Dear Reviewer,

Thank you very much for your constructive feedback and for the great interest you have shown in our work, which will help us improve the manuscript.

Summary 

This manuscript describes the evolution of stable water isotopes in water vapor measured during an atmospheric river event at Dome C in Antarctica. The event led to a strong increase in surface temperature, humidity and δ18O during 3 days in December 2018. Using model tendencies from the general circulation model LMDZ6iso, the authors disentangle the influence of different processes on the humidity and isotope signal. They find that surface sublimation accounts for nearly 70% of the humidity and δ18O increase during the event, while large-scale advection accounts for 30%. Outside of the event, surface sublimation and condensation dominate the δ18O variability and advection is less important. Cloud formation, precipitation sublimation, and shallow convection play a minor role in general. The results highlight the need for accurate representation of fractionation during sublimation in isotope-enabled models.

This is really nice work. The paper is well-written, the results are interesting, and the figures are nicely designed. The methodology is sound except for one caveat, which is that it does not account for the influence of processes that decrease the specific humidity on the isotope signal in the final result. While these processes do not contribute any moisture, they could still alter the isotope ratios (e.g. cloud formation would decrease δ18O in the vapor). I think it may be possible to include them too, which would make the method even more useful (see general comment 2 below). Overall, I find this study a very valuable contribution to the field and would recommend publication after addressing the comments below.

General comments

1) Cloud condensation and precipitation sublimation are currently treated as one process, but they differ substantially in their impacts on both the water vapor and isotope budget. One process is a sink of vapor, the other is a source, and they involve different kinds of isotopic fractionation which imply that one is not simply the inverse of the other in terms of isotope concentrations. I therefore think it would be good to treat them as separate processes in the method. If the model does not allow this, a straightforward way to distinguish them would be to assign every dq/dt < 0 from this tendency to cloud formation and dq/dt > 0 to precipitation sublimation.

In LMDZ, we have one tendency output for the “large-scale condensation” routine which manages both cloud condensation and precipitation sublimation. However, the calculation of positive and negative isotopic tendencies is not handled symmetrically. We will add a sentence in the text to provide more details on the different fractionation schemes applied during condensation and during the sublimation of precipitation.

2) For the final result, the method currently only considers processes that increase the humidity content of the atmosphere. Here it seems that the total isotope budget is still approximately closed (Figure 11d), but there could be locations / situations where the humidity-decreasing processes matter more, e.g. in the tropics or midlatitudes, where convection is more important than in Antarctica. Dütsch et al. (2018) came up with a process attribution for both humidity-increasing and humidity-decreasing processes, which allows to fully close the isotope budget. This was for Lagrangian trajectories, but it could be applied equally well to Eulerian tendencies. The attribution for humidity-increasing processes is the same (equation 11 here corresponds to equation 8 in Dütsch et al., 2018), but there is an additional term for humidity-decreasing processes (equation 9 in Dütsch et al., 2018). I think adding this would make this tendency-based method even stronger and more generally applicable.

We fully agree on the importance of considering negative tendencies. It was indeed an issue for interpreting our cumulated tendencies, that we solved by focusing on the sources of vapour. What may be confusing is that, although the contents are labelled by processes, we actually trace the origin of the vapour associated with these processes. In this context we identify the vapour coming from surface sublimation as sourced from surface snow, and vapour from positive tendencies of advection as sourced from large-scale moisture transport, such as the final vapour content is the mixing of vapour content from the different sources. In this framework, negative tendencies means vapour is removed from all sources. 

This methodology is useful to trace the origin of water but the downside is that the impact of water removal is distributed on the isotopic composition of each water source and is not traced back to the process responsible for water removal. We agree that this part of the analysis is missing. We plan to (1) clarify the denomination for our “water sources” and find a denomination that is not the same as for the associated processes; and (2) add the analysis of cumulative tendencies (both positive and negative) alongside the source analysis, in order to assess the contribution of all processes to the isotopic changes. Finally, in our case, both the water and isotopic budgets are numerically closed, meaning that when summing all individual water and isotopic tendencies, we obtain the simulated specific humidity and δ¹⁸O signals.

3) Since surface sublimation is the most important contribution to δ18O in this study, and there is increasing evidence that it involves fractionation (e.g. Ritter et al., 2016; Wahl et al., 2021), I think there should be some discussion on how the results would change if this was included in LMDZ6iso. The overestimation of surface snow d-excess is likely related to the missing fractionation during sublimation. How could this affect the vapor, and (how) would it change the relative contribution of the processes? If feasible, a sensitivity test where fractionation is added to sublimation in LMDZ6iso would be very helpful.

