Modelling the small-scale deposition of snow onto structured Arctic sea ice during a MOSAiC storm using snowBedFoam 1.0.
- 1WSL Institute for Snow and Avalanche Research SLF, Davos, Switzerland
- 2CRYOS, School of Architecture, Civil and Environmental Engineering, EPFL, Lausanne, Switzerland
- 3Thayer School of Engineering, Dartmouth College, Hanover, New Hampshire, USA
- 4USACE-CRREL Alaska Projects Office, Fairbanks, Alaska, USA
- 5NOAA Physical Science Laboratory, Boulder, Colorado, USA
- 6Cooperative Institute for the Research in Environmental Sciences, University of Colorado Boulder, Boulder, Colorado, USA
- These authors contributed equally to this work.
- 1WSL Institute for Snow and Avalanche Research SLF, Davos, Switzerland
- 2CRYOS, School of Architecture, Civil and Environmental Engineering, EPFL, Lausanne, Switzerland
- 3Thayer School of Engineering, Dartmouth College, Hanover, New Hampshire, USA
- 4USACE-CRREL Alaska Projects Office, Fairbanks, Alaska, USA
- 5NOAA Physical Science Laboratory, Boulder, Colorado, USA
- 6Cooperative Institute for the Research in Environmental Sciences, University of Colorado Boulder, Boulder, Colorado, USA
- These authors contributed equally to this work.
Abstract. The remoteness and extreme conditions of the Arctic make it a very difficult environment to investigate. In these regions, the wind has a substantial effect and redistributes a large part of the snow, which complicates precipitation estimates. Moreover, the snow mass balance in the sea ice system is still poorly understood, notably due to the complex structure of its surface. Quantitatively assessing the snow distribution on sea ice and its connection to the sea ice surface features is an important step to remove these uncertainties. In this work we introduce snowBedFoam 1.0., a physics-based snow transport model implemented in the open source fluid dynamics software OpenFOAM. We combine the numerical simulations with terrestrial lidar observations of surface dynamics to simulate snow deposition on a piece of MOSAiC sea ice with a complicated structure typical for pressure ridges. The results demonstrate that a large fraction of snow accumulates in their vicinity, which compares favorably against terrestrial laser scans. However, the approximations imposed by the numerical framework together with potential measurement errors (precipitation) give rise to quantitative inaccuracies. The modelling of snow distribution on sea ice should help to better constrain precipitation estimates and more generally assess and predict snow and ice dynamics in the Arctic.
Océane Hames et al.
Status: final response (author comments only)
-
RC1: 'Comment on gmd-2021-254', Anonymous Referee #1, 05 Oct 2021
In Modelling the small-scale deposition of snow onto structured Arctic sea ice during a MOSAiC storm using snowBedFoam 1.0 the authors present a new software package, snowBedFoam. This software is a snow transport model that uses the CFD package OpenFOAM.
In general the manuscript reads well, but could do with a few clarification passes.
My first concern is the lack of consideration for a sublimation flux. Sublimation of blowing snow and of snow packs has been identified as a major contributor to mass loss in many environments (Mott 2018 ) . The authors do not describe the humidity of this location. Indeed, perhaps it is sufficiently humid that such an approximation is warranted. However, this is never described nor justified in anyway. I see that the MOSAiC companion paper Wagner (2021; TCD) notes low sublimation fluxes for this period, perhaps 6%. This seems derived from modelling studies and not observations. The simulation period is short and perhaps sublimation is negligible. However, that has not been demonstrated and to wave it away, especially when the model is noted to have deviations from observations, seems problematic. I would like to see a sublimation sink added and evidence that, in a distributed context that this process is indeed negligible.
The authors extensively cite Bagnold 1941 without ever noting that it is a sand transport study. I would encourage the authors to cite the vast body of literature surrounding the early work of adopting this work for blowing snow such as, e.g., Budd, 1966; Schmidt, 1982; and Pomeroy, 1990.
My second concern is shown in Fig 5. Specifically the long, low velocity streaks of u*. Ivanell (2018) and Wagenbrenner (2019) have a discussion of similar streaks in RANS models. These areas may have substantial impacts on blowing snow simulations and deserve attention. I would like to see, at a minimum, a description and placement in the literature of these features and if the authors think they are real. Wagenbrenner (2019) identifies that they are somewhat dependent upon the upwinding scheme used. Do the authors think that is the case here?
