the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 24 Apr 2026
The importance of alpine blowing snow for cloud processes
Abstract. Numerical models are known to fail in reproducing the large gap that exists between measured ice nucleating particle and ice crystal number concentrations in alpine regions. Improvements have been made by adding different sources of secondary ice production mechanisms into the models. Blowing snow has been identified as an additional possible source of ice particles. Driven by this assumption, we investigate the effect of blowing snow particles using the numerical model CRYOWRF, in which a new saltation scheme has been implemented to better represent the boundary conditions necessary for the blowing snow equations. First, ice crystal number concentrations are compared with measured data from Jungfraujoch, in the Swiss Alps, showing the importance of secondary ice production, blowing snow and microphysics scheme. Then, erosion and deposition patterns are also analyzed, as well as the influence of blowing snow on precipitation. It is shown that our implementation of blowing snow dynamics improves significantly the match between observed and simulated cloud particles.
- Preprint
(5888 KB) - Metadata XML
-
Supplement
(761 KB) - BibTeX
- EndNote
Status: open (until 05 Jun 2026)
- RC1: 'Comment on egusphere-2026-2132', Anonymous Referee #1, 12 May 2026 reply
-
RC2: 'Comment on egusphere-2026-2132', Vincent Vionnet, 22 May 2026
reply
Review of the paper “The importance of alpine blowing snow for cloud processes” by Viaro et al.
In this paper, Viaro et al. investigate why, in alpine terrain, atmospheric models fail in simulating the large difference between the number of ice nucleating particles in clouds and the corresponding ice crystal number concentrations. This topic has already been studied in a large body of literature. In this study, the authors rely on the recently-developed CRYOWRF model that couples the WRF atmospheric model with the SNOWPACK land surface scheme. CRYOWRF includes a double-moment blowing scheme to simulate the transport of wind-blown snow particles in the atmosphere. Modifications to this scheme were recently proposed and are considered in this paper. A parameterization of secondary ice production was also added to one of the cloud microphysical schemes available in WRF. The authors tested various configurations of CRYOWRF at 1-km resolution centered around the Jungfraujoch (JFJ), in the Swiss Alps and used measurements of ice crystal number concentration, ice water concentration and liquid water concentration to evaluate the model. Their results show the importance of secondary ice production and blowing snow as sources of ice crystal number concentrations that need to be taken into account in models when simulating cloud dynamics in alpine terrain.
The topic of this paper is relevant for the community studying cloud dynamics in alpine terrain and its impact on precipitation formation and snow deposition on the ground. However, at this stage, the paper suffers from several major limitations that need to be corrected before it can be considered for publication. These limitations are listed below as general comments and are followed by specific and technical comments.
General comments
1.The title of the paper is “The importance of alpine blowing snow for cloud processes” and can be considered as misleading. Indeed, in their paper, the authors show that including blowing snow particles when computing ICNC improves the agreement between simulated and observed ICNC (as clearly stated in the last sentence of the Abstract). However, the authors barely investigate how the cloud dynamics and associated precipitation formation is affected by the interactions between blowing snow and cloud processes (at the exception of Section 3.6, see below). At the moment, in CRYOWRF, the only interaction between the cloud microphysics scheme and the blowing snow scheme occurs via the sublimation term in the blowing snow scheme and related changes in specific humidity and temperature with potential feedback on the cloud microphysics scheme. The importance of this feedback needs to be properly quantified in this study. For example, the authors could quantify (i) how relative humidity is modified due to the presence of blowing particle in the lowest levels of the atmosphere, (ii) what is the influence on cloud variables (number content, mass concentration, …) and (iii) what is the final impact on simulated snowfall. Simulations could be carried out with and without the blowing snow sublimation term in the blowing scheme. At the moment, the authors only show on Figure 12 a map of difference of cumulative precipitation with and without blowing snow equations using the MOR scheme. No further analysis is carried out. It would be interesting to know (i) how big this difference is relative to the cumulative precipitation during the event, (ii) how does it vary with the complex topography of the region, (iii) how does it compare with the difference in cumulative precipitation generated by two different cloud microphysics schemes (MOR and ISHMAEL).
Adding more interactions between the cloud microphysics scheme and the blowing snow scheme would be a natural next step (as mentioned in the conclusion) but I believe that the current version of CRYOWRF has already enough potential to produce a meaningful first study that really highlights “The importance of alpine blowing snow for cloud processes”. At this stage, it seems that the authors have a somehow unique model with a very strong potential to carry out detailed analysis of complex snow-atmosphere interactions, but they do not really use the potential of this model in the study.
Another illustration of this lack of analysis concerns the potential contribution to ICNC of surface hoar forming at the surface of the snow cover and being transported in the atmosphere when the wind speed is sufficient. Such process is mentioned in Lloyd et al (2015) as a potential origin of ice crystals at JFJ. Viaro et al. explicitly mentions in their paper (L 225 -227) the fact that CRYOWRF can simulate all these processes. It is a unique feature for an atmospheric model. Therefore, it would be very valuable it the authors could quantify if surface hoar formation is simulated in the CRYOWRF experiments and what is the importance of this process.
