Overall:

This paper describes a new model scheme, in a suit of the CryoGrid community model (version 1.0. Under review at GMD), to simulate the temporal evolution and the vertical distributions of ground ice content by calculating ice segregation (excess ice) when cold and thaw consolidation when warm, and associated ground heave and subsidence. The model incorporates soil mechanical processes, soil hydrology, and soil freeze/thaw physics. The authors conducted a series of proof-of-concept examinations of the new scheme with respect to climatic (i.e., thermal, and hydrological) conditions, soil types, external loadings, and sedimentation, which demonstrated reasonable performance of the model. It is not so simple to evaluate an additional module when the base model is still under review, however, the reviewer found that the manuscript is moderately well organized and written. Yet, some elaborations and clarifications regarding the points raised below will improve the manuscript before being published in the Cryosphere journal.

Issues:

• Ll. 144-146: When it is referred by a general term “soil”, is it assumed that the saturation (in terms of volumetric water content?) is equal among the constituents (e.g., mineral, organic)?
• L.146: How do you justify the threshold value of 50%? Some reference or practice examples would be helpful.
• L. 154: Does “grain” refer to mineral only, or include organic matters in the module (or CryoGrid) terminology?
• L. 174, l. 211, and l. 454: “section 2.1” is used for a reference, however, section 2.1 includes many subsections. It is more reader-friendly to provide more specific reference (such as at l. 241 “section 2.1.2”).
• Ll. 205-207: It is not clear how pore water pressure is updated with respect to eqs. (4) and (9).
• Figure 3: What does the solid straight blue line in the figure (with a filled reverse triangle on the left shoulder) denote? If it explains something it should be noted either in caption or text. Otherwise, should be removed.
• Ll. 257-258, 278: Description of the range of temperature variations and the period of measurement are somewhat awkward. It is assumed to be meant something like “The observed ground temperature at a depth of 20.75 m varied between -9.0 and -7.9 C during the period of 2007 to 2016.”
• Ll. 262-263, l. 281: Description is somewhat sloppy. The forcing data should have been taken from “the CCSM4 outputs simulated under the RCP8.5 scenario”, or “the RCP8.5 run by CCSM4”. Why there is no reference on this model or its CMIP5 outputs?
• L. 292: It is not clear the “key difference” from what is discussed.
• L. 297: Like the above issue, it is not clear from what the upper 0.15 m was unchanged.
• L. 346: it is not so clear what is meant by “the annual downward thawing in summer deepens.”
• L. 361: Is this necessary to mention “that are forced with data from this location”? Any specific reasons?
• Table 3. Why is no value given for maximum snow depth in the B-clay scenario? Does this mean snow depth is unlimited?
• Ll. 401-408: It is difficult to read changes in ground surface height from figures 6, and 1-2 in Supplement B so that it is not easy to follow the statement and discussions.
  
  Further, for the sake of reader-friendly discussions on evolution and relative impacts of ground heave (subsidence) and ice segregation (thaw consolidation), it is suggested to plot segregation ice and surface height changes together (possibly overplotted in figures 7, 8, and 10?).
• Figures 6, 9, 11 and 1-4 in Supplement B: The vertical axis of altitude seems to be the model surface height (say, absolute altitude). Unless it is necessary (e.g., in laterally coupled cases), it would be easy to follow the height changes if shown by a relative height with reference to the initial surface altitude.
  
  Also, the near-surface zones such as active layer is very small, leading it difficult to read and compare especially for the argument of thermal gradients (e.g., ll. 401-408, ll. 420-425). Elaborations will be very helpful.
• Ll. 416-417: “In model scenario S-clay-rain50, the permafrost does not degrade to the end of the 21st century and no talik develops as less energy is consumed for thawing and melting of soil water under drier conditions” (underline by the reviewer). I am puzzled how this argument makes a sense.
• L.419: “1908s” -> “1980s”?
• Figure 7, ll. 418-425: A general feature of Figure 7 looks that the B-clay run produced more segregated ice compared to S-clay run after the 1980s except for the oscillatory decadal periods of comparable segregated ice formation in the 2020s, 2050s, and 2080s. It does not seem a simple warming phenomenon. The argument developed in this paragraph are not well-founded or convincing.
  
