The paper presents the results of a search for a specific dynamical pattern in a very large radar dataset of PMSE. The structures that might be related to mesospheric bores are characterized by their occurrence, dimensions, velocities etc. A focus is on events with large vertical velocities. The study is very valuable in determining how often such structures occur and what their properties are. I have a number of minor comments and questions that I hope help to improve the paper.

I suggest to add more detailed definitions and method descriptions. One example is the vertical extent (l. 307). Is this the distance from the lowest to the highest altitude with significant PMSE detection? Is it the same as h1 (Fig. 7)? I also didn't find the resolution in altitude and time, i.e. the size of the "PMSE-pixels". This is an important information for someone who tries to replicate the work. As the patterns were detected manually, it is crucial to describe the process as detailed as possible for someone to repeat it. How exactly were meteor occurrences and system-induced artifacts discriminated from "good" data (l. 119)? How often did these artifacts occur? Please give more details on how the manual search was carried out (e.g. 20 min periods, +- 15 m/s, look for a minimum distance between h0 and h1, or for simultaneous up- and downdrafts of x m/s amplitude?). Some of it was described around l. 344, but it should be put into a dedicated paragraph in the Method section. Were there up- and downdrafts that were not simultaneous but slightly shifted? It would also be interesting to see a compilation of all or the most extreme events, e.g. those with a widening of 7 or beta of 6. Maybe this can be shown in a supporting information, or a zip file. It could be helpful for future research.

The discussion leaves the impression that the work is unfinished, as it is a bit vague and unfocused (e.g. "the significance is yet to be determined", or the vague comparison with Hozumi et al., 2019). Suggestions for analysis that would be required to explain the origin of the structures are made but not carried out. How many events have simultaneous lidar data, and does that indicate temperature inversion layers? (For further investigations, it could be helpful to provide a list of the dates and times of the 707 events). Is there data on horizontal wind? It says "kilometer-scale" in the title, but no horizontal dimensions of the structures are estimated in the text. Inter- and intraannual variation of the occurrences of the structures are shown but remain fully unexplained. The diurnal variation of the occurrences of the structures is shown together with the diurnal variation of PMSE (Fig. 6), but the relative occurrence would be more interesting. Is there, or isn't there a significant diurnal or semidiurnal variation left when accounting for PMSE occurrence? Sect. 4.3 should be rewritten based on relative occurrences. Attention should be paid to units or resolution when stating occurrence rates. It is better to say "structures were observed on 33 out of 100 days on average"  instead of "33%" (l. 245), because when counting occurrence or non-occurrence not on scales of days but on scales of 20 min or less, the occurrence rate is much smaller than 33%.

Comments by line number:

l. 11 please add the total duration of the events, is it about 200 hours?

l. 17 "highlighting their extreme nature": if so, you could mention that the distribution is non-Gaussian

Fig. 4 The examples show different numbers of oscillations, e.g. three in Fig. 4a. I would thus label t0, t1, t2,… (and not t0, t0, t0) to include a counter for this number in addition to your "d = duration" which is actually the period of one oscillation.

In Fig. 4c, only a slight variation can be seen in the evolution of the upper boundary, but the velocity measurements show a clear oscillation for several periods. The amplitude of these oscillations could also be an interesting parameter (it can be seen in the color, but a color bar is missing).

l. 120 Modelling is mentioned briefly, but it doesn't become clear what exactly was expected in terms of occurrence rate from the modelling.

l. 185 in addition to the mean and standard deviation, the kurtosis and skewness could be stated, that indicate in what way the distribution differs from a Gaussian

Fig. 6b please calculate and give the percentages of the one-, two- and three-oscillation classes in the text, e.g. 80 %, 15 % and 5 %, and the same for the high-velocity subset

Fig. 8a A logarithmic y axis might show better the extreme events with low counts

Fig. 8a Please add the Gaussian to Fig. 8a. Then one can directly see where it differs. I expect to see long tails, i.e. the extreme values are much more frequent than if the distribution would follow a Gaussian.

In Fig. 8b you could show the Gaussian with the same vertical axis as on the left (so not scale it to fit in the window). So the peak will be way outside the plot, but then you can compare the tails, which is the interesting part. I think it is fine to show 8a with log y axis and 8b with linear y axis from 0 to 60 counts.

Fig. 8b if the figure shows the histogram of the "maximum vertical velocities of varicose-mode events", then the total number should add up to 707. There are however many more. Is this the maximum w per profile? If so, what is the temporal resolution of a profile?

Minor:

l. 29 remove ", e.g. the so-called"

l. 30 change "high" to "mid" (latitudes). NLC occur at high latitudes, but can only be visually observed in twilight (not at night!) and thus from a mid latitude viewing towards north. At high latitudes the sky is too bright for naked-eye observers to see them.

l. 66 the reference to Taylor et al., 1995 should be added here as well

l. 72 sentence with "tradionally" and "outliers" is double

l. 76 delete "even"

l. 105 the last part of the sentence is missing

Fig. 1 remove "of the highlights" in the first sentence of the caption

Fig. 3 The labels a) and b) are missing in the figure.

Fig. 3 For a vertical beam, range equals altitude, then I would write "altitude" on the y axis instead of range, and "vertical velocity" instead of "radial velocity".

Fig. 3 Is it Fig. 3a or b that the velocities are missing? Please check that the text is correct

Fig. 3 Please add the date of this observation

Fig. 4 Is it intentional that the colors appear somewhat unsharp? If not, my hint is that it is related to a problem with resolution when converting to or from postscript.

Fig. 4 please add a colorbar

Fig. 4 dot missing at end of caption

Fig. 4 caption second line: change downwards to downward

l. 165 "2.5% in June and July, 0.3% in May and 1.0% in August"

l. 193 remove "in" after "observed"

Fig. 7 Horizontal axis numbers are easier to read

Fig. 8 Labels a and b are missing in figure

l. 195 in a number of places, you explain about the more intense colors and thicker outlines of the bars for the values above 15 m/s. I would maybe mention this in the first caption where it applies (Fig. 5 I think) stating that it is done like this in the following figures also, and maybe once in the text, but not over and over again.

l. 200 delete"," after "found"

l. 200 delete "also"

l. 202 no brackets around the two values

l. 211 occurr -> occur

l. 295 delete "relatively narrow (". Just give the numbers. Some readers might not consider 5.4 km to be narrow. Or, if you want to stress it, you should say compared to what the width is narrow, e.g. compared to other radars of that type if that is true and potentially important e.g. to find out if the structures can or cannot be found in other radar datasets.

l. 398 "heightest" -> highest