Thiebaut et al. present an interesting study on turbulence measurements with profiling lidar. The idea to place two lidars with a yaw angle offset and combining them to a lidar with more beams is interesting and innovative. The plots are well prepared and the manuscript is well written. At mutiple points I feel that the description of the results and the methods are a bit unprecise. The comparison to other lidar configurations and techniques on the contrary goes a bit too far in my opinion, because it cannot be justified with the results of the experiment. I thus suggest the manuscript for publication only after major revision. General and specific comments are given below.

General comments

- The database could be better described. A brief statement on wind speed span, median and mean is given, but for example no information about the distribution of wind direction. Is that dataset statistically significant? I think it is, but it is not shown.

- I am not very convinced about the comparison with the 6-beam method. You cannot easily compare the datasets and the lidar parameters are quite different. The comparison of errors and uncertainties on that basis is not sound.

- Some details of the results are not explained in enough detail. For example, the effects for cross-wind variance and how they depend on wind direction, or the differences in the spectra. Why is the low frequency not the same for all methods, why does the variance method have more high frequency noise etc.

Specific comments

p.1, l.21: I do not think that you can say that so generally. There are a lot of people who do VAD with pulsed lidars as well. It has advantages, especially for turbulence retrievals, too.

p.2, l.48: Eberhard et al. 1989 requires a full VAD at 35.3° and provides TKE and the covariances, not the single component variances. Later studies by Smalikho, Stephan, Wildmann and Päschke showed that this method is very accurate, if the lidar "intra-beam" volume averaging effects are corrected in the retrieval. 

p.3, l.63f: There has recently been a release of a commercial 6-beam lidar https://halo-photonics.com/lidar-systems/beam-6x/, https://halo-photonics.com/lidar-systems/beam-6x-windpower/. It also does not take 15s for an instantaneous measurement any more. I saw further down the manuscript that you discuss this instrument, but i think it would be fair to mention here already.

p.4, l.112: Despite the fact of the foundation being quite impressive, I am not sure how relevant it is for this study. Information about the wind conditions at the site (wind rose, etc.) could be quite interesting instead.

p.7, l.159: Also for the collected dataset, some statistics would be helpful here: wind rose, histograms, of wind, turbulence, stability for example.

p.8, l.185ff: Please provide the thresholds for the despiking

p.10, Eq.13&14: $U$ remains the absolute velocity?

p.11, Eq.19: You switch here from vector notation to Einstein notation (I think), without explaining it. That could be confusing for readers and should be explained.

p.13, l.310: I assume that the "virtual kinematic heat flux" was calculated using the sonic vertical velocity and sonic temperature? Thus not directly the virtual temperature. Could be confusing if you use the same symbol as for the average virtual temperature from the WXT530.

p.15, Tab.2 and p.16, Fig.5: Comparing the methods with completely different datasets is not sound. You would have to make sure that you have the same amount of data for all sorts of wind bins, wind sectors, stability classes, which I assume is not the case here!?

p.16, l.355f: can you explain why the spectra differ at low frequencies? 

p.22, ll.434ff: I think the ideas and comparison to other lidar configurations are a bit superficial and not exactly based on results from this study. I recommend to skip them and focus more on the direct findings of the new variance method. 

p.22, l.457: Intermittent turbulence is especially observed in stable boundary layers with strong shear. A violation of homogeneity assumptions cannot be deirectly associated with neutral and unstable conditions alone. Neutral conditions can be perfectly homogeneous over flat terrain, stable conditions can be non-homogeneous with only slightly complex terrain. I think you should be a bit more precise what you mean here.

p.23, l.477: I think this dependency on wind direction should be shown explicitly.

p.24, l.497: You should at least mention that a two-lidar setup doubles the cost at this point.