The authors have used the RACMO2.3 model and a downscaled, long-term dataset to investigate temporal and spatial differences in precipitation and extreme precipitation events (EPEs), and their role on Surface Mass Balance (SMB) at three different ice core sites in Dronning Maud Land, Antarctica.

The paper is well structured and flows logically from each section. The introduction and motivation outline the gap in our knowledge and the importance of investigating precipitation and extreme events. The discussion is also comprehensive and provides a different perspective on the results. However, moving some of the discussion points from the discussion to the results or introduction would provide the reader more trust in the results and the use of the model for assessing the precipitation. In addition, the results could be separated into spatial and temporal distribution to better aid the understanding. I do think major revision is required, as some of the results are not clearly explained or presented. I am also unsure about the use of the multiple datasets and how they compare or validate the observations and each other. I believe the study sheds light on an important topic, and showcases the difficulties of using relatively lower-resolution models to investigate SMB at site-specific observations, especially in a geographically complex region. I think the conclusions are valid and important, but the results need more clarification and justification before it can be published.

Comments:

Your introduction is very comprehensive, and your motivation is clearly outlined. No comments for the introduction.

Major revision:

Section 2.3: More information is required for this product. Add the time period to line 140 - 'Extending the time period covered by RACMO' is not enough for the reader to assess the length of time. Even though you later say the ESM was used from 1850-2014, it is not clear if the downscaled final product also uses this same timeframe. Which version of high-resolution RACMO is used? 2km or 5.5km? What resolution is the downscaled dataset? The reader shouldn't have to read Ghilain et al. 2022 for this information, given that it is important to the study. The downscaled product uses RACMO data, but there are large differences in the outcome, especially for the EPEs, but this isn't reflected in results or discussion – you should discuss this.

Which variables are you using? Snowfall or precipitation? Are these synonymous or comparable between the downscaled dataset and RACMO? Does total precipitation = snowfall in this region, or are their times of rainfall in summer?

Section 3: There are major differences between the simulated SMB by RACMO and the ice core records. This is the first thing reported and then makes it very difficult for the reader to trust that RACMO is going to be used for the rest of the study. The justification for using it comes in the discussion, but you should consider moving this earlier, and perhaps bolstering this justification further. Almost 50% of the ice core SMB is not represented in the model – if precipitation is the main component, are you convinced that the model is representing the precipitation properly? Whilst models are always wrong, there is additional model justification and testing which is presented in the discussion which could perhaps come early in the results to bolster the reason for continuing to use RACMO despite the consistent, large underestimation. It would be ideal to see more comparison of the key variables such as precipitation with other observations.

The discussion section regarding complexity of SMB in ice cores should perhaps be moved to the introduction or results. In my case, I am very familiar with RACMO and the SMB analysis in the polar regions, but haven't used ice cores as observations before. Therefore, a straight comparison of the ice core and the model seems like a bad idea, given how poorly RACMO captures the ice core observations. However, I do see value in continuing the study to analyse the RACMO data and assume that it can be a tool for showing SMB differences in time/space.

With the downscaled data, it isn't even capturing the long-term trends found in ice cores (section 3.2.2), which makes it even more challenging to justify using. If it isn't capturing long-term trends, which typically models can capture, how do you know it is capturing any spatial variability?

The authors give a short analysis of the blowing snow contribution – which is significant, but then it is not investigated further or mentioned in relation to the spatial differences in SMB between the three sites. Perhaps more emphasis should be given to this investigation, especially as you conclude that precipitation/extreme precipitation is not responsible for the spatial differences between the sites. If there is a blowing snow modified version of RACMO, could you investigate (briefly) the differences in the model output with and without this modification?

Figure 3: It would be useful to change this to better allow the reader to compare the products and the locations. Figure 3 is busy and it is too hard to compare SMB from RACMO and observations, this could be a separate figure to the other components. In addition, it is hard to compare the locations when there's a lot happening in one figure. Similarly, it would be useful to get one figure where ice cores, RACMO and downscaled data are presented together. Statements like 'in contract with ice core records' (Line 203) are hard to check in the figures, when long-term SMB from ice cores is on figure 2, satellite-era RACMO is on figure 3 and long-term downscaled data is on figure 4.

