paper is well drafted, and its purpose and major theses are cleared presented. The obtained results are optimal and well explained. There are no specific areas that need to be addressed in terms of corrections.
Minor comments:
Figure 1: Please explain the acronym DSM (line 65) and ODP (line 85) before using them in this figure.
Section 3.4 Satellite laser ranging:
If available, some technical features of the Laser system, such as Azimuth and Elevation precision (arcsec), range precision (cm), etc., might be of great interest to readers.
Please move Figure 6 before the Conclusions section.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
Overall, the manuscript is very well written and high quality. There are certainly great effort made to conduct this research. Few minor concerns that i have in mind is as follows:
1) in section 3.2, you described three orbit determination approaches for LEO satellites, and you mentioned reduced-dynamic approach produces the highest quality. But subsequently, you compared dynamic approach against kinematic approach saying that dynamic approach is favored. It sounds a bit contradictory for me as a reader;
2) should you be able to provide more mathematical formulas with regard to theoretical background? There are lots of references provided, which is great. It would be more convenient and easier to follow for the reader to fully understand if there are a few detailed equations with related to your reduced dynamic approach, e.g., orbit modeling, GNSS observation equation, etc;
3) in Fig 4, it appears that WEGC-Bernese (JPL) and WEGC-Bernese (CODE) and AIUB-Bernese results are at the same level (1-2cm difference) for CHAMP, can you please explain more why for GRACE-A, AIUB-Bernese is not at the same level as WEGC-Bernese CODE and JPL?
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
