In this work, wind measurements in the mesosphere-lower thermosphere region between two instruments using very different techniques are compared. One is a meteor radar with large horizontal observational volume; the other is based on interferometry of airglow emission layers that has high horizontal resolution but lower vertical resolution with changing average emission layer height. From three months of data in polar winter, correlation analysis between wind magnitudes and direction have been carried out, subsequently refining the technique of comparing measurements from both instruments. In the most sophisticated, airglow layer height was variable and radar winds were weighted to match a supposed Gaussian emission layer profile. It was found that the wind magnitudes derived from airglow were a factor of 0.7 of the winds derived from the meteor radar. One possible reason for the airglow winds to be underestimated was explored in some detail, i.e. a modulation of the airglow layer brightness in height due to small-scale gravity waves in a way that gravity wave phases associated with lower horizontal wind are favoured. The paper is well written and explained and the work is very valuable as wind is a critical quantity that is difficult to measure in this altitude range. Validating different techniques is very important. I have a number of minor comments that might help to improve the manuscript and try to summarize the main points before giving a list by line numbers.

Correlation coefficients are calculated but a thorough error analysis with significance levels and confidence limits is lacking. The intepretation is often only qualitative, e.g. "significantly larger", "match well", etc. But to interpret the physics of a tiny increase between two correlation coefficients or lack of it might not make sense if both are insignificant. E.g. I would have expected higher R values in Fig. 7 compared to Fig. 6, but there is no significant improvement, rather a decrease. What is the interpretation then? That over two or three months, the airglow layers cannot be approximated by a Gaussian? But shouldn't over a long enough time frame an approximation by a Gaussian become reasonable? Maybe it shouldn't be interpreted at all, because all R values are insignificant, I am not sure. It is also crucial if the observed differences in wind magnitude are significant or not. Uncertainties should be addressed in more detail. It is also relevant to show that the meteor radar sufficiently samples the large observation volume in order to proof that the larger-magnitude winds compared to the ERWIN measurements are not due to inhomogeneous horizontal wind structure and sampling issues with the radar. The number of meteors has to be evaluated for this.

There is a mismatch between text and figures regarding the dataset (in l. 63 it is stated that data is from Dec 2017 and Jan 2018. The plots show also data of Feb 2018). Please state clearly how long and large the data set is, i.e. time range, number of hours, number of cloudy days removed.. (around l. 170). When the instruments were installed in 2006 and 2008, respectively, and have always been operational, why use only two months of data?  Is the used dataset sufficiently large? Why was this time period chosen, and what are the implications? How were conditions, was there a stratospheric warming? A paragraph should be added about the expected background. What are wind magnitudes expected in this region? Do seasonal averages of the wind magnitude and direction match? What are the dominant processes? Is the variability mostly due to tides? I also would encourage to look closer at the data. E.g. in Fig. 12, I wonder why the values of the green "MR Gaussian" line is sometimes larger than the others when the former constitutes a vertical average. Can the variance then truly be larger? Also, for the meteor radar it might be worth looking into the extreme values e.g. around 15 January - are they truly that large, are they related to some event, or is there some problem with the data?

Regarding the mismatch of wind magnitudes and variances, I have several questions. The authors argue that the averaging should have produced comparable magnitudes and variances, but I am not convinced this is true (l. 348, l. 390). The footprint of the instruments is still different and together with structures of wind below the resolution and unsufficient sampling, this could well result in different variances. Naively I would have expected another outcome, that the ERWIN variances are larger than the meteor radar variances, when the latter are averaged over a larger volume. We know from visual polar mesospheric cloud observations that there is considerable variability on the km scale which likely transfers to the wind field. An uneven spatial sampling due to low meteor numbers might increase the variability of the meteor winds. I also wonder if there is a possible effect due to the horizontal distance due to ERWINs viewing geometry?

