Review of: “Spatial distribution and controls of snowmelt runoff in a sublimation-dominated environment in the semiarid Andes of Chile” by Álvaro Ayala, Simone Schauwdecker and Shelley MacDonell.

This paper presents an interesting case study of a catchment in the Andes, which is a snowmelt-dependent region in which sublimation plays a significant role on the snow cover and water balance. The paper builds on previous studies focussing on modelling performance and underlying snow processes. The authors perform an elaborate analysis on the hydrological importance of the processes occurring in the Corrales catchment, Chile. In general, this is a well-written manuscript. However, parts of the manuscript require some additional attention, so that the overall quality of the manuscript improves. As such, I advise the paper to be revised before publication. Below I have stated more general and specific comments, which I hope the authors consider to be constructive.

General:

Results:

The results contain a lot of information and figures, all of which are important. However, it is sometimes hard to make the connection to the other results for me as reader. Each figure is treated separately, and not always clearly connected to previous results. To illustrate, almost each paragraph starts with “In figure x, we compare…” or “Figure x shows …”.

I would advise the authors to focus on the point you are trying to make and try to make in-text connections between the separate figures based on the general story. This results in a storyline in which the figures are a helpful tool instead of treating the results as a point-by-point discussion of the figures. Another option would be to merge the results in the discussion, however that is also not sufficiently done currently.

Figures:

All the figures used in the manuscript are important, and significantly contribute to the manuscript. However, multiple figures are rather unconventional. For example, some figures miss an x-axis and/or y-axis label or contain a strange diagonal line through the colorbar. Also, it seems that part of the figures consist of multiple loose figures, which are not all aligned. I encourage the authors to re-do part of their figures, so that these look more professional. (See the specific comments for examples).

Data and code:

I am happy to see that the data used in this manuscript can be found online. However, I highly encourage the authors to also publish their code used for the data analysis. This would make the research more align with the FAIR principles and also accessible for interested readers.  

Specific comments:

• L71-74: The definition of snowmelt hotspots is not completely clear for the reader, especially when reading the paper for the first time. The second sentence could also refer to the areas where snow surface sublimation dominates over snowmelt.
• L101 – 116 and Figure 1: Is discharge data also available? It seems like that based on Reveillet et al., (2020) (L92-94). This would be beneficial for the understanding of the reader, especially when discussing the hydrological importance. (Also later on in combination with Fig. 8)
• L128: Could you briefly elaborate on what this simple method entails? In general, I agree that this method is in the Supplementary materials.
• L224-225: Is there a specific reason why you do not consider rain-on-snow events in the snowmelt runoff variable? Previous studies have shown the significant effect rain-on-snow events can have on runoff. Based on Figure 8, I see that rain especially takes place in summer and autumn, during which temperatures are around 0 ^oC and snowfall also takes place, which could result in ideal conditions for rain-on-snow events to generate relatively high runoff, partly from the snowpack.
• Tables 3 & 4: In these tables the input parameters are presented for the simulations. But it is unclear for me if this results in two “types” of simulations. Do you perform one base simulation (Table 3) and the ensemble runs (Table 4). Or do you vary the parameters in Table 4 as input in the simulations (Table 3)? In the former case I don’t understand where you use this “base” simulation. If the latter, couldn’t these tables be combined?
• L245-246: Could you elaborate on the physical meaning of the slope and curvature wind distribution weights?
• L261: Perhaps I misunderstood the definition of SP, but doesn’t a value of 0.2 mean that there is only 20% of the time snow present during the ablation period? If that is the case 0.2 seems to me also mostly snow-free.
• Figure 3: In the text, you refer to valley bottoms and ridges (L255), but it is hard to come to same conclusions based on your figures. Would it be an option to add isohypses to a and b? Additionally, I would advise to add labels to the colorbars, and add a y-axis label to the c and f figures.

• L269-270: I recommend to include the equation used to compute the coefficient of variation and explain how you compute these terms. This will leave no space for any uncertainties on how you computed these.
• L285-294: The verification of the model simulations partly is performed based on a single observation site. The authors compare snow depth and SWE observed at Tapado with the modelled version of these variables representing the entire grid. Is there any evidence on how representative the measurements are for the entire catchment? How complex are the surroundings of that specific measurement site in relation to the entire catchment? Is the measurement site at a wind-exposed or wind-sheltered place?
• L291: What do you mean with the Geonor sensor? I suspect that is the precipitation measurements based on table 1?
• L308-309: How do you compare the satellite-derived indices with the model-derived indices? Do you use the model values exactly at the moment of the satellite overpass? Or do you average the model values over a certain period?
• Figure 6: Are SA and SP Wayand the observations? Additionally, I would recommend to add a 1:1 line and the equation of the trendline, so it is clear that the absolute values do not match. Also out of curiosity, is there a reason why you do not force the fit through [0,0] (i.e. leave out the intercept). Theoretically, the simulations should be the same as the observations, so would justify removing the intercept.

