As commented by the reviewers in the discussion phase, the approach of the paper is based on previous work by other groups, particularly the publications of Frey et al. (2009) and Erbland et al. (2013). However, the data are from a different area and extend deeper into the snowpack. They are a valuable addition to previous datasets and challenge existing hypotheses. I am satisfied that most of the issues raised during the discussion phase have been addressed by the authors in their revised submission. Despite the continued reservations of Referee #1, I consider the work of high scientific quality and sufficient novelty to be publishable in Atmospheric Chemistry and Physics, subject to a few revisions as outlined below.

Page and line numbers are based on "acp-2014-826-author_response-version2.pdf".

To ensure traceability of results and long-term access, please include the full dataset (values of data points) as tables in the supporting information.

Jan Kaiser
Editor Atmospheric Chemistry and Physics


Calculation of Rayleigh fractionation constants
In the ACPD version of the paper, apparent fractionation constants were calculated for non-overlapping individual depth ranges as well as entire profiles. Both approaches are defensible because the Rayleigh fractionation equation applies to any part of the data domain, for which its assumptions hold. In other words, it is possible to calculate "local" fractionation constants by taking the derivative of ln(1+delta) with respect to ln f, where f is a suitable transformation of the mass fraction into a quantity of dimension 1. Such a transformation could be as simple as dividing by the base unit (i.e. "ng g-1"); it doesn't have to be the "initial" mass fraction.

Due to the prominence Rayleigh fractionation calculations take in this paper, the axes of Fig. 5 should be log-transformed to ln(1+delta) and ln(w/(ng g-1)) and the corresponding linear fit equations should be shown on the figure. For consistency, the corresponding Rayleigh plots for P1 to P3 should be added as additional panels and the related data in Table 2 should be referenced in the caption. The current version of Fig. 5 (with delta on the y and w on the axis) is not needed to follow the discussion, but it could perhaps appear in the supporting information.

The strong correlation between ln(1+15delta) and ln(w/(ng g-1)) in P3 (0-20 cm, 0-40 cm and 0-60 cm) as well as for the 25-100 cm depth horizon in P1 (ACPD version of the manuscript) could perhaps be taken as evidence for postdepositional processing. I didn't see this being discussed in the manuscript; could you please add this?


Section 4.1.2:
The mass-balance approach is an overestimate of the exchange because of the way it is calculated here. You assume that photolytic fractionation and exchange take place sequentially. However, both occur simultaneously. This means that the exchange should not be considered to take place between water and the hypothetical final nitrate delta after photolysis, but between the continuously changing isotope delta of nitrate and water. Because the actual nitrate isotope delta is always lower than the hypothetical final isotope delta, less exchange is required to explain the observed isotope delta. Of course, such a revised estimate relies on the same assumptions as the present calculation (e.g. whether epsilon(18O, photolysis) applies and what its value is). For the example presented in this section, the revised calculation gives 11 % less exchange (43 % instead of 54 %).


Influence of stratospheric chemistry (abstract line 41, conclusions lines 773-775, section 4.4, esp. lines 745-747):
Mass-balance calculations (Savarino et al. 2007) and direct measurements of stratospheric nitrate (Moore 1974) show that its delta(15N) is between 1 and 19 ‰. This does not preclude isotopic fractionation during transport, deposition or indeed after deposition, which could enhance delta(15N). You suggest that this is unlikely because you expect similar effects for delta(18O) and Delta(17O), but as you write yourself (line 712), recycling in atmosphere or snow can "reset" the oxygen isotope signature. Also, the winter 2011 data in P1 show even higher delta(15N) values without associated high Delta(17O)/delta(18O). Indeed, the apparent Rayleigh fractionation constant of -62 ‰ for the 25-100 cm section of this profile might be taken as indicative of postdepositional processing.
l 745-747 I don't think the claimed stratospheric influence is clear from the data, especially when seasons prior to 2012 in P1 as well as P2 and P3 were included in the discussion. High Delta(17O) values can also be produced in the troposphere, as the mechanisms for nitrate production and the isotopic signatures of the reactants are similar in troposphere and stratosphere, especially under Antarctic winter conditions.

Fig. 3: There is a datapoint missing near the surface; please include it; you may want to break the axis as done similarly for P6.

Fig. 4: Could you please add the corresponding delta(18O) vs. delta(15N) plots for P1 to P3?

Fig. 8: Please add two additional panels, one showing delta(15N) vs. w for the entire P1 data set; the other showing delta(18O) vs. delta(15N), similar to Fig. 2 in Savarino et al. and discuss the interannual variations.

"app" in epsilon_app should not be in italics.
2/32, 3/104, 4/137, 9/328, 17/684 & Fig. S1 caption: Please add space before units (cm, ml, K, DU).
many locations: Please add space before "‰" and "%" signs.
3/78 & 10/404: Please use equilibrium arrows (right-left harpoons).
5/164: Please include a meaningful physical quantity, e.g. amount (n), in the isotope ratio definition, i.e. n(15N)/n(14N), etc.
5/174: Please list the standard uncertainty of delta(15N, USGS 35) and delta(18O, USGS 32).
USGS 32, USGS 34 and USGS 35 carry no hyphens.
6/211: Please state which NaCl salt was used, evidently not Fisher Scientific's product.
6/220: Difference between what?
9/326 & 13/513: "cross" symbols in Eqs. 3 and 6 are not needed. Indices should not be in italics. The exponential function should preferably be written as e^(-cx).
9/329: The unit for actinic flux should be "cm-2 s-1 nm-1". "photons" are not units.
9/337: For consistency with the rest of the text and Eq. 2, the non-primed quantity symbol (j) should refer to the light isotopologue and the primed symbol to the heavy isotopologue (j').
11/422: The current understanding of isotopic fractionation during volatilisation is limited, but ignorance should not be taken as "little evidence". Please delete or revise this sentence.
15/620: Please expand "MSA" and add "mass fractions" after "Na+".
18/759: In line 391, you use delta(18O, H2O) = -60 ‰.
Table S3: epsilon should be called isotopic fractionation as elsewhere in the manuscript.
Fig. S1 caption: Which statistical test does p refer to?

Dear Dr Shi

many thanks for your revised manuscript, which I am happy accept for publication subject to a few minor revisions as detailed below.

Kind regards
Jan Kaiser


Using a data repository is fine, but please include the doi of the data - I could not find them in the Brown repository yet. Alternatively, you are welcome choose a more convenient file format than PDF (tab-delimited ASCII), zip it and include it as supplementary information with the manuscript.

l. 39: Please add "alone" after "processes". Independent of the snowpack measurements, atmospheric NOy concentrations and the aerosol isotopic signature (albeit not measured here) are strong hints that loss of nitrate from the snow phase does occur.
l. 163: "with n representing the amount of substance".
l. 212: It is still unclear which of Fisher Scientific's NaCl products were tested and which one was chosen. Please be specific.
l. 423: To avoid following "however" with "nevertheless" please change "Nevertheless, it is" to "This makes it".
l. 655: Please cite Moore (1974) and add their measured value for delta(15N) of stratospheric nitrate to your discussion. I am not aware of their concentration or isotope measurements to have been refuted. Their other nitrate isotope deltas (e.g. for tropospheric nitrate, rain and seawater) are not unlike those obtained by more modern methods. If you know more, please include this information in your discussion.