In the discussion of the proposed gravity-wave-phase mechanism I was at first confused about the relation of airglow brightness and horizontal wind. The latter is derived from the Doppler shift and thus independent of the absolute airglow layer brightness. The trick here is, I think, the change of airglow brightness along the imaged column, because only the integral is measured and the height information is lost. If there then is a dependence on gravity wave phase, this could introduce a bias. It is a relatively complicated argument and needs to be explained carefully to the reader. As the argument was extended to the meteor radar, it is not immediately clear to me how the occurrence of meteor trails should depend on the gravity wave phase, please explain. Also, please state explicitly for what type of waves this effect is relevant, i.e. vertical wavelengths below 5 km and horizontal wavelengths between 5 and 60 min (?).

I suggest to use the term "observational filter" instead of the here used "instrument filter" (l. 94 and later). For the latter, the reader could imagine some filter built physically into an instrument. The meaning intended however is of the capability of a technique to observe only part of the relevant spectrum of the process that is to be studied, e.g. by a limitation of range or resolution. I think this is better expressed as "observational filter".

I suggest a sub-structuring of section 4 regarding the selected methods that are applied to the data before comparison: Around line 193, "The first", "the second" are mentioned, but "the third" is missing. I suggest to make subsection "4.1 Method A" in l. 211, "4.2 Method B" in l. 240, "4.3 Method C" in l. 262 or similar.

The figures can be improved. Some are hard to read because there is too much data in it such that no good visual comparison is possible (e.g. Fig. 2 and 9), some are redundant and can likely be removed (Fig. 3 and 4), and some can be extended.

l. 1 "winds in the upper mesosphere" it is clear to an expert that meteor radars measure wind in that altitude range, but it should be mentioned in the beginning for non-expert readers

l. 5 "at all three heights" please mention the heights explicitly, maybe in the brackets in line 4

l. 5 Michelson and Fabry-Perot are mentioned. Please make more clear in the abstract how many instruments are used, and what their names and techniques are

l. 6 "higher accuracy/higher cadence" higher than what? The meteor radar? The past? Please state this clearly

l. 8 "airglow and radar winds" I understand the meaning, but more correctly, it is wind measured by means of meteor radar or Doppler shifts of airglow emissions.

l. 65 please make the difference between Michelson intereferometers and Fabry-Perot more clear. Do you refer to previous work/literature that used Fabry-Perot interferometers, and the difference in this work is the use of Michelson interferometers?

l. 107 how are the accuracies determined?

Fig. 1 could be extended by adding three Gaussians centered at the relevant heights for ERWIN on the left (and the green shading removed) and the height distribution of meteor from the dataset used in 3 km bins on the right. That would make it very clear and helpful.

l. 123 please make clear what the 5 min cadence refers to. Are measurements in the five directions "simultaneously" (l. 114, that is 5 min of data every 5 min for each direction), or is it "sequenced through" (l. 123)? How much time does the calibration take?

l. 136 what are "140 bin quadrants"? Please explain more about how the uncertainties are determined. Do they change with time?

l. 180 please give the values of the determined variances

l. 209 what is meant by "generally complement the MWR winds well"?

Fig. 2 does this figure show ERWIN winds filtered with a 90 min running mean? I find the type of visualization with the arrows very difficult. There is so much variation on short time scales, it is impossible to compare the time series. Maybe consider making eight color plots for the radar and the three transmissions and meridional/zonal?

l. 220 how was the correlation coefficient determined? Please state significance levels and confidence intervals.

