Review of "Waves in SKRIPS: WaveWatch III coupling implementation and a case study of cyclone Mekunu"

General comments

In this manuscript the authors assess the performance and sensitivity to different parameterizations of a regional atmosphere-ocean-wave coupled model in simulating cyclone Mekunu in Arabian Sea. The authors first compare the performance of an atmosphere-ocean-wave fully coupled simulation, an atmosphere-ocean coupled simulation, and a standalone atmosphere simulation. They conclude that both versions of the coupled simulation give better results than the standalone atmosphere simulation. They further examine the sensitivity of the coupled simulation to different options of Langmuir turbulence parameterization and found that the simulation results including the mixed layer depth and sea surface temperature are sensitive to the choice of Langmuir turbulence parameterization. The authors also report the sensitivity of the simulation results to different choices of ocean surface roughness parameterization in the appendices. 

In general this is an interesting study. The results are helpful for improving our understanding of the atmosphere-ocean-wave coupling during cyclones, and are useful for the development of regional atmosphere-ocean-wave coupled models. While the manuscript is easy to read, I think it can be significantly improved by a careful revision.  

One of my major concerns is that the focus of this study doesn't seem clear to me. If I understand it correctly, the focus of this study is to assess the effects of ocean surface waves by incorporating a wave model WaveWatch III into a regional atmosphere-ocean coupled model SKRIPS, using cyclone Mekunu as an example. If this is the case, the comparison with a standalone atmosphere model WRF seems to distract the readers from the focus. Also, coupled model has more skill in simulating cyclones than standalone atmosphere model may not be entirely new. I'd suggest the authors focus more on the impact of  ocean surface waves by the coupling with a wave model. In this sense it would be better to examine in more detail what are the impact of including the effects of Stokes forces, Langmuir turbulence, wave modulated wind stress and ocean surface roughness seen by the atmosphere as introduced in Section 2 on simulating cyclone Mekunu. The presentation of the results in Section 4 is very brief and is not focusing on the effects of waves in my opinion, whereas Section 5 only discusses the impact of different options of Langmuir turbulence parameterization, which is only one of the wave effects included in this coupled model. So the section title of both sections are very confusing. In addition, the results of different sea state dependent surface roughness closures are presented only briefly in the appendices, which is also confusing to me why the authors choose to present these materials there.

Another major comment is on the result of VR12-MA, one of the Langmuir turbulence parameterizations tested in this study. The authors found that using VR12-MA makes the simulated mixed layer depth shallower and sea surface temperature warmer in the cyclone wake than the simulation without Langmuir turbulence parameterization. This result is not intuitive as it is expected that Langmuir turbulence enhances the vertical mixing and deepens the mixed layer. The authors provide a possible explanation in Section 5.2 by examining the regionally averaged vertical profiles, which is very interesting. This may highlight a deficiency of KPP which uses a bulk Richardson number to determine the boundary layer depth, which might be sensitive to the structure of the velocity and buoyancy profiles. I'd suggest the authors to look closer to this issue, perhaps by plotting the time evolution of these profiles in Figure 11 at a point on the cyclone track and check how these profiles change as the cyclone passes by. I guess VR12-MA would still give stronger deepening of the mixed layer depth during the cyclone, but the mixed layer depth may be shallower after the cyclone for reasons suggested by the authors.   

I'd also appreciate it if the authors could provide more detailed discussion on the results. My impression is that the authors presented a lot of figures showing the results, but the corresponding description and discussion in the text are rather brief.

Specific comments

L5: Why comparing with a standalone atmosphere model? The difference would be dominated by the effect of including an active ocean model? Why not comparing with the coupled model without the wave component?

L9: Is Langmuir turbulence the only way through which the effects of waves are included? It might be helpful to mention what wave effects are included in the coupled model.

L22: "Intensity" -> "Intensity of TCs"?

L37-38: Is Langmuir turbulence the only way impact of surface waves is implemented in this study? I know this becomes clear in section 2. But it would still be helpful to discuss at least why Langmuir turbulence is emphasized here.

L68: What "surface boundary fields" are exchanged here?

L70-71: Not sure what do the authors mean here... Why online regridding is not needed? If online regrinding is not needed, why implementing it? 

