This manuscript is improved from the prior version. The use of English is significantly improved, increasing the readability, and some analytical discussions have been improved. However, the discussion needs increased clarity and the needs to be consistent with the magnitude of observed effects. The discussion and conclusions also present ideas that are not tested or proven by the results and thus need to be more focused towards what was actually observed.
The manuscript and data show very valuable analyses of various salt sources (frost flowers, brine, snow) and their relationship to aerosol particles. The data clearly show that most of the source samples have chemical compositions similar to sea salt, with some large (e.g. sulfate) and smaller (near error estimates) deviations from sea salt composition. The aerosol particle samples show both a sea salt component, and other interesting behavior (e.g. Mg2+/Na+ ratios differ from sea salt). The current manuscript makes too much of a point of some of the small deviations from sea salt composition and uses language that is too strong for the small effects seen, and thus needs revision.
While some effects are large (e.g. sulfate is depleted by a large factor (~5) compared to sea water ratios), many of them are small. Therefore, it is necessary to give a well described error analysis that accurately uses appropriate words. Error estimates are included, but for example in Figure 6, the error bars are not specified as 1-sigma or otherwise. The analyses of sea water (discussed further below) are not compared to literature values, and a quick comparison seems to show deviations of 10-20% from literature, which is on the order of potential analysis error, particularly when propagated into ratios. At the same time as there are apparent differences from literature sea water, small effects (10-30%) are being described in the text as "other sea salt ratios .... increased remarkably" (page 7, line 32). Another example is "molar ratios to Cl- in some brine samples markedly exceeded the seawater ratios." (page 10, line 17), which is used to describe 20-30% changes on average (Table 1), or maybe a few outlier points on Figure S2. A third example is "Therefore, this drastic change of sea salt ratios on 26 February..." (page 11, line 11), which is used to describe a 10-30% change. Note also that all of these changes depend upon the chloride analysis (the denominator of the ratios). The authors have some very clear effects and should discuss them more, and also have some small effects, which need more error discussion and need weaker wording. For the weaker effects, a rigorous discussion of analytical errors is needed. A t-test could be used, but often errors are not truly normally distributed (they are non-Gaussian), so effects that are indicated to be "significant" in a statistical senses but are not far from the level of significance are not to be over interpreted.
The idea that blowing snow may be a source of sea salt aerosol is still not truly considered in this manuscript. Instead the authors stay with the idea of the prior manuscript that sea salt aerosol essentially come from frost flowers. The manuscript appears to show photographic evidence for blowing snow. Figure 3, panel f shows the "condition of old sea ice on 2 March, immediately after the storm", which is clearly scoured of snow. Photographs on earlier days (Figure 3, panels (a), (c), and (d)) clearly show snow on the sea ice. Apparently this snow was blown away, and as snow blows, it sublimes, producing aerosol particles. Open sea water can also lead to production of sea salt aerosol.
The new "schematics of sea-salt cycles..." Figure (12), has some good ideas in it, but does not belong in the conclusions (it is a discussion point and some of the text related to it are not reasonable conclusions of the present study). Some aspects of this discussion that are taken too far by the authors include the following. In the "Initial stage", bubble bursting releases particles from the surface microlayer of the ocean. This microlayer is organic-rich, and could potentially include inorganic counterions selected by the organic species. In the second stage, it is posited that "ikaite-like and mirability particles are released from frost flowers and brine on sea ice into the atmosphere". What physical release mechanism is being proposed here? For the frost flowers to be depleted in sulfate with respect to chloride, there must be a physical separation of brine (depleted in sulfate) from mirabalite crystals. The normally discussed mechanism is that the mirabilite is left in the brine on the sea ice, and the sulfate-depleted brine migrates up the frost flower. That would leave the mirabilite further from the atmosphere and seems to not be compatible with production of mirabilite particles in the atmosphere. The text needs to posit a mechanism for this effect, and it should be discussed, as the results of the manuscript don't appear to prove such a mechanism exists. The "third stage" indicates that it is concluded that "iodine is released into the atmosphere through heterogeneous reactions and Br- is released slightly or non-significantly under dusk conditions. The more solar radiation, the more reactive halogens can be released from frost flowers and brines." This statement is way too far for the results shown. Remember that the results show slight enhancements in Br- and I- in the older frost flowers. The argument for I- release is that I- is less enhanced than Br-, but again these enhancements are small and analytically suspect in themselves, their differences are even smaller. Lastly, how can the study that says bromide is not released conclude that with more solar radiation there is release of bromide? In the fourth stage, the lack of release of particles by wind from frost flowers is discussed to somehow conclude that Mg-enriched particles are released by wind?
More specific comments:
The text has been modified, but the abstract has been modified very little. Line 18 of page 1 discussed heterogeneous SO4-- formation that is not really discussed in the text anymore.