• L299-316: In this paragraph (also in the discussion), you refer multiple times to the R² as correlation. Formally, R² is the coefficient of determination and not the correlation. Yet, obviously, both are closely related. Additionally, the numbers in the text are not exactly the same as the numbers in the figures.
• Figure 7: What do the different markers mean? Am I correct to interpret these as different stakes?
• Figure 9: I would advise to use the same colorscales for the maps and polar plots. Also, In the colorbars of the maps, some strange diagonal dashed line is present. Lastly, I suspect the caption is not complete.
• Figure 10c: why is there a message in the figure? I agree that this is an important message, but this can also be inferred without the message (and is also stated in the text).
• Figure 12: it is hard to assess which areas are positive and which are negative, due to the chosen colorscales. Also, I suspect the caption is incomplete.
• L433-447: The authors start this paragraph by stating that the model results are in good agreement with the distributed datasets. I only partly agree with them. The R² shows indeed relatively good scores, but this is not the case for the absolute values, which shows that the simulations underestimate the indices at least by a factor 2. I would recommend the authors to also mention the performance based on absolute values and put both these performances in perspective to previous studies. For example, is this known to be a common case with SnowModel? And is there an explanation for these mismatches in absolute values?
• L473-L485: This would be a nice place to discuss the dominant processes that you found in the Corrales catchment and what could be the cause of the snowmelt hotspots. However, you do not go into depth, and only briefly touch upon “the large spatial variability of the physical processes that control snowmelt runoff”. I encourage you to elaborate more on what you found, which could serve as an overview of your findings merged into one story. Discussing this, would allow you to also compare your results with other regions in the world, especially where sublimation also plays a significant role.
• L424-459: I miss a discussion on how well SnowModel generally performs based on the previous studies and how this could relate to your results. For example, could it be the case that SnowModel often overestimates snowmelt in specific parts of a catchment? A discussion on this would clarify whether you actually found snowmelt hotspots or are looking at the modelling uncertainty.
• L486-488: It is unclear what you mean here? What part of the results do you refer to?

In the manuscript “Spatial distribution and controls of snowmelt runoff in a sublimation-dominated environment in the semiarid Andes of Chile”, the amount of snowmelt and the snow sublimation are quantified for a catchment of 78 km² using the SnowModel and two years of measured meteorological data. The paper aims to present the spatial distribution of snowmelt and snow sublimation processes and to do that, it defines so-called ‘snowmelt hotspots’ in the catchment. The study concludes that 50% of the snowmelt occurs in only around 20% of the catchment.

Overall, I found the study interesting, mostly well written and the topic well-suited for HESS. However, I have some doubts about the novelty of the study. It is not surprising that snowmelt is spatially heterogenous, and the paper also cites quite some studies that already looked at the contribution of sublimation exactly at this location. In the introduction it is written that these studies rather focus on models and uncertainty, rather than hydrological importance. However, in this study the ‘hydrological importance’ part is unfortunately not so clear either: only a short statement about the implication of snowmelt hotspots on recharge areas is given. I think the study would clearly benefit from describing more explicitly the added value of this study in the introduction, discussing in more depth the implications for snow science (in semi-arid areas) and explain more explicitly the hydrological importance of the findings.

My other main concern is the presentation of the results and the figures. Sometimes units are not described, the same color bars for everything are confusing, text is added at strange places and captions are not always informative enough. The manuscript presents a lot of figures, which are sometimes only described in very few sentences in the results section and rather in a disconnected way. It would be helpful if the figures and the text together form a story and are answering a question or research gap that is presented in the introduction.

Please find below more detailed comments:

L14: ‘satellite-derived …….product’ – suggest to leave out, because it comes also a few sentences later

L17: ‘absence and persistence’ – maybe add the season?

L19: here the characteristics of the snowmelt hotspots are shortly summarized. Maybe you could indicate which of these elements are likely also applicable to other regions/catchments. The following sentence about “we suggest that snowmelt hotspots play a key hydrological role” is a too strong statement for the abstract as this is not shown in the current study. I suggest to reformulate.