L73-74: Might be helpful to be specific on what inputs and outputs are included...

L81: By "Langmuir turbulence parameters" do the authors mean "Langmuir number"?

L99-100: So in addition to the surface Stokes drift mentioned on L80, the integrated Stokes transport is also needed to approximate the Stokes drift profile, right? Also,  the same authors (Breivik et al) have an updated method to approximate the Stokes drift profile (Breivik et al., 2016), essentially requiring the same information from WW3. Any comments on why not using this newer method?

L117: Might be helpful to write out the equation here rather than referring the readers to an equation in Li et al., 2019?

L117: "using" -> "uses"?

L119: The entrainment flux is also affected by the enhanced turbulent velocity scale, right? 

L126: Same as above, might be helpful to write out the equation here?

L147-148: Which other models are used in this study. Was there a comparison of different options?

L173-L185: Are the boundary conditions of atmosphere, ocean and waves consistent with each other?

L193: What do the authors mean by "derive skill from boundary conditions"

L208-209: I didn't follow this sentence.

Section 4: The purpose of this section is a bit confused. If the purpose is to validate the simulation results of the coupled model (which seems to be suggested by the section title "Results"), more details and discussions on the comparison among the three sets of simulations (CPL.AOW, CPL.AO, and ATM.DYN) seem appropriate. If the purpose is to provide a background information for the discussion on the wave effects, the authors might need to be explicit on that.

L250-251: Might be helpful to be specific on what is better and what is worse in CPL.AOW than CPL.AO.

L271: "Fig 5(b)" -> "Fig 5(c)"?

Figure 4 caption: What do the black and red dots mean?

L297: Switch the order of "cool the SST" and "deepen the MLD"?

L299: Nudging to what?

L300-301: Perhaps more reasoning of why nudging is necessary here deserves more clarification.

L318-321 and L325-326: Do the authors mean that the vertical mixing of momentum is too much in VR12-MA, which reduces the vertical gradient of ocean current and reduces vertical mixing of tracers like temperature? It is not clear to me why an enhanced vertical mixing of momentum coexists with a reduction in vertical mixing of temperature. It might be helpful to elaborate on why this is the case.

Figure 9, 10: The results of VR12-MA are not intuitive to me. According to Section 2.3, VR12-MA also includes the effects of Langmuir turbulence on  enhancing the vertical mixing. Then why the MLD gets shallower and SST gets warmer along the cyclone track than the case without Langmuir turbulence? Are the snapshot plotted at the time indicated by the red dot?

Figure 11: Why the mixed layer depth (dashed lines) is different in panel (d) from other panels?

References

Breivik, Ø., J.-R. Bidlot, and P. A. E. M. Janssen, 2016: A Stokes drift approximation based on the Phillips spectrum. Ocean Modelling, 100, 49–56, https://doi.org/10.1016/j.ocemod.2016.01.005.

General Comments:

            The manuscript presents a set of experiments aimed at demonstrating the impact of 1) coupling between the atmosphere/ocean, 2) further coupling the atmosphere/ocean model to a wave model and 3) various parameterizations of Langmuir turbulence using tropical cyclone Mekunu as a case study. The paper is clear, well written and provides a great level of detail on the various model formulation settings which is very useful for readers who want to explore similar experimentation.

            The authors demonstrate significant improvement to the mean and RMSE values of cyclone central pressure, wind speed and latent heat fluxes through coupling. Though as the authors point out, the CPL.AOW model does not outperform the CPL.AO model. It would have been nice to see some discussion on why this might be. There are additional figures comparing the evolution of SST and MLD in the CPL.AOW simulation and that in the HYCOM analysis; however, there is little discussion of these figures. For instance, I’m surprised the HYCOM analysis has a great decrease in SST but smaller decrease in MLD.

            The second section of “Results” (Section 5 for some reason) examines the impact different parameterizations of Langmuir turbulence have on SST and MLD. The authors demonstrate that the LF17 and LF17-ST experiments produce more accurate changes in SST with greater cooling and greater decreases in MLD relative to an experiment without Langmuir turbulence included. Interestingly, they also find that the VR12-MA experiment produces worse results than running without Langmuir turbulence. They attribute this fact to a reduction in turbulent shear and ocean mixing by the VR12-MA scheme.

