The paper "Atmospheric Chemistry Experiment (ACE) Winds" by Wyatt et al. presents a new version (v5.3) and validation of ACE Line-of-sight (LOS) wind speeds. Covering altitudes from 18 to 140km, ACE LOS winds are a valuable data set because direct wind observations in the upper stratosphere to lower thermosphere are quite sparse. ACE v5.3 winds are compared against MIGHTI, meteor radars, MERRA2 reanalysis, HWM14, and WACCM-X. It is found that the new data version is much improved over v5.2. For example, biases are much reduced at altitudes below 50km.

Overall, the paper is well written and recommended for publication in AMT after addressing a few minor comments as detailed below.

SPECIFIC COMMENTS:

(1) l.44-51: Suggestion for future work:

You should compare ACE LOS winds with the Japanese JAWARA reanalysis that was developed in the frame of the ICSOM project. Unlike MERRA2, JAWARA assimilates data even in the upper mesosphere and therefore provides quite realistic winds even at altitudes somewhat above 100km.

https://jawara.nipr.ac.jp/home

Sato, K., Tomikawa, Y., Kohma, M., Yasui, R., Koshin, D., Okui, H., et al. (2023), Interhemispheric Coupling Study by Observations and Modelling (ICSOM): Concept, Campaigns, and Initial Results, Journal of Geophysical Research Atmospheres, 128(11), e2022JD038249, doi:10.1029/2022JD038249.

Koshin, D., Sato, K., Watanabe, S., & Miyazaki, K. (2025), The JAGUAR‑DAS whole neutral atmosphere reanalysis: JAWARA​, Progress in Earth and Planetary Science, 12:1, https://doi.org/10.1186/s40645-024-00674-3.

(2) Is there a reason why you are not comparing to TIDI winds?

TIDI has a much longer dataset than MIGHTI.

(3) l.59-66: You should specify whether data from a free WACCM run were used in your study, or whether WACCM was at least nudged to analysis/reanalysis at low altitudes. This information is important because, depending on setup, WACCM data may be closer or less close to the real atmospheric state.

(4) In L.119 you mention that after homogenization ACE altitude profiles are shifted as a whole to match an analysis of the Canadian weather model at Environment and Climate Change Canada.

Please provide some information how large these shifts typically are.

(5) l.203: What do you think is the reason for the mentioned outliers?

(6) l.207: Where do you think the sunrise/sunset biases come from?

Could this be some thermal drift of the satellite, or an effect of stray light that would be different between sunrise/sunset?

(7) Another bias becomes evident in Fig.6b. At 100km ACE sunset winds are offset by 40m/s with respect to the radars. As measurements are coincident, this should not be an effect of atmospheric tides.

Do you have any idea where this offset comes from?

(8) In Fig.7a ACE v5.3 shows a strong 50m/s jump at 50km, not seen in V5.2, or MERRA2.

How often do such effects occur?

Do you have any explanation for this effect?

(9) About Fig.10a:

It is quite encouraging how well ACE captures the general global circulation patterns!

You should also mention that in September/October the winds in the tropics at 50km (eastward) and 80km (westward) are opposite. This is as expected from the vertical structure of the semiannual oscillation (SAO). See, for example, Ern et al. (2021), their Figs. 2 and 3.

It is also notable that the westward winds at 80km in the tropics are much stronger than in MERRA2. From Ern et al. (2021), Fig.2 it looks like MERRA2 winds are strongly damped above 65km.

(10) Data availability section is missing.

TECHNICAL COMMENTS:

l.26: atom oxygen -> atomic oxygen

l.228: able compare -> able to compare

The paper “Atmospheric Chemistry Experiment (ACE) Winds” shows the updated Atmospheric Chemistry Experiment Fourier Transform Spectrometer (ACE-FTS) v.5.3 winds.  In the paper they compare these winds to the previous (v.5.2) winds as well as the Ionospheric Connection Explorer (ICON) Michelson Interferometer for Global High-resolution Thermospheric Imaging (MIGHTI) winds in close temporal and spatial proximity. Along with data comparison, the paper compares the new ACE wind data to MERRA-2, HMW14, and WACCM-X model winds.  The new winds are improved but still show a 15 m/s bias above 80 km, but the bias is down to 5 m/s below 50 km. Overall the paper is well written and recommended for publication after a few minor comments.