L41: shouldn’t it be the other way around? i.e. decreases the energy and therefore lowers the temperature?

L66: “From another perspective” – not clear what is meant here

L116: here I wondered why the groundwater and hydrological data are not used in this study?

L132-133: Maybe shortly explain how the measured precipitation relates to the mean annual precipitation given for La Laguna earlier in the manuscript (i.e. why 3 times higher, elevation?)

L142: somewhere in this paragraph or earlier, suggest to explain winter/summer and the corresponding months

Section 3.2: it would be good to explain why a model is used when daily SWE maps are also available. Now the reader has to guess based on the Results section. The same explanation is lacking for the SA and SP indices, why are these indices relevant for this study?

3.3 Not a great name for a section and confusing as it still describes another snow product. It would be useful to describe what different information can be obtained from the SA and SP indices and the SCA

L239: why 5 mm and in table 3 1mm?

4.2: is there any routing of the snowmelt runoff?

Table 3: why is the precipitation lapse rate 0?

Table 4: how were these values determined? And is the precipitation correction coming on top of the 30% increase (bias correction) in the precipitation measurements?

Figure 3: some suggestions:

• Add row names with Snow absence and snow persistence, and add in caption the period for which this was calculated (months, years)
• Could one color bar be used for all graphs? Why are color bars for c and d smaller? Could the bars be made such that the colors are more intuitive, i.e. the same color to indicate snow accumulation “hotspots” (high SP values and low SA values?)
• Add axis labels on the axes and not at the top
• What are the white areas in a?
• What are the numbers in b and e? elevation?
• Add explanation about the units of each graph in the caption, i.e. indicate that it is a fraction of time in a,b, d and e and percentage of space in c and f.

L269: “interannual median” – per pixel or for the catchment?

Figure 4, similar issues as with Figure 3. In particular, please provide units in the graphs.

L290: where does “also” refer to?

Figure 6 – it would be more logic to have a maximum y-axis of 100% in figure a.

L307-309: Why was the comparison not based on simulations at the same time as the satellite data? This would be a more fair comparison and indicate if the model is under- or overestimating snow persistence. The same comment for the SWE estimates in the next lines.

Figure 7: What is meant with “to help comparison” in the caption? Why is there only one stake for 2020 at the beginning of the season? Why is the first value not the measured value but the simulated value.

Figure 8: some suggestions:

• Add more x-axis labels
• Indicate somewhere that these are stacked bars
• During summer snowfall events (Jan 2020) it looks like the ice melt flux is largest – why is that?

L361: “where snowmelt is more important than sublimation” – if values are above 50% than this is nowhere the case?

L365: 4.3 mm3 a-1 – to give a hydrological meaning to the number, wouldn’t it make more sense to give it in the same units as P, i.e. mm per catchment?

Figure 9 – Please introduce some consistency in the color bars and ranges. Also the headers could be improved which are now sometimes over two lines and sometimes centered but not always.

L368 “total ablation” and L365 “total ablation” – please reformulate as they refer to something different. This also applies to the conclusion, point 1 which is now a bit ambiguous

L370-375 how are the location of the snowmelt hotspots determined? Is there a clear ranking of which cells are included and which not? The line in Figure 10 c looks rather linear, at least between 10 and 30% of the area?

Figure 10: please remove the text from the figure C

L387: Do the SP and SA values refer to the observations or the simulations?

L392 “consequently” – Please explain this last sentence in more detail

Figure 11: why are cumulative distributions used here? Please also add in the caption which data is in the graphs. For example for the maximum snow depth, is it the maximum depth over the whole simulation period?

The role of Figure 12 in the study does not become clear from the text.

L453: “similar AWS forcing” – what is meant here, as the lines before just explain that there was a different availability of AWS data. “Similarly” – similar to whom?

L477: “In this direction…..this type of environment” – it comes across as if this sentence does not fit the text

L487: “We here show that …” – please describe more explicitly what is the case here in this study, also referring to figures.

L491: “where large part of snowmelt is generated” – repeat where this is and if this is general for semiarid Andes

L507: can ice also sublimate?

In section 6.2, I was expecting a discussion about the “glacier hotspots” too, as they turned out to provide even more melt, but have an even smaller area. It is shortly mentioned, but what is the hydrological implication of glacier hotspots versus snowmelt hotspots?