  Also, it is suspected that the vertical lines showing the end of the spinup periods may be off by 10 years or so if that for the S-runs ends in 1969 and that for the B-run ends in 1989.
• Figure 10. The scenario names are different from those found in Table 3 or text.
• Ll. 478-480: I wondered the relative contribution of ice and sedimentation can be quantitatively assessed if the density of ice is constant (and it is apparently assumed so in this module).
• Ll. 535-538: It is not clear which figures in supplement B would support the argument.
• Ll. 554-555: “Due to the soil characteristics of the clay, mobile soil water occurs next to the ground ice and is pressed out due to higher total stress” It is not clear whether this statement is meant to be general explanation of the real world, or the result description of the model. Similar ambiguity or confusion was found sporadically in the text.
• Ll. 565-566: “Therefore, the time period available for the formation governs today’s segregated ice content.” This is another example of the above issue.
• L. 621: It is not clear what “fields sites affected by carbon-rich soils” mean or denote.
• L. 642: It is not clearly stated which part of the manuscript support the claim of being able to “lead to improved simulations of ice-rich ground responding to a warming climate.”
• Ll. 650-652: “Climatic conditions, which lead to large gradients…” It is not general “climatic conditions” but warm and moist conditions that derive. It needs to be specific and precise. It is suggested to check the overall manuscript from this perspective.

Juditha Aga et al presented a new modeling capability added to CryoGrid model to simulate ice segregation and thaw consolidation considering proof-of-concept scenarios. While the capability is important and of interest to the modeling community of permafrost regions, especially considering ground heave and subsidence evolution under a changing climate, I have a few concerns that are listed below:

Major comments:

1. A better motivation for the study is needed in the Introduction section; why should one care about ice segregation; where in the Arctic they form; what observations show us; what has been done already to address ice segregation (not Earth system models but small-scale models). Field imagery (and/or a schematic) of ground heave and subsidence can help set the stage as ground heave and subsidence are linked with ice segregation and thaw consolidation.
2. Lots of details have been provided about the model in the Methods but they are hard to follow, especially the description of CryoGrid class (modules) needs a better workflow; a schematic would be helpful to understand what parts of the existing CryoGrid have been modified and what new parts are added.
3. The manuscript mainly focused on Samoylov (S) scenarios and not Bayelva (showed just one B-clay). If the focus is more on one site, it would make sense to drop the other to not confuse the readers. Samoylov’s climate can be made synthetically warmer as the authors are performing proof-of-concept simulations and not validating the model.
4. While I personally believe projections are not needed here to demonstrate the capability, since they are there how do they compare against simulation without the formation of ice segregation?
5. I have also some reservations about the reference date of August 31 when using different climates, this might be fall freeze-up time for one site but not for another site. So, the comparison that on August 31 one site showed this much ice segregation and another showed that much is not very convincing.
6. I would suggest just focusing on the historical climate (1000s of years) and showing how thick ice lenses (segregated ice) formed in the past – at least show the simulated segregated ice is in some comparison with the field observations. Showing the formation of 2-3 cm segregated ice is not very appealing given it is a modeling paper and the main focus is ice segregation and thaw consolidation, so the authors need to explore it further.

Minor comments:

L2. What does “very ice-rich soils” mean?

L24: Sibiria should be Siberia

L2531-34: the formation of segregation ice needs a clear description. Cryosuction can even happen from top to bottom.

L43: “thermokarst and subsequent ground subsidence” does it mean thermokarst happens first and ground subsides afterward? What is thermokarst?

L71: “contrasting permafrost sites” in terms of what?

L94: relative humidity or specific humidity? Please specify

Section 2.1 Model description. A schematic here can really help understand all the processes in the model and how these processes are linked together. While the authors have explained the model in the text, it is still confusing for those who are not very familiar with the model.

