This work applies the WRF-Chem model to the Arctic and compares with observations, from January and February, 2014, in an effort to evaluate and improve the model’s capabilities to represent the atmospheric aerosol over the Arctic with a particular focus on the modelling the main chemical components of sea spray.

Comparisons of simulations from a “Control” version of the model and from an updated version (“HEM_NEW)” are drawn with coarse particle observations of nitrate, sulphate, chloride, ammonium and sodium from Alert, Villum and Zeppelin.   The domain, d1 shown in Figure 1, is simulated at a resolution of 100x100 km.  Then, ‘Control’ and ‘HEM_NEW’ comparisons are made with fine particle observations from Simeonof and Gates of the Arctic, for the same suite of chemical components.  In this case, the model domain (d2) is nested within the d1 domain, and the simulations are a resolution of 20x20 km.  (Note: in Figure 1, Simeonof is shown outside of d2, but in Figure 3 it appears that the model resolution is the same for both Simeonof and Gates of the Arctic; please clarify).    Subsequently, comparison of ‘Control’ and ‘HEM_NEW’ are drawn with observations of both super-micron and sub-micron particle composition measurements from the Barrow Observatory within d2 at 20x20 km resolution.  The differences between “’Control’ and ‘HEM_NEW’ could be more clearly delineated, perhaps in a Table.  I understand them to be the addition of marine organics, the addition of a sea-surface temperature component for sea spray emissions, application of satellite data to improve the whitecap fraction and the addition of a sea-salt sulphate component.

In section 5, Table 3 is a bit confusing, but it appears that ‘HEM_NEW’ run at 20x20 km resolution over Alaska becomes the “Alaska_Control” and “NEW_Alaska” is ‘Alaska_Control’ with updated dry deposition code, a local source of marine organics, a modified sub-micron particle dependence on wind-speed and increased resolution of sea-ice fractions.   

Overall, the paper is an important document of justified changes to a model, made to improve its ability to represent the Arctic aerosol.  The paper is long, a bit repetitious in a few spots, and reasonably well organized.  It could benefit from a careful look at little details, including some of the figures.  I have one major concern and a number of minor comments.

Major 

1.    Sea-spray aerosol (SSA) is a major component of this paper, and I agree that it is an important topic.  However, Arctic Haze is in the title and nss-sulphate has been the major component of Arctic Haze.  The deficiency in modelled submicron sulphate at Barrow is substantial (Fig. 4b and 11), but little discussion is given to it.  The sulphate time series is relatively flat in Figure 4b, which seems unrealistic.  Sub-micron aerosol at Alert, Villum and Zeppelin is neither shown nor discussed, aside from the reference to OM at Alert on line 429.  My concern here is that if the model does not simulate Arctic Haze well, that transport of sea spray from more distant sources may be a problem as well.  If so, this could jeopardize your conclusion about open leads as the major source of SSA.  On the other hand, sub-micron nss-SO4= at Alert during January and February, 2014 was unusually low, and may be more consistent with your simulations.  Do you have any references that indicate the model does well with Arctic Haze, or would you show comparisons of submicron sulphate from the model with observations at Alert, Villum and Zeppelin?  This issue needs to be dealt with in the paper.  

Minor comments

2.    Line 4 – Concerning organics, you might consider here the paper by Mungall et al.: Microlayer source of oxygenated volatile organic compounds in the summertime marine Arctic boundary layer, P. Natl. Acad. Sci. USA, 114, 6203, https://doi.org/10.1073/pnas.1620571114, 2017.

3.    Line 16 – Maybe reduced biases, instead of improved biases.

4.    Line 31 – “mid-latitude”

5.    Line 55 – Also important to further knowledge about the vertical extent of SSA.

6.    Line 110 – Instead of “All the various processes”, maybe “The well-known processes” or “Basic processes”

7.    Line 112-113 – What about dust?  It can be an important factor in the Arctic.

8.    Section 3.1 – With respect to Alert, I suspect the coarse-particle chemistry is based on the difference between high-volume (essentially TSP) filters collected outside and submicron filters collected inside.  The high uncertainties may be related to uncertainty in cut size of the submicron filters.  Also, the high-volume filters are susceptible to blowing snow, which may (on occasion) bias concentrations a bit high.

9.    Line 224 – “long-raNge”

10.    Lines 228-229 – The sea around Alert is also frozen in winter.  

11.    Figure 2 – I suggest only showing the weekly-averaged model points to enable the scale be expanded to show the proper comparisons better.

