|Review of revised manuscript: "Snow model comparison to simulate snow depth evolution and sublimation at point scale in the semi-arid Andes of Chile" by Voordendag et al.|
The authors provided a thoroughly revised manuscript, with most issues properly addressed. There are a few remaining issues though, which I think the authors should take into consideration before the manuscript can be accepted for publication. Please find those below.
The most important issues are with the new section "Idealised simulations". First, I'm not sure if idealised is the best term, since it mainly concerns the reconstruction of precipitation from SD and/or SWE. So maybe the title should reflect that. The main confusing part for me here is what is considered optimal or idealised. In Fig. 4, the thick line indicates the most optimal settings. I noticed in that figure for the SNOWPACK results that the drop in SWE towards the end of September in Fig. 4c is much smaller than in Fig. 3c. So there are clearly other settings being used, and it is unclear to me how the results can be so different when the manuscript suggests that both figures are created with a kind of optimal/best settings. It has an impact on the precipitation reconstruction from snow depth, since SNOWPACK will add precipitation to match the observed snow depth after this drop that occurred at the end of September.
Some minor comments regarding Section 4.1:
- L237: If possible, please provide some details of the crashes (temperature related, model bugs?)
- L249: It looks like that there was a high wind speed event mid-July. That certainly could have resulted in some erosion in reality, which is not simulated by the models. I think that that is as much an explanation as the overestimation of PSWE at the end of June.
- L251-252: As I mentioned earlier, it looks like melt is overestimated (since there is a lot of runoff simulated around this time when looking at Fig. 6c,d). So then the models need to simulate precipitation to bring the SWE/SD back to observational levels.
- I struggle to understand the legend in Fig. 3, particularly when it comes to what the blue line represents, since it includes both "uncorrected precipitation" and SD/SWE, where I assume it's actually corrected precipitation by using the model's ability to assimilate SD/SWE?? I think at least the figure caption should be improved here. The figure caption should also explain why SnowModel has multiple blue and red lines.
- I strongly recommend to add the undercatch correction from Wolff et al. (2015) in Panels 3e and 4f.
-L3/4: "While many studies focus on evaluating these uncertainties, issues still arise, especially in environments where sublimation is the main ablation process." "issues still arise" is very vague language, please improve this sentence with some concrete examples.
-L60-65: Instead of only listing those studies, please provide details of what those studies found (particularly the results relevant to this study).
-L127: "includes *the* MeteoIO preprocessing library"
-L143: "new snow depth" This is somewhat confusing, since new snow depth can also be interpreted as the increase in depth from precipitation. Maybe "remaining snow depth"?
-L157: "models were calibrated". The examples given indicate that it is not a calibration. If the soil albedo is set to 0.15 based on measured albedo, I would not consider that a calibration. I interpret calibration as trying different values and see what matches SWE or SD best.
-L256: overestimation of what exactly?
-Fig 3: the x-tics in (a) do not align with the x-tics in the other subfigures. Please correct, since the figure suggests a common x-axis.
-L275: I actually struggle to see this in Fig. 4a,g.
-L308: "e.q." -> "e.g."
-L328: Since for SnowModel, you talk about clusters, I think it's appropriate to talk about clusters here too. Instead of using the language of "range between 1.41 and 2.96", since it's not really a range. When I interpret it correcly, SNOWPACK produces two clusters in sublimation rate, based on roughness length.
-L339: "Z0" -> "z0"
-L365: Since the main objective of the study seems to be to quantify sublimation, and sublimation is most strongly impacted by roughness length (see the dicussion on the clusters found in model results), I think that should be discussed before albedo.
Throughout manuscript: instead of talking about with and without precipitation uncertainties, I suggest writing including and excluding precipitation uncertainties. I think that makes it more clear.
-L356: "such as grain size and snow surface area". It's actually called "specific surface area", but please note that SNOWPACK currently does not consider SSA specifically, since it's microstructure model is constructed based on totally different parameters. I suggest writing "grain size and microstructure".
-L379: "as observations are biased". It's bit vague which observations are meant here. I assume authors mean that deriving density from SWE/SD measurements is biased over direct density observations using manual measurements?
-L495: Please update the SNOWPACK repository link. It seems to have changed recently.