Original Submission

 HSRL-2 Retrievals of Ocean Surface Wind Speeds

1.1. Recommendation

Major revision

1. Comments to Author:

Overall opinion: This paper demonstrates the HSRL-2 retrieval of ocean surface wind speeds based on the HSRL-2 measurement principle relying on the wave-slope variance determination and dedicated campaign of comparison with the airborne and dropsonde data. The topic you cover is a key for understanding the response of ocean surface reflectivity to changing conditions of ocean and elucidates previously unknown aspects on ocean surface wind speed using HSRL. You also have nearly perfect instrumentation to address this issue. However, this study is not ready to be accepted due to the following setbacks: the plain title that carries very little information about the exact, less broad scope or the actual content of the study, the lack of quantitative information in the abstract (you are not convincing by simply reporting four numerical arguments as your main proofs), poorly structured methodology with significant gaps (did you specify the angular specifications of HSRL-2 system for instance?) within and most critically, unconvincing “results” section due to poor structural choices and omitted numerical arguments in some cases when you speak about correlation. Please revise the manuscript using the comments below.

2.1. Comments:

Title: The title is overly-general. It sounds like it’s a white paper on HSLR retrieval on ocean surface winds. First, I am not sure whether it is a good idea to use unexplained acronyms even in the journals like AMT oriented on a very specific niche of experts. Moreover, what have you exactly done with HSRL-2 retrievals, introduced them? Evaluated? For how long period of time, etc? Your title does not reflect this idea and is therefore not a good title choice.

Abstract: The abstract is not convincing in the present form because of the following issues:

• No Justification. Why it is important to understand, for instance, sea surface reflectance or surface wind speeds over ocean using HSRL? Which research gap you will close by bringing new knowledge on this topic? The statement, explaining this aspect should start your abstract from my point of view if you think about general readers.
• Poor structure: Unclear role of two research flights that were introduced after you reported some results in numerical form. It’s uncommon structural decision for abstract of a peer-reviewed study to juggle between the results and methodological descriptions. Both introduction and main text leaves the feeling that two sections of the scientific analysis are poorly connected to each other, please address this aspect.
• Scientific value: You reported four numbers numerically in an abstract of the study, called “HSRL-2 retrievals”. A simple question here: can we actually make any conclusions about HSRL-2 retrievals based on four numbers? You may reconsider the structure of your manuscript by minimizing methodological information in brief way and by nailing down your numerical arguments about efficiency of HSRL-2 retrieval.
• You implied the wind speed over ocean behind the term “wind speed” but not articulated in sufficiently for a general reader by quickly resorting to the term without mentioning “ocean”. There is no value in wave-slope parametrizations over land, right? Please either emphasize once for the entire manuscript that you mean wind speed over ocean or always stick to this term, please.

Introduction: Several non-critical issues here:

• You spoke about CALIPSO retrievals of surface wind speeds using Cox-Munk principle but forgot about one of the seminal works on this topic (Josset et al., 2008 about synergy of CALIPSO and cloud radar data to retrieve AOD from this relationship). So, your speculations about correction of ocean signal by available AOD information from CALIPSO should definitely point out to this study and mention please that this idea has been already proposed and utilized by researchers.
• CM54 as a term is actually worse choice than just plainly referring to the Cox-Munk principle as ‘Cox-Munk’ or ‘Cox and Munk parametrization’ from the reader point of view. You save very little space by introducing an unnecessary acronym that has to be searched in the manuscript by a reader. Moreover, Cox-Munk principle is quote known in this expert field, so professional readers are more familiar to this term rather than CM54
• Equations are not common for the introduction of peer-reviewed studies, but I’ll leave this decision up to the editors.
• CALIPSO night data have higher quality from signal-to-noise ratio standpoint, but you said that only daytime data is available for AOD retrieval for correcting ocean signal (see Line 61). Clarify this aspect please.
• The indicative paragraph of the introduction is poorly structured (the last paragraph of the introduction). A reader should get a clear idea about your research aim, but this paragraph might actually confuse a reader by excessive information inserted here. Please follow the structure: methodology shortly, research aim, site/period. All aspects requiring extensive referencing can be mentioned before this paragraph.

Methodology: This section should be revisited as well:

• Provide explicit and extensive description of your HSRL system before you introduce the concept of the ocean surface return retrieval. Without knowing the specifications of your HSRL system, it is impossible to judge whether the calculus you choose and, for instance, crude assumption about Fresnel reflectance (if we are speaking about non-nadir angles) is valid for your study or not.
• From my point of view, the structure of your methodology should be: instrument, campaign, calculus, correction/collocation procedures. Now it’s: “very short instrumental description, calculus, instrument a bit again, campaign, correction/collocation procedures, trivial statistical wiki”, but it is up to you.
• I think sub-section 2.5 is redundant for the journal like AMT. These statistical approaches are common knowledge for atmospheric research.
• You mentioned MERRA-2 reanalysis (Lines 423-425), but never described this data in the methodology. If you really used this data, this is a critical oversight, making your study unreproducible and therefore not suitable for peer-review. If you did not use it, it’s confusing why you suddenly show us SST background map here like it will be an important aspect of your study later.

Results: I am afraid that the figures 1 – 8 and corresponding description alongside the tables are nothing but chaos from peer-review study point of view. Think about readers, would it be convenient for them to comprehend the material in this way? Please follow the structure “Description of figure 1, Figure 1, then Description of Figure 2, Figure 2, … etc” It does not mean you need to make every plot as a separate figure but speak out what you show us. Also, Tables are wrongly formatted. See the AMT requirements for border formatting when it comes to tables. Also, please ensure that you report quantitative findings consistently without over-relying on qualitative terms.