You are absolutely right about the importance of accounting for fractionation during sublimation. Following this study, we introduced isotopic fractionation during sublimation in the LMDZiso model, which is the focus of a separate paper currently under preparation. This new study provides a detailed analysis of surface fluxes and discusses in particular the December 2018 event. In that work, we show that during this event, the presence of strong winds could have triggered a drifting snow episode (which was indeed observed in the field), and that this suspended snow may have sublimated without isotopic fractionation. In other words, during this event, we believe that the model in its current version captures the observed isotopic peak, but for the wrong physical reasons.

We will add the following sentences to the manuscript to highlight this limitation of the model (l.393):

We note, however, that in the model there is currently no isotopic fractionation during sublimation. While numerous field studies have demonstrated that the sublimation flux is fractionating (Steen-Larsen et al., 2013; Casado et al., 2016; Madsen et al., 2019), the introduction of a fractionation scheme in the model could reduce the contribution of sublimation. Nevertheless, the analysis of the total water budget suggests that surface sublimation makes a substantial contribution to the observed specific humidity anomaly.

Moreover, when working with isotope-enabled AGCMs, the positive bias observed in surface snow is mainly generated by the parameterization of supersaturation (the lambda parameter in the relation Si = 1 - lambda (T - Tfreeze)), before being further influenced by post-depositional processes, as shown in the supplementary material of Dutrievoz et al. (2025).

We will take all the following specific comments and figure comments into account in the revised version of the manuscript.

Specific comments:

L7: LMDZ6iso is sometimes LMDZiso and sometimes LMDZ6 or LMDZ. I would always use the same term.

L26: ice core -> ice cores

L80: These -> This

L98: are -> were

L107: Within 98th percentile? Not above?

L109: Repetition? IWV and vIWVT were mentioned before already.

L300: reveals -> indicates (particle density is not necessarily equal to moisture sources)

L318: A word is missing in this sentence

L368 & L379: Condensation should be sublimation? (Because it is treated as a vapor source)

Figure comments:

Figure 1: Residence time would be a better description than number of particles (if this is what is plotted). Did you run FLEXPART for this study or take data from someone else? If the latter, please add a reference.

Figure 2: Rename „cloud & precipitation sublimation“ to „condensation and sublimation“ (or „cloud condensation and precipitation sublimation“), also everywhere else in the text. Otherwise it sounds like cloud sublimation.

Figure 3, caption: „outside 1.5 times the inter-quartile range“ -> do you mean inside?

Figure 7: The label for temperature is still in French.

Figure 10: I don‘t see the dashed line. Is it behind the black line?

References

 Dütsch, M., Pfahl, S., Meyer, M., & Wernli, H. (2018). Lagrangian process attribution of isotopic variations in near-surface water vapour in a 30-year regional climate simulation over Europe. Atmospheric Chemistry and Physics, 18(3), 1653-1669.

Ritter, F., Steen-Larsen, H. C., Werner, M., Masson-Delmotte, V., Orsi, A., Behrens, M., ... & Kipfstuhl, S. (2016). Isotopic exchange on the diurnal scale between near-surface snow and lower atmospheric water vapor at Kohnen station, East Antarctica. The Cryosphere, 10(4), 1647-1663.

Wahl, S., Steen‐Larsen, H. C., Reuder, J., & Hörhold, M. (2021). Quantifying the stable water isotopologue exchange between the snow surface and lower atmosphere by direct flux measurements. Journal of Geophysical Research: Atmospheres, 126(13), e2020JD034400.

Dutrievoz, N., Agosta, C., Risi, C., Vignon, É., Nguyen, S., Landais, A., ... & Prié, F. (2025). Antarctic water stable isotopes in the global atmospheric model LMDZ6: From climatology to boundary layer processes. Journal of Geophysical Research: Atmospheres, 130(5), e2024JD042073.

Jouzel, J., & Merlivat, L. (1984). Deuterium and oxygen 18 in precipitation: Modeling of the isotopic effects during snow formation. Journal of Geophysical Research: Atmospheres, 89(D7), 11749-11757.

Hoffmann, G., Werner, M., & Heimann, M. (1998). Water isotope module of the ECHAM atmospheric general circulation model: A study on timescales from days to several years. Journal of Geophysical Research: Atmospheres, 103(D14), 16871-16896.

Dear Michelle,

Thank you very much for your valuable comments and for the interest you have shown in our work; they will be very helpful in improving our manuscript.

Summary

This study by Dutrievoz et al. presents a detailed analysis of the isotopic signature of an atmospheric river (AR) that crossed the polar plateau and reached Dome C in December 2018. The event caused a spike in near-surface temperatures, specific humidity, and water vapor δ18O. Using a combination of surface/near-surface observations and the LMDZ6iso model, the authors quantify the impact of the event and the relative contributions of large-scale and localized processes to the surface water vapor mass budget. Their finding that surface sublimation is the primary source for the specific humidity and δ18O increase during the event highlights the important influence of boundary layer processes on the isotopic signature of Antarctic ARs.