Fig. 3 and 4. I see little value in these plots, as Fig. 5 summarizes the results. It would be an option to remove them. If kept, significance levels and confidence intervals should be added.

l. 229 it is not "roughly Gaussian".. Please add the Gaussian fit and parameters in the figure.

l. 237 R=0.77 might not be an excellent correlation

l. 266 please motivate the use of a 2-day window. What processes act on these scales?

l. 279 consider adding a line plot of the six time series of emission heights and SABER (in one plot) that might visualize better the agreement  ("basic pattern") in addition to the figures of Fig. 8

l. 283 what is meant by "variations of close to +- 5km" and "total variation close to 10 km"?

l. 287 this is a very short paragraph and makes a somehow unfinished expression, as does Fig. 9. It is hard to read anyting useful from this plot. I cannot see that the "agreement is good" and "match on average".

l. 320 please re-evaluate whether approaches are good based on significance levels and confidence intervals

l. 329 "wind uncertainties" please add the magnitude of the uncertainties in the respective earlier section

l. 361 "is significantly" -> "is about an order of magnitude"

l. 361 remove "temporal" in "the temporal variability"

l. 361 please add an interpretation to this finding. I guess it makes sense as ERWIN implies a vertical average. The conclusion then would be that there is significant variability on vertical scales between 3 km (radar vertical resolution) and about 5 km (ERWIN vertical resolution)?

l. 363 Again, I am not convinced that comparable variability can be expected due to the averaging.

l. 366 the magenta line of Fig. 12 left panels could be added to the right panels for direct comparison as the y axis have different scales.

l. 378 is this sentence true? It is not the height-resolved MR winds against the temporal variability of the ERWIN winds. It is the vertically averaged MR wind against the (inherently) vertically averaged ERWIN winds.

l. 429 what is meant by "For lambdad_z >~ Delta z the radar height averaging should accomodate the gravity wave variations"? In that case, there will be a bias?

l. 447 "The background profile is unlikely to have significant small scale variations" doesn't that depend on the definition of the background profile? It was defined to be constant in l. 413, so by design it does not have small scale variations. Also with real data, it depends on the separation into perturbations and background how smooth the background is.

Eqn. 4 Can you give expressions for the two summands on the right. It is then easier to understand how the first summand depends on alpha as aruged in line 450.

l. 510 "The periods and wavelengths of importance" please give the values, are they periods above 5 min (the measurement cadence) and below 1 h (the radar temporal resolution) and vertical wavelengths below 5 km (the width of the airglow layer), is this correct?

l. 512 what is meant by aliasing issues, please explain.

l. 512 "with observation geometry relative to gravity wave phase fronts" do you mean the tilted beams?

l. 521 "world leading measurement accuracy" this should be shown with the relevant citations of comparable instruments in the introduction or second section

l. 545 "As noted above, it is unlikely that radar winds are biased positively" Where was this noted? Please repeat the argument. Why is it unlikely?

Minor changes:

l. 355 "In this figure, …" two sentences can be removed

l. 349 "spatial averaging" -> "vertical averaging"?

l. 414/415 this is a complicated sentence, please rephrase.

l. 423 Delta z is not defined

l. 464 there are no panels a, b and c labeled in Fig. 13

l. 532 add "(Fig. 10)" as reference for the value 0.7

Fig. 6 please add the selected altitudes to the figure labels or caption

Fig. 8 it is hard to see the difference between the magenta and red crosses

Fig. 8 why are the top of the profiles 0.0 (green)? Is there not data? Then please consider using white color.

Fig. 12 "nominal winds": please add "(at 5 min resolution)"

Definition of abbreviations: l. 3 "ERWIN" is not defined at this stage; l. 62 "MWR" not defined. "Meteor wind radar winds" is double, maybe just "MR winds"?; l. 87 "MLT" not defined.