Section 3.3: More information is needed on how you calculate the thresholds – does each location have its own 95% threshold – e.g each grid cell which represents the ice core location has a value? Or are you using an area average for all locations? Are the thresholds re-calculated for the downscaled data? Later on, you use ERA5 too – are you calculating the thresholds again for ERA5 data, or simply using ERA5 to extra data on specific dates, which have been above the threshold from RACMO?

Section 3.3.2: I think it is a good idea to look at the synoptic situation during these events, but I question the addition of ERA5 data (also because it is not listed in your data section). This is another dataset, which is not compared to RACMO, the downscaled data or ice cores. Whilst synoptic conditions are generally well captured in ERA5 and most models, you are looking at specific dates of these events. How do you know that ERA5 also captures the EPEs which RACMO is seeing? Precipitation is a difficult variable, even in higher resolution reanalysis products. Perhaps it is better to look at the synoptic conditions in RACMO, rather than introduce an additional dataset and therefore additional uncertainties in the conclusions. If you do stick with ERA5 – there should be some discussion of its useability in this region and how it compares to RACMO. In addition, do you select the data from ERA5 based on dates in RACMO, or do you re-calculate the 95% threshold with ERA5 data?

Section 3.4: The geographic and synoptic set up doesn't particularly align with the characteristics of foehn winds. With the Peninsula, a long, high ridge prevents the air from flowing around the obstacle and therefore forces it over 2000m– this is what creates the foehn winds. However, in the case of ice rises, the airflow could flow around the obstacles, given their size, and likely not create the warm and dry leeside conditions. The definition of a foehn wind is also the warm and dry lee slope winds, and not the reduction of precipitation down wind.  Instead, you're perhaps referring to orographic precipitation characteristics, such as rain shadow. I wouldn't introduce the foehn effect here, as it doesn't really apply.

Section 3.4.1: Can you find a different name for the EPEs in this section? Up until now, you have defined EPEs as extreme precipitation events with a 95% or 98% threshold for the value of extreme. However, in this section, EPEs now mean 'percentage of cases where the other sites receive more precipitation than their site-specific EPE threshold'. It is then confusing to try and interpret the results – especially the relative wet/dry of the locations. However, this definition does answer a question I had earlier about whether thresholds were site-specific. With the current definition, I am left confused about whether IC is drier or wetter than other sites during EPEs.

Whilst the differences in EPEs and negative anomalies per location from RACMO (table 4) seem significant, the differences between the sites in the downscaled data seems negligible or insignificant. Have you run any statistical tests on these results? The neg.anom for EPEs at IC and EPEs at TIR are very similar, and the two data sets do not agree with each other. The results here focus on the RACMO set, but you don't discuss the lack of consensus among the datasets. This could be because the downscaled data includes a longer time period, but as stated in your earlier results, there is no long-term trend in the data for precipitation, SMB or EPEs, so this perhaps doesn't answer it.  I am really not convinced with section 3.4.1 I understand your hypothesis and perhaps the method of trying to look at it, but the results are quite confusing.

Section 4: This first paragraph about ice core complexity should go in the introduction – throughout the results, I am concerned with how RACMO is representing observations, but this section gives me pause about the ice cores as observations. This level of complexity regarding SMB from ice cores should come earlier, especially for readers who are not experts in ice core interpretation.

If 2.2km RACMO was found to be more representative of the ice cores than 5.5km RACMO, why not use the higher resolution one? Or is the 2.2km RACMO the downscaled product you have used?

Minor:

Section 2.1: Can you provide the elevation of the ice rises – this becomes fairly important for your discussion on foehn winds and the loss of moisture across the trajectory.

Line 165 and 168 say the same thing.

Section 3.2.1 – is this really interannual variability section, or is it more spatial variability? Apart from the first line, the rest of this section is about the different locations.

Line 182: What do you mean by opposing signals in ice core records? Is this figure 2? Apart from TIR which has a decreasing trend, they don't seem to have opposing signals. This is hard to tell in figure 3 too.

Table 3: caption says it is contribution to the total annual precipitation, which you also confirm in line 252, however in line 259 you say that EPE variance accounts for 2/3 of the SMB variance. So is the variance SMB and the average contribution annual precipitation? Different variables are used between RACMO (annual precipitation) and downscaled data (snowfall) – are they comparable?

Line 327: I don't understand this sentence – where are the observed global atmospheric pathways observed?

