|This paper titled, Characterizing the dynamic movement of thunderstorms using VLF/LF total lightning data over the Pearl River Delta region, describes the use of lightning location data to characterize thunderstorms in various ways. The authors provided detailed discussion of the methodology related to the storm clustering algorithm, with nice visuals to clearly illustrate what was done. The figures are generally well made and professional looking. However, due to my major concerns with the paper that I outline below, I believe there should be major revisions done. If the authors can address my concerns, I do believe this work can be a positive contribution to the literature.|
I am concerned that the authors have not done a thorough literature review for the background of this study. They state that no other studies have used lightning data to study thunderstorm characteristics, which is simply not true. Below I list several after just a few minutes of online research. I think the authors need to take the time to properly review the literature to understand what research has been completed in this domain.
My biggest concern with the paper is that much of the discussion relates to statistics (max, min, median, etc.) and makes some generalized claims relating to these quantities. However, the sampling size of this study is extremely small with only 8 storms. Because of this, I think the authors need to refrain from making generalized claims.
There seems to be no discussion about whether the results could be influenced or biased by the methodology. The most obvious example relates to the storm velocities (Figure 6). In my opinion, the large variability is more an indication that the 12-min windows are simply too large to properly resolve the fast, but smoothly changing velocity of the storms. There should be more discussion about how there may be biases introduced by the data and/or the methodology.
Finally, I found that the Discussion and Conclusions section to be significantly lacking discussion of the results. It was often difficult to follow what the authors were trying to point out. The authors need to spend more time expanding the discussion.
See my line numbered comments below for more details on my major concerns.
Comments by line number:
Line 15: first distribution should be plural
Line 51: hail should be singular
Lines 58-60: I actually disagree with this statement and feel that the authors did not do a thorough literature review before making this statement. After a few minutes of investigation, I found multiple papers that specifically look at the lightning characteristics over the course of individual thunderstorms. Here are a few that I was able to find within a few minutes of online searching.
Soula, S., & Chauzy, S. (2001). Some aspects of the correlation between lightning and rain activities in thunderstorms. Atmospheric research, 56(1-4), 355-373.
Williams, E., Boldi, B., Matlin, A., Weber, M., Hodanish, S., Sharp, D., ... & Buechler, D. (1999). The behavior of total lightning activity in severe Florida thunderstorms. Atmospheric Research, 51(3-4), 245-265.
Wang, C., Zheng, D., Zhang, Y., & Liu, L. (2017). Relationship between lightning activity and vertical airflow characteristics in thunderstorms. Atmospheric Research, 191, 12-19.
Line 80: The use of heights for classifying IC versus CG flashes is problematic and can lead to many misclassified CGs as ICs. Combining this with waveform classification will improve this, however, this is very little information provided regarding the details or the efficacy of the algorithm. Can the authors provide some literature review as to how accurate this method is as well as the expected false classification rate?
Line 82: This type of network does not inherently detect lightning flashes. Because it is an VLF/LF network, it is detecting charge motion occurring from current pulses. Therefore, there must be some pulse clustering occurring to clustering these pulses into flashes. This clustering methodology is important to this analysis since it can impact the number of flashes and the flash rates. Therefore, the authors need to describe the flash clustering algorithm used in the paper.
Line 89: Cai et al., **2019**
Line 109: wording here is unclear, please rephrase.
Lines 137-142: There seems to be no way for a thunderstorm to split and may affect the results since this does happen in real thunderstorms. Have the authors considered this? I think this is worth discussing in the paper.
Lines 170-171: Check grammar.
Line 190: This phrasing makes it seem like the thunderstorm initiation and dissipation is decided by radar data. Is that true? If so, the previous methodology made it sound like the lightning was the only deciding factor for clustering flashes into thunderstorms. If not, please rephrase.
Line 191: You should state explicitly that the IC and CGs are for the storm only, not the entire dataset.