            Overall, the manuscript is straightforward, easy to follow and presents some interesting results that I feel many in the scientific community will find useful. I feel the manuscript only requires minor revision with perhaps a bit more discussion in the Results section and addressing the comments below.

Specific Comments:

Sort of a general note, but I’m surprised the first mention of using an ensemble comes on Line 195. I would think this would have been mentioned in the abstract or perhaps further up in the Methodology section.

Section 2.3: It might be useful to have some description on why these three Langmuir parameterizations were selected. Especially since the Results section shows that the LF17 formulations are quite similar in their impact and the VR12-MA simulations are substantially different.

Line 196: “small random perturbations to the initial SST (<0.01 ◦C) at every grid point in the coupled model.” Are they actually random or is there some amount of spatial/temporal correlation?

Line 234-235: The language here does not reflect that there are three model being compared. Unless the purpose it to only discuss the coupled model simulations.

Figure 3: Interesting that the models are somewhat indiscernible until ~day 3. Suppose this illustrates the time necessary for the ocean/waves to begin to have a meaningful influence on these metrics.

Lines 261-263: Not sure what’s being said here.

Line 288: Please describe why this figure shows that the wave height is sensitive to the wind speed.

Line 298: Would be useful to provide more description on why spectral nudging is necessary.

Figure 9/10: Perhaps I’m missing something, but in the figures with spectral nudging the red dot isn’t centered on the largest SST/MLD differences. Should it be?

Lines 339-342: Having trouble connecting this phrase to figure 11. Perhaps demonstrate explicitly if it is seen in the figure

Technical Comments:

Line 41: “the Antarctica”?

Line 167: Maybe just personal preference, but I would rephrase to “The ocean-atmosphere model is not coupled to the wave model”. Same for #3 (Line 169).

Line 176: Perhaps linearly interpolating “to” not “between”?

Line 249: “Despite CPL.AOW better simulates…” Believe some words missing here.

Line 267: Figures 5b, c. Not 5a.

Line 271: Fig. 5c. Not 5b.

Line 277: Fig. 6b, c. Not just Fig. 6b

Figure 7 Captions don’t match the figure.

Review of “Waves in SKRIPS: WaveWatch III coupling implementation and a case study of cyclone Mekunu” from Sun et al.

The manuscript described the integration of the WW3 wave model into the SKRIPS couple framework. The cyclone Mekunu is chosen as study case in order to validate and compare a set of coupled and uncoupled simulations. Moreover, different Langmuir parametrizations have been implemented in order to investigate the impact of the wave to ocean coupling with an emphasis on SST and mixed layer depth analysis. Different surface roughness parametrizations are also compared to estimate the impact of waves in the coupled system. The simulations are validated against observations of the cyclone’s characteristics and SST from drifters. Overall, the coupled simulations show better skill in reproducing the cyclonic event. The sensitivity to wave coupling on the characteristics of the cyclone is not found to be significant. However, the impact on SST and mixed layer is shown with a decrease of 0.5 degC and a deepening of the mixed layer of up to 20m in the wake of the cyclone compare to the simulation without Langmuir turbulence. Although, one of the parameterizations, VR12-MA, showed counterintuitive results with weaker SST cooling and shallower MLD. This is explained looking at vertical profiles showing a reduced horizontal shear velocity when using that specific scheme.

General comments:

Overall this is a good paper and well written. Although some of the results of the study are well known (i.e. coupled model generally have better skill in modeling cyclone events) I believe the emphasis is also put on the technical implementation of the wave model into the SKRIPS framework and of the different Langmuir and roughness parameterizations. For that reason, I believe it would be a good fit for GMD journal. However, one of the major issues in my opinion is that the manuscript needs more clarity on certain aspects and more discussion of the results which would be beneficial to the overall paper. As well as maybe a reorganization/clarification between sections 4 and 5 where the goals are sometimes unclear or confusing whether it aims to compare coupled and uncoupled, the impact of the different parametrizations or some validation of background ocean state. Some findings of the study are really interesting but often times too briefly discussed or even some figures not discussed at all. Also, as an optional comment the discussion on the appendices A and C on the roughness parametrizations may fit well in the main text.