This brings me to my last concern – a lack of uncertainty analysis. I would like to see the authors quantify the impact of any meteorological forcing, e.g., their input precipitation and the values in Table 2. Certain values can have massive impacts, e.g., friction velocity threshold, and it would be useful to understand how sensitive the model is to these parameters. It is not 100% clear to me how the simulations were done. It seems to be developing a steady-state (?) simulation of 1000 s at which point the wind speed is reduced? How sensitive are the results to this (spin up?) 1000 s period?
The large temporal periods over which the simulation is run from mean values of wind is concerning or at least requires further discussion. The temporal scales that impact blowing snow are quite small, < 1 s (Aksamit & Pomeroy, 2016, 2018), although at 15 m to 1 h r scales, mean shear-stress models tend to be successful. However it is not clear how successful a many-hour mean wind structure is for representing these features. To me it seems a mis-match to run a sub-metre spatial model, but drive it with many-hour mean windflow that we know doesn’t represent any of the wind structure known to drive blowing snow events.
In summary I would recommend this for major revisions. It has the potential to be a unique contribution to the blowing snow literature, however I do not believe to be there in its current form.
References
Aksamit, N. O., & Pomeroy, J. W. (2016). Nearâsurface snow particle dynamics from particle tracking velocimetry and turbulence measurements during alpine blowing snow storms. The Cryosphere, 10(6), 3043–3062. https://doi.org/10.5194/tcâ10â3043â2016
Aksamit, N. O., & Pomeroy, J. W. (2018). Scale interactions in turbulence for mountain blowing snow. Journal of Hydrometeorology, 19(2), 305–320. https://doi.org/10.1175/jhmâdâ17â0179.1
Budd, W. F. The drifting of nonuniform snow particles. Studies in Antarctic meterology 59–70 (1966) doi:10.1029/ar009p0059.
Ivanell, S. et al. Micro-scale model comparison (benchmark) at the moderately complex forested site Ryningsnäs. Wind Energy Sci 3, 929–946 (2018).
Gerber, F., Mott, R. & Lehning, M. The importance of near-surface winter precipitation processes in complex alpine terrain The importance of near-surface winter precipitation processes in complex alpine terrain. J Hydrometeorol (2019) doi:10.1175/jhm-d-18-0055.1.
Pomeroy, 1990 Saltation of Snow WATER RESOURCES RESEARCH, VOL. 26, NO. 7, PAGES 1583-1594, 1990
Schmidt, R. A. Properties of blowing snow. Rev Geophys 20, 39–44 (1982).
Mott, R. et al. Orographic effects on snow deposition patterns in mountainous terrain. J Geophys Res Atmospheres 119, 1419–1439 (2014).
Mott, R., Vionnet, V. & Grünewald, T. The Seasonal Snow Cover Dynamics: Review on Wind-Driven Coupling Processes. Frontiers Earth Sci 6, 197 (2018). Wagner 2021, https://tc.copernicus.org/preprints/tc-2021-126/tc-2021-126.pdf
Vionnet, V. et al. HighâResolution Large Eddy Simulation of Snow Accumulation in Alpine Terrain. J Geophys Res Atmospheres 122, 11,005-11,021 (2017).
Wagenbrenner, N. S., Forthofer, J. M., Page, W. G. & Butler, B. W. Development and Evaluation of a Reynolds-Averaged Navier–Stokes Solver in WindNinja for Operational Wildland Fire Applications. Atmosphere-basel 10, 672 (2019).
Page 1
L1 > “In these” This sentence is not clear
L2 > has a substantial effect
On what?
L3 > a large part of the snow
Of the snow mass? Clarify what this means
L3 > which complicates precipitation estimates.
Oh so this is precip undercatch?
L5 > to remove these uncertainties
It’s not clear what these are. Do you mean surface precip?
L8 > on a piece of MOSAiC sea ice
MOSAiC domain?
L8 > On a piece o
L9 > terrestrial laser scan observations?
L11 > Could help to better constrain precipitation estimates
This is not clear from the abstract why this is the case and it remains un answered.
L16 > seems to respond
Word choice for respond
L16 > very
17 > an important reduction
Significant reduction?
Page 2
L 20 > : the i
Start new sentence.
L29 > Also, Arctic precipitation estimates could be signifi- 30 cantly improved by an accurate assessment of the snow deposition on sea ice
How? It is well known that snow depth on the ground doesn’t equal snow fall due to sublimation
L33 > Connecting the snow mass balance to snowfall
What about sublimation?
L35 > that gets
That is
L38 > the influence of each of these processes
Influence where? Certainly these have been well constrained in many studies
L45 > To our knowledge, such spatial observations of snow deposition are non-existent in the literature for Arctic sea ice.