Conclusions: This is a section with little value for reader in the current form. Please revisit your conclusions by thinking about two aspects: harmonizing this section of the manuscript with key aspects such as research aim, methodology, key results, etc; second, think about implications you give for future studies. Good and bad lessons are both valuable, use them in the best way to inform the reader. Specifically:

• Foremost, I think you should report your correlations for each wind speed range directly. I see correlations only over the entire ranges, which might be misleading due to highly variable correlations depending on the wind speeds. Stick to quantitative reporting please without switching to qualitative remarks, behind which, a reader cannot discern an actual low correlation for high wind speeds. Such correlations are also key, essential lessons learned from your analysis, making your study more valuable.
• I suggest to remove a remark on the comparison between the efficiency of CM-54 and Hu-08 models because you never set up such research aim; there was no ultimate analysis on this, right? However, note that some other expert might thinks, well, once the authors state that they can actually ESTIMATE surface reflectance, why they did not go all the way by comparing Cox-Munk, Wu 90, Hu 08, Li-2010 etc, parametrizations QUANTITATIVELY if they finally get their hands on the ground truth using HSRL when it comes to sea surface reflectance?
• Line 496. A very strange remark appeared here. You used the word “novel” for the first time in the conclusions, you never ever articulate it before the conclusive remark is made. Which leads to conclusion, have you been surprised that your method is novel once you reached conclusions? Explain the novelty of your method in a comment below and provide required explanations in abstract as well.

Minor comments:

• Line 64 Errors can create errors is a bad word choice
• Line 99 Section 1 = Introduction
• Line 104 This approach, as well as Hu-08 piece-wise approach for U derivation; both will not work with non-nadir lidar systems (see works of Josset et al 2010; about the equation for non-nadir retrieval of ocean surface reflectance and ongoing works of Labzovskii et al., on non-nadir retrievals of ocean surface reflectance from Aeolus which measures at >35 degree incidence).
• Line 108 Is your system similar to CALIPSO to assume this Hu approach works with your HSRL system?
• Line 115 Structural setback of your methodology is evident here. You talk about HSRL, then calculus, then HSRL again. Can you first extensively discuss your HSRL system and only then, to justify that Hu/Josset/… etc approaches demonstrated for CALIPSO are applicable in your case?
• Line 128. Broken out = separated?
• Line 159 “Arb units”?
• Line 165 Please address this structural setback here and elsewhere in the manuscript. In peer-reviewed studies you normally first describe the figure, and then show it below.
• Line 170 Here, your structural setbacks become critical, please split your methodology into: description of HSRL and explanation of the calculus + if needed, correction procedures for HSRL.
• Line 175. Not true, it also depends on the wavelength and the incidence angle of lidar (see works of Li et al., 2010 on pre-launch Aeolus demonstrator; Josset et al., 2010 work on non-nadir retrieval of Beta_surf and Labzovskii et al. ongoing Aeolus works on applying this principle to Aeolus setup with UV wavelength and >35 angle incidence). Add references if needed for justification and readers’ interest.
• Line 176. Once again, this assumption depends on the setup of your lidar system which is a mystery after the incomplete methodological description. Not true for non-nadir systems (sensitivity starts from U = 12 m/s if your incidence is > 30 degrees according to Aeolus pre-launch works of Li et al. 2010 for instance).
• Line 180. This is unsupported surmise because (a) we do not see any results, confirming this statement, (b) nor a reference, where a reader can get familiarized with this common expert knowledge.
• Line 188 Do you need to account for aerosol transmittance a s well or you mean only molecular transmittance?
• Line 198 “Unique to the HSRL-2…” rephrase this sentence please, it is ambiguous in the current form. Below, “to do so” -> replace to any structure, more common for academic English.
• Line 205 Is it actually true, can TSR be constant in real conditions over considerable spatio-temporal range??
• Line 213 Ideal -> replace to some other word or justify why it is “ideal” here.
• Line 380 I think you mean that CM54 and Hu08 demonstrate high bias, no? Also, why to report all-range-encompassing results, while you literally made piece-wise detailed analysis for every range? Please explain. I think this without this aspect, you can more clearly report bias/correlations/any other issues between HSRL and dropsondes right in your abstract as well without being confused to make additional comments on higher or lower correlation at some wind speed range. I hope my comment is clear here.
• Line 395 What is winter and summer for you?
• Lines 400… I did not see where you discuss the complete lack of correlation at high wind speeds seen at Figure 8b for instance. Can you navigate me please?
• Line 424. Which day, can you reiterate in the text as well for convenience?
• Line 427 (and also 440, 456). “Significant” is a statistical term for peer-reviewed studies; it requires some arguments on statistical significance. If you did not mean that, use “substantial” as your word instead. Also, add a reference about common surface temperature gradient in this area. Think about general readers and argumentation, please.
• Line 430 Sounds like this flight should have been described in the methodology or?
• Line 436 Explain to a general reader which physical principle is the fundament of this HSRL ability please. Mention actual spatial resolution which is used to resolve such gradient.
• Line 443 One would argue that 3.32 m s-1 is not agreement but discrepancy, e.g. substantial bias, no?
• Line 447. Yes, it might be difficult during high cloud fraction conditions. Thus, this method is also constrained like Hu et al. 2008 method relying on clear atmospheric conditions for retrieving surface reflectance from ocean?
• Line 477 Perhaps, you might be even more assertive here? Do you think we really can use HSRL-2 wind speed retrievals in such conditions? I mean fair, transparent recommendation would work the best for your own benefit here from my point of view. This is important lesson learned, very valuable.
• Line 485/6 “Results being compared”, can you simplify the wording here and elsewhere in conclusions please. Scientific analysis encourages brevity and clarity over vagueness and wordiness.