In their manuscript Yang et al. present a correction algorithm for rotor-induced wind biases measured by a 3D anemometer placed on top of a rotary unmanned aerial vehicle. The authors used CFD simulations of an UAV under different wind conditions and UAV flight speeds to develop their algorithm. The decision to use CFD in this study allows for testing UAV under a large variety of wind conditions and with detail which would not be available if e.g. the study was conducted in an wind tunnel. The study highlights the benefits of measuring 3D wind using the UAV. This is especially important for studies in which UAVs are employed to measure (point-source) pollutant plumes to estimate emission strength, for these knowing the accurate wind speed and direction is crucial. This study brings a valuable contribution to having more accurate wind measurements from UAV platforms.

However, I do have a few major concerns regarding the methodology as well as thoroughness in discussing the possible drawbacks of their approach that I hope the authors would address.

Specific comments:

1. The authors based their study on results from a large set of CFD simulations of a flow around a UAV with rotating blades. For these simulations they used the Solidworks model. What I lack is a reference to literature in which the model is described and also a short description of how exactly Solidworks works, what are the governing equations, why the model was chosen and what are the implications of those choices.
2. In Section 2.4 the authors describe their flow as “turbulent and laminar flow with turbulence intensity of 0.1% and turbulent length scale of 0.012 m” (in a 3.3X3.3x3.3 m3 domain) both of those parameters point to a laminar flow in the atmosphere. Can the authors provide a Reynolds number for their simulations? Then it would be easier to understand what kind of a flow they had in the simulations. If the flow is indeed laminar or low on turbulence, the authors should provide discussion on the implications for their study.
3. In their results authors present time averaged wind fields from which they derive their correction algorithm. However, the actual measured wind is highly fluctuating and dependent on the atmospheric stability and, if close to the surface, the surrounding orography and obstacles. How do stationary results relate to turbulent atmospheric measurements? Additional discussion on influence of turbulence on UAV wind measurements would be beneficiary.
4. I believe the simulations in this study had neutral stratification (it is not explicitly stated). However, these conditions are rarely encountered in the real atmosphere. How applicable is this algorithm for when, for example, the atmosphere is unstable and the vertical wind component is much stronger? Can authors provide a discussion on potential errors which may arise when applying this algorithm.
5. The authors simulate flow at 30 m and 1000 m heights and recommend their correction algorithm for UAV flights below and above 500 m, respectively. However, wind variability is highest within the atmospheric surface layer (typically below 500 m). Why were only these two heights chosen? Would additional simulations closer to the surface give more realistic assessment of near-surface conditions? The influence of surface roughness and stability effects should be discussed.
6. In connection to the previous comment: the authors validate their results by measuring wind using the UAV and comparing results with the measurements from a near-by meteo tower which measures winds at 3 different heights. Firstly, they exclude from their analysis the wind measurements at 30 m due to the influence of orography even though their correction algorithm is based on simulations at 30 m. I think this should be properly addressed in the text. Secondly, the results are presented in Fig. 9 which seems to be showing wind speed and direction comparison at only one height. If the measurements are somehow aggregated, it should be clearly stated how. If not, then the height at which the measurements are compared should be indicated in the text and why the second is left out should be discussed. Preferably wind at all 3 heights should be compared and shown in the results section. Lastly, the authors state that their results show especially good fit for higher wind speeds (they define them as >= 3 m/s, why this threshold?). From Fig 9. It seems that for approximately 20% of data this condition is met. And they present two sets of data of about 15 min with 5 s sampling frequency. Can the authors comment on the statistical robustness of these conclusions and possible implications.
7. As I mentioned above I miss general discussion of limitations and drawbacks in this study. Neither in results or in the conclusions sections are these properly discussed.

Minor comments:

Line 130 (And table 1.) If the UAV can only operate under the true WS of <= 18 m/s, then doesn’t the e.g. fly speed of 14 m/s with 10.7 m/s tail wind exceed that?

Line 140 I think it should be clarified how both (wind) speeds were imposed in simulations. The domain seems too small to have the UAV to actually be moving through the simulations. But if I am wrong this should be clarified.

Line 149-164 It would be good to show the forces acting on the UAV in a diagram. Also it would be good to have a diagram of the UAV, the projection surfaces and the angles which are mentioned in the equations (1)-(5).

Line 165 This paragraph is very difficult to decipher. Is “that” in the first sentence extra?

Line 167 Since the authors make a clear distinction between the many different wind vectors (wind from UAV movement, wind from the simulation, measured wind at the anemometer) they should take care to clarify each time to which wind they are referring to.

Line 172-180 Can the simulation specifications be shown in a table for easier following?

Line 188 What is the depth of the rotating cylinders and why was it chosen?