2. The paper follows the classical structure of a scientific journal, but several sections need to be rewritten and improved before the paper can be considered for publication.
- Introduction: the first paragraph is already very specific. I recommend the authors to start with a paragraph with a larger scope (see for example the introduction in Georgakaki et al. (2022, 2024) on a similar topic). In addition, more references are needed to support certain statements in the introduction (L 4-5 for example).
- Section 2.1: the description of the measurements used to evaluate the model is totally missing. In particular, the authors should explain (i) which instruments they are using among those that were deployed during the 2014 CLAVE campaign, (iii) what are the associated uncertainties and (ii) why they selected this time period for their simulations.
- Section 2.2.1: since the impact of the SIP scheme is evaluated in detail in the results section, I recommend the authors to add a description of this parameterization in their paper. Referring to Appendix B of Sotiropoulou et al. (2020) (S20) is not sufficient. Indeed, S20 implemented a parameterization of collisional breakup (BR) in the Gettelman and Morrison (2015) microphysics scheme developed for global models. S20 focused on the simulations of Arctic clouds. In this paper, Viaro et al. are focusing on clouds in alpine environment using the Morrisson et al. (2005) scheme. Therefore, details are needed about their implementation. It can be placed in an Appendix if it adds too much content to the text.
3. The authors compare simulations with the MOR and ISHMAEL schemes available in WRF. They show that the two schemes simulate large differences in ICNC with ISHMAEL being closer to the observations than MOR when SIP is not considered. Adding SIP to MOR brings the results closer to ISHMAEL. The authors conclude the “changing primary ice production treatment in models does strongly impact cloud characteristics” (L189-190). To better understand this point, it would be good if the authors could explain which features of the two schemes are responsible for this difference. In the current manuscript, the authors do not provide details about these two schemes and how they treat solid particles in the clouds. The authors also mention that adding a collision break up mechanism in ISHMAEL could change the results and potentially lead to an overestimation of ICNC (L 230-231). In this context, what does it mean in term of the real-world importance of the physical process of collision break up if its representation needs to be added or not in could microphysics schemes?
4. Section 3.5 describes the erosion and deposition patterns simulated by the blowing snow scheme in CROYWRF at 1-km resolution. This section has limited added value since sub-kilometric resolutions are needed to simulate any meaningful patterns of erosion and deposition in alpine terrain (e.g. Mott et al., 2010; Vionnet et al., 2014). Haddjeri et al. (2024) have shown how a resolution of 250 m is already coarse. Therefore, I recommend the authors remove this section and add more content related to the importance of blowing snow on cloud processes (see general comment 1).
5. The comparison of the different configurations of WRF and CRYOWRF is purely based on the visual comparisons presented on Figure 5, 6 and 7. It would be useful for the reader to have a Table that summarizes the error metrics (bias, RMSE and R2) for the simulations of temperature, relative humidity and wind speed by the different model configurations T.hat would support the statement: “All simulations …. agree well with observation” (L176). For the cloud variables, the author could use error metrics and/or statistics (cf for example Table 3 in Georgakaki et al. (2022))
6. The simulation of blowing snow occurrence in a model such as CRYOWRF strongly depends on the surface snow conditions simulated by SNOWPACK. SNOWPACK is a detailed multilayer snowpack scheme with several prognostic variables for each snow layer. This makes the initialization of the snowpack conditions challenging when starting a simulation in the middle of the winter (25 January 2014 at 00:00 in the study). The authors should explain (i) which strategy they are using to initialize SNOWPACK and (ii) what is the impact on the simulated blowing snow occurrence during the CRYOWRF simulations.
Specific Comments
L 15: Explain in one or two sentences what is the Hallett–Mossop process.
L 30: Mention at which height above the surface are taken these wind speeds.
L45-46: for which purposes have Saigger et al. (2024) highlighted the necessity of including blowing snow within atmospheric models?
L80: in the model configuration, what is the height of the lowest atmospheric level and the number of levels in the lowest 100 m of the atmosphere. Such information is very useful to know when simulating blowing snow in alpine terrain.
L84-85: Could the authors add the corresponding references for the MODIS 15s and Copernicus 3s datasets?
L 86: A reference is missing for the RRTGM scheme. The acronym should also be detailed.
L 137- 154: I understand well the importance of code management and the challenges associated with models relying on different libraires and schemes available through different repositories. Nonetheless, the description in this paragraph is very technical and does not focus on the main topic of this paper. Since this is not a publication in Geoscientific Model Development, I recommend the authors to move this paragraph to an Appendix.
L 156 – 165: This paragraph describes several methodological aspects related to model evaluation (comparison of simulations with observations) and model configurations (choice of the PBL scheme and associated vertical discretization). It should be move to Section 2 that describes the methods.