Is this not a summary of Trujillo 2016? Please clarify
L 54 > which we perform in this study.
Page 3
L54 > spatial variability of snow deposition around complex terrain
To be clear you mean blowing snow transport?
L54 > is not yet fully understood Grünewald et al., 2010)
Is there a more recent reference ?
L 55 > Multiple model approaches exist that try to describe it
If the previous section is indeed blowing snow then this is missing many references. But what is ‘it’ referring to?
L 63 > on a piece of sea ice
Is this the technical definition? Please be more specific
L 63 > Several data sets
Datasets of what?
L 63 > MOSAiC (Multidisciplinary Drifting Observatory for the Study of Arctic Climate
Define this earlier.
L 65 > The first one
Data set
L 79 > This article
Consider“manuscript”
L80 > second stage,
“Section”
L 80 I would like to see scientific questions and hypothesis testing!
Page 4
L87 > data sets employed here
“Used” here
L87 > employed
L 97 > that were successively operated
Do you mean observed?
L 101 > was placed as high as possible
How high is this?
L103 > were recorded relatively
Relative
L103 > Details about the use of TLS for sea ice measurement
Does that describe this dataset or in general?please clarify.
L 104 > and interpolated
Spatially? What method?
L 104 > Several correction
To the interpolated or the raw point clouds?
L 105 > is a last step, the point clouds were aggregated
Is this the QGIS step?
L 110 > with a constant snow density value of 210 kg.m−3
Was this an in situ observation on the ground or fresh snowfall?
L 111 > spatial variability of snow density
Please note what processes impact this ie is this snow compaction or from blowing snow?
Page 5
Fig 1 > during a helicopter flight
This is the first mention of a held. I thought this TLS was from the ship? Please clarify or remove.
L 120 > correction of orientation
I don’t understand what this means. Please clarify what corrections were done
L 122 > among other instruments
Please list the source of all observations.
L125 > that this data
This = Radar derived precip estimates in general or the data used here?
L127 > by releasing particles
Are all the particles same size? I think it is noted they have a distribution later but just note it here.
Page 6
L 147 > walls
I know what you mean but I would like you to-clarify what you mean walls in a natural context to aid a non-domain modeller in reading this manuscript.
Page 7
L179 > which was added within the core code
By the authors? Ie is this part of the scientific contribution?
Page 8
L188 > (Bagnold, 1941)
Is this really the best reference? Bagnold 1941 is sand processes, not snow.
There are more recent descriptions of the blowing snow transport outside of wind tunnels such as:
Aksamit, N. O. & Pomeroy, J. W. Near-surface snow particle dynamics from particle tracking velocimetry and turbulence measurements during alpine blowing snow storms. The Cryosphere 10, 3043–3062 (2016).
that should be referenced.
L195 > Bagnold, 1941)
Bagnold 1941 is sand, not snow. Odd to include it in this list. Would suggest replace it with the early efforts to adopt Bagnold to blowing snow that I list above.
L200 > a threshold value defined as (Bagnold, 1941)
It is not clear to me why a sand-grain threshold is used. Where is the constant A from? Is this a snow value or a sand value?
L 207 > empirical parameter set to 1.5 (Doorschot and Lehning, 2002),
I had assumed equation (11) was referring to eqn (8) in Doorschot and Lehning, (2002) but they note a value for C_{ae}=1. I don’t see any other C_{ae} in that manuscript so please double check this value.
Page 9
L 235 Note in this section the distribution and mean cell size of the hexes
Page 10
L 243 > with STereo Lithography (STL) input data
I know this is a file format but I think this should be clarified
L245 > the vertical grid spacing dz ranges between
Are you using log spacing? Alternatively how were the vertical layer thicknesses decided?
L 255 > corresponding cell of the connected periodic plane
Maybe note this results in snow being added on the upwind domain boundary.
L 256 > at the wall
I would describe this for the reader who is interested in this approach but does not know what a wall means in this numerical context
L258 > set to a null value.
Null = zero in this context? If so just say zero for clarity
L 261 > for a neutral flow
Define what you mean by neutral flow – I assume neutral stability?
Page 11
L 273 > as much as possibl
Page 12
L 290 > time of 1000s
1000 s
L290 > particles aloft in the air got
“were” not got
L290 > through the deactivation of the fluid driving force in snowBedFoam
This is not clear to me exactly what this mean. Please clarify what ‘deactivation’ means for a wind velocity. Are developing a particle steady-state and then turn the wind off and let it settle?