Line 195-216 This whole section is a bit confusing to read due to the amount of different velocity definitions. What exactly are global domain and subdomain? What is the difference between Vg and Vs exactly? If they are mean velocities, what is the averaging time?

Line 197 What does “convergence of the simulation results” mean exactly?

Line 204-214 Arguably, a schematic here would help to follow exactly what is happening with the velocity components in different reference frames.

Figs 3-5. It is unclear which wind velocity is shown exactly in these figures. Is it movement of the UAV + air wind speed? Why are patterns for headwind and tailwind flipped? Why is crosswind similar to headwind? Might be good to have a “base state” simulation when there is no air or UAV movement, just the propellers.

 Line 275 A “for the cross-wind conditions” is missing from the sentence?

Line 280 Can the authors discuss why only in the crosswind there has to be height distinction for the vertical wind speed?

Fig 6 and 7 Why Du_x is fitted to ux_sensor and Du_y is fitted to uy_sensor but Du_z is fitted to ux_sensor?

Line 370 Wrong reference to definition of A and B? Should be eqs. (24) and (25) instead (17) and (18)?

Line 415 Wrong reference? Equation (23)?

Line 411 “small low bias” can the authors quantify this?

Line 412-417 I would say all three wind speeds in Fig 10 show predominantly northernly winds. The main difference is that Vo has about 15% more wind going to the south than the wind measured at the meteo tower Vt. Vr indeed does not have those 15% bias in comparison to the Vt, however Vr has around 15% higher NW wind component that is not present in Vo and Vt. Is the Vr wind then really showing a much better agreement with the tower than Vo?

Fig 9 Does the meteo tower measure the instantaneous wind every 5s or it gives 5s averages? Also, why impose first the running mean on the UAV measurements before making 5s averages?

Fig 10 The caption does not match the labels (a), b), c)) in the figure.

 First following the review criteria, free comments below:

1. Does the paper address relevant scientific questions within the scope of AMT?

A: Definitely. The paper deals with improving the quality of rotor UAS based atmospheric wind measurement.

2. Does the paper present novel concepts, ideas, tools, or data?

A: The paper does not break new ground; the basic approach is known from oter researchers. However, it is a valuable contribution to transforming more complex methods into practical and operational approaches. From the reviewers point of view this is a very often underestimated aspect in research.

3. Are substantial conclusions reached?

 A: Yes.

4. Are the scientific methods and assumptions valid and clearly outlined?

A: Generally yes, however see suggested major revision below.

5. Are the results sufficient to support the interpretations and conclusions?

A: While the computational part is fine, the experimental validation needs some more description and explanation.

6. Is the description of experiments and calculations sufficiently complete and precise to allow their reproduction by fellow scientists (traceability of results)?

A: Similar comment as on question 5.

7. Do the authors give proper credit to related work and clearly indicate their own new/original contribution?

A: As far as crosschecked: Yes.

8. Does the title clearly reflect the contents of the paper?

A: No. The calculation is done in 3D but the output shown covers only the horizontal component. The reader might expect a 3D wind vector as output.

9. Does the abstract provide a concise and complete summary?

A: The abstract is adequate.

10. Is the overall presentation well structured and clear?

A: Yes. The paper is straight forward structed.

11. Is the language fluent and precise?

A: The work is written in an easily understandable, perfectly adequate English.

12. Are mathematical formulae, symbols, abbreviations, and units correctly defined and used?

A: No mistakes found.

13. Should any parts of the paper (text, formulae, figures, tables) be clarified, reduced, combined, or eliminated?

A: Clarification: See following suggestions for revisions

     No reduction, combination or elimination suggested.

14. Are the number and quality of references appropriate?

A: Yes.

15. Is the amount and quality of supplementary material appropriate?

A: No supplementary material given at the moment.

Suggestions for major revisions / questions:

Lines 67 to 100: While the main conclusion is correct, the authors avoid extracting uncertainty numbers for the measured wind states from the references.

Lines 135ff: The reviewer does not understand, why a matrix of  ground-speed and wind-speed variations was used, instead of varying the airspeed? In other word: How is the groundspeed feed into the simulation?

Line 193, fig. 2, resp. ch. 2.4: The computational domain is described in detail, but I miss a reasoning for the chosen parameters.   

Line 382ff, ch. 3.5: The flight tests are not described adequately in quality and quantity. Thus the data basis of the uncertainty numbers given in line 399 – 401 is not clear. A link to the quite nice fig. 6 and 7 and the uncertainties is missing.      

Suggestion for minor revisions:

Line 249ff, fig. 3 to 5: I suggest a uniform scaling for the color bar.

Fig. 9: Instead of V0, VR, VT better write a full word description into the legend.      

Keep researching!