L 159: Since model results are extracted at the nearest grid point to JFJ it would be good to know what the elevation of the model at this grid point is and to compare it with the actual elevation of JFJ. A difference in elevation can create systematic biases (for example for the simulation of air temperature).
L 180-182: it is not clear at all why the authors mention the HICAR model here. If they want to mention the potential of HICAR for similar studies on the importance of blowing snow for alpine cloud processes, I recommend mentioning it in a paragraph in a section dedicated to a discussion about their results.
L 186: can the author provide an estimate of the additional computational cost when using the ISHMAEL scheme.
L 193-194: since this section focuses on primary ice production, I recommend the authors to move this sentence to the next section about SIP and the importance of blowing snow particles.
Figure 5: On this figure, it is not clear if the observations and the model results are considered at the same frequency.
L 207: it is not easy to identify on Figure 7 the period that corresponds to temperatures around -15 °C. It could be useful to add on Figure 7 the observed air temperature (as Fig. 3 in Georgakaki et al. (2022)) or to add a figure that show ICNC as function of temperature for the observations and the different model configurations.
L 216: Same as my previous comments, it is not easy to find on Figure 7 the two time windows of velocities stronger than 10 m s−1. Could the author highlight them on Fig. 6 and 7? Maybe a figure that shows ICNC as a function of wind speed for the observations and the different simulations could nicely highlight the fact that simulations without blowing snow fail at capturing ICNC for the strong wind speeds.
L 225: What is the threshold wind speed for blowing snow initiation associated with surface hoar in SNOWPACK?
L 232-237: given the fact that multiple sources contribute to the observed ICNC and all the uncertainties associated with modelling of blowing snow, the authors cannot conclude that their newly implemented saltation scheme is better than the previous one based on the comparison of their simulations with observed ICNC. For example, the wind speed is strongly overestimated by WRF between 26 Jan 00:00 and 26 Jan 12:00, this will automatically result in an overestimation of the blowing snow intensity simulated by the model.
L 275-277: I understand that the authors want to promote HICAR which is a very promising model to study snow/atmosphere interactions, but I do not understand why references to HICAR are added several times in the paper. As mentioned in a previous comment, I recommend the author to build a solid Discussion section where references to HICAR could be added.
Technical Comments
Text
L1: Maybe mention “Atmospheric models” instead of “Numerical models” which is rather vague.
L4: Again “numerical model” is vague, the “snow-atmosphere coupled model CRYOWRF” would give more information to the reader.
L24: Define the acronym “SIP”
L 170: I am not sure that we can say that the wind modulus “is directly related to blowing snow”. Maybe you could say “which influences blowing snow occurrence and intensity.”
L 259: “Wavy patterns” is a non-scientific description and should be rephrased
Citation: https://doi.org/10.5194/egusphere-2026-2132-RC2
Model code and software
CRYOWRF-WRF WSL-CRYOS https://gitlabext.wsl.ch/atmospheric-models/CRYOWRF
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 253 | 85 | 26 | 364 | 23 | 17 | 16 |
- HTML: 253
- PDF: 85
- XML: 26
- Total: 364
- Supplement: 23
- BibTeX: 17
- EndNote: 16
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Review of “The importance of alpine blowing snow for cloud processes” by S. Viaro, A. Sigmund, E. Thomas, and M. Lehning submitted to ACP for publication
Many studies have shown that the number of ice crystals observed in boundary layer clouds is much larger (by orders of magnitude) than the available ice nucleating particles. This paper investigates whether blowing snow can be an effective secondary ice production mechanism that increases the number of ice crystals in cloud systems in alpine regions. Nested WRF simulations with an innermost 1km nest are run using 2 different land surface models, 2 different microphysics schemes, and with and without blowing snow. The ISHMAEL microphysical scheme predicts the evolution of ice habits and density while the Morrison scheme uses 3 categories of ice with fixed habits and densities. There are many other differences between the two schemes as well, specific to this study, the secondary ice production mechanism by rime splintering following the parameterization of Hallett and Mossop (1974) is included in both schemes but implemented differently. An additional secondary ice production mechanism of ice-ice collisions is included in a sensitivity study with the Morrison microphysics. Blowing snow is modeled with a two-moment scheme but importantly, only interacts with the microphysics through the water vapor source-sink term.
Summary: This study shows that blowing snow can reach heights that impact cloud processes, but the studies of the different microphysics and SIP mechanisms are very limited, and I think additional studies need to be done to identify what processes cause the different results. Also, to understand how blowing snow impacts cloud processes the blowing snow ice crystals need to be included in the microphysics. The conclusions are overstated or not supported by the model studies. For example, in the conclusions it says, “When blowing snow particles are included in these schemes, the overall effect on the results is comparable to a simulation with a standard microphysics scheme that includes secondary ice production.” Is this meant to describe the MOR_noBS and MOR_BSorig runs? I don’t see this result in the model studies presented and discussed. It also says in the conclusions, “We also confirm the importance of adding secondary ice production mechanisms to improve the estimation of ice particles”, which is not supported by the model studies.
Main Comments:
Minor comments: