General Comments:
While previous studies have established the relationship between polarimetric variables and lightning, in this paper authors have suggested the potential application of ZDR and KDP columns identified using 3d mapping columns method to predict lightning activity. This study initially presents a detailed analysis of a single storm with multiple cross sections of polarimetric and microphysical variables. Additionally, it includes timeseries and cross-correlation analysis of lightning frequency, ZDR and KDP columns, radar-derived hydrometeor variables and microphysical processes of 14 additional storms in a variety of CAPE environments ranging from 400 to up to 2500 J Kg-1.
This revised manuscript has addressed all previous comments, but there is still substantial room for improvement in the revised content. The recommendation is to reconsider this manuscript after the authors address the following concerns.
Specific comments:
1 Section 2.1 needs a brief comment on how storm cases were selected. Were there only 15 isolated lightning producing storms in two months period (between 29 May 2016 to 27 July 2016) in the GZ radar and LFEDA domain? This explanation is important to justify the authors’ claim, ‘lightning is not observed in the absence of a ZDR column and a KDP column is not observed without ZDR column’ (line 594-595). Since this statement may not be universally true given the analysis of limited number of storms, the claim, needs to be rephrased to include additional factual words such as limited number, isolated and over South China and can also potentially be combined with lines 597-599.
2 Line 451-508: The discussion of the case study using Figure 7 needs more clarity. Following comments are presented in the context of Figure 7.
Line 458-460: Was the overlapping of ZDR column with reflectivity core persistent between 17:48 CST and 17:54 CST? If yes, it would be beneficial to include 17:54 CST in Figure 7 to explain ZDR column structure prior to initiation of lightning.
Line 460: The beginning of the riming process is not ‘obvious’ if we ignore Figure 7 e2 since both the ZDR [b2] and KDP [c2] values above melting level are quite low indicating rarity or absence of large supercooled raindrops. Based on the presence of riming process in both e1 and e2, it is possible that it was initiated before 18:00 CST.
Line 461-462: The increased reflectivity and its altitude are not sufficient to explain intensification of the updraft, especially with collapsed ZDR column. Although cold cloud processes may have increased, low ZDR [b2] and low KDP [c2] and hailstones [d2] in high reflectivity region above melting level together indicate that observed high reflectivity could be due to dry hailstones present in the examined volume. On the contrary, weakening of updraft and strengthening of downdraft may have occurred considering collapsed ZDR columns and heavy rain below freezing level after 18:06 CST until 18:24 CST. This further makes referring lightning peaks to the impulse in the conclusion (line 764-765) ambiguous as strengthening of updraft is questionable as discussed in this case.
Line 490-491: Referring to panel b1, the ZDR column ‘partly’ overlaps with the reflectivity core.
Lines 496-498: Authors tried to justify the second impulse event with invigoration of updraft due to freezing of raindrops in mixed phase region. This hypothesis (also identified as mixed-phase invigoration or cold-phase invigoration) is widely criticized in the literature (Varble et al., 2023: https://doi.org/10.5194/acp-23-13791-2023) due to the relatively low magnitude of latent heat of fusion. The increased loading of updraft due to large hydrometeors compensate the marginal increase in updraft strength due to release of latent heat. So, observed changes in updraft strength in this case cannot be justified if it were due to natural intensification of storm due to meteorological factors or due to cold-phase invigoration.
Line 504-506: Following argument for Lines 496-498, the presence of ZDR column within reflectivity core cannot be justified as indicator of imminent invigoration. For example, ZDR column was within reflectivity core at 17:24 CST [Figure 4a] but it disappeared at next scan at 17:30 CST [Figure 5a] which would not be expected from a strengthening updraft. A strengthening updraft should be able to support and perhaps deepen the ZDR column. Further, since ZDR column is where supercooled liquid drops and wet graupel are lifted above melting level and reflectivity core is where collision and fallout typically occur, ZDR column within reflectivity core can thus be an ideal indicator of active storm electrification and not invigoration.
3 How did authors determine the beginning and end of storm cloud lifecycle? Please classify if there is any reflectivity threshold considered to identify a particular radar scan time as the beginning and end of storm lifecycle. Objective identification of these points is important to justify the cross-correlation analysis results. As a suggestion, time series alignment can be well compared if matched with peak updraft strength or first lightning occurrence. Line 770: Figure 18 captions can be discussed in more detail. Use the observed data to make cloud lifecycle more definitive by including discussion of formation, mature and dissipation phases.
4 Is autocorrelation removed from each time series before cross-correlation as ZDR column height, graupel mass, and lightning flash rate etc. can show strong temporal autocorrelation during storm lifecycle. The cross-correlation without this type of correction may overestimate the strength of relationship.
5 Line 595-597: Presence of both ZDR and KDP columns at same instance may not necessarily result in highest flash frequency. Although this statement is true for case #11, other cases (#1, 2, 3, 4, 5, 7, 8, 11, 12, 13, 14) also have both these columns co-present at multiple times but with only limited flash count. In case #8, the highest lightning frequency was observed in the absence of both ZDR and KDP columns. No lightning was observed when these two columns were co-present in this case. Therefore, authors can try to accommodate such exceptions in their language while stating any such observations.
6 Line 771-773: Again, following earlier argument for lines 451-508, ZDR column within reflectivity core may not always indicate convective invigoration. At 17:24 CST [Fig 4a] and at 17:42 CST [Figure 4b] the ZDR column is within the reflectivity core, but it was not followed by imminent invigoration and onset/increase in lightning activity.
7 Line 783-785: The statement “high temporal resolution observations of phased-array radar may decrease the uncertainty.” is too general. Please rephrase it to be more specific. Will shorter volume scan time increase the lead time of more than 6 minutes or decrease it? Also add at least one more sentence on why and how you think it will increase or decrease it. Radar volume sampling depends on scanning strategy adopted, while it is good to comment on phased-array radar, there are still a lot of conventional dish radar out there which can benefit from this study. Therefore, a short explanation will be helpful.
8 The conclusion and discussion section will need to be updated based on the comments above. Extrapolating conclusions to all storms in general terms introduces ambiguity and does not account for exceptions (refer to specific comments #5, #6, #7, #8). Therefore, only specific conclusions based on the analysis of 15 observed storms are encouraged. Similarly, discussion can be made more direct and clearer by mentioning specific microphysical or dynamical variables and processes considered in the study instead of referring to them with these general terms.
Technical Corrections:
1 Line 506, 779: Word Kdr should be KDP.
2 Line 767-768: This line should be a part of Figure 13 caption.
3 Figure 5: Mention case #1 either on figure or in caption and indicate total flashes and CG flashes similar to Figures 10, 11.
4 Figure 7 (d1-d7), Figures 10, 11 (c1-c7): The full names for acronyms of hydrometeor types are not found in the text. Please mention it in the figure captions or at line 455.
5 Figures 10, 11: All subpanels in the figure need to be replotted as the text including labels is not visible when zoomed in.
6 Figure 12: Please explain the meaning of positive/negative lags in the captions.
|
The authors analyze a single case study of an isolated thunderstorm over land to the northeast of Guangzhou, China. Analysis of differential reflectivity and specific differential phase columns over the lifecycle of this storm allows the authors to analyze how the cloud microphysics lead to lightning, and the lead time that polarimetric radar allows in inferring the onset of lightning.
The case study largely repeats previous findings. There are a few valuable advancements in analysis methods (column identification methodology; inferring supercooled rain water content using a method from the late 1990s / early 2000s; lead time calculations by different methods). There are also some process inferences related to different pathways by which lightning might be produced that could be valuable and clarifying, but the universality of which is hard to judge on the basis of a single case study.
The authors have therefore engaged substantively in an ongoing tradition of analysis of polarimetric radar and lightning signals, with fair-to-good scientific significance, and good scientific and presentation quality. Below I note additional areas that could improve the manuscript, adding some missing information and clarifying the interpretation.
Major comments:
The authors do a nice job of reviewing the literature. I wanted to also mention our just-published paper, Bruning et al. (2024, 10.1175/MWR-D-24-0060.1), which pursues a very similar analysis on a large sample of storms. The authors’ detailed look at the time-series perspective here is valuable (and something we did not yet do), and I would be interested to see where this study fits in the distribution of lightning and polarimetry of storms sampled by Bruning et al., which were probably similar small, isolated, subtropical storms.
355-57: It is not clear to me that the Zh signal is better related to lightning. – for instance, the Zh signal is quite noisy, while there is a very clear max in high LWC values just before each of the peaks in lightning that is much less noisy – and the authors conclude later that the LWC signal is the most robust. So this claim confused me.
369: how does the collapse of the column result in an increase in lightning if graupel (which is thought to be necessary for electrification) is inferred as decreasing or absent in the column? Further discussion of the process would be valuable here; there are some hints in the discussion/conclusion section here, but I felt that further information and data was needed to verify the interpretation of the two different pathways to lightning the authors have identified.
393: note, however, that the correlation with Zdr is relatively large and increases (0.6) for about 20 min before the maximum in lightning, but falls off rapidly by 12 min after the lightning increases. From a practical point of view, the timing of the maximum correlation is less important than a trend toward confidence for lightning, and so in that sense the Zdr signal is more helpful.
423-6: These correlation coefficients do not seem different enough to allow the authors to say one is best, especially on the basis of a single case study. Values all >0.8 are quite high for each of these variables.
449: After studying the lead times and identifying and emphasizing a 6 min lead time in their results section, the authors return to quoting the 36 min lead time in their conclusions, which does not seem supported by the detailed analysis the authors undertook. Of course, the 36 min lead is there in the data, but it is not well-correlated to lightning. Many moderately vigorous storms will produce a small Zdr column without going on to produce lightning. Likewise on 476-477, I would be reluctant to forecast lightning on the basis of a 36 min lead - that cell is simply one to keep an eye on for future lightning.
Fig. 12: the authors indicate that no Kdp column was present in their data, but do not show Kdp in Fig. 6. I would like to see further data on this, as it may explain the relatively fewer cases in Bruning et al. (2024) that had Zdr columns and lightning but did not have a Kdp column.
Minor comments:
31: The grammar implies lightning flashes can be detected with polarimetric structures; this is not directly possible. The polarimetric signatures are proxies for lightning with some associated error. Please rephrase.
37: “establish” — this study is not the first to use this method, as many of the authors’ citations show. “Build on” or “improve” would be a better choice, since “establish” implies that the authors have made a pioneering advancement. There are some thoughtful adjustments to past methods here, but they are incremental refinements.
124: “later” - do the authors mean a time scale immediately following the Zdr column (~5 min) or subsequent updraft pulses in a multicellular sequence (~20-30 min per cell)?
125: “attempted to determine the constraints of“ should be “attempted to constrain”
140-141: “therefore the correlation coefficient … was not high.” What does “therefore” mean here? It typically indicates that a conclusion has been reached, so the facts supporting the conclusion need to be stated first. They seem to be in the sentence following “therefore”.
175-181: what Kdp calculation method was used? Kdp is a very noisy measurement, and so is very sensitive to algorithm design and configuration choices.
220: here and throughout the paper, melting level is preferable, since melting always begins at this level for any hydrometeor but freezing might not.
221: What are other parameters (CAPE, etc.) of this sounding? They would be helpful in placing this storm in the context of other environments globally.
227: “automatically” should be “automatic”
271: A new sentence should start after “(Figure 2e,f)”.
281: “resulted by” should be “resulting from”
447: I suggest dropping “inappropriate”. Any algorithm choice requires some judgment, and reflectivity thresholds have a sound physical basis and are in wide use. Of course, using fewer or improved variables and thresholds is also good, and in that way the authors have made a nice methodological contribution, but “inappropriate” is unnecessarily harsh.