Review “Climatology and Surface Impacts of Atmospheric Rivers on West Antarctica” by Michelle Maclennan and co-authors.

This manuscript investigates the climatological conditions and the surface impacts of atmospheric rivers (ARs) in West Antarctica. The author first uses reanalysis model output (MERRA-2) in combination with an AR detection tool to examine the contribution of ARs in this region from 1980 to 2020. Then for a more detailed and smaller scale perspective the authors present a case study of three successive ARs on Thwaites Glacier in February 2020, for which they use reanalysis data, in-situ measurements and a firn model. Finally, the authors discuss how ARs may change in a future climate. 

The manuscript is well written with clear figures. It is an interesting and relevant study within the scope of TC. The idea and methods are not completely new, it builds on existing knowledge from ARs in Antarctica and previous firn modeling efforts. By combining large scale model output and in-situ measurements, the results are a useful contribution for understanding the climatology and impacts of atmospheric rivers in West Antarctica. Despite being a topic of interest, there are some minor aspects especially regarding the contribution/purpose, goals stated in introduction,  methodology and results that might be better represented. I elaborate on this in the comments below, which follow the order of the manuscript. 

General comments/questions: 

1. Contribution/purpose of this study: an elaborate introduction about AIS mass balance and atmospheric rivers is given. However, the articulation of the purpose and contribution of this study in the introduction can be improved. Articulate more clearly what is new in this study compared to previous work, the added value of this study, what is already known about ARs this region (What have Willle et al. 2019 & 2021 found about ARs in West-Antarctica, e.g. how many per year/trend)? Also the reason why is chosen for this region (Lines 123-127) would be more suitable for this part of the introduction.
2. Contrasting impacts on SMB: In Lines 76-78 you state that ARs have contrasting impacts on SMB, and that it is therefore important to study them from both large-scale climatological perspective and a case study. With contrasting impacts on SMB, I understand that you mean snowfall, melt or temperature? However, the melting (and temperature) part is not studied from the large-scale climatological perspective. Nevertheless, melt could be important on e.g. Abbot ice shelf. I think it would be good to explain in the introduction that and why the focus of the large-scale climatological perspective is on precipitation. This is probably also why you use the vIVT detection algorithm. 
3. Lines 85-86 “Finally, we discuss the results in the context of how ARs contribute to the present mass balance of the AIS and how their frequency and precipitation may change in future climate scenarios.” I don’t see where the future frequency and precipitation is discussed in the manuscript? You do discuss a potential increase in melt related to AR events. Perhaps use: “Responses and impacts of atmospheric rivers to climate change by Payne et al. (2020)”, and the fact that there is an ongoing increase over time of current AR events, which might continue (Fig. 3a). 
4. The discussion is strong and very interesting, one thing that might be added is some comparison to previous findings about ARs on the WAIS, which is mentioned above as well. (E.g. Wille et al. 2019 & 2021).
5. Lines 373-375: “Limited by 1.5 years of in-situ data.”  I wonder why you only look at 1 AR familiy event, while there are multiple AR events each year? 

Specific comments/questions: 

1. Lines 63: Why is this unique for Antarctic ARs? Is this not the same for Greenland ARs?
2. Lines 66: They carry much moisture, but does the fact that the AIS is a desert not also play a role in the importance of ARs?
3. Line 72: The study of Neff et al 2014 is about Greenland.
4. Line 73: “ARs act to increase Antarctic SMB, as they cause significantly more snowfall than surface melting”. Can you give a reference for this statement?
5. Lines 81-82: “to provide key insights on in-situ conditions” this can be rephrased. In-situ is often only used to describe the way a measurement is taken, maybe replace by local conditions.
6. Line 93: basal channel?
7. Section 2.2: Is there a reason why you chose MERRA-2 instead of ERA-5? Perhaps you can add that both reanalysis products give similar results in Wille et al. 2021. 
8. Line 135: Not over the AIS but over the WAIS.
9. Line 153: Actually, you use three different approaches, also the GNSS measurements.
10. Section 2.4: I think it can be clarified how the firn modelling works. Perhaps add that snowfall is assumed to occur when measured snow height exceeds the modeled snow height. Strictly speaking, there can also be snowfall when the observed snow height remains stable e.g. if there is snowfall in combination with densification, sublimation or melt.  The difference between the observed and modelled snow height is then added to the snowpack, which can be converted with the fresh snow density to accumulation. 
11. Line 190-191: Perhaps start the results section with one sentence describing the kind of results you are going to show in Figure 2, as an introduction to the reader.  Also refer to panels of figures if that is the case, so Figure 2a e.g in Line 193.
12. Lines 190-191: the 3.2% in combination with the reference to figure 2 is a bit confusing as I don’t see 3.2% in the figure. Also I suggest to first give the definition of frequency of ARs and then discuss the numbers.
13.  Figure 3: Would it not also be interesting (and possible) to have a third graph with the amount of precipitation from ARs over time? 
14. Caption Figure 5: I suggest to add that this figure is about Thwaites Eastern Ice Shelf.  
15. Line 228: The temperature decrease is not only in the winter right?
16. Line 228: perhaps omit: “from the mean 24 hrs before landfall to the mean 24 hrs after landfall.” as this should already be clear.
17. Caption Figure 6: Line 1 omit repetition of “on TG”.
18. Line 281: Should “different spatial resolution” not be “low spatial resolution”?
19. Line 281: Perhaps include in method section 2.4 that you calculate surface melt from the SNOWPACK model.
20. Caption Figure 8: Line 2: atmospheric conditions are used “to” force SNOWPACK.
21.  Line 306: Improve the reference formatting.  
22. Line 320: From Wille et al. 2021 I understand that 10% of total snowfall comes from AR events. The percentage of extreme precipitation events explained by ARs depend on the threshold but is 10% lower in West Antarctica than East Antarctica (where it is 25-45%).
23. Line 356-357: “As surface-based temperature inversions are least developed in austral summer, the baseline surface temperatures before AR events are nearest the melting point in summer.” And also because it is simply warmer in summer?