12.    Figure 6 and like figures – Can you place the labels horizontally above or below the globes?

13.    Figure 3 - What is the difference between the blue crosses and the grey dots in SO4= in Figure?

14.    Lines 401-403 – This may be too simple a solution.  The aqueous phase is unlikely at Alert in winter, unless in fine haze particles (possibly quite acidic).  Transport to Alert through liquid clouds at more southern latitudes is possible, but then some sunlight might be a factor too.

15.    Line 407 - It is not so easy to see this for NH4+, since the concentrations of NH4+ are quite small.  What is the uncertainty in the NH4+ measurements?

16.    Line 415 and Figure 4b - It is very difficult to tell from this figure that lower concentrations are better represented by HEM_NEW.

17.    Line 418 – Modelled sub-micron OA…

18.    Line 429 – parenthesis after “2” not needed.

19.    Line 455 – “difficulties in capturing sub-micron SSA and nss-SO4= during wintertime…”

20.    Line 462 – Demonstrations of the importance of wet and dry deposition goes back further than 2007.  Maybe “the importance of the formulation of wet and dry removal…”?

21.    Figure 7 caption – “sub-micron aerosol mass concentrations” instead of aerosol mass concentrations for sub-micron”.

22.    Line 514 – not be expected

23.    Line 517 – “in clean regions of the Arctic” is a little too broad.  Wasn't it near Zeppelin?  Also, clean regions of the Arctic is contradicted by Arctic Haze.

24.    Lines 621-622 – Maybe “Overall, the Alaska-New January simulations of sub-micron Na+ and Cl- are an improvement over the CONTROL, but still ...”

25.    Line 667 – “not a significant source of SSA”

26.    Lines 726-728 - This is misleading, unless you can show that submicron aerosol at other sites compares well.  

27.    Lines 729-730 - Not to underestimate the potential of aqueous-phase processes, I feel that the third reference to metal-catalysed aqueous-phase oxidation, without a comprehensive discussion, over-emphasizes the potential of this process.

28.    Lines 766-769 – By radiative effects in the darkness of winter, I assume you are suggesting that SSA contributes to changes in longwave radiation or is somehow involved with ice crystals.  A little more detail would help, including a reference or two.

The manuscript by Ioannidis et al. reports model simulations over the Arctic with the aim of improving SSA predictions. The study is a large effort trying to examine the impact of various recent parameterisations on the model output and agreement with the observations.

Despite a significant effort by the authors it falls short of advancing the understanding of processes and their impact on model output. Just for a start, the paper title emphasizes Arctic haze while the main stated purpose is to advance SSA predictions. I am not suggesting that Arctic haze is not important, but the current version if a mix of everything: mostly sea spray, but frequently interspersed by acidic components and anthropogenic sources. That does not help to deliver focused conclusions, because there are already too many issues related to sea spray alone.

However, the most significant problem is related to the implementation of the study. What was the purpose of using an outdated sea spray source function – Gong97, which is not even based on observations, but is rather a mathematical extension of an even older, although pioneering at a time, study of Monahan, combined with the slightly more contemporary observational data of O’Dowd et al. 1997? Being instrumentally limited, even O’Dowd et al. size distribution stopped at 0.1um in diameter when there are plenty of recent papers providing evidence of large numbers of sea spray particles down to 10nm (L. Cravigan et al. 2015 JGR, A. Schwier et al. 2015 ACP, J. Ovadnevaite et al. 2014 ACP, W. Xu et al., 2022 Nature Geoscience just to name a few). Surprisingly, none of the above papers are mentioned despite a clear contextual value. That is even more surprising given the choice of the state-of-the-art WRF-Chem model. If sea spray model output so outdated and diverging with observations, how can anyone trust inferences on marine organic matter, ssSO4 and processes taking place in internally mixed particles?

I wonder why the authors did not make an attempt to compare with size distribution data. That is a critical aspect, because focusing on mass balance does not advance process understanding and produces little value when it comes to studying aerosol-cloud interactions and predicting CCN. Mass balance is dominated by supermicron fraction in sea spray while the number is dominated by the submicron fraction.

The paper is very long and well organized, but suffers from lack of focus. For example, take dry deposition section. There is no context of reason given why the impact of dry deposition is explored given the fact that the model generally underestimates SSA and larger dry deposition is only making matters worse. The other example is sulphate. Although there is a clear distinction between ssSO4 and nssSO4, the model comparison only deals with total SO4 making all findings or inferences vague as those species come from entirely different sources. The authors devote a large section on neutralization factor, but use total sulphate in calculations despite the fact that only nssSO4 is making aerosol acidic. By contrast, ssSO4 is balanced by sea salt cations.