Typos: l. 1 "two" is double; l. 10 0.3 m/s; l. 147 double "in"; l. 216 missing word: ", and all those"?; l. 239 please add units to the values for slopes and intercept; l. 316 missing word "cardinal wind direction"?; l. 395 double "the"; l. 413 verb is missing; l. 553 "fields"; l. 556 double "a"

A copy editor may help with hyphens and comma, I just give some examples:

Use of hyphens: wind-measuring instruments, field-widened, airglow-weighted, gravity-wave-airglow-brightness-weighting (?, maybe better change that sentence..), gravity-wave-associated-wind, layer-weighted, space-based, ground-based, multiple technique -> multi-technique, airglow-radiance-weighted, three-body, …

Comma: l. 11: On average, …; l. 39 Often, …; l. 114 remove comma; l. 167 remove comma; l. 235 "Figure 6, contains" remove comma, l. 250 "the height, and thickness" remove comma; l. 328 "comparisons, need" remove comma; l. 544 "two techniques, implies" remove comma; l. 573 "these, are" remove comma

"Wind comparisons between meteor radar and Doppler shifts in airglow emissions using field widened Michelson interferometers" by Kristoffersen et al. deals with a comparison between two completely different wind measurement techniques in the mesosphere and lower thermosphere region. Their approach is more advanced than other existing comparison studies between meteor and optical methods by introducing the effect of airglow brightness weighting, which I have long been interested in but have had no opportunity to test myself. I enjoyed very much reading the manuscript and do not have a major concern about what have been done in the present study.

However, the obtained results are still not conclusive and there could be more to be done. I would like the authors to even step forward and to think about the following approach additionally, which will simplify the comparison setting and provide further insight into the difference of the two techniques with less restriction. A major obstacle that makes the comparison difficult is the difference in temporal and spatial resolution/averaging between the two techniques. Although the authors use 90 min averaged and some hundred-km horizontally averaged winds in this study, I would directly use the 5 min ERWIN winds and instantaneous meteor echo radial velocity without conducting such temporal and spatial averaging in order to avoid/minimize unwanted effects caused by small scale gravity waves described in the discussion and conclusion.

The number of meteors detected within the FOV of the ERWIN may be small, but by using meteors within an area such as 10 degree diameter centered around the ERWIN FOV the number will be enough to make a direct comparison with only a minimal spatial averaging because you have two month long data set. Some minor correction of radial wind values for meteor echoes will still be necessary considering the elevation and azimuthal angles of meteor echoes. Further, a frequency spectral analysis of ERWIN radial velocities for individual four FOVs will also be what should be tried to see how the spectral values behave, especially in the high frequency range. Because of their large off-vertical angles most contribution to the radial winds are from horizontal winds and the effect of vertical winds to the spectra will be negligible.

I strongly recommend that the authors try such approach and make the present study even more attractive to potential readers.

MINOR COMMENTS

Abstract

 line 3

   ERWIN is not defined.

 line 10

   0.3/s > 0.3 m/s

lines 85-86

  "They consist of ..... cross section"

  The meaning is not clear. Grammatical correct?

line 138

  "Kristoffersen et a. (2021)" should be

   [Kristoffersen et al., 2013]

line 143

  one "]" is missing.

line 147

  one "in" is redundant.

lines 166-167

  "[ ]" is missing.

Figure 2

  Information on direction is missing.

Figure 8

  Red and magenta are confusing.

  White can be a better choice for one of them.

lines 342-343

  "2-D fits to many meteors to set the zero"

  The meaning is not clear.

Figure 12

  It is hard to tell apart the lines in the left panels.

Equation 2

  delta_z is not defined.

  delta_z/lambda_z in front of sin is reversed if my calculation is correct.

Equation 4

  dz is missing in the numerator.

  One ")" is missing in the denominator.

lines 447-448

  A conjunction is probably missing.

line 456

  I believe that "sigma approaches zero" should be "sigma approaches infinity" or "lambda_z approaches zero". If not, the description in this line is not immediately evident to me.

Figure 13

  What alpha and delta_u are used for the evaluation?

  The letters "a,b,c,d,e and f" are too small and hard to recognize.

  Figure caption: "pi" should be in Greek.

line 482

  Lots of expressions such as "small vertical scale gravity wave horizontal variability" are seen in the manuscript. They are not so easy for a non-native English speaker like me to follow, being vague in what is the subject/object or adjective. Expressions with some moderate use of prepositions would be preferable.