Initial Study of Adaptive Threshold Cycle Slip Detection on BDS/GPS Kinematic Precise Point Positioning during Geomagnetic Storms

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Detection and repair of circumferential jump has always been a difficult problem for GNSS to realize high-precision positioning and navigation, and many scholars have already conducted in-depth research on this problem. Electromagnetic storms have a large impact on the navigation and positioning signals, which seriously affects the realization of high-precision navigation and positioning. Based on the above problems, the manuscript investigates in detail the weekly hop detection and repair of the global positioning system during magnetic storms, which has very important scientific value and significance. The methodology of the manuscript is clear and the arguments are rigorous, but the manuscript is not innovative enough, and the quality and level of the manuscript should be improved by increasing the analysis of the previous research results to reflect the innovation of the manuscript. It is recommended that the manuscript be published after minor revision..

Introduction: The manuscript investigates the advantages and disadvantages between the adaptive thresholding method and the fixed thresholding method, but there is no introduction or comparison of previous research results on the two methods in the introduction.

Although the manuscript takes the localization accuracy as an indicator in the validation results, it is not enough to show the innovativeness of the manuscript. Therefore, the authors are recommended to dig deeper into the innovation points to fully reflect the innovation of the manuscript.

There are many studies on circumferential hop detection and repair during magnetic storms, and it is suggested that the manuscript add a discussion section that analyzes and compares the results of previous research with the results of the manuscript and briefly describes the limitations of the results and the direction of future work

Author Response

Please see the additional

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

11)    What is I in Eq. (11)? What is the relationship with I_1 and I_2?

22)    In Table 1, why n_i of BDS is so larger than that of GPS? Assuming R=1, n_i of BDS is 30 times larger than that of GPS. Eventually the GF adaptive cycle slip threshold for BDS is much larger than GPS. Is this reasonable? Why? do you have any principal behind this?

33)    What is the unit for n_i in Table 1, subsequently what is the unit for U in Eq. 12 and 14. Is it meter? If so, it is very big for BDS satellites. when R = 30 as in the experiment, n_5 could be more than 4 meters for BDS. This equals not applying GF cycle slip detection at all.

44)    What is cycle jump rate? How is it defined?

55)    I suspect that the authors did not make clear why the PPP solutions are bad in the magnetic stormy period. The ambiguity resets due to cycle slip should not make very large (meter level) fluctuation if there are many other satellites still keeping tracking. If there are cycle slips but do not reset the ambiguity term, the position error would be large as well. The fluctuation in position domain looks like the position engine reset completely?

66)    The so-called adaptive thresholds look like just let the cycle slip go through the quality control and continue to make use of those carrier phase.  

77)    What are the criterions to use five thresholds, respectively?

Comments on the Quality of English Language

The English presentation is at the intermediate level. Basically the description is OK, but sometimes it is too length. Maybe the authors should send the manuscript to be smoothed by professionals.

Author Response

Please see the additional

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

In my opinion, the work “Initial Study of Adaptive Threshold Cycle Slip Detection on BDS/GPS Kinematic Precise Point Positioning During Geomagnetic Storms” is low quality and could be much better organized. I am also not convinced the study is innovative. Consequently, I recommend the editor reject the work in its current form. To me, resubmission is feasible, but only after significant corrections.

 

1. English should be improved.
2. “They are mainly based on adding cycle slips to the data artificially and then conducting experiments to get the cycle slip detection success rate.” This statement is not true
3. Although I agree that a typical GF threshold for high latitude is incorrect, the new algorithm is given a highly complex form, and thus, it must be clarified. The main points that should be improved here are:

- why do we need some scaling factor for low-elevated data (Equation 13). The authors do not mention any reason. It would be good to show this necessity by analyzing temporal differences of L4 data. The dependence of factor k on elevation should also be plotted.

- looking at Table 1, we see a different threshold for GPS and BDS (for BDS seems to be much higher, but I do not see any units). Why? The impact of the ionosphere is expected to be very similar for MEO satellites in both cases. For BDS GEO satellites, it could be different (Equation 15), but this is not shown. In my opinion, the authors should support the selection of thresholds with some example plots showing the temporal variations of GF, including data from different BDS satellites

- the values of the threshold in meters would be much more straightforward for readers

- what are the units of R? 

- if the results provided by strategy 5 are the most stable (according to the following sections), a higher threshold could be better.  

                                                                                                                                        

4. The description of storms should be oriented on ionosphere variations (using, for example, ROT or ROTI) or indices highly correlated with ionosphere state, for instance, Ae for high latitudes. The current form does not show any connection between the storm and the state of the ionosphere. Selecting a few stations and illustrating actual ionosphere conditions with ROT or ROTI would be good. Such plots can be considered as a preliminary analysis connected with your thresholds. Furthermore, I expect some connection between your results and IMF Bz/Dst, but the remaining two indices (Kp and plasma speed) are unnecessary for your work and, thus, can be removed.
   

