Evaluating Total Electron Content (TEC) Detrending Techniques in Determining Ionospheric Disturbances during Lightning Events in A Low Latitude Region

Round 1

Reviewer 1 Report

The topic of this research is interesting. Authors analyzed a huge volume of GNSS-data and familiar with modern literature in this research field. However, I fill that the paper needs more work in order to make it clear and readable.

First of all, the paper lacks the graphical part. Some the key plots must be included into the main text. They have to show GNSS network location and relative GNSS sites distribution. The example (s) of original (unfiltered) VTEC series should be presented in comparison with filtered by polynomial and Savitzky-Golay fit TEC series. We have to see on the same plot, for example as different subplots, time series for site-satellite pairs showing TEC variations during non-lightning and lightning days (events) in order to see differences between them. It is very inconvenient to work with so huge amount of plots given in supplementary materials.

There are some other questions:

1) From the paper it is not clear what disturbances caused by lightning events authors are looking for: some ionospheric response on individual lightning events like separate peaks or oscillations or some background (mean) TEC level change? May be something else?

2) What GNSS site (s) was (were) selected for analysis and plotting in supplementary materials? Lines 120-122 declare that many site-satellite pairs were analyzed. However, we cannot see any site notations in DTEC time series plots.

3) Why authors show all satellite DTEC plots for selected day? Some of them are empty or cover time interval which is not interesting for analysis. Is it possible to show most informative DTEC time series only?

4) What unites were used for DTEC time series plotting? Please, note about it in the text.

5) What GNSS satellites constellation was used - GPS? Please, clarify in the text.

6) In results section frequency term is often used. What kind of frequency is analyzed there?

7) The discussion about DTEC amplitude change after application of different order polynomial or  Savitzky-Golay fit seems obvious because the fitting order controls the band-pass frequency and amplitude value. Higher order polynomial approximation will give lower amplitude and higher frequency output signal. In order to form the best filter for your application you have to know what kind of disturbances you are looking for and in what frequency band they have to be expected. From the Table 3 and conclusion section one can see that you recommend the filter which provides minimal RMSE value of DTEC. However, from the paper text and figures is not clear why this filter is the best for the lightening events (days?) distinguish.

8) Table 2 is also just show the some filters can/cannot resolve for lighting events. However, where the criterion of this selection is given in the paper? Please, clarify this item.

Please, format you reference list and complete some references.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

This communication paper shows an evaluation of Total Electron Content (TEC) detrending techniques to determine ionospheric disturbances during lightning events. This study was done in the Southern China Coastal area using Hong Kong local GNSS network. The paper reads well and can be accepted after minor revision. 

1. "Evaluating TEC detrending techniques in determining ionospheric disturbances during lightning events: Case study of Southern China Coastal area using Hong Kong local GNSS network"-I would suggest to revise the title and elaborate TEC. No need to mention the case study in the title.
2. Line 15: Elaborate GNSS.
3. Line 23: Re-write
4. Introduction is very general. A number of literature review is required. 
5. Section 2: Data should be first and then methods.
6. Advantages and disadvantages/limitation of each method should be mentioned.
7. line 125: Re-write
8. Avoid the word "plots'
9. line 345-46: No need

Author Response

"Please see the attachment

Reviewer 3 Report

Referee’s report on

“Evaluating TEC detrending techniques in determining ionospheric disturbances during lightning events: Case study of Southern China Coastal area using Hong Kong local GNSS network”

by Louis Osei-Poku et al.

The intent of this manuscript is to compare two algorithms for isolating perturbations in GNSS TEC to distinguish the effects of lightning from the effects of ionospheric structure created by other means, including TIDs. Although they give summary tables of some statistics of the perturbations determined by the two methods, it is impossible to judge the accuracy of the determinations or the validity of those statistics because all plots of the detrended TEC time series are relegated to a supplemental information file, seemingly in an undigested format, with output for every event processed. The authors need to choose some typical cases for presentation in the manuscript itself, to insure that such analyses become part of the journal record. It is not sufficient for the authors to dump every case in a file to be archived; they need to curate the data they have assembled. They can present the cases that support their thesis, point out the ones that don’t, and strengthen their argument by showing why those cases are exceptions.

Author Response

"Please see the attachment

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Authors made some changes and improvements in the manuscript. Number of figures is large now.  Nevertheless, some key plots are missing: region of the study, GNSS-network configuration, IPP-tracks surface projections, plot illustrating the processing methodology. Instead of a huge number of figures 3-27 a limited set of the most prominent plots could be shown. All other similar figures must be moved into supplementary materials. The present version of the paper has a lot of typo and formatting problems.

As for the content of the paper it is not clear what figures show the TEC variations caused by lightning events? All days selected for the analysis are characterized by more or less lightning activity. Authors do not provide us in the section 2.4 a clear quantitative criteria for selection of DTEC anomalies generated by lightning activity. From the presented DTEC time series for the non-lightning/lightning days it is impossible in my opinion to detect any TEC anomalies which could be clearly correlated with lightning events as can be seen from the paper of Kumar et al., 2017 (doi:10.1002/2017JA023914). This paper deals with the same region and operates by the same GNSS-network.

According to my opinion figures 28-32 show absence of linear correlation between lightning counts and filtered DTEC data. The DTEC time series also do not show clear lightning events signatures. At the current stage I cannot support authors' statement about SGF 6-120 better work with respect to the polynomial approximation. I cannot see clear evidences about it from the figures.

I am suggesting to authors deeply work on graphical  part of the manuscript and support their conclusions by numerical and visual evidences presented in clear and readable form.

Please, see other comments and questions in the manuscript as sticky notes.

Comments for author File: Comments.pdf

Author Response

Please see the attachment 

Author Response File: Author Response.pdf

Reviewer 3 Report

the manuscript can be published as it is.

Author Response

Thanks for your valuable suggestions to improve the manuscript and onwards acceptance for publication