Lines 192-194: I do not agree with this statement. It is true that the CG rate usually peaks near the middle to end, but so does the IC rate. Maybe the authors meant to say that the IC:CG ratio is highest at the beginning of the storm? I am not sure that is true either, but to show that you could add a curve for the IC:CG ratio on the plots as well.
Line 196: "visions to the horizon" does not make sense. Maybe "The footprint, trajectory, and flash density of thunderstorms" is more accurate?
Line 225-229: I find the discussion of the velocity results to be somewhat biased in that there is no discussion that they could be a result of the method or sampling of the methodology. When observing the motion of a storm in real-time, there is no apparent drastic changes in motion of typical storms. They generally move quite smoothly. I believe the result that the velocity seems to vary so much is more related to the methodology. It could indicate that the 12-min windowing is sampling the motion of the storm too slowly, resulting in these results not being able to resolve the proper smooth changes of the storm.
Lines 268-272: These references all relate to radar data and seem inappropriate to reference the characteristics used for lightning data. As I mentioned above, there are previous papers that have used lightning data to estimate thunderstorm characteristics.
Line 286: Please re-state what the mean was for this study so that the reader does not need recall what it was or to go back and find it.
Line 287: 6-minute time interval? I thought the time intervals for this paper were 12-minutes.
Line 289: This is not a discussion, this is simply stating what others have found. Please include discussion on how this compares to the current study. It sounds like the storms in this study had an average area equal to the maximum area of a supercell (which is defined as an extremely strong thunderstorm). That seems very unlikely. This raises several concerns related to the methodology of the paper. The authors need to provide much more detailed discussion.
Line 294-295: How is this conclusion found? Once again, there is not enough discussion provided for the reader to follow this statement.
Line 304: "Owning" should be "Owing"
Line 318: Are the authors sure that the direction angle are in the same reference frame as this study? Storms in the USA also generally move from West to East.
Figure 1: The colormap label is confusing. Seems to me the top two row colors are exactly the same or extremely similar. Please consider redoing it.
Figure 2: 16:00 is missing. Is that intentional? That is confusing and should be stated in the caption for readers to understand.
Figure 4: "during the thunderstorm process", it is unclear to me what you mean by this. Do you mean the lightning associated within each chosen thunderstorm? please rephrase for clarity.
Figure 5. Similar to Figure 4, you should add the Storm ID so that readers can easily compare. Also, the caption is wrong. The horizontal is actually longitude, while the vertical is latitude.
The work reveals the characteristics of thunderstorm dynamic movement based on eight thunderstorms evenly distributed around the Pearl River Delta (PRD) region in southern China from May 17 to May 23, 2014. They found that the storms initiated in the west of the PRD region, then moved to the east and disappeared after the thunderstorm travels around 106.5 km in longitude. In addition, there are two kinds of distribution to depict the property of valid area, which are one-peak distribution and two-peak distribution. This manuscript used five parameters and found, affected by rivers in the Pearl River Delta region, the motion of storms shows a distinct pattern, as the spread of direction distributes tightly in the range of 0°- 90° and 270°-360°. The overall structure and layout of the manuscript is clear and the experimental design is reasonable. I will suggest it to be accepted after addressing my comments below.
L189: The definitions of one-peak distribution and two-peak distribution in the manuscript are vague, and more detail is suggested.
L196: “Figure 5(c)(d) is not in full accord with one-peak distribution ... after the highest peak”, is that means there may be three kinds of distribution of thunderstorm valid area in the whole evolution processes? Does it need to be considered separately?
L202, 220: The finding that velocity does not match precisely with the VA , which is differ from common cognition about thunderstorms, is mentioned several times in the manuscript. How is the common cognition (example references) and what causes this?
L8, 250: “was” should be replaced by “were”.
L142: The full name of “PRG” should be elaborated.
L187: “3.2” should be replaced by “3.3”.
L241, 245: “duation” should be replaced by “duration”.
L243: “setted” should be replaced by “set”.
L247: “aviod” should be replaced by “avoid”.
L274: “acitivity” should be replaced by “activity”.