5. What does “Dual-frequency Non-ionospheric” mean in Table 2?

6. What is exactly given in Table 3, RMSs (Equation 18)? According to this table, the deteriorating impact of storms at low latitudes is rather low, but in Figure 3, it is shown that the error for this region may exceed 1 meter. Was such a strong effect observed only for a few stations? If yes, please explain why.

7. In my opinion, the section devoted to the carrier-to-noise ratio should be removed. 
   

8. I do not understand the conclusion that “the positioning accuracy of most of the stations has returned to the average level”. Even with 100% improvement, which was not observed for the best strategy 5, the accuracy could not be as good as for quiet time (see Table 3). Thus, if I am not wrong, the accuracy is still worse. This should be shown and explained.

9. According to Figure 11, even for an adaptive GF threshold, the cycle-slips rate for station MAC1 is close to 50% for a few hour period, and this does not influence PPP accuracy. It looks extraordinary. First, I would expect a lower ratio for each epoch. To me, it suggests still too low threshold or some problems with data. On the other hand, If the cycle slips were real, we should observe the deterioration of PPP accuracy. In Figure 11, one can also see some jumps in PPP accuracy (best visible for IQAL station). The reason for their occurrence needs some clarification.   

 

Comments on the Quality of English Language

I suggest carefully reading the work, some sentences are difficult to follow.

Author Response

Please see the additional

Author Response File: Author Response.pdf

Reviewer 4 Report

Comments and Suggestions for Authors

Manuscript:

Initial Study of Adaptive Threshold Cycle Slip Detection on BDS/GPS Kinematic Precise Point Positioning During Geomagnetic Storms

submitted by:

Xing Su, Jiajun Zeng, Quan Zhou, Zhimin Liu, Qiang Li, Zhanshu Li, Guangxing Wang, Hongyang Ma, Jianhui Cui and Xin Chen

 

 

Satellite-based precise positioning, both static and kinematic, is ensured by new methods. These include reference stations, radio and internet-transmitted corrections (RTK/RTN), and the PPP method. Both single-system and multi-system receivers are used. In the article the Authors propose   adaptive threshold for cycle slip detection to improve the kinematic PPP accuracy during the magnetic storms.

 

The structure of the article is correct and well thought out. The introduction includes a literature review. The theoretical part provides a detailed mathematical description of TurboEdit cycle slip detection methods, and GF adaptive thresholding model. The research was conducted during three magnetic storms. The results of the research were compared with calm periods and presented graphically. The conclusions, on tha basis of the research, are clear and evident.

 

I have no significant substantive comments. 

Author Response

Please see the additional

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

My previous remarks and comments are well addressed. But there are some small errors. for example:

1) Eq.(10) is still wrong. I_1 = I_2?

2) L109, L112, only use surnames for citation, please use the proper citation format.

3) In table 1, threshold of GPS is three times large than BDS, when R=1, Why? what is the principal behind? when 0.03 for GPS, while 0.0092 for BDS, does the fourth decimal 0.0002 makes sense? what is the difference between 0.001 and 0.0011 in statistics sense? are the authors geodesists?     

  

Comments on the Quality of English Language

it needs some modifications in general.

 

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

Unfortunately, I still have to disagree with many statements in the revised manuscript; thus, I believe it should not be published. Below, I listed the major concerns: 

 

1. The selection of different GF thresholds for particular systems or satellite types should be clearly described. The GF combination eliminates geometry for all satellites; thus, we should expect the same threshold value as practical. I could imagine a slight difference for GEO/IGSO satellites due to different impacts of plasma motion, but this is not a key factor. Consequently, equation 15 is difficult to understand. The authors should still provide plots with GF threshold as a function of elevation.     
2. If I am not wrong, the threshold for the 5th strategy equals approximately 0.20 m (30-second sampling interval), corresponding to ~ 2 TECu for the GF combination. This is too small, considering the ionospheric variability at high latitudes. Even plots given in answers show ROT values (1-minute intervals) at the level of 10 TECu. On the other hand, it is hard to believe in ROT values over 20 TECu, and here, I would expect cycle slips. The separation of both effects is a challenging task.
3. The natural consequence of low thresholds is a large and, to me, unreal number of cycle-slips. 50% of data with cycle slips has to deteriorate PPP GNSS positioning strongly, or at least part of them are not authentic cycle slips.

 

Since I do not see substantial progress, I recommend rejecting the work again.

 

Author Response

Please see the attachment.

Author Response File: Author Response.docx