General Comments

This is a really interesting paper that investigates the role of atmospheric rivers on the surface mass balance of the West Antarctic ice sheet, for which there is a clear knowledge gap. The authors have obviously put in a lot of work into this and strived for high standards.  The Introduction is nicely written/researched, and the study nicely put into context with previous work.  However, Section 2 on data and methods is difficult to follow as it is disorganised / disjointed and contains sometimes unnecessary text – the paper would really be improved if this section could be organised better.  Section 3 is a well written description of the comprehensive analysis. The figures are clear and appropriate. I was slightly uncertain about Figs 2 and 3, as the results mentioned in the text did not seem to be in the same range as Fig. 2, and the justification for the trend 1995-2015 was not clear in Fig. 3 – also a possible explanation for these trends seems to be missing.  But despite that the authors have obviously put a lot of work into this analysis.  The study includes a very comprehensive, well researched, and well considered Discussion section which does a very good job of contextualising the results. To summarise, I think this is an excellent study, but would benefit from addressing some of the comments below, especially related to Section 2 which really needs to be much clearer / linear – especially given the complexity of the analysis and the number of data sets and the incorporation of both climatological and case study analysis.   

Specific Comments

+ The motivation for the case study in the Introduction is not that clear. I understand that it is included as it can be investigated in more detail using the in-situ observations, and so complements the more broader scale climatology work.  But this is not that well explained and comes across as rather disjointed.  Please strengthen this justification.

+ Section 2.1 is labelled ‘observations’ but has quite a few sentences describing the method, including the SNOWPACK model which is mentioned before described in its own dedicated subsection later on.  I find this rather unstructured/confusing/disorganised and would suggest that a dedicated methodology section would help the reader.   And in general, please choose appropriate sub-headings and stick to the appropriate content for these headings.  

+ Lines #98 - #100: More details on the reanalysis are required such as their appropriateness / representativeness of the AIS, and even just spatial resolution are necessary. Also, the reanalysis are compared with the in-situ observations on Thwaites, but there is no explanation for whether this is appropriate. For example, whether the in-situ observations are representative of a wider area that is comparative to the reanalysis grid boxes.  

+ Section 2.2: Please see comment above about discussing reanalysis data before it is properly introduced. Another comment here is that you state that the datasets are ‘regularly gridded’, so is that in terms of lat/lon?  Also, much of the text in this section again seems rather inappropriate and better placed elsewhere. For example, mention that ‘this region’ has experienced large acceleration in recent years should surely have been clarified in the Introduction and no need for repetition. Finally, its not really clear why MERRA is used for one purpose (as opposed to ERA5) and ERA5 only used for comparison with MERRA during the case study.      

+ Section 2.4: This section is labelled SNOWPACK firn modelling but the opening sentence discusses precipitation from reanalysis. Please restructure these sections much better.

+ Section 2.5: Its not clear why surface height changes using interferometric reflectometry is necessary given that the in-situ observations also mention snow height.  Can you please clarify?

+ Section 3.1, first paragraph: 1) The value given is 3.2% but Figure 2 only shows AP frequency values from 0 to 0.8%? So its not at all clear how this value was calculated. 2) Please clarify how the uncertainty value is computed? 3) Similar to the above, its not clear where the value of 28.7% comes from as this is not the range in Figure 2.

+ Figure 3: Is the large variability of AP events connected to the large variability in the Amundsen Sea Low / large interannual variability in cyclone frequency in this region (Simmonds and Keay, 2000)?

+ Figure 3: 1) Can you please justify why the range 1995 to 2015 was chosen?  Bluntly, was this cherry picked to get a significant correlation? What if you shifted the range by 1 or 2 years, how does the trend change and its significance? 2) There doesn’t seem to be any mention of what could be causing the positive trend in AR events – this is also noticeably absent from the Discussion. For example, could this be due to decadal changes in the Madden-Julian Oscillation (Hsu et al., 2021; Science Advances) which occurred in the late twentieth century and early twenty-first century?   