Figure 8: ERA5 data?

Line 356: change 'excludes' to 'rejects'.

Line 438-439: Perhaps include that this is a conclusion from a model, not from observations.

This study makes use of carefully obtained datasets to discuss a recently important topic in Antarctic Climate science, the role of extreme events on driving the spatial and temporal variability of Surface Mass Balance.

The article structure is organized in a comprehensive way, with good progression. The manuscript reads well. Most conclusions are sound and supported by the data, but some affirmations remain to be contextualized, clarified, or sourced. The analysis and figures are of good quality. However, I think that the model data is not fully used, and even though principal results are already included in the article, there is room for further improvements suggested later in this comment. Finally, I am not competent enough to correct all, but I noticed some grammar issues that may result in inaccurate or even incorrect sentences.

In that regard, I support its future publication in The Cryosphere, but I think it would greatly benefit from major revisions before publishing.

1. Major comments and important revisions

Mistitled sections

I picked up numerous cases of sections with titles that did not correspond to the content of the section. Sections are also unequal in length, so it would be possible to group ideas in sections of comparable length before deciding on an appropriate title.

Title: The article title emphasizes ice cores and does not mention RACMO, whereas the results section of the article has exactly 5 lines about ice core results (148-152) on a section of 192 lines (148-339). It is totally ok to study the components of SMB in a model in order to better understand ice cores composition, but the title should be a brief summary of what is in the study, not a summary of the final objective. In fact, it is deceiving to read “observed in three ice cores” the “role of (…) extreme precipitation events” when the actual results about EPEs rely on modelling.

L147: May use a more descriptive title for the subsection

L178: Section named interannual variability appears to treat spatial variability across the three sites, at daily resolution. The title is not suited.

L200: “Multi-decadal variability from downscaling”. This section has two sentences related to the title, and the rest discusses spatial variability and annual variability.

L288: distribution (of what?). Frequency distributions of precipitation anomalies would be a more informative section title

Introduction

The first two paragraphs of introduction, about contextualization, are inaccurately describing or entirely missing core mechanisms.

L33: “Basal mass loss” sounds like the only mass loss was to basal melting, whereas the general agreement is that glacier discharge is the main driver of ice loss (Davison et al., 2024; Diener et al., 2021). Basal melting can induce more ice loss, notably through reduction of buttressing effect and acceleration of ice discharge (Pritchard et al., 2012), but is not the main driver of Antarctic mass loss. (also mentioned L37)

L46: “Synoptic-scale dynamics correspond to short-lived intrusion of maritime air (…)”: Worded like that, the synoptic scale dynamics always cause high precipitation, where in fact it depends on the scale and intensity of the synoptic system, and intense precipitation are localized. Smaller rates of precipitation are also observed at a distance from the center of the system, but still associated with it.

L50: “Large-scale dynamics correspond to the southward moisture transport from lower latitudes due to large-scale atmospheric modes of variability”: There appears to be a confusion between large-scale circulation (average circulation) and large-scale variability (variability of the circulation). This specific sentence is unclear if not incorrect. Since this paragraph appears to describe the general mechanisms leading to precipitation, please detail variability in a different paragraph.

L52-55: I really do not like the separation of thermodynamic and dynamic, since we could classify "interaction of winds with topography" as adiabatic changes in the air parcel (as in foehn wind), which is fundamentally thermodynamics (this is also mentioned by the authors L55)

The second half of the introduction is well written and introduces the problematic adequately.

Lack of justification for the downscaling dataset

The downscaling dataset provides useful insights onto the long-term evolution of precipitations. However, it is used throughout the article to create some statistics about precipitation events and spatial variability, along with RACMO2, and with almost no difference in most cases. Since it relies on RACMO2 to perform the downscaling, I would have appreciated a discussion on this aspect, to understand why it is relevant for sections discussing spatial variability, since its results do not differ from RACMO.

Do you want to verify if spatial variability is stable through time? In that case you may have to compare different epochs, using temporal subsets of the downscaled dataset.

Moreover, the contribution of EPEs to total SMB derived from the downscaled dataset are almost absent from the discussion, although I was expecting this to be one of the main results from this article after reading the introduction. This may also relate to the too low confidence in the downscaled dataset compared to ice core SMB, but I think both are worth considering (see the other hypothesis that I give for model-data discrepancy).