Page 2, line 13: This says that sea salt is an ice nucleus (IN), which seems surprising given that most IN are not soluble. Give a reference or cut this text.
page 2, line 23: Again it is said that salt is an IN.
page 2, lines 32-38: This section is about fractionation in a section nominally about "modification". Move to the other section about fractionation. A clearer description of how "fractionation" differs from "modification" would also be useful.
page 3, section 2.1: Give a clearer definition of new, young, old, very old ice.
page 5, line 7: There is a shift here from mass based concentrations to molar based ratios. Make that shift more clear in the text.
page 6, section 3.3:
This section is valuable for presenting the ion ratios in various ice types. It should be made more clear that these ratios are all on a molar basis, but that the ratios were taken using mmol/L concentrations. The slopes of the correlations are unitless, but the intercept has the unit of the concentration, and thus is important. This intercept is often trivially small, but for two cases, the intercept may have an effect, specifically the Mg2+ correlation to Na+ in snow shows a very different slope than the other species, but also an intercept that is similar to actual snow Mg2+ snow concentrations. The brine Cl- to Na+ comparison may also be affected by this offset. It also needs to be noted that ion ratios (calculated in other places) are equivalent to slopes of correlation plots only when the intercept is zero. When the intercept is non-zero, the slope of a correlation plot will differ from the ratio of the ion concentrations. On page 6, line 40, the authors indicate that ratios to Na+ are consistent with sea salt fractionation (by mirabalite precipitation), which they are, but these changes are relatively small because there is much more Na+ than SO4-- in sea water, so sulfate limits the removal of Na+. Therefore, what should be pointed out is that SO2-- is depleted compared to either Na+ or Cl-, which is clearer evidence of mirabalite precipitation. The magnitude of these changes appear consistent.
The treatment of sea water ratios and their errors is still lacking in this discussion. Specifically, Table 1 shows the ratio of Br- to Na+ and Cl- from literature (Lide, 2005). However, the analyses of sea water are used for the other ratios. When I look up sea water ratios to Na+, I find the following values, which are then compared to the analysis results from seawater (n=2) in Table 1, which is in parentheses following the literature values: K+/Na+ = 0.022 (0.020), Mg2+/Na+ = 0.113 (0.091), Ca2+/Na+ = 0.022 (0.020), Cl-/Na+ = 1.164 (1.227), and SO4--/Na+ = 0.060 (0.0613). These differences are on the order of 10-20%. They could be true differences between the sea water sampled as compared to standard sea water, or they could be small analytical errors. However, it is important that the present study used only a literature value for the Br-/Cl- ratio and then compares to analytical results to say that Br-/Cl- ratio is larger in frost flowers than in sea water. the magnitude of the enrichment is ~20%, which is comparable to the differences between the sampled sea water and literature sea water. Therefore, it is not clear to me that the "result" of bromide enrichment is a true result. Given the magnitude of errors, this result should be a minor one rather than a serious highlight of the results.
page 7, section 3.4:
This discussion is interesting and clearly reiterates the point of sulfate being depleted in frost flowers but not in brine. The effects on other ions are relatively small, but the wording of the discussion does not reflect the small magnitude of these changes and still lacks a complete error discussion. For example, page 7 line 32 indicates that Mg2+/Cl-, K+/Cl-, and Ca2+/Cl- "increased remarkably". These increases are on the order of 10-20%, and not too different from reported error bar magnitudes. The wording needs to be more reflective of the actual magnitude of the effect. Additionally, the errors are indicated as "analytical errors" in the caption of Figure 6, but that is not a clear indication of the type of error. If these are 1-sigma error bars, that should be noted. The results section does not point out the large difference (factor of 5) present in sulfate ratios, but instead focuses on small changes that appear near error estimates. Additionally, in the start of this section (lines 17-18 of page 7), the text should more clearly make a correspondence between the terms "aged, young, fresh" frost flowers and the sampling sites.
page 9, section 3.9:
The observation of Mg2+ enhancement in aerosol particles is an interesting one. Figure 11 should include a sea water ratio line (0.113 from literature, but 0.091 from sea water analysis in this work). Given that Figure 11 shows no wind speed, it is very challenging to confirm the statement "In conditions of blowing or drifting snow and strong winds, the Mg/Na ratios and the standard deviations decreased in both modes...". This needs to be more clearly presented. The storm of 1 March is noted, but the figure shows day of year. The text should indicate the DOY of this storm.
This section indicates that Mg/Na ratios decreased (towards sea salt ratios) during blowing snow and strong winds. However, page 13, line 20 indicates that Mg is enriched during strong winds. These two statements appear to conflict.
page 10, lines 12-36 are mostly speculation, and their only basis is results that are probably close to analytical error limits.
page 11, line 11 overstates the "drastic change". Also, lower in this section, lines 23 and 24 indicate significant and non-significant changes that seem similar to the eye and have no statistical basis.
page 11, line 36: I have no clue how these molar ratios were generated; please explain more, or more likely cut this section.
page 11, line 42: A tiny difference is being used to justify a broad statement of "likelihood that I was released". This is really wild speculation. Iodide is not even a conserved species in sea water, so drawing this line is not straightforward from the literature. This wild speculation continues through page 12, line 16.
page 13, line 20: This statement conflicts with page 9, section 3.9.
page 14, lines 1-6 are truly conclusions from this work. Other true conclusions should be added to this section to make a new conclusions section. The reminder of the new "conclusions" should be moved to a discussion, as these points are really a discussion of potential ideas about sea salt aerosol formation related to sea ice. As discussed earlier, this discussion should be narrowed to what is defensible from the observations in this manuscript and/or points already in the literature (with citation of those literature sources). The current discussion lacks appropriate citation.