L106: define the sub/superscripts.

L128: "table" should be Table

L168: This is a model development paper, instead of referring the reader somewhere else please provide the definition here.

L247: So Bayelva is warm and maritime, and Samoylov is cold and continental? The climate is mentioned for one site but not for the other.

L251: “In summer, precipitation and evapotranspiration balance each other.” It is not clear if this statement is only for this particular site or not. Please clarify. I am not sure if this is true across the Arctic, it depends on the climate.

L255-265: many abbreviations without definitions.

L267-268: 870 mm rainfall and 668 mm snow, so the total is about 1500 m annual precipitation. Is this for Samoylov? This is a lot different than what other studies have reported for example Liljedahl et al. (2016)

L298: “undecomposed organic material features coarse pores” how old is this organic material? Also, is this just an assumption or the organic material is undecomposed at those sites?

L317: 1100 m: +10.2C, does this deep soil temperature also come from borehole measurements?

L320: what is the total depth of the soil column, Table 2 lists properties for 0-9 m but L317 provides soil temperatures with depths of 0-1100 m.

L318: where did this geothermal value come from? Any reference?

Table 2: what is the residual saturation?

L330-335: This is totally confusing, the authors mentioned considering the undecomposed organic matter and now they started talking about peat which has totally different hydraulic and thermal properties. Please explain whether are you still using sandy soil properties for the organic layer and how this part (L330 onwards) is related to L298. I would assume most of the organic material at Samoylov is (partially)decomposed.

L342: “effect of different soil types” on what?

Table 3. If out of 12, 11 scenarios are for the Samoylov site then what is the purpose of including Bayelva site? Both sites should be treated similarly or just remove Bayelva site from the manuscript and focus on one site.

L365: Polygon centers are not representative of the entire polygon. Active layer thickness varies among troughs, rims, and centers. Although ALT varies across microtopographic locations, it would be reasonable to compare it against the average ALT (average of rims, troughs, and centers).

L370-374: Have you looked at the comparison of the observed and simulated snow depths to support your reasoning for the mismatch? Have you compared the forcing data to any nearby climate station? Also, since you have taken the observed soil temperature data from a polygon center, was the polygon center inundated during the fall freeze-up? So it is not just the forcing (snow depth) that could affect freeze-up.

L382: this is confusing again, at L317 it says -7C at 5 m depth, here is it -9C at 5 m depth. Why is that?

Figure 5: How do these results differ from those of Nitzbon et al. (2020)? Other than you not considering the ice-rich zone

Figure 6: What is the focus of this figure, ice segregation or ALT or talik? The section heading says Formation of ice segregation, but the ice segregation and its formation are hard to see (visually) which makes it difficult to understand what is going on. A better comparison here would be to compare simulations with and without the formation of ice segregation and how much will it affect project ALT.

L421-423: this contradicts the formation of more segregated ice in the future. Figure 7 does show that B-clay leads to more segregated ice.

In section 2.2, the authors talked about different simulations, but it is not clear what type of simulations are planned in this work. Please clearly state the set of simulations or summarize them in a table. A simulation description is lacking. It would help better follow the description if this section is split into field sites and forcing data.

Section 3.4. Table 2 shows that the top 15 cm has identical soil properties. What is actually the S-peat scenario? How does it differ in terms of peat thickness in the soil from other scenarios (soil types)? Peat has different thermal and hydraulic properties (L344) than other soil types and leads to shallower ALT, but the thickness of the peat is important. If in summers, the peat keeps the ALT colder than in winters it helps prevent escaping heat from the soil. I am not sure if the entire soil column is peat in the S-peat scenario. If the entire 9m (or so) column has peat, is it realistic?

L586: “complex lateral processes”? Unless you are considering 3D simulations at a larger scale, which are mostly not feasible for permafrost regions, what complexity can lateral processes bring? My guess is even using CryoGrid with three tiles should not make it complex, but just an additional process.

- check for typos

- Map showing the location of the field sites is missing