Lines where the use of the downscaled dataset is not sufficiently justified:

L138-145: This section needs further description. Is RACMO2.3 run through the entire period ? or are daily climate states statistically linked on the observation period and then back-casted onto the longer simulation period? This doesn't need to go as much in detail as the cited article for downscaled dataset creation, but now please include enough detail so that we can understand the general method that was used to produce this dataset.

L142: “high-resolution RACMO simulation” which simulation? the same as previously described?

L143: is there a downscale simulation for each ESM simulation?

L196 “the results align with RACMO2.3”: I was somehow expecting so since the downscaled dataset is a statistical tool using RACMO2.3

L313: “FK is more subject to large precipitation anomalies than the other sites, a finding confirmed by the downscaling results”: same as above.

Affirmations not enough supported by the data presented

L198 – 199: “confirming the absence of spatial variability in annual precipitation across the three sites”

This is an inaccurate shortcut:

Table 1 presents spatial distribution of event occurrence, and the distribution of precipitation amount during co-occurring events. You did not show the distribution of precipitation amount vs precipitation rate for each site, which could be very different from one site to another.

Simply said, you show here that it snows simultaneously (on about 80% of days) and that more than 95% of snowfall occurs on these 80% of days. In reality, one site could have very few variations in annual snowfall compared to the other two sites, that your statistics shown here would not capture, because they are daily and event-related statistics.

You should either reword your sentence to match what your data supports, or do the actual figure to support what you are writing. An example of a figure that you could possibly make is a distribution of annual precipitation for each site (relative to the mean accumulation), similar to your fig. 7 but not restricted to EPEs.

L205-206: “Overall, Table 2 confirms the absence of significant spatial variability in precipitation across the three sites.”

Spatial variability is too vague in this case. I would prefer a "concurrent positive (or negative) annual SMB anomalies across the three sites", referring to the timing of anomalies, which is what your analysis reveals.

Again, same as Table 1, it would be possible and interesting to study the range of anomalies to answer the following question: Do stronger anomalies occur simultaneously at the three sites? (Not only showing positive or negative anomalies, but also comparing their values.) This could be simply presented by comparing sites by pairs, with scatter-plots of accumulation anomalies. Plus, I think it might highlight that spatial variability at extreme precipitation rates is greater, as extreme precipitation is usually very localized. If this is the case it is an important result that would fit perfectly in your study.

L222: “percentile (98th) as a proxy for atmospheric rivers”

While it is true that atmospheric rivers can induce EPE, the definition of another percentile of EPE (98%) does not match the definition of atmospheric rivers given in the introduction, which needs to be a spatial feature with intense meridional moisture flux. If you want to link the 98^th percentile with atmospheric river, you need further justification, either by citing literature proving (or showing here with other analysis) that 98th percentile of precipitation can be considered as resulting from atmospheric rivers in most cases.

The article by Wille et al. (2021) attributed *half* of EPE days within the 99th percentile to atmospheric rivers, which means that another half is not related to atmospheric rivers. This proportion decreases for 95th and 90th percentile, which means that the 98th percentile that you used here should include atmospheric river-related EPE, but also include many other events not related to AR.

L292-293: “to test the hypothesis of a drier air mass reaching the other sites after precipitating at the first site”, L303 “due to a reduced moisture availability in the air mass following intense precipitation at the IC site”, and L316 “hypothesis of a drier air mass reaching the other sites after precipitating at the first site”

Given that the three sites are coastal, a single air parcel precipitating at the three sites would require alongshore (zonal) transport, which goes against synoptic conditions presented in fig. 6. If you want to show that moisture is reduced after precipitation at IC site, you would have to show that the air parcel has gone through these sites exactly using air parcel tracking algorithm (e.g. hysplit or flexpart), or at least wind maps.

Another explanation for the difference for IC site would be related to its particularities: IC is a dome summit on island (as opposed to coastal crests for the two others), therefore there are more "angles" from which the precipitation could form. This would be apparent not only on parcel tracking, but also on moisture flux maps, which have much thinner spatial variability than geopotential anomaly maps.

If you look closely, you can even see that effect on your Fig. 6, where the core of <-10 hPa of SLP anomalies is geographically more restricted for EPE at the IC site, meaning that the low-pressure system can be displaced and still produce EPE.

