see attached

This is an interesting work dealing with the micromechanical modelling of the rheology of snow with various microstructures, mostly described by the solid volume fraction. The paper fits the scope of EGU sphere. There are a number of points that are worth to strengthen in the actual manuscript, therefore I recommend publication after major revisions (details below) have been taken into account.

• Equ (1) and (2): please indicate which strain-rate and stress components represents \sigma and \dot\varepsilon. Line 32, and also later in the manuscript, I am not expert in snow rheology but using the term “Glen’s law” for snow is misleading. Glen’s flow law is used for ice, which is considered incompressible and therefore it is the deviatoric stress that enter into the formula. Here, snow deformation is not isochoric, so it departs from Glen’s flow law used in glaciology.
• Line 44-45 “Rheological models, for instance, the Maxwell, Kelvin, or Burger models”: the link with previous sentence is not clear. I think the authors should write that equ 1 and 2 are for steady state creep, whereas the Maxwell, Kelvin or Burgers models account for the transient creep regime.
• Line 115 The authors should clearly indicate the bibliographic references for the tomographic images used in fig 1, either in the figure or in its legend. Same for fig 5
• Line 124 explain what SSA means
• Line 131 “Air was modelled as an infinitely soft elastic medium”: having infinite mechanical contrasts in the spectral (FFT) scheme often prevents the model to converge. Please detail.
• Lines 150-151 for a vertical compression, \varepsilon_{zz} should be negative, not positive
• Section 2.1.2 the rheology used at the crystal scale should be clarified. I guess it is of Maxwell type, i.e. without considering transient creep response at the crystal scale as in Suquet et al 2012?
• The microstructures indicated in fig 2 and 3 show very few contacts between the crystals. This should quite heavily influence the overall rheology. It would be interesting that the authors add a comment on this, and provide a statistical quantification of the contact area, compared for example with the Voronoi microstructures used in the Vedrine (Acta Mater 2025) paper.
• Line 158, what do you mean with “The numerical integration of crystal plasticity is computationally expensive”? Provide details?
• Line 171 I don’t understand why you speak about “an elastic regime”. Here there is no threshold for the activation of plasticity, so viscoplastic deformation should occur whatever the prescribed stress level.
• Along the same line, I think you should not speak (line 172 but also elsewhere in the manuscript) about a “yield stress”, which refer to some stress threshold to activate plasticity. A better word is probably “flow stress”?
• I find lines 201-204 unnecessary (delete?)
• As written, equation (6) is not a “minimization problem”
• Lines 232-234 are not clear. What do you mean with “BC have a negligeable effects”? Effects on what? BC effects is generally considered with care specially because they can significantly affect model results.
• The word “frustration” is often used in the manuscript, but it is not common in the field of micromechanics. Please detail what is meant with it.
• Line 245 it could be interesting to compare your results with the lower bound (uniform stress within the specimen) as it also leads to n=2 too. How far is the stress heterogeneous (ex. Standard deviation) wrt to volume fraction?
• 256-260 I have no idea from where comes the statement “The solid fraction sensitivity m relates to the heterogeneity of viscoplastic deformations within the sample” which sounds weird to me. Please explain or provide the bibliographic reference. And next sentence, the “m” value computed for linear behaviour (such as thermo-elasticity) is probably very different from the “m” computed in the viscoplastic regime. Could you compare both?
• Line 270, “suggesting that under loading, the microstructure tends to optimize itself”?? You mean that there is some microstructure evolution during deformation, but this is not considered in your computation, so what might be the effect? And I don’t see the link with the over- or under-estimation of the reference stress indicated in the previous sentence.
• Section 3.2 Why do you call this section “experimental data-driven model”?? I don’t understand what it means. And why “data-driven”?
• Fig 8 and lines 338-354. First, I don’t understand the figure, as you represent in colour the R values (differences between V25 and Br92 strain-rates) but the points and dash lines are stress vs volume fraction. Why do you say then that” This agreement supports the validity of the V25 model” (line 349) as you don’t show that V25 reproduces the Col de Porte data but only that it is in agreement with Br92 (without showing that Br92 reproduce Col de Porte data)? Same for the last sentence “Br92 over-estimates the strain rate for Φ > 0.08 and stresses smaller than those observed at col de Porte, Br92 under-estimates the strain rate for very low-density samples (Φ < 0.08) and for stresses higher than the "standard" ones” as you don’t show direct comparison between modelled stress and field stress.
• Fig 9, 10, 11, you write line 368 “This result aligns with observations” but the observations are not shown…
• Line 404, the “crystalline structure” of ice Ih is very well known (space group P 63/mmc, etc).
• Lines 416-421 I don’t understand why you state that “our model is one-dimensional” as your FFT computations and microstructures are 3d, so you deal with full 3d problems…
• Line 441, “In fact, the linear viscosity η, a function of solid fraction (which itself is a function of stress), can artificially introduce non-linearity”? Do you really mean that the volume fraction depends on the stress? Perhaps you mean cumulative strain?
• Line 455 the indicated elastic behaviour is not isotropic but transverse isotropic
• A good point is that the Vedrine et al. paper [Acta Mater, 2025] that is often cited in this manuscript has been accepted for publication.