+ Could the pressure patterns / anomalies responsible for Ars be compared to the analysis of Scott et al. (2019; Journal of Climate) , which uses ERA5 and a cluster technique to identify dominant circulation patterns.  Perhaps this would be appropriate for the Discussion section.   

Minor / Technical Corrections

+ Line #9: 3 -> three

+ Line #9: Please give the year of the case study.

+ Line #13: I assume the accumulation value is water equivalent. Maybe state this.

+ Line #26: As written this states that all mass loss is from the WAIS, which is not the case as the Peninsula region has surely also lost mass.

+ Line #28: This statement requires a reference.

+ Line #33: TG is undefined.

+ Line #40: What about evaporation? With increasing surface melting this will become increasingly important. For example, Bromwich et al. 2011 J. Climate showed that sublimation and evaporation combined accounted for around 25% of the precipitation term.

+ Line #56: Mention of ‘on the order of the Amazon River’ is confusing. Do you mean the actual river? Is this a type of AR?  Are you referring to spatial size?  I’m afraid that this comparison is not that helpful so please revise.  

+ Line #78: Maybe clarify this sentence a little regarding ‘rely on reanalysis’.   For example, by saying ‘In this study, we rely ….’

+ Line #103: Its not clear whether by observations you are referring to the in situ observations or the reanalysis. See specific comment above. Please clarify your methodology/approach in a dedicated section.

+ Line #146: Is there justification for the 12 hour threshold?

+ Line #206: Its not clear how these average surface pressure maps during AR events are computed.  See comments above. Presumably you identified the ARs and then did calculated a composite of these events. But this really needs to be made clearer.

+ Figure 4: The stippling wasn’t really obvious. Could this be made clearer?

+ Line #219: 1) So you are creating a distribution of the temperatures.  Perhaps this needs a little more explanation. 2) What does ‘all seasons’ mean?  Figure 5 only shows the seasonal breakdown?

+ Line #225: Maybe state melting point of snow/ice.

+ Line #291: 2 -> two

+ Line #317: Again, what is the uncertainty mentioned here. Is it one standard deviation? Please clarify.

General comments

The manuscript by Maclennan et al. presents a study of atmospheric river events and their effects on the climatology and surface mass balance in West Antarctica. First, the MERRA-2 and ERA5 reanalysis products are used to quantify the frequency, trends, and large-scale effects of ARs on precipitation in the period 1980 to 2020. Then in-situ observations from weather stations at Thwaites Glacier are used to reconstruct accumulation and firn conditions during a series of AR events in 2020. Finally, the possible future effects of increasing AR intensity and frequency on surface conditions and surface mass balance in the areas are discussed.

The paper provides a good background of ARs in West Antarctica and their effects on surface mass balance, and the large-scale study is combined with the in-situ data into a very interesting discussion. The topic is timely and the paper is suitable for The Cryosphere.

The only minor issue is that the presentation of the work should more clearly state the goal of the investigations as well as summarize the findings more clearly. As I see it, the main strength and new contribution of this paper is that the authors combine the large-scale reanalysis products with detailed in-situ data. Thereby, they are able to qualify the discussion of the future impacts much more convincingly than from reanalysis products alone. This message should be communicated more clearly. The discussion section is strong, but I suggest that the Discussion and Conclusion section is divided into two, so there is a separate conclusion section in order to communicate the findings more clearly.

Detailed comments:

Page 1: the abstract is far too long. The length should be 250 words (see instruction in the TC). Remove sentences that are essential background or discussion.

Page 2-3: The introduction contains the motivation and background on ARs. However, the purpose of the study is not clearly stated. Rewrite the last paragraph to start with “In this study, we… This would also make it clear from the start how this paper differs from earlier studies by including the in-situ data, and why these data are included.

Page 2:, line 33: I don’t think “TG” has been defined, please do so.

Page 4: AMIGOS – include a reference to define what AMIGOS is, it is not enough to include it in the title of section 2.1

Page 4, line 100: add “s” to sensor, and change “is” to “was”.

Page 5: Perhaps explain a little more clearly why you focus on the precipitation and use the vIVT algorithm to detect the ARs. The precipitation effect is most important at present, but this could perhaps be made more clear here, or stated earlier in the motivation.

Page 7, figure 1: Indicate the 80degS latitude at the figure to the left. This would be helpful later in the discussion. What is the black outline in the middle figure?

Page 10, line 231: Please define “ASE”.

Page 12, line 278: remove “and”.

Page 12, line 280: The spatial resolution of the reanalysis product could both mean that it does not resolve variations within the grid cell, and also that some larger scale patterns are not resolved properly. It could be relevant to mention both.

Page 14, line 306: please correct the reference.

Page 14, line 312: Add the 80degS latitude to figure 1, see comment above.

Page 17: I miss a conclusion section to summarize the findings clearly and provide an outlook.