Hereafter are the details of the specific’s comments.

Specific comments:

Line 80: It could be useful to reference the section 2.4 when mentioning the “momentum flux terms due to waves”, it feels a bit too vague otherwise.

Line 86: Maybe mention here that details of the calculation of the momentum stress is given later on, otherwise one can wonder why you’re not mentioning it here. 

Section 2.4: Mention that the momentum terms are output from WW3. Is the total air-side stress calculated inside MITgcm ? If yes, why don’t send the total air-side stress calculated within WRF which was calculated using wave information ? And thus what kind of bulk formulae is used in MITgcm to calculate the air-side stress (line 83-84) ? (similar remark can be made for heat flux). I think that a bit more clarity on how the surface flux are calculated and exchanged could be useful.

Line 147: In which scheme did you implemented these parametrizations ? Aren’t some of these already available in WRF ?

Line 147:150: Since you implemented several roughness parametrizations, any reasons why you didn’t implemented COARE3.5 which is more recent and as also formulation of surface roughness based on sea state variables ?

In Appendix A, line 373-374: please describe this options using the Charnock from WW3.

Table 1: Is the surface layer scheme used in WRF the same as PBL, MYNN ?

Line 207: Is a coupled frequency of 120 seconds necessary? An Hourly coupling or 30 min coupling would allow to capture the diurnal cycle as well, any reason behind the choice of the coupling frequency? Also wondering if the computation time is impacted by this high coupling frequency ?

Line 210-211: In line with my previous comment, I found it not clear how the surface stress is calculated. On Figure 1, WW3 send the surface roughness to WRF to get the stress but here it is mentioned that the Charnock coefficient is used. Is it actually the Charnock coefficient that is sent to WRF or the roughness length? Please clarify and modify Figure 1 if necessary.

Section 3.2: What are the models frequency outputs (if different from one model to another)? It would be useful to add it here.

Section 4: The name of the section, “Results”, is fairly generic.  The goal of this section needs some clarification, it includes some results between coupled simulation and some more validation part, i.e. section 4.2 or 4.3 which resemble more to a validation of the wave field and no impact of waves on air-sea interaction or between simulations are discussed yet.

Line 250: Any hypothesis as why the CPL.AOW would give “worse” results than the CPL.AO ?

Line 271: “Fig. 5(c)”

Line 271: Is this a known cold bias from HYCOM ? Any reason as why HYCOM would show stronger SST cooling ?

Line 272:273: Does this mean it is in agreement with some observations ? Please clarify this statement.

Section 4.2: the results of the MLD differences in HYCOM showed in Figure 6c are not discussed, such as any hypothesis as why HYCOM would show shallower MLD while stronger cooling. Overall this section could benefit from more explanation of the figures and results showed.

Figure 7: Caption does not match the actual figure, please clarify.

Line 284: You probably meant Figure 7b or Figure 7c.

Line 291: Please precise Appendix C.

Line 304: Please precise Appendix B. Also, in the text Appendix C is cited before Appendix B so maybe the order of the appendices could be revised.

Line 389: “which are”

Line 391: Please be more explicit than “CPL” here as on the figure they all have different names and none is CPL

Line 395-396: Please specify the region, as in the Figure A4 it looks like there are some regions where the wind speed and the latent heat flux are weaker compared to CPL.CHAR.

Figure A4: “parameterization”; Also what do you mean by “without parameterization”, isn’t CPL.CHAR parameterized using the Charnock coefficient from WW3 ?

Section 5.1: The Figure 8c showing HYCOM SST compared to drifters is not discussed although it looks like the cooling along the track is the closest compare to the drifters, is that right? However, in Section 4.2 it seems like the SST cooling in HYCOM was too strong which was one of the reasons leading to a too low intensity of the cyclone simulated ATM.DYN, please clarify or comment on that. Here again, this is interesting results and a bit more discussion on these findings could be beneficial.

Figure 11: Looks like the mixed layer depths (dashed lines) are different in panel (d) than the others, please correct/clarify this.