The manuscript is thorough, interesting, and well-written. I have included some major comments, as well as minor comments, below. Overall, I think this will be a valuable contribution to the field.

Major comments

I find the unique origin of this event quite interesting, as discussed in Section 3.3, and I think the results may be improved by including more context on how rare or common this type of cross-polar plateau AR path is. For example, the origin of the AR over the south Atlantic and its path across the polar plateau may yield a different isotopic signature than an AR that originates over the Indian Ocean and arrives at Dome C from the north instead of the south. It would also be helpful to know more about why you chose this event specifically. There is a discussion of some other examples of recent AR events, but less information on the climatology of ARs at Dome C and why this one was selected, aside from the pronounced meteorological impacts (or perhaps that is the main reason why?). This also impacts the interpretation of the results in the paper, and whether the conditions observed are representative of other ARs in the region.

Regarding the analysis of this event, it represents the first atmospheric river observed at Dome C for which isotopic vapour data are available. Moreover, according to the analysis of Petteni et al. (2025), at Dome C, the large majority of the back-trajectories originate from the Indian Ocean, as expected for this part of the East Antarctic Plateau (Sodemann and Stohl, 2009).

We have added the following sentence to clarify this point (second line of the discussion):

This is the first atmospheric river event observed at Dome C with available isotopic vapour data, and the only one recorded over the 2009–2019 period originating from the Atlantic sector (Petteni et al., 2025).

l.444: Moreover, an isotopic analysis of these events using a Lagrangian approach, combined with a process-based decomposition similar to that of Dütsch et al. (2018), would provide a complementary perspective on the processes involved during the transport of atmospheric rivers.

In terms of the order of the results, I also think it may make more sense to introduce the large-scale drivers of the event (section 3.3) before diving into the local scale (boundary layer and surface) characteristics and impacts. This would help to contextualize the event, orient readers to the location of interest and the unusual path of the AR and set up the discussion of the different contributors to the isotopic signature of the event.

Yes, you are absolutely right — I will therefore begin the Results section with “Large-scale conditions during the event.”

Finally, on the topic of large-scale drivers, since ERA5 and MERRA-2 have been used more frequently to diagnose the synoptic patterns associated with Antarctic AR events, I think it would be helpful to assess how LMDZ compares to at least one of these atmospheric reanalyses during the event, especially given its comparatively coarse resolution. Are the spatial patterns comparable? Does LMDZ exhibit similar intensities in temperature and specific humidity to ERA5 or MERRA-2?

The simulation we used is nudged to ERA5 for temperature and wind, with a relaxation timescale of 12 hours. We will add the tropospheric profiles from ERA5 in Figure 5.

It is mentioned briefly in the discussion section (lines 427-432), but I would be curious to know more about the bearing blowing snow may have on the method and results presented in this study, especially given that this event was marked by elevated surface wind speed. I think the study would benefit from more discussion on how blowing snow may impact the surface snow isotopic signatures during the AR event as well as the vapor isotopic budget (does the sublimation of blowing snow have an influence on the observations? In what ways might blowing impact the contributions of the various processes to observed δ18O and d-excess? How would this affect or add to the range of uncertainty of the model comparison to the observations?)

Indeed, this event is characterized by very strong winds, which triggered drifting-snow observed in the field. In an effort to reduce the amplitude bias of the modelled diurnal cycle, we have improved the parameterization of surface isotopic fluxes in LMDZiso in an article currently in preparation. However, we have found that the isotopic composition of evaporation switches from fractionating before the event to non-fractionating during the event, which might  highlight a potential isotopic signature of drifting snow. In the context of our present study, this process appears to be well reproduced by the model, but for the wrong physical reasons, as our evaporation is always non-fractionating. We are currently pursuing a more detailed investigation of the impact of drifting-snow events on water vapour isotopes, although this goes beyond the scope of the present paper.

We have added the following sentence in the discussion to clarify this point (after the bracketed part taken from the article):

[At such intensities, drifting snow can be lifted several hundred metres above the surface (Palm et al., 2011), where it may readily sublimate, potentially changing the isotopic composition of atmospheric water vapour (Wahl et al., 2024).]

Consequently, the sublimation of airborne snowflakes may exhibit a distinct isotopic signature than that of sublimation flux from the surface.

[As this process is not currently represented in LMDZiso, understanding its contribution to the surface isotopic signal would be a valuable next step.]

Regarding the use of “diurnal cycle”, “daytime” and “nighttime” – while it is clear from the time series shown in the study that a diurnal cycle is present and influences the boundary layer conditions before, during, and after the AR, I also find it a bit confusing to read about “nocturnal” or “nighttime” conditions when this time of year is marked by polar day. Because these descriptors are heavily used throughout the text (for example, lines 250-257, 340-343, and 385-388), I would recommend adding a sentence to the methods or results (early on) that clarifies the usage of these terms.

Yes, we’ll do the changes. 

I will take all minor comments below into account:

MINOR COMMENTS

(organized by line number)

4 – “3-fold increase” instead of “times”

19 – “Only specific meteorological conditions driven by the atmospheric river can explain these strong interactions” – this implies that only ARs and not any other weather events can be responsible for a combined sublimation and large-scale advection signal in the isotopic signature. Is that the intent? May be helpful to briefly back this up with why that’s the case.

36 – temperature “anomaly” instead of “increase”

40 – “were multiplied by 3” is awkward phrasing

40 – this paragraph switches between past and present tense in describing the event. Would recommend picking one and sticking to it.

43-47 – very long sentence, would break into two

51 – use “AR”

Figure 1 – the core of the AR over the southern Atlantic is white, exceeding the darkest purple/black value at the max of the colorbar. Would recommend extending the maximum value of the colorbar.

Figure 1 – what is the uncertainty in the FLEXPART 10-day back trajectories? Do the particle numbers shown in the figure capture the full range of possibility for back trajectories or do they represent a mean/ the most likely back trajectories?

58 – rephrase to “large-scale transport only accounts for about half…”

60 – “AR”

124 – what is the domain that LMDZ6 is run over? Do you have a map of this that can be included in the supplementary material?

162 – which period are you evaluating the bias over? (ex. “mean cold bias of…”)

172 – what is your hypothesis for why the modelled amplitude of vapour δ18O is so high? As a general comment, when the larger biases are mentioned in the paper, I think it would provide helpful context to readers if there is a hypothesis for why they occur, perhaps in the discussion section.

207-210 – nice explanation

242 – regarding the 13-year AWS record – do the AWS observations continue through the present, or did they end in 2018? If they continue, I think it would be helpful to include the entire record through the present-day. If they do not, is there an alternative record you could use to put this event in context of the present climate? Given the occurrence of the March 2022 event, and a number of other extreme events in recent years, I think it would be helpful to know how the December 2018 event fits in this context. It is clearly a stand-out event over 2006-2018, but how about 2018-2024?

249 – how did you decide on the alternative values for roughness length?

Fig 3 – small note that the panels appear to be different sizes/are not aligned on the top edge. Also, the orange bar for the median could be thicker to make it easier to see.

270 – include value for standard deviation when mentioning “significant spatial variability” in the surface snow δ18O

286 – was the event studied here the only AR event in December 2018? (as in, do the periods 1-18 December and 24-31 December represent only non-AR conditions? Were there any other moisture intrusions during this month that may not quite meet the threshold for an AR?)

295 – don’t need “However”

304 – do you think this points to the need for AR detection tools that operate across the polar plateau? I would be interested to hear more about the authors’ view on this in the discussion section. Clearly, this AR had a large surface impact that occurred outside of the footprint detected by both the IWV and vIVT tools.

Fig 6 – I like the satellite imagery shown here, I think it does a great job showing the cloud plume associated with the AR crossing the polar plateau. Regarding the left-hand figures, I would recommend adjusting the colorbar slightly. Currently, it is difficult to ascertain the peak IWV values associated with the AR, as these peak well above the maximum on the colorbar. At the same time, the IWV contour levels are perhaps too coarse for the lower IWV values associated with the region of the AR that crosses the polar plateau – in panel (c), it doesn’t look like elevated values of IWV quite reach Dome C.

322 – add “spatial” to “anomalies”

Fig 7 – I would strongly recommend marking the regions of statistically significant anomalies for each panel in the figure. You could make the purple contour a solid line (rather than dashed) to make it stand out more, and I would specify that these data are from LMDZ in the figure caption. Also, a small note that the colorbar label for panel (a) is en français.

324 – “850hPa geopotential height” instead of “over sea-level”?

329 – what is happening in this region with the strong negative anomaly in relative humidity?

391 – “These results…” – this is a cool finding, and I would recommend emphasizing it more throughout the text (esp. the abstract and introduction), as it is good motivation for the study.

Data availability statement – given the range of observational data, AR catalogues, etc. presented in this study in addition to the model output, it would be appropriate to expand this statement to describe access to all the datasets used.

References:

Sodemann, H., & Stohl, A. (2009). Asymmetries in the moisture origin of Antarctic precipitation. Geophysical research letters, 36(22).

Petteni, A., Casado, M., Leroy-Dos Santos, C., Landais, A., Dutrievoz, N., Agosta, C., ... & Stenni, B. (2025). Air Mass Origin Effects on Antarctic Snow Isotopic Composition: An Observation and Modelling Study. EGUsphere, 2025, 1-26.