Specific Comments:

In Figure 8 the MERRA-2 profile between 40 and 60 km jumps around 10 m/s between altitudes.  This is surprising for averaged model data over 3000 profiles.  Can the authors double check that these in fact are coming out of the MERRA-2 data and if real, add a statement to the paper explaining why this is occurring.  

Why does Figure 9 not have a legend like the rest of the Figures? The same strangeness in Figure 8 is seen in the differences as well.  

On line 286 the paper states that “It is not reasonable for a model to predict wind patterns within the polar vortex during winter…”.  There are models that reasonably predict this wind pattern (MERRA-2 for example).  This is just a reference to HMW14 and not models in general?

It appears that in Figure 12, there is an error in plotting.  Figures 12a-c are the same as Figures 12d-f.  Please fix this and make sure the discussion of this figure in the body of the work is still correct with the correct figures. 

This study presents a detailed comparison on the ACE Winds v5.3 product with various other data sources and models of horizontal neutral winds in the mesosphere and lower thermosphere. The paper is clearly written and logically structured, and in my view, requires only minor revisions before publication. In particular, the authors should aim to make the comparisons more quantitative, rather than relying primarily on visual assessments when discussing improvements between the v5.2 and v5.3 products. My specific comments are as follows.

Specific Comments:

The title could be made more informative by indicating the specific updates or changes that the paper addresses. As written, it is hard to guess the details of the study contained in the paper.

Overall, the introduction currently reads as a list of different datasets and models that describe winds across various atmospheric regions. As written, it is somewhat difficult for the reader to discern the main focus of the paper. Certain details—such as the spatial resolution of specific models—feel out of place for an introduction, while broader contextual framing is missing. The introduction should more clearly establish what this paper contributes and why it matters within the broader landscape of neutral wind research.

In the discussion of Figure 2, a brief description of the satellite’s orbit and precession would also help orient readers attempting to interpret the figure.

Lines 200–201: The terms “similar agreement” and “better agreement” should be supported with quantitative similarity metrics rather than qualitative, visual judgments. This comment applies throughout the paper wherever such comparisons are discussed. Since a quantitative comparison was performed for the MERRA-2 results, a comparable level of quantitative analysis would also be expected for the MIGHTI, meteor radar, and other model comparisons.

The ±15 m/s adjustment described is somewhat unclear. In Figure 4, it appears that 15 m/s has been added to the sunrise case and subtracted from the sunset case to produce the pink curve. Unlike the individual example in Figure 3, the average agreement between MIGHTI and ACE appears to worsen at sunrise between versions 5.2 and 5.3; if so, this should be explicitly acknowledged and, if possible, explained. The authors should also specify how the ±15 m/s offset was determined—was it chosen by eye or by optimizing an objective measure of agreement? The latter approach would strengthen the analysis and make it more reproducible. Moreover, rather than saying “There is a sunrise–sunset bias of around ±15 m/s,” the phrasing should clarify the directionality, e.g., “At sunrise (sunset), there is a bias of approximately –15 m/s (+15 m/s).” This clarification should be applied in the Figure 4 caption, the abstract, and anywhere else this statement appears.

Technical Corrections:

Line 26: Typo. Should read “atomic oxygen”

Line 64: The acronym CESM2 is used without being defined earlier in the text.

In Figure 2, the x-axis should be limited to 0–360 rather than extending to 400 to avoid confusion.

In the description of the MIGHTI data, the authors should additionally cite Englert et al. (2023), which describes the v05 MIGHTI wind product used in the comparisons.

Line 190: specify that the winds retrieved are horizontal vector winds, not the full 3D wind vector.