the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Simulating the effect of natural convection in a tundra snow cover
Abstract. As a straightforward solution to improve the one-dimensional physics-based multi-layer snow model SNOWPACK, a C++ interface is implemented for a tight coupling between SNOWPACK and OpenFOAM with the aim to investigate natural convection in real snow covers. OpenFOAM simulates convection in two dimensions based on SNOWPACK snow profiles and feeds the convective vapor fluxes back to SNOWPACK. Among different snow covers, significant convection events that are well synchronized with cold events are numerically observed by the SNOWPACK-OpenFOAM coupler for a herb tundra permafrost at Bylot Island only if we neglect wind slab formation for surface layers. We find significant sublimation in downward flow regions and that convection generates a low-density path extending vertically almost through the whole snow column. The strong footprint of convection on snow density and temperature with significant lateral variations up to 90 kg m−3 and 5 K respectively makes a consistent representation in the SNOWPACK one-dimensional profile difficult. But the most prominent effect of convection on the snow density profile, namely a low density foot and high density top, corresponds qualitatively to observations. Further work is required to adapt physical parameterizations in conventional snow models to represent this effect and its interaction with snow settling and metamorphism.
- Preprint
(17859 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2025-3035', Kevin Fourteau, 06 Aug 2025
-
RC2: 'Comment on egusphere-2025-3035', Anonymous Referee #2, 05 Sep 2025
This is a review for “Simulating the effect of natural convection in a tundra snow cover”. Overall I really enjoyed this manuscript. Over the past few years I’ve read the various papers regarding the importance of estimating the near-ground low-density layer, and how difficult this is to simulate. The numerical scientist in me really likes the full-complexity model approach to attempt to bound the behaviour. A “what will this take” type of approach is great to see.
I have two broad critiques, and both should be easily addressed.
First, this paper absolutely needs a set of research questions to guide the model development. Indeed the goals allude to this around L65. But it needs to be research questions otherwise it’s “I ran the model and it did Y”. I raise this because I think this paper sells itself a bit short in the discussion and conclusion. These are really valuable insights into a) the difficulty and b) importance of this process. I use models like Snobal and FSM in most of my work because it is computationally fast enough to deploy at a high resolution over millions of km. I’m left a bit unsatisfied as a snow modeller as to the next steps. Indeed you note some, but I mean practical next steps for the community. OpenFOAM is a massive hammer that is not going to be deployed over anything but a small point-scale domain (as done here) due to the computational and IC costs. So, I would really like a discussion on how, by answering these RQs, the authors can advance the more “applied” science. First, it’s clear that even though massive computational effort was used, the results are still not correct (e.g., density profiles). So, how do the authors think these results can inform parameterizations in other less complex models? Is it possible to parametrize around this without the full physics of OpenFOAM? I get very excited with these results as they are a clear step, but they are bespoke. What’s next in advancing the snow modelling communities characterization and simulation of these layers?
Second, this paper needs a tighten w.r.t how some of the sections are presented. I guess it should have been obvious early on, but it wasn’t that this was a 2D (x,y) model instead of a 1D col model with the processes added. I mean, it’s obvious as I write this, but when I was reading the paper it didn’t immediately jump at me. I think that there ought to be a small schematic just showing the (numerical) experimental setup in the main text to lead the reader through what is obvious to the authors (perhaps an opportunity to graphically detail the layer coupling?). I found the equation at l135 to come out of no where, and either walking the reader through this more or cutting it would be my suggestion.
Specific comments follow:
w/c = word choiceL1: “straightforward” OpenFOAM is almost never considered simple!
L3 convection of what?
L4 “feeds” w/c
L5 “numerical observed“ -> simulated?
L5 “coupler” remove? w./c -> model?
L6 “only if [..] surface layers” perhaps remove
L7 “downward” w/c -> vertical?
L9 “in the SNOWPACK” insert “non-coupled” or similar to clarify
L11 “this effect” what effect? be specific
L12 “and its interaction[…]” within snowpack or the surface wind model?
L17 “i.e” -> i.e.,
L22 “finite” as opposed to infinite distances?
L22 “and changes in”
L24 “significant” w/c as significant implies statistical significance
L25 “convection” within the snowpack
L27 “convection of water vapour” a small blurb about what this is might be helpful
L28 “convectional” do the authors mean conventional?
L29 “Subarctic” -> subarctic
L34 love the ecological tie in. super important
L38 “SNOWPACK, as a […]” remove as
L48 “direct” for the snow modeller not versed in numerical methods you might contextualize direct versus alternatives.
L48 “accuracy between […]” versus what?
L50 I find this sentance awkward and difficult to fully comprehend. It doesn’t lead well into the next sentence either, so it’s not super clear what the authors wish to convey.
L52 remove “, such as those […] Island”
L53, L54 “no physically accurate” … “rough approximation” suggests it’s at least partially physically accurate ??
L55 “natural convection” as opposed to?
L61 “is not possible” why?
L70 turn these into research questions that will guide and support this manuscript but also inform the next steps
L72 strong disagree that this is a straightforward approach.
L80 maybe it’s difficult to do but I think a figure illustrating this section would be helpful
L81 “internal elements” w.c -> computational elements?
L91 OF=OpenFoam, SN=Snowpack needs to be in the text
L105,120 could be shorter and tighter
L105 “only feedback” I think this shoudl be re written in the vernacular of couplers e.g., flux exchange between the models etc
L135 This eq comes out no where I found, and it’s complex enough to need a lot of time to read
L135 “presented as” is confusing. was this modified from Jafari 2022 into this? or something else?
L138 Jv is missing definition
L164 Courant of 5. This feels a bit adhoc of a choice. How sensitive are the results to this choice?
L164 PIMPLE = ??
L167 “faster” It might be faster but is the result right? Providing the wrong answer faster isn’t interesting. I think your text suggests it’s fine w.r.t error, but this can be tightened up a lot.
L175 “computer runtime” suggest you use the HPC vernacular of “wall clock” or similar. You need to distinguish between core hours and wall clock either way.
L177 “small differences” define small, e.g., \eps < ??
L205 Above I note that explaining what this convection process is would help the reader. e.g., including the info from here would be good. “It’s a process that is dependent upon thermal gradients and high density, ….”
L2XX these plots are really nice
L2XX i wonder if fig 3 and 4 could be made a 4 pane single plot?
L239 I am quite surpsied and how wrong the density still is!
L239 Figure 7 caption, define H and \rhosum If it’s the same snow height as fig 5,6, please use the same y axis label
FIgure 8, 9 need axis labels
Figure 8, I don’t know that I really had a prior to what I expected this to look like but the symmetrical convection cells were perhaps not that. The figures are very cool and informative, but they need a tighter description to lead the reader though — 2D snowmodels are pretty rare.
L286-9 great result
L309 “benchmark” w/c/ observations?
L333 It would be great to have a simple concluding sentence somewhere in here where you can attribute X% uncertain to ignoring this process. A nice quotable sentence for other papers to cite, to really hit home “ignoring this, even if not perfect, is costing X% in uncertainty”
L335 Open science and data section?Citation: https://doi.org/10.5194/egusphere-2025-3035-RC2
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
286 | 19 | 11 | 316 | 10 | 12 |
- HTML: 286
- PDF: 19
- XML: 11
- Total: 316
- BibTeX: 10
- EndNote: 12
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1