L 290 > the preferential deposition
Preferential deposition arises due to terrain impacts on local meteorological conditions, causing increased deposition on the leeward slopes and decreased deposition on the windward, and is typically a critical process in mountain terrain (e.g., Gerber et al., 2019; Mott et al., 2014; Vionnet et al., 2017). A few places in the manuscript it seems like blowing snow process and the deposition of suspended snow to be called ‘preferential deposition’. I would like to see this tightened up so the reader is not potentially confused.
L 300 > (280m range gate)
280 m, but also what is a range gate?
L 302 > threshold formulation (Bagnold, 1941),
Ah, so this is where A is defined. Please add this to the description of eqn (10).
Page 14
L320 > Additional limitations arise from the forcing of the model.
It is not totally clear to me not why simulate the whole time series? I realize ‘compute’ is offered up, but how prohibitively long would it be? It’s probably outside the scope of this project (but perhaps not) however it would be interesting to know how much worse these assumptions made the model output v. running the model for the entirety of the observation period
L327 > Figures 4 and 5 report the results
Legend needs units (even if its in the caption)
L327 > extrapolated areal
What does “extrapolated” mean in this context?
Page 15
L336 > in blue in the surface friction velocity plots (Fig. 5).
These low friction velocity streaks require more description and a quantification if the authors believe they are ‘real’.
Page 16
L369 > Quantitatively, our model appears only partially successful
I would like to see RMSE + CV for the domain.
L 387 > may be multiple reasons for the
Density assumption is not addressed here.
Page 17
L391 > used four averaged values for wind speed and direction in OpenFOAM to represent a one-week period of measurements.
I am not I surprised this didn’t when we know blowing snow is at higher temporal scale!
L392 > many specific wind conditions
Would be good to muse on if the neglected conditions are similar to those shown by Aksamit 2016, for instance. I would also explicitly note “specific wind conditions” to include, e.g., gusts, etc.
L 420 >
Could some of this be due to ignoring sublimation? Figure 6 (left) suggests an over estimation of deposition in areas such as due north of the middle of B, on the flatter (?) section. The elevation (?) isolines between these two figures are different though, making a qualitative comparison difficult.
Page 18
L425 > tendency to overestimate precipitation
By how much, exactly?
Page 20
L496 > snow distribution patterns were accurately captured
You note the following in the. results section “we observe that the quantitative performance of our o model is not optimal” and Fig 6 shows substantial differences.
L498 > enhanced deposition
Blowing snow deposition, correct?
L 505 > performance
Word choice – ensure it is clear you mean the accuracy of the model output and not the computational performance of the model
Page 26
I would like to see units on the colorbars
Page 28
extrapolated snowBedFoam 1.0
I don’t understand what extrapolated means for these results
-
RC2: 'Comment on gmd-2021-254', Anonymous Referee #2, 03 Jan 2022
The authors present a blowing snow model based on OpenFoam and some newly developed snow transport-related works, which is important for the snow process research area. The manuscript is well structured and easy to read, and all the figures are well presented. However, there are serval issues that need to be fixed before being accepted:1. The finest vertical grid is 0.1m, while the saltation layer of blowing snow is almost the same order of magnitude. It means almost all the interactions between snow particles and airflow are in the first of the vertical grid, which indicates that the wind velocity estimation on particle location is important. Could the authors add more detail about the wind velocity approach where the snow particles located.2. What are the time steps for airflow and particle,respectively? Since this model considers the splash process, the time step for the particle should be very limited to a small value.3. The period of time is a week, which is really long for blowing snow evaluation. The authors may also need to discuss the effects of other snow processes on snow distribution, such as the thermal processes.
- AC1: 'Comment on gmd-2021-254: responses to referee 1', Océane Hames, 11 Mar 2022
- AC2: 'Comment on gmd-2021-254: responses to referee 2', Océane Hames, 11 Mar 2022
Océane Hames et al.
Data sets
MOSAiC Met City preliminary (Level 2) wind data Matthew D. Shupe et al. https://doi.org/http://dx.doi.org/10.16904/envidat.223
Terrestrial Laser Scan Data David Clemens-Sewall https://doi.org/10.18739/A2DZ03304
Model code and software
snowBedFoam 1.0. Océane Hames, Mahdi Jafari, Michael Lehning https://doi.org/http://dx.doi.org/10.16904/envidat.223
Océane Hames et al.
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
800 | 182 | 27 | 1,009 | 8 | 7 |
- HTML: 800
- PDF: 182
- XML: 27
- Total: 1,009
- BibTeX: 8
- EndNote: 7
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1