Conclusions are a compilation of mostly speculative statements. Some of them are pure speculations as to suggesting what was not included in the model and what impact it would have had if it was (anyone’s guess really), some are speculative in a sense that no quantitative support is given. If something improved the simulations, then by how much? Was better agreement marginal or significant? Same criticism applies to Abstract where there is a single number to illustrate the results.

Although the value of the paper would not dramatically improve, a reorganisation of the paper by dropping anthropogenic components and by providing better context, reasoning and the outcome of specific parameterisations would make it more readable and useful.

Other comments as they appeared

Line 2. Models will always have difficulties reproducing observations, because they are just model approximations. Model agreement does not need to be perfect if it captures key processes consistently.

Line 8. The statement that the model overestimates sea salt is contradicted by all further results where model consistently underestimates sea salt.

Line 10. Sea salt sulphate is part of sea salt, why is it included second time? Please clarify if not.

Line 21. Breaking waves (open ocean or open leads) are the main source of SSA, how can it be missing?

Line 55. How come open ocean be a new SSA source if it was always the main one? The statement is either completely wrong or should be rephrased/clarified.

Line 129. Sea spray is not just Na and Cl. Sea spray is sea salt (including sea salt sulphate) and primary organic matter. Those early SSA functions were derived from physical particle measurements, not chemical measurements. This aspect is crucial for comparing model and observations which often report just Na and Cl and rarely sea salt sulphate, other major ions and almost never OM.

How Gong et al. source function was translated into Na and Cl emissions? Was it necessary at all, because it is rather straightforward to split sea salt into major components, like Na, Cl, ssSO4, etc. 

What was the rationale to improve model simulations by using three decades old source function alongside with newly discovered SSA sources from open leads and frost flowers?

Line 187. It is crucial to acknowledge and discuss sampling loses in sea salt observations, because models do not take into consideration of sampling inlets and other sampling artefacts. Sampling losses are mostly related to super-micron range, but depend on specific inlets or samplings ducts.

Therefore, given the fact, that the model is generally underestimating observations, sampling losses would make underestimation even more dramatic.

Figure 2. HEM_NEW or CONTROL simulations were not introduced up to this point.

Line 217. How could Fuentes parameterisation for sub100nm be fully utilized if Gong97 or its extension based on O'Dowd97 stopped at 100nm?

Line 224. 20% of ss-SO4 is not a small fraction.

Line 233. I am not sure that anthropogenic Na was properly estimated given Cl depletion in anthropogenic air masses, which tends to be translated into excess Na and thus anthropogenic source. The authors do not discuss Cl/Na ratio which would be informative and also considering derivation of ssSO4 as 0.25xNa.

Line 259. What was the degree of neutralization considering the fact that NH3 neutralizes the stronger sulphuric acid and only then the nitric acid?

Line 278. The authors missed to mention earlier pioneering studies of Blanchard 1976, O'Dowd et al. 2004 Nature and later Ovadnevaite et al. 2011 GRL, 2014 JGR. The authors have a justified liberty of choosing parameterization, but the acknowledgement of earlier studies is advisable.

Line 316. Wind speed is indeed an oversimplified dependency most importantly not accounting for increasing and decreasing wind speed relationships. This aspect as well SST dependence together with a new sea salt source function accounting for the sea state was introduced by Ovadnevaite et al. in 2014 ACP.

Line 331. Na, Cl, ss-SO4 and OM does not capture full sea spray as Mg, Ca, K are missing. How was the ss-SO4 fraction of 9.9% obtained, especially that 7% is arising from basic sea water composition?

Line 356. Basic inorganic chemistry mandates that NH3 is first neutralising sulphuric acid (stronger acid) and producing either ammonium bisuphate or sulphate. Only a leftover producing ammonium nitrate which is pretty stable in polar areas.

Line 360. f=1 is representing fully neutralized aerosol, not "more". More importantly, only nss-SO4 (and NO3) should be used for assessing neutralization, because ss-SO4 is balanced by other cations in sea water (salt).

Line 378. Sulphate and nitrate mass is conserved and has nothing to do with chlorite depletion reactions as model simulation do not output masses of specific salts like Na2SO4 or NaNO3 to attempt Na mass balance.

Line 480. If model underestimates SSA in general, higher dry deposition only makes matters worse. Clearly larger dry deposition is studied not for the better agreement, but simply for theoretical reasons. Better justification and context is needed in this section.

Line 558. SSA - win speed relationship is nonlinear (power of 2-3) with Gong97 on a high end of values. What is the purpose of comparing with linear dependences?