L356-357: “This finding excludes the hypothesis of decreasing precipitation intensity along the air mass trajectory” this sentence may need adjustments depending on the corrections you plan regarding my previous comment.

L317-319 “All these observations further support the previous hypothesis that neither precipitations nor EPEs explain the contrasting SMB trends in the three ice cores and instead point to the influence of global atmospheric pathways”

First, the hypothesis you made earlier was the contrary: “A particularly high precipitation event at one site might result in abnormally low precipitation at the two other sites, possibly explaining the different SMB trends observed in the three ice cores” (L279-280). This does not further support it, this disconfirms it.

Second, you still have 40% of EPEs that do not occur simultaneously, how could you rule out that they do not play a role in the differences observed between ice cores? You would have to show that there is no specific temporal pattern for the non-simultaneous EPEs.

Third, what is “global atmospheric pathways” and how can the similarity between local EPEs explain less difference than global phenomena?

L326: “global atmospheric pathways are predominantly observed”

I do not think I understand what you mean by this. Global atmospheric pathways would generally refer to atmospheric cells or jet streams, which are not presented in this study’s result.

L386: “the suitability of the 5.5 km products in this study.”

Suitable in this study for which purpose? for site-to-site comparison? I think it needs a little more justification than just comparable correlation coefficients. The average values you described just before and on Fig. S4 seem quite different, so maybe the temporal variability of precipitation and precipitation intensities can vary quite much between the two spatial resolutions. In addition, if you have high annual SMB correlation, it is hard to say anything about EPEs which are daily events.

L390-392: “However, this temporal variability appears to be underestimated in model outputs in comparison to the ice core observations, highlighting a limitation in the model representation of SMB interannual variability”

Underestimation of temporal variability in models is one possible explanation that the authors explored well.

However, the temporal variability of ice cores could also be increased by non-climatic factors. Several studies have shown that: e.g. due to snow erosion-redeposition effects, creating uneven layers of accumulation (Karlöf et al., 2006; Münch et al., 2016; Münch and Laepple, 2018). A single ice core is usually overestimating the variability because of this effect. A study of intercomparison of 76 ice cores (Altnau et al., 2015), although inland DML, showed that signal to noise ratio for an ice core SMB is about 0.4 on shelves (meaning 60% of the observed variability is noise), or even less for continental sites (or even more noise). Similarly radar studies point to an overestimated SMB in single ice cores (Cavitte et al., 2023).

Given that the SMB is one of your datasets, you need to develop and present this second hypothesis in your discussion, to present the limits of this dataset.

 2. Minor comments and technical corrections

L21: This sentence sounds like RACMO2.3 was further downscaled, which is not the case. Please rephrase.

L22: please include numbers on the actual contribution of precipitation and EPE-related precipitation to SMB in the abstract

L35: The Clausius-Clapeyron describes the moisture content in the air. The effect on precipitation is indirect, so this sentence should be reworked.

L44: I am unsure where this classification of "three mechanisms controlling precipitation" comes from.

L95: Additional Information that could be included: Ice thickness at each ice rise, Synthetic Aperture Radar-based ice velocity (just to confirm that value at the crest is approximately zero; SAR ice velocity maps are not necessarily required, but could be cited)

L102: “annual layer thickness” I guess you mean water-equivalent annual thickness? It needs to be specified here.

L103-104: I understand that the correction is for ice flow divergence, thinning the layers over time, which is different from compression effects as would be inferred from the second half of this sentence. These two sentences could be reworked to use more specific wording: (1) Correction of compression effects using ice density profiles, and (2) Correction of thinning effects related to divergent ice flow at ice rises

L111-117: This paragraph may be more suited for results/discussion section, not methods?

L126: I had a hard time figuring out this "respectively". I am unsure about the grammar, but for me it would be more natural to place it at the end of the section (or at the beginning), not in the middle. Here I stopped to "respectively" and read as if SU is sublimation again, and ER is erosion deposition.

L127: I do not understand how the sublimation of drifted snow (SU_s) and erosion (SU_ds) are counted. If snow is eroded than sublimated, does it count in SUds? in the case that it is eroded but deposited elsewhere it does in ERds? but what is the difference from the viewpoint of ice core site, since it is just lost to erosion in the beginning, why not group them for simplification?

L148: RACMO is not black on Fig. 3, and there is no grey line for ice core records.

L150: FK, TIR, and IC were previously referred to using Fk17, TIR18 and IC12. May need to homogenize notations.

L203: “in contrast with ice core records” please ref fig. 2 here, or include the ice core records on the Fig. 4

L217: I would prefer to rephrase this sentence to avoid starting with "Besides". This would be acceptable if "besides" refers to the previous paragraph, but this is not the case here.

L220: you could merge the two sentences by replacing “a certain percentile” by “the 95^th percentile”. Please also precise if this is the percentile for the corresponding model grid cell.

L240-251: please move the cross-site correlation comparison after the description of results from Table 3. Discussion about the values (L252-262) should come before cross-site correlations.

L238 (Fig. 5 caption): “variability” I would rather use the term uncertainty to describe the internal variability of downscaling products, to differentiate from temporal or spatial variability.

L247: “EPE impacts are more localized compared to the average conditions”. This is an important result from this study. I think it deserves to be discussed more, in light of the maps you made, and also be repeated in Conclusions.

L270: (Suggestion) It would be interesting to have also meridian moisture flux maps, that would enable discussion about atmospheric rivers. Such maps would be more relevant if shown on a smaller domain, similar to fig.1 for example. In that case, comparison of 98th and 95th percentile may highlight spatial differences, explaining the lower correlations for 98th percentile precipitation across sites given line 244. Furthermore, cross-site differences may appear more clearly at a finer scale.

 L307: please explain what you mean by “evolve in the same direction”: Geographical direction?

L311: As the figure title says, this is frequency distribution “of precipitation” not “of EPEs”.

L323: “Colors of the distribution correspond to each ice core” not to each ice core (this is not ice core data) but to the model data corresponding to each ice core site.

L350 “a better understanding of the SMB is crucial to improve the projections” It would be good to precise here how this work can contribute to better understand SMB. Importance of extreme events is one aspect. The strength of your model is also its small scale capable of capturing more realistic SMB (as in fig.8 of Ghilain et al., 2022), and this may be mentioned at some point. In particular, how high spatial resolution affects the representation of EPEs and their contribution could be of particular importance on coastal location with important terrain slope. You open this topic L363 but do not develop further or discuss your data, only citing literature.

L352: “simulations indicate no spatial variability” please precise that you are talking about trends; the average value for each site is different, which is a form of spatial variability.

L364: “snow redistribution” I was expecting an example of precipitation pattern, and I got snow redistribution. These two processes that admittedly depend on topography, but since you study the different components of the SMB, I would have liked a clearer separation between them.

L365: “could enable downscaling reanalysis” consider replacing “enable” by “justify”?

L371: “difference observed between the 5.5 km RACMO2.3 product and the IC12-derived SMB record (~265 mm w.e. yr^-1).” Does this mean 5.5km RACMO has still a large margin for improvement? Please comment on your model here.

L376: How about comparison of SMB in the 2 km scale RACMO and the ice cores? Are the SMB getting more similar to ice core data in the 2 km product?

L378: “the three SMB time series” please precise “annual SMB” if this is the case.

L380: you can recall that FK and TIR are the two sites connected to the continent with grounded ice sheet (here, but also at other points of discussion)

L388: “would increase further if the ice core SMBs were adjusted to reflect the consistently lower SMB” Are you considering to adjust ice core SMBs? I do not understand this phrase. It would be possible to change corrections made to ice cores to produce accumulation record, but it needs more contextual discussion.

L390: “the temporal variability is well represented by ice-core records (Cavitte et al., 2022).” It might be good to remind here what are ice-core records compared to in Cavitte et al. 2022, to confirm that temporal variability is "well represented"

L397: “contribution of blowing snow sublimation increased by 52 %” it may be good to precise that this means reduced SMB for Antarctica.

L407: “the contrasting SMB records” again you need to precise <temporal variability of SMB>, because the relative contribution of EPEs to average SMB (fig.5) follows roughly the same distribution as average ice core total SMB.

L408-409: “a common regional atmospheric circulation pattern” may be replaced by more detailed description such as “a dipole of Low-pressure West/High Pressure East of the ice core site”

L414-415: “only representative of a limited area, approximately 200 to 500 meters in radius around the drilling site, and consistently exhibit lower SMB values than the mean across the entire ice rise” This affirmation can be easily sourced to previous works in this area, please cite them here

L430-451 The Conclusions are OK, but may need to be reworked to match the corrections made, emphasize more on current study’s results, and give numbers, especially about EPEs: how much do they contribute to mean SMB, to SMB interannual variability, what atmospheric pattern lead to EPEs… This should answer the original question of “are EPEs driving the variability at these sites?”. Maybe you can also point out that the EPEs impact is currently underrepresented at Antarctic scale, due to the technical aspect of fine scale modelling and coastal topography, which is one of the challenges that could actually be tackled in the short term, by doing circumpolar coastal modelling. There could also be conclusions regarding the potential of SMB to compensate for ice loss in the future, given that models seem to underestimate coastal accumulation based on your results.

Supplement, Fig. S4: how do you explain the relative difference lower than -100% for IC? Wouldn’t 100% reduction just become zero?

References

Altnau, S., Schlosser, E., Isaksson, E., and Divine, D.: Climatic signals from 76 shallow firn cores in Dronning Maud Land, East Antarctica, The Cryosphere, 9, 925–944, https://doi.org/10.5194/tc-9-925-2015, 2015.

Cavitte, M. G. P., Goosse, H., Matsuoka, K., Wauthy, S., Goel, V., Dey, R., Pratap, B., Van Liefferinge, B., Meloth, T., and Tison, J.-L.: Investigating the spatial representativeness of East Antarctic ice cores: a comparison of ice core and radar-derived surface mass balance over coastal ice rises and Dome Fuji, The Cryosphere, 17, 4779–4795, https://doi.org/10.5194/tc-17-4779-2023, 2023.

Davison, B. J., Hogg, A. E., Moffat, C., Meredith, M. P., and Wallis, B. J.: Widespread increase in discharge from west Antarctic Peninsula glaciers since 2018, The Cryosphere, 18, 3237–3251, https://doi.org/10.5194/tc-18-3237-2024, 2024.

Diener, T., Sasgen, I., Agosta, C., Fürst, J. J., Braun, M. H., Konrad, H., and Fettweis, X.: Acceleration of Dynamic Ice Loss in Antarctica From Satellite Gravimetry, Front. Earth Sci., 9, https://doi.org/10.3389/feart.2021.741789, 2021.

Ghilain, N., Vannitsem, S., Dalaiden, Q., Goosse, H., De Cruz, L., and Wei, W.: Large ensemble of downscaled historical daily snowfall from an earth system model to 5.5&thinsp;km resolution over Dronning Maud Land, Antarctica, Earth System Science Data, 14, 1901–1916, https://doi.org/10.5194/essd-14-1901-2022, 2022.

Karlöf, L., Winebrenner, D. P., and Percival, D. B.: How representative is a time series derived from a firn core? A study at a low-accumulation site on the Antarctic plateau, Journal of Geophysical Research: Earth Surface, 111, https://doi.org/10.1029/2006JF000552, 2006.

Münch, T. and Laepple, T.: What climate signal is contained in decadal- to centennial-scale isotope variations from Antarctic ice cores?, Clim. Past, 14, 2053–2070, https://doi.org/10.5194/cp-14-2053-2018, 2018.

Münch, T., Kipfstuhl, S., Freitag, J., Meyer, H., and Laepple, T.: Regional climate signal vs. local noise: a two-dimensional view of water isotopes in Antarctic firn at Kohnen Station, Dronning Maud Land, Climate of the Past, 12, 1565–1581, https://doi.org/10.5194/cp-12-1565-2016, 2016.

Pritchard, H. D., Ligtenberg, S. R. M., Fricker, H. A., Vaughan, D. G., van den Broeke, M. R., and Padman, L.: Antarctic ice-sheet loss driven by basal melting of ice shelves, Nature, 484, 502–505, https://doi.org/10.1038/nature10968, 2012.

Wille, J. D., Favier, V., Gorodetskaya, I. V., Agosta, C., Kittel, C., Beeman, J. C., Jourdain, N. C., Lenaerts, J. T. M., and Codron, F.: Antarctic Atmospheric River Climatology and Precipitation Impacts, Journal of Geophysical Research: Atmospheres, 126, e2020JD033788, https://doi.org/10.1029/2020JD033788, 2021.