Performance Assessment of PPP Surveys with Open Source Software Using the GNSS GPS–GLONASS–Galileo Constellations

Round 1

Reviewer 1 Report

Corrections are given in the PDF document.

Comments for author File: Comments.pdf

Author Response

Please see the attachment

Best regards 

The authors

Author Response File: Author Response.pdf

Reviewer 2 Report

• A good number of previous works on the same topic is presented in the literature, but it is not explained how this work is related to them. What is the innovation brought by this work?
• The introduction does not state the scientific goal of the authors. It is clear that multiple constellations can improve the accuracy of positioning, but the authors could have tried to describe whether the obtained improvement were expected (according to some theoretical model, for example) or why there are some configurations that yield different results (e.g. stability of GPS+Galileo is better than Glonass+Galileo).
• In the experiments GNSS data have been collected, and processed with an open-source software and using existing models for the troposphere. What has been developed by the authors apart from this procedure?

Author Response

Please see the attachment

Best regards 

The authors

Author Response File: Author Response.pdf

Reviewer 3 Report

The paper "Accuracy of PPP surveys with open source software using the GNSS GPS-Glonass-Galileo constellations" focus on evaluating the performance of multi GNSS PPP processing with the open source software library RTKLIB. The topic is timely and the use if RTKLIB is of great interest to many users as an easy accessible alternative to the proprietary solutions provided by the GNSS manufacturers.

However the paper has some major limitations that heavily reduce the impact of the presented results. For that reason I will not focus on any details, but rather on the main weaknesses of the paper.

General remarks:

1. All qualitative metrics are presented in the coordinate domain, thus making it very difficult to assess the implications of changes made in the observation domain.
2. All qualitative metrics are expressed in terms of precision, rather than accuracy as the title suggests.
3. The paper lack somewhat focus on what it actually wants to test and there is a lack of reasoning for the chosen experimental design.

Introduction

The introduction chapter gives a very wide background to results from various PPP processing experiments conducted earlier. However, it is somewhat hard to see the relevance of all these papers to the chosen experiments. I suggest the authors focus the introduction more towards the techniques actually used in RTKLIB.

Experimental design

It is hard for the reader to see the role of the receiver in test B. There is a need to a better explanation why you choose to include these observations in the analysis to the reader.

Currently the RTKLIB project is continued on GitHub as version 2.4.3 b33. However, as there exist many versions of the software there is a need to state the exact version for others to be able to reproduce the results, e.g. 2.4.2 p13

The performance of PPP-AR provides no information to the reader besides that it is "experimental" and "unstable", even though this is listed as one of the three main topics to investigate in the paper.

Results an discussion

The outlier detection based on changes in the estimated coordinates must be reconsidered as it very inefficient to detect observation errors in the coordinate domain. Outlier detection in the observation domain is already implemented in RTKLIB.

Fig 8-11 are unreadable and needs to be reformatted.

Author Response

Please see the attachment

Best regards 

The authors

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

The authors have satisfactorily addressed my previous comments.

Author Response

Please see the attachment.

Best regards 

The Authors.

Author Response File: Author Response.pdf

Reviewer 3 Report

General:

Thank you for the revised version of your paper. The paper looks much better now in terms of language and level of details, but the general critical arguments concerning the experimental design is still not fulfilled. To my opinion the proposed idea is very interesting to the reader, but the chosen approach to quantify the actual performance makes the novelty of this paper rather low.

RTKLib:

Note that your chosen version of RTKLib (i.e. 2.4.2 2013/04/29) is completely obsolete and is only listed for archive purposes. This goes especially for the Galileo part of the software which was very immature in 2013.

The current 2.4.2 p13 version is developed on GitHub.

Figures:

The figure 8-11 are still unreadable in the sense that it is not possible to read the fonts in the figures. Including a legend doesn't help the readability. However, note that you can produce these figures yourself from the result files in RTKLib. That way you can better control the layout of your figures.

Author Response

Please see the attachment.

Best regards

The Authors